This page intentionally left blank

Contemporary Chinese Politics

Contemporary Chinese Politics considers how new and diverse sources and
methods are changing the study of Chinese politics. Contributors spanning three
generations in China studies place their distinct qualitative and quantitative meth-
odological approaches in the framework of the discipline and point to challenges
or opportunities (or both) of adapting new sources and methods to the study of
contemporary China. How can we more effectively use new sources and ­methods
of data collection? How can we better integrate the study of Chinese politics
into the discipline of political science, to the betterment of both? How can we
more appropriately manage the logistical and ethical problems of doing political
research in the challenging Chinese environment? In addressing these questions,
this comprehensive methodological survey will be of immense interest to graduate
students heading into the field for the first time and experienced scholars looking
to keep abreast of the state of the art in the study of Chinese politics.

Allen Carlson is an associate professor in Cornell University’s Government

Department. He is the author of Unifying China, Integrating with the
World: Securing Chinese Sovereignty during the Reform Era (2005) and the
coeditor (with J. J. Suh and Peter Katzenstein) of Rethinking Security in East
Asia: Power, Identity and Efficiencies (2004). His articles have appeared in the
Journal of Contemporary China and Pacific Affairs.

Mary E. Gallagher is an associate professor of political science at the University

of Michigan and the director of the Center for Chinese Studies. She is also a
faculty associate at the Center for Comparative Political Studies at the Institute
for Social Research. She is the author of Contagious Capitalism: Globalization
and the Politics of Labor in China (2005), and her articles have appeared in
World Politics, Law and Society Review, Studies in Comparative International
Development, and Asian Survey.

Kenneth Lieberthal is a senior Fellow in Foreign Policy and in Global Economy

and Development and also is director of the John L. Thornton China Center at the
Brookings Institution. He has written and edited fifteen books and monographs,
most recently The U.S. Intelligence Community and Foreign Policy: Getting
Analysis Right (2009) and (with David Sandalow) Overcoming Obstacles to
US-China Cooperation on Climate Change (2009). He is also the author of
about seventy periodical articles and chapters in books and has published in the
New York Times, Washington Post, Los Angeles Times, Financial Times, South
China Morning Post, and numerous other newspapers.

Melanie Manion is a professor of political science and public affairs at the

University of Wisconsin–Madison. Her publications include Retirement of
Revolutionaries in China (1993) and Corruption by Design (2004). Her articles
have appeared in the American Political Science Review; Comparative Political
Studies; Journal of Law, Economics, and Organization; and China Quarterly.
Contemporary Chinese Politics
New Sources, Methods, and Field Strategies

Edited by
Allen Carlson
Cornell University

Mary E. Gallagher
University of Michigan, Ann Arbor

Kenneth Lieberthal
Brookings Institution

Melanie Manion
University of Wisconsin–Madison
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore,
São Paulo, Delhi, Dubai, Tokyo, Mexico City

Cambridge University Press

32 Avenue of the Americas, New York, ny 10013-2473, usa
www.cambridge.org
Information on this title: www.cambridge.org/9780521155762

© Cambridge University Press 2010

This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.

First published 2010

Printed in the United States of America

A catalog record for this publication is available from the British Library.

Library of Congress Cataloging in Publication Data

Contemporary Chinese politics : new sources, methods, and field
strategies / edited by Allen Carlson ... [et al.].
   p. cm.
Includes bibliographical references and index.
isbn 978-0-521-19783-0 – isbn 978-0-521-15576-2 (pbk.)
 1. China – Politics and government – 2002– 2. China – Politics and
government – 2002– Methodology. 3. Political science –
Methodology. I. Carlson, Allen, 1968– II. Title.
jq1510.c68 2010
320.951–dc22    2010018862

isbn 978-0-521-19783-0 Hardback

isbn 978-0-521-15576-2 Paperback

Cambridge University Press has no responsibility for the persistence or

accuracy of urls for external or third-party Internet Web sites referred to in
this publication and does not guarantee that any content on such Web sites is,
or will remain, accurate or appropriate.
Contents

Tables and Figures page vii

Contributors ix
Abbreviations xi

Introduction 1
Allen Carlson, Mary E. Gallagher, and Melanie Manion

Part I Sources
1 State-Generated Data and Contentious Politics in China 15
Xi Chen
2 Why Archives? 33
Neil J. Diamant
3 The Central Committee, Past and Present: A Method
of Quantifying Elite Biographies 51
Victor Shih, Wei Shan, and Mingxing Liu
4 Experimental Methods and Psychological Measures
in the Study of Chinese Foreign Policy 69
Peter Hays Gries
5 Internet Resources and the Study of Chinese Foreign
Relations: Can Cyberspace Shed New Light on China’s
Approach to the World? 88
Allen Carlson and Hong Duan
6 Information Overload? Collecting, Managing, and Analyzing
Chinese Media Content 107
Daniela Stockmann

Part II Qualitative Methods

7 The Worm’s-Eye View: Using Ethnography to Illuminate
Labor Politics and Institutional Change in Contemporary China 129
Calvin Chen
v
vi Contents

8 More Than an Interview, Less Than Sedaka: Studying

Subtle and Hidden Politics with Site-Intensive Methods 145
Benjamin L. Read
9 Cases, Questions, and Comparison in Research
on Contemporary Chinese Politics 162
William Hurst

Part III Survey Methods

10 A Survey of Survey Research on Chinese Politics:
What Have We Learned? 181
Melanie Manion
11 Surveying Prospects for Political Change: Capturing Political
and Economic Variation in Empirical Research in China 200
Bruce J. Dickson
12 Using Clustered Spatial Data to Study Diffusion: The Case
of Legal Institutions in China 219
Pierre F. Landry
13 Measuring Change and Stability over a Decade in the
Beijing Area Study 236
Mingming Shen and Ming Yang, with Melanie Manion
14 Quantitative Research and Issues of Political
Sensitivity in Rural China 246
Lily L. Tsai

Reflections on the Evolution of the China Field

in Political Science 266
Kenneth Lieberthal

Glossary 279
References 283
Index 311
Tables and Figures

Tables

3.1 Tracking Chen Yun’s positions in the 1950s page 55

4.1 Structures of national identity: Pattern matrix loadings
for principal axis factor analysis with Promax rotation
for American and Chinese samples, 2009 79
4.2 Descriptive statistics: Correlations, significance levels,
means, standard deviations, and scale alphas and Ns
for 2009 Beijing sample (minimum N = 156) 81
6.1 Percentage of 100 samples falling within one and two
standard errors of the population mean in two newspapers 115
10.1 Probability sample surveys on Chinese politics resulting
in publications analyzing original datasets 186
10.2 Mainland partner institutions for probability sample surveys
on Chinese politics 191
11.1 The trade-off between goals of development among officials
and entrepreneurs 211
11.2 Perceived threats to stability among private entrepreneurs
and local officials 213
11.3 Distribution of private entrepreneurs in local people’s
congresses 214
11.4 Probit regression: Determinants of private entrepreneurs
in people’s congresses 216
12.1 Conditions for institutional diffusion 230
12.2 Impact of court adopters on the mean propensity
to go to court 231
12.3 Probit estimates of going to court in civil, economic, and
administrative cases 233

vii
viii Tables and Figures

13.1 Questions asked annually in the Beijing Area Study 239

13.2 Overview of sampling and survey implementation
in BAS first decade 243
13.3 Incomplete interviews due to unmet requirements
in BAS first decade 244

Figures

3.1 Average age and 25th and 75th percentile age of

Central Committee members, 1921–2006 61
3.2 Average birth year and 25th and 75th percentile birth
year of Central Committee members, 1921–2006 62
3.3 Percentage of Central Committee members with Long March
experience and the number of Long Marchers 63
3.4 Average education level of Central Committee
members, 1921–2006 64
3.5 The share of Central Committee members with common
experience with Hu Yaobang and Hua Guofeng, 1970–1990 66
4.1 International anxiety as a function of nation, domain,
and frame, 2006 samples 74
4.2 International pride as a function of nation and frame,
2006 samples 75
4.3 Final Beijing path model, 2009 sample 82
6.1 Percentage of Xinhua news articles among all articles
about the United States published in the People’s Daily,
September 2001–July 2002 118
12.1 The ILRC survey of China (2003–2004) 222
12.2 Example of a county map displaying township boundaries
and basic infrastructure overlaid to the Google Earth model,
with grids coded in KML displaying three TSUs in each
sampled township 224
12.3 Example of a spatial sampling unit (half-square minute)
drawn in a low-density rural area in western China 225
12.4 Example of spatial sampling units in an urban area 226
12.5 Impact of the combined presence of court adopters in the
community and the respondent’s level of trust in the courts
on the probability of adopting courts as a dispute resolution
venue in a civil case 234
Contributors

Allen Carlson is an associate professor in the Government Department at

Cornell University.
Calvin Chen is an associate professor of politics at Mount Holyoke College.
Xi Chen is an associate professor of political science at Louisiana State
University.
Neil J. Diamant is an associate professor of political science and Asian law
and society at Dickinson College.
Bruce J. Dickson is a professor of political science and international affairs at
George Washington University.
Hong Duan is a Ph.D. candidate in the Government Department at Cornell
University.
Mary E. Gallagher is an associate professor of political science and director of
the Center for Chinese Studies at the University of Michigan.
Peter Hays Gries is the Harold J. and Ruth Newman Chair in U.S.-China
Issues and director and associate professor at the Institute for U.S.-China
Issues at the University of Oklahoma.
William Hurst is an assistant professor of government at the University of
Texas at Austin.
Pierre F. Landry is an associate professor of political science at Yale
University.
Kenneth Lieberthal is a senior Fellow in Foreign Policy and in Global Economy
and Development and also is the director of the John L. Thornton China
Center at the Brookings Institution.
Mingxing Liu is an associate professor at the China Institute for Educational
Finance Research at Peking University and research Fellow at the China

ix
x Contributors

Economics and Management Academy at the Central University of Finance

and Economics.
Melanie Manion is a professor of political science and public affairs at the
University of Wisconsin–Madison.
Benjamin L. Read is an assistant professor in the Department of Politics at the
University of California, Santa Cruz.
Wei Shan is a postdoctoral Fellow at Duke University.
Mingming Shen is a professor of political science and director of the Research
Center on Contemporary China at Peking University.
Victor Shih is an assistant professor of political science at Northwestern
University.
Daniela Stockmann is an assistant professor of political science at Leiden
University.
Lily L. Tsai is an associate professor of political science at the Massachusetts
Institute of Technology.
Ming Yang is an associate professor of political science and associate director
of the Research Center on Contemporary China at Peking University.
Abbreviations

APA American Psychological Association

APSA American Political Science Association
BAS Beijing Area Study
BB bulletin board
CASS Chinese Academy of Social Sciences
CATA Computer Aided Text Analysis
CC Central Committee
CCP Chinese Communist Party
CFI Comparative Fit Index
CICIR China Institute of Contemporary International Relations
CIIS China Institute of International Studies
CNKI China National Knowledge Infrastructure (database)
CNNIC China Internet Network Information Center
CPPCC Chinese People’s Political Consultative Conference
CSE collective self-esteem
CYL Communist Youth League
FBIS Foreign Broadcast Information Service
FSU Final (spatial) Sampling Unit
GPS Global Positioning System
HSM half-square minute
ILRC Institutionalization of Legal Reforms in China
IQRM Institute for Qualitative Research Methods
IRB Internal Review Board
IWEP Institute of World Economics and Politics
KML Keyhole Markup Language
KMT Guomindang
MFA Ministry of Foreign Affairs
MOE Ministry of Education
MR military region
MVD Ministry of the Interior (Russia)
NBS National Bureau of Statistics

xi
xii Abbreviations

NFI Normed Fit Index

NPC National People’s Congress
NSF National Science Foundation
PAF principal axis factoring
PC party congress
PLA People’s Liberation Army
PPS probability proportionate to size
PSC Politburo Standing Committee
PSU Primary Sampling Unit
RA research assistant
RC residents’ committee
RCCC Research Center for Contemporary China
RMSEA Root Mean Square Error of Approximation
RWA right-wing authoritarianism
SASAC State-owned Assets Supervision and Administration
   Commission
SEC State Education Commission
SIIS Shanghai Institute of International Studies
SIM site-intensive method
SMA Shanghai Municipal Archives
SOE state-owned enterprise
SSU Secondary Sampling Units
TLI Tucker-Lewis Index
USC Universities Service Centre
Introduction

Allen Carlson, Mary E. Gallagher, and Melanie Manion

At a workshop at the University of Michigan in November 2006, three genera-

tions of scholars met to discuss and debate the study of Chinese politics and
how new and diverse sources and methods are changing the field. This volume
is the culmination of that workshop. Drawing on diverse research experiences,
we present a wide range of sources, methods, and field strategies for the study
of Chinese politics in the new era. As political scientists, we place our distinct
methodological approaches in the framework of the discipline and point to
particular challenges or opportunities (or both) of adaptation in the context
of contemporary China. With the main focus on methodological concerns and
the discovery of new data sources, the chapters in this volume are also richly
substantive illustrations that demonstrate how to adapt method to context
innovatively and appropriately. Thus, this book illustrates the benefits of the
emerging cross-pollination between China studies and the broader discipline.
Three major themes emerged from our workshop discussions: (1) how to
effectively use new sources and data collection methods, (2) how to integrate
the study of Chinese politics into the discipline of political science to the bet-
terment of both, and (3) how to deal with logistical and ethical problems of
doing research in a challenging environment. In this Introduction, we discuss
these themes in the sections below in the context of the initial workshop,
the substantive chapters in this volume, and the field more generally. As only
sporadic attention has been paid to the nuts and bolts of the study of Chinese
politics, we hope this volume will spark future debates and other publications,
conferences, and graduate training on research design and methodology in
challenging fieldwork sites. We recognize that this volume joins an existing
ongoing debate (Baum, 2007; Harding, 1994; Heimer and Thøgersen, 2006;
Manion, 1994; Perry, 2007 and 1994b; Shambaugh, 1993; Wank, 1998).
Collectively the following chapters illustrate that although much has changed

The editors would like to thank the American Council of Learned Societies, the Chiang
Ching-Kuo Foundation, the Center for Chinese Studies at the University of Michigan, and the
Department of Political Science at the University of Michigan for their generous support.

1
2 Allen Carlson, Mary E. Gallagher, and Melanie Manion

in the realm of studying Chinese politics, many of the fundamentals that pre-
vious scholars learned about this endeavor still apply. Language skills and
familiarity with China strike us as remaining core prerequisites for scholars
wishing to make sense of any given aspect of Chinese politics. Moreover, local
knowledge – that is, knowing China – is increasingly insufficient. Each of the
contributors to this volume has also utilized a wide variety of research skills
in his or her work. These skills cover a broad set of approaches to politics and
include the use of sophisticated quantitative techniques, the production and
utilization of survey data, the application of new technologies, searching out
and making use of previously closed archival sources, and even conducting
quasi-experiments. Although such approaches cover many tools in the politi-
cal science kit, and are illustrative of the impressive and at times conflicting
directions in which the study of Chinese politics is headed, all contributors
to this volume have made use of such methods with a common purpose in
mind: to amplify their ability to describe and explain key aspects of politics
in contemporary China. As such, the volume shows the rewards of bring-
ing together scholars with diverse backgrounds, yet who share a collective
commitment to pushing both China studies and the discipline forward in an
inclusive and mutually beneficial manner. Thus, although the volume focuses
on mainland China almost exclusively, we believe that the methodology and
research design strategies presented here are relevant to scholars in many other
places around the globe.

An Abundance of Riches? Dealing with Data

The study of China within the discipline of political science has changed
dramatically over the past thirty years, reflecting in many ways the events
and transformations that have occurred in Chinese politics. From a period
of near total isolation from one’s subject of study when China was closed off
from Western scholarship, to a new period in which China’s engagement with
the world has become a source of wonder, political scientists studying China
have gone from a dearth of sources and data to an overwhelming abundance.1
Moreover, this recent surge in the access that scholars have to a staggering
array of sources relating to the Chinese state represents a rather fundamental
change in the way in which scholars come to know China. In other words,
although the use of new methods is laudable and receives a good deal of atten-
tion in this volume, it is also clear that the contributors have been able to apply
more advanced social science techniques to the study of Chinese politics only
as they have gained access to a historically unprecedented wealth of informa-
tion within China relating to domestic politics, foreign relations, and national
security.

1
See Baum (2007) on the generations of political scientists studying China in the post–World
War II period.
Introduction 3

The increasing diversity, amount, and complexity of data on Chinese poli-

tics require that scholars pause to think about and debate how to use the
data effectively and responsibly. The availability of new, often more system-
atic, data presents researchers with new opportunities not only to use these
data effectively but also to combine these riches with more established data
sources. Such opportunities can increase the external and internal validity of
our arguments and also effectively bring the Chinese case to bear on debates
in comparative politics and political science. Several of the chapters in this
volume deal explicitly with strategies for using multiple methods and sources
to achieve these goals. Lily L. Tsai discusses field strategies to improve the
quality and reliability of survey data. Pierre Landry applies new methods
of statistical sampling using Global Positioning System (GPS) spatial tech-
nology, allowing him to show the patterns of legal diffusion and the actual
mechanisms of changing popular opinion toward China’s legal system. Victor
Shih, Wei Shan, and Mingxing Liu present a new database on members of the
Central Committee of the Chinese Communist Party, which permits more
systematic analysis of China’s key political elite. Neil Diamant and Xi Chen
demonstrate the value and increasing accessibility of state-generated data in
Chinese archives.
Changes in sources and data also impact how we interpret and evaluate
older methodologies and sources. Although few China scholars would accept
the comparison to the “Kremlinologists” of old, some of what we study still
necessarily includes attention to and analysis of important facets of Chinese
politics that do not lend themselves to systematic study. Although the elite
politics examined by Shih, Shan, and Liu in Chapter 3 comes immediately
to mind, other important social phenomena, such as collective violence, cor-
ruption, tax evasion, and ethnic conflict, are also critical research topics that
must be studied with limited, often flawed data, and in a political atmosphere
that at times entails overt government suppression and at other times astound-
ingly effective self-censorship on the part of informants, local officials, and in-
country colleagues. Many of the authors here provide detailed explanations of
how they deal with important topics that can yield insufficient or flawed data.
Xi Chen’s typology of “upstream” and “downstream” state-generated data
provides helpful strategies on how to interpret and assess the reliability and
accuracy of government reports and statistics regarding contentious actions
by Chinese citizens. He also details how clear knowledge and familiarity with
the bureaucratic structure and politics of the government units in charge of
monitoring citizens’ collective action are critical for measuring accurately the
value of one’s data. Calvin Chen’s ethnographic study of Chinese factories
provides a critical look at the seamy side of the “workshop of the world”
on China’s southeastern coast. In addition, Neil Diamant’s contribution illus-
trates how digging deeper into previously unavailable archival sources can
shed new light on past events and challenge conventional wisdom about poli-
tics during the early years of the PRC. Many of these strategies should be a
more visible part of any political scientist’s toolbox in challenging fieldwork
4 Allen Carlson, Mary E. Gallagher, and Melanie Manion

locales, in China and elsewhere. Daniela Stockmann utilizes the vast wave of
new print media sources that have flooded China during this period. Although
China’s newspapers and wire reports may be dismissed by some as doing no
more than reproducing official rhetoric, Stockmann shows that via the use of
new technology, specifically Yoshikoder, it is possible to find in these publica-
tions a great deal of new information about trends in contemporary Chinese
politics. Allen Carlson and Hong Duan turn to Internet resources related to
Chinese foreign policy and national security, and although they find that there
is less here than initially meets the eye, these sources too are promising.

Comparative Politics and Comparing China

The study of Chinese politics in recent decades has also been profoundly influ-
enced by the political science discipline. This is evidenced not only by the
ongoing lively debate about appropriate methodology but also by renewed
attention to placing area studies in the broader context of comparative politics.
Increasingly and appropriately, graduate school training in political science
requires acquisition of strong methodological skills, offering new opportuni-
ties for students of Chinese politics. At the same time, application of these
methods requires sensitivity in the field to take into account the different con-
text – a developing economy, an authoritarian polity, and an Asian culture.
There is then a delicate balancing act to be maintained in the training of
graduate students working on China. Increased knowledge of methods (quan-
titative or qualitative) is a must, but so too are language and cultural training.
It is now clear that both of these skills are required, that is, walking on two
legs, to conduct successful research on Chinese politics. The nuances of such
adaptations are not commonly acquired in the classroom.
Moreover, in recent years, the study of specific places (especially single coun-
tries but even specific regions) has been deemphasized in the field of political
science. Whether or not this constitutes progress, comparative politics, and to
a certain extent international relations and security studies as well, now aspire
to develop theories and arguments that can be investigated in and applied
to any locale. Ideally, theories with the greatest amount of breadth should
be developed to explain important political and economic transformations,
for example democratization, rapid economic growth, efficient public goods
provision, and ethnic peace and conflict. In most places around the globe,
comparative work has absorbed traditional “area studies,” and specialists on
a single country or region are encouraged and rewarded professionally when
they show their ability and inclination to go “cross-national.” Although cross-
national comparison has long been a hallmark of comparative politics, the
methods of comparison have changed as better data have become available,
as many recently democratized countries have produced electoral data wait-
ing to be analyzed, and as computing programs have become more powerful.
Combined with the growing emphasis on quantitative research methods in
graduate school training, studies in comparative politics increasingly employ
Introduction 5

large-N datasets and sophisticated statistical analysis to compare countries.

For various reasons, these methods are often regarded as more effective and
powerful than comparative case studies or other small-N comparisons (but
see Schrank, 2006).2
In general, specialists on Chinese politics have not gone down this path,
or at least have not gotten very far down this path. There are several reasons
that the study of China remains somewhat apart from these broader trends.
They include the problem of making relevant comparisons, China’s significant
internal diversity, and the challenge of finding or producing high-quality data
for cross-national comparisons. Given that these problems are not unique to
China and are often present in many other regions and countries, the strate-
gies used to enhance comparison in China may also be applied elsewhere.
The countries most commonly compared to China in earlier periods either
collapsed or democratized (or both) during the 1990s as socialism failed in
countries from Eastern Europe to Central Asia. The end of the Soviet Union in
particular complicated the previously active field of comparative communism.
Although comparisons between China and Eastern Europe, Central Europe,
and the former Soviet Union continued into the 1990s and beyond, apt com-
parisons have become more difficult as the political systems have diverged
markedly. Even in the resurgent authoritarian states of the former Soviet
Union, a number of the political systems remain more democratic and more
open than the one-party state of the Chinese Communist Party. As many stud-
ies across the globe now focus on elections and party politics, China’s one-
party system and dearth of competitive elections beyond the grassroots level
leave China out of many cross-national studies. In some cases, the availability
of systematic electoral data has redirected research away from questions that
cannot be probed this way. As Lieberthal notes in the conclusion, research
questions should be developed that are interesting and relevant rather than
simply because they can be answered through available data. This exclusion
of important questions, and by extension some countries, because the data are
not comparable to those available in developed democracies is regrettable. In
the Chinese case, both quantitative and qualitative research on grassroots elec-
tions, and semicompetitive elections at other levels, have yielded important,
perhaps pathbreaking, insights into the nature of elections in nondemocratic
societies (e.g., Manion, 1996; Shi, 1999a; Tsai, 2007b). However, because
Chinese data remain difficult to integrate into mainstream comparative poli-
tics research on elections (which is overwhelmingly drawn from democratic
countries), the Chinese case does not have a large impact on the field.
Because China does not conform to the path hewed by other socialist states
that experienced socialist breakdown – first economic, then political – before
2
We cannot do justice here to the ongoing, vibrant debate on the strengths and weaknesses of
different methodologies. Despite the increased reliance on quantitative methods in political
science over the past two decades, qualitative methods have enjoyed a renaissance of sorts.
These include comparative case studies, process tracing, ethnography, and others. All of these
methods are highlighted in this volume.
6 Allen Carlson, Mary E. Gallagher, and Melanie Manion

moving on to democratic transitions with varying success, the range of coun-

tries to which China can be appropriately compared remains unclear, particu-
larly as cross-national work often aims to classify countries by regime type.
China’s extraordinary economic successes in the past twenty-five years place
it solidly among the tigers of the developing world. Therefore, comparisons to
the economies of its East Asian neighbors across time or to Brazil, Russia, and
India (the other “BRICs”) today are becoming more common. In its politics,
however, China’s atypical path of sustained authoritarian rule by an unre-
formed Communist Party presents researchers with problems of both theory
and method. To what other nations should China be compared? How should
we accurately code China’s regime type in large, cross-national studies? How
can we avoid the ontological goal of democratic transition when most of our
theories treat democracy in some form as the normal state of politics? In other
words, how can we examine China for what it is rather than for what we hope
it to become?
Second, comparative research on China as a single entity often masks the
remarkable and sustained regional diversity within China itself. Although
aggregate statistics for China demonstrate its economic success, the rapid
decline in poverty, and the impressive numbers in rural-to-urban migration,
urbanization, and industrialization, they often mask the huge and growing
regional inequalities. Chinese coastal cities are now reaching the levels of devel-
opment and standards of living of some of their wealthy developed neighbors
whereas the interior still struggles with high levels of poverty, illiteracy, and
underdevelopment. Economic diversity is matched by cultural, linguistic, and
social diversity that in some cases approaches differences between countries in
other parts of the globe. As William Hurst argues in this volume, scholars of
comparative politics should be encouraged to pursue any kind of comparative
research that yields interesting theoretical and empirical findings. Subnational
comparisons can be as fruitful as some of the cross-national research that
is so highly esteemed in the discipline. Indeed, many of the authors in this
volume utilize China’s rich internal diversity to test hypotheses or to explore
causal mechanisms of general theories in political science about the nature of
economic development, changes in state-society relations, or developments in
civil society.
Finally, even though the quantity and quality of Chinese data have
improved markedly over the years, it remains difficult to find high-quality
data that are easily comparable to data compiled in other countries by inter-
national agencies, national governments, academic organizations, or com-
mercial companies. The Chinese government regards much information as
politically sensitive and continues to obstruct the collection of systematic
political data, broadly defined. In some cases, the government manipulates
data for political purposes, which makes it difficult to be confident about the
accuracy of government figures. As a result, statistical data from China are
rightfully regarded with a healthy degree of skepticism by many researchers.
Numbers, although of utmost importance in studying Chinese politics, have
Introduction 7

to be placed in context, used only when it is clear how they were generated
and for what purposes. Indeed, tracing the origins of statistical data gener-
ated within China is an exercise that can reveal a great deal about how the
modern Chinese state works, as Chen and Diamant show in their chapters on
state-generated data in Chinese archives. Although China is surely not unique
in this regard, the Chinese Communist Party’s attention to the importance of
both information and organization can translate into tight controls over sur-
vey research and access to some government documents and certain archives.
Even some regulations and laws are official secrets. This control continues to
limit the creation of accurate datasets, which in turn reduces the integration
of China into comparative studies.
Although the study of Chinese politics is not wholly integrated into the
subfields of comparative politics and international relations as a whole, the
contributions to this volume show that engagement with the theory and meth-
ods of the discipline are now the norm for political scientists who conduct
research in China today. The problems discussed in this Introduction and
the sense of distance between the field of Chinese politics and the discipline
are diminishing as scholars trained in the historical, cultural, and institu-
tional context of China deploy standard methods of social science research.
The chapter by Peter Hays Gries exemplifies this trend. His scholarship is
grounded on a particularly close read of Asian culture yet is informed by
a highly critical understanding of work in the vein of political culture that
has forwarded rather unsustainable generalizations about differences between
East and West. Rather than simply perpetuating this mythical divide, Gries
explores recent advances in the field of political psychology and then conducts
a series of social experiments in Asian and American settings to determine the
degree to which his subjects “see” the world differently. His findings are then
applied to developing a more rigorous frame for analyzing the role of leader-
ship psychology in the interaction between states (including the United States)
in the Asian security sphere.

Political Research in Challenging Environments

China’s sustained authoritarianism also presents political scientists with
logistical, ethical, and political problems when undertaking research that
touches on sensitive topics, uses new and innovative data collection methods,
or reaches results that may be unsettling or dangerous for powerful domes-
tic interests. As in many other places around the world, studying politics in
China is still difficult, at times dangerous, for researchers and research sub-
jects alike, and is wrapped up indelibly with the practice of politics. This work
highlights the value added of making use of new sources and methodologies.
However, it became clear during our discussions at the workshop that along-
side such accomplishments, there is a need for a more candid discussion of the
trade-offs when doing fieldwork in difficult locations or on sensitive topics.
Although as social scientists we strive for robust internal and external validity,
8 Allen Carlson, Mary E. Gallagher, and Melanie Manion

we are routinely presented with situations that require compromise. There are
risks in the study of Chinese politics. Researchers who strive to gain access to
data that are considered to be “internal” (内部), or related to state secrets, may
put themselves at odds with the Chinese state. This is particularly so when the
research delves into areas of political sensitivity in China (topics that include
ethnic minorities, democratization, religious freedom, etc.). At the same time,
carrying out interviews, conducting surveys, and working with officials to
gain access to archival sources may also put one’s subjects and colleagues in
harm’s way. Thus, although we do not intend to overemphasize these chal-
lenges, at the same time the challenges confronting political scientists working
in China extend beyond the issues of increasing explanatory power.
The logistical and ethical problems of doing research in China have become
more complicated as our research access and opportunities have widened.
As with the treatment of foreigners generally in China, foreign scholars are
now much more autonomous from their official sponsors and somewhat bet-
ter integrated into Chinese society at large. In addition to a more receptive
environment for scholarly work, there are new avenues for collaboration with
mainland scholars and a better infrastructure for large-scale projects – sample
surveys, archival research, and construction of large databases. Such integra-
tion and more frequent collaboration require additional attention to the ethi-
cal problems of social science research, including protection of informants,
attention to the needs and concerns of local collaborators, and striking a
balance between the requirements for human subject protection in Western
universities with the more informal approaches often taken by scholars work-
ing in the field. Several of the chapters here provide effective strategies to
mitigate the problems that occur when one is doing research on sensitive top-
ics. Lily L. Tsai examines interviewing techniques that may reduce response
errors or misunderstandings between survey enumerators and respondents,
particularly with sensitive questions about local government performance,
clan relations, or the enforcement of unpopular policies such as birth control
or tax collection. Bruce J. Dickson shows the importance of the local partner
to ensure on-the-ground cooperation with the survey team. Local partners
and colleagues better understand how topics can be presented to reduce politi-
cal sensitivity, limit self-censorship, and encourage support by local officials.
Benjamin L. Read’s reliance on “site-intensive methods” allows him to gather
information and participant-observation experience at the grassroots level in
urban China, the critical point where citizens encounter the state most often
and most intimately. Without a considerable amount of time and energy spent
intensively studying a few places, Read argues, we often miss the hidden and
subtle aspects of power in an authoritarian regime.

Road Map
One of the more exciting developments in much of the recent scholarship on
Chinese politics is the exploitation of different sources of evidence and multiple
Introduction 9

research methodologies in a single study. Indeed, although we have organized

the following chapters according to their main methodological themes, a num-
ber of them draw from research that illustrates multiple research methods at
their best. We hope this volume contributes further to this development and
to fruitful collaborative relationships among scholars to exploit more fully the
new sources, methods, and field strategies in investigating important ques-
tions of Chinese politics.
The chapters in Part I focus on new sources for the study of Chinese poli-
tics. Chapters 1 and 2 are companion pieces. In Chapter 1, Xi Chen examines
the promise and pitfalls of utilizing xinfang (信访 i.e., petitioning) archives.
This chapter then meticulously outlines a road map of the kinds of documents
available, where they are located, and how accessible they are likely to be, and
also presents a series of strategies for maximizing the chances of successful
use of such materials. More broadly, Chen assesses the reliability of Chinese
archival data. Neil J. Diamant’s Chapter 2 echoes and expands upon this con-
clusion. With a focus on the utility of making more direct use of open archival
sources, he crafts a compelling case for broadening the temporal scope of
Chinese politics, that is, for bridging the divide between historians and politi-
cal scientists. Building on this observation, he demonstrates how newly avail-
able archives that detail aspects of the personal lives of veterans can provide a
new understanding of broad issues in Chinese politics, including controversial
questions such as citizenship and patriotism.
Whereas Diamant and Chen examine the Chinese state in the past, and do
so largely at the level of local politics, in Chapter 3, Victor Shih, Wei Shan, and
Mingxing Liu look to the present and turn attention to the study of elite poli-
tics. Shih et al. show how bringing more rigor to the study of elite politics has
its own rewards. More specifically, they develop a comprehensive database of
the Chinese leadership dating back to the founding of the Chinese Communist
Party in 1921. They argue that previous studies of Chinese elite politics lacked
such a detailed foundation for studying how and why the careers of China’s
leaders progressed (or declined).
Chapters 4 through 6 redirect attention to how the consideration of new
data sources can deepen understanding of Chinese politics. In Chapter 4, Peter
Hays Gries looks for sources in an entirely new direction, mainly by turning
to experimental methods and psychological measures to study Chinese foreign
policy. In so doing, he seeks to push the discussion of political culture’s role
in China’s emerging relationship with the rest of the world beyond the earlier
flawed work in this vein. More specifically, he first outlines the approach he
developed in two separate psychological studies. He then utilizes this work to
inform a broader discussion of the challenges and limitations of experimen-
tal work and psychological measures in the study of Chinese foreign policy.
In Chapter 5, Allen Carlson and Hong Duan examine the apparent surge in
cyber activity related to Chinese foreign policy. They argue that this develop-
ment has been poorly understood by researchers, and, ultimately, has tended
to be overhyped in the field. Rather than finding a revolutionary development
10 Allen Carlson, Mary E. Gallagher, and Melanie Manion

unfolding in China’s Internet space dedicated to foreign affairs and national

security, they uncover a limited set of new sources in this terrain. Although
these data are valuable to scholars, they appear to be less extraordinary than
initially expected. In Chapter 6, Daniela Stockmann casts a broad net to reach
a somewhat different conclusion. She explores the vast number of Chinese
news media sources that are now available online and contends that they pro-
vide scholars with new opportunities to conduct content analysis across media
sources, across regions, and over time. Stockmann explores the problem of
choosing the appropriate sample size for content analysis of Chinese daily
newspapers. Drawing on insights from communication methodology, she
compares the effectiveness and efficiency of various sample sizes for content
analysis in the Chinese context. Although focusing on sample size, Stockmann
also includes suggestions for sampling frames for content analysis that involve
comparisons across media sources, across regions, and over time.
Part II focuses on qualitative methods in the study of Chinese politics.
Qualitative methods of varying kinds have been the hallmark of the study of
Chinese politics since American researchers were allowed back into the field
in the early 1980s. This volume builds on this strong tradition, but the authors
also look beyond the study of Chinese politics to the larger discipline, demon-
strating how the study of politics in China can contribute to larger debates on
the nature of state-society relations in authoritarian regimes (Read), workshop
politics in a rapidly developing economy (Calvin Chen), and the comparative
political economy of unemployment (Hurst).
In Chapter 7, Calvin Chen demonstrates how ethnography is a useful tool
to understand contemporary Chinese politics. Although some scholars con-
sider the approach inadequate to meet such a challenge, Chen suggests that
ethnographic research can go beyond the simple provision of “thick descrip-
tion” and afford a stronger grasp of the multiple and sometimes hidden fac-
tors that trigger, sustain, or obstruct change. In focusing on and dissecting
developments in communities and institutions at the micro level, ethnogra-
phy provides a means for generating deeper insights into how macro-level
forces influence the interactions and lives of ordinary Chinese citizens and
vice versa. Indeed, this approach can aid conceptual development and refine-
ment not only by offering an empirical “reality check,” but also by identifying
and evaluating the factors that contribute to the social and political outcomes
in reform-era China that we seek to explain.
In Chapter 8, Read expands on this approach by arguing for research
designs that incorporate “site-intensive” research. His methodology, used in
a project on the changing nature of residents’ committees in urban China,
combines an ethnographic approach that is broader than a single case but
still not a large-N study. Read argues that this approach is integral to politi-
cal science as it allows researchers to develop new hypotheses, expose causal
mechanisms, and even falsify existing hypotheses in the literature. His argu-
ment builds on the wider literature that has employed these methods, under
different names and in other subfields, including American politics. Read also
Introduction 11

shows that in authoritarian and some cultural contexts where politics is often
hidden, these methods are even more important.
In Chapter 9, Hurst argues for greater attention to subnational variation
across China to develop research questions and effective research designs.
Building on a long tradition in Chinese studies to divide the Chinese polity
into distinct political economies, he develops a cross-regional research project
to explain the wide variation in unemployment policies and outcomes. He
discusses the superiority of this approach over single-city case studies, which
are not broadly representative, and large-N survey projects, which are often
hampered by political sensitivity and unavailable data. Hurst also makes his
case by engaging the literature in comparative politics on case studies and the
comparative method.
The chapters in Part III consider the place of survey research methods in
the study of Chinese politics. In a political environment that remains (at best)
officially skeptical about the enterprise, the number of representative sample
surveys on Chinese politics nonetheless has grown substantially in the past
two decades: political scientists trained and based outside mainland China
conducted only two such surveys in the 1980s, but the number of surveys
increased more than tenfold in the 1990s and continues to rise steadily. In
Chapter 10, “A Survey of Survey Research on Chinese Politics,” Melanie
Manion reviews these surveys and their products, and evaluates their achieve-
ments, with attention to their cumulativeness, contributions to knowledge,
and fit in Chinese area studies. Chapters 11, 12, and 13 present findings as
they discuss methods of exemplary original survey studies of Chinese politics.
In Chapter 11, “Surveying Prospects for Political Change,” Bruce Dickson dis-
cusses how his surveys of private entrepreneurs and local officials make use of
variations across the span of time, geographic region, and key individual-level
characteristics to illuminate in the unusual Chinese context a major question of
comparative politics: the role of capitalists in democratization. In Chapter 12,
“Using Clustered Spatial Data to Study Diffusion,” Landry demonstrates the
advantages of spatial sampling using GPS technology to analyze the diffusion
of formal legal institutions through networks of small communities across
the country. In Chapter 13, “Measuring Change and Stability over a Decade
in the Beijing Area Study,” Mingming Shen and Ming Yang, with Melanie
Manion, examine the first decade of the Beijing Area Study (BAS), an ongoing
annual representative survey of Beijing residents, conducted since 1995. The
measures of change and stability in the unique longitudinal perspective of the
BAS allow the authors to gauge the underlying temper of Beijing residents,
isolating general patterns of confidence and satisfaction with particular trends
of discontent. In Chapter 14, “Quantitative Research and Issues of Political
Sensitivity in Rural China,” Tsai draws from her qualitative fieldwork and sur-
vey research in the Chinese countryside to discuss the implications of political
sensitivity for the process and products of our research on Chinese politics,
and for our colleagues and interview subjects who remain after we leave to
begin the work of evaluating evidence and analyzing data.
12 Allen Carlson, Mary E. Gallagher, and Melanie Manion

Finally, in a concluding chapter, Kenneth Lieberthal reflects on the evolu-

tion of the China field in political science. Lieberthal describes the obstacles
confronting the study of Chinese politics over the decades and celebrates the
current methodological richness and institutional maturity of the field. At the
same time, Lieberthal cautions scholars about new problems that will require
our ongoing attention if we are to continue to contribute to a broader and
deeper understanding of Chinese politics.
Part I

Sources
1

State-Generated Data and Contentious Politics in China

Xi Chen

China specialists in the reform era are blessed with much better ­accessibility
to data than previous generations (Perry, 1994b: 704–713). In particular,
much information that was collected and maintained by governmental agen-
cies has now become available to researchers. However, because of scholars’
unfamiliarity with such sources, uncertainty about collecting them, and long-
existing skepticism of their reliability, so far very few scholars have taken full
advantage of this opportunity.1
In China, many topics on state-society relations are regarded as “sensitive”
by the government. Consequently, researchers are faced with various restric-
tions on their empirical inquiries. Those who study popular contention often
find themselves in a difficult situation: interviews or surveys are difficult to
arrange, and little information can be found in the media. Under such circum-
stances, governmental materials, if available, often provide the best sources
for the study of state-society interactions. After all, for a long time the party-
state not only monopolized political power but also tightly controlled the flow
and distribution of critical information.
This chapter will address two major concerns about collecting and using
state-generated data: accessibility and reliability. The accessibility of such data
is often underestimated. In the reform era, the Chinese government has actu-
ally relaxed its control over archival materials. Many documents are published
or are available to the public in national or local archives. Various govern-
mental agencies are also willing to share certain materials with researchers.
However, the entire archival system is still severely underinstitutionalized, and
researchers still face considerable uncertainty about what might be found.
There are also some pitfalls and biases in state-generated data. The reli-
ability of governmental data has long been a concern when studying one-
party authoritarian regimes. It is generally believed that such political systems
seriously constrain the flow of information, not only between the state and

1
Two notable exceptions are Perry (1994a), and Walder and Su (2003).

15
16 Xi Chen

society but also within the bureaucratic system. This will certainly affect the
data we collect. In order to evaluate the reliability of governmental materials,
we need to understand the political processes by which such materials are gen-
erated. As some sociologists have noted, “the more we know about the struc-
ture and sources of bias in our data, the better prepared we will be to avoid
erroneous interpretations of its patterns” (Earl et al., 2004: 77). Therefore,
this chapter also examines the structures and processes in which state agents
produce archival materials. In particular, we will discuss two sorts of con-
straints: features of the political system and the process of government-cit-
izen interactions. As Charles Tilly suggests, social researchers need not only
develop theories explaining the phenomena under study but also develop the-
ories explaining the generation of evidence concerning the phenomena (Tilly,
2002: 248–254).
This chapter mainly draws on my own experience of collecting and using
data from the petitioning ( 信访 ) system. During my 2002 field research, I
visited four petitioning agencies in Hunan province, including one provincial-
level bureau, two prefecture-level bureaus, and one county-level bureau. On
my 2008 field trip, I visited several local archives in Guangdong, Hunan, and
Hubei. Data generated by other state agencies may not be similar to the peti-
tioning data in every respect, but a discussion of petitioning data can certainly
help us think about how best to work with state-generated data in general.

Two Advantages
Compared with interviews, surveys, and newspapers, state-generated data
have several advantages for the study of contentious politics. Here I will high-
light two of them. First, state data can bring the state into focus for our under-
standing of state-society relations. Second, they make it possible to conduct
event analysis, which has made significant contributions to social movement
theories in recent decades.
Most scholars of contentious politics in China have focused their empirical
studies on social groups rather than on the government. This is understandable,
as it is considerably easier to conduct interviews or surveys involving societal
actors. To be sure, it is possible to interview or survey government officials,
but such interviews or surveys are not easy to arrange and the researcher also
faces the difficulty of eliciting candid and informative answers. Those who
study popular contention often find a stark contrast between the attitudes of
protesters and the attitudes of government officials. Most protesters are eager
to talk to researchers – one of their greatest grievances is that no one listens
to them. By contrast, government officials are generally reluctant to engage in
conversations of any substance on sensitive topics with researchers, especially
researchers from foreign countries. With more officials being reprimanded
for talking to the media and to others, government officials in most places in
China have become increasingly cautious.
State-Generated Data and Contentious Politics in China 17

Governmental data may be a shortcut. If available, they can provide

rich information at a relatively low cost. In particular, they can shed light
on the government as an actor in contentious interactions. This is especially
important considering that the dominant theoretical paradigm of social
movements – political process theory – largely attributes dynamics and out-
comes of popular mobilization to the features of the state and their changes.
Governmental data, including meeting minutes, leaders’ instructions, com-
munications among agencies, and investigation reports, can help us open the
black box of the bureaucratic system. For instance, the governmental data
that I collected in Hunan province enabled me to conduct systematic studies of
governmental responses to popular contention (Chen, 2009; Chen, 2008).
Governmental data not only help us treat the state as an actor but also
show us how the state views the relationship between the government and
its citizens. Rich governmental archival materials often provide us with a
feeling of “seeing like a state” (Scott, 1998). Like social scientists, the state
is curious about some social facts and often spends abundant resources to
investigate them. State perspectives can help to enhance our understanding
of social phenomena. Of course, this is not to suggest that such perspectives
are necessarily superior to academic views, but they are often inspiring and
illuminating. Take the study of the rising trend of popular contention as a case
in point. To explain the dramatic rise of popular contention in China since
the 1990s, many studies focus on specific social groups, such as workers or
peasants. This is not a problem in itself. But some of these studies tend to over-
emphasize group-specific conditions and grievances and fail to understand
that such social change is a cross-sectoral phenomenon. Government data can
help researchers avoid such narrow perspectives since they are concerned with
change in the society as a whole. The petitioning system has tracked the rise of
collective action for every notable social group. For example, in Hunan prov-
ince, petitioning data show that social protests on a variety of issues, from
pension arrears to environmental pollution, grew substantially and simulta­
neously from 1994 to 2001.
Although governmental data do have distinct advantages over interviews
and surveys, they work best when used together with other data. The state
shares a sense of curiosity with social scientists, but its main interests are
somewhat different. For example, when the state reports the behavior of pro-
testers, it does so mainly for the purpose of surveillance, and it is very likely
to ignore some facts that researchers may find interesting. No matter how rich
and comprehensive the governmental materials are, researchers often still find
them inadequate for systematic studies. It is therefore necessary for them to
collect additional data. Yet, researchers can benefit greatly from governmental
data when conducting interviews or surveys. For example, with a complete
list of petitions and protests from a certain period, researchers can select and
locate interviewees and respondents in a rigorous way. In addition, govern-
mental data can also help researchers frame questions more effectively. In
18 Xi Chen

fact, interviewees are generally more willing to talk candidly when researchers
appear to have a good grasp of the issues. This is especially true for interviews
with government officials.
Another major advantage of Chinese governmental data is that they make
event analysis possible. Although archival data are often associated with
qualitative methods, they are also suitable for quantitative research or mixed
methods, such as content analysis and event analysis.2 For researchers of con-
tentious politics, event analysis is a particularly important research tool. The
methods of collecting and processing data have contributed significantly to
the shifting of theoretical paradigms in the field of contentious politics. The
two most important theories in recent decades – the resource mobilization
and political process models – have substantially benefited from the method of
event analysis, which Charles Tilly and his associates pioneered in the 1960s.
Based on extensive protest event data collected mainly from newspapers, a
number of scholars have conducted cross-regional and cross-period quantita-
tive studies. As the Tillys remarked more than thirty years ago, this method
can provide a sounder procedure than “piling example on example, citing
informed observers, or reporting strong impressions” (Tilly, Tilly, and Tilly,
1975: 16).
A critical procedure in this method is to compile event catalogs. Such cata-
logs provide information about occurrence, timing, intensity, sequencing,
outcomes, and other aspects of claim-making events. With such information,
researchers can answer some important questions about the influence of tim-
ing, organizational dynamics, and the political environment on the rates of
collective action and on the success of social movements (Olzak, 1989: 120).
According to Tilly, analysts can use such catalogs in three ways: aggregate,
incidence, and internal regularity. First, they can aggregate counts or selected
aspects of events into overall measures for times, places, or social categories,
and then attempt to explain variation over time, place, or social category.
Second, they can also study incidence by examining “whether distinguishable
features of the phenomenon measured by event catalogs co-vary with charac-
teristics of settings, participants, or associated events.” Finally, they can use
such catalogs to search for internal regularities, such as recurrent sequences
or causal links among apparently separate events. The last task is particularly
demanding for the data. Analysts have to “break down and recombine narra-
tives of episodes and descriptions of their settings into elements that analysts
can then reassemble into representations of the associations or causal connec-
tions they have theorized” (Tilly, 2002: 252).
Most scholars have used newspaper data rather than governmental data. In
their study of Germany, Italy, and France, the Tillys found that “a continuous
run of a national newspaper is a somewhat more reliable source (and a more
practical one) than any major archival series we have encountered, a much
more reliable source than any combination of standard historical works, and

2
For a good example of content analysis, see Shapiro and Markoff (1998).
State-Generated Data and Contentious Politics in China 19

superior to any other continuous source it would be practical to use” (Tilly,

Tilly, and Tilly, 1975: 16). Indeed, newspaper data are often better than gov-
ernmental data in terms of scope and reliability. We can hardly expect to find
governmental archival materials that can match such newspapers as the New
York Times in these two respects.
However, there are circumstances in which governmental archival materials
may be a better source. For example, in his study of strikes in Russia, Graeme
Robertson mainly relied on a protest dataset compiled from unpublished daily
text reports of the Interior Ministry (MVD). In order to evaluate the data’s
reliability, he compared them with the published data and found that “they
are a considerable improvement on published strike statistics” (Robertson,
2007: 790). This is because the MVD has both motivation and resources
for writing reliable reports, which are primarily used for national security
purposes, compiled by officials responsible for actively monitoring levels of
disruption in the areas under their supervision. Consequently, “comparisons
between the MVD data and official sources show the same macro-trends
over time while suggesting higher levels of activity than published sources”
(Robertson, 2007: 791).
Likewise, in contemporary China where the media are under very strict
constraints regarding reporting popular collective action, governmental data
are clearly superior to newspaper data. To be sure, we can still find sporadic
reports of popular collective action in the international and domestic media as
well as on the Internet. Careful researchers may collect possibly hundreds of
such reports. However, these reports can hardly be used in a systematic way as
most such events took place in settings that are unrelated. A more useful strat-
egy is to conduct an analysis of a smaller scope but with more solid evidence.
For instance, because the event catalog in City Y, which includes all reported
events in an internal circular for ten years, provides a variety of informa-
tion about protest events, such as the number and identity of the participants,
forms of action, protest tactics, grievances, ways of framing, and government
response, it can be used to answer a wide range of questions. With a statistical
study of this dataset, I examined which protest tactics enhance the likelihood
of a substantial Chinese government response (Chen, 2009).

Where and What to Look For?

Scholars can access state-generated data from three main sources: (1) pub-
lished materials, (2) national or local archives, and (3) other central or local
governmental agencies. In addition, there are some less common ways to
locate governmental documents. For example, governmental materials can
sometimes be garnered from petitioners. Petitioners are often well aware that
such materials may enhance their bargaining power. Despite the government’s
efforts to withhold critical information, on occasion petitioners have managed
to obtain governmental documents. Even many classified internal documents
may be circulated among petitioners. Petitioners may sell them or share with
20 Xi Chen

other petitioners for free. Some documents are highly informative. A notable
example is an internal speech by Deng Pufang, chairman of the Federation of
Disabled People in China, in March 2000. Handicapped petitioners in many
provinces used this document to exert pressure on local governments in order
to obtain favorable policies.
Accessibility is a big advantage for governmental publications. For the field
of contentious politics, the most useful among such publications are the vari-
ous county gazetteers and provincial yearbooks. One reason that such pub-
lications can be quite informative is that the boundary between publications
for internal use and publications for external use is sometimes vague and fluid.
Take the county gazetteers, for example. Although they were originally com-
piled for internal use, since the 1980s they have been available for public use
(Thøgersen and Clausen, 1992; Vermeer, 1992). A local governmental docu-
ment states that the primary reason for compiling gazetteers ( 地方志 ) was
“to provide the local leaders with scientific information so that they can make
correct decisions” (Thøgersen and Clausen, 1992: 165). Shortly after publica-
tion, however, the gazetteers were freely circulated in China. The vagueness
as to the audience allows for some degree of openness.
Some local governments are more open than others. As Walder and Su
note, the interpretation and implementation of general regulations on infor-
mation openness is primarily a matter for the local authorities (Walder and
Su, 2003: 80). If we compare the Henan yearbooks with the Hunan yearbooks
in the 1980s and 1990s, we find that those from Henan include much richer
information about petitions and appeals. Of course, even the most informa-
tive publications omit information that can be found in local archives or other
governmental agencies. Hence there is a trade-off: governmental publications
are most accessible, but they are the least informative on sensitive issues.
Accessibility to local archives is usually not a problem. The Chinese govern-
ment has evidently improved its archival system since the 1990s. The Archives
Law was amended in 1996, and a new version of the guidelines was promul-
gated by the State Council in 1999. Most, if not all, local governments at the
county level and above have set up local archives. In recent years, most local
archives have not only increased their collections but have also opened more
materials to the public. Of course, despite the Chinese government’s efforts to
institutionalize the archival system, local archives vary considerably in their
interpretation of what should be included in their archives. Thus, it is still
highly unpredictable as to what is maintained in the local archives and what
researchers are allowed to read and photocopy. Some local archives maintain
only very old and incomplete materials because other governmental agen-
cies are reluctant to transfer their archival materials to them. Even for those
archival materials that are supposed to be open, some special restrictions
may be imposed. For example, in the archives of Zhijiang City, Hubei prov-
ince, researchers who want to look up archives from a specific agency need to
obtain the approval of that agency. Therefore, if a researcher wants to read
archives that were transferred from the petitioning bureau, he or she needs
State-Generated Data and Contentious Politics in China 21

to be approved not only by the archives but also by the petitioning bureau.
The situation is even more complex for foreigners. In some local archives,
however, researchers may find relatively new and quite complete archives. In
my trip to the municipal archives of Yiyang City, Hunan province, in 2008,
for instance, I was astonished to find that many “sensitive” internal govern-
mental documents from the petitioning bureau dating as late as 2000 were
open to the public. Among the most valuable was an internal circular called
“Petitioning Express,” which was originally intended to be provided only to
top local leaders.
Archival materials in other governmental agencies are more difficult to col-
lect. Generally, the higher the level of the agency, the more reluctant it is to
open its filing cabinets to researchers. Therefore, it is more difficult to collect
materials from provincial agencies (and even more so from the central govern-
ment) than from prefectural-level or county-level agencies. Scholars who are
lucky enough to access archival materials directly from governmental agencies
can expect to find a variety of data. The following is a brief description of what
can be found in the cabinets of the petitioning bureaus: (1) registration forms
for letters and visits; (2) case files that include petition letters, meeting min-
utes, and investigation reports; (3) governmental reports on petitioning events;
(4) periodic summaries and analyses of petitions, work reports by petitioning
officials, and speeches by local or central leaders concerning petitioning work;
and (5) statistical data compiled by the petitioning bureaus. Regarding the
extent to which these materials have been processed, they can be roughly clas-
sified into raw materials (such as petitioning letters), partially processed prod-
ucts (such as reports of events and summaries of petitions), and well-processed
products (such as work reports and statistical data). They are also officially
classified into short-term, long-term, and permanent archives according to
how long the government intends to keep them. Petitioning bureaus can keep
the long-term and permanent archives in their own offices for a certain period
of time and then send them to the local archives. According to the Archives
Law, county-level bureaus can keep their archives for ten years before trans-
ferring them to the local archives. In practice, however, some bureaus may
keep them for an even longer period. Most raw materials are regarded as
short-term archives and will be disposed of within a year or so. Consequently,
they cannot be found in the local archives.
Each of these five types of materials has its strengths and weaknesses. From
the first type of material, registration forms, researchers can establish a com-
prehensive database about all petitions addressed to the petitioning bureaus.
Yet the information about each case is limited. There is little information
about how the petitioners made their claims and about how the government
responded. The second type of material, case files that provide rich informa-
tion on each case, is suitable for more in-depth case studies. With the petition
letters, registration forms, leaders’ written instructions, meeting minutes, and
investigation reports on file, one is able to reconstruct the entire process of
contentious interactions.
22 Xi Chen

The third category of material is especially valuable for event analysis.

It usually takes two main forms: (1) special reports sent to local leaders
that are titled “important petitioning issues for review,” and (2) a special
internal circular published by almost every petitioning bureau at each level.
It is worth noting that most petitioning bureaus simultaneously publish
two different types of circulars. One is purely for internal use whereas the
other is more public. For example, in Tianjin City, the first type is called
Petitioning Information ( 信访信息 ), and the second Tianjin Petitioning
( 天津信访 ), which is a monthly journal. The internal circular is usually much
more informative than the public circular. In City Y, the internal circular is
issued regularly, about two issues each week. Yet during important periods,
such as during the local people’s congress annual meetings, it is published
almost daily.
The fourth and fifth types of materials are well processed by petitioning
officials to inform local leaders or the upper authorities about the basic facts
regarding petitions and protests. They often include a large amount of infor-
mation that is convenient to use. But they have two shortcomings. First, as
well-processed materials, they are more vulnerable to distortion. Second, they
are not processed as academic research and therefore cannot be used directly.
For example, many statistical forms use a variety of official jargon and many
basic terms are not clearly defined. Therefore, researchers need to interpret
these materials carefully before using them for academic research.

The State, Information, and Petitioning Data

Like their accessibility, the reliability of governmental data is also cause for
concern among scholars. Compared with other data, it is well recognized that
using governmental data requires special caution. As Paul Starr rightly points
out, “Administrative data are also particularly sensitive to vagaries of bureau-
cratic policy or procedure unrelated to the external social phenomenon they
may be taken to measure. For example, changes in budgets, personnel, and
internal incentives and controls such as quotas for cases opened or closed are
likely to influence administrative measures of crime, illness, and other forms
of civil deviance” (Starr, 1983: 30). Similarly, Earl et al. argue that reporting
may be particularly susceptible to error and bias when information about an
event is collected from authorities (or participants) because “these actors often
have a stake in how the event is portrayed” (Earl et al., 2004: 73). Unlike news
agencies, the governmental agencies that produce the data are also actors in
state-society or intrabureaucratic interactions. In this sense, using governmen-
tal data requires more caution than using newspaper data.
A variety of events in the history of the People’s Republic of China (PRC),
from the Great Leap Forward to the recent severe acute respiratory syndrome
(SARS) crisis, suggest that China specialists may have more reason to be cau-
tious. Even Chinese leaders seem to be well aware of the unreliability of gov-
ernmental data. Yao Yilin, the director of the General Office of the Central
State-Generated Data and Contentious Politics in China 23

Committee from 1980 to 1982, expressed his unhappiness with the existing
channels of information:
The most difficult thing for a leadership unit to do is to collect accurate information
at the basic level. Various departments also conduct their own investigations but their
investigations sometimes have a departmental bias or partiality. Now the reports from
the localities are written after repeated deliberations and they have been perfected so
that you cannot see anything in them (Huang, 1995: 832).

The tendency toward distortion or even falsification of information in the

Chinese political system has been well discussed by academics. For exam-
ple, Richard Baum and Alexi Shevchenko argue that “lower-level officials in
Leninist systems have a strong incentive to lie to their superiors, hiding assets
and under-estimating capacity, the quality of information available to leaders
in such systems is generally poor” (Baum and Shevchenko, 1999: 337).
However, we should not exaggerate the problems of governmental data.
After all, except for a few tumultuous periods, the mechanisms for informa-
tion generation and circulation have worked at least well enough to sustain the
political system. In order to accurately evaluate the quality and reliability of
government information, we need to examine the institutional structures and
political processes for information generation and circulation. In particular,
we need to determine why and how the state agencies have produced the data
that we are going to use.
Why does the Chinese government produce so much petitioning informa-
tion? Every state, especially a modern state, is hungry for information about
social facts. This is why James Scott claims that the legibility of the society is a
central problem in statecraft (Scott, 1998: 2). Similarly, Paul Starr notes, “that
the word statistics comes from the same root as ‘state’ testifies to an impor-
tant stimulus of development: the demands of the modern state for social and
economic intelligence” (Starr, 1983: 15).
Different regimes may have different strengths and weaknesses for collect-
ing information. Authoritarian regimes such as China have three advantages.
First, they often have strong power to simplify and standardize the society
to make it more legible (Starr, 1983: 15). Second, some regimes, especially
totalitarian regimes, have a strong capacity to mobilize ordinary people to
provide information. Third, their deep penetration into the society also facili-
tates surveillance. In China, grassroots institutions such as work units have
long worked as state agents to supervise the population (Walder, 1986).
However, these advantages can sometimes turn into disadvantages. For
instance, deep penetration of the society and an all-encompassing structure
may actually hinder the collection and flow of information. When most social
institutions are under its direct control, it is difficult for the party-state to find
any independent sources of information. Moreover, the lack of protections of
freedom of speech and of freedom of the press further exacerbate this prob-
lem. The media, supposedly the most important source of external informa-
tion, usually remain silent on critical issues.
24 Xi Chen

Small wonder then that many authoritarian regimes rely heavily on citizen
complaints as sources of information. In order to redress their grievances,
petitioners have a strong incentive to provide information to the state. Also,
because most complaints only bear on trivial issues, relatively few political
risks are involved in making complaints. Therefore, Liu Shaoqi, former presi-
dent of the People’s Republic, once commented that he felt more comfortable
reading petition letters than reading bureaucratic reports (Diao, 1996: 157).
Generally, the state collects two types of social information: one focuses
on the subjective state of the population whereas the other focuses on purely
objective facts, such as demographic and economic statistics. The informa-
tion at stake in the SARS crisis was mostly objective. For example, how many
SARS cases had been found? The information provided by petitioners is not
confined to objective facts and also may include subjective facts. It is especially
important for the state to understand the feelings and opinions of its citizens.
The party-state in China collects information from petitions and appeals for
three main purposes: responsiveness, accountability, and surveillance. Those
authoritarian regimes that care about political responsiveness work hard to
collect information about popular preferences, albeit in a different way from
that of liberal democracies. When little reliable information is available from
elections and public opinion polls, petitions and appeals constitute an impor-
tant source of information about popular preferences. Another purpose is to
hold governmental agents accountable. With few alternative sources, such as a
reliable media, the party-state must identify the misdeeds of its agents through
negative feedback from petitioners. For example, various statistics indicate
that this channel has provided clues for at least 70 percent of the corruption
cases taken by legal enforcement agencies in any period of the PRC.3 Finally,
the party-state can also garner information about the most volatile segments
of the population by monitoring the petitioning activities. This is because
many petitions or protests have the potential to develop into more rebellious
action and violence.
Although petitions and appeals are regarded as external information, by the
time they reach the party and government leaders, they have already been pro-
cessed by the bureaucracy. Therefore, like media information in China, such
so-called external information sources are also subject to the constraints of
bureaucratic politics. At least two features of the political system may affect the
tendency for information falsification and distortion: alternative information
sources and responsibility systems. When the leadership relies heavily on one
particular information source and has no alternatives, the situation can be best
described as “information dependence.” Wang Shaoguang, writing on China’s
fiscal system, attributes information manipulation to such a dependence. As he
explains, “the center has to rely upon agents spread widely across the nation
for information. In a hierarchical system, those at the bottom have little choice
but to pursue their objectives by manipulating the supply of information to the

3
For example, see Diao (1996: 39).
State-Generated Data and Contentious Politics in China 25

center. Therefore there is always the danger of information transmitted to the

center being distorted” (Shaoguang Wang., 1995: 87). Furthermore, there is a
variety of local responsibility systems in China that impose rewards and sanc-
tions based on the information generated within the bureaucracy. In particular,
it is widely accepted among Chinese officials that “numbers affect positions.”
Therefore, we can make two assumptions. First, a higher level of infor-
mation dependence implies a stronger tendency for information distortion.
Second, when the information is more closely tied to political responsibility,
the tendency for information distortion is higher. Fortunately, judging by
these two criteria, information distortion has not been a big problem for most
governmental data from the petitioning system since the 1990s. The petition-
ing system is only one of many information channels. As to claim-making
activities by ordinary people, there are at least five agencies that routinely
provide such information to leaders: (1) the police department, (2) the Social
Stability Maintenance Office ( 维稳办 ), (3) the Politics and Law Committee
( 政法委 ), (4) the petitioning system, and (5) the 610 Office for activities
related to Falungong followers. In addition, one of the most striking fea-
tures of the petitioning system is that there are many “skip-level” petitions.
Petitioners may deliver their petitions not only to local governments but also
to their superiors. Because the petitioning system is not the only agency pro-
viding information, and petitioners often send information directly to upper-
level governments, it is more difficult for the petitioning agencies to hide and
distort information.
Second, the connection between petitioning data and political conse-
quences does not create particularly strong incentives for information dis-
tortion. Although the Chinese government holds local governments strictly
responsible for handling petitions and protests, the mere occurrence of peti-
tion and protest events does not warrant reprimand; political consequences
result only when such events are handled poorly. To be sure, at times many
local governmental policies have viewed high rates of petitions and complaints
as a negative reflection of the performance of local leaders. In recent years,
however, there has been an attempt to detach responsibility from the local
government for petitions and protests. This became a trend particularly after
Premier Zhu Rongji’s speech in 1996 when he claimed that a higher frequency
of petitions and appeals was not necessarily a negative indication.4 Rather, it
might suggest that local leaders have a more democratic work style and that
ordinary people trust them. Of course, leaders in different localities still have
different interpretations. For instance, some types of petitioning activities –
such as repeated petitions over a long period – may indicate that the local
governments are unresponsive or otherwise flawed. In general, however, the
motivation for manipulating information about petitioning activities in gov-
ernmental reports for internal use is not overwhelming.

4
Comrade Zhu Rongji’s Opening Speech at the Fourth National Conference on Xinfang Work,
1996. This is a document kept by the Xinfang Bureau of County H in Hunan province.
26 Xi Chen

This is not to suggest that we can ignore the information distortions caused
by bureaucratic politics. As has been widely observed, protesters tend to over-
report, whereas governmental agencies tend to underreport. Protesters often
exaggerate the number of participants in order to demonstrate their strength or
legitimacy. By contrast, governmental agencies, such as the police, are likely to
underreport events or participants in order to reduce the impression of social
disorder. This rule also applies to the petitioning system. For example, when
I visited the petitioning bureau in County H, Hunan province, I noticed that
not every large-scale protest had been reported as required. An official tried to
justify such behavior: “That event should not count as an event with over fifty
petitioners since most of the participants were only onlookers. They came to
the county seat to go shopping, and only unintentionally gathered before the
county government.”5
Indeed, a general discussion of the reliability of petitioning data is insuffi-
cient since some types of data are more reliable than others. For instance, raw
materials in petitioning bureaus suffer relatively minor problems. Registration
forms or case files are seldom distorted. By contrast, those well-processed data
are more likely to be manipulated.
Bureaucratic constraints are not the only challenges to information genera-
tion and circulation. The contentious interactions between government and
petitioners may also affect the process. Historically, it is not uncommon for
the state’s search for social information to result in political contention. For
example, many premodern censuses in Western societies met with strong resis-
tance (Kertzer and Arel, 2006: 665–666; Starr, 1983: 15). Also, as mentioned
above, one of the foci of struggles between government and petitioners in
China is information about relevant policies and about the petitioning activi-
ties of other groups. Therefore, consideration of the possible impact on social
actors may also constrain the process of data generation.
For instance, the method of categorization can affect the behaviorial pat-
terns of the petitioners. Petitioning data use categories such as “collective
visits,” “skip-level visits,” “repeat visits,” and “abnormal visits” to identify
petitioning activities. Such categories in fact constitute the basis for norms
concerning permitted and forbidden activities. Indeed, the methods of data
collection often involve trilateral interactions among upper-level authorities,
lower-level authorities, and petitioners. When upper-level authorities imple-
ment a certain statistical method, they must often consider the impact on
the two other actors: their subordinates and the real or potential petitioners.
They are thus often caught in a dilemma – data collection for the purpose of
­exerting pressure on lower authorities may be exploited by the petitioners, and
yet the neglect of such statistical work may reduce the sense of responsibility
of the lower-level authorities.
The Hunan Provincial Bureau once experimented with abolishing the
counting of repeated petitioning activities by so-called long-term petitioners.

5
Interview, July 2002.
State-Generated Data and Contentious Politics in China 27

Such activities had been regarded as a negative factor since a large number of
repeated petitions to the upper authorities was thought to indicate the fail-
ure of local governments; the provincial bureau’s counting of such activities
therefore exerted pressure on the local governments. However, this method
was then exploited by savvy “long-term petitioners.” They understood very
well that each time they petitioned the provincial government their action
reflected negatively on the local government. Therefore, to prevent such activi-
ties, the local governments were under greater pressure to negotiate with these
petitioners. Conversely, if these activities were not counted, they would not
have such leverage. To eliminate the incentive for long-term to visit the upper
authorities, the Hunan Provincial Government decided not to count such
activities. Yet the policy lasted for only a few weeks, and the previous practice
was reinstated. Although the intended effect had largely been achieved (weak-
ened incentives for petitioners to appeal to the upper authorities), the policy
had a strong and undesired side effect: lower-level authorities began to work
less diligently to prevent long-term petitioners from petitioning to the upper
authorities. In sum, the process of data generation about contentious politics
is itself a part of contentious politics. A good understanding of intrabureau-
cratic and government-citizen interactions is therefore helpful for us in evalu-
ating the availability and quality of the data.

The Petitioning System and Its Adaptations

After a discussion of the general political process of data generation, we
can now shift our focus to the specific governmental agency that collects,
processes, and maintains petitioning archives. A closer examination of the
petitioning system, especially its organization and operating procedures, will
further help us evaluate the quality and reliability of petitioning data. In gen-
eral, petitioning bureaus with strong incentives and abundant resources can
produce reliable and high-quality data, but the converse may also apply. The
evolution of the petitioning system has been a process that integrates institu-
tional continuity and change. On the one hand, the petitioning system’s basic
functions and its relationship to the local leadership have hardly changed. On
the other hand, since the dramatic rise of popular contention in China in the
1990s, the petitioning system has undergone remarkable adaptations. In order
to handle petitions and protests more effectively, the organizational structures
have been adjusted and some new procedures have been implemented. Such
continuities and adaptations have had a significant impact on the data.
Petitioning agencies still play a role in the bureaucratic system as it was
designed more than fifty years ago. As part of mass-line politics, the peti-
tioning system was designed to facilitate communications between party
leaders and the masses. Therefore, petitioning agencies are highly dependent
on party-state leaders. Indeed, they were never meant to fulfill any function
other than that of assisting the leaders. The primary goal of their information
work is to serve the local leadership. Each year the information materials in
28 Xi Chen

the petitioning bureaus are evaluated, and the officials who produce “useful
information” are praised or rewarded. The primary criteria for evaluating
“useful information” are whether the materials have attracted the attention
of the local leaders and whether written instructions have been issued based
on them. In the relationship between 条 / 部门 and 块 / 政府 , 块 is much more
important for the petitioning agencies (Lieberthal, 1995; Unger, 1987).
Such dependence of petitioning bureaus on local leaders has caused con-
siderable variation in the quality and style of petitioning work across locali-
ties. In those areas where local leaders care more about the petitioning work,
petitioning agencies enjoy more resources and tend to produce better data. In
contrast, petitioning bureaus in some localities are clearly poorly equipped
and staffed. In one county bureau that I visited, no official was specifically
assigned to handle petitioning information. Although most of the original
case files were maintained, few reports or statistical data were produced.
More important, local leaders also tend to have different understandings
about whether information on petitions will affect their performance evalu-
ations. Those leaders who still regard a high rate of petitioning events as a
negative factor will discourage petitioning bureaus from providing accurate
information.
Overall, however, since the dramatic rise of collective petitioning activities
in the 1990s, the petitioning system has been strengthened. It has obtained a
higher status in the bureaucracy and it enjoys more resources. For example,
in the late 1990s, the Hunan Provincial Government elevated the rank of offi-
cials in the provincial petitioning bureau by half a level. Consequently, the
bureau chief, whose rank had previously been one-half level lower than that
of other department chiefs, began to enjoy the same rank. In addition, since
the late 1990s, petitioning bureau chiefs have been concurrently appointed the
deputy secretary generals of the government at the county, city, or provincial
levels.6
In addition, there have been several remarkable adaptations of the proce-
dures for information collection and processing. First, since the mid-1990s the
petitioning system has set up procedures for reporting urgent information.
For example, in Hunan province, “must report” issues include (1) collective
petitioning of over fifty participants that has been addressed to the county
government or above; (2) potential collective petitioning activities of over one
hundred people; (3) information about collective petitioning events to the pro-
vincial government or Beijing; and (4) events that evolved from ordinary peti-
tioning into “explosive mass incidents” ( 突发性群体事件 ) that have disrupted
normal work, production, business, and daily order. The former three should
be reported within the same day; the latter should be reported instantly. The

6
It seems that this is a measure that was adopted nationwide. In Zhejiang province, for exam-
ple, all petitioning bureau chiefs at the city and county levels had been concurrently appointed
deputy secretary general or deputy director of the General Office before the end of 2001. See
Zhejiang nianjian (2002: 72).
State-Generated Data and Contentious Politics in China 29

report should include the time, location, number of participants, forms of

action, claims, governmental responses, and trends ( 动态 ). When necessary,
the local petitioning bureau should send follow-up reports.
Second, petitioning agencies are also required to screen “elements of insta-
bility” periodically. One of the tasks of the petitioning agencies is to provide
information about the targets of social control. There are two main tar-
gets: long-term petitioners ( 老户 ) and collective petition organizers. Before
the rise of collective petitions, long-term petitioners were the primary targets.
Since the mid-1990s, however, the primary target has shifted to collective peti-
tioners. Screening “instability factors” is not temporary work. The petitioning
system tried to institutionalize it as standard procedure. It was thus written
into the Regulations on Letters and Visits in 2005.
Lists of “instability factors” are usually made shortly before important
occasions: New Year’s Day, the Spring Festival, the “Two Conferences,”
National Day, and so on. These events are targeted because it is generally
believed that collective petitions are more likely to take place on such occa-
sions and that they will have a relatively greater impact on society. Therefore,
local governments have often designated such periods as “specially protected
periods” ( 特护期 ). In some years, petitioning agencies have been especially
busy during these occasions. In 1999, for example, besides the regular “impor-
tant periods,” there were other special events during which collective petition-
ing activities were to be strictly prevented: the crackdown on Falungong, the
tenth anniversary of the crackdown on the student movement, the handover
of Macau to China, and the fiftieth anniversary of the founding of the PRC.
In that year, the municipal bureau and seven county-level bureaus in City Y
compiled 43 lists in which they identified 739 “instability factors.”
Third, the procedures for compiling statistical data have been substantially
improved and standardized since the mid-1990s. Provincial bureaus have
established some standard forms that all bureaus at the county and city levels
must fill out each month. In City Y, for example, the petitioning bureaus of the
five counties and two districts must complete the standard forms each month
and send them to the municipal bureau. When the municipal bureau receives
the information from the counties, it will complete the same standard forms
with the information about the entire city. When the various cities send their
forms to the provincial bureau, the provincial bureau will fill out the same
forms with the information about the entire province and then send them to
the national bureau.
The quality of statistical work by petitioning agencies has improved con-
siderably since the mid-1990s. More extensive information has been collected,
and the methods have become more refined. For example, events with five to
twenty-nine participants are differentiated from events with over thirty par-
ticipants on the statistical forms of collective petitions. Similarly, among non-
collective petitions, petitions with one to two participants are differentiated
from those with three to four participants, which are called “group petitions”
( 群体访 ). Previously, this subtle differentiation did not exist.
30 Xi Chen

Finally, petitioning agencies have begun to produce better analytical reports.

Although the statistics can speak for themselves, petitioning agencies regularly
write all kinds of reports to help the leaders understand the general situation.
Many of these reports are based on statistics. There are special reports, which
may focus on a special period, area, or issue, and comprehensive reports,
which survey the situation about all issues for the entire area. The quality of
the reports has also improved considerably since the mid-1990s. Most reports
before the mid-1990s were superficial and written in a journalistic style. The
reports after the mid-1990s are more extensive and solid. They not only pres-
ent and analyze statistical data but also provide some background and even
offer suggestions for tackling problems.

Implications and Suggestions

The above analysis of the process of data generation carries some encourag-
ing messages for researchers. The authoritarian system has not created fatal
problems for governmental data. Many governmental materials, especially
raw or semiprocessed materials in the petitioning agencies, do not suffer from
serious distortion. The regime has also proven to be very adaptive. When
the leadership was alerted to the rise of popular contention, the organiza-
tions and procedures of the petitioning system were substantially improved.
Consequently, the petitioning system has begun to produce richer and better
data. Furthermore, the archival system has become more open.
This is not to suggest that governmental data are readily available to every-
one. Some conventional strategies for conducting fieldwork in China are still
very important. For example, it is helpful for foreign researchers to be affili-
ated with a Chinese university or research institute. Personal connections are
especially important for collecting governmental archival data. Furthermore,
researchers are advised to pay more attention to lower level agencies, which
are generally more open than higher level agencies. To be sure, higher level
agencies usually maintain materials of a larger scope. However, for some “sen-
sitive” topics, data from a county government may be very valuable. Finally,
because different local governments may have quite different policies, it is
important to try different agencies and locations. Often valuable data cannot
be found until researchers have explored a number of possibilities.
But the above analysis also reveals some problems. Because information
work in the petitioning system (as well as in most other public agencies in
China) generally has not been well institutionalized, there are remarkable
variations in the available data across time and region. Differences in the
scope and content of collection often make it difficult to conduct cross-time
and cross-regional comparisons.
In addition, the irregular variations make the biases in the data somewhat
unpredictable. Fortunately, there are some measures to partially remedy the
problem of uncertainty. First, we can seek some indicators to measure the
quality and reliability of the data. As Walder and Su’s (2003) study of county
State-Generated Data and Contentious Politics in China 31

gazetteers indicates, the quantity of information is often an indictor of its

quality. This is because the quantity reflects the approach of the local leaders
regarding information collection and processing. Where the leaders stress the
importance of information work, governmental agencies tend not only to col-
lect more information but also to collect it more carefully. Other indicators
may also be helpful. For example, if a bureau has received an award for its
information work, usually the data it has produced are of higher quality. Of
course, this may also raise questions about bias in the data.
Second, we can investigate the reliability. One advantage of using contem-
porary data is that it is more convenient to examine how these data have
been produced. Government officials in charge of compiling materials can
sometimes be interviewed. We can also evaluate the reliability by comparing
different sources. Since governmental data include multiple sources, they can
be used for cross-checking. For example, important petitioning events can be
reported in different forms: they will be reported in the “important petitions
for review,” the internal circulars, the reports on “screening elements of insta-
bility,” and the periodic reports. In addition, the reliability can be evaluated
by examining outside sources. For example, we can compare governmental
data with information from interviews. In this sense, triangulation of differ-
ent sources is especially useful.
Third, problems are less serious if we use upstream sources. The more the
data have been processed, the more biased they tend to be. As analyzed ear-
lier, analytic reports and statistical results are likely to be more biased. When
upstream information sources, such as petition letters, registration forms,
and original reports about the events, are available, researchers are advised
to process the data by themselves as this can yield more reliable outcomes.
The statistical data in the petitioning system have multiplied in recent years.
Although they are rich and very informative, there are some risks and dif-
ficulties in using them. The definition and measurement of key categories are
often vague and inconsistent. For example, the term “collective visits” for
some years referred to petitions delivered by three or more participants, but
in other years it referred to petitions delivered by five or more participants. In
addition, the official jargon, which is very common in statistical data, may not
only confuse researchers but may also lead to inconsistent information. This
is because officials from different agencies may interpret it in different ways.
For example, petitions related to illegitimate fees imposed by local schools are
sometimes classified as issues of “ 三乱 ” but at other times as “issues of peas-
ant burdens.” It is uncertain whether these vaguely defined and sometimes
overlapping concepts are interpreted consistently by local petitioning cadres.
In contrast, with upstream materials researchers can control the methods of
codification and aggregation.
These risks and difficulties also apply to information sources from higher
level governmental agencies. Higher level agencies usually enjoy much better
resources and facilities and therefore produce obviously much better archival
materials. However, it should be noted that most of the aggregate data come
32 Xi Chen

from below. As Leo Orleans suggested long ago, “the greater the number of
administrative plateaus which serve as resting places for statistics as they are
moved up the line, the less accurate are their figures” (Orleans, 1974: 51).
Of course, even upstream sources have some selection or description biases.
For instance, in the petitioning bureaus we can find files only on those cases
that have been regarded as important by the government. Similarly, inter-
nal reports of the petitioning system are more likely to cover protests that
took place near a government compound or that caused considerable disrup-
tion. It is important for us to evaluate and state explicitly these biases in our
research.
Fieldwork in China is often like an adventure. No good map is available for
the hidden treasure of state-generated data. There are also pitfalls along the
road. For those undeterred adventurers, however, other explorers’ reflections,
such as those in this chapter, may sometimes provide valuable clues.
2

Why Archives?

Neil J. Diamant

Like many of the contributors to this volume, I began graduate school roughly
a decade after China began its “reform and opening-up” process. At the
time, the field of Chinese politics was, temporally speaking, reasonably well
demarcated. Those of us in political science departments were expected to
say something novel and significant about the post-1949 period. Within this
rather narrow range, however, an even more contracted time span – post-
1978 politics – drew the largest number of students, as well as generated
the most interest among the broader public and departmental colleagues.
To be sure, this temporal border was somewhat porous,1 but because it was
also vigorously patrolled by tough-minded historians, who were reluctant to
define the 1949–78 period as “history,”2 the “present-ism” in the study of
Chinese politics was difficult to ignore. More recently, thanks to the grow-
ing interest in China’s “rise,” the gravitation of political science toward the
present has become even stronger. Students of politics overwhelmingly con-
centrate on events in the last thirty years, whereas historians, with some
exceptions, still focus on the pre-1949 period, leaving the other three decades
of the People’s Republic of China (PRC) politics (1949–78) in a disciplinary
twilight zone. In this chapter I suggest that presentism has not been cost-free
in terms of our ability to understand Chinese politics and that these costs
can best be recouped by broadening the temporal scope of Chinese politics
and making archival research a far more integral part of scholarship than is
currently the case.

1
For example, political scientists such as Elizabeth Perry, James Tong, Joseph Fewsmith,
Andrew Nathan, and David Strand crossed it, survived, and made their names concentrating
on pre-1949 politics.
2
The Chinese history field still does not have its equivalent of a Strand, Nathan, or Perry –
someone who wrote an award-winning study of the 1960s or 1970s. In contrast, American
historians can write books about the civil rights movement or “Berkeley in the ‘60s” without
any special justification.

33
34 Neil J. Diamant

These musings about disciplinary boundaries were stimulated by Roderick

MacFarquhar and Michael Schoenhals’s study of the Cultural Revolution,
Mao’s Last Revolution (2006). The book is exceptional in several respects.
The authors collected fascinating data from a dizzying array of sources, among
them Red Guard personal diaries, broadsheets, handbills, big-­character post-
ers, and minutes from Politburo meetings. It is difficult to imagine a more
dogged pursuit of data on such a sensitive topic. Unsurprisingly, the book is
jam-packed with politics (which they view as driven by personal and fam-
ily agendas, with “institutions” and “organizations” playing bit roles). At
the same time, Mao’s Last Revolution is atheoretical; I doubt it would pass
muster as a “job talk” in most American political science departments, let
alone provide the ballast to cross the tenure threshold. Last is the topic itself.
Although some historians and sociologists and their graduate students have
been encroaching upon the Cultural Revolution, it would be the intrepid grad-
uate student in political science who would venture into the mid-1960s to pen
a dissertation about civil-military relations in, say, Sichuan province. As if
snubbing their noses at the fads and scientific pretensions of political science,
MacFarquhar and Schoenhals present a balanced political history of those
fateful years. The extended commentary on the credibility of sources, the dis-
cussion of how to weigh a particular piece of evidence, and their digging for
valuable data nuggets would make both authors highly attractive candidates
on the history job market!
But aside from these speculative scenarios – both authors have received
tenure at well-regarded institutions – Mao’s Last Revolution is noteworthy
for something even more important, and directly related to the discussion of
the role of archives in the study of politics: The authors’ ability to marshal
thousands of new sources on a topic that in many libraries occupies only
several shelves demonstrates the exciting possibilities to generate new knowl-
edge about critical events in PRC politics (with legacies lasting decades) if
researchers take up the challenge of shifting the temporal frame backward
toward the prereform era. Equally significant, Mao’s Last Revolution was
researched without access to the central party archives, the “mother-load”
of Cultural Revolution material. Imagine the scope of data, and new insights
about politics, that could be gleaned from a year-long immersion in this and
other archives, or if local archives opened their doors to researchers examin-
ing other seminal political events in the PRC, such as land reform, the Korean
War, the Great Leap Forward, or diplomatic relations with the USSR! These
sources would activate tremors in the field; to stay intellectually honest we
would have to revise our lectures, textbooks, and “consensus” histories of the
period (such as the Cambridge History of China, vols. 14 and 15). We would
also have to recalibrate our interpretation of the post-Mao period, particu-
larly if scholars offer “path-dependent” and historically grounded explana-
tions for political and economic phenomena, or alternatively, argue that the
present represents a sharp break from the past.
Why Archives? 35

Fortunately, we do not need to overstretch our imaginations to envision

this scenario. In recent years, some scholars of the PRC, like some of their his-
torian counterparts,3 have taken advantage of archives to press for significant
revisions of the received wisdom about a host of topics. Mark Frazier’s (2003)
thesis on the Republican-era origins of the work unit (单位) system, Julia
Strauss’s (2002) work on early 1950s political campaigns, Eddy U’s (2003)
study of Shanghai teachers, Patricia Thornton’s (2007) analysis of efforts to
control corruption during the Qing dynasty, the Republican period, and the
PRC, Elizabeth Perry’s (Perry and Li, 1997; Perry, 2006) studies of labor, and
my own on the 1950 Marriage Law (Diamant, 2000) all relied extensively on
newly accessible archival data. These archive-based studies, I argue, should be
the jumping-off point for a much-needed modification of the received wisdom
about politics in the PRC. However, as the situation currently stands, histo-
rians willing to “push” history up to the 1970s and 1980s, and not political
scientists, most of whom are reluctant to “pull” politics back to this period,
will be the main beneficiaries from the rapid declassification of materials now
under way in PRC archives at all levels, sans the Center; to wit: most of the
scholars named above are not employed by political science departments at
large American research universities. This would be unfortunate. Given the
short history of the PRC, and that much of what we have learned about its
politics is based in the “pre-archival era,” it is far too soon to relegate the foun-
dational years and critical events of the prereform PRC to “history.” Think
about this: what would the field of American politics look like if anything that
happened prior to the Carter administration was considered “historical,” or
if political scientists eschewed research in the Nixon Presidential Library and
Museum because colleagues and others were more fascinated by the Obama
administration? Most Americanists would consider these propositions fairly
ridiculous.
In this chapter, I want to make the case for the vital utility of archives to
the study of Chinese politics, past and present. I will do this by examining
some documents that provide an unprecedented look at the politics surround-
ing a high status, but personally vulnerable, population: the wives/fiancées of
People’s Liberation Army (PLA) soldiers during the 1950s and 1960s. Even
though these old memos and investigation reports (the proverbial pot of gold
for historians) represent a smidgen of what is available in archives, they can
open a window onto larger, more conceptual, “political science–friendly”
issues – in this case, patriotism and nationalism. They also call into question
much of the conventional wisdom about the high status of the PLA and its
personnel during this period.
Before moving onto the nitty-gritty of this topic, however, I want to expand
a bit on archival research. What are some of the pitfalls? What documents are

3
The fields of Ming, Qing, and Republican legal history have been thoroughly revitalized by
archival research. See the works by Philip Huang, Kathryn Bernhardt, Melissa Macauley,
Matthew Sommer, and Janet Theiss, among others.
36 Neil J. Diamant

available, and how should researchers use them? As we will see, the volume
and level of detail in archival documents can stimulate us to look at old issues
in new ways, and draw attention to aspects of the state and society that are
easily brushed under the carpet in sources that pay more attention to elites.

Archival Research: Pitfalls and Opportunities

Unlike many methodologies in political science (or the social sciences more gen-
erally), there is little by way of official guides for conducting archival research
in China. Like the National Archives in the United States, some archives in the
PRC publish their own guidebooks (指南) – the Shanghai 指南, for instance,
fills 850 pages – but most do not. Wa Ye and Joseph Esherick (1996) compiled
a useful overview of archives based on information in county gazetteers, but
this volume already is somewhat dated, does not include information on mate-
rials that might be made available upon request, or the contents of district
(区) archives in cities. Thankfully, Esherick and others have created a Web
site with updated information on various archives4 as well as a very helpful
compendium of “Archive Users’ Experiences.” It is also possible to write to
an archive prior to one’s visit to request information and to ask if foreign
scholars can access certain materials. For sensitive topics, however, it is prob-
ably best to work through a research institution or university; connections
(关系), of course, help a great deal, as can reciprocal exchange opportunities
and conferences. For instance, archivists at the Yunnan Provincial Archives
were receptive to my requests for material because their director had recently
been hosted by the National Archives.
The successful mining of archives, much like other methodologies, depends
upon many factors. The background, education level, and cosmopolitanism of
archive personnel can make a significant difference in access. The topic mat-
ters as well: given the many political events in PRC history that are considered
politically sensitive, how one frames a subject and presents it to Chinese hosts
and archivists can affect the outcome of research. Generally, it is important to
frame one’s subject in as noncontroversial language as possible. For example,
a study of land reform might be pitched as a study of changes in land-holding
patterns. Since land-holding patterns were partially shaped by land reform,
there is greater justification to request these materials. I framed my study of
veterans and military families under the more general rubric of “providing
preferential treatment for the worthy.” It is also important to arrive in China
with ample financial resources since it often takes time to collect enough
material to produce an article or book-length study. As might be expected, it
is beneficial to develop and cultivate ties with archive officials. And last but
not least, archival research requires patience. Sometimes the best materials
suddenly surface after several months of persistent requests. I definitely rec-
ommend taijiquan or yoga for relaxation.

4
See http://orpheus.ucsd.edu/chinesehistory/chinese_archives.htm, accessed August 2009.
Why Archives? 37

Since any one of these factors could be problematic, researchers should

always have a backup plan in case one archive does not allow access or sim-
ply not enough access to justify an extended stay. In Shanghai, Beijing, and
other areas that have had more exposure to researchers (who occasionally
pressure archivists for more openness), it is possible to procure documents
without worrying too much about surcharges and added fees. The same can
be said for counties that have hosted foreign researchers; one definitely should
not assume that “rural” implies greater conservatism. In Shanghai, research-
ers can search a database of “open archives” (开放档案) based on keywords.
Because all of the declassified documents will appear on the screen, research-
ers do not have to search for material by work unit, which is how most data
are organized. For instance, typing in 复员军人 (veterans) will bring up docu-
ments from Civil Affairs, the Shanghai Garrison, the Labor Bureau, factories,
and the party committee, among others. This is not the case for archives with
fewer patrons and resources, however. Even large provincial archives, such as
Shandong and Jiangsu, still use folders labeled with the titles of documents,
access information, and page range. When the documents are found in the
indices, researchers have to write them down on a form and present them to
the officials in charge of the reading room, who, in turn, pass them up to the
director of “user services” (利用处).
This is when things can get interesting, and messy. In some instances, offi-
cials will deny the request, citing “privacy rights” or some other reason. In
others, materials will be furnished, but researchers will not be permitted to
photocopy them. This can be a minor or major inconvenience, depending on
how much time one has to pore over the materials. In this event, time-rich doc-
toral candidates, or senior faculty on sabbatical, will have an advantage over
those who have flown in during semester break or two weeks in the summer.
Despite this problem, the struggle is usually worthwhile: some documents can
be very revealing about post-1949 politics,5 small pieces of evidence from one
area can confirm findings from other regions and thus enhance generalizabil-
ity, and the battle might make it easier for subsequent researchers.
These problems, however, are not necessarily universal. Archive officials
will be less skittish when presented with topics that are perceived as less
“political.” But even with more sensitive topics, such as veterans and mili-
tary families, it is not a foregone conclusion that photocopy requests will be
categorically denied; sometimes one has to make repeated visits to the same
archive to procure more and better materials.
Unfortunately, Chinese officials are not the only ones raising obstacles to
successful political science–oriented archival research. Unlike historians for
whom archival research is a badge of professional métier, political scientists
in the China field (less so in American, Latin American, or European poli-
tics, for whom forays into history are common) who rely on these sources

5
For instance, in Shandong I found the minutes of a meeting of county chiefs. Many told their
superiors about uprisings in the province led by Korean War prisoners of war and veterans.
38 Neil J. Diamant

might encounter a number of discipline-related hurdles. First, archival data

can be very detailed and descriptive, which can militate against “parsimony”
in explanation, the gold standard for some political scientists. Second, despite
the dismal record in predicting major political events, political science tends to
reward “theoretical innovation” more than the discovery of new information
per se. Third, archival research is associated with “historical” or “qualitative”
methods, which, in some quarters, are more suspect than methods such as
survey research. Finally, there is the presentism issue. With all the news about
China “rising,” many political scientists, like the lay public, want to know
what is happening in China now. The rosier contemporary scene – not the
brutal and poor Maoist period – is “hot.” As a result, topics that get defined as
“historical” and that rely on methods drawn from “history” need to be legiti-
mated in ways that more contemporary topics using quantitative methodology
do not.6 My point here is not to dissuade scholars from using archives but
rather to note that the incentives for engaging in intensive archival research
are not well aligned with the professional incentives in political science (with
the exception of the most cosmopolitan departments). For the budding histo-
rian the opposite situation prevails; history departments praise the researcher
who pries open an archive. This incentive structure is highly problematic,
since archival documents, when used in conjunction with other sources, can
shine a bright light on many of the dark corners of PRC politics.

What’s in There Anyway?

Once having gained access to archives and having limited time and money,
what sort of documents are worth pursuing with rugged determination, and
which can be given up with little loss of sleep or increased blood pressure?
The criteria I have used most frequently in requesting materials are (1) the less
censored the better; (2) the more local, the better,7 and; (3) get everything with
the words 调查 (survey) or 检查 (inspection) in it. These guidelines encompass
a fairly wide range of documents, so it is worthwhile to get into the specif-
ics: what can researchers expect to find in archives, and why are these materi-
als more useful than other sources? Below is a brief survey.

Minutes of Meetings
Minutes can be a very useful source, since, unlike more official sources such
as newspapers, they are usually unedited and uncensored. I am not certain at

6
For this reason, archival research is somewhat more risky for doctoral candidates in Chinese
politics than for established scholars. This can be problematic, because, as noted above,
it is these researchers who tend to have the time required to plumb through the archival
materials.
7
This is important because most indices only provide the title of the document, so searching for
keywords referring to a specific place, or “minutes,” can save a great deal of time and hassle.
Why Archives? 39

which level of government it is mandatory for a secretary to take notes, but the
practice is widespread. In my own research, I have seen transcripts of meetings
of work teams discussing the implementation of the Marriage Law, judges’
deliberations, Civil Affairs officials learning about the marriage registration
process, and mediators’ sessions with veterans embroiled in disputes with local
officials. My guess is that many factory archives include comparable materials
for workers’ conflicts. Like reading recordings of Kissinger’s phone calls to
foreign leaders in The Kissinger Transcripts, minutes provide an unvarnished
representation of how the state operates, a central concern in political science.
At the same time, reading minutes is labor-intensive, particularly for those not
accustomed to reading handwritten Chinese with various shortcuts for char-
acters. It is probably best to have these materials photocopied because time is
a precious resource.

Investigation Reports
These sources are probably the most valuable. First, many satisfy my “local
rule” because investigations occur at all levels of the party-state apparatus.
Trade unions, factory party committees, courts, hospitals, and military units
have all been ordered, at one point or another, to investigate something.
When the State Council ordered an investigation (on veterans in 1956–57,
for example), investigation teams were formed nationwide, which means that
archives in Shanghai and Yunnan will have materials on the same problem.
This can facilitate interesting comparisons between locales. These materials
may be printed or handwritten. Second, investigations have been common.
To our immense benefit, the PRC has been obsessed with research. Although
their findings are never couched in terms other than Marxist or Maoist, the
quotes from officials and ordinary people provide a candid look at the inter-
face between state and society. Third, because investigation reports have never
been intended for outside eyes, they are not heavily censored, if at all. Even
better, one can sometimes read multiple drafts of the same report to see what
parts have been excised.
At the same time, investigation reports should be handled gingerly since
they are almost always focused on problems. To assess just how extensive
these problems are, researchers should look for corroborating evidence from
other sources (interviews, the press, gazetteers, or other documents in the
archives).

Work Reports
Most of us have complained at one time or another about bureaucracy, but
researchers should always remain grateful that bureaucrats have meetings and
routinely issue reports about what they are up to. PRC archives are full of this
type of document; they were issued quarterly, biannually, and annually, often
40 Neil J. Diamant

in multiple drafts. Work summaries (总结报告) tend to be broad-stroked but

also include muted criticism of an agency’s work, broken down by subcatego-
ries. (For instance, Civil Affairs will be broken down into sections on disaster
relief, refugees, local elections, veterans, and minorities.) Compared to investi-
gation reports, work summaries have more empty verbiage (空话) and provide
a shallower sense of context and the specifics of a problem. Because of this,
they can be read fairly quickly. Work summaries generally begin by providing
basic information, continue with two or three examples of success (typically
model individuals praising the party/Mao), and then shift into critique mode.
The transition between sections two and three is clearly marked by the word
“however” (但是). Weighing which of this information is closest to the truth
of the matter can be a bit challenging: the “achievements” sound exagger-
ated, but the “problem” section can also be overly harsh. I have generally dis-
counted examples of “models” in the achievement section and have waited for
further confirmation from other sources to determine whether the “problems”
in these summaries were in fact pervasive.

Policy Documents
Since archives are repositories of official activity, policy documents unsurpris-
ingly represent a substantial share of their content. Although they are not as
comprehensive as central archives, researchers can still find most or all impor-
tant policy documents relevant to research in large provincial archives or in
urban archives that have provincial status. In some respects, archives at the
latter level are better for looking at issues of policy implementation because
they receive key policies and reports from the center (these sometimes include
materials from other provinces), devise their own policy documents, and col-
lect materials from their districts, suburban areas, and rural counties. If one
does not have the time to work in a rural archive, or would rather avoid the
hassle of securing permission and letters of introduction, it is still possible to
conduct a rural/urban comparison in a large municipal archive.
But even without this sort of comparison, archives open a wide window
into officialdom, allowing political scientists to see the state “in action.” In
addition to policy documents, there is also the more mundane back-and-forth
correspondence between agencies, letters and complaints from individuals
about problems that are too small to be included in work summaries, and
answers to queries submitted by lower level officials. The downside of all this,
of course, is that it is fairly easy to be overwhelmed by the quantity of data.
It is advisable to come prepared with a clear idea of what you are looking for
prior to a visit to an archive.
But enough said of these “how to” matters! In what remains of this chap-
ter I want to provide a glimpse of how I am trying to bridge the gap between
the more theoretical concerns of political scientists and the “informational”
strengths of archives (more prized by historians). Although my focus happens
to be on patriotism and how it relates to military service, I would venture that
Why Archives? 41

many of the “large” issues that concern political scientists – power, justice,
status, distribution of resources, and so forth – can be addressed as fully and
provocatively with archival documents as with any other source.

Patriotism, in Everyday Practice

Patriotism, as well as its sister concept in the scholarly literature, national-
ism, has been frequently evoked in the study of Chinese politics. It has been
used to explain the victory of the Communist Party over the Guomindang,
the split between China and the USSR in the late 1950s, and the resilience
of the CCP after the collapse of Marxism-Leninism in the late 1980s and
early 1990s.8 Although used in a number of different ways, the literature on
patriotism and nationalism has been consistent in (1) focusing on writing by
elites, rather than more popular sources; and (2) focusing on musings, ideas,
and ideologies, rather than on action or behavior. These two items are not dis-
connected: because elites are those who record their sentiments, feelings, and
thoughts about a topic in any depth, the literature as a whole is far more likely
to focus on elite cognition than on other facets of patriotism. Unsurprisingly,
the search for evidence about patriotism and nationalism has gravitated in
this direction. I argue below that this conceptualization is problematic, and
that archival materials, by focusing our attention on behavior and everyday
interactions among people, shine a far brighter spotlight on what patriotism
and nationalism actually mean in practice. Archival materials help ground
these concepts – and subsequent theories – in people’s experiences (gener-
ally a research topic affiliated more closely with social historians and cultural
anthropologists). To the extent that archives zero in on political experiences,
we can narrow the gap between what Clifford Geertz (1983) and others have
called “experience-near” and “experience-distant” concepts in political sci-
ence.9 In my view, studies of nationalism and patriotism lean far too heavily
toward the latter.
How would one know to what extent citizens in China, or anywhere else for
that matter, are “patriotic”? Any assessment of this question must begin with
at least a rudimentary effort to define the term. This isn’t easy: patriotism is a
value-laden and politicized term. Despite the eminent political theorist George
Kateb’s warning that patriotism is “inherently disposed to disregard moral-
ity” and a “grave moral error” whose source is “typically a state of mental
confusion” (2000: 901, 909), American surveys repeatedly show that most

8
In China, for example, many have cited the rise of “nationalism,” 民族主义 but in Chinese
the more appropriate term, and the one used in propaganda and education, is patriotism
(爱国主义).
9
According to Geertz, experience-near concepts are ones that “someone . . . might himself natu-
rally and effortlessly use to define what he or his fellows see, feel, think, imagine and so on.”
In an “experience-distant” concept, “specialists of one sort or another . . . forward their sci-
entific, philosophical or practical aims.” Love, he suggests, is an example of the former and
“object cathexis” the latter.
42 Neil J. Diamant

people consider themselves to be “patriotic,”10 probably because its antonym

connotes dishonorable behavior; few people relish being labeled “unpatriotic”
even if they do not agree with government policies. Political baggage, how-
ever, should not deter us from trying to get a better sense of what the term
involves; it should not be any different from the key concepts in comparative
politics that scholars often debate, such as “democracy” or “corruption.” A
reasonable place to begin, it seems to me, is the most common and seem-
ingly straightforward definition of it, and the one that was translated into
Chinese: “love of country-ism.” “Love” is sometimes padded with qualities
such as loyalty or pride, “an attitude of sentiment and devotion” to a state
or nation (Fletcher, 1993: 17, 140), an “acquired sentiment” (Druckman,
1995: 58), “ongoing civic concerns” for fellow citizens (Tamir, 1997: 21–33,
37), or “identification with others in a particular common enterprise” (Taylor,
1989: 166). Maurizio Viroli (1995: 1) suggests that historically patriotism has
been used to strengthen or invoke “love of political institutions and the way of
life that sustain the common liberty of a people,” whereas Walter Berns argues
that patriots in the traditional (“Spartan”) sense are “citizens who love their
country simply because it is their country” (2001: 10, 65).
From the perspective of archival research on how officials and ordinary
people treat the wives of military personnel on active duty – those who are
actually fulfilling what the government itself deems a “patriotic duty” to
serve – these definitions of patriotism are problematic. First, to the extent
that “love,” “sentiments,” and “devotion” can be considered emotions, it is
fairly commonsensical that all emotions vary in intensity across time and the
“object” of desire or affection. What do they actually mean in the context of
states or nations? Should we “love” a country like a father loves a son or like
a teenager loves his girlfriend? Second, and more problematic, they are all low
thresholds for claiming or assigning patriotic status to any individual or group.
Is it really enough for someone to simply express sentiments such as “I love my
country” to be considered patriotic? How should we view someone who says
“I love my country” in the morning, but that same afternoon has an affair
with the wife of a special operations officer risking his life abroad? “Loyalty,”
for that matter, is also too easy. As Morton Grodzins (1956) argued over fifty
years ago, the overwhelming majority of citizens are “loyal” simply because
they do not actively join ranks with a country’s sworn enemies. Loyalty is a
passive sentiment (which, of course, usually makes charges of “disloyalty”
highly suspect). On a day-to-day basis, love, loyalty, and sentimentality do
not demand any sort of behavior from citizens; their excessively low thresh-
old is somewhat comparable to a regime that claims to be a “democracy” just
because people have rights in the constitution.

10
In a 2003 survey (1,200 people contacted by phone) conducted by the Institute of Politics
at Harvard University, almost 90 percent of Americans considered themselves to be either
“somewhat” or “very” patriotic. The results were the same in 2002. These rates may be
higher because of September 11. See http://poll.orspub.com. This database is available only
via subscription.
Why Archives? 43

This problem also applies to the concept of “nationalism,” particularly

when scholars adopt, either implicitly or explicitly, Benedict Anderson’s wildly
popular definition of nationalism as an “imagined community” based on the
perception of some sort of shared identity11 – a “recognition of similarity” in
the Chinese word for “identity” (认同). Imagination, like identity, essentially
is a cognitive process – it “tells” us that we share a commonality with fellow
citizens, much like “emotions” and “sentiments” also occur at this “intra-
cranial” level. Even among scholars who question whether Anderson’s causal
argument can be applied to an underdeveloped country like China, or whether
his focus on the “nation” as the sole object of loyalty and identity is sufficient
given the prominence of regionalism in many areas of the world,12 few have
noted this “threshold” problem: one becomes a “nationalist” either through
an “act” of imagination (can the firing of neurons count as an act?) by sug-
gesting, usually in speech or writing, that one has this identity, or what Adam
Smith, in A Theory of Moral Sentiments, calls “fellow feeling” with other
citizens, or that other citizens should “awaken,” and/or embrace a higher
degree of cultural or ethnic homogeneity.13 But, as Ronald Krebs insightfully
comments, “Nationality is assuredly an imagined construct, but that does not
imply that the mechanisms by which its boundaries are drawn and redrawn
are best grasped through models emphasizing cognition and mental creativ-
ity” (Krebs, 2006: 11).
The methodological consequences of this conceptualization are far-reaching.
Much of the work on patriotism and nationalism tends to focus on the short-
term protests of elites14 or their sentiments, ideas, ideals, and ideology, since
they are the ones who leave the most coherent records of their musings.15
Jonathan Unger, for example, argues that Chinese peasants “had little notion
of China as a whole, let alone being attuned to the nationalist sentiments that
were developing among the educated classes in China’s urban areas” (Unger,
1996: xv). The rise of “nationalist sentiments” or “popular nationalism” in
China has become the conventional wisdom in much of the recent work on
China’s international relations.16 Journalists often use the “trope” of rising
nationalism as well.17
11
For an application of this notion to American patriotism, see Bodnar (1996).
12
See many of the contributions to Unger (1996b). Also see Goodman (2004).
13
Viroli (1995) argues that nationalism focuses on defending or reinforcing the “cultural, lin-
guistic and ethnic oneness” of a people.
14
It is very common in Chinese history to speak of patriotic movements or protests, much like
Chinese demands for democracy are often expressed in movements rather than in institutions.
See Strand (1990).
15
For a sophisticated approach to nationalism using intellectuals’ writings, see Fitzgerald (1996).
16
For example, Gries (2004) examines the writings of a new cohort of young, fourth generation
“nationalist” intellectuals as well as sporadic incidents of outrage against Japan and the United
States. He does not consider whether penning a book or participating in a short-lived, almost
risk-free, protest should qualify someone as a “nationalist.” In my view, such activities might
qualify as “fleeting anti-American sentiments,” but not as nationalist in a meaningful sense.
17
Journalists have adopted this “trope” of rising nationalist “sentiment” uncritically. See
“Balancing Act: A Survey of China,” The Economist, March 25, 2006.
44 Neil J. Diamant

In my view, to be even moderately useful, concepts must include and

exclude certain behaviors, qualities, and attitudes. Self-definitions, claims of
“love,” “devotion,” and “loyalty,” are insufficient since they tend to include
most everybody. I would suggest the following: to be meaningful, patriotism
should incorporate at least two of the following four criteria: sustained action,
moderate to long-term commitment of resources, some courage, and what
a reasonable person would consider a sacrifice (not just a “willingness” or
“readiness” to sacrifice, which many can easily claim (Druckman, 1995: 58).
These dimensions of patriotism are not new by any stretch of the imagination,
maybe just unfashionable in an age that is highly materialistic, focused on the
rights of individuals and groups, and largely devoid of any integrative expe-
rience such as conscription or some other form of national service. Usually
promoted by military leaders (far more than intellectual elites) in the context
of mass protracted wars (Porter, 1994), they harken back to a long tradition of
republican or martial patriotism which stresses the notions of “self-sacrifice
for the good of all” (Bodnar, 1996: 7; Krebs, 2006: 28), a “framework of
duty” (Ben-Eliezer, 1995: 265) that treats military service “as the preeminent
civic obligation and identifies the good citizen as one willing to die on the
battlefield for the political community.” Much like the emphasis on ma’asim
(deeds) in scripture, it implies that speech, cognition, right-minded thinking,
sentiments, love, and imagination do not a functioning society, nation, or state
make. Political theorist Michael Walzer, for example, argues that “men are
bound by their significant actions, not by their feelings and thoughts; action
is the crucial language of moral commitment” (Walzer, 1970: 98). But such
commitments cannot be fleeting. “The nation’s existence is a daily plebiscite,”
notes Ernest Renan in “What Is a Nation?” which he defines largely in terms
of actions and the memories created around them, especially shed blood: “the
culmination of a long past of endeavors, sacrifice, and devotion” (O’Leary,
1999: 4–5). Nor is this perspective necessarily Western. Korean nationalists at
the turn of the century penned eloquent essays on “virtues,” such as bravery
in action and the “sacrificial spirit,” that they saw in both Western and Asian
civilizations, citing figures as diverse as Jesus Christ and Bismarck and Zhuge
Liang (Tikhonov, 2007: 1030). One of the most revered figures in Chinese
civilization, Qu Yuan (c.340 b.c. – 278 b.c.), achieved iconic status as a “patri-
otic poet” because he sacrificed himself for the principle of loyalty and oppos-
ing corruption in government; Yue Fei, a Song dynasty general (1103–1142), is
currently revered as a patriot because of his loyalty and for devoting years of
his life and taking significant risks in combat for the state. Of course, from the
early 1960s until the end of the Cultural Revolution, soldiers of the People’s
Liberation Army were praised for precisely these qualities as well, as were the
families that supplied them to the state.
From this reconceptualization of patriotism we can extract several method-
ological prescriptions, particularly regarding the sources that might be used
in research. Definitions of patriotism and nationalism that emphasize abstract
ideas (e.g., imagination) over the institutional and behavioral will necessarily
privilege elite sources. Less attention will be paid to the smaller scale actions
Why Archives? 45

of the lower classes, such as soldiers, veterans, and military families – the
populations that usually pay a significant price for service on behalf of the
patriotic causes the elite support in rhetoric. (For instance, the family of a sol-
dier from a peasant family sacrifices the labor of a healthy young man.) The
emphasis on love and imagination also shifts our analytical lens away from
interactions among people – the actual treatment of those whose sacrifices
and commitments have been elevated to “patriotic” status. In other words,
claiming “love of country” should be analytically differentiated from com-
mitting sufficient budgetary resources to the military (proper armor), veterans
(favorable legislation), or supporting military families.
This is precisely where archival sources should be mined. Archives, far
more than articles and speeches by elites (many of whom are in the employ of
the state), enable us to part the velvet curtain that hides and protects national
and heroic narratives and peer at the “backstage” of patriotism where people
interact. They both ground and fill out abstract ideas in messy reality and
local interpretations. Archival sources – particularly the obscure memos, min-
utes, and investigation reports – do not draw our attention to various expres-
sions of pride in country (“mouth patriots”), flag waving at sporting events,
or words that reveal “mental constructs” about the nation, but instead zero
in on how flesh-and-blood people behind patriotic discourse were treated in
more mundane or “everyday” circumstances: in villages when military fami-
lies needed help with the harvest; when veterans needed time off from work
to see a doctor. In short, if patriotism can be conceptualized as Renan’s “daily
plebiscite,” we can get a better sense of its depth and content by examining
documents that hone in on the everyday and can explain the gap between offi-
cial discourse and how the people valorized in it were treated in “everyday”
interactions with fellow citizens.
The treatment of military families at the height of pro-military, nationalist
discourse can serve as a useful test for the significance of patriotism and the
utility of archival research in China. Did the valorization of the PLA as the
most patriotic and revolutionary of political institutions in the years prior to
the Cultural Revolution (when many were convinced that China would be
attacked) translate into positive interactions with the wives and fiancées of the
soldiers in military service? Treating military families well, unlike attending a
short-lived, state-sponsored protest, may not be easy – which is precisely why
it serves as a better test of patriotism. Military families need assistance with
harvests; wives and fiancées are left without much protection, requiring other
men to stave off their sexual temptations; local financial assistance requires
sacrifice. If one were to read newspapers, memoirs, or even to observe the
behavior of students during the Cultural Revolution, one could deduce that
military families were treated well. After all, Red Guards paraded around in
military fatigues, spoke in military-inflected vernacular, and organized their
units along military lines.18 Does a more inductive and grounded approach

18
For an interesting analysis of the military influence on the Red Guards, see Perry and Li
(1993).
46 Neil J. Diamant

challenge this fairly widespread view about the status of the military as a
patriotic institution and its personnel as exemplars of patriotic commitment?
Let us take a brief look at how archival data challenge sources that rely more
heavily on the views and short-lived and dramatic outbursts of the more edu-
cated classes.

Military Dependents (军属)

In and around Shanghai, the city well known for its leftist radicalism prior to
the Cultural Revolution, a flurry of confidential investigation reports threat-
ened to tar its revolutionary reputation. Officials reading these reports in the
mid-1960s were probably shaking their heads at the more literal, very ris-
qué twist on the so-called flesh and blood relationship that was said to exist
between those who patriotically served in the military and the state: cadres
and others were having affairs with the fiancées and wives of PLA soldiers. In
a report that was furnished by Civil Affairs to the Shanghai Party Committee,
this phenomenon – known as “ruining a soldier marriage” – was deemed
“quite serious,” affecting soldiers in urban and rural areas (SMA1, 1964:13).
In 1963 alone, the report found, some 253 cases had been “discovered” in
the greater Shanghai area, including 193 in the rural suburbs and 60 in the
city proper. Most of the perpetrators held positions of authority; in the sub-
urbs, 57.4 percent of them were cadres at the township, village, or subvillage
(“production team”) levels; Communist Party and Communist Youth League
members – those most exposed to education about the importance of patrio-
tism and the heroic role of the PLA – figured prominently (39 percent) in the
overall statistics, although blame was also appropriately apportioned to “bad
elements, hooligans, and hoodlums” (SMA1, 1964:13).
Ruining soldiers’ marriages was not simply the result of a few bad eggs with
inappropriate sexual appetites, however; there were larger, “structural” causes
at work as well. According to the report, military families’ poverty, which
should have been alleviated by government programs and ­community-based
assistance, combined with officials’ access to resources, created the opportu-
nity for sexual mischief. Local officials “got close” to PLA wives by express-
ing “concern about them,” and this developed into an exchange of food and
entertainment for sex. That the women already were married to soldiers
“defending the nation” apparently made little difference. In the Jiangnan
Shipyards in Shanghai, a CCP member surnamed Ye was having sex with
a Mrs. Yang, whose husband was in the army. To encourage Mrs. Yang to
divorce her husband, Ye invoked the high probability of defeat if China were
to be attacked: “Do you want to become a counterrevolutionary when Chiang
Kai-shek attacks the mainland and returns here?” “You’ll be a widow if you’re
married to a soldier.” Some of these relationships were long term. In Songjiang
county, a son of a “counterrevolutionary” surnamed Zhang “has been hav-
ing sex with a woman surnamed Wu since her husband went into the army in
1956.” Lacking effective birth control measures or perhaps not caring enough
Why Archives? 47

about this, Wu had become pregnant three times by Zhang, and with his help
managed to secure a divorce. Divorces and pregnancies were not unusual –
fifty-four babies were born from these Shanghai relationships, with another
thirty expected in 1965 – and thirty-two soldiers were divorced, in addition
to thirty-two who had their engagements broken off because of adulterous sex
(SMA1, 1964: 14).
Peasants were well aware of this problem. The praise heaped on the PLA
by state propagandists for patriotically defending China against external ene-
mies could not counter the more real, tangible domestic threats faced by PLA
soldiers. In contrast to hundreds of memoirs penned by urban students who
stressed the attractiveness and patriotic glory of military service in the 1960s,
archival documents tell of villages in which youth were warned, “If you’re
married, you’d better not join the army.” Soldiers who got wind of these prob-
lems requested leave to “visit relatives” but were mainly interested in making
sure that village officials had not already seduced their fiancées and wives.
Unsurprisingly, municipal officials were concerned because recruiting and
military morale were adversely affected, but they did not take much action to
change the situation. When confronted, local officials claimed that they were
not responsible (“If the soldiers don’t lodge a complaint, there’s nothing we
can do about it”), blamed the PLA soldiers’ wives, or argued that sex between
PLA wives and villagers was “just an ordinary sex problem” which did not
merit further attention. The archival evidence is mixed on the issue of whether
this lax attitude spread to the upper echelons of the party. In one account,
courts and party organizations rendered 146 verdicts on these cases, with
close to half (43 percent) of the perpetrators ordered to undergo “criticism and
education,” and the rest were punished more seriously (28 percent receiving
criminal sentences, ranging from one to five years and more). Because level-
ing an accusation against a party official was (and remains) no small matter
and sex-related offenses can be particularly embarrassing, these cases might
represent a fraction of what actually occurred in villages; “lumping it” was
probably common (SMA1, 1964: 13–14).
But, however many officials’ heads were shaking and pens scrawling, there
is little evidence that the situation improved much between 1964 and 1966, the
peak years of pre–Cultural Revolution hyping of the PLA. A nationally circu-
lated “Situation Report” authored by the Political Bureau of the Guangzhou
Military Region in March 1966 (SMA2, 1966: 24–25) and distributed nation-
wide (and therefore available in the Shanghai archives) notes that sixteen
cases of adultery had occurred “only recently”; six were classified as involving
“seduction/rape,” whereas eight were said to have been consensual: the sol-
diers’ wives had “low political consciousness and were attracted to material
possessions” (SMA2, 1966: 24).
According to this report – classified as top secret – this phenomenon was
not limited to South China; Shandong, Shanxi, and other northern provinces
reported comparable problems. In Shandong, for example, a soldier’s wife
named Wen was raped by the village accountant, a man surnamed Wang. Wen
48 Neil J. Diamant

reported Wang to the county police, but because “they did not handle the case
expeditiously,” the accountant “snuck into her bed again.” Fighting Wang off,
Wen grabbed his leg and held on as he dragged himself from the cooking area
toward the door, but he escaped by bashing her finger. Word of this incident
leaked out. From his military base, Wen’s husband dashed off a letter to pro-
vincial-level public security officials, who dispatched an investigation team.
Upon arriving in the village, however, they found that Wang already had been
tipped off by other officials and escaped. His disappearing act did not last
long. Soon enough, he was back in the village, where he once again confronted
Wen: “I’m back. Let’s see what you can do about it!” (SMA2, 1966: 24).
The Guangzhou Military Region’s report seems to have had some impact,
at least at the level of pushing paper from one bureaucracy to the next.
During the Cultural Revolution’s first summer of 1966, the Supreme Court
and Ministry of the Interior issued a directive to their subordinate units to
look more seriously into the problem of “ruining soldiers’ marriages,” since
many soldiers – whose sacrifices and patriotism were being heralded across
the land – were “unsatisfied” with the situation. In response to this directive,
the Shanghai High Court and Bureau of Civil Affairs convened a joint meet-
ing, which was also attended by officers from the Shanghai Garrison (SMA3,
1966: 28).
Despite the urgent tone in the correspondence leading up to this meeting,
garrison officials were already well aware of this problem – they had spon-
sored a large-scale investigation of twenty-eight work units in Nanhui county
only two years earlier and had reported on the situation in 1965. That earlier
investigation, also preserved in the archives, found that of those twenty-eight
units, twenty-three (82 percent) had soldiers whose wives or fiancées were
involved with other men, with sixty-one cases in total. Garrison officers fin-
gered civilian party officials as the primary perpetrators (a bit over 50 percent
of the cases) but also blamed the courts (whose educated officials were better
at “imagining” the nation) for not taking the issue seriously; the verdicts were
“too lenient” they complained, and “many objections” were raised to them.
One example they provided involved a township accountant, a man surnamed
Su, who was having affairs with two PLA wives (separately or in a ménage-à-
trois is not clear). Local officials were aware of Su’s indiscretions but decided
to look the other way. The case came to light only when one of the soldiers’
wives “drowned her baby in the toilet of a bathroom at the commune seat
when there was a meeting going on” (SMA3, 1966: 29–30).
The impact on soldiers was unmistakable. Some heard about their wives
and immediately went AWOL; some “immediately fainted on their beds and
did not get up for several days.” Soldiers circulated in the village to encour-
age the fathers of draft-age boys to quiz recruiting cadres about “whether
you’ve got your eyes on my daughter-in-law again” (SMA3, 1966: 34). In that
politically hot summer of 1966 the contrast in images could not have been
starker: at the same time that Red Guards were convincing themselves that
they were Chairman Mao’s “little soldiers” (as we know from photographs,
Why Archives? 49

memoirs, and propaganda), real soldiers were going AWOL, their families
were up in arms about cadre abuses, PLA wives were being raped and aborting
babies, and the Shanghai Supreme Court complained of a “very dim concep-
tion of national defense and supporting military families” (as revealed by the
archives) (SMA3, 1966: 38).

Conclusion
How these data are interpreted will, of course, depend on the choice of the
analytic lens through which we view them. Perhaps it was a failure of admin-
istration, seen in the lack of effective supervision over local cadres, or even a
reminder of the power of sex to undermine state goals. Perhaps poverty under-
mines patriotism, which requires time, patience, and resources to cultivate
(unless, of course, one sees these as “natural” political phenomena, which I do
not). Certainly there are elements of truth in these arguments.
In my view, however, the larger and weightier issue was the state’s failure
to cultivate patriotism that was meaningful to the extent that it shaped the
behavior of ordinary people and officials, despite numerous efforts in rhetoric,
education, and propaganda. In significant ways, this should not be surprising.
In many other comparative contexts, nationalism and patriotism have been
strongly linked to the military experience, since the military experience, far
more than most civilian and economic institutions, encourages selflessness,
sacrifice, courage, loyalty, and the pursuit of collective goals. Nationalism and
patriotism have also emerged in the context of universal conscription and mass
wars, but China has never had a national army – only party armies. Citizens,
therefore, never fought as a “nation-in-arms” (neither the Nationalists nor
the Communists were in a position to institute a national draft during World
War II; most conflicts were limited wars, or border wars; educated elites rarely
served in combat, even during the Anti-Japanese War). Moreover, militaries,
especially in the context of mass wars, have been one of the only institu-
tions in society that have brought together people from many social classes,
provided opportunities for upward social mobility for the lower classes, and
generated sympathy for their plight among elites.19 In China, the poor treat-
ment of military dependents suggests that the PLA still had low status, and
this, in turn, had adverse consequences on the development of patriotism and
nationalism.
Virtually none of this chapter’s archive-driven “counter-narrative” about
the treatment of PLA dependents has appeared in Western scholarship, and
certainly not in recently published gazetteers in the PRC, valuable as these
are to scholarship on Chinese politics. In part, this can be attributed to the
operationalization of key concepts: if patriotism is mainly about love and
sentiment, and nationalism focuses on cognition, we will tend to look for
sources that document thoughts and feelings, not action and the treatment of

19
I develop this argument further in Diamant (2009).
50 Neil J. Diamant

others – which are better revealed in archival sources whose texts are closer
to the ground. It can also be explained by more prosaic reasons: it is more
convenient to study elites simply because they write and publish – note the
cottage industry of former Red Guard memoirists and essayists in journals,
magazines, and blogs about China’s position in the world. Although useful for
studying how nationalism and patriotism are understood by elites at a partic-
ular moment in time, such sources tell us far less about how these elites actu-
ally treat their compatriots, particularly their social “inferiors,” which, in my
view, is the more important and meaningful gauge of patriotism and national-
ism. The contrast between the elite “take” on the military and the view from
the archives cannot be starker than in the case I present here. If the PLA and
its personnel had high status, and military service was considered patriotic,
why were officials so worried about recruitment and morale? Why did judges
give lenient sentences to those who violated soldiers’ marriages? Why did so
few pitch in to help military families avoid poverty? The elite-based narrative
cannot easily accommodate the new evidence from the archives.
If even this small cache of archival documents – roughly eleven pages in
total – can raise some intriguing questions, think what would happen to the
field of Chinese politics if thousands of such documents were brought to light.
The impact on pedagogy and research would be difficult to avoid. Would we
be able to teach students that the PLA was venerated prior to the Cultural
Revolution? Could we portray intellectuals as the primary victims of Maoist
rule if archival evidence reveals thousands of “red” and “heroic” veterans
committing suicide in the 1950s? And should we continue to divvy up recent
Chinese political history by “campaigns” – which mostly affected intellectu-
als – if sources provide us with far more information about what happened
in the temporal interstices between them? Much as archival research and oral
history about sexuality and divorce have revealed continuity between the
Maoist and reform periods, I expect that more archival research will com-
plicate any arbitrary or stark division between pre- and post-reform China
(Diamant, 2000; Honig, 2003).
Herein lies the rub. Even though archival sources have the potential to
modify previous interpretations of key political events, to suggest new ways of
understanding the inner workings of the state and its relationship to society
over time, and to provide much-needed insight into the foundational compo-
nents of state-building and citizenship – traditionally the strong suits of the
political science discipline – they might go untapped by political scientists
because of the incentive structure in the discipline and the prevailing pre-
sentism among many China scholars in the social sciences. I hope that this
chapter, by presenting just one effort at definitional innovation with a smatter-
ing of new information, will persuade more scholars and students to delve into
the archives and not be deterred by the risks. It would be rather ironic if most
new insights about key events in Chinese politics were supplied by historians
rather than by political scientists, but, sadly, this seems to be the direction in
which we are heading.
3

The Central Committee, Past and Present

A Method of Quantifying Elite Biographies

Victor Shih, Wei Shan, and Mingxing Liu

The political elite was an early focus of Western analysis of the People’s
Republic of China, partly because information about the elite was relatively
available in the opaque political system. As China opened up, information
about all aspects of Chinese politics became increasingly available, thus
decreasing the relative importance of elite studies in China. In this chapter, we
first argue that elite studies continue to be important and necessary for under-
standing authoritarian regimes like China. Further, we introduce a dataset
of Central Committee (CC) members that combines traditional elite studies
with new coding and statistical methodologies. The data are used to trace sev-
eral basic characteristics of the CC through the entire history of the Chinese
Communist Party (CCP), and to measure the relative influence of Hu Yaobang
and Hua Guofeng in the CC. We further discuss the potential of this dataset
for providing more systematic evidence of how elite characteristics and elite
conflict affect policy and political outcomes in China.

Elite Studies in Authoritarian Regimes

At a time when China was sealed off from the rest of the world, Western
scholars of China relied mainly on the official press and in some cases mili-
tary sources from Taiwan to make scholarly inferences. As the heroic deeds
and – at times – the “errors” of senior leaders were often the foci of official
press coverage, Western scholars naturally made extensive use of this relatively
abundant source of information, thus creating a heavy emphasis on the elite.
Today, China scholars have a wide range of sources available to them, rang-
ing from internal government documents to interviews to survey instruments.

We would like to thank Yang Bo and Li Qiang, both graduate students at Peking University,
for carefully and accurately coding the data used in this chapter. They have both graduated
into government service and will soon become the subjects of our inquiry! We further thank
Nancy Hearst at the Fairbank Center Library for valuable help during the final stage of data
collection.

51
52 Victor Shih, Wei Shan, and Mingxing Liu

The proliferation of new sources has given rise to the study of grassroots-level
political phenomena in China and a decline in the relative importance of elite
studies. However, the authoritarian nature of the Chinese regime provides
compelling reason to continue elite studies.
Perhaps the most important reason to focus on the elite in authoritarian
governments is that power in these regimes tends to be concentrated in the
hands of one or a few leaders, whose preferences, beliefs, and actions can
have a profound influence on political and economic outcomes. In a rigor-
ous study using leaders’ natural deaths as an exogenous variable, economists
Jones and Olken (2005) find that autocrats have a much more pronounced
impact on a country’s growth than their counterparts in democracies because
they can directly influence fiscal and monetary policies as well as adjust politi-
cal institutions. Being familiar with the disastrous outcomes of the Great Leap
and of the Cultural Revolution, scholars of China hardly need convincing
of the importance of individual leaders (Dittmer, 2001; MacFarquhar, 1997;
Schram, 1989; Schwartz, 1966). This approach leads scholars to examine the
writings and biographical background of key leaders in order to uncover the
“thoughts” and experience that guide their action (Schram, 1989; Schwartz,
1966). Alternatively, with the help of in-depth historical analysis, scholars puz-
zle out how preferences or “thoughts” guide leaders through complex political
situations to achieve a set of objectives (Fewsmith, 1994; MacFarquhar and
Schoenhals, 2006). Following earlier works on the elite in Western democra-
cies (Aberbach, Putnam, and Rockman, 1981), China scholars are also begin-
ning to uncover the preference of the political elite via survey instruments
(Dickson, 2003).
Another reason for elite study in authoritarian regimes is that the selector-
ate, that is, those with the power to directly affect leadership selection, is a
much smaller share of the total population than in a typical democracy (Bueno
de Mesquita et al., 2006). Thus, even if authoritarian leaders have some pro-
clivity to provide public goods, they have strong reasons to direct policies
toward fulfilling the interests of supporters in the narrow selectorate so as to
maintain power (Kang, 2002; Shirk, 1993; Svolik, 2005; Tullock, 1987: 17).
A long-standing literature in China studies has built on this understanding of
authoritarian politics, and insights developed in this literature have yielded
highly robust predictions of policy outcomes (Bachman, 1991; Manion, 1993;
Pei, 2006; Shirk, 1993). In this framework of analysis, understanding the
backgrounds of the political elite is also important because interests are often
determined by institutional affiliation or past experiences (Lieberthal and
Oksenberg, 1988).
The general comparative literature also focuses on the dynamic interactions
among the elite because these interactions often lead to palpable political or
economic outcomes. For example, Ramseyer and Rosenbluth (1998) demon-
strate that fierce factional rivalry between the Meiji elite in Japan led to the
formation of competitive alliances with forces outside of the oligarchy and to
the eventual collapse of the oligarchic arrangement. Although elite rivalries
The Central Committee, Past and Present 53

are readily observable in democracies because rivals often belong to different

parties and the media provide extensive coverage of competition, this is not
the case in most authoritarian regimes. Outside observers are often unaware
of competition until someone has been unseated or a coup has occurred.
In studies of Chinese politics, elite rivalry became a focus of inquiry after
the outbreak of the Cultural Revolution, which saw Mao systematically purg-
ing nearly all of his former colleagues in the Politburo Standing Committee. A
seeming “round-table” arrangement among the top leaders instantly slid into
a frenzy of mutual accusation and purging (Teiwes, 1993). Clearly, though dif-
ficult to observe, elite conflict is an important aspect of authoritarian politics.
The Cultural Revolution has spawned an enormous literature that tries to
understand its origins (e.g., MacFarquhar, 1997), manifestations at the local
and grassroots levels (e.g., Forster, 1990; Perry and Li, 1997), and its long-
term impact on society (e.g., Walder and Su, 2003).
In sum, elite analysis remains important in studies of authoritarian regimes
because the preferences, values, and background of the elite, and the often
competitive interactions among the elite, continue to play a crucial role in
shaping political and economic outcomes. With the proliferation of new tools
and sources of information in formerly hermit regimes like China, instead of
abandoning elite studies, we should use these new sources to expand the scope
and depth of them.

The Central Committee Database

In the tradition of elite analysis, this project focuses on the characteristics of
an institutionally defined group of the elite – the Central Committee members.
Although the Central Committee is by no means the universe of the power
elite in China, one can reasonably argue that most officials holding important
positions are CC members. There are obvious exceptions – for example, Deng
Xiaoping and Chen Yun in the 1990s – but even they had been CC members
for a long time. Chen Yun to this day holds the record as the longest serving
CC member (1930–1987). CC members further wielded real power by control-
ling specific bureaucracies (i.e., provinces, ministries, military regions, and
so forth) and by exerting influence on the selection of the top leaders of the
regime (Shirk, 1993).
Given the importance of Central Committee members, it behooves us to
know as much as possible about them. Indeed, various studies already examine
the traits of CC members during various periods of time (Bo, 2004a; Cheng
Li, 2001; Lee, 1991; Li, 1994; Nathan and Gilley, 2002). Although these stud-
ies give us intimate understandings of subsets of the CC elite during various
periods, we still lack an overall sense of how members of this elite body have
evolved over the more than eighty-five years of CCP history. Furthermore,
without data on CC members over time, it remains difficult to make causal
inferences on how membership characteristics and the overall makeup of the
CC affected political and policy outcomes and vice versa. This has produced a
54 Victor Shih, Wei Shan, and Mingxing Liu

field of study rich in theory but remarkably lacking in quantitative assessments

of how the power elite influence policy outcomes. To the extent there has been
work (Huang, 1996; Landry, 2008a; Shih, 2004; Su and Yang, 2000), it mainly
focuses on a subset of the CC elite, the provincial and municipal leaders.
In the following, we first describe the conceptual underpinnings of a quan-
titative dataset that tracks the careers of all CC members from the First Party
Congress to the Sixteenth Party Congress. We then provide an account of how
we implemented the coding, the problems we encountered, and the solutions
we devised to deal with these problems. In order to develop a comprehensive
database of CC members, one has to overcome several conceptual and practi-
cal problems. First, the end-product must allow users to generate various indi-
cators of interest without having to hand-code additional information from
printed sources. New information on CC members only needs to be added to
the existing database. This is a tall order for two reasons. First, beyond basic
characteristics, such as birth year, education level, year of obtaining party
membership, and so forth, CC members typically rotated through a series
of positions over the course of their careers that are challenging to track.
Furthermore, many CC members, especially senior officials, concurrently
served in different positions, which need to be expressed in a quantitative
database without generating confusion.
Inspired by work done by Adolph (2003) on the career trajectories of cen-
tral bankers, this database overcomes the above difficulties by coding the
positions, start year, and end year of nearly all the positions held by CC mem-
bers throughout their careers, rather than merely those positions held by CC
members at the time that they served in the CC. This produces a database
where every row is a CC member. For every position a CC member held, three
columns are dedicated to coding it – a numerical code to describe the position,
the start year, and the end year. In this manner, if one adds enough columns,
one can track their movement over time as well as the positions they held
simultaneously. The example in Table 3.1 contains an illustration of how the
database codes Chen Yun’s State Council career in the 1950s. As one can see
in Table 3.1, the multiple positions held by Chen Yun simultaneously do not
create any confusion under this coding scheme. We clearly see that Chen Yun
served as chair of the Finance and Economic Committee, which overlapped
with his duties as vice premier and minister of commerce. The positions held
by Chen Yun are arranged in sequential order, although they need not be.
With this data configuration, we are able to make logical inquiries such as:
locate all vice premiers who served between 1949 and 1970

or, since we coded the CC members’ tenure in a number of different posts:

locate all CC members from the Twelfth Party Congress who had prior military expe-
rience and also served as the party secretary of Sichuan

For the ease of coding and data processing, we further broke down the positions
into various categories: People’s Liberation Army (PLA) posts, State Council
 The Central Committee, Past and Present 55

Table 3.1. Tracking Chen Yun’s positions in the 1950s

Name Chair, Start End Vice Start End Minister of Start End
Finance and Year Year Premier Year Year Commerce Year Year
Economic
Committee
Chen 1421 1949 1956 1025 1949 1975 1321 1956 1958
Yun

posts, provincial posts, National People’s Congress (NPC)/Chinese People’s

Political Consultative Conference (CPPCC) posts, court positions, university
posts, and positions in the CCP itself. For party experience, we coded the mem-
bership of the Central Committee and the positions in the central party organs
separately. For CC membership, we coded both full and alternate membership,
as well as the sessions (from the First CC to the Sixteenth CC). For party posi-
tions, we first specified the central leadership positions, such as party chairman/
general secretary, Standing Committee of the Politburo membership, member-
ship on the Politburo, and alternate membership on the Politburo. Then we
scored the composite units of the central party organization. In addition to
the core organs such as the Secretariat, the Department of Organization, the
Department of Propaganda, and so forth, we also coded the leadership in party
newspapers and journals, such as the People’s Daily, the Guangming Daily,
and Red Flag, as well as leadership positions in party-directed mass organiza-
tions such as the All China Women’s Association.
For each position, we assigned a four-digit number to represent it. The first
three digits denote the state or party organ in which the position is located.
The last digit represents the level of the position. The score 3021, for example,
breaks down to 302, which stands for General Political Department of the
People’s Liberation Army, and the final digit – 1 denotes the highest level
in that department – the chief of the General Political Department. In most
cases, we used a “1, 3, 5, 7” system in which 1 represents ministerial positions,
3 represents vice ministerial positions, 5 represents departmental positions
(司, 局, 厅), and 7 represents those positions below departments. With this
coding system, we were able to quantify the promotions of all CC members
throughout their careers.
We additionally included some demographic variables, which facilitate
analysis of the background of CC members. We coded basic demographic
details including birth year, gender, party membership year, level of education,
ethnicity, princeling status, and whether and when the person was purged or
rehabilitated. For education, we specified the levels of education and, if they
attended university, the specific university they attended. To track the changes
in the school networks among CC members, we created a list of colleges and
universities on the basis of the Inventory of National Universities on the Web
56 Victor Shih, Wei Shan, and Mingxing Liu

site of the Ministry of Education (MOE) and employed the official school code
as our scores for the universities.1 We further traced their majors in univer-
sity and graduate school, when information was available. We additionally
divided the CC members’ careers before they entered any government offices
into major categories, including employment as workers, teachers, soldiers,
and so forth. With this variable, we obtained an image of the person prior to
entering politics.
An important aspect of the career trajectories of many CC members was
their service in various units before 1949. Because it is too unwieldy to code
their participation in specific military units, we coded their participation in
the Long March, their main base area before the Long March, and their main
base area after the Long March. For base area experience, since the bases
shifted substantially over time, we decided to code them according to major
regions and revolutionary experience, which were relevant during the various
periods of the revolution. These regions’ revolutionary experience included the
central Soviet area in Jiangxi, the Shaanxi-Gansu-Ningxia area, the northeast
area (Manchuria), the North (Northern China and Xinjiang and Qinghai,
except Shaanxi-Gansu-Ningxia), the South (Southern China except Jiangxi),
the New Fourth Army, the Eighth Route Army, the Hubei-Henan-Anhui-
Northern Jiangsu base area, the Southwest (including Tibet), and overseas
work. We created additional dummy variables recording crucial experiences
during the revolution, including experience in the Anyuan mine in the 1920s,
experience in Jinggangshan, teaching or matriculation at the Anti-Japanese
University, and military experience on Taiheng Mountain.
For the experience during the 1946–49 Civil War, we coded the field armies
(Swaine, 1992). Field army designations contain extremely useful information
about the loyalty of some CC members because field armies were formed out
of existing units – for example, the Eighth Route Army and the New Fourth
Army – and had fairly consistent leadership. We coded five field armies: the
Northwest Field Army (or First Field Army), the Central Field Army (or
Second Field Army), the East China Field Army (or Third Field Army), the
Northeast Field Army (or Fourth Field Army), and the North China Field Army
(or Fifth Field Army).
With the inclusion of demographic and career variables, the dataset cur-
rently has 1,604 observations (rows) and 261 variables (columns). As new
positions of interest arise, it will be relatively easy to append additional codes

1
There are 1,607 universities and colleges on the MOE Web site. We selected the top 102 schools
according to the 2003 university ranking by netbig.com (there are ties in the ranking so we
selected 102 instead of 100 schools). In addition, we added Peking Union Medical College, the
Chinese Academy of Sciences, the Chinese Academy of Social Sciences, and the Central Party
School which are not included on the MOE list or on the netbig ranking but are prestigious in
their fields. Also, we included the Southwest Union University and the Anti-Japanese Military
and Politics University. These two do not exist today but were influential before 1949 and
produced a number of CCP leaders. Finally, two general categories are also included: military
academies and other schools. In total, there are 110 values in this variable.
The Central Committee, Past and Present 57

or columns to describe them. This configuration of the database makes it

­easily expandable.

Implementing Coding
The coding itself took place in several stages. We first hired a research assis-
tant (RA), a master’s candidate at the School of Government Management at
Peking University, to make a survey of the evolution of party, State Council,
and PLA organizations over time since we needed to assign a code to nearly
every position across these three institutions. After making a complete cata-
log of the positions we potentially wanted to code, we selected a subset of
them and began coding. The main source we used was the Dictionary of Past
and Present Central Committee Members (Dictionary) (Central Organization
Department and Party History Research Center, 2004). We supplemented it
with various Western and Chinese sources, as well as sources available on
the Internet (Bartke, 1997; Bartke and Institut für Asienkunde [Hamburg,
Germany], 1991; Lamb, 2003; Mainland China Research Center, 2006;
Jianying Wang, 1995). To fill in the missing data, we also conducted some
in-depth historical research, collecting biographical details especially on early
revolutionaries (e.g., Cai, 1995; Kou, 2008; Xu, 2005).
To ensure integrity in our coding, we hired two RAs – both master’s candi-
dates at the Peking University School of Government Management – to code
the data. Each RA coded the entire CC membership independently of the
other. We first composed a coding manual and trained both RAs in our coding
concept, and one of them made a trial coding of a 3 percent random sample
out of the Dictionary to check the feasibility of the coding manual. We then
revised the manual according to the problems revealed by the trial coding,
and both RAs began to code. We entertained scanning the Dictionary into
digital format, but it did not prove necessary as the RAs soon became quite
proficient in using the paper text. Over the following two years, as the cod-
ing progressed, new problems emerged and we continued to revise the ­coding
manual. Despite these changes, the fundamental logic of the database remained
the same – position, start year, end year. All three authors of the chapter
monitored the coding progress of the two RAs, both in person and through
the Internet.
After both RAs completed the coding, we had them cross-check their
entries with each other. If their codes matched, we left them as they were.
This applied to the vast majority of the cases. If there were any discrepancies
between the two, they noted them down, and the three authors “judged” the
merits of the two versions. This laborious process was completed in July 2006,
when we obtained a preliminary dataset. We then found that there were quite
a few missing data points, especially concerning the start and end years for
various positions. To fill in the missing data, the authors conducted further
research to fill in as many of the blanks as possible. By October 2006, we had
fewer than 400 missing data cells in a dataset of over 417,000 cells.
58 Victor Shih, Wei Shan, and Mingxing Liu

Problems and Solutions

After we began coding, problems emerged, which we dealt with in various
ways, some more satisfactory than others. First, as briefly discussed earlier, we
discovered that it was extremely difficult to track the pre-1949 military experi-
ence of various CC leaders. Units were created and disbanded in days, whereas
others were destroyed and reformed under the same name in a totally different
place. Furthermore, there was a blur between civilian positions and military
positions during the revolutionary period. To a large extent the party was no
more than a leadership group of the army. After all, the base areas were not
only territorial areas administrated by the party but also military bases for
the Communist army. Thus, before 1946, we coded CC members’ experience
mainly according to their base areas.
This problem became especially serious in the 1945–49 period when the
establishment of geographical administration overlapped with military units.
With the information we were able to collect we had no clear picture of the
relationship between the specific PLA units, the field armies, and the base
areas. For example, for a vice director of the Political Department of the
Shanxi-Chahaer-Hebei Military Area in 1945–48, Cai Shupan, do we code
him as an officer within the North China Field Army or as a local cadre of the
Shanxi-Chahaer-Hebei Base Area? Was the military area a subordinate unit
of the field army, or an administrative unit of the base area? Because we had
insufficient information on these issues, we instructed our coders to adhere
to the following principle: although military positions were usually coded as
field army experience, when we were unsure of the field army designation of
a military position but were more certain about the geographical location
of this unit, we coded the position according to the base area and left blank
the field army designation. The second major problem we encountered was
that there were simply too many positions in the Chinese bureaucracy due to
the multiple levels of government. Every ministry has departments (司, 局),
which in turn have bureaus (处), which in turn have sections (科). Every prov-
ince has both functional departments (厅) and geographical jurisdictions,
including prefectures, counties, township (brigades), and villages. Of course,
within each geographical jurisdiction, the various functional departments are
replicated. An attempt to code every single position in the vast bureaucracy
would make coding extremely unwieldy. Instead, we anchored our coding to
the ministerial/provincial level. That is, for all positions below or at the minis-
terial/provincial level, we identified them as a position under that ministry or
province at one of four levels: below the departmental/prefecture level (司, 局,
厅, 地 ), at the departmental/prefecture level, vice ministerial/provincial level,
or full provincial/ministerial level. We made some modifications to the pro-
vincial coding to distinguish between functional departments and the various
geographical subunits and to distinguish between the governor and the party
secretary. We also devised a similar coding scheme for military regions and
the main departments and branches of the PLA.
The Central Committee, Past and Present 59

One drawback of this approach is that we do not track movement of CC

members through various prefecture-level cities or departments within minis-
tries. For example, the dataset specifies that a CC member (say, Li Yuanchao)
served as a prefecture or county mayor or party secretary in Jiangsu, but does
not specify that this person was in fact the party secretary of Nanjing. We
feel that this coding strategy does not forgo too much information, especially
if one is interested in analyzing elite politics. Usually, promotions above the
departmental/prefecture level signal one’s entrance into national politics.
The third major problem we encountered was the frequent changes in the
State Council bureaucracy and shifts in the military region structure. For
example, the First Ministry of Machinery was merged into the Ministry of
Machinery before folding into the State Economic and Trade Commission
and eventually the State-owned Assets Supervision and Administration
Commission. In general, we adopted a principle of successor. That is, in prin-
ciple, all codes of the party and state organs are based on their latest manifes-
tations after 2002. Since we have tracked the evolution of all organs, we gave
agencies in the past the same scores as their successor agency in the current
government. For example, we coded both the Ministry of Domestic Trade and
the Ministry of Foreign Trade and Cooperation as the Ministry of Commerce
because the former two were incorporated into the latter in 2003. This of
course leaves out important information, especially in the State Council,
where many top leaders emerged out of the now-abolished machinery or pet-
rochemical ministries. Toward that end, we created three general categories
to score experience in these agencies: the machinery ministries (including the
former 1st, 2nd, 3rd, 6th, 7th, and 8th Ministries of Machinery), the petro-
chemical ministries (including the former Ministry of Petroleum Industry and
Ministry of Chemical Industry), and light industry ministries (such as the for-
mer Ministry of Textile Industry). We likewise applied the successor principle
to military regions and party organs. A good case in point was the Xinjiang
Military Region (MR). This MR became subordinate to the Lanzhou MR, so
we coded positions in the Xinjiang MR with the values of the same positions
in the Lanzhou MR. We also applied this principle to code provincial revolu-
tionary committees during the Cultural Revolution.
Work experience in enterprises (企业) and institutes (事业) after 1949
also presented a messy problem. During the planned economy era, all enter-
prises were supervised by governmental organs and behaved like a part of
their supervising organs (Naughton, 1996). They had the same bureaucratic
ranks as departments or bureaus in the supervising agencies. In our coding,
we first tried to link enterprises to their supervising party-state organs because
the career paths of many party leaders, especially the technocrats, began in
these enterprises. In some cases, finding out the supervising agencies became
extremely difficult, if not impossible. We also obtained a list of enterprises
currently managed by the State-owned Assets Supervision and Administration
Commission (SASAC) and coded these enterprises and their recognizable
“ancestors” as under SASAC or various line ministries. This still left many
60 Victor Shih, Wei Shan, and Mingxing Liu

enterprises uncoded. We then created a variable that records the function of

an enterprise: service units (事业单位), transportation, electronics/informa-
tion, electricity/hydrotechnics, petroleum, mechanical/steel, light industry,
and others. We followed this coding with the start and end years of the CC
members’ tenures in these enterprises.
Despite these problems and the less-than-perfect solutions we devised, we
still have a firm grasp of the career trajectories of CC members once they took
administrative office. We also have nearly perfect information on their party
and state posts. For the vast majority of CC members, the solutions we devised
gave us a fairly clear picture of their early careers. Most important, with the
basic logic of our dataset, if a research project compels it, we have the flex-
ibility to incorporate additional information into the dataset to address new
issues.

Annual Indicators
Given the dataset, what is one to do with it? As a first step, we generated
a number of annual indicators that provide year-to-year descriptions of the
Central Committee as a whole. It has become a fairly regular exercise to derive
various characteristics of the CCP elite over certain periods of time or for
various party congresses (Bo, 2002; Bo, 2004b; Cheng Li, 2004, 1994, 2000,
2001). These exercises have yielded important insights into the social charac-
teristics and preferences of the CCP elite.
Although one can tabulate various characteristics of CC members for cer-
tain years, it becomes difficult to hand-count the number of Long March
veterans – for example – over half a century. Furthermore, previous works
tabulating the characteristics of CC members covered only a small number of
years, especially during the years of the party congresses (PCs) (Baum, 1998;
Dittmer, 1983; Saich, 1992; Starr, 1976; Wich, 1974). Major changes often
occurred during PCs, but shifts in CC membership also took place between
congresses. For example, as the analysis below reveals, the 1985 National
Party Conference convened by Deng had a major impact on the age and edu-
cation structure of CC members. Finally, as one examines an increasing num-
ber of characteristics, hand-counting becomes unwieldy. Even fully mapping
CC members’ characteristics for one year becomes a laborious exercise (Bo,
2004a).
Because this dataset records the start and end dates of most positions held
by CC members, it becomes a relatively simple matter to derive various char-
acteristics of CC members over time. Granted, devising such indicators still
requires a careful crafting of complex logical statements in statistical software
(Stata 9). As mentioned above, there are some drawbacks to the dataset, par-
ticularly the inability to code many characteristics before 1949. Nonetheless,
as the exercises below reveal, a thorough examination of the annual trends of
these characteristics provides further empirical support for some of our core
intuitions and generates new insights about CC members.
The Central Committee, Past and Present 61

25 30 35 40 45 50 55 60 65 70
age

1920 1930 1940 1950 1960 1970 1980 1990 2000

year
Mean Age Age, 75th Percentile
Age, 25th Percentile

Figure 3.1. Average age and 25th and 75th percentile age of Central Committee
members, 1921–2006.

Age Structure of CC Members over Time

From a Western perspective, the CCP is often known for its gerontocracy.
However, the analysis here reveals that the CCP’s tendency to have an elderly
elite mainly stemmed from the dominance of the May Fourth generation over
the party for several decades. Both Mao and Deng, in their own ways, car-
ried out remarkable campaigns to rejuvenate the elite of the party. By the late
1980s, a system to regularly replace older officials with younger cohorts was
firmly in place.
In Figures 3.1 and 3.2, we see that the initial CCP elite did not come from
the May Fourth generation, who were born around 1900 or so. In fact, in
the early 1920s most CC members from the First to the Third CC were sub-
stantially older than that. The Comintern clearly did not trust eager twenty-
somethings with the early CCP, preferring more experienced revolutionaries
such as Chen Duxiu and Dong Biwu. Once the May Fourth generation took
over the party in 1934, however, it remained in charge until the Ninth PC in
1969 (Figure 3.2). Thus, the nearly continuous rise in the average age of CC
members from 1934 to 1969 seen in Figure 3.1 was produced by the same age
cohort controlling the CC during that period, as seen in Figure 3.2. In that
period, the average age of CC members rose linearly from 33.7 to 55, with a
minor rejuvenation at the 1956 Eighth Party Congress.
In addition to many other things, the Cultural Revolution represented a
substantial rejuvenation of the party. The 1969 Ninth Party Congress saw the
formal removal of many of the May Fourth generation from the CC, causing
the average age of CC members to plummet from 64 to 55. This spelled the
end of the May Fourthers as the dominant generation in the CC. Although
the average age of CC members crept upward toward the end of the Cultural
62 Victor Shih, Wei Shan, and Mingxing Liu
1960
1940
1920
age
1900
1880

1920 1930 1940 1950 1960 1970 1980 1990 2000

year
birth year, 25th percentile birth year, 75th percentile
Mean birth year of CC members

Figure 3.2. Average birth year and 25th and 75th percentile birth year of Central
Committee members, 1921–2006.

Revolution as many veterans were rehabilitated into important positions, the

average CC member nonetheless came from the 1914 cohort instead of an
earlier cohort. Clearly, many CC members from the May Fourth generation
never returned to active duty.
The rejuvenation effort continued relentlessly into the Deng era (Manion,
1993). The Twelfth PC in 1982 saw the displacement of the 1914 cohort with
those born around 1919, which brought the average age of CC members from
66.5 to 62. The most momentous rejuvenation effort engineered by Deng and
Chen Yun rivaled even that carried out by Mao. The 1985 National Party
Conference and the Thirteenth Party Congress removed wholesale the genera-
tion of leaders born before 1919 in favor of much younger officials from the
cohort born after 1929. This brought the average age of CC members from
64.5 in 1984 down to 58 in 1987. Thereafter, the age structure of CC members
stabilized into the regular pattern of late 50s at the time of a party congress,
followed by a gradual rise between congresses toward the low 60s. By these
indicators, Deng Xiaoping and Chen Yun were successful in institutionalizing
mechanisms to continuously rejuvenate the CCP.
Another way to track the successful rejuvenation of the party is to fol-
low Long March veterans in the CC through time. As Figure 3.3 shows, the
number of Long Marchers in the CC rose steadily through the late 1930s and
1940s and enjoyed a dramatic rise at the 1956 Eighth PC. Although the rapid
expansion of the CC in 1956 saw a slight decline in the share of CC members
who were also Long Marchers, these veterans nonetheless occupied over 60
percent of the seats in the CC. The dominance of the Long Marchers in the CC
was maintained until the Cultural Revolution, when the Ninth PC inducted
many from a younger generation into the CC. This is despite the fact that the
 The Central Committee, Past and Present 63

80

80
number of Long Marchers
% of CC Long Marchers
60

60
40

40
20

20
0

0
1935 1945 1955 1965 1975 1985 1995
year
% of CC Long Marchers number of Long Marchers

Figure 3.3. Percentage of Central Committee members with Long March experience
and the number of Long Marchers.

Ninth CC included the highest number of Long Marchers. The high number
of Long Marchers in the CC in 1969 is not surprising since many commanders
of the military regions took over as heads of the various provincial revolution-
ary committees, granting them passage into the CC.
Lin Biao’s purge in 1971 led to a substantial decline in Long Marchers in
1972 and 1973 as his suspected followers were systematically removed. Long
Marchers then enjoyed a brief revival with Deng’s rehabilitation in the late
1970s and early 1980s, but policies enacted by Deng and Chen, in addition
to illness and death, produced extremely sharp drops in both the number
and share of Long Marchers in the CC at the Twelfth and Thirteenth PCs.
Although many Long Marchers remained healthy into the early 1990s, the
Fourteenth PC did not induct any Long Marchers into the CC, thus spelling
an end to the Long Marchers’ influence in formal politics. Without an active
policy to pressure Long Marchers into retirement, the Long March generation
likely would have enjoyed a few more years of formal influence.

Education of CC Members
How did the education level of CC members evolve over time? As seen in Figure 3.4,
political events clearly had a substantial impact on the average education level of
CC members. In Figure 3.4, the 0 to 3 scale on the Y-axis represents less than
a high school education, high school education (or equivalent), college educa-
tion, and graduate school education. As the figure reveals, the CC began as a
highly educated body dominated by cosmopolitan intellectuals. After the CCP
split with the Guomindang (KMT), the party inducted into the CC many labor
leaders who had participated in anti-imperialist strikes in the mid-1920s, which
led to a plummeting of the average education level of CC members to well below
 64 Victor Shih, Wei Shan, and Mingxing Liu
3
2.5
2
mean level of education
1.5
1
.5
0

1920 1930 1940 1950 1960 1970 1980 1990 2000

year
Figure 3.4. Average education level of Central Committee members, 1921–2006.
Note: On the Y-axis, o denotes less than high school or equivalent; 1 denotes high
school or equivalent; 2 is college or equivalent; 3 is graduate level education.

high school level. Many of these new CC members turned out to be highly unre-
liable and either defected to the KMT or absconded with party funds.
As the party shifted toward a rural focus in the early 1930s, however, the
average education level of CC members climbed back toward the mid-point
between high school and college. The CCP likely learned not to place too
many responsibilities on the shoulders of the uneducated. Although the CCP
elite actively recruited peasants during the Jiangxi Soviet period, few of them
were immediately inducted into the CC. This accords with Benton’s (1992)
finding that the southern guerrilla bases in the early 1930s often saw splits
between the intellectual “leftist” elite sent down by the party center and
the local guerrilla forces of mostly uneducated peasants. Instead of induct-
ing uneducated peasants into the party center, the average education level of
CC members remained the same throughout the Long March and the Yan’an
period, a mix of high school and college-educated leaders. Some of the peas-
ant fighters recruited in the 1930s, especially those who had participated in
the Long March, were finally inducted into the CC at the Eighth PC in 1956,
which caused a sharp drop in the education level.
The average education of CC members dropped below high school level
during the Cultural Revolution as Mao introduced a mix of veteran peasant
fighters and mass representatives into the CC at both the Ninth and Tenth
CCs. The lowering of the average education level at both the Eighth and the
Ninth PCs likely was a manifestation of Mao’s “mixing in sand” tactic of fill-
ing the ranks of the elite with loyal, though not necessarily capable, leaders
(MacFarquhar and Schoenhals, 2006: 333).
The Central Committee, Past and Present 65

After the fall of the Gang of Four, the average education of CC members
slowly recovered, although the pace was much slower than one would expect.
Again, due to the active retirement policy pursued by Deng and Chen, the aver-
age education level of CC members climbed from nearly high school level just
before the Twelfth PC to over half-way toward college level by the Thirteenth
PC. As the norm of meritocracy was established, the average education level
of CC members climbed with each subsequent party congress. Although the
education trend of CC members supports the notion of a transition from revo-
lutionary cadres to technocrats (Lee, 1991), this transition did not have a clear
effect on the elite makeup of the CCP until well into the Deng era. The average
education level climbed above college level at the Fifteenth PC in 1997 and
continued its ascent at the Sixteenth PC. The membership of the CC is now
among the most educated elite in the world.

Factional Influence
Finally, this dataset is useful to derive indicators of factional influence. If
one believes that factions are formed on the basis of shared native place as
well as common education and work experience (Lieberthal and Oksenberg,
1988: 156), then these biographical data would enable us to infer the share of
CC members with factional ties with top leaders. The purpose of this exercise
would be to track the relative balance of power between various top leaders,
which would also provide an indication of how fragmented the party elite is
at any particular moment.
In the past, a main challenge of coding factional ties was the enormous
amount of work entailed in uncovering how the numerous members of the CC
were tied with Politburo Standing Committee (PSC) members. To be sure, it
is a relatively simple task to find out all the Tsinghua University graduates or
all the Hunan natives among CC members at any given time, and important
work on factions has been done on the basis of these simpler biographical
coincidences (Li, 1994). Nonetheless, it becomes much more difficult to track
job coincidences to capture all CC members who had worked in the same unit
as a PSC member, especially one with a rich career. Taking Hu Yaobang as
an example – he was a native of Hunan and a graduate and instructor of the
Anti-Japanese University. However, he also held the following positions before
becoming general secretary of the CCP: a senior cadre in the Communist Youth
League between 1936 and 1937 and between 1952 and 1966; a senior cadre
in the Central Military Commission between 1939 and 1946; a commander
in various units of the North China Field Army between 1946 and 1949; a
member of the Southwest Military and Political Committee between 1953 and
1956; one of the party secretaries of the CC Northwest Bureau between 1964
and 1965; the vice president of the Central Party School between 1977 and
1982; the head of the Central Organization Department between 1977 and
1978; and finally the third secretary of the Central Discipline and Inspection
Commission between 1979 and 1982.
66 Victor Shih, Wei Shan, and Mingxing Liu
.35
.3
.25
.2
.15
.1

1970 1975 1980 1985 1990

year
Share of HYB Supporters in CC Share of Hua Supporters in CC

Figure 3.5. The share of Central Committee members with common experience with
Hu Yaobang and Hua Guofeng, 1970–1990.

To find all CC members who had worked with Hu at the Twelfth PC, for
example, one would have to carefully examine the biographies of all CC mem-
bers elected at the Twelfth PC. If one wanted to track those who had worked
with Hu over time, one would have to examine every CC member’s biogra-
phy over a period of time. This exercise would be even more laborious if one
wanted to compare Hu’s influence with that of Hua Guofeng, whose followers
presumably would be quite different from Hu’s. Although it took us quite a
while to code the CC dataset, because we have the start and end years of most
of the positions held by CC members, we can deploy a computer algorithm (in
Stata) to search for CC members who had worked at the same time and in the
same unit with Hu Yaobang. We can do this for every year during Hu’s career,
but especially during the years when he served as party general secretary.
Although it is by no means a simple exercise, the availability of the data and
the use of a computer algorithm make comparing the influence of various PSC
members in the CC a manageable task.
In Figure 3.5, we present a comparison of the share of CC members who had
common experiences with Hu Yaobang and Hua Guofeng, two contenders for
power in the late 1970s. As one can see, Hu Yaobang’s wide-ranging experi-
ence gave him a large advantage over Hua Guofeng throughout the entire
period, even before the Eleventh PC. Hu’s wide-ranging experience afforded
him shared experiences with a much greater proportion of the CC throughout
much of the 1970s and 1980s. In contrast, Hua Guofeng served only in pro-
vincial positions in Shanxi, Hunan, and Guangdong, as well as a short stint
as the minister of public security before becoming the party helmsman. Thus,
although the Eleventh Party Congress in 1977 allowed Hua to induct some
The Central Committee, Past and Present 67

cadres with shared experiences into the CC, his influence in the CC still paled
in comparison with Hu’s, who shared experiences with roughly 30 percent of
the CC in 1977.
Why couldn’t Hua have blocked Hu followers from entering the CC at
the Eleventh Party Congress? Even if we discount Deng Xiaoping’s backing
of Hu Yaobang, it would have been very difficult to do so given Hu’s wide-
ranging experience. In order to block all possible Hu followers from the CC,
one would have had to exclude cadets at the Anti-Japanese University, officers
of the North China Field Army, and senior cadres in the Communist Youth
League for much of the 1950s and 1960s. As these cadres made up some of
the best human capital in the party at the time, it simply would have been
unimaginable to exclude them all. Although one’s influence in the CC does not
entirely determine one’s political fortune, the data in Figure 3.5 are certainly
consistent with Hua’s eventual fall and the elevation of Hu to the position of
party general secretary.
To be sure, Hu later suffered his own fall from power at the hands of a
powerful coalition between Deng Xiaoping and Chen Yun (Fewsmith, 1994).
Beyond pushing for a liberal line against Deng’s wishes and causing instability,
Hu’s own influence also began to slip by the mid-1980s due mainly to the death
and retirement of many in his cohort. This may have made removing him from
the party general secretary position easier for Deng and Chen. However, one
can see that even after his demotion and the retirement of many of his followers
at the Thirteenth PC in 1987, Hu remained an influential figure in the CC.

The Next Step

In this chapter, we introduce a new database of all Central Committee mem-
bers that uses an unconventional logic of coding biographies. Instead of focus-
ing on time and sequence, the coding scheme focuses on positions, followed
by when the CC members served in a given position. This logic of coding elite
biographies provides enormous flexibility both to generate various indicators
of elite characteristics and to allow for the expansion of the dataset in various
directions. As a preliminary step, we developed a few basic time-series trends
in the CC. We hope this chapter will pave the way for other researchers to
code other elite bodies in China and elsewhere. For example, it would be fruit-
ful to code all PLA generals, not all of whom were CC members. Likewise, one
could code members of the Central Discipline and Inspection Commission to
examine their role in Chinese politics.
With the CC data, one can conduct three kinds of analyses. First, with these
annual indicators, one can conduct time-series analyses, especially given the
numerous data points provided by the database. Elite characteristics can be
used as either dependent or independent variables. For example, certain types
of political crises might increase the number of PLA representatives in the CC.
Likewise, shifting characteristics of the CC, such as the average education
level of CC members, might explain the pace of adopting new policies.
68 Victor Shih, Wei Shan, and Mingxing Liu

At the other extreme, one can conduct individual-level analyses and treat
every CC member as an observation. With this dataset, one can infer the char-
acteristics that would earn CC members promotions to the Politburo level.
The individual-level data can be augmented by various economic and political
indicators to provide a comprehensive look at the factors that drive elite pro-
motions. Instead of focusing only on regional administrators (Landry, 2008a;
Li and Zhou, 2005), a general theory of the factors that drive elite promotions
across the various segments of the CCP regime can be tested rigorously.
Furthermore, this dataset also allows the researcher to segment the CC
population into sectors or geographical regions and to correlate characteris-
tics of regional or sectoral elite with various policy outcomes. Regional elite
indicators, including regional representation in the CC, provincial factional
affiliations, and the average education level of provincial elite, can be used to
explain a host of regional economic and policy outcomes. With the develop-
ment of these various indicators, we hope to make quantitative studies of the
Chinese elite a less burdensome task.
4

Experimental Methods and Psychological Measures

in the Study of Chinese Foreign Policy

Peter Hays Gries

Do concerns over “face” play a greater role in Chinese than American foreign
policy? What is the nature of Chinese national identity? Can it be empirically
measured and compared to other national identities, and is it consequential
for foreign policy outcomes? For instance, how do Chinese patriotism and
nationalism compare with, say, American patriotism and nationalism?
This chapter argues that experimental methods and psychological mea-
sures provide valuable tools to the political scientist interested in answering
such questions. Experiments have long been the first choice for establishing
causality in the hard sciences. The social sciences are catching on, with psy-
chology and behavioral economics leading the way. Political scientists, led
by Americanists interested in voting behavior, are beginning to follow suit
(Druckman et al., 2006). International relations scholars have taken notice.
As Rose McDermott (2006: 356) has recently argued, “experiments offer a
unique opportunity to make a clear causal argument … which is why it has
been differentially adopted by the hard sciences, psychology, and behavioral
economics as the gold standard method of choice.” It is the random assign-
ment of subjects to experimental and control conditions that allows ana-
lysts to be confident that variation between groups of subjects on dependent
measures was “caused” by variation in the independent variables that were
manipulated. By contrast, the majority of quantitative work in political sci-
ence, which is based on research designs that are correlational in nature, can-
not confidently make causal claims.
The postwar histories of comparative politics in general and China studies
in particular make the idea of utilizing psychological measures in the study of
Chinese foreign policy particularly contentious. In the 1960s and 1970s, mod-
ernization theory emerged as a major intellectual paradigm, with the concept
of “political culture” at its core. Unfortunately, “political culture” was often
used as a residual variable to explain what the theory could not otherwise
explain. Its deterministic conclusions – certain nations could not democratize
because of their “backward” cultures – also came under sustained criticism.

69
70 Peter Hays Gries

In the China field, Lucian Pye (e.g., 1968) took “political culture” in a psy-
choanalytic direction, put the entire Chinese people “on the couch,” and
declared the Chinese race to be stuck at the anal stage of development. Such
arguments rubbed many younger China scholars the wrong way. Many of the
Vietnam generation, often organizing around the Committee of Concerned
Asian Scholars, revolted against Pye and modernization theory. “Political cul-
ture” and psychology in general have largely been taboo topics in the China
field ever since.
The China studies taboo on the study of political psychology is extremely
unfortunate. Psychology in general and social and cross-cultural psychology
in particular have much to offer the political scientist interested in Chinese
politics and foreign policy today. In addition to valuable theory, psycholo-
gists have developed reliable instruments to measure a variety of phenomena
central to the study of Chinese foreign policy. For instance, Luhtanen and
Crocker’s (1992) collective self-esteem (CSE) scale, developed to study gender,
ethnic, and other social identities, can easily be adapted to study constructs
like Chinese and American patriotisms, as is done in the second study in this
chapter. Instead of “I am proud to be a woman or black,” the researcher can
use “I am proud to be Chinese or American” and take advantage of the hard
work in psychometrics Luhtanen and Crocker undertook to develop a robust
16-item CSE scale. The methodology of psychometrics, furthermore, provides
the China scholar with the desire to study new or China-specific phenomena,
such as U.S. policy preferences, with the tools to develop their own internally
reliable measures to tap such constructs.
To illustrate these methodological and measurement issues, I first intro-
duce selected results from two separate studies. The first shows how an
experimental design and survey data can be used to test the popular assump-
tion that Chinese are more driven by “face” in their foreign policy than
Americans. The second shows how psychological measures and exploratory
factor analysis can be used on survey data to inductively uncover differing
Chinese and American structures of “patriotism” and “nationalism.” It then
uses the statistical technique of path analysis to uncover a possible model
of the consequences of variations in Chinese patriotism and nationalism for
the perception of U.S. threats and even U.S. policy preferences. In the final
section, I draw on these two case studies to discuss the challenges and oppor-
tunities of experimental work and psychological measures in the study of
Chinese foreign policy.

“Face” and Foreign Policy: An Experiment

Are Chinese more concerned about “face” in their foreign policy than
Americans? This question was addressed in one part of a larger study (Gries,
Peng, and Crowson, under review) exploring symbolic and material gains and
losses as determinants of (in)security in international relations.
Experimental Methods and Psychological Measures 71

Design
A pair of student surveys was implemented in the United States and China
in 2006. The design included both experimental (random assignment) and
quasi-experimental (natural group) variables. It also included both between
and within subjects designs. In other words, students’ responses were com-
pared to each other’s as well as to their own responses to other questions.
A 2 (domain) by 2 (frame) by 2 (level) by 2 (nation) factorial design with 16
conditions was employed, thus requiring a large sample size (N = 521). But this
complexity allowed for the analysis of four key issues underlying the security
studies debate over the fundamental determinants of (in)security in interna-
tional affairs.

Independent Variables
The core of the design is an experimental 2 by 2 involving domain (mate-
rial/symbolic) and frame (gains/losses). This portion of the design is a pure
between-subjects experiment, with student participants randomly assigned to
one of four conditions: (1) material gain, (2) material loss, (3) symbolic gain,
and (4) symbolic loss. Each condition was operationalized with a set of sce-
narios that participants read and that differed only on the issues of domain and
frame. For instance, “You have been dating your boy/girlfriend for over three
months and realize that you love him/her. You decide to take a risk and tell
him/her that you love him/her. He/She responds by saying that he/she doesn’t
love you anymore and wants to break up” was a symbolic loss scenario.
Both the materialist and symbolic security studies camps rely on an analogy
with individual human needs. Materialists assume that states, like ­individuals,
prioritize survival. Symbolic analysts posit that states are driven by higher
human needs for belonging and esteem. Both camps thus appear to share
the assumption that the dynamics of security and insecurity are the same at
the individual and international levels. To put this assumption to the test, we
included a third variable in our design, level, by adding to the individual-level
scenarios a parallel set of scenarios at the international level. For instance,
“Sports analysts now predict that China will double the American medal
count at the 2008 Beijing Olympics. In their view, China will be the only
sports superpower in the 21st century” was an international symbolic loss
scenario for American subjects.
The final independent variable, and the one of central interest here, in the 2
by 2 by 2 by 2 design is nation (U.S./China). Orientalist notions of a Chinese
obsession with “face” persist today and have a direct bearing on the issue
of symbolic and material gains and losses. The Chinese, both Western (e.g.,
Smith, 1890) and Chinese (e.g., Ho, 1976; Hu, 1944) sources have long told
us, are culturally predisposed to be sensitive to issues of “face.” Americans,
meanwhile, supposedly disregard face in favor of a more rational calculation
of their material self-interest.
72 Peter Hays Gries

Hypothesis: Chinese are more sensitive to symbolic gains and losses than Americans,
and Americans are more sensitive to material gains and losses than Chinese.

To put this hypothesis to the test we first adapted the original English-
language survey to the Chinese perspective. For instance, in the Chinese ver-
sion, the material-gain condition of the energy scenario read, “A Chinese oil
company has just purchased monopoly rights to drill in the two largest oil fields
in Africa, beating out a U.S. company.” This reverses the words “Chinese”
and “U.S.” from the U.S. material gain condition, thus making the content
of the U.S. material-gain version the same as the Chinese material-loss con-
dition, and the U.S. material loss the same as the Chinese material gain. We
then translated the adapted survey into Chinese and then back-translated it to
ensure comparability. (For instance, “一家中国石油公司刚刚打败美国一家公司,
购买了在非洲两个最大油田的独立钻井的权力.”)

Dependent Measures
Each of the scenarios used to tap the different conditions was followed by a
battery of emotional response items. Each was on a 1 (strongly disagree) to 7
(strongly agree) seven-point Likert-type scale. At its most fundamental level,
security means the absence of concern or anxiety. We therefore constructed an
anxiety score by averaging the self-reported responses to the “I feel worried”
(我感到担心) and “I feel afraid” (我感到害怕) items.
Davis Bobrow (2001: 4) has perceptively noted that “threat centered work
provides rich ground for security dilemma spirals of action and reaction, mea-
sure and countermeasure.” He thus urges that the study of threats be balanced
with the study of opportunities. To balance our negative anxiety measure with
a more positive one, we decided to supplement it with a single item measure of
pride, “I feel proud” (我感到骄傲).

Participants and Method

Because our 2 by 2 by 2 by 2 design entailed 16 conditions, and we desired
at least 30 students per condition (actual M = 32.56), a large sample of 521
university students (284 female, 215 male, and 22 who did not indicate their
gender) was recruited to participate in the study on a voluntary basis in spring
2006. Among these students, 240 were American at the University of Colorado
and 281 were Chinese at Peking University. Participants ranged in age from 17
to 32 (median age = 20), and a t-test revealed that the American students (M
= 20.58, SD = 4.44) were only slightly older than the Chinese students (M =
19.88, SD = 2.23), t = 2.27, p = .024.
We tested the Chinese and American participants in 15-minute sessions.
The experimenter told the participants that the purpose of the study was
to assess their reactions to eight scenarios. After assuring the participants
that their responses would remain anonymous, the experimenter adminis-
tered the survey packets. The participants filled out a series of questionnaires
Experimental Methods and Psychological Measures 73

individually. After completing the packet, the participants were thanked for
their participation, debriefed (i.e., informed that none of the scenarios that
they had read were real), and released. The ethical standards of the American
Political Science (APSA) and the American Psychological Association (APA)
were strictly followed during data collection and analysis.

Selected Results
To see whether nation had any impact on our dependent measures, we ran
a series of four three-way (frame X domain X nation) analyses of variance
(ANOVAs). The first, with individual level anxiety as our dependent vari-
able, revealed the main effects of gain/loss and material/symbolic, but not of
nation. Losses (M = 5.17) produced much more anxiety than gains (M = 3.34),
F (1, 495) = 302.68, p < .001. And material scenarios (M = 4.84) produced
more anxiety than symbolic scenarios (M = 3.68), F (1, 495) = 121.88, p <
.001. The effect size of gain/loss (η p 2 = .38) was massive: about twice that of
material/symbolic (η p 2 = .20).1 The p value for nation (p = .49), however, was
not even close to statistical significance. None of the two-way interactions was
statistically significant either. The three-way interaction of nation, domain,
and frame was statistically significant, F (1, 495) = 12.66, p < .001, but not in
any obviously meaningful way, and the effect size, η p 2 = .025, was quite small.
The mean overall levels of anxiety were also very close for both the United
States (M = 4.27) and China (M = 4.18), suggesting that there was no method
effect impacting the results. At the individual level, in short, the evidence over-
whelmingly suggests that the Chinese and American respondents’ self-reports
of anxiety were not significantly different.
When we ran a second three-way ANOVA on international-level anxiety,
however, moderate national differences began to emerge. Overall, Chinese
participants (M = 3.54) reported higher levels of anxiety after reading the
international scenarios than did the American students (M = 3.03). There were
main effects of gain/loss, material/symbolic, and nation (all ps < .001), with
effect sizes of η p2 = .18, .15, and .03, respectively. All the interactions were
significant as well, although the effect sizes were small. Figure 4.1 reveals that
although the overall effect size, η p2 = .02, of the three-way domain by frame
by nation interaction, F (1, 492) = 11.78, p = .001, was on the small side, the
Chinese participants (M = 3.87) reported much more anxiety in the interna-
tional symbolic-loss condition than the U.S. participants (M = 2.43).
A three-way ANOVA on individual-level pride revealed the main effects
of gain (M = 4.74) over loss (M = 1.91), F (1, 499) = 632.08, p < .001, η p2 =
.56, symbolic (M = 3.64) over material (M = 3.01), F (1, 499) = 31.49, p <
.001, η p2 = .06, and nation, F (1, 499) = 3.95, p = .047, although the effect

1
Partial eta-square (η p2) provides a global index of the size of observed differences in means.
Small and medium effects are represented by values around .01 and .06, respectively. Large
effects are represented by values around .14 or greater.
74 Peter Hays Gries

nation = U.S. nation = China

5.00 5.00
Estimated Marginal Means of Anxiety

4.50 4.50

4.00 4.00

3.50 3.50

3.00 3.00

2.50 2.50

2.00 2.00

gain loss gain loss

GainLoss GainLoss

Material/Symbolic
material symbolic

Figure 4.1. International anxiety as a function of nation, domain, and frame, 2006
samples.

size for the latter, η p2 = .01, was very small. The only statistically signifi-
cant interaction was gain/loss and material/symbolic, F (1, 499) = 25.88, p <
.001, η p2 = .05. Both American and Chinese students reported significantly
more pride in personal symbolic gains (M = 5.39) than in material gains (M =
4.14), with symbolic losses (M = 1.94) and material losses (M = 1.88) virtually
indistinguishable.
A three-way analysis of variance (ANOVA) on international-level pride
revealed the main effects of both gain/loss, F (1, 497) = 275.61, p < .001, and
material/symbolic, F (1, 497) = 12.73, p < .001, although the effect size of the
latter, η p 2 = .03, was dwarfed by that of the former, η 2 = .36. Although there
was no main effect of nation, there was a statistically significant interaction,
F (1, 497) = 41.83, p < .001, between nation and gain/loss, with a medium
effect size, η p2 = .08. As displayed in Figure 4.2, compared to the Americans,
the Chinese reported both higher levels of pride with national gains (China M
= 5.18; U.S. M = 4.18), and lower levels of pride with national losses (China
M = 1.99; U.S. M = 2.78). Indeed, subtracting the losses scores from the gains
scores reveals that the Chinese participants (3.19 difference) were over twice
as impacted by national gains and losses as the American participants (1.4
difference).

Discussion
Are Chinese more sensitive to symbolic gains and losses than Americans, and
Americans more sensitive to material gains and losses than Chinese? The evi-
dence from our experiment is mixed, but revealing. At the individual level,
American and Chinese students were virtually indistinguishable when it came
to their self-reports of anxiety and pride in response to symbolic and material
gain and loss scenarios. For instance, both Chinese and American students
 Experimental Methods and Psychological Measures 75

5.18 (.13)
5.00
Estimated Marginal Means

4.18 (.14)
4.00
nation
U.S. China
3.00
2.78 (.15)

2.00 1.99 (.13)

gain loss
Gain / Loss
Figure 4.2. International pride as a function of nation and frame, 2006 samples.
numbers in chart = means number in parenthesis = standard deviations

took more pride in symbolic gains than in material gains. This suggests that
scholars should be wary of Orientalist and Occidentalist notions of deep-
rooted cultural differences, such as the idea that Chinese have an inordinate
cultural sensitivity to issues of “face.”
National differences did emerge, however, when we shifted from individual
to international scenarios. As Figure 4.1 reveals, Americans reported much
lower levels of anxiety in response to national symbolic losses than did the
Chinese participants. And as Figure 4.2 shows, Chinese were over twice as
sensitive to gain/loss as Americans when it came to national pride.
Two questions arise from these international-level findings. First, were the
Chinese levels of national pride and anxiety high or were the American levels
low? In other words, is this finding evidence of a Chinese oversensitivity to the
plight of their nation, an excessive concern with China’s national “face”? Or
is it evidence that Americans can more easily disassociate themselves from the
fate of their nation, or that they can kid themselves into believing that they
don’t care? Further experimental work is needed to clarify this issue.
Second, why did our nation variable produce these international-level dif-
ferences? Are they a product of the distinction between individualist and
collectivist cultures, such that Chinese have more of their psychological well-
being invested in the good of their groups? Alternatively, could these differ-
ences have historical origins, with the Chinese experience of victimization
at the hands of Western imperialism during the “Century of Humiliation”
making them more sensitive to their international status? Or are they simply
the product of the current balance of material power, such that Americans
have less to worry about or to take pride in, confident in U.S. global preemi-
nence. Chinese, by contrast, may be more anxious simply because they are
confronting the reality of an American military superpower that is ambivalent
about China’s rise. Although the use of an experimental design allows us to
76 Peter Hays Gries

confidently state that nation “caused” these differences in our international

anxiety and pride scores, further research is needed to uncover the mechanism
of causation.

Structures and Consequences of Chinese National

Identity: Psychological Measures and Methods
What is the nature of Chinese patriotism and nationalism? How do they differ
from American patriotism and nationalism? And what impact do they have on
Chinese foreign policy attitudes? To explore the structure and consequences
of Chinese national identity, two surveys were conducted in the United States
and China in spring 2009 (see Gries, Zhang, Crowson, and Cai, 2010). Using
psychological measures, exploratory factor analysis, and path analysis, we
found that although American patriotism and nationalism were empirically
similar, they were highly distinct in China, with patriotism aligning with a
benign internationalism, and nationalism with a more malign blind patrio-
tism. Chinese patriotism, furthermore, had no impact on the perception of
U.S. threats or U.S. policy preferences, whereas nationalism did. The role of
nationalist historical beliefs in structures of Chinese national identity was also
explored, as well as the consequences of historical beliefs for the perception of
U.S. threats. Selected methods, results, and discussion are presented below to
demonstrate the utility of psychological measures and methods to the study of
Chinese foreign policy.

Participants and Procedures

The surveys were completed by 512 Chinese and Americans in spring 2009,
among whom 161 Peking University undergraduate students in interna-
tional relations filled out a three-page hard-copy survey in February, and 351
American adults from around the country took an online survey in March.
Both surveys began with an explanation that the survey was about the rela-
tionship between personality and international relations, and that the data
collected would remain confidential.
The Beijing sample included slightly more women (N = 89) than men (N
= 69), and more students from the “masses” (群众) (N = 95) than CCP party
members (党员) (N = 61). Due to a clerical error, age was not requested, but
a survey of the same Peking University class the previous semester revealed
a median age of 20. Of these students, 44 claimed to have grown up in the
countryside and 114 claimed an urban upbringing.
The national American Internet sample was very well balanced, with slightly
more men (N = 177) than women (N = 174), and slightly more Democrats (N
= 130) than Republicans (N = 121) and Independents (N = 100). Ages ranged
from 18 to 69, with a mean age of 33.54 (SD = 14.20). In terms of ethnicity,
the sample was 81.2 percent white, 3.7 percent African American, 2.3 ­percent
Experimental Methods and Psychological Measures 77

non-Chinese Asian American, 2.3 percent Latino/a, 4.3 percent Native

American, and 6.3 percent “other.”

Measures
Unless otherwise noted, the questions that composed the following scales were
on seven-point Likert-type scales, ranging from 1 (“strongly disagree”) to 7
(“strongly agree”). Both samples responded to the national identity items. The
Americans were not asked the nationalist history, U.S. threat perception, and
U.S. policy preferences items.
National Identities. Twelve items were utilized to tap four distinct types of
national identities discussed in the literature on patriotism and nationalism.
Higher values on these scales indicate greater patriotism, blind patriotism,
nationalism, or internationalism. The Chinese language versions of these
national identity items are listed in the note to Table 4.1.
Patriotism. Three items adapted from Luhtanen and Crocker’s (1992)
collective self-esteem scale were used to tap the positive love of one’s own
country. They were “I’m glad to be Chinese/American,” “I often regret that I
am Chinese/American” (reverse coded), and “Being Chinese/American is an
important reflection of who I am.”
Blind patriotism. Shatz, Staub, and Levine (1999) distinguish “blind” from
“constructive” patriotism, arguing that the former represents an unquestion-
ing allegiance and intolerance of criticism. We adapted three items: “China/
America is virtually always right,” “Chinese/American foreign policies are
almost always morally correct,” and “I support my country whether its poli-
cies are right or wrong.”
Nationalism. Kosterman and Feshbach (1989) argue that nationalism goes
beyond a positive love of one’s own country (patriotism) to a belief in the
superiority of one’s own country over others. We adapted three items, “China/
America is the best country in the world,” “It is NOT important for China/
America to win international sports competitions” (reverse coded), and “In
view of Chinese history and democracy, it is only natural that China lead East
Asia/the U.S. lead the world.”
Internationalism. Kosterman and Feshbach (1989) further distinguish
nationalism from internationalism. We adapted three items: “The alleviation
of poverty in very poor countries like Haiti is their problem, not ours” (reverse
coded), “Our children should be taught to support the welfare of all of human-
ity,” and “Our foreign policies should pursue the greatest good internation-
ally, and not just pursue the Chinese/American national interest.”
Nationalist History. Beliefs about the nature of China’s past encounters
with the outside world may impact the nature of Chinese national identity
(Gries, 2004) as well as beliefs about the intentions of other countries in
the present. We therefore included four items tapping beliefs about two dis-
tinct Chinese historical encounters with the outside world: (1) the “Century
78 Peter Hays Gries

of Humiliation,” and (2) the Korean War. The two “Century” items were
taken directly from the series preface to the multivolume “Never Forget the
National Humiliation” (毋忘国耻) history book series (1992): “China’s early
modern encounter with Western imperial powers was a history of humiliation
in which the motherland was subjected to the insult of being beaten because
we were backwards” (中国近代与西方帝国主义列强的历史就是祖国蒙受奇耻
大辱落后挨打的惨痛史) and, “China’s early modern encounter with Western
imperial powers was a heroic struggle by the Chinese people against imperial-
ism” (中国近代与西方帝国主义列强的历史就是中国人民不甘屈服于帝国主义及
其附属的英雄斗争史). The two Korean War items were “China won the War
to Resist America and Aid Korea” (中国在抗美援朝中得胜了) and “The War to
Resist America and Aid Korea was a heroic moment in Chinese history” (抗美
援朝是中国历史上的英雄时刻).
Threat Perception. Perception of the threat that the United States poses to
China was tapped with four items, composed of two possible subscales. Two
addressed military threat: “A growing American military is bad for China” (美
国军队的发展对中国无益) and “The recent increase in U.S. defense spending
undermines Chinese security” (最近美国国防开支的增长威胁中国安全). Two
addressed humiliation threat: “American criticisms of Chinese ‘human rights’
are really just attempts to humiliate China” (美国政府批评中国 “人权问题”
实际上是在羞辱中国) and “American support of Taiwan and Tibet is really
about insulting the Chinese people” (美国支持台湾和西藏是在羞辱中国人民).
U.S. Policy Preferences. Three items were developed to tap respondents’
preferred U.S. policies. They were “The Chinese government should adopt
tougher foreign policies toward the U.S.” (中国政府应该对美国采取更强硬的
外交政策), “The best way to deal with the U.S. is to build up our military
and seek to contain U.S. influence throughout the world” (应对美国的最好
方式是增强我国的军备，削弱美国在世界范围的影响), and “If the U.S. threatens
China, we should use military force against them” (如果美国威胁中国, 我们
应该用军事力量对美国进行反击). Higher values indicate desires for tougher
Chinese policies toward the United States.

Results: Structures of Chinese and

American National Identity
To compare the structures of Chinese and American national identities, we first
conducted exploratory factor analysis on our American and Beijing samples.
Principal axis factoring (PAF) was conducted on both samples, followed by
Promax rotation with Kaiser normalization to aid in the interpretation of the
factors. Table 4.1 displays the results and includes all loadings greater than 0.35.
PAF on the American sample produced two factors with eigenvalues greater
than 1 (5.18 and 1.51, respectively).2 As Table 4.1 reveals, both were clearly

2
Eigenvalues represent the weight of the loadings on a factor. Generally, an eigenvalue of at least
one is seen as a necessary but not sufficient condition for viewing a factor as independent.
Experimental Methods and Psychological Measures 79

Table 4.1. Structures of national identity: Pattern matrix loadings for

­principal axis factor analysis with Promax rotation for American and
Chinese samples, 2009

American Chinese

Factor 1: Factor 2: Factor 1: Factor 2:

Patriotism / Internationalism Patriotism / Nationalism /
Nationalism Internationalism Blind
Patriotism
Patriotism 1 .908 .466
Patriotism 2r .829 .811
Patriotism 3 .797 .696
Blind Patriotism 1 .405 .351 .546
Blind Patriotism 2 .433 .388 .669
Blind Patriotism 3 .660 .614
Nationalism 1 .455 .782
Nationalism 2r .390
Nationalism 3 .780 .601
Internationalism 1r –.612 .639
Internationalism 2 –.725 .500
Internationalism 3 –.553
Eigenvalues 5.18 1.51 3.26 2.28
Factor intercorrelation .558 .269
Scale α (N) .88 (9) .66 (3) .75 (5) .78 (5)
Mean (SD) 4.43 (1.16) 5.27 (1.11) 5.63 (1.08) 3.92 (1.17)
Note: Factor coefficients are shown only if > 0.35. Reverse coded items denoted with an “r” and italicized.
Differences in American/Chinese versions are underlined:
Patriotism
1. “I’m glad to be American/Chinese.” 我很高兴自己是中国人 。
2r. “I often regret that I am American/Chinese.”我经常遗憾自己是中国人 。
3. “Being American/Chinese is an important reflection of who I am.” 我觉得作为中国人对我的自我认同很
重要.
Blind Patriotism
1.“American/Chinese foreign policies are almost always morally correct.” 中国的外交政策基本上都是正义
的。
2.“America/China is virtually always right.” 中国的决策几乎都是正确的 。
3. “I support my country whether its policies are right or wrong.” 无论我国的政策对错与否, 我都予以支
持。
Nationalism
1.“America/China is the best country in the world” 中国是世界上最好的国家 。
2r.“It is NOT important for America/China to win international sporting competitions.” 中国赢得国际体
育竞赛并不重要 。
3.“In view of America’s/China’s history and democracy, it is only natural that the U.S. lead the world/China
lead East Asia.” 鉴于中国具有悠久的历史,光辉的文明,中国的然应该领导东亚 。
Internationalism
1r. “The alleviation of poverty in very poor countries like Haiti is their problem, not ours.” 诸如海地这样的
贫穷国家所面临的问题应该由他们自己解决,与我们无关 。
2. “Our children should be taught to support the welfare of all of humanity.” 我们应该教育我们的子孙后代不
仅为中国而为全人类的福祉做贡献 。
3. “Our foreign policies should pursue the greatest good internationally, and not just pursue the American/
Chinese national interest.” 我国外交政策应当追求国际主义而不是只追求中国的国家利益 。
80 Peter Hays Gries

interpretable. All nine patriotism, blind patriotism, and nationalism items

loaded most strongly on the first factor, which has been labeled “patriotism/
nationalism.” The three internationalism items loaded on the second factor,
labeled “internationalism.” The two factors intercorrelated quite highly at r =
.56 and the internationalism items all loaded negatively. Along with our second
eigenvalue of just 1.51, this suggests that our two factor solution was close to
being a single factor solution. In short, although patriotism, blind patriotism,
and nationalism are conceptually distinct, our data suggest that empirically
American patriotism and nationalism go together and even approach being
part of a single dimension set against internationalism.
PAF on the Beijing sample also produced two factors with eigenvalues
greater than 1 (3.26 and 2.28, respectively). Table 4.1 reveals that both were
clearly interpretable with no cross loadings. All three patriotism and the
first two internationalism items loaded on factor one, labeled “patriotism/­
internationalism.” All three blind patriotism and two of the nationalism items
loaded on factor two, labeled “nationalism/blind patriotism.” The two factors
intercorrelated at just r = .27, indicating that these two dimensions of Chinese
national identity are largely orthogonal or independent of one another.
The differing structures of American and Chinese national identities
revealed in Table 4.1 are truly striking. Where patriotism – love of country –
and nationalism – belief in the superiority of one’s country over other coun-
tries – go together in the American sample, they do not go together in the
Chinese sample. Instead, patriotism in China is associated with international-
ism and should thus be understood to be more benign than American patrio-
tism. In other words, the more patriotic an American is, the more nationalistic
she or he also tends to be. In China, however, patriotism and nationalism
do not necessarily go together, such that a highly patriotic Chinese is just as
likely to be low on nationalism, and a very nationalistic Chinese to be low on
patriotism.

Results: Consequences of Chinese Patriotism

and Nationalism
To explore the consequences of Chinese patriotism and nationalism we first
constructed scales for each, as well as for the history, threat, and policy vari-
ables. The scale reliabilities and Ns are reported in the last two columns of
Table 4.2, which reports the descriptive statistics for the Beijing sample. The
Cronbach’s alphas ranged from fair (α = .71) to good (α = .82) internal reliabili-
ties, giving us confidence that the specific survey items composing each scale
tapped the same underlying construct.3

3
Cronbach’s alphas range from zero to one, with higher scores indicating greater internal reli-
ability of the measure; .60 is generally seen as the minimal acceptable alpha. Alphas tend to
be higher for longer scales, so our alphas of .71 and .73 for scales of just two and three items
might actually be interpreted as quite good.
Experimental Methods and Psychological Measures 81

Table 4.2. Descriptive statistics: Correlations, significance levels, means,

­standard deviations, and scale alphas and Ns for 2009 Beijing sample
(minimum N = 156)

Variables 1 2 3 4 5 6 M SD α N
1. Patriotism/ – .19* .38** .20* .29** .13 5.63 1.08 .75 5
Internationalism
2. Nationalism/ – .44** .32** .38** .35** 3.92 1.17 .78 5
Blind Patriotism
3. Nationalist – .39** .45** .38** 4.71 1.15 .75 4
History
4. Military Threat – .34** .32** 4.29 1.21 .71 2
5. Humiliation – .44** 4.31 1.51 .82 2
Threat
6. U.S. Policy – 3.81 1.26 .73 3
Note:
** Correlation is significant at the .01 level (2-tailed).
* Correlation is significant at the .05 level (2-tailed).

The means and standard deviations for all six of our scales are also
listed in Table 4.2. The Beijing sample reported much more patriotism/
internationalism (M = 5.63) than nationalism/blind patriotism (M = 3.92).4
Given a scale midpoint of 4, we can say that, overall, the Beijing sample was
very patriotic but quite balanced in terms of nationalism. The means for
nationalist historical beliefs (M = 4.71), military threat (M = 4.29), and humili-
ation threat (M = 4.31) were just above the scale midpoint of four, whereas
U.S. policy preferences were just below the scale midpoint (M = 3.81), suggest-
ing a good balance on all of our scales.
Finally, Table 4.2 also reports the zero-order correlations among our six
scales. With the exception of the lack of a relationship between patriotism and
U.S. policy, all of the correlations were statistically significant and positive.
And with the exception of the relationships between patriotism and national-
ism, and patriotism and military threat, which correlated at just p < .05, the
remaining correlations were highly significant (p < .01) and substantial in size,
ranging from r = .29 to r = .44.
Given that the zero-order correlations do not account for collinearity, we
decided to use path analysis to better understand the precise relationships
among our variables. Path analysis has a number of advantages over mul-
tiple regression, such as the ability to model mediated relationships among
variables, as well as the ability to evaluate the global fit of a model contain-
ing those mediated relationships. We used AMOS 17.0 with full information
4
An independent sample t-test revealed the difference between the means to be both statisti-
cally significant and very large, t (160) = 15.14, p < .001.
82 Peter Hays Gries

R 2 = .18 d2

U.S. military
threat .16*
patriotism / .31***
internationalism .18*
.18*
.37***
R 2 = .25

nationalist tougher
.18*
history .17** U.S. policy

.44*** .35***
.31***
R 2 = .25 d1
nationalism /
blind patriotism
U.S. humiliation
.23** threat
d3

Figure 4.3. Final Beijing path model, 2009 sample.

*** Correlation is significant at p < .001; ** Correlation is significant at p < .01;
* Correlation is significant at p < .05

maximum likelihood estimation to first test a fully saturated model in which

patriotism, nationalism, and nationalist historical beliefs were treated as cova-
rying exogenous variables predicting U.S. military and humiliation threat,
which in turn predicted policy preferences. After removing the statistically
insignificant paths, the model displayed in Figure 4.3 emerged as the best fit
for the Beijing data.
We examined the fit of our path model based on the χ 2 test, the χ 2 /degrees
of freedom ratio, the Comparative Fit Index (CFI), the Tucker-Lewis Index
(TLI), the Normed Fit Index (NFI), and the Root Mean Square Error of
Approximation (RMSEA). Nonsignificant χ 2 values and χ 2 /df ratios < 2 or
3 are considered reasonable indicators of a close model fit. Conventional cut-
offs for a close model fit are CFI, TLI, and NFI values greater than .95 and
RMSEA values less than .06 (see Kline, 2005; Schumacker and Lomax, 2004).
Our final model in Figure 4.3 was a very good fit for the Beijing data, with a
nonsignificant χ 2 value of p = .338, a χ 2 /df ratio of 1.135, a CFI of .997, a TLI
of .983, a NFI of .976, and an RMSEA of .029.
The most striking aspect of our Beijing path model is that when control-
ling for nationalism and nationalist history, patriotism had no impact on
perceptions of U.S. military or humiliation threat or U.S. policy preferences.
Nationalism, by contrast, had a strong impact on U.S. policy preferences, both
directly and mediated through perceptions of U.S. military and humiliation
threats. Indeed, these three paths combined to account for a full 25 percent
of the variance in U.S. policy preferences.5 From a foreign policy perspective,
5
The “d1” in the path model represents prediction error, or the 75 percent of the variance in
preferred U.S. policies not accounted for by the variables included in the model.
Experimental Methods and Psychological Measures 83

therefore, Chinese patriotism appears decidedly benign, whereas Chinese

nationalism appears potentially malign in its consequences.
It is also noteworthy that nationalist historical beliefs covaried strongly
with both patriotism (r = .37) and nationalism (r = .44), providing strong sup-
port for the argument that beliefs about the national past and national identi-
ties in the present are mutually constituted. Nationalist historical beliefs also
strongly predicted both perceptions of U.S. military (r = .31) and humiliation
(r = .35) threat.
Finally, Figure 4.3 reveals that perceptions of a humiliation threat had a
much greater impact on U.S. policy preferences than did perceptions of a mili-
tary threat. Indeed, squaring the partial coefficients reveals that whereas per-
ceptions of a U.S. humiliation threat accounted for a full 9.6 percent of the
variation in U.S. policy preferences, perceptions of a military threat accounted
for just 3.2 percent of that variation. Those interested in the determinants of
China’s U.S. policy, therefore, would be wise to consider not just the objective
balance of military power but also the subjective realm of identity and affect.

Discussion
Allen Carlson (2009) has lamented the lack of rigorous measurement in stud-
ies of Chinese national identity and the failure of scholars to place Chinese
nationalism in a broader comparative framework. I agree. This case study
should demonstrate that the rigorous measurement of constructs like patrio-
tism and nationalism is not only possible but that such constructs can be part
of an explanatory social science.
Neil Diamant (2009: 18–23; this volume, Chapter 2) has argued that there
is a “threshold problem” in studies of popular nationalism such as my own
(Gries, 2004): the patriotism of the self-styled “fourth generation” of urban
Chinese does not entail sufficient sacrifice or commitment to rise to the level
of true patriotism. Compared to the veterans he studies who have genuine
“patriotic standing,” Diamant dismisses the urban youth who have been at the
forefront of the last decade of popular nationalist protests as inconsequential
“café latte” nationalists. Those like myself who have studied the rise of this
popular nationalism, furthermore, foster “China threat” discourse.
Diamant’s “threshold” approach to patriotism/nationalism is both con-
ceptually and empirically problematic. Conceptually, where should one draw
the line? What level of “sacrifice” is sufficient to be included in his “patriot”
category? (Or what level of latte drinking is sufficient to be dismissed as an
unpatriotic “elite”?) Empirically, reducing concepts like patriotism or nation-
alism to an either/or binary does violence to the complexity and variability
of the concepts. For patriotism or nationalism to be useful in a social science
that seeks to be explanatory, we should seek to maximize rather than mini-
mize the variation that is empirically measured. Variables should vary – as
much as possible. For instance, without the variability of each of our survey
items, whose responses were on 1–7 Likert-type scales, and without five items
84 Peter Hays Gries

tapping both patriotism and nationalism, increasing each scale’s internal reli-
ability, it is unlikely that we would have been able to empirically distinguish
between Chinese patriotism and nationalism or to uncover the unique conse-
quences of each.
Diamant is also misguided to dismiss the nationalism expressed by China’s
young netizens (网民) and street demonstrators as inconsequential. This study
has shown that individual differences in “trait” or enduring levels of nation-
alism impact both perceptions of U.S. threat and preferred U.S. policies. It
is likely that temporary or “state” levels of nationalism have similar conse-
quences. Thus when incidents like the 1999 Belgrade bombing or the 2001
Hainan Island plane collision temporarily inflame anti-American national-
ist sentiments, Chinese perceptions of U.S. threat likely increase, along with
Chinese desires for tougher U.S. policies. During such crises, therefore, inflamed
Chinese nationalism could have very serious consequences for Chinese foreign
policy, even if temporarily inflamed levels of nationalism dissipate later, as
Diamant rightly notes. In short, variations in Chinese nationalism, whether
between individuals or across time, appear to be related to variations in both
threat perception and even foreign policy preferences, thus warranting further
research rather than dismissal.
Diamant’s most serious charge is that studies of popular Chinese national-
ism, like my own, foster “China threat” discourse. His logic is one of guilt
through association: “China threat” proponents frequently refer to the rise of
Chinese nationalism to support their arguments; therefore, those who study
Chinese nationalism are complicit in the “China threat” project. This logic is
problematic: once scholarship has been published, no scholar can completely
control its use or misuse. Diamant’s charge could also have a chilling effect
on scholars, who too often cede the public sphere on sensitive political topics
to nonexperts. I would argue that we need to encourage more, not less, aca-
demic work and policy outreach on highly consequential topics like Chinese
nationalism.

Conclusion: Opportunities and Challenges

In the conclusion to their edited volume New Directions in the Study of
China’s Foreign Policy, Thomas J. Christensen, Alastair Iain Johnston, and
Robert S. Ross (2006: 387) rightly note that Chinese foreign policy studies
have been a “consumer but not a producer of theory and methods.” I hope this
chapter has made the case that experimental methods and psychological mea-
sures offer great promise not just for deepening our substantive knowledge of
Chinese foreign policy but also for contributing to the theoretic development
of the field of foreign policy analysis. Just as cross-cultural psychology was
well positioned at the margins to challenge and reshape the universalism of
much of mainstream social and cognitive psychology (see Nisbett [2003] for
an overview), China scholars are well positioned to challenge and reshape
theories inductively derived from the Western experience that do not travel
well to China.
Experimental Methods and Psychological Measures 85

Experimental methods, furthermore, are theoretically neutral. Whether

you are a structural realist focusing on the balance of material power like Bob
Ross, or a constructivist focusing on sociological theory like Iain Johnston,
you can benefit from the rigorous causal explanatory power that experiments
bring to empirical work.
The same cannot be said for psychological measures. If you agree with
rationalists like Bob Ross (2001: 395) that “common arguments about mis-
perceptions in policymaking … do not apply to the U.S.-China conflict,” you
will have no need for psychological measures. If, on the other hand, you agree
with Robert Jervis (1976) that perception and misperception are central to the
relations among nations, psychological measures may well be indispensable
for those who wish to move beyond theory to the empirical examination of
Chinese foreign policy.
Psychological measures developed in the West, however, do not always
travel well to China. Many core concepts in the psychological literature,
such as conservatism, are so bound up in the Western liberal tradition that
they simply cannot travel to China. For instance, the widely used right-wing
authoritarianism (RWA) scale has idiosyncratic items like “Atheists and oth-
ers who have rebelled against the established religions are no doubt every
bit as good and virtuous as those who attend church regularly” that clearly
would not resonate with Chinese subjects.
Another problem with using Western psychological measures in China is
that the scales tend to be one dimensional, requiring that negative items be
reverse-coded. Cross-cultural psychologists have found that due to a greater
tendency toward dialectical rather than categorical thinking, Asians are more
likely than Westerners to simultaneously hold contradictory attitudes (e.g.,
Spencer-Rodgers, Peng, and Wang, 2004). Forcing their responses onto unidi-
mensional scales, therefore, is likely to reduce their reliability. For example, a
Chinese respondent to a stereotyping or prejudice scale might simultaneously
rate Americans as both highly “friendly” (a positive attribute) and highly “dis-
agreeable” (a negative quality), thus reducing the reliability of a unidimen-
sional scale. Greater multidimensional scaling in such a situation would be
needed. Similarly, in our second case study, separate items tapping beliefs that
China was a victim and beliefs that China was a victor during the “Century
of Humiliation” cohered, despite their apparent contradiction. Such empiri-
cal findings point to the importance of a fundamentally inductive orientation
toward the data gathered from Chinese surveys.
Another challenge is that psychological constructs like threat perception are
notoriously difficult to measure, especially when using hard-copy or Internet
surveys that rely on self-reporting. People are not always honest, even with
themselves, about their actual views and emotions. Through “impression man-
agement” or “self-presentation” techniques, we often seek to orchestrate the
images we present to ourselves and the world (Goffman, 1959). With sufficient
funding, future research could also use physiological techniques, such as mea-
suring blood pressure and galvanic skin conductance as indicators of anxiety
and threat perception. A more affordable approach is sentence unscrambling
86 Peter Hays Gries

tasks and other methods that can reveal subconscious or implicit levels of var-
ious attitudes and affect.
At a broader level, all statistical studies confront two challenges: junk-in-
junk-out and gold-in-junk-out. First, regardless of the rigor of the statistical
analysis, if the original data are poor, the results will be of limited value. This
problem has been well documented in the case of the famous “correlates of
war” database once widely used in conflict studies. The coding of even basic
issues such as distinguishing between interstate and civil wars or the exact
onset or termination dates of hostilities was found to be highly problematic.
In the case of the type of political psychology conducted here, good internal
reliabilities are vital to the discovery of robust patterns of associations among
our variables. For instance, had the internal reliabilities of our measures been
lower, the fit statistics for our path model would not have been good enough
to give us confidence in the pattern of associations depicted in the model.
Second, even excellent survey data, if not properly interpreted, yield little of
value. Statistics do not speak for themselves. One problem is that political sci-
entists overemphasize statistical significance testing (p values) at the expense
of effect sizes (η p2) and the interpretation of the practical significance and
meaning of their statistical results. The focus on statistical significance testing
is particularly problematic when political scientists use preexisting datasets,
such as the American National Election Survey or the Chicago Council on
Global Affairs surveys that generally have very large Ns. With such datasets,
it is rare for variables to be completely uncorrelated; p values are a direct
product of sample size. Another problem with statistical significance testing
is the systematic failure to report statistically insignificant findings. Empirical
falsification, as Karl Popper (2001) has noted, is vital to the accumulation of
knowledge. Statistical non-findings can also prove highly instructive, such as
the finding reported in the first case study above that American and Chinese
students responded similarly to the individual-level scenarios of symbolic and
material gains and losses, allowing us to question Orientalist notions of a
uniquely Chinese obsession with “face.”
Much statistical work also suffers from a lack of attention to the interpreta-
tion of the meaning of statistical results. This is likely due to the limits of corre-
lational data and the challenges of interpretation. Correlational designs cannot
yield causal explanations. Even our path model presented in Figure 4.3, although
an excellent fit to the data, did not prove a causal relationship. It is always pos-
sible that there are other configurations of paths that would fit the data equally
well or better. Interpretation can also be challenging even with data resulting
from experimental designs. For instance, whereas we can feel confident that
nation (American vs. Chinese) was the cause of the differing levels of anxiety
displayed in Figure 4.2, the precise mechanism of causation remains unclear.
Our interpretation of the results, therefore, must remain tentative. In nascent
fields like the political psychology of international relations and Chinese foreign
policy studies where so little is yet known, translating even rigorous experi­
mental findings into coherent causal arguments remains a daunting task.
Experimental Methods and Psychological Measures 87

The list of challenges, frankly, is humbling. It is not easy to do this type of

research. Mastering psychological and international relations theories, exper-
imental and survey designs, psychometrics, and statistical analysis are just
the beginning. Another challenge surrounds data collection. Finding partners
in China is the critical first step. It is vital to work with groups of people
who trust one another and who share common interests and goals. I prefer
the coauthoring and sharing of data to purchasing data, although that is not
always possible if one seeks a nationally representative sample. Often more
challenging are human subjects review boards at U.S. universities – these are
difficult to begin with and frequently have misperceptions about China that
hinder project approval. Once approval is secured, the procedural details are
numerous but manageable. Hard-copy questionnaires must be stored in secure
areas, devoid of any identifying information. Electronic data files must be
password-protected on computers to ensure participant confidentiality.
A final challenge for political psychologists is that peer review in political
science journals will always be challenging because of an arbitrary prefer-
ence among most reviewers for the external rather than the internal validity
of research designs. This is likely due to the dominance of American politics
and Americanists’ focus on voting behavior, where external validity is obvi-
ously of paramount importance. Furthermore, most political scientists have
a myopic view of external validity: all they can understand is random sam-
pling. Replication is a fundamental principle of the scientific method. Indeed,
the entire discipline of psychology is built on the cross-validation of research
using independent samples. And yet I have experienced reviewers who refuse
to accept results replicated across four or more independent samples. They
were trained in random sampling and are dismissive of other approaches to
external validity. Clearly scholars need to be careful about generalizing from
nonrandom samples, but there are times when the advantages of measurement
and internal reliability outweigh the costs to generalization (e.g., Nicholson-
Crotty and Meier, 2002). Research design should follow from research goals,
not dogma.
In my view, such challenges to using experimental methods and psychologi-
cal measures must be overcome if we are to better understand the determi-
nants of security and insecurity in U.S.-China relations – and avoid another
bilateral conflict. China and the United States fought twice in the latter half
of the twentieth century, and the United States could easily be dragged into
another conflict involving either Taiwan or China-Japan relations at the onset
of the twenty-first century. As Figure 4.3 makes clear, psychological concepts
like humiliation have a powerful impact on the foreign policy preferences of
individual Chinese. Rationalist approaches to IR, such as neorealism and
neoliberalism, therefore, must be supplemented by approaches that take into
account the intersection of politics and psychology. The stakes are simply too
high to cover our eyes and simply hope for the best.
5

Internet Resources and the Study of Chinese

Foreign Relations
Can Cyberspace Shed New Light on China’s
Approach to the World?

Allen Carlson and Hong Duan

For decades, students of Chinese foreign policy were confronted with a

stark dearth of information relating to China’s position in the international
arena. In the 1960s, Allen Whiting’s seminal work, China Crosses the Yalu,
while drawing from the author’s extensive governmental experience, refer-
enced only a limited pool of official Chinese sources to describe Beijing’s
stance on the Korean peninsula. In the 1970s, the main source of information
for Samuel Kim’s book, China, the United Nations and World Order, was
Chinese votes in the United Nations General Assembly. A few years later,
A. Doak Barnett’s short, but influential, The Making of Foreign Policy in
China, was also marked by the use of a limited number of sources. In con-
trast, in the late 1980s, a new generation of scholars began to gain access
to a somewhat broader set of data. The best examples of this trend were
Tom Christensen’s consideration of newly available documents relating to
the Korean War, followed by Iain Johnston, David Shambaugh, and Robert
Ross’s utilization of extensive interview data. However, this being the case,
the general informational frame for researching Chinese foreign policy has
remained relatively static since the early 1990s, with researchers repeatedly
making use of the same limited set of sources (a handful of Chinese-language
journals dedicated to international politics,1 official statements, a smattering
of 内部 documents, and interviews with a small circle of foreign policy elites).
In comparison, it is widely perceived that the study of Chinese foreign policy
is now poised to enter a new era. Indications of such an incipient development
are purported to be found in the expanding availability of new data. The most
prominent of these sources is generally seen to be the Chinese Internet, which
appears to contain a treasure trove of new information related to China’s

The research and writing of this paper was generously supported by Cornell University’s LaFeber
Fellowship, a fund that supports collaborative research between Government Department fac-
ulty and graduate students. Ben Brake, a Cornell graduate student, provided additional research
assistance for this chapter.
1
The most important of these are 国际问题研究,世界经济与政治，现代国际关系.

88
The Internet and the Study of Chinese Foreign Relations 89

foreign relations. The question then becomes how significant is the Internet
to the study of Chinese foreign policy?
In taking up this issue, this chapter consciously seeks to reorient focus on the
Chinese Internet away from its much discussed potential to contribute to soci-
etal transformation in China and issues of censorship and freedom of speech.
Instead, the chapter concentrates on the more prosaic, yet, for researchers,
more elemental issue of whether this space contains new information that can
contribute to our understanding of China’s foreign relations.2 On this score it
contends that the returns to date are mixed. On the one hand, as Part One of
the chapter will show, the main sites in China dedicated to foreign policy issues
contain less new data than many may suppose to be the case. Indeed, many
of the official statements and elite analysis that have been posted in Chinese
cyberspace are not original content but rather reproductions of work that has
been previously published in more conventional formats. On the other hand,
as discussed in Part Two of the chapter, a number of incremental advances
in the collection and analysis of data related to China’s foreign affairs can be
made within this new space. Thus, the conclusion contends that Internet mate-
rials can be used to supplement conventional resources, but they do not appear
to be in a position to supplant the interviews, journal articles, archives, and
news sources that have come to form the core of the field of Chinese foreign
policy studies over the last two decades.

Part One: Mapping China’s Foreign

Relations–Related Cyberspace
This section describes the main contours of Chinese foreign policy cyberspace.
It defines this terrain as including those sites that consistently contain a signifi-
cant amount of content related to China’s foreign relations. Such an exercise is
valuable simply in terms of providing an initial guide to those with an interest
in traveling more extensively within this area.3 Along these lines, the carto-
graphic presentation that follows is divided into four parts: a consideration of
the main media and news sources that cover Chinese foreign policy and inter-
national affairs, a survey of the Web pages maintained by China’s Ministry of

2
For extensive consideration of the societal impact of the Internet on Chinese politics see
Taubman (1998), Hartford (2000), Chase and Mulvenon (2002), Yang (2003), Shie (2004),
Zhou (2005a), Tsui (2005), Kluver and Yang (2005), Saunders and Ding (2006), Damm and
Thomas (2006), Tai (2006), Zheng (2008), and McKinnon (2008). For a skeptical discussion
of the broader argument that cyber activity can have real-world impact, see DiMaggio et al.
(2001), Thelwall and Smith (2002), and Langman (2005).
3
This being said, a limited number of previous efforts have been made to create similar maps;
among the best of these in the security field are Fravel (2000) and Zhou (2005b). More
broadly, Rebecca McKinnon has led a series of groundbreaking workshops that explore vari-
ous emerging issues in Chinese cyberspace. The most recent such workshop, “China and the
Internet: Myths and Realities,” was held in 2008 in Hong Kong. Archives of the workshop are
available at http://jmsc.hku.hk/blogs/circ.
90 Allen Carlson and Hong Duan

Foreign Affairs (MFA) and top research institutes and universities, a review of
the most important academic sites which exist primarily in cyberspace (rather
than as Internet-based affiliates with real world institutions) and are dedicated
solely to foreign policy issues, and a survey of Chinese Bulletin Boards (BBs)
and blogs that focus on international politics.

Chinese Cybermedia Outlets and Their

Coverage of International Affairs
Over the last decade the most prominent and rapidly expanding aspect of
Chinese cyberspace related to foreign policy issues has been the emergence of
online media sources. This trend was led by Xinhua and People’s Daily, each
of which established major Chinese news portals. However, alongside such
sources, new comprehensive Web portals such as Sina Net, Sohu Net, and
Netease have also opened news channels and started providing news online.4
As a result, and following developments elsewhere in the world, the Internet
in China has increasingly become an important source of both domestic and
international news. The extent of this development is indicated by a recent sur-
vey conducted by the China Internet Network Information Center (CNNIC).
This report found that by the end of 2007 the total number of Chinese Internet
users had reached 210 million, and that reading news online was the fourth
most frequent type of activity among Chinese Internet users (CNNIC, 2008).
Indeed, 73.6 percent of Internet users who responded to the survey claimed
they read online news; moreover, 20 percent said that the first thing they do
on the Internet is to read online news (CNNIC, 2008: 16). In other words,
the new cyber media has become a crucial conduit for the dissemination of
information related to China’s foreign affairs. Four main sites are of specific
interest.

1. People’s Net ( 人民网 ) http://www.people.com.cn

People’s Net is the most visible of the main media sites and is run by the offi-
cial newspaper of the Chinese Communist Party, the People’s Daily. The site
is organized into a number of issue specific sectors. For example, it contains
channels dedicated to news about the party,5 the government,6 the National
People’s Congress,7 and the Chinese People’s Political Consultative Conference.8
In addition to the regular news channel, more detailed and categorized news
is also available on channels with titles such as politics, world, military,
and Taiwan. Moreover, full-text versions of the print edition of most of the

4
For Sina Net visit http://www.sina.com.cn/; Sohu Net’s site is http://www.sohu.com/; Netease
is found at: http://www.163.com/ .
5
See http://cpc.people.com.cn/GB/index.html.
6
See http://gov.people.com.cn/GB/index.html
7
See http://npc.people.com.cn/GB/index.html.
8
See http://cppcc.people.com.cn/.
The Internet and the Study of Chinese Foreign Relations 91

newspapers and magazines published by the People’s Daily Press, including

人民日报 (domestic and overseas editions), are also accessible via this site and
archived back to 2000.9
Beyond these general channels, People’s Net also contains an impressive
array of sites dedicated primarily to issues related to international politics.
For example, it maintains a channel that catalogs all the speeches and talks of
foreign ministry spokespersons dating back to 1997. In addition, the site also
repeatedly invites Chinese foreign policy experts to discuss topical issues and
to conduct online dialogue with Internet users. The list of scholars who have
participated in such forums is constantly expanding and includes some of the
most influential students of foreign policy within China. Among those who
have been involved are Wang Jisi, Yan Xuetong, Jin Canrong, Shi Yinhong,
and Liu Jiangyong.10

2. Xinhua Net ( 新华网 ) http://www.xinhuanet.com

Xinhua Net is run by the government-owned Xinhua News Agency, and, in
many respects, it is similar to People’s Net. It also contains a wide variety of
dedicated news channels. For example, in its “Data Channel” there is informa-
tion about the structure of the Chinese government, the CCP, nonruling par-
ties, civil associations, China’s economic development, legal system, human
rights, education, national defense, and diplomacy.11 The site also offers full
texts of the print edition of the newspapers and magazines that belong to
the Xinhua News Agency, including Xinhua Daily Telegraph. Moreover, like
People’s Net, Xinhua Net’s world channel invites government officials and
scholars to talk with Internet users about particularly controversial issues relat-
ing to China’s foreign relations. The transcripts of these online talks are then
posted on the site in a channel entitled “International Interview.”12 Beyond
such content, Xinhua Net also places a relatively prominent emphasis on links
to other sites. For example, its homepage highlights links to each of the other
major Chinese news portals and to thirty-five regional news portals.13

3. China Net ( 中国网 ) http://www.china.com.cn/

China Net differs from the other sites discussed in this section in that although
it is a news portal, it is directly sponsored by the State Council’s Information
Office. China Net publishes both world news and, perhaps of greater inter-
est, news about Chinese foreign policy. Along these lines, a main source of

9
 Until recently, the Global Times, a popular weekly newspaper with a focus on international
issues and published by the People’s Daily Press, was available as well. However, this publica-
tion launched its own Web site in September 2007 and has become a separate news portal
(http://www.huanqiu.com/ ). As such, it is no longer accessible via People’s Net.
10
For a list of the topics and transcripts, see http://world.people.com.cn/GB/8212/115071/
index.html.
11
See http://news.xinhuanet.com/ziliao/2004–11/04/content_2177717.htm.
12
See http://www.xinhuanet.com/world/gjft.htm.
13
See http://www.xinhuanet.com/dfwl.htm.
92 Allen Carlson and Hong Duan

content on this site is reports about Chinese diplomacy, exchanges with other
countries, foreign trade, and cross-Strait relations. As abundant as such cov-
erage is, it is necessary to note that the site’s Chinese-language version con-
tains significantly more information than its English-language mirror site. For
example, its 国情, or “National Conditions,” channel contains an encyclo-
pedic directory of other Chinese-language sites and archives related to both
national defense and diplomacy but lacks an English-language companion of
comparable scope.14 The same holds true for sections of the site that pro-
vide “statistical reports” (统计公报),15 annual reports on global politics, and
reports on national security.16

4. China Military Online ( 中国军网 ) http://www.chinamil.com.cn/

China Military Online is run by 解放军报 and acts as a conduit for the cyber
publication of the People’s Liberation Army’s (PLA’s) own position on both
general issues in world politics and specific military concerns. The site has
both an English- and Chinese-language homepage (as do most other Chinese
news sites). As in the case of China Net, on Chinese Military Online there is
a noticeable difference in the content of these two pages. The English site is
clearly organized into a number of distinct channels, including the follow-
ing topics: national defense, army building, political works, logistics, mili-
tary diplomacy, armaments, information technology, science and technology,
military training, disaster relief, frontier and coastal defense, and the history
of the PLA. The Chinese site is basically organized in the same fashion but
contains significantly more information. For example, it offers full texts of
the print edition of 解放军报 and the other newspapers and magazines affili-
ated with the PLA. This list is expansive and includes publications that range
from the authoritative, such as PLA Daily and China’s National Defense, to
both popular and technical publications like China’s Militias, PLA Pictorial,
Military Correspondents, and Global Military Affairs. In most cases, back
issues of these papers have been archived online dating to 2004. In addition
to being a valuable warehouse for the Chinese military press, the site also
contains an extensive survey of the structure and organization of the Chinese
military. Along these lines, it includes a detailed database introducing China’s
current military leadership, its military regulations, armaments, and other
information.17

Web sites of the Ministry of Foreign Affairs, Universities,

and Research Institutes
Over the last decade China’s Ministry of Foreign Affairs (MFA) and each
of the main foreign policy research institutes and universities with major

14
See http://www.china.com.cn/aboutchina/node_6175014.htm.
15
See http://www.china.com.cn/economic/zhuanti/06gongbao/node_7014958.htm.
16
See http://www.china.com.cn/zhuanti2005/node_6087279.htm.
17
See http://www.chinamil.com.cn/site1/database/index.htm.
The Internet and the Study of Chinese Foreign Relations 93

international studies programs have launched their own Web sites. Although
these sites were initially quite skeletal and still vary quite significantly in terms
of content, they have expanded in recent years and have become a durable
facet of China’s foreign policy-related Internet. Five of these sites, along with
a cluster of university Web pages, merit specific mention.

1. Ministry of Foreign Affairs ( 中华人民共和国外交部 )

http://www.mfa.gov.cn/
The Chinese-language version of this site contains rich information regard-
ing China’s diplomacy.18 The “diplomatic developments” channel consists of
a comprehensive and current news archive on high-level diplomatic activities
and MFA press briefings.19 The “resources” channel is composed of foreign
policy speeches by Chinese leaders and high-ranking officials, Chinese for-
eign policy communiqués, and a list of treaties.20 In addition, the site also
posts detailed information about the MFA itself. The “ministry” channel lists
each ministerial-level official’s name, biography, and the issue areas within
his or her portfolio.21 This channel also offers information about the foreign
ministry’s organization. Not only are all the general departments under the
ministry listed, but each department’s contact information and the names of
its directors are also provided. It is also of note that the site is well connected.
It lists links to Chinese missions overseas, other central government organiza-
tions, local governments and local foreign affairs offices, and major Chinese
Web news portals.

2. Foreign Policy and International Affairs Research

Institutes of the Chinese Academy of Social Sciences
As China’s “highest academic research organization in the fields of philoso-
phy and social sciences,” the Chinese Academy of Social Sciences (CASS) is
a vast organization that employs almost 3,000 researchers and encompasses
31 research institutes and over 50 research centers (CASS, 2009). Although
the majority of academic activity and policy analysis that takes place at CASS
falls outside the scope of this chapter’s focus, a number of China’s most prom-
inent foreign policy and international relations research institutes are located
here. Whereas in the 1990s learning about these centers and gaining access to

18
The MFA site also has versions in English, Russian, French, Arabic, and Spanish. The English
version (http://www.fmprc.gov.cn/eng/ ) is somewhat different from the Chinese version. To
begin with, the Chinese version tends to be more current. Moreover, the “countries” channel
is organized somewhat differently on the two mirror sites. Whereas the Chinese version of
this channel (http://www.mfa.gov.cn/chn/pds/gjhdq/gj/ ) offers mainly political and economic
profiles of individual countries and regions, the English version (http://www.fmprc.gov.cn/
eng/gjhdq/ ) focuses on their bilateral relations and bilateral exchanges with China, which is
probably more useful for students of China’s foreign policy.
19
See http://www.mfa.gov.cn/chn/pds/wjdt/.
20
See http://www.mfa.gov.cn/chn/pds/ziliao/.
21
See http://www.mfa.gov.cn/chn/pds/wjb/.
94 Allen Carlson and Hong Duan

researchers working in them required repeated phone calls, letter writing, and
personal visits, today all of the main CASS institutes maintain Web sites that
contain detailed data on the organizational structure as well as the contact
addresses of their hosts. However, beyond such information, these sites are gen-
erally quite sparse. For example, one of the most significant of the CASS sites
is maintained by the Institute of World Economics and Politics (IWEP) which
introduces the institute’s history, its organization, and its leading researchers
(including resumes, e-mail addresses, and a list of major publications).22 Yet,
beyond these data, the few articles and papers posted on the page have been
previously published. The Institute of American Studies page is organized in
a similar fashion and contains only a handful of articles published by the
institute’s researchers.23 However, the contact information on the site is not as
comprehensive as that which appears on the IWEP Web page. The Institute
of Asia-Pacific Studies site has a complete list of information regarding its
researchers, but beyond such data, this site is also quite skeletal.24 Of even less
substance is the site maintained by the Center for China’s Borderland History
and Geography Research, which fails to provide detailed information on the
researchers employed there or to post research publications.25

3. China Institute of International Studies (http://www.ciis.org.cn/)

Although the China Institute of International Studies (CIIS) is the main
research institute of the MFA, the site which it maintains contains relatively
little substantive information. Indeed, the CIIS Web page does little more
than provide a scant outline of the organization, brief biographies of its main
researchers, and links to the table of contents of each issue of 国际问题研究
(CIIS’s flagship journal).

4. China Institute of Contemporary International Relations

(http://www.cicir.ac.cn/)
Like the CIIS site, the China Institute of Contemporary International Relations
(CICIR) Web page is not especially informative. It does little more than offer
brief introductions to each of its research divisions and provide the table of con-
tents and abstracts of articles published in 现代国际关系 and 国际资料信息.

5. Shanghai Institutes of International Studies

(http://www.siis.org.cn/)
This site is more informative than the CICIR site. It offers introductions to
the Shanghai Institutes of International Studies (SIIS) itself and to each of
its research divisions. It also provides biographical information on each of
the institutes’ researchers. Moreover, it contains introductions to research

22
http://www.iwep.org.cn/.
23
http://ias.cass.cn/.
24
See http://iaps.cass.cn/.
25
See http://chinaborderland.cass.cn/.
The Internet and the Study of Chinese Foreign Relations 95

projects SIIS has conducted and lists of books and articles published by its
researchers.

6. University Sites
The handful of Chinese universities with significant international relations
programs now all maintain their own Web sites. However, such sites are,
generally, of minimal substance. For example, the main pages of Peking
University’s School of International Studies,26 Renmin University’s School of
International Studies,27 Fudan University’s School of International Relations
and Public Affairs,28 Tsinghua University’s Institute of International Studies,29
and China Foreign Affairs University30 all consist of little more than lists of
each program’s main organizational attributes and faculty members.

International Relations Professional Web Portals

and Scholarly Discourse
The number of reports and traffic on sites produced directly by the Chinese
media and on official sites of the main foreign policy institutes is staggering.
Yet, it is also the very volume and scope of the sites discussed above that
makes navigating them so daunting. In addition, as the main media sites cover
such a wide variety of issues, their discussion of foreign policy tends not to
be particularly well focused and is largely episodic (driven by the latest news
cycles and government directives). Moreover, with only limited exceptions,
these outlets rarely expand their coverage of either China or the world beyond
the limits set in official government statements. In contrast, in recent years a
handful of more specialized academic sites dedicated solely to foreign affairs
and international politics have come online in China. Although these sites are
less traveled, they possess a singular focus and have shown the potential to
stretch (if not directly challenge) the discursive limits for the discussion of for-
eign policy set by China’s leaders. As such, it is these nascent sites that appear
to be of even greater value to students of China’s foreign relations.
Before delving into specifics, it is important to emphasize that these portals
share a number of common characteristics. To begin with, each is designed
to cater to those with a pronounced interest in the study of international rela-
tions in general and Chinese foreign policy more specifically. They are clearly
not intended for a broad readership, and, in light of the fact that all of the sites
also have much more content in Chinese than in English, they seem to be cre-
ated mainly for consumption within China. Beyond such broad issues, these
sites share a number of additional similarities. First, they are all organized in a

26
http://www.sis.pku.edu.cn/.
27
http://sis.ruc.edu.cn/.
28
http://www.sirpa.fudan.edu.cn/.
29
http://166.111.106.5/xi-suo/institute/index.htm.
30
http://www.fac.edu.cn/.
96 Allen Carlson and Hong Duan

similar fashion. Content is divided into substantive issue areas, including top-
ics such as international relations theory and more policy-oriented discussions
involving national security concerns, Chinese diplomacy, and more detailed
studies of China’s main bilateral relations (particularly with the United States
and Japan). Second, all of the sites have tended to highlight Western interna-
tional relations scholars (including Morgenthau and Waltz, among others).
Indeed, it is worth noting that works by established contemporary Western
scholars are not only available but are also reviewed, and even download-
able on these sites. Third, these portals tend to contain extensive reviews of
the leading scholars and research institutes working within China. Finally,
the sites are all quite interlinked to each other (while also providing linkages
to sites outside of China as well). Although the list of such portals is likely
to expand over the next several years, at present two sites are of particular
importance.31

1. IR China ( 中国国关在线 ) http://www.irchina.org

IR China was founded by the Institute of International Studies at Nankai
University located in Tianjin. Zhang Ruizhuang, a senior scholar at Nankai,
is widely perceived as having been a prime force behind the development of
the site, but it is governed by an academic board composed of established
scholars from dozens of prominent foreign policy–related think-tanks, aca-
demic departments, and institutions within China. Among those in this orga-
nization from Beijing are Jia Qingguo, Qin Yaqing, Men Honghua, and Li
Shaojun. Shanghai participants include Yang Jiemian, Zhu Mingquan, and
Su Changhe. Indeed, this group encompasses virtually all of the best-known
scholars working on international relations in China and can generally be seen
as a shorthand list of who is inside the inner circle of China’s foreign policy
elite. In this sense, IR China is clearly the best established and influential of
the Chinese IR Web portals.
Such influence is underscored by the scope of content featured on the site.
The creators of the site have collected approximately one hundred Chinese
scholars’ works in various forms – journal articles, commentary pieces pub-
lished in newspapers, interviews, conference papers, book chapters, and oth-
ers. Such an array of sources has not been matched on any other Chinese site
and as such further sets IR China apart from other portals. Nonetheless, the
value of these postings may not be as great as they would first appear. Indeed,
upon close inspection the majority of publications that are placed on the site
are simply duplications of articles that were previously posted elsewhere. For
example, of the first five articles posted on the English version of the site’s
“International Observation and Hot Issues” page in the spring of 2009, all

31
When this chapter was first written, a third site, Chinese IR Study Network (中国国际关系
研究网) http://www.sinoir.com/, appeared to be poised to develop alongside these two sites.
However, underscoring the transitory nature of the Chinese Internet, the link to this site is
now inactive.
The Internet and the Study of Chinese Foreign Relations 97

were reproductions of articles that had appeared in print.32 The first five arti-
cles in the lead subject category on the site’s Chinese-language page, “学科
建设,” at this time were also all reprints.33 In other words, not much of the
content on this site is new. Instead, it mirrors quite closely the discussions that
have been featured in recent years in the main academic journals published in
China dedicated to foreign policy and international relations. In this sense, the
site appears to be less innovative and pathbreaking than initial consideration
might lead one to believe it to be.
Nonetheless, IR China also features two characteristics that are of interest
to students of China’s foreign relations. First, it is self-consciously concerned
with the development of the field of international relations in China. The chan-
nel mentioned above, “学科建设”, or “constructing the discipline,”34 contains
more than 100 scholarly articles on the study of international politics. Second,
it is widely connected to other sites related to the study of international rela-
tions in China. It offers links to dozens of universities and research institutes.35
Beyond these academic links, the site also provides linkages to major Chinese
foreign affairs and international politics journals and magazines.36

2. TECN Academic Net ( 天益学术网 )

http://www.tecn.cn/academic/index.php.37
In contrast to IR China, TECN Academic Net is administered not by a formal
academic institution but by a group of individuals. The site covers a variety of
fields within the social sciences and humanities, and its international relations
channel is especially impressive. The organizers of the site have established a
network of top Chinese scholars and worked to then post specific examples of
their works online. As with IR China, much of what is published simply cop-
ies articles that have previously appeared in various print sources. However,
the overlap on this site is not as pervasive as is common on other Web sites.
Indeed, quite a few of the works posted on TECN are originals. Of even
greater interest to students of Chinese foreign policy, TECN’s cyber papers
have frequently pushed the envelope established by official sources for the dis-
cussion of sensitive issues in international politics. Among the most prominent
examples of this trend are the following papers: Zhang Wenmu’s article on
the American “Tibetan Plan” and its failure,38 Wu Xinbo’s article on the U.S.
and East Asian integration,39 Wang Jisi’s article on the historical lessons of the
32
See http://www.irchina.org/en/news/hot.asp?cataid=25.
33
See http://www.irchina.org/news/xueshu.asp?cataid=22.
34
See http://www.irchina.org/news/xueshu.asp?cataid=22.
35
See http://www.irchina.org/xueke/inchina/jigou.asp.
36
See http://www.irchina.org/xueke/inchina/kanwu.asp and http://www.irchina.org/guancha/
link.asp.
37
Since late July 2009, this site has been inactive, leading to speculation that the site has been
closed. This is unfortunate for it contained relatively more original/unpublished works by
scholars than other available sites.
38
See http://www.tecn.cn/data/detail.php?id=18165. See footnote 37, above.
39
See http://www.tecn.cn/data/detail.php?id=18243. See footnote 37 above.
98 Allen Carlson and Hong Duan

Soviet-U.S. hegemonic competition and the road to China’s rise,40 Xu Yan’s

talk on China’s military buildup and national security,41 Song Wei’s article on
Diaoyu Island and China’s policy toward Japan,42 and Liu Junning’s article on
nationalism.43

IR-Focused Bulletin Boards and Blogs

Whereas media portals and academic Web sites largely represent an exten-
sion of the sources of which students of Chinese foreign policy have been
making use since the 1980s, in recent years cyberspace has also emerged as
a forum for the expression of broader public opinion about foreign policy
and international affairs. Indeed, although views from the general popula-
tion about foreign policy were almost entirely opaque through the end of the
1990s, nonspecialists in China are now using the Web to discuss China’s place
in the world. The most pronounced facet of this development has been the
emergence of bulletin boards and blogs that focus on international relations.

1. Bulletin Boards
Most Chinese Web portals provide bulletin boards. To post on BBs, Internet
users have to register; however, many BBs allow unregistered Internet users
to read postings. Moreover, it is now quite apparent that a large number of
Chinese netizens frequently post on such sites. For example, according to a
survey report released in January 2008, 35.4 percent of respondents reported
that they had posted information or replied to postings on online forums
and/or bulletin boards (CNNIC, 2008:54). Many of the sites discussed above
maintain BB pages. Three BB pages are of particular note.44
The most prominent BB is maintained by People’s Net and named, 强国, or
Strong Nation forum. The 强国 forum had its beginnings in May 1999 when
the then online edition of 人民日报 launched its first bulletin board – “the pro-
test forum” – to facilitate Chinese Internet users’ publication of their opinions
on, and protests against, the bombing of the Chinese embassy in Belgrade
during the course of the Kosovo war.45 This temporary site was then replaced
by 强国论坛, a forum that rapidly developed into a lasting virtual community.
It now contains dozens of categorized lists and has more than 680,000 regis-
tered users.46 Most recently, 强国论坛’s already prominent profile in Chinese

40
See http://www.tecn.cn/data/detail.php?id=5947. See footnote 37 above.
41
See http://www.tecn.cn/data/detail.php?id=2216. See footnote 37 above.
42
See http://www.tecn.cn/data/detail.php?id=364. See footnote 37 above.
43
See http://www.tecn.cn/data/detail.php?id=6810. See footnote 37 above.
44
Note that Chinese Military Online also maintains an expanding BB. See http://bbs.chinamil.
com.cn/site1/gwgfsq.
45
For a more detailed report, see http://www2.qglt.com.cn/fuwu/dt/hm99/hm9905.html.
46
See http://bbs1.people.com.cn/. A widely held belief is that a large number of unregistered
Internet users read postings in this community, particularly during periods when interest in
world affairs is most pronounced within China.
The Internet and the Study of Chinese Foreign Relations 99

cyberspace was boosted when Chinese president Hu Jintao visited People’s

Net and conducted an online dialogue with Internet users via this forum.47
The second set of BBs of special significance is maintained by China Net,
which hosts a number of discussion forums dedicated to foreign policy issues.
The most significant of these is the 和平论坛 (Peace Forum) that contains
forty subdiscussions, five of which focus on international affairs, Taiwan, and
­military/security concerns. In 2008, the forum had more than 120,000 regis-
tered members.48
The final BB of note is maintained by the MFA. The forum, entitled
中国外交论坛, is only maintained in Chinese. Like the other BBs, those wishing
to post on this page must register, but guests can monitor the forum without
taking this step. The forum, although appearing to be more closely monitored
by the authorities than either 强国论坛 or 和平论坛, is particularly interesting
because it has often been mentioned among the new technologies that China’s
leaders have begun to develop to gauge public opinion within China on for-
eign policy–related issues. In addition, MFA officials regularly log onto the
forum to answer questions from forum participants.
Although knowledge of these BBs is useful, it is essential for any student of
Chinese foreign relations to also consider the overall popularity of such forums
within the broader space of the Chinese Internet. However, gauging interest in
these sites is not a particularly straightforward process. To do so it is neces-
sary to first locate the most visited Chinese Web sites, which, according to a
2007 survey conducted by CASS’s Research Center of Social Development,
include Sina Net and Baidu.com, and the top Web search engines in China
were Sohu Net and Netease (CASS, 2007: 39). In addition, according to China
Web sites Ranking, a specialized Web Sites–ranking site sponsored by the
Internet Society of China, the top five most visited comprehensive Web sites
include Baidu.com, Tencent (the official Web site of QQ, the most popular
instant messaging software in China), Sina Net, Sohu Net, and Netease.49
Using this information as a guide, the authors of this chapter examined recent
postings on 中国外交论坛 50 and 和平论坛 51 and compared them with the num-
ber of postings on popular BBs on the portals mentioned above. This survey
found that these two forums have not attracted all that much Internet traf-
fic. In other words, in terms of visits and postings, both of these forums lag
far behind mainstream BBs. This finding suggests the limited influence and
appeal of such BBs in China, and should stand as a cautionary observation for
any scholars who attempt to draw broad generalizations about Chinese public
opinion via use of the messages that are posted on them.

47
For the transcript, see http://www.people.com.cn/GB/32306/33093/125024/index.html.
48
See http://forum.china.com.cn/ciicbbs.
49
For this ranking, see http://www.chinarank.org.cn/top500/Rank.do?r=1213942617523.
50
See http://forum.china.com.cn/ciicbbs/thread.php?fid=71.
51
See http://bbs.fmprc.gov.cn/board.jsp?bid=6.
100 Allen Carlson and Hong Duan

2. Blogs
Since the concept of blogging was introduced into China in 2002, the number
of blogs in Chinese cyberspace has expanded at a rapid clip. Indeed, accord-
ing to a recent CNNIC survey, by the end of November 2007, the number of
bloggers in China was about to reach 47 million, among whom almost 17 mil-
lion were considered to be active bloggers (CNNIC, 2007: 9–10).52 With the
rapid growth of blogs, it may not be a stretch to say that an alternative, indi-
vidualized, less-censored, and freer outlet of views and opinions has emerged
in Chinese cyberspace. However, the existing evidence seems to imply that
this form of online expression has not yet extended much in the direction of
China’s foreign relations and national security. To begin with, compared with
other topics, “international affairs” and “foreign relations” are far from a
popular focus of discussion for most bloggers or blog readers. In fact, accord-
ing to the CNNIC survey, the majority of blog postings in Chinese cyber-
space are bloggers’ records of their own experiences and life stories (CNNIC,
2007: 18). This is not to say that there are no Chinese netizens who blog on
international politics, but rather that BB forums have been a more popular
format for the discussion of these topics.53

Part Two: Making Use of Chinese Foreign

Affairs–Related Cyberspace
The previous section of this chapter treats the Chinese foreign affairs–related
Internet as a relatively static entity. Admittedly, this is a somewhat artificial
approach as it portrays a dynamic terrain as if it were unchanging and thus
it precludes any consideration of the ebbs and flows that are constantly at
play. However, as so little is known within the China-watching community
about this facet of the Chinese Internet, mapping out this territory represents
a necessary first step in beginning to explore its potential value for the study of
Chinese foreign policy. As noted above, such a space, especially when viewed
as an inert form, does not contain as much new or groundbreaking data as
one might expect. Thus, compared particularly to the access scholars and
analysts gained starting in the late 1980s to Chinese foreign policy elites,
Internet sources do not appear to offer a significant new pathway for con-
ducting research in the area of Chinese foreign relations. However, scholars
are also ill advised to simply ignore these resources, as they do contain useful

52
Also note the growing number of English-language sites that track blogging activity in China;
in particular, see http://www.virtual-china.org/, eastwestsouthnorth at http://www.zonaeu-
ropa.com/archive.htm, and the University of Heidelberg’s Digital Archive for Chinese Studies,
http://www.sino.uni-heidelberg.de/dachs/.
53
Moreover, compared to experts and scholars in other disciplines, IR scholars and experts
have been slow to take advantage of this new channel for publishing their work. Few estab-
lished IR scholars write blogs. Although a handful of scholars have blogged, including Pang
Zhongying (http://blog.sina.com.cn/m/pangzhongying), most of these blogs have not been
active in recent years or have stopped altogether.
The Internet and the Study of Chinese Foreign Relations 101

data. In general, the significance of these virtual sources is located in the ease
with which they may be accessed (in comparison to print resources or elite
interviews) and in the fact that they constitute a database on China’s foreign
relations that is in motion (rather than stationary). That being said, the fluid-
ity of this terrain, and the breadth of the materials posted within its limits,
can be challenging. In light of such promise and perils, there are then at pres-
ent three primary approaches that researchers may consider to maximize the
research potential of China’s foreign affairs–related cyberspace.

Treating Web Pages as Windows into the MFA,

Think-Tanks, and Research Institutes
Even though the Internet sites maintained by the major foreign policy insti-
tutions in China do not contain an abundance of new data on Chinese
foreign policy, they still constitute a useful source for the initial stages of
conducting research on China’s foreign relations. The main utility of such
sites is that they contain readily accessible outlines of the organizational
structure of both the formal institutions and informal ties that form the
core of China’s growing community of foreign policy elites. On the first of
these fronts, that of organizational structure, transparency has been incre-
mentally increasing ever since the late 1970s. However, as recently as the
mid-1990s, it was possible to track institutional development only through
painstaking library-based research, interview data, and reliance on a hand-
ful of authoritative secondary sources. In contrast, by monitoring the Web
pages these organizations host, it is now a relatively straightforward exer-
cise to locate the individual departments within these institutions, the for-
mal responsibilities with which they are charged, their personnel, and even
their publication records. For example, when studying the MFA’s Web page
in the spring of 2009, it was a simple endeavor to note that the organization
had established a new department, the Department of Boundary and Ocean
Affairs, and to learn that the director of the department was Ning Fukui.54
Looking beyond the MFA, the structure of each of the main foreign policy–
related institutes may also be traced quite easily online. For example, as
noted, the State Council’s CICIR Web site lists each of the organization’s
departments and department chairs.55 The same holds true for the sites
maintained by the MFA’s Chinese Institute of International Studies and the
various CASS research institutes.
It is also possible to track movement by individual scholars between vari-
ous universities and research institutes via the Web. Thus, by accessing Peking
University’s Web site, it can be confirmed that Zhang Qingmin, a promi-
nent foreign policy analyst and Sino-U.S. watcher who had been affiliated
with China Foreign Affairs University for much of his career, had become a

54
See http://www.fmprc.gov.cn/eng/wjb/zzjg/bianhaisi_eng/.
55
See http://www.cicir.ac.cn/tbscms/html/jgsz_En.asp?rid=jigou_en.
102 Allen Carlson and Hong Duan

member of the faculty at Peking University (Beida).56 Of even greater utility,

Zhang’s Web page at Beida contains contact information that ten years ago
would have been available to a researcher only after extensive phone calls
and the use of preexisting social networks. Moreover, Zhang is by no means
unusual in this regard. Indeed, over half of the professors listed on Beida’s
site provide contact information on their personal Web pages. Virtually all
of the scholars with pages on Tsinghua’s Institute of International Studies
Web site include their contact information.57 Although the Web pages main-
tained by the plethora of think-tanks in Beijing and Shanghai are not quite as
proficient in providing similar information on individual researchers, each of
these sites does maintain institutional e-mail addresses and phone numbers.
In short, although some of the data posted on these sites appear to be dated,
and even when current there is no assurance that meetings or interviews will
be forthcoming, at the very least it does greatly ease the process of making
initial contact with scholars working in China. This development represents a
significant change in the manner in which elite interviews, long a staple in the
field of Chinese foreign policy studies, can be arranged.

Using New Technologies to Track Changes in Internet-Based

Collections of Print Sources
The use of elite interviews and Chinese-language sources made an obvious
contribution to the study of Chinese foreign policy in the 1990s. However,
as the vast majority of this work was heavily qualitative in nature and often
lacked any direct consideration of issues of data collection or methodological
concerns, it was also rather open to challenges regarding the representative-
ness of the views that were culled from such sources. Iain Johnston helped lead
the field past such perceived limitations. More specifically, Johnston (1996)
developed a social scientific frame for the qualitative analysis of elite Chinese
views on foreign policy and national security issues in his work on China’s
historic realpolitik strategic culture and its imprint on the more recent past
(particularly Mao’s approach to military conflict). Although much of this
work focuses on the issues of cognitive mapping and symbolic analysis, in
other publications Johnston repeatedly returns to the technique of conducting
content analysis (searching out specific terms and key phrases) within large
open-source datasets. Over the last ten years, a growing number of scholars
followed Johnston’s move and made use of content analysis in attempts to dis-
cover new trends and tendencies within Chinese foreign policy statements and
analysis. However, such endeavors were exceedingly time-consuming and dif-
ficult to replicate due to a pair of limitations. First, as few of the sources that
were of interest to researchers were available in electronic form, print data

56
See http://www.sis.pku.edu.cn/web/Teacher_Browse.aspx?ID=101. For a full list of those
working at Beida see http://www.sis.pku.edu.cn/web/Teacher.aspx.
57
See http://rwxy.tsinghua.edu.cn/xi-suo/institute/english/faculty/faculty.htm.
The Internet and the Study of Chinese Foreign Relations 103

had to be coded by hand (a process that introduced the likelihood of human

error). Second, when data first became available via CDs and other electronic
formats, search engines for exploring them were rather rudimentary. As a
result, until recently the broader use of content analysis techniques has been
limited. However, over the last several years, the rapid expansion of online
collections of both Chinese- and English-language sources, and the emergence
of new technologies for tracking the data contained within them, have gone
a long way toward overcoming such difficulties. Indeed, these dual develop-
ments have now made content analysis, a task that was once particularly oner-
ous, a relatively straightforward task.
Daniela Stockmann’s chapter in this volume explores the use of many of
these new technologies with reference to the process of conducting content
analysis of Chinese media sources. In light of the comprehensive nature of
Stockmann’s survey, this chapter will not dwell at length on these issues.
However, two particular programs are of note. First, the Yoshikoder pro-
gram, which allows users to conduct relatively quick and sophisticated counts
of the frequency with which specific terms appear in texts, may be utilized
to examine electronic collections of official Chinese foreign policy state-
ments.58 Second, the China National Knowledge Infrastructure (CNKI) data-
base maintained by EastView in the United States, a foundational resource in
Stockmann’s own work, is of special use. To begin with, CNKI, via its “Core
Chinese Newspapers Database,” contains a readily accessible, and search-
able, collection of official statements relating to Chinese foreign policy. More
important, CNKI’s China Academic Journals Full-Text Database, especially
its subcollection “Politics/Military Affairs/Law,” contains a comprehensive
archive of elite analysis of Chinese foreign policy and international affairs.
Indeed, whereas gathering articles from the major Chinese foreign policy jour-
nals, let alone publications from provincial universities, was once a difficult
and time-consuming chore, now all of these materials are available online
for any subscriber to CNKI. Moreover, as CNKI’s search function is easy to
navigate, the site readily lends itself to rudimentary content analysis of such
publications. For example, the rapid emergence and subsequent decline of the
peaceful rise (和平崛起) debate in China can easily be traced through a search
of article titles appearing in CNKI’s “Politics/Military Affairs/Law” subsec-
tion. In 2002, no articles with 和平崛起 in their titles were in the database;
in 2003, there were 8 (included Xia Liping’s influential contribution on this
subject in 国际问题研究), whereas the following year there were 201 articles,
and in 2005 there were 307 – but by 2008 only 72.

Employing New Technologies to Track Changes in Cyber Terrain

The techniques discussed in this section so far may generally be seen as exten-
sions of conventional research strategies. Each promises to ease the collection

58
This technology can be downloaded at http://www.yoshikoder.org/.
104 Allen Carlson and Hong Duan

of data related to Chinese foreign policy, but neither breaks new ground in
terms of recognizing and managing the notoriously turbulent realm of cyber-
space. Indeed, until recently the shifting and transitory nature of this space
has largely defied categorization and measurement across the social sciences
(not only in the China field) as the technology to map this terrain was so
underdeveloped. However, during the last several years a growing number of
rudimentary research tools have been invented that are designed to more sys-
tematically track Internet data. Through exploring the utility of these tools,
scholars can begin to develop more accurate understandings of the flows of
information and networks that have formed in China around foreign policy
issues. Two specific new technologies appear to be particularly promising.
The first of these tools is the program Touch Graph, which uses a Java
application to track the way Internet sites are linked.59 The free download-
able version of the program provides users with a relatively easy to use device
for observing how popular sites are by presenting links in an active graph
matrix. It is also possible to make use of an online version of the program with
reference to Google’s online search engine. For example, via utilization of
the Google application of Touch Graph the connectivity of the Chinese MFA
was examined for this chapter. This was accomplished by pasting the MFA
site’s URL60 into Touch Graph’s search function, and the program produced
a graph that succinctly captured the centrality of this site in Web discussions
of Chinese foreign policy. The graph showed the MFA’s Web page to be quite
closely linked with the Chinese media, state, and the main Web page of the
PRC. Beyond governmental sites, a similar test was run on China’s premier
academic Web site dedicated to foreign policy issues, IR China. Once more
Touch Graph’s technology underscored the connectivity of IR China, particu-
larly with reference to organizations within China, such as the Ministry of
Education’s University of International Relations61 and IWEP, and also with
worldwide foreign policy research initiatives.62
The second tool is Technorati, which is used to collect online aggregate
data by tracking over 100 million blogs and the linkages between them.63
This program was originally designed to search English-language blogs and
its scope and ease of use make it a powerful tool. Thus, it is possible to utilize
Technorati technology to search the number of times a topic is discussed or a

59
This program is available at http://www.touchgraph.com. Ben Brake made a significant con-
tribution to the research and writing of this paragraph.
60
See http://www.fmprc.gov.cn/eng/.
61
See http://www.uir.cn.
62
Other programs which perform comparable functions include (but are not limited to)
SocSciBot, which is described at length by the leading information scientist Mike Thewall
in his handbook on online network analysis, available at http://linkanalysis.wlv.ac.uk/;
UCINET’s social network analysis/cultural domain analysis software: http://www.analytict-
ech.com/; and VOSON’s peer-produced tools for social science research of online networks,
available at http://voson.anu.edu.au/.
63
Available at http://technorati.com/chart/.
The Internet and the Study of Chinese Foreign Relations 105

word is used during a given time period, and garner a better sense of which
issues are attracting the most attention in cyberspace at any given point in
time. For example, using the program’s online site, a search was conducted
of the frequency with which all English-language blogs mentioned the phrase
“North Korea” during a 180-day period in the spring of 2009. The resulting
graph traced a surge of blog activity related to this term during a time period
that neatly maps with the escalation of tensions on the Korean peninsula.
The significance of such capabilities for students of China’s foreign rela-
tions is even greater now that the program also allows for the searching of
Chinese language–based blogs. As an early test case of such utility, during
a thirty-day period in the fall of 2007 the Technorati program was used to
track the frequency with which the Chinese blogosphere contained postings
containing the Chinese characters (布什) in President Bush’s name. During
this period, over 6,000 blogs made use of this combination of characters. In
contrast, in a subsequent search for the same term during a 180-day period in
the winter of 2008–2009, Technorati generated data that revealed that Bush’s
name had, not surprisingly, disappeared from the Chinese blogosphere. Of
perhaps greater interest, a parallel search of the Chinese term 朝鲜 (North
Korea) was also conducted during a period that mirrored that used for the
English-language search discussed above, and produced a similar pattern of
activity.
In sum, Touch Graph and Technorati should do much for those wishing
to make use of Chinese cyber data in their studies of China’s foreign policy.
Nonetheless, as useful as such programs appear to be, there remains a par-
ticularly steep challenge facing any scholar who seeks to employ Internet post-
ings as an indication of Chinese popular sentiment regarding foreign affairs.
Directly stated, broad questions loom about the overall representativeness
of views posted online as such a forum is available only to those who have
Internet access and who also engage in Web discussions. Although the sheer
number of Internet users in China has been increasing rapidly, this segment
of the population remains largely urban and educated, and the larger rural
population and less-well-educated population might have been left out.

Conclusion
The significance of the issues considered in this chapter is particularly acute
since over the last several years a new round of debate about China’s place on
the world stage has begun to unfold. Although such a discussion is far from
new, this is the first time it has taken place during a period when Chinese
Internet-based sources are readily available. Thus, a more pointed reprise to
the query posed in the Introduction involves asking whether reliance upon
such sources has afforded researchers with greater insight into the nature of
China’s current rise.
In this vein, it is first readily apparent that the use of Internet resources
in analyzing Chinese foreign affairs at least has made for fuller, richer
106 Allen Carlson and Hong Duan

descriptions of Beijing’s policy decisions and the motives underlying them.

However, not surprisingly, it has done little in regard to creating a consen-
sus about China’s rise. For example, on the one hand, China security expert
M. Taylor Fravel recently surveyed Chinese military Web sites to construct an
argument about the posturing of PLA forces in China’s frontier regions. Fravel
uses these sources to support the claim that there is a decidedly defensive
rationale informing the deployment of Chinese troops within China (Fravel,
2007: 701–737). On the other hand, Paul Mooney, an established journalist,
surveyed other facets of Chinese cyberspace in 2005 and concluded that the
Internet was “fanning the flames of nationalism” (Mooney, 2005).
Such contrasting observations point to the remarkable diversity of Internet
resources related to Chinese foreign relations. They are also illustrative of the
fact that these new data have done little to bring analysts together in regard to
the portrait that is now painted of China’s rise. This being the case, it is too
much to ask of any type of source, particularly one as fluid as Chinese foreign
policy cyberspace, that its value be gauged primarily by the degree to which it
can create consensus among those who make use of it. However, at the same
time, it is not unfair to critically examine the ways in which scholars have
utilized such new resources. Unfortunately, on this score the majority of the
existing work in the foreign policy field was designed with little reference to
methodological concerns or consideration of basic issues regarding the collec-
tion and use of empirical data. In other words, although there is much promise
in mining cyberspace for the study of Chinese foreign relations, so far schol-
ars have failed to fully realize its potential. This chapter, then, is designed to
provide researchers with the main coordinates of this terrain and with sugges-
tions on how to develop more effective research approaches to it.
In closing, as China’s foreign relations–related Internet continues to expand
in the coming years, the researcher’s ability to make use of it will likely con-
tinue to be hobbled by numerous difficulties. Chief among these, the appear-
ance of data in cyberspace tends to be fleeting, and the volume of content
published there is massive. In addition, questions over the reliability and
representativeness of the information that appears on the Chinese Internet
(especially with reference to issues of access, firewalls, and censorship) will
continue to pose vexing problems for researchers. Moreover, especially in the
realm of BBs and blogs that make use of rather stylized, colloquial versions
of standard Chinese, language issues will remain a challenge for non-native
speakers. Finally, as few students of Chinese foreign policy are trained to make
use of advanced computer programs and software, technological challenges
will persist. Nonetheless, most of these obstacles may be overcome through
close monitoring of this terrain, coupled with a utilization of the strategies
discussed in this chapter. As such, the Chinese Internet is poised to emerge as
an increasingly useful tool in the study of China’s foreign relations.
6

Information Overload?
Collecting, Managing, and Analyzing
Chinese Media Content

Daniela Stockmann

Only two decades ago the main information source for China scholars abroad
was the Foreign Broadcast Information Service (FBIS), an open intelligence
source of the CIA. FBIS collected, translated, and disseminated available news
and information from Chinese media sources. During the Cultural Revolution
most foreign researchers were not allowed to conduct fieldwork in China; the
five-times-weekly FBIS report thus constituted one of the main sources for
information about events in the mainland. Since original media sources only
became available during the reform era, those interested in Chinese media
reporting usually relied on the information selected and translated by FBIS.
Aimed at the U.S. intelligence community, FBIS did not draw an unbiased and
representative sample of the Chinese news, but, at the time, it constituted one
of the best information sources available to the research community abroad.
Today, the situation could not be more different. Since China’s opening up
to the outside world, we have the opportunity to spend extensive time in the
country reading, listening, and watching the same news as Chinese citizens.
Even when we are not in China, we can access a large number of Chinese media
sources electronically using the Internet, cable, or satellite dishes. In addition
to increased access, we also have the opportunity to store this information.
We can mail abroad newspapers, magazines, and DVDs from China, scan
documents, download files, and record television and radio programs. Most
of this information is stored electronically. Once preserved, the data gathered
in countless hours sit on our hard drives, waiting for analysis. Instead of read-
ing the FBIS report, we simply access a folder on a computer and immediately
have data available for research. What a wonderful world for doing research
on China – or so it would seem.

For insightful comments and suggestions I would like to thank Allen Carlson, Iain Johnston,
Will Lowe, Zhang Jie, Jonathan Hassid, Jessica Weiss, and Jamie Reilly. For sharing data I am
grateful to Ku Lun-wei from National Taiwan University and Deborah Cai from the University
of Maryland. Many thanks as well to Wang Mingde for research assistance.

107
108 Daniela Stockmann

The amount of information available inside as well as outside of China has

increased tremendously over a relatively short time. Inside China, the media
sector grew by about 100 percent between 1978 and 2008.1 Outside China,
information about the country has increased as well. A growing number of
Chinese publications aimed at propagating China’s policies abroad have been
established since the 1980s.2 There is also more and more news about China in
the foreign press.3 Today, it is a challenge to keep track of the news on China
simply because there is an enormous amount of information out there that
cannot all be processed simultaneously.
A second challenge is that the available information turns over rapidly.
Some newspapers close while others are newly founded; electronic search
engines suddenly appear on the Internet but disappear quickly. Therefore,
many China specialists have started to become collectors of data, not only
accumulating data directly related to their current research projects but
preserving other interesting materials for fear that the information will
never be found again. As a result, we are not only dealing with the chal-
lenge of trying to handle and process the information out there but also with
large chunks of data stored in offices and on hard disks queuing up to be
analyzed.
None of these developments is necessarily specific to China. One aspect of
globalization has been an increase in information and a more rapid pace of
information turnover. The success of search engines – such as, for example,
Google – demonstrates that there is a demand for services that assist people
in locating information. Similarly, decreasing prices of technology that allows
storage of information combined with the higher speed of the Internet and
the increased opportunities for file-sharing have resulted in huge chunks of
unorganized data on many computers. Software firms are therefore in the
process of developing file-storage systems that facilitate data management on
computers. Hence, what we observe in China is part of a broader trend. What
is specific to China, however, is the rapid pace with which we have moved
from information scarcity to overabundance.
China scholars have had difficulties adapting to the changes resulting from
the information revolution. Up to the present, there has been little scholarly
discussion about the methods and techniques that can be employed to cope
with the rapidly increasing and changing information environment. In this
chapter, I explain the usage of electronic aids to collect, manage, and analyze
data originating in the Chinese news media. Since electronic aids to analyze

1
Newspapers increased from 186 to 1,943, periodicals from 930 to 9,549, television stations
from 32 to 287 (excluding guangbo dianshitai), radio stations from 100 to 263, and Web sites
from 0 to 2,878,000. See http://www.gapp.gov.cn; http://www.drcnet.com.cn; http://www.
stats.gov.cn; http://www.cnnic.cn; http://number.cnki.net; http://press.gapp.gov.cn.
2
The China Daily, an English newspaper, was established in 1981. The People’s Daily Overseas
Edition was founded in 1985 (information taken from each newspaper’s Web site).
3
See, for example, Ethan Zuckerman’s Global Attention Profiles, at http://h2odev.law.harvard.
edu/ezuckerman/, accessed January 24, 2008.
Information Overload? 109

Chinese texts qualitatively have so far not been developed, the focus of this
chapter is only on content analysis. The main advantage of relying on the tech-
niques described in this chapter is that they assist us in conducting research in
a systematic fashion. Random sampling automatically leads us to ask questions
about the nature of the data available for our research and the generalizability
of our conclusions to China as a whole. Content analysis software enables us
to remain consistent across a large number of texts. Therefore, electronic aids
help us to gain a more accurate understanding about China and to strengthen
our conclusions about where China is heading.
The explanations laid out in this chapter are relevant to those planning to
use content analysis in their work. Content analysis quantitatively analyzes
messages, broadly defined (Neuendorf, 2002). It is therefore different from
qualitative text analyses, such as, for example, discourse analysis or grounded
theory (Charmaz, 2006; Wood and Kroger, 2000). Content analysis best
answers questions of who, where, how many, how much, and the relationship
between specific variables. Although these questions are raised across differ-
ent subfields in Chinese politics, so far content analysis has been primarily
used to examine Chinese media content. Such studies are conducted for two
reasons: first, researchers use the press as a social indicator – as a window to
understand opinions of intellectuals (see, for example, Gu, 1996; Johnston
and Stockmann, 2007; Li and White, 1991). Second, scholars aim to under-
stand the causes and consequences of propaganda through the news media
(see, for example, Esarey, 2009; Hassid, 2007; Stockmann, forthcoming-a;
Stockmann and Gallagher, forthcoming; Wang and Tan, 2008).4 However,
content analysis can also be applied to other sources. Political scientists have
used content analysis of speeches, letters, party manifestos, textbooks, gov-
ernment bills, and court rulings to study beliefs of political elites (Burden and
Sanberg, 2003), citizen opinions (Lee, 2002), the political ideology of parties
(Laver and Garry, 2000), international conflict (Bar-Tal, 1998), political agen-
das (Martin, 2004), and the relationship between different legislative institu-
tions (Kilwein and Brisbin, 1997), to name just a few examples. Apart from
the media, content analysis can therefore be applied to a wide range of texts.
In Chinese politics we now have access to diverse sources, many of which
are in electronic format, such as, for example, political speeches, government
documents, court rulings, governmental Web sites, textbooks, lectures, and
Chinese scholarly work. Although this chapter focuses on the Chinese media,
the techniques explained here can also be applied to other sources relevant for
the study of Chinese politics.
This chapter will proceed as follows. In the first part, I provide an overview
of electronic sources of the Chinese news media, focusing on Web sites and
archives that provide useful tools for collecting news media sources. Next, I lay
out sampling techniques for content analysis and show how to investigate the

4
Before the reform period, content analysis of content was also used to understand elite politics
(Walder, 1979).
110 Daniela Stockmann

representativeness and efficiency of the sampling size. Subsequently, I explain

techniques of data management and analysis. In particular, I explain the use
of a software program for content analysis of texts in simplified Chinese char-
acters. I conclude by discussing the limitations and potential of the use of
electronic sources in content analysis for further advancing our understanding
of Chinese politics.

Electronic Sources of the Chinese News Media

When collecting news articles, researchers of advanced industrial democracies
can usually rely on data archives, such as, for example, Lexis Nexis or the
Television News Archive at Vanderbilt University in Tennessee. China schol-
ars are not so fortunate. Most datasets for research on Chinese politics are
not archived. However, China scholars are usually willing to share existing
datasets when personally contacted. In recent years, a number of datasets of
Chinese news content have been created, which are, in principle, available for
further research. One of the first datasets was collected by the University of
Maryland for the U.S.-China Security Review Commission (Maryland Study,
2002). This dataset sampled newspaper articles on the United States in six
newspapers between 2001 and 2002. Stockmann (2007) created a dataset of
all articles relevant to labor law in three newspapers in Chongqing during a
period of three months in 2005. Esarey (2009) collected information about
propaganda in newspapers in Shanghai, Beijing, and Guangzhou. Hassid
(2007) analyzed aggressiveness in twenty-six newspapers between 2004 and
2006. More recently, Stockmann (2009) examined reports on the United
States in three newspapers in 1999 and 2003. The usefulness of these exist-
ing datasets for further studies of Chinese news media content is, however,
limited. All of these existing datasets sample news articles published only in
newspapers, excluding other media sources; they also encompass relatively
short periods of time; and most deal with specific issues, such as legal news
reporting about labor or international news reporting about the United States.
These features make it difficult for other researchers to use the data for fur-
ther inquiry. Original research involving the Chinese news media will there-
fore often require the collection of news reports for content analysis. Below I
explain which electronic sources may be used for data collection.

Search Engines of News Web Sites

The obvious place to sample from the Chinese news media is the Internet.
Many (but not all) traditional media make their news reports available online.
CCTV, for example, has a search engine for transcripts as well as video-clips
of previous reports going back to 2000. Similarly, newspaper content can
often be searched online starting as early as 2000. Online news Web sites that
selectively display news articles by other news organizations, for example,
Sina, allow researchers to see which articles were displayed on the Web site
Information Overload? 111

on a particular day (going back to 1998). Finally, Xinhua news reports can
be searched online from 2000 on. A list of useful electronic search engines of
influential national and local media sources, such as those mentioned above,
is available on www.daniestockmann.net. When using these search engines,
however, researchers should be aware of the following problems.
First, news content displayed on Web sites of traditional media outlets,
including newspapers, magazines, television, and radio, is not necessarily the
same as that broadcast or printed in traditional media outlets. Some articles
are published in online editions only, and media staff have more leeway in writ-
ing these articles (He and Zhu, 2002; Stockmann, forthcoming-b). The choice
of a search engine depends on whether researchers want to explore online
news content or news content in traditional media. As a rule of thumb, search
options directly available on the Web site’s “home” tend to search online news
content, whereas search engines of broadcast or printed news content can be
found through links named “advanced search” or “electronic edition.”
Yet even when located correctly, search engines of traditional media out-
lets differ in terms of how they search and return news content. Some search
engines allow full-body searches based on keywords, returning txt-files,
whereas others allow searches based on publication date, returning pdf-files.5
These differences may be irrelevant when scholars aim to select articles solely
based on publication date since all search engines share this feature. Most of
the time, however, we are looking for specific keywords in the body of a text.
In these cases, we may introduce a selection bias in our data collection. Media
outlets in certain regions are more likely to use search engines that return txt-
files. For example, most newspapers in Beijing use search engines that allow
keyword-based searches and return txt-files, whereas those in Chongqing are
usually limited to searching by date only, and they return pdf-files. Therefore,
we should keep in mind that we may introduce systematic biases into our
research when relying on one kind of search engine as opposed to another.6
The same is true for solely relying on online news Web sites to collect news
reports. Clearly, those traditional media outlets that have online news editions
with functioning search engines are also the ones that are more profit-oriented
and commercialized. For example, local metro and evening papers are more
likely than local official papers to have online editions. And newspapers from
the more developed areas on the east coast are more likely to be present on

5
These features do not necessarily go together, but in practice they often do.
6
Inaccurate results may also result from three kinds of measurement error: first, they some-
times return texts that do not match actually printed articles 100 percent. To double-check,
researchers can compare a sample of results retrieved electronically with broadcast or printed
texts. Second, the same article is sometimes returned multiple times, in which case the
researcher needs to correct the number of “hits.” Finally, error may be introduced when search
engines differ with respect to the techniques used to identify keywords. The same article may
be a “hit” in one search engine but not in another. To hold techniques for keyword searches
constant, one can select search engines that simultaneously search a number of media outlets.
Newspaper conglomerates sometimes offer this option.
112 Daniela Stockmann

the World Wide Web than those located in the western inland regions (He and
Zhu, 2002). To get a representative sample of news content of a particular
region in China, it is therefore advisable to rely on additional means to col-
lect news reports. Some of these means are also available electronically, as
explained in the next section.

Other Electronic Sources

Selection bias due to search engines and the media outlets’ presence on the
Internet can be partially overcome using other electronic resources. To my
knowledge, publicly available databases for television and radio do not exist in
China. My focus in this section is therefore on traditional print media.
Most official papers can be accessed through the China National Network
(CNKI.net). In total, the “China Core Newspapers Full-Text Database”
(Zhongguo zhongyao baozhi quanwen shujuku) includes over 700 newspa-
pers, starting from 2000. Most newspapers in this database are so-called offi-
cial or party papers; evening and metro papers are rarely found. Nevertheless,
this dataset complements online news Web sites. More complete but also more
costly is a newspaper database by Apabi.7 In addition, the People’s Daily can
be searched from 1946 to the present using the “Renmin Ribao Full-Text
Archive,” accessible through many libraries.
Electronic versions of many newspapers that cannot be found on the World
Wide Web are available on CD-ROM. This option is particularly interesting
for researchers who intend to do time-series analysis. For example, the Beijing
Evening News can be searched on the Internet beginning from 2005, but can
be purchased on a CD-ROM for all years between 1997 and 2004.8
Electronic databases as well as CD-ROMs constitute additional electronic
sources that can be used to complement data collection based on the search
engines of news Web sites. Although these sources help us overcome a selec-
tion bias when relying solely on news Web sites, the correction may only be
partial. In order to collect representative material, researchers may still need
to go back to hard-copy versions of print media or tape-recordings of broad-
casting media, followed by subscriptions to these materials into electronic
formats when using computer-assisted content analysis. A smaller random
sample of hard-copy material can be used to assess the magnitude of the selec-
tion bias, thus allowing the researcher to make corrections (Heckman et al.,
1998). However, there is no doubt that the electronic sources available greatly
facilitate data collection. Since the exact nature of these electronic sources,

7
Available at www.eastview.com/Online/AsianProducts.aspx. A free (but more limited) version
of Apabi is available at www.press.idoican.com.cn, accessed May 26, 2009. Another newspa-
per database called WiseNews appears to be using CNKI.net. See www.wisers.com/corpsite/
global/en/products/wisenews.html, accessed January 8, 2010.
8
Contact information and prices for the Beijing Evening News and Beijing Daily are available
at http://www.bjd.com.cn/com/2001gp.htm, accessed January 24, 2008.
Information Overload? 113

especially news Web sites, changes rapidly, however, my final advice for data
collectors is to plan to finish the data collection within a short period of time,
especially when relying on the Internet. If data collection extends over several
months, some useful Web sites may have already been lost in the data jungle
of the World Wide Web.

Drawing Representative and Efficient Samples

Given the vast amount of information available to China scholars today, it is
hardly possible to examine all materials on a particular research topic. Unless
one’s research question is narrow enough to limit the number of documents
to one that is manageable, the best solution is to randomly select a sample
from the “population,” the set of units about which the researcher wishes
to generalize. Nonrandom, so-called nonprobability, samples should be used
only if no other options exist since we cannot generalize findings from non-
random samples to a population. For content analysis, the unit of analysis is
often a message (broadly defined), but depending on the research question
the researchers may also break it down into message components or choose
a broader unit, such as, for example, media sources. Once the population is
defined, it must serve as the basis for the sampling.9
At present, two kinds of sampling techniques are primarily used by China
scholars when analyzing media sources. The first one is called systematic
random sampling which selects every xth unit from a list of all units in the
population (also called the sampling frame) or in some flow of occurrence
over time. For example, the Maryland Study of Chinese news content about
the United States selected the issue of six newspapers every second day over
a period of several months (Maryland Study, 2002). A big disadvantage of
using systematic random sampling when working with media sources is that
the format differs depending on the day of the week. For example, the Beijing
Youth Daily publishes a special edition on Mondays, which is especially
popular among Beijingers (CPCR, 2005). If this periodicity matches the skip
interval of systematic random sampling, the representativeness of the sample
may be negatively affected. Therefore, studies employing systematic random
sampling often lower the skip interval and thus increase the sample size. This,
however, may result in oversampling: researchers analyze more data than are
needed to retrieve estimates representative of the population.10 To avoid this
problem, Esarey (2009) and Stockmann (2009) have used a second sampling
technique, called constructed-week sampling, for the analysis of daily news-
papers. According to this technique, all weeks during the period of interest to

9
For a detailed introduction to different definitions of populations and sampling for content
analysis, see Neuendorf (2002).
10
If the size of the population and the desired sample size are known, researchers can divide
the population size by the sample size to calculate the skip interval to avoid oversampling
(Neuendorf, 2002).
114 Daniela Stockmann

the researcher are numbered, and subsequently one Monday, Tuesday, and so
on is randomly selected until one, or several, weeks are constructed (Lacy et
al., 2001; Stempel, 1952). The procedure works similarly for weekly publica-
tions (Riffe, Lacy, and Drager, 1996) and online news sites (Wang X., 2006).11
Constructed-week sampling is especially attractive for research that involves
multiple years because it relies on a small sample size while, at the same time,
retaining representative results.
Yet how do we know how many constructed weeks to sample? Ideally, we
would like to draw a sample size which is “just right.” That is, an efficient
sample size is achieved at a point when increasing the number of cases will not
significantly reduce the sample error, whereas decreasing the number will sig-
nificantly damage the representativeness of the results. In the American con-
text, media scholars have found that two constructed-week samples constitute
the most efficient sample size for daily newspapers (Lacy et al., 2001). To test
whether the same sample size applies to the Chinese context, I compared the
results of the Maryland Study’s systematic sample, which oversampled the
number of cases, with the results of constructed-week samples over a period of
ten months.12 If the estimates based on the constructed-week samples are close
to the estimates drawn from the over sampled systematic sample, we can be
confident that these findings are also valid estimates of the population data.
In doing this comparison, I was particularly interested in differences
between different types of newspapers. Therefore, I chose the Beijing Youth
Daily as an example of a nonofficial paper and the People’s Daily as an exam-
ple of an official paper. Official papers include papers under direct supervision
of state units (such as, for example, the Worker’s Daily); nonofficial papers are
evening and metro papers that are run with a stronger commercial orientation.
Chinese urban residents prefer reading nonofficial papers (Stockmann, forth-
coming-a). For each newspaper, the Maryland Study allowed me to assess how
many articles about the United States were published per day and the average
tone in these articles per day.13 Treating the systematic sample as the popula-
tion, I examined the likelihood that the sample means of these variables in the
constructed-week samples fall within one and two standard errors of the pop-
ulation means. According to the Central Limits Theorem, 68 percent of the
sample means should fall within one standard error of the population mean
and 95 percent of the sample means should fall within two standard errors of
the population mean. Accordingly, a sample size is considered effective only if

11
Riffe, Lacy, and Drager (1996) argue that stratification by month, followed by simple random
sampling of two days per month, is the most efficient sampling method for weekday TV net-
work news in the United States.
12
The Maryland Study includes about 46 percent of all articles published during ten months
(September to December 2001; February to July 2002); thus it is highly representative of the
population.
13
These variables were assessed by trained coders. More detailed information about coder
training and measurement can be found in the Maryland Study (2002) or in the online appen-
dix available at www.daniestockmann.net.
Information Overload? 115

Table 6.1. Percentage of 100 samples falling within one and two standard
errors of the population mean in two newspapers

People’s Daily Beijing Youth Daily

Number of Average Tone of Number of Average Tone of

Articles about Articles about the Articles about Articles about the
the U.S. (per day) U.S. (per day) the U.S. (per day) U.S. (per day)

1 s.e.a 2 s.e. 1 s.e. 2 s.e. 1 s.e. 2 s.e. 1 s.e. 2 s.e.

1 week 20% 32% 22% 39% 22% 35% 15% 34%
2 weeks 32% 51% 29% 55% 25% 55% 25% 49%
3 weeks 28% 56% 39% 66% 35% 60% 32% 62%
4 weeks 47% 72% 44% 69% 51% 78% 37% 68%
5 weeks 65% 90% 49% 75% 46% 75% 48% 83%
6 weeks 53% 90% 57% 81% 53% 78% 47% 83%
7 weeks 48% 86% 47% 87% 56% 86% 47% 88%
8 weeks 60% 91% 50% 89% 65% 92% 60% 88%
9 weeks 60% 87% 62% 93% 62% 96% 68% 94%
10 weeks 69.00 100% 75% 98% 72% 96% 70% 96%

Population 6.43 0.18 9.65 0.15

Mean34
(s.e.) (0.32) (0.07) (0.41) (0.05)
{s.d.} {3.81} {0.81} {4.83} {0.60}

Source: Maryland Data.

a
   By Mean; s.e. = standard error; s.d. = standard deviation.
b
  Tone was assessed based on trained interviewers, ranging between +3 (highly positive) and –3
(highly negative). See the online appendix at www.daniestockmann.net for details. Results do
not differ significantly when using sampling weights that account for the sampling design (first
drawing a systematic sample from the whole population and then drawing a constructed-week
sample). Results can be retrieved from the author at [email protected].

its sample means distribution meets these standards. As displayed in Table 6.1,
ten constructed weeks fulfill this requirement when 100 samples are randomly
drawn from the Maryland Study data. This sampling size did not differ much
between the People’s Daily and the Beijing Youth Daily. When it comes to
daily newspapers, constructed-week sampling is preferable to systematic sam-
pling as it significantly reduces the sampling size, in this case by 50 percent,
while, at the same time, remaining representative.14
Do you always need to draw ten weeks when using constructed-week sam-
pling? The answer depends on the goal of the study. My example here is most
relevant for those who intend to generalize from a small number of media
reports to media reporting over an extended period of time, in this case about
a year. If the purpose of the study is to examine how media reporting changed

14
The Maryland Study collected 140 out of 303 days; 10 constructed-weeks sum up to 70 out
of 303 days.
116 Daniela Stockmann

over a short period of time, such as, for example, in response to a specific
event, the population of articles may be small enough that sampling is not
necessary. Furthermore, my example relates to international news reporting,
especially with respect to the United States. When researchers are interested in
features with a lower variance in day-to-day news reporting than the standard
deviations displayed in Table 6.1, they can reduce the sample size.15 For any
other topics that share the same variance in coverage to ten articles per day,
ten constructed weeks should be sufficient.
Now that I have explained how to go about collecting samples, let us con-
sider the next step in conducting content analysis: managing and analyzing
the data.

Managing and Analyzing Data Using

Digital Technologies
At first glance, seventy days do not seem to represent news reporting in one
newspaper during almost a whole year. Yet once we consider that each day
includes more than one article, our dataset of ten constructed weeks may soon
contain about 4,000 articles dealing with one issue area – reports related only
to the United States. All of these articles will need to be managed, read, coded,
and analyzed.
A common way to deal with the problem of analyzing large datasets is to
organize a number of people to help with the coding and content analysis.
Although this makes a lot of sense, it also creates potential problems in terms
of consistency since different individuals may use different standards to eval-
uate news content. For example, one person may interpret an article as some-
what negative whereas the other may find it more neutral in tone. Therefore,
scholars have placed much emphasis on the training of content analysis per-
sonnel and have developed mathematical techniques to test coder interreliabil-
ity (Perreault and Leigh, 1989; Riffe, Lacy, and Fico, 1998).
Consistency may also be ensured by using digital aids. For many languages
there are qualitative and quantitative software programs that can be used for
content analysis. So far, there is only one software program that is able to
recognize and analyze simplified Chinese characters. Yoshikoder is an open-
source software available for free on the Internet.16 In addition to Chinese,
it is available in many other languages and thus allows for comparative con-
tent analysis across different languages. More specifically, the program may
be used qualitatively to facilitate comparisons of texts that mention specific
keywords. Researchers can upload a number of articles simultaneously and

15
This also explains why Lacy et al. (2001) decide that two constructed weeks are sufficient: the
variables they examine are characterized by low variance.
16
The software runs on both Windows and Macintosh operating systems. It can analyze txt-
files in UTF-8 format. For general information, see www.yoshikoder.org; for information on
how to run Yoshikoder in Chinese, see www.daniestockmann.net.
Information Overload? 117

color keywords of interest. This function comes in handy when comparing the
usage and framing of certain words in a single or several texts. However, the
program’s strength lies in its quantitative content analysis. The software can
count specific characters or groups of characters, either in the whole text or
within a certain distance to a keyword. This allows the quantification of the
use of certain concepts or categories of words. Below, I use two examples to
illustrate how these functions may be used in practice.

Assessing the Sensitivity of News Content by Counting Keywords

Researchers of the Chinese media agree that the state generally exerts a con-
siderable amount of control over Chinese media content. However, govern-
ment control over news reporting in China changes over time. In addition to
major political events during which media reporting may be restricted, such
as, for example, during the SARS crisis in 2003, each year the Chinese media
undergoes a regular cycle of tightening and loosening. At around the time of
the Spring Festival and the meeting of the National People’s Congress (usu-
ally between late January and March), reporting is supposed to be positive.
Toward the end of the year there is more room for criticism (Stockmann,
2007). The dynamic nature of space for news reporting potentially creates
problems when sampling news content. For example, in my study of news
reporting about the United States, I was interested in the tone of the news
content. Yet if, by chance, my constructed-week sample overrepresented the
period during the Spring Festival, perhaps my results were overly positive.17
Yet there is a way to account for these changes of space for news reporting
over time in the Chinese media.
In my content analysis I am using the percentage of Xinhua articles pub-
lished on the same topic on a particular day as a means to control for sensi-
tivity over time. Xinhua articles are sometimes reprinted in newspapers as
must-carry news. In other words, the Propaganda Department requests that
Xinhua articles be published instead of reports written by the newspaper’s
own journalists. Explicit requests are rare today, but editors still feel that they
are on the safe side by publishing Xinhua articles when reporting about sensi-
tive issues. Handbooks distributed by newspapers to journalists for training
purposes also recommend that journalists rely on Xinhua material when cov-
ering certain issues (Stockmann, 2007). Researchers of the Chinese media can
therefore use Xinhua reporting as a proxy for sensitivity over time.
Note that I look at changes in the percentage of Xinhua articles over time,
not whether an article is equivalent to a Xinhua report. Especially when it
comes to international news reporting, newspapers are somewhat dependent

17
This problem is not easily addressed when employing stratified sampling. First, we do not
know precisely what news reporting in China would look like if journalists were not to self-
censor. It is also difficult to set a precise date marking the beginning and end of periods dur-
ing which articles are more tightly or loosely controlled.
118 Daniela Stockmann
Percentage of Xinhua Articles Per Day

100%
NPC
90% Spring Festival
80%
70%
60%
50%
40%
30%
20%
10%
0%
Sep Oct Nov Dec Feb Mar Apr May June July
Time
Figure 6.1. Percentage of Xinhua news articles among all articles about the United
States published in the People’s Daily, September 2001–July 2002 (no data for January
was available).
Source: Maryland Data.

on the Xinhua News Agency as a source of information. Most newspapers do

not have permanent correspondents abroad. As a result, editors often publish
Xinhua reports, even when the Propaganda Department has not imposed any
restrictions. Yet in an effort to attract readers, many newspapers are making
an effort to write their own articles. As a result, nonofficial papers, like the
Beijing Youth Daily, are actually less likely than official papers, such as the
People’s Daily, to simply reprint Xinhua reports.18 In this case, reporters sup-
plement Xinhua reports with alternative news sources, such as, for example,
the Internet and the newsletter of the U.S. embassy. During major events, such
as the 2003 invasion of Iraq, media institutions may also temporarily send
reporters abroad (Stockmann, 2007). Therefore, newspapers are partially
dependent on the Xinhua News Agency. Simply using a dummy variable for
being a Xinhua report will be insufficient to assess sensitivity over time. Only
when looking at changes in the percentage of Xinhua articles published on
the same topic will we be able to identify periods during which editors reprint
Xinhua articles to meet the demands of the state.
Figure 6.1 confirms that this measure indeed represents a valid indicator
for sensitivity over time. The Maryland data allowed me to investigate how
Xinhua news reporting developed over a period of almost one whole year,
starting in fall 2001 and ending in summer 2002, excluding only August 2001
and January 2002. In Figure 6.1 the x-axis indicates time, the y-axis shows
the percentage of Xinhua articles related to the United States published each
day. As expected, the percentage of Xinhua articles increased during periods
of high sensitivity. Immediately after September 11, space for press report-
ing was relatively open until foreign correspondents in China reported that

18
Compare Figure 6.1 with Figure A1 in the online appendix at www.daniestockmann.net.
Information Overload? 119

Chinese media coverage was not very sympathetic.19 The state determined that
this attitude was not constructive and tried to replace it with a more sympa-
thetic view.20 After the government stepped in, media reporting related to the
United States became highly sensitive (Stockmann, 2007). After September
11, the percentage of Xinhua articles increased visibly when reporting about
the United States, covering such topics as politics, economics, culture, soci-
ety, sports, and entertainment. In mid-November, sensitivity went back to
average levels at around 50 percent of the articles per day. In addition to this
period of tension in Sino-U.S. relations, sensitivity increased during the period
of the Spring Festival and the meeting of the National People’s Congress. In
February and March 2002, the percentage of Xinhua news articles increased
to about 60 percent. By late April, space for news reporting had loosened
again, continuing into the summer, although events related to the American
war on terrorism in June and July somewhat increased sensitivity again.21 The
same pattern is found when observing the percentage of daily Xinhua reports
in the Beijing Youth Daily. Overall, the measure responds in sensible ways to
changes in the space for news reporting about the United States.
To reproduce this measure using content analysis software, it is necessary
to account for whether an article constitutes a Xinhua report. Fortunately,
Chinese newspaper articles usually cite Xinhua as the source, especially
when an issue is sensitive. Editors have an incentive to print this information
since they can hardly be held responsible for content that has been super-
vised by someone else. To quickly find a reference to Xinhua in an article,
I simply asked Yoshikoder to count the term “新华” for me. If Yoshikoder
did not find a match, the article was coded as “zero.” If there was a match,
it was coded as “one.” The number of reprinted Xinhua articles was then
divided by the total number of articles published on the same day, and each
article published on that particular day received the respective proportion
in my dataset.22
Although assessing the degree of sensitivity on an issue may be interesting in
itself, the percentage of daily Xinhua reports may also prove useful as a control
variable. When entering this measure into a regression analysis, the relation-
ships between other variables of interest can be investigated while controlling
for changes in sensitivity over time (see, for example, Stockmann, 2009).
19
Urban youth and netizens expressed a fair amount of “schadenfreude” immediately after 9/11
(Chen S., 2004; Guo, 2002). For an example of foreign correspondent reporting, see CNN,
“China Tries to Keep Tight Lid on Anti-U.S. Feeling,” September 14, 2001.
20
For example, universities were required to show a movie that portrayed the Americans in a
more sympathetic light (Chen S., 2004).
21
In June 2002, the government of the Afghan Transitional Administration was established. In
July 2002, the Chinese press started to discuss a potential American military intervention in
Iraq.
22
Using proportions may be preferable to percentages if all independent variables are coded
to run from “zero” to “one.” This allows for interpretation of the constant in regression
analysis.
120 Daniela Stockmann

In this section, I used Xinhua reports as an example to illustrate how

Yoshikoder may be a useful tool to count keywords. This function may also
be used to examine other features of news reports. For example, researchers
interested in how the Chinese media frame certain issues can identify a num-
ber of keywords associated with a particular frame. In my own work on the
framing of legal news reporting about labor issues, I grouped terms that were
associated with workers and their legal representation, such as “employee”,
“lawyer”, “plaintiff”, and so forth, and terms that stood for businesses, such
as “employer”, “factory”, “corporation”, and so forth. By counting the num-
ber of times that these terms were used in news articles, I could assess whether
news reports were written primarily from the perspective of workers or of
employers (Stockmann, 2007).
In addition to counting individual terms of groups or keywords, Yoshikoder
can also be used to count cross-references between groups of keywords. This
feature is used, for example, when assessing the tone of news reporting, which
I explain in the next section.

Assessing the Tone of News Reporting by Counting Cross-References

How can a computer pick up the tone of a news story? Many words we use
can be grouped into having positive or negative connotations. Once words are
grouped into “positive,” “negative,” and “neutral,” computers can recognize
and count these words around the specific concepts of interest. The trick, of
course, is to categorize the language in a meaningful way. Fortunately, China
specialists do not need to start from scratch. Numerous psychologists and
linguists have developed “dictionaries” to measure tone as well as other cat-
egories in text documents (Stone, 1997; Stone et al., 1966). A large number of
dictionaries are available for computer-aided content analysis in various lan-
guages.23 Up until recently, we lacked a complementary dictionary in Chinese,
but in 2005 linguists at National Taiwan University created the first collection
of positive and negative words (Ku et al., 2005).24 These terms can be entered
into the Yoshikoder software program as two separate categories and then
compared with the terms immediately surrounding other keywords. Although
a mainland Chinese version would be desirable, in practice I have found that
the results based on the Taiwanese version match my qualitative reading of
mainland Chinese texts fairly accurately.
For example, I was interested in studying the differences in legal news report-
ing related to labor issues between official and nonofficial papers. In addition

23
See, for example, the homepage of the General Inquirer at http://www.wjh.harvard.
edu/~inquirer/, accessed January 26, 2008.
24
This dictionary is composed of a translation of the General Inquirer and a collection of
colloquial terms used on the Internet. Its validity was pretested using a method to extract
positivity/negativity from radicals in Chinese characters. The dictionary can be retrieved in
traditional Chinese characters from Ku Lun-Wei at [email protected].
Information Overload? 121

to the Chinese dictionary of positive and negative valence words, I also created
groups of synonyms for all main actors involved in labor disputes, including
agents of the state (for example, “government” or “administration”), the law
(for example, “contract” or “regulation”), employee (for example, “worker”
or “migrant worker”), and employer (for example, “company” or “boss”).25
I then counted the relative frequency of positive and negative words within a
distance of eight words to these synonyms.26 Subsequently, the article’s tone
was measured by subtracting the number of negative words from the number
of positive words in this semantic space surrounding all synonyms for the
same concept.27 Somewhat surprisingly, nonofficial papers actually turn out
to be more neutral in tone than official papers. Greater balance in positive and
negative expressions create distance from overly positive reports associated
with the propaganda of official papers and aids in creating perceptions among
readers of nonofficial papers as a credible information source. The data also
revealed a stronger pro-business orientation of nonofficial papers. Yet report-
ing in favor of employers does not necessarily conflict with being in favor of
employees: nonofficial papers continuously pointed out the positive treatment
of workers when reporting positively about specific corporations; they were
pro-business and pro-worker at the same time. My findings demonstrate that
the message of the state and commercial orientation can mutually reinforce
each other (Stockmann, 2007).
Cross-referencing does not necessarily have to be used to assess the tone of
news reporting. Researchers have developed dictionaries for language associ-
ated with particular institutions, identities, and values, to name just a few
examples.28 As of 2009, only a limited number of such dictionaries existed
in Chinese, including dictionaries for sensitive keywords as well as Chinese
foreign politics (Hassid, 2007; Stockmann, 2009). If shared, the creation of
new dictionaries would add to the current applications of Yoshikoder and
therefore further improve computer-aided text analysis in Chinese studies.
The examples above illustrate that Yoshikoder is a useful tool for assessing
Chinese news content. One of the key advantages of the program is that it
increases consistency throughout the process of analyzing Chinese-language
materials. Compared to human coding Computer Aided Text Analysis (CATA)

25
These groups are called “categories” in Yoshikoder; individual keywords are called “patterns.”
For useful suggestions on developing categories, see http://www.wjh.harvard.edu/~inquirer/
developing_new_categories.htm, accessed January 26, 2008.
26
In Yoshikoder, first make a “concordance,” followed by a concordance report.
27
Since the number of matching synonyms for a concept of interest affects the number of nega-
tive and positive words found in the text, I recommend to either normalize this measure of
tone by this variable or to control for this variable in the statistical analysis. In some cases,
the length of the article (counted by the Yoshikoder) may be preferred.
28
To develop your own dictionary, first you must define words consistent with the conceptual
definition for that construct. Then you need to add variations on those root terms. Finally,
check for inappropriate variations and for words too ambiguous to be validly included. For
more examples of dictionaries, see http://www.wjh.harvard.edu/~inquirer/homecat.htm,
accessed February 20, 2008.
122 Daniela Stockmann

offers reliability and standardization. However, relying on software to analyze

texts also has disadvantages. CATA is less nuanced and flexible than human
coding as it cannot pick up contextual information: the computer locates co-
occurrences while ignoring the semantic relationships between words that co-
occur (Roberts, 1989). For example, a 1999 report in the People’s Daily states
“以美国为首的北约袭击中国大使馆，是对中国主权和民族尊严的粗暴侵犯” (The
surprise attack on the Chinese embassy in Belgrade by the United States as the
main actor is a sincere violation of China’s sovereignty and national dignity).29
In this sentence, the United States is the country taking the initiative and China
is the victim. Negative terms such as “surprise attack” and “sincere viola-
tion” describe the actions by the United States rather than by China. However,
when analyzing the tone surrounding the terms “United States” and “China,”
Yoshikoder will find one negative term (“surprise attack”) close to the United
States and one positive and three negative words (“dignity, “surprise attack,”
“sincere,” and “violation”) close to China. As a result, the tone surrounding
China will be measured as more negative than the tone describing the United
States. Therefore, CATA’s assumption that co-occurrences correspond to rela-
tionships may in individual cases inaccurately measure relationships between
different groups of words. In practice, I have found that the average tone of a
Chinese text picked up by the software program corresponds to my qualitative
reading of the text, even if my qualitative assessment of individual sentences
occasionally differs.30
In this context, it is important to keep in mind that the use of software does
not substitute for reading. When using computer-assisted content analysis it
seems as if one could simply insert a document into a machine that subse-
quently returns numbers in response to a research question. This idea seems
tempting, since mindless counting of keywords is a waste of time. Numbers
are only of substantive value when they match the concept they intend to
measure. Therefore, anyone working with Yoshikoder or any other content
analysis software should frequently check to determine whether the software
is picking up all keywords and whether important synonyms or categories
need to be added.31 And, of course, counting keywords can never substitute for
the in-depth understanding one gets about the interpretation of current events
in China from reading the news.

Content Analysis and Chinese Politics

What, if anything, can researchers of Chinese politics gain from using con-
tent analysis? This question was vividly discussed in the 1960s and 1970s.
In a study of Chinese press reporting during the Indo-Chinese border crisis

29
See “A Sincere Threat to World Peace,” People’s Daily, May 12, 1999.
30
A possible solution to this problem is semantic and network text analysis (Roberts, 2000).
31
To improve the software, please report problems to the developer of Yoshikoder (see www.
yoshikoder.org).
Information Overload? 123

of 1962, Liao and Whiting (1973: 81) propose that content analysis offered
an “appropriate method of research in which availability of data is a prob-
lem or where the investigator is heavily dependent on documentary evidence.”
They stress three key advantages: first, the method made optimal use of scarce
information; second, it provided a “check on impressionistic research” and
permitted replication by other researchers; and third, it uncovered patterns
that were not intuitively obvious and therefore provided insights on their own
(Liao and Whiting, 1973: 97). However, their claim met with strong criti-
cism. Friedman (1975: 538), not convinced that content analysis added to the
understanding of Chinese politics, retorted: “the Liao and Whiting model
ends up tied to a mechanical model which cannot handle the open, subjec-
tive, reflexive realities of politics and politicians.” In an earlier publication,
Oksenberg (1964: 605) goes further, warning of the dangers inherent in the
new methods: “Increased use of computers, quantitative content analysis, and
other advanced research techniques cannot eliminate the problems [associated
with source bias]; the danger is that they may camouflage them.” Revisiting
the topic, Walder (1979: 570) agrees that content analysis must begin with
an assessment of possible sources of bias but argues that its strength lies in
exposing “to the critical reader the set of assumptions and decisions that lead
to a particular conclusion – information that is not made explicit in other
methods.” Walder concludes the discussion on a positive note, emphasizing
transparency as a key advantage of content analysis.
Not all the points raised thirty years ago for and against content analy-
sis are still relevant today. In contrast to China experts in the 1970s, we no
longer face a shortage of data sources. The information revolution brought
about many opportunities for research on Chinese politics. Most important,
China scholars no longer need to think about how to make optimal use of
scarce resources – a question that Liao and Whiting considered as key in the
study of Chinese politics. Instead, a central question in developing a research
design today is how to narrow down the information so the research is man-
ageable but still representative of developments in China. It is therefore useful
to develop sampling techniques that produce representative results. In this
chapter, I have demonstrated that ten constructed-week samples constitute
an efficient sample size for collecting international news reports about the
United States during a period of about one year. Inferences can be drawn
from these findings on daily newspaper content for other topics if researchers
have some understanding of the nature of the variation in newspaper report-
ing. However, they cannot easily be transferred to radio or television reports.
To advance research in Chinese politics through the Chinese media, further
tests of sampling techniques are needed with respect to other issue areas as
well as other media sources. When conducting these tests, researchers can
follow the procedure outlined in this chapter, whereby I have relied on a
dataset that oversampled news reports and compared the results to a more
efficient sampling technique. More broadly, studies that investigate sampling
124 Daniela Stockmann

are important for addressing the question of how to reduce the information
available on China while simultaneously drawing accurate inferences about
China’s development.
This central question relates to Oksenberg’s worry about source biases
when using content analysis in Chinese politics. At present, such a bias is com-
mon in studies of Chinese media content. Researchers overwhelmingly rely
on official national channels and media outlets located in China’s more devel-
oped eastern regions when analyzing Chinese media content.32 These biases in
contemporary research are unrelated to research methodology. When investi-
gating Chinese news media content, scholars tend to use qualitative research
methods, and the few who rely on quantitative techniques have so far made
little use of computer software programs. Therefore, the existing bias in the
use of sources has not resulted from using quantitative research techniques or
electronic aids as Oksenberg feared. As Walder argues, source bias represents
a problem when conducting research no matter which research methodology
is employed. At the same time, Oksenberg’s observation that electronic aids
limit the range of sources that are used in practice by the researcher remains
true today. In this chapter, I have shown that electronic sources provide easy
access to media outlets from more developed regions and underrepresent
less-developed regions. Nevertheless, to date China scholars have not fully
explored the opportunities that electronic sources offer to extend knowledge
of the Chinese media. They could use electronic sources to extend knowl-
edge about magazines, television, and radio; about local media outlets; and
about the media environments beyond Beijing, Shanghai, and Guangzhou.33
Oksenberg’s point is still relevant, but considering the nature of source biases
in contemporary research, electronic aids embody more opportunities than
limitations.
Given the predominant use of qualitative interpretation of Chinese media
content, Friedman’s criticism of content analysis still resonates among many
contemporary China scholars. There is no doubt that content analysis lacks

32
Among 31 articles that systematically analyzed content of Chinese media sources, 65 per-
cent focused on newspapers, 23 percent on television programs, and only 13 percent on Web
sites. I am not aware of any works on radio broadcasting or periodicals. Among studies that
analyzed the content of newspapers, 85 percent studied newspapers that circulate nationally;
55 percent local newspapers in China’s leading newspaper markets of Beijing, Shanghai, and
Guangzhou; and only 20 percent local newspapers in other regions of China. Despite the fact
that official papers are not popular among Chinese newspaper audiences, 95 percent of all
studies that examined newspapers chose to analyze official papers, though 65 percent also
sampled non-official papers. Regarding television, all programs selected for content analysis
were broadcast on CCTV. Among four articles that examined online content, only one sam-
pled online Web sites, most of them located on the east coast. For references to these studies,
see the online appendix at www.daniestockmann.net.
33
Although transcripts of radio and television reports are available online, the field would profit
from developing electronic aids to transcribe additional audio and video material into text
files. Voice recognition software available for Chinese (such as Via Voice) may be a useful
start to build new software.
Information Overload? 125

the in-depth and detailed understanding of meaning that qualitative research

methods can extract. When interpreting texts, qualitative researchers take
into account the context of the text, including detailed information about
the source, message, channel, and audience associated with the text (see, for
example, Roberts, 1989). Content analysis loses much of this context, but it
also yields information that is often not exposed in other methods. Content
analysis encourages transparency and consistency, thus producing replica-
ble, reliable, and generalizable results (see, for example, Neuendorf, 2002).
Because of these complementary strengths and weaknesses, recent literature
recommends a mixed approach to analyzing messages, supplementing quan-
titative with qualitative research methods (Duriau, Reger, and Pfarrer, 2007).
When using computer-assisted content analysis, I have argued that the extrac-
tion of qualitative meaning should always be the standard against which to
judge the quality of the measurement. Computers are just a tool to help human
coders carry out their work more consistently and thus produce more reliable
results. Therefore, electronic aids such as Yoshikoder should be regarded more
as a helpful supplement rather than as a substitute for qualitative text analysis.
By assisting in analyzing texts in a systematic fashion and by allowing us to
include a great variety of sources, electronic aids help us gain a more complete
and accurate understanding of Chinese politics.
Part II

Qualitative Methods
7

The Worm’s-Eye View

Using Ethnography to Illuminate Labor Politics and
Institutional Change in Contemporary China

Calvin Chen

In her 1994 review of the state of the Chinese politics field, Elizabeth Perry
notes that the normalization of Sino-American relations in 1979 presented a
new generation of American political scientists with an opportunity to “pursue
a new brand of scholarship, based upon field work and documentary materials
available only in the PRC.” Moreover, given the “extraordinary diversity that
makes up the Chinese political experience” and the maturation of the field,
she argued that Chinese politics might finally “be fertile soil for the formula-
tion of new analytical approaches” (Perry, 1994b: 704, 712).
Since the publication of Perry’s article, China has not only accelerated its
economic reforms but also has inaugurated a new generation of national lead-
ers, joined the World Trade Organization, and become the world’s second larg-
est economy. These profound changes, brought about by a matrix of global,
national, and local forces, are opening up more avenues for research, dissolv-
ing and reconstituting old boundaries, and forcing us to reconsider how we
understand issues that have been central to the field of Chinese politics. For
example, how does state action impact economic and social transformation,
especially in areas where the “reach of the state” has been historically weak
(Shue, 1988)? What forms can increased political participation take and how
might they contribute to greater political stability and legitimacy? How are
market-oriented activities reshaping institutions, practices, and social struc-
tures, especially at the local level? In what sense can reemerging and evolving
social identities provide a basis for effective collective action? The Chinese
case offers a valuable perspective on these and other questions and may poten-
tially deepen our knowledge of how similar processes affect citizens in other
regions of the world.
For many political scientists, using ethnography to explore such issues
seems inadequate or inappropriate. Although Clifford Geertz (1973: 3–30)
helped make “thick description” a tolerable, even legitimate enterprise within
the social sciences, most contemporary political scientists, whether quantita-
tive or qualitative in their orientation, are nomothetic in their aspirations and
regard ideographic narratives about local contexts to be of limited value in

129
130 Calvin Chen

the search for lawlike regularities or general explanations of significant social

and political phenomena. It is true that such approaches as process-tracing
(George and Bennett, 2004) and comparative-historical analysis (Mahoney
and Rueschemeyer, 2003) do draw attention to the value of local observa-
tions in explaining outcomes in particular cases. For the most part, however,
political scientists depend on replicable methods to develop causal inferences
or parsimonious models that purport to explain uniformities or variations
across time and space. In the process, the discipline has left very little space for
efforts to develop deep understandings of the meanings that individuals and
groups attach to their practices, choices, and environments.1 Certainly, there
are inherent trade-offs in applying different methods in the social sciences, but
the discipline can only benefit from a greater appreciation of these trade-offs
and a better balance between various kinds of nomothetic and ideographic
approaches (Sil, 2004).
In contrast, many China scholars, as well as many area specialists, read-
ily appreciate the potential contributions of ethnography. In spite of this, the
applicability of ethnography has been limited at best, given that area special-
ists in political science are aware that this work cannot constitute a “test”
of a hypothesis or provide an empirical foundation for general theory. In the
case of China, there is an added incentive to eschew ethnographic scholarship
because ethnography, even in the post-Mao era, remains largely a hit-or-miss
affair. Securing approval from Chinese authorities for such study can be time-
consuming and burdensome: the process is rarely consistent or transparent.
Even when official approval is granted, a host of challenges awaits, including
scheduling conflicts, information gaps, access restrictions, and a reluctance
or fear on the part of locals to speak with outsiders. Given such problems,
the benefits of hewing to a more conventional research strategy that links
general models to measurable variables seems to outweigh the potential risks
of conducting ethnographic research. In the process, however, the prospective
intellectual gains of ethnographic study in and of itself and as a means of gath-
ering data in larger studies are lost. Consequently, many important questions
are left unexamined because they cannot be readily represented in the form of
standardized variables or easily subjected to replicable methods of empirical
analysis (Sil, 2004).
While recognizing the constraints and pitfalls of ethnographic research in
China, this chapter attempts to highlight how such research can improve our
understanding of Chinese politics. Although more variable-oriented approaches
may offer important insights, I argue that the expanded use of ethnographic
research is critical to developing a more comprehensive approach for identify-
ing and exploring different dimensions of Chinese politics and society. I sug-
gest that ethnography can, in fact, shed new light on central issues in politics
and aid the development of more empirically grounded, nuanced, and rigorous

1
Michael Burawoy makes a similar point in his review of early anthropological studies. See
Burawoy (2000: 9).
The Worm’s-Eye View 131

arguments. Although ethnography places heavy emphasis on “thick descrip-

tion,” it is important to note that in the course of depicting the complexity
surrounding particular people in a particular place and time, ethnography
also affords a stronger grasp of the multiple and sometimes hidden factors
that trigger, sustain, or obstruct change. In focusing on the specific and the
concrete, social scientists can avoid exaggerating the significance of a priori
deductive reasoning as well as the overcommitment to and misapplication of
theoretical constructs and categories.2 Indeed, I agree with James Scott when
he asserts that social scientists in general and political scientists in particular
are more likely to make lasting intellectual contributions to debates within the
field if they balance their preference for what the ancient Greeks called techne,
or settled knowledge, with a greater respect for metis, or practical knowledge,
the facts that defy easy categorization (Scott, 1998: 319–323). Paying closer
attention to outliers and anomalies would not only precipitate deeper under-
standings of local concerns and practices but also provide new opportuni-
ties to reexamine, refine, and challenge the assumptions undergirding existing
theories and concepts.
Ethnography alone cannot fully resolve these problems, but it does offer a
more holistic understanding of interactions and processes that often escape
scholarly analysis and establishes a stronger basis for understanding what
Charles Ragin terms “multiple conjunctural causation” (Ragin, 1987). In
this sense, ethnography is less at odds with current trends in political sci-
ence and China studies that stress quantitative and variable-oriented research
approaches than one might imagine; indeed, more ethnographic research can
arguably augment efforts to capture the interactions and effects of macro pro-
cesses in micro settings and vice versa. In this way, we might not only improve
our ability to test one hypothesis against alternative explanations but also
establish an epistemological “middle ground” that fosters greater dialogue
among scholars across disciplines and even advance the cumulation of knowl-
edge in the social sciences (Sil, 2000a: 166).

The Ethnographic Eye

The advantages and pitfalls of ethnographic research have been the subject
of intense debate in the social sciences. Long associated with anthropologists
and sociologists, this research approach has effectively opened up both seem-
ingly familiar neighborhoods and faraway places to scholarly analysis. As
Michael Burawoy points out, pioneering works like W. I. Thomas and Florian
Znaniecki’s The Polish Peasant in Europe and America (1918) and Bronislaw
Malinowski’s work (1922) on the Trobriand community set the stage for
the rise of major ethnographic studies that later became the hallmark of the
Chicago School (Burawoy, 2000: 7–11). These works examine how ordinary
actors perceived and responded to their world as it was beset by unprecedented

2
See Emerson, Fretz, and Shaw (1995: 111).
132 Calvin Chen

changes in patterns of economic production, social organization, and political

authority. Other studies reveal how more remote communities remained rela-
tively unaffected by such factors. All offered a vital corrective to elite-oriented
analyses by focusing considerably more attention on nonelite individuals and
groups – their thoughts, strategies, and actions. Through extensive conver-
sations with locals and the scholars’ own lived experiences as members of
the studied community, these accounts expanded scholarly understandings of
the ways in which they coped with, and even resisted, the seemingly invisible
macro forces that enveloped them.
Despite such breakthroughs, social scientists often criticize ethnography as
a research methodology for its dependence on anecdotes and impressionistic
evidence rather than on rigorous empirical analysis or parsimonious theoreti-
cal models. The seeming fixation of many an ethnographer on gathering an
ever-growing collection of fine details, whose value in generating broader gen-
eralizations about political dynamics is puzzling to some, only reinforces the
impression that ethnographers are little more than storytellers. This is exacer-
bated by popular perceptions of ethnographers as dashing figures cut from the
same cloth as the celebrated medical anthropologist Paul Farmer, who seems
to exude in equal measure daring, idealism, and toughness. Individuals like
Farmer achieved rock star status not only because of their personal courage
and uncompromising commitment, but more important, because they prac-
ticed what Farmer himself calls an “anthropology [that] concerned itself less
with measurement than with meaning” (Kidder, 2003: 72) and appealed to a
wide readership by revealing how a common humanity still exists in even the
most uncommonly wretched and forgotten places.
Other critics point out that the process of ethnographic data collection
rarely, if ever, accords with the strictures of the scientific method, which are
presumed to apply equally to quantitative and qualitative analysis (King,
Keohane, and Verba, 1994). From this perspective, ethnographers are seen
as relying too much on serendipitous encounters and personal observations,
with little concern for selection bias or sampling errors. Because ethnogra-
phers emphasize understanding how multiple factors, material and ideational
at both the individual and structural levels, combine to produce specific out-
comes in a given context, they do not focus on the causal significance of spe-
cific variables in producing a distribution of outcomes. Seeing this as a sign
that ethnographic scholarship lacks rigor, critics contend that the conclusions
drawn from these works remain at best atheoretical (and possibly subjective)
interpretations of particular circumstances. Even proponents of “analytic nar-
ratives” (Bates et al., 1998; Levi, 2004), although seemingly intent on com-
bining theoretical models and deep knowledge of particular cases, bypass
the complexities posed by the multiplicity of local narratives and ultimately
turn to the logic of strategic games for their causal analyses (Sil, 2000b: 375).
Efforts to push ethnography in a more theoretical direction have not been
helped by the poststructuralist turn in some quarters of anthropology and
sociology where the apparent adoption of a more relativistic epistemological
The Worm’s-Eye View 133

stance has called into question the possibility of value-neutral representations

by the scholar-observer in the field (Rosenau, 1992). Given these develop-
ments, it is not surprising that many political scientists are skeptical that the
often delightful anecdotes collected by their colleagues can be anything more
than just “stories.”
Such criticism is not totally unjustified. Ethnographers concentrate so much
on capturing the singular uniqueness of specific people in specific circum-
stances that they sometimes fail to bring out the full theoretical implications
embedded in their work. However, the perceived gulf between ethnogra-
phers and political scientists is often overstated. Here, it is important to note
that ethnography is not and has never been totally divorced from theory.
As Clifford Geertz points out, the ethnographer “confronts the same grand
realities that others – historians, economists, political scientists, sociologists –
confront in more fateful settings: Power, Change, Faith, Oppression, Work,
Passion, Authority, Beauty, Violence, Love, Prestige; but he confronts them
in contexts obscure enough . . . to take the capital letters off them” (Geertz,
1973: 21). To be sure, ethnographers could and should address theoretical
issues more directly and deliberately, but the larger point here is that the raw
empirical data contained within ethnographic accounts are indispensable to
the development of sophisticated explanations. Although ethnographers are
justifiably reluctant to generalize on the basis of one study, it is clear that such
studies can nevertheless ground and extend the import of theory by testing its
relevance in new or previously understudied sociopolitical contexts.
Moreover, ethnographic accounts are valuable because they often force
scholars to reevaluate developments and processes with a fresh perspective.
The anthropologist George Marcus warns against sticking too closely to
established categories and modes of reasoning, for “if there is anything left to
discover by ethnography it is relationships, connections, and indeed cultures
of connection, association, and circulation that are completely missed through
the use and naming of the object of study in terms of categories ‘natural’ to
subjects’ preexisting discourses about them” (Marcus, 1998: 16). Similarly,
when Kevin O’Brien conducted research on local people’s congresses in
China, he found that by adopting a “flexible approach to question choice,
question order, and follow-up inquiries,” it became obvious when “[a ques-
tion] had been conceptualized incorrectly or [when he] had posed a dilemma
that did not exist or missed a dilemma that did” (O’Brien, 2006: 36). This
does not imply by any means that ethnographers should or actually disregard
key concepts within their respective disciplines; neither should it lead us to
conclude that ethnographers are undisciplined in their research methodology.
If anything, this search for “cultures of connection, association, and circula-
tion” compels us first to clarify how we define and apply preexisting concepts
informing our theoretical frameworks and then to test their analytical utility
anew when new data are introduced. For O’Brien, making “mid-course cor-
rections” or changes to the original questions or concepts as a result of con-
ducting field research actually allowed him to refine concepts that extended
134 Calvin Chen

from a “Western understanding of political reality to a place where it did not

apply” (O’Brien, 2006: 30,37). Indeed, by uncovering the systemic regularities
that are intertwined with the particularities of highly diverse settings, eth-
nographic studies demonstrate how the application of an inductive approach
alongside a deductive one can enrich social science inquiry.

Ethnography and Chinese Labor Politics

When opportunities to conduct ethnographic research appeared in conjunc-
tion with China’s post-Mao reforms, scholars were eager to undertake new
research but were confronted with a daunting set of challenges. Some involved
practical concerns like securing access to field sites, dealing with possible gov-
ernment surveillance, and protecting respondents from potential harassment.
Such constraints often made it difficult for many researchers to engage in
anything more than what Thomas Gold (1989) has called “guerrilla interview-
ing.” Moreover, given China’s sheer size, ethnographers struggled mightily to
find field sites that were, on one hand, relatively open to scholarly research
and, on the other, representative of a cross-section of the country. Although
these issues persist, scholars have nevertheless managed to conduct critical
research that has filled in major gaps in how we understand major develop-
ments in contemporary Chinese society.
One of the key debates in the field is the nature and trajectory of China’s
industrial transformation and its impact on the multitude of workers who toil
on the shop floor. Some recent works on Chinese labor have been especially
illuminating in regard to the sometimes murky world of factory work. Ching
Kwan Lee’s Gender and the South China Miracle, for example, highlights
how “diverse patterns of production politics emerge as conjunctural outcomes
of the state, the labor market, and differentiated deployment of gender power”
(Lee, 1998: 160). Her work shows how global economic forces and state objec-
tives combine with such factors as native-place ties, age, and gender to shape
the authoritarian management strategies and grueling work conditions in the
two electronics plants she researched in southern China. Similarly, Pun Ngai’s
Made in China (2005) explores how female workers(打工妹) who anchor the
factory workforce defy their male superiors despite the trauma they experience
while working in an alienating, industrial setting. Both highlight the hidden
battles and behind-the-scenes politics that employees engage in on the shop
floor. Using the testimony and experiences of female workers, Lee and Pun
successfully draw out micro-level data to show conclusively that production
politics involves a more complicated, sometimes unseen set of negotiations
and processes that simultaneously takes place on individual and collective
planes. It is these processes that ultimately mold workers’ ability and willing-
ness to challenge work-related injustices and seek redress.
In comparison, my research on institutional origins and practices in rural
enterprises was conducted in two townships in Wenzhou and Jinhua prefec-
tures (Zhejiang) that were quite different in many respects from those Lee
The Worm’s-Eye View 135

and Pun studied in southern China. Nevertheless, I also came upon a com-
plex set of organizational dynamics that encompassed a sometimes confus-
ing mix of formal requirements and informal practices. In order to untangle
these strands, I first toured all of the major plants of Wenzhou-based Phoenix
and Jinhua-based Jupiter,3 shadowed key managerial staff throughout their
work day, labored alongside production workers on the assembly line, and
finally conducted multiple open-ended interviews with various personnel. In
Wenzhou, I stayed in one of the company dormitories and ate with employees
in the canteens. Over time, I became a part of the community and, as such,
was never asked for gifts or payments.4
My extended stay allowed me to ask key questions about the genesis and
implementation of company policies as well as the sources of resistance to them
and their eventual evolution. More important, the data I collected prompted
me to revise initial hypotheses regarding the linkage between financial incen-
tives and employee compliance with managerial directives as well as the enter-
prise’s overall profitability and efficiency. As a consequence of new discoveries
I made through my interactions at these companies, I began to explore the
organizational landscape differently, focusing on how official goals often
diverged from the informal norms and practices that bind all members of the
enterprise community together and how management struggles to bring both
into closer alignment.
Scholars employing variable-oriented research approaches to study similar
questions often find the incorporation of developments and data that fall out-
side the purview of their original research design a challenging task. An aston-
ishing example of this emerged during my 1997 research stint in Wenzhou.
That summer, a pair of Chinese researchers from the Chinese Academy of
Social Sciences (CASS) arrived to conduct surveys of employee attitudes and
to investigate management practices at several local firms. During their week-
long visit to Phoenix, they met with a “representative” group of managerial
staff for a few hours to discuss their work experiences and the challenges
they saw in working at a private firm. Not surprisingly, with their superiors
observing the proceedings from the corner of the conference room, every staff
member either made positive statements expressing a deep appreciation for the
opportunity to better his or herself or said nothing at all. Who in their right
minds, some participants remarked afterward, would state their true feelings
and risk jeopardizing their jobs for the sake of a survey? Individually and col-
lectively, they decided it was safer to express what they thought the research-
ers and their superiors wanted to hear.

3
Phoenix and Jupiter are fictitious names I created to protect the identities of the companies and
their employees.
4
I did give many respondents small tokens of appreciation just before I left the field. These con-
sisted of lapel pins, buttons, and school pennants and were symbolic of my appreciation and
affection for them.
136 Calvin Chen

To their credit, the CASS researchers recognized the predicament of the

employees and tried to counteract their reticence by attempting to elicit more
candid responses through anonymous surveys. What they did not realize was
that the employees saw the survey forms as an onerous burden; the partici-
pants had already spoken at earlier meetings and saw the surveys as another
chore packed onto their already heavy work schedules. Toward the end of a
lunch break one day, I overheard several members of the office in which I was
based laughing about the responses some had written. When I asked them
what was so funny, they told me that they had engaged in a bit of mischief. On
questions regarding age and gender, for example, they deliberately gave the
wrong answer. If the person was male, he instead wrote “female.” If he was
actually 25 years old, he instead wrote “52.” On other questions, they simply
didn’t respond or gave simple “yes” or “no” answers, or even worse gave
nonsensical answers. I never found out how extensive such behavior was but
there were rumors that staff in other offices responded in a similar fashion. It
seems they did so partly because they were upset about wasting their time on a
seemingly meaningless task and partly because this became a chance to relieve
some of their frustration and boredom without undermining themselves. The
CASS researchers were not responsible for such mischief but this undoubtedly
distorted the data they collected and, eventually, the findings of their study.
Although this example is undoubtedly extreme and hardly representative
of all sample and survey-oriented studies, it nevertheless highlights major
differences between more deductive and theory-driven research approaches
and ethnography. For example, social scientists aiming to develop arguments
that apply to a broad range of cases often cannot avoid making simplifying
assumptions about complex social phenomena, whereas ethnographers seek
and insist upon greater empirical validation for their assertions. Moreover,
although variables and the relationships between them are often rigorously
defined in more theory-driven studies, this type of work is less effective in
uncovering how and when specific factors contribute to the generation of a
particular result (Ragin, 1987: 54–56; Wedeen, 2004: 301). Hence, because
the CASS researchers could not fully explore the magnitude or causes of
employee mischief, the potential significance of employee frustration on their
analysis was lost.
Even though the CASS researchers undoubtedly had little time to conduct
a more comprehensive study, they remained relatively indifferent to explor-
ing possible explanations falling outside the realm of their original research
design. When, for instance, I asked the researchers whether they planned to
hold more candid conversations with employees outside of the company’s
offices or workshops, neither felt it was necessary. Instead, by retiring to their
hotel room in the evenings to analyze the data they had already gathered, they
gave up on opportunities to uncover new information through additional dis-
cussions in less formal settings like over meals or cups of tea. Their attitude
paralleled the behavior of economists, as described by Truman F. Bewley:
The Worm’s-Eye View 137

In economics there has been a tendency to suppose that you shouldn’t ask people what
they’re doing or what their own motives are because they’ll deceive you, or they won’t
know what their own motives are, they don’t understand them, they’ll exaggerate
their own roles, and glorify themselves, and so on. So what you do is put a distance
between yourself and the subject matter. This can be healthy in some circumstances,
but it can also be unhealthy. It’s like treating the economy as if it were like some dis-
tant galaxy and then you are really stuck with using only very indirect theoretical and
statistical methods to understand it. I think those are useful but I think you need to
have more immediate experience (Bewley, 2004: 382).

Put another way, sticking strictly to their original research design reduced
their engagement with “a great variety of people in different circumstances”
and made it more difficult for them to see a larger and more logically coherent
story (Bewley, 2004: 383). To be sure, this illustration is not meant to high-
light how employees at Phoenix were all disingenuous and uncooperative nor
is it intended to denigrate the serious efforts of the CASS researchers. What it
does underscore are the limitations of a research approach that, however well
conceived, is less attuned to uncovering, identifying, and investigating new
leads. By contrast, an ethnographer instinctually seeks to take into account
the unspoken and the unseen alongside the “official transcript”5 and to use
that data to weigh the significance of what had been presented in a more pub-
lic and high-risk forum.
The implication here is that devoting greater attention to understanding the
relationships in which enterprise employees are enmeshed as well how actors
define and understand the challenges before them – the process of “meaning-
making” as Lisa Wedeen (2002: 717) calls it – will greatly enrich our stud-
ies. In the previous example, employees gave positive responses because they
believed that was what the researchers expected and because they feared ret-
ribution from their superiors if they said anything that could potentially tar-
nish the company’s image. Their answers masked deep-seated resentment and
frustration with the way their superiors treated them. Again, the problem is
that given the predilection of survey research methodology for parsimony and
rigor, much of the “noise” that could be important in explaining social and
political phenomena is filtered out of the analysis before its full significance
can be ascertained. Unfortunately, the CASS study did not contain a mecha-
nism by which such factors could be incorporated and considered.
Ethnography is distinctive and valuable not only because it can uncover
hidden dynamics, but because it emphasizes “moving pictures” over “snap-
shots.” What Paul Pierson’s (2004: 2) observation means is that in studies
that are geared more toward theory-testing, political and social developments
often appear to be “frozen in time.” One-shot surveys certainly reveal a tre-
mendous amount regarding singular moments, but long-term participant

5
The term comes from the work of James Scott. The implication is that accurate reads of power
relations in any context require in-depth studies of what goes on behind the scenes, beyond the
public record or official transcript. See Scott (1990).
138 Calvin Chen

observation captures a fuller sense of process, of how we get from point A to

point B. For ethnographers, history not only matters, but, more important, is
seen as central to the development of broader and deeper understandings. In
the case of Phoenix as well as other enterprises, without past work practices
and authority orientations serving as a reference point, it becomes inordi-
nately difficult to understand why some members of the company’s manage-
rial staff became emboldened to undermine their superiors’ directives, for
such patterns are indicative of other underlying disagreements and conflicts
at work.
On this score, my research reveals that this behavior is largely connected
to changes in organizational structure and management practices, ones that
undercut a more personal and trust-based approach in favor of one that is
more impersonal, hierarchical, and routinized, that is, bureaucratic. The con-
centration of authority in the hands of a select group of managers and the
concomitant erosion of personal initiative and control among the enterprise’s
lower ranking members is a major source of friction and strain on all employ-
ees: the survey respondents’ mischief-making was a means of venting their
frustration over being saddled with what they perceived to be time-wasting
activities. It was also a way of exacting a small measure of revenge on their
superiors without having to suffer any fallout for their actions.

Ethnography, Organizational Change,

and Institution-Building
In addition to providing more accurate understandings of employee atti-
tudes and motivations, ethnographic research provides vital insights into
the nature and fault lines of organizational dynamics in China. At both
Phoenix and Jupiter, for instance, enterprise executives actually invited local
party representatives to establish branches of party-affiliated mass organiza-
tions – the trade union, the Women’s Federation, and the Communist Youth
League (CYL) – during the early to mid-1990s. Why did enterprise leaders
unexpectedly bring the party back in when the consensus in scholarly and
policy-making circles was that the party should stay out of enterprise affairs
whenever possible? Rather than simply assume that the reemergence of party-
affiliated institutions in these companies meant that politics was once again
in command, my ethnographic approach shows that the party is now in the
unusual position of supporting rather than supplanting the directives and
policies of management. Moreover, employee commitment to party ideals has
also waned considerably, further undermining a possible party resurgence.
When I attended an enterprise-wide party meeting at Jupiter in 1998, it was
clear that most of the members were uninspired by socialist pronouncements.
Despite the exhortations of party officials and company executives to work
hard for the development of the country, several young men present snick-
ered and laughed (most of the women were quiet and more respectful) and
even ridiculed some of the speakers for their verbal gaffes and doubted their
The Worm’s-Eye View 139

sincerity and commitment. At Phoenix, employees were less boisterous and

critical than their Jupiter counterparts, but they too were unmoved by the
goals and ideals articulated at party-sponsored events. In 2004, Phoenix held
a two-day retrospective commemorating the centennial of Deng Xiaoping’s
birth. Although all senior staff were present, almost no one paid close atten-
tion to the proceedings. In fact, I was amazed when two of my respondents
(one was actually a key organizer) answered my telephone calls during the
meeting in a low whisper for fear of being too disruptive. They told me after-
ward that this was regrettable but unavoidable since they were still respon-
sible for critical and pending decisions.
What participant observation and ethnography also uncovered was that
the turbulent course of rationalization and the increased number of nonlocal
migrants who dramatically altered the workforce compositions of both firms
during the late 1980s and early 1990s had precipitated unprecedented social
conflict within the enterprises. As management in both firms sought to build
upon its earlier success by routinizing and expanding operations, it undercut
prevailing work practices that emphasized personal initiative and norms of
reciprocity and trust. Employees, especially locals (本地人) who had worked
at these companies since their founding, became increasingly frustrated by
what they saw as an overly strict work regimen; they also resented heightened
job competition from migrants, both skilled and unskilled. To them, their
hard-fought gains and status were coming under siege, producing anxiety as
well as sporadic but growing resistance. Increasing class and social tensions
within the workforce were accelerating the loss of skilled personnel and a
decline in productivity. Thus, by bringing the party back into the workshops,
enterprise executives hoped to use the party’s extensive experience in organi-
zation to bridge social fissures they had failed to resolve rather than to resus-
citate the party’s preeminence within the enterprise.
Similar efforts to quell internal dissension have been documented in such
classic accounts as Richard Cyert and James March’s A Behavioral Theory
of the Firm (1992) and more recent studies like Kathleen Thelen’s How
Institutions Evolve (2004). Although neither is an ethnographic study of orga-
nizational behavior, each meticulously traces how different actors can build
coalitions to articulate and advance their own interests and those of their
allies. As new challenges and interests emerge, these coalitions may try to hold
their partnerships together through the reallocation of “slack resources”6 or
choose to remake their alliances. At Jupiter and Phoenix, executives adopted a
similar strategy with the party serving as its “front man.” They used company
profits to provide a variety of collective benefits to all employees, ranging from
the inexpensive – bars of soap, containers of cooking oil, cases of soda – to the
more costly – new clothes, day trips to local attractions, free yearly medical
examinations, and subsidized meals. All were designed to convince employees

6
“Slack resources” refers to resources that exceed what is needed to attain organizational goals.
These include but are not limited to capital and labor.
140 Calvin Chen

that life at the enterprise could be highly rewarding and would remain so
as long as all devoted themselves to becoming “good workers” (Edwards,
1977: 147–152).
To win the hearts and minds of employees, union representatives and mem-
bers of the CYL also worked tirelessly to organize a vast array of events that
entertained as well as promoted a strong company identity that superseded
workplace rivalries and native-place allegiances. Most of these took the form
of “matches of skill” (比武) and employees generally accepted these with
enthusiasm. For instance, production line workers were encouraged to take
part in contests measuring how quickly and skillfully they could assemble the
items they were charged with churning out each day. Similarly, office staff
members were challenged on their typing, accounting, and even their com-
munication skills. Later, speech contests in which employees were commonly
invited to speak about how much they had grown personally and profession-
ally since their tenure at the company began were added.
For participants and onlookers alike, speech contests can cause both anxi-
ety and excitement. For some entrants, there was enormous pressure to excel;
for audience members, office pride and prestige were on the line. One female
respondent who worked in Phoenix’s public relations office was particularly
nervous about the tryouts. If she did not advance to the final round, she felt
she was not only embarrassing herself but also letting down her office-mates.
In contrast, although the less well-spoken and less-experienced competitors
knew they did not have the necessary polish to advance, they were neverthe-
less determined to share their experiences with the judges and the audience. In
fact, they delivered some of the most memorable lines, ones that drew rousing
applause because of their earnestness and joy. A middle-aged female custo-
dian, for example, relayed how other employees treated her with great respect
and consideration despite her low status in the company. She commented: “All
I do is keep the restrooms clean – it’s not a very important job. But whenever
I run into others, they always ask me how I am doing and tell me not to tire
myself out. They are always looking out for me. No one cared about me at
any of my previous jobs.” Although other employees did not always share such
sentiments, they could still appreciate how she felt.
Even as members of the Phoenix community recognized that factory life
had many shortcomings, many also understood just how far the company had
come. These events provide employees with a venue for releasing pent-up frus-
tration and to come together without worrying about workplace rivalries or
competing social allegiances. Others feel that progress on breaking down ste-
reotypes and eliminating at least some discriminatory employment practices
has been achieved. A twenty-something female member of Phoenix’s Central
Accounting Office who is a local, stated flatly:
The company has gotten better about trusting nonlocals with important jobs – it
wasn’t always this way. In my accounting department, for example, almost one-third
of the staff are nonlocals. When I first arrived in 1996, all positions were exclusively
The Worm’s-Eye View 141

reserved for locals.. . . But now they’ve changed their views and can accept these staff-
ing changes.7

Similarly, an outsider in his early thirties who recently had risen in 1998
from quality control inspector to assistant manager of a division expressed
positive sentiments: “Locals don’t discriminate against outsiders as much as
before; in fact, most of our workers and mid-ranking personnel are outsid-
ers. The leaders have more respect for workers and pay more attention to
their needs. There is still some friction on occasion but it’s not as bad as in
the past.”8
The party also seeks to boost the morale of newer and younger workers
through their affiliates. According to the current CYL secretary,
the reason we organize events is so that employees don’t feel like there is no life beyond
work. Of course, their main reason for being here is to earn money – I understand that.
But their time here should be more than going to work in the morning and returning
home at night – what’s the point of that? . . . I always tell members that should they
leave the company, they can take these experiences and memories with them. It will
remind them of good times.9

Despite their long work days, many members remain enthusiastic about orga-
nizing and participating in CYL events. One male in his mid-twenties com-
mented that organizing events was “a kind of training” (锻炼) and a great way
to meet people. It’s an opportunity for us to develop our talents.”10 Similarly,
the Women’s Federation (妇联) representatives have concentrated on improv-
ing promotion opportunities for female employees and raising awareness
of stress induced by work and personal relationships (or lack thereof). The
Women’s Federation head, a woman in her early thirties, noted that just lis-
tening to women’s concerns is an important service because the “employee is
sometimes embarrassed or reluctant to talk, especially if the representative is
male. So often they call me directly when we talk on the phone. What we do
is listen carefully to them. After the worker is done describing her complaint,
we analyze the situation and then lay out the consequences of taking a certain
action. We try to help them find the best solution to their problem, but it’s very
difficult. There is only so much we can do.”11
These findings both complicate accepted understandings of how organiza-
tional control is established and provide a significant opportunity for revising
such theories. The developments at Phoenix reveal that this process does not
simply work in a top-down fashion, but instead it is the result of complicated
negotiations involving all segments of the enterprise community. To be sure,
employees are reluctant to use anything other than “weapons of the weak”

7
Respondent #94, personal interview, August 11, 2004.
8
Respondent #42, personal interview, August 18, 2004.
9
Respondent #128, personal interview, August 12, 2004.
10
Respondent #132, personal interview, August 17, 2004.
11
Respondent #87, personal interview, August 7, 2004.
142 Calvin Chen

(Scott, 1985) in their struggles against managerial power, a clear acknowledg-

ment of their subordinate status vis-à-vis management. Nevertheless, manage-
rial attempts to restore labor stability and social peace within the organization
represent a clear recognition that different thresholds of legitimate behavior
not only exist but that they are vital to maintaining social cohesion. In this
sense, these findings are consonant with but may also help refine the theo-
retical import of Chester Barnard’s concept of “zones of indifference” as well
the insights offered by the moral economy perspective (Barnard, 1968; Scott,
1976). They also suggest that in addition to material incentives, employees
view how the company treats them as individuals with an especially critical
eye and will constantly recalibrate their demands accordingly. That is to say,
the “zone of indifference” may not be as large as we once imagined, nor the
floor for trust and reciprocity as low as we once thought.
Indeed, informal factors – membership in kinship and native-place net-
works – have a profound influence on how authority is constructed and
perpetuated at the highest levels of the enterprises. Managers are still
inclined to select candidates with whom they are most familiar and see as
more trustworthy. As one outsider technical specialist at Phoenix suggested,
“Management is scared that if we are given more authority and responsibil-
ity, we’ll learn all of the company’s strategies and secrets. They’re afraid this
information will fall into the hands of competitors if we leave.”12 In 1998, few
outsiders served as plant vice managers at Phoenix; in fact, only one outsider
made it to the position of factory manager, and this was only by virtue of
his long-standing personal friendship with the enterprise founder. Although
a 1998 “purge” of factory-level managers removed a significant number of
distant relatives and locals from the managerial ranks, closer relations, like
the founder’s brother and brother-in-law, remain firmly ensconced in high-
ranking positions.
This situation reflects the continued influence of what Mark Granovetter
calls “strong ties” (1973: 1360–1380), which, in this case, remain nested
in the kinship and social networks that anchor the enterprise’s personnel
core. As long as this group remains on the scene, their prominence in deci-
sion making is not likely to wane soon, despite their somewhat diminished
numbers. In contrast, the persistence of informal status distinctions was
even more pronounced at Jupiter. The managers of each of the enterprise’s
most profitable subsidiaries are members of the president’s inner circle. For
example, his son and son-in-law are general managers of critical divisions as
is his niece’s husband and another close relation who once bailed him out of
trouble.13 Even at lower levels, similar patterns of domination by locals are
evident. At the factory manager rank, 142 of 158 managers hailed from the
local prefecture; of the remaining sixteen positions, nine were from areas

12
Respondent #29, personal interview, August 21, 1997.
13
Respondent #110, personal interview, June 30, 1998.
The Worm’s-Eye View 143

outside the prefecture but within Zhejiang province and seven were from
outside the province.14
These developments not only suggest that the process of establishing a
hegemonic factory regime (Burawoy, 1985) rests on decision-making power,
informal status differences, and the distribution of slack resources, but, more
important, they underscore how managers link and manipulate such factors
that can enhance or inhibit their ability to mobilize collective action and
preserve their own interests. Incorporating these insights into the fields of
Chinese politics and organization theory and institutions allows us to bet-
ter understand how the variability of these factors produces and reproduces
consequential structures that we often take for granted (Nohria and Gulati,
1994: 550–551).

Conclusion
The field of Chinese politics has made enormous strides in recent years. The
availability of new research materials, improved access to archives, as well as
new information from field studies have allowed scholars to explore and bet-
ter understand the contours and ramifications of China’s current transition
from socialism. These data have also provided an opportunity for scholars to
apply a number of different research methodologies in the hope of developing
more sophisticated theories of political behavior and change.
Although most political scientists consider ethnography to be a rather lim-
ited research approach, it offers a powerful and underutilized means of inves-
tigating and understanding larger trends and developments in Chinese social
and political life that cannot be easily detected or explained through formal
models, statistical approaches, or broad cross-national comparisons. The dis-
tinct utility of ethnographic research stems not only from its careful attention
to how specific actors think and behave in a particular place and time, but,
perhaps more important, from an appreciation of how a complex, intercon-
nected set of forces and processes collectively generates the outcomes we seek
to understand. As vitally important as untangling these strands is to deepen-
ing our knowledge of political phenomena, such efforts sometimes appear to
undermine scholarly attempts at developing more rigorous and practical expla-
nations. In focusing on particular causal mechanisms while often neglecting
others, competing research traditions a priori overlook the interrelationships
between individual-level and structural forces and how they produce conse-
quential shifts in both political and social phenomena. Understanding the
complexity of these linkages requires the use of a much wider range of lenses
and in that spirit, ethnography offers a fresh perspective and critical insights
that are not otherwise possible.
In grappling with the different meanings that individuals and groups accord
to their actions, practices, and worlds, the China field is poised to reexamine

14
These data are drawn from Jupiter’s 1997 internal reports on its workforce.
144 Calvin Chen

and revise not only the assumptions animating existing theories, but more
important, to reevaluate the accepted wisdom on how and why specific out-
comes are determined. In the process, we are restoring a sense of richness and
depth that comes from the close study of specific people in particular circum-
stances. Integrating the “extraordinary diversity that makes up the Chinese
political experience” (Perry, 1994b: 712) into scholarly debates promises to
fuel increased exchange and expansion of knowledge in the social sciences.
8

More Than an Interview, Less Than Sedaka

Studying Subtle and Hidden Politics with
Site-Intensive Methods

Benjamin L. Read

The field of Sinology has benefited greatly in the past twenty years from a
rich array of studies based on ethnography and participant observation.1
Anthropologists, not surprisingly, have led the way (Bruun, 1993; Chen,
2003; Fong, 2004; Friedman, 2006; Gladney, 1996; Hertz, 1998; Jacka, 2004;
Jankowiak, 1993; Jing, 1996; Judd, 1994; Kipnis, 1997; Litzinger, 2000; Liu,
2000; Murphy, 2002; Notar, 2006; Perkins, 2002; Pun, 2005; Rofel, 1999;
Schein, 2000; Watson and Watson, 2004; Yan, 1996, 2003; Yang, 1994; Zhang
2001). A number of sociologists also embrace these methods (Calhoun, 1994;
Chan, Madsen, and Unger, 1992, 2009; Farrer, 2002; Lee, 1998).2 Whatever
its disciplinary origin, and whether the specific topic is interpersonal relation-
ships or village governance or migrant entrepreneurs, much of this work is
strongly political in orientation. It hardly seems possible to come away from
such books without acquiring immensely valuable knowledge about the work-
ings of power in China, particularly at the local level.
Several political scientists have applied versions of these research tech-
niques in China as well.3 But these methods have a more problematic rela-
tionship with the discipline of political science. Though members of our field
have long employed ethnography and participant observation (which I group
together below under the term “site-intensive methods”), such approaches have
never been seen as mainstream. The discipline is now going through a pro-
cess of energetic debate over what constitutes important knowledge and what
methods usefully contribute to such knowledge. Practitioners of other qual-
itative methods began some time ago to weigh in with articulate statements

Parts of this chapter were published in the Fall 2006 issue of the newsletter of the APSA orga-
nized section on Qualitative Methods.
1
This is to say nothing of earlier work by Fei Xiaotong, Sidney Gamble, and others. Also, this
chapter deals only with English-language studies.
2
I owe thanks to Elizabeth J. Perry, Li Zhang, and Ethan Michelson for bringing to my atten-
tion some of the items mentioned here.
3
In addition to the works discussed later, see Blecher and Shue (1996), Friedman, Pickowicz,
and Selden (2005), Hurst (2009), Steinfeld (1998), K. Tsai (2002), (2007).

145
146 Benjamin L. Read

explaining how what they do fits into the process of building and testing polit-
ical science theory. A new generation of ethnographers and participant observ-
ers has recently emerged as well (Schatz, 2009b). As Schatz notes, members
of this group differ among themselves, with those identified as interpretivists
challenging multiple aspects of the ontology and epistemology of mainstream
political science. Others, he writes, carry out ethnographic work in a qualified
neopositivist mode, one that “uses attention to detail to generate middle-range
theories, that considers cumulative knowledge a possibility worth pursuing,
and that is optimistic about the scholar’s potential to offer contributions”
(Schatz, 2009a: 14). While I applaud the idea that politics should be studied
from many perspectives, not merely those fitting within scientific paradigms,
what I have to say here falls within the latter category. I argue that political
scientists of many stripes can and should appreciate the value of the fieldwork
techniques under discussion.
This chapter draws on examples from research on China as well as other
locations. Most of what it has to say is not specific to any one country. It
makes an argument that site-intensive methods form a particularly appropri-
ate and often necessary part of the apparatus used to study China, though
the argument pertains just as well to other non-Western and other politically
repressive settings. I wish to develop the theme that those who use ethnogra-
phy and related techniques need to build bridges with one another rather than
digging moats as is sometimes the case. This means cultivating greater aware-
ness of what specialists of other regions are doing, among other things.
The revitalization of ethnographic work in political science that I am advo-
cating – building on existing studies – has two components to it. The first con-
cerns how we, the practitioners, conceptualize our work, what we do in the
field, and how we write up and package our findings. We should strive to come
together on accounts of this methodology that can guide what we do and help
make the research process as efficient and productive as possible. The second
concerns the discipline more generally. To some extent, bringing site-intensive
methods to full fruition in political science requires efforts to shape the disci-
pline itself. It is necessary to make more clear to other political scientists how
these methods fit into the methodological world with which they are familiar.
Only then will this type of work, whether on China or anywhere else, achieve
its maximum potential.
This chapter particularly concerns trade-offs that are inherent in site-based
research, notably the balancing of breadth (studying more units, maximiz-
ing variation among them) and depth (getting the most validity, richness, and
understanding out of each unit). Although by no means uniquely so in the social
sciences, political scientists are obliged to confront issues of generalizability.
Our work is expected to speak in abstract terms to conceptual questions of
broad interest to our peers, and the explaining of variation is prized. Sometimes
the most appropriate research strategy lies at one or the other end of the spec-
trum: training microscopic scrutiny on a single locale, as in James Scott’s (1985)
famous study of the Malaysian village he called “Sedaka,” or applying thin
More Than an Interview, Less Than Sedaka 147

measures to a large number of sites or respondents, as in much survey research.

Often, though, there are reasons for pursuing a compromise approach in which
a small or medium number of units is observed in ways that obtain some frac-
tion of the depth that can be achieved in traditional ethnography. Although
there can be no formula that resolves this dilemma neatly for all projects, this
chapter attempts to conceptualize and clarify some of the factors involved.
Finally, I happen to be a committed proponent of applying multiple or
“mixed methods” to attacking empirical questions in political science. In my
own work on neighborhood organizations in China and Taiwan, for example,
I have adopted an approach that pairs the deeper insights from participant
observation and interviews with thinner but more extensive survey data. Some
of the other research considered in the pages that follow is also multimethod,
and the benefits of exploiting more than one type of information will be evi-
dent in the discussion. However, ethnographic and participant-observation
methods need to be understood and justified on their own terms as well as in
combination with other approaches.

Site-Intensive Methods
The methods that generally go under the labels ethnography and participant
observation occupy a somewhat awkward place in political science (Bayard de
Volo and Schatz, 2004). Our discipline lays claim to prominent, if rather iso-
lated, examples of scholarly work based on these methods – with perhaps the
two most widely read being Fenno’s Home Style (1978) and Scott’s Weapons
of the Weak (1985). A subset of empirical researchers has always been drawn
to them, going back at least as far as the immediate post–World War II gen-
eration and presumably earlier (Banfield, 1958). They are discussed on the
occasional conference panel. They are actively employed in much exciting
research today, by themselves or in conjunction with other methods (non-
China examples include Adams, 2003; Allina-Pisano, 2004; Bayard de Volo,
2001; Cammett, 2005, 2007; Galvan, 2004; MacLean, 2004, 2010; Roitman,
2004; Schatz, 2004; Straus, 2006).
And yet, these methods remain marginal. I think it is fair to say that they
only occasionally crop up in methodology curricula. Even within the world
of the American Political Science Association’s organized sections on qualita-
tive and mixed methods and the stimulating ferment that has been fostered
in recent years, they have shown up so far as a distant cousin. Very little in
this vein appeared on the 2006 Institute for Qualitative Research Methods
(IQRM) syllabus, for example. Two of the most important recent books on
qualitative methods, although immensely useful to ethnographers and par-
ticipant observers, also seem to have been written without these approaches
particularly in mind (George and Bennett, 2004; Brady and Collier, 2004).
What are the reasons for this marginality? Let me first mention a few obvi-
ous ones. Acquiring the skills needed to use these methods and then applying
them is time-consuming and costly, particularly when research is conducted
148 Benjamin L. Read

in foreign-language settings. They are thus rather difficult to recommend in

good conscience to the average graduate student who is under pressure to
minimize time to completion. Also, other methods skills are much more in
demand within the discipline as a whole.
But clearly there are other reasons as well. Practitioners are split by apparent
differences in epistemology, such as the previously mentioned divide between
“interpretivists” and “positivists,” which looms large in some recent accounts
(Burawoy, 1998; Schatz, 2006; Yanow, 2003). There is also a hesitance on
the part of many to go beyond describing what they did in their own research
to suggesting sets of general procedures for others to follow. Efforts within
political science to spell out the benefits of these methods for building and
testing theory have so far been limited. Finally, and perhaps most important,
there has been no real push to build a coalition behind the critical apprecia-
tion, application, and teaching of ethnography and participant observation.
I believe an appropriate step forward would be for researchers who use
these methods – either exclusively or in combination with other methods –
to work toward building a coalition or users group within the discipline.
This enterprise would start by identifying common ground in a related set of
approaches to the gathering of sources, evidence, and data. This would, one
hopes, cross-cut and set aside underlying epistemological divides.
An umbrella term bringing together a set of related methods might be use-
ful. One such term would be site-intensive methods (SIMS), referring to the
collection of evidence from human subjects within their ordinary settings,
where their interaction with the surroundings informs the study just as the
researcher’s own questioning does. This implies the need for a deeper engage-
ment with a site, context, locality, or set of informants than is obtained
from, for instance, telephone surveys or some types of one-time interviews.4
This term would subsume most of what is referred to as “ethnography” and
“participant observation,” and perhaps some forms of other practices, such
as focus groups. It would also highlight the diverse forms that this research
takes, including studies that strive for a high degree of depth in a single locale
along with those that aim for breadth as well, and projects in which SIM is the
main course, so to speak, as well as others where it is more of a side dish.
I somewhat reluctantly suggest some new term like SIMS in part because I
am not convinced that other categories are up to the task of bringing together
the most useful and productive coalition of researchers within political sci-
ence. “Ethnography” on its own has the advantage of a long pedigree within
the social sciences and a voluminous methodological literature, especially
in anthropology. But it may have drawbacks as well. Some versions of the
anthropological model may set the bar too high in implying that months or
years of immersion are required to obtain insights. It may connote a holistic
orientation according to which the entirety of a community or locale must be

4
In fact, some kinds of surveys and interviews require considerable stage-setting and trust-
building; see Posner (2004).
More Than an Interview, Less Than Sedaka 149

comprehended to make sense of any one part. It should be pointed out that
the meaning and practices of ethnography today are undergoing evolution
and sharp debate; by no means do all practitioners see ethnography as limited
to these forms.5 Nonetheless, the term seems to me problematic as applied to
projects employing shorter stints of fieldwork and less encompassing modes of
information gathering.
“Participant observation” would also seem to be a possible umbrella term.
Yet here too, the boundaries implied by this concept may not be coterminous
with what it is we want to bring together. On the one hand, the researcher
“participates” in other kinds of research, such as straight-up interviews.
Conversely, to some it may imply that only through a long-term process in
which the researcher becomes a part of community life can full or meaningful
participation be achieved. Some may feel that this covers only a subset of the
practices comprised by ethnography whereas others see the two concepts as
interchangeable.
As noted earlier in this chapter, two studies, by Fenno (1978) and Scott
(1985), may be the most widely read examples of site-intensive methods in politi-
cal science. Reflecting on these books in particular provides an opportunity to
consider what it might take to promote an initiative within the discipline that
incorporates and promotes both types of research. Viewed from one perspec-
tive, they can be seen as strongly contrasting, perhaps almost polar opposites
in their approach. Scott describes his project as a “close-to-the-ground, fine-
grained account of class relations” (Scott, 1985: 41) in a Malaysian village,
population 360, which goes under the pseudonym of Sedaka. Situating his
methods in the ethnographic tradition of anthropologists (Scott, 1985: xviii,
46), Scott states that he spent at least fourteen months in Sedaka, interview-
ing, observing, and taking part in village life.
Fenno’s (1978) work was motivated by questions concerning the relation-
ship between politicians and those they claim to represent. “What does an
elected representative see when he or she sees a constituency? And, as a natu-
ral follow-up: What consequences do these perceptions have for his or her
behavior?” (Fenno, 1978: xiii). His approach was to spend time in the com-
pany of members of the U.S. House of Representatives in their home districts.
He famously characterizes his research method as “largely one of soaking and
poking – or just hanging around,” and situates it explicitly within the tradi-
tion of participant observation as practiced by sociologists and other political
scientists; ethnography is not mentioned, as far as I can tell (Fenno, 1978: xiv,
249, 295). In the text of the book and its long methodological appendix, Fenno
candidly and rather self-deprecatingly explains his modus operandi of accom-
panying politicians wherever they would let him tag along, building rapport,
recording their remarks, and asking questions when possible.6 Clearly this was
5
Marcus (1998), for example, provides arguments in support of multisited studies, albeit with
reservations.
6
He also discusses his methods in a 1986 APSR article and other essays, all reprinted in Fenno
(1990).
150 Benjamin L. Read

a far “thinner” form of engagement with a research milieu than was Scott’s
village study. Relative to a single-site project, Fenno traded depth for breadth,
studying eighteen different representatives and thus obtaining substantial
variation on characteristics such as party affiliation and seniority (Fenno,
1978: 253–254). The total time he spent with each representative ranged from
three working days to eleven, averaging six (Fenno,1978: xiv, 256), and on
some of those days the research subject was available only part of the time.7
Despite their differences, these books can readily be seen as belonging to
a common category. Both scholars were propelled into the field by strongly
theoretical motivations – theories of hegemony and false consciousness in
one case, and theories of representation in the other. In both instances, the
researchers identified an empirical subject of key importance where (at least
as they portrayed it) existing accounts relied on assumptions that needed to be
tested or fleshed out through on-the-ground study.

When Should Researchers Use Site-Intensive Methods?

What are the circumstances under which these approaches are particularly
valuable – sometimes even necessary – for political research? One way to view
it is that such methods are especially valuable when what we are studying is
subtle (for example, relationships, networks, identities, styles, beliefs, or modes
of action),8 and when what we are studying is hidden, sensitive, or otherwise
kept behind barriers that require building trust, waiting to observe unguarded
moments, or otherwise unlocking access.9 Scott’s and Fenno’s projects serve
again as examples. In both books, to simplify somewhat, the fundamental
subject of study was individuals’ perceptions (peasants’ views of class relations
and politicians’ views of their constituents). These perceptions are both subtle
and, most of the time, hidden. This fact made forms of research like surveys
and short interviews unworkable and necessitated strategies involving trust-
building and over-time observation.
Subtlety, as considered here, is in part a relative concept. The ways in which
site-intensive methods are best applied to a problem may change significantly
depending on the existing state of theory. The success of a project like Scott’s,
in the way that he framed it, hinged on the scholarly readership accepting false
consciousness as a viable explanation or foil. How a specific application of
site-intensive methods is pitched depends in part on the sophistication of exist-
ing hypotheses and data. If everyone agrees that X 4 has a positive relationship
with Y when X1, X2, and X3 are controlled for, and the debate is over the

7
One could go on about the contrasts. Scott is, of course, particularly attuned to the voices and
experiences of the subaltern, whereas Fenno does not conceal an often admiring sympathy for
the elites whom he studies.
8
In fact, most concepts I can think of in the realm of politics contain subtleties that probably
deserve up-close scrutiny. At the same time, this is not at all to deny that less-intensive meth-
ods can also shed light on these topics.
9
On hidden aspects of politics, see also Scott (1990) and Kuran (1995).
More Than an Interview, Less Than Sedaka 151

magnitude of the coefficient, whether the relationship is linear or quadratic,

and whether an interaction term with X5 is required, then merely observing
that X 4 and Y appear linked does not provide as much added value as it would
have earlier in the research cycle. But by no means are site-intensive methods
appropriate only for little-studied topics or when existing theories are crude.
To continue the above analogy, up-close fieldwork could still usefully point
out that X 4 is more complex than we thought it was and requires new mea-
sures; that, in real life, the causality seems in fact to run from Y to X 4; or
that the relationship is conditioned by other factors that previously had been
ignored entirely.
This leads to one answer – in the affirmative – to the question of whether
such methods have special applicability in a country like China. It would be a
mistake to argue that sociopolitical phenomena in China are somehow “more
subtle” than those elsewhere. But one might point out that China has long
produced events and things that, at least initially, either confounded or else
fit only problematically within the social science categories created in North
America and Western Europe. A list would include the CCP’s rise to power;
the Cultural Revolution; local associations; patterns of interest representa-
tion; modes of contention and grievance articulation; rapid economic devel-
opment under Communist Party rule; the transition from socialism; and even
the PRC’s current regime type. The question, “What is this phenomenon I
am looking at in China a case of?” is not always easily answered.10 This has
created stumbling blocks but also opportunities for innovative contributions.
Thus, as in the study of other non-Western systems, empirical phenomena
in China need to be analyzed carefully before determining their relationship
with existing concepts, and site-intensive methods are well suited for this.
The other part of this answer concerns “hidden” subjects. It should be
emphasized again that even in open, liberal political systems, certain impor-
tant topics remain hidden, such as a politician’s true perceptions of his or her
constituents. In China, barriers created by state controls lead, in some circum-
stances, to obstacles to access and concerns about the validity of the data that
are obtained. China today is not as open as, say, Taiwan, nor as closed as, say,
North Korea or Myanmar. Particular institutions or individuals vary greatly
in how available they are. Occasionally, one can obtain what is needed from a
government agency just by walking through the front door. For some topics, a
short phone or questionnaire survey conducted by a stranger may be adequate.
But a common configuration is one of formal barriers and informal porous-
ness. In these circumstances, valid or usefully detailed information cannot
be acquired in a single visit or conversation but can be obtained through a
more patient approach. The proportion of settings in which relatively intensive
strategies are required to obtain any valid information is large in China com-
pared to countries with more open political institutions and stronger norms

10
I thank Laura Stoker of the University of California, Berkeley, for insisting that students ask
themselves the “What is this a case of” question.
152 Benjamin L. Read

of transparency and disclosure. Conversely, the rewards to approaches that

involve trust-building and over-time information gathering are correspond-
ingly large there and in other closed or semiclosed political systems.
For these reasons, certain kinds of projects require extensive cultivation of
research sites whether or not they involve traditional aspects of ethnography,
such as direct observation of human behavior. For example, Melani Cammett,
in her project on Hezbollah in Lebanon, went through an elaborate process of
establishing links with the organization in order to lay the groundwork for a
series of interviews with its members.11 Relatedly, studies like Whiting (2001)
and Remick (2004), although not ostensibly ethnographic in nature, clearly
relied on building rapport and networks within the two or three localities
upon which they were based.
In my own work, I have used site-based methods as one way to study the
state-fostered neighborhood organizations of China and Taiwan. These are
known as Residents’ Committees (RC) or Community Residents’ Committees
(社区居民委员会) in the former and neighborhood heads (里长) and block cap-
tains (邻长) in the latter. My research in China took place in seven different
cities but focused on the city of Beijing. My goal there was to study several
neighborhoods in different parts of the city and in a variety of socioeconomic
settings. I wanted to choose my sites more or less at random. I also wished
to avoid the “model” neighborhoods that each district establishes; in major
cities, local and international visitors alike are routinely shepherded to such
showcases. Equipped with a letter of introduction from my “host unit,” the
Chinese Academy of Social Sciences, but otherwise without special permis-
sion, I knocked on the doors of RC offices, identified myself, described my
research, and asked to come in and chat.12 In cases where the committees were
receptive, I made repeated follow-up visits and eventually established a set of
ten neighborhood sites in this fashion.13 I spent about thirty half-days (morn-
ing or afternoon) with two of the committees, and made twelve to fifteen trips
to most of the others over fourteen months of dissertation fieldwork (1999–
2000) and during three shorter follow-up stays (2003–2007).14 On each visit,
I sat in the offices, chatted with the staff, and watched and listened as they
went about their work.
The most fundamental purpose of my research was to understand the power
dynamics of neighborhood communities and the kinds of relationships that

11
As of June 2009, papers from this project were under review but not yet published.
12
In two cases, initial contact was made informally through acquaintances, and in two other
cases I took a more official route involving a personal contact at a Street Office but also obtain-
ing formal permission via the city government and my host unit’s foreign affairs office.
13
In five neighborhoods I was turned away, either by the RC itself or by government officials.
In two of these cases I was accompanied by a police officer to the local police station (派出所)
and told not to return to the neighborhood. In two of the ten sites, after six successful visits
the committees were told by the authorities to discontinue contact with me.
14
In three of the ongoing sites I offered to teach English to staff members or their children, and
for several months I ran a weekly language class in my apartment.
More Than an Interview, Less Than Sedaka 153

state-sponsored organizations at this ultra-local level have with the residents

they notionally represent. These topics are subtle in that they feature complex-
ity and nuance, and in that they vary across places, people, and situations.
They are hidden, or at least potentially hidden, in that RCs or constituents
might not speak frankly to strangers about these relationships, and in that
interaction between them generally leaves no publicly available record. Direct
observation does not automatically overcome these challenges. But by build-
ing familiarity and trust, by probing gently, and by happening to be present
as the committees dealt with all manner of situations and problems, I could
learn much about this point of contact between state power and urban society.
The RCs thus exemplify the pattern described above: there are formal impedi-
ments to studying them, but it is often (though not always) possible to obtain
entrée, especially via informal channels.

How Do Site-Intensive Methods

Contribute to Theory?
In some circumstances, as the previous discussion indicates, gathering any
valid data at all requires building ties of trust and familiarity with one’s
research subjects and the institutions in which they are embedded. This alone
is a powerful rationale for site-intensive methods. But apart from this, it may
be constructive to spell out at least four ways in which ethnographic work
contributes to the building of social science concepts and theory.
First, ethnographic work can inductively generate new hypotheses. For
Fenno (1978), the choice to conduct his field research by “soaking and pok-
ing” was in part driven by his conceptualizing the project as breaking trail
through wholly uncharted territory. He argues that political scientists had
previously all but ignored representatives’ understandings of their districts
and how they behaved there (Fenno, 1978: xiii). Given this theoretical tabula
rasa, a “totally open-ended and exploratory” approach was needed:
I tried to observe and inquire into anything and everything these members did. I wor-
ried about whatever they worried about. Rather than assume that I already knew what
was interesting, I remained prepared to find interesting questions emerging in the
course of the experience. (Fenno, 1978: xiv)

Whether it was absolutely necessary for Fenno to cast the project in this way
seems open to question. In the course of acknowledging previous research, he
states that “political science studies conducted in congressional constituencies
have been few and far between,” but goes on to cite no fewer than eight books
and articles on this topic as just the ones that have been “the most helpful to
me” (Fenno, 1978: xvi, n. 9). Moreover, I fear that Fenno’s heavy emphasis on
the unstructured nature of his project may have helped lead many members
of the discipline to believe that participant observation is necessarily specula-
tive and free-form, useful, at best, only for the earliest and most preliminary
phases of a research cycle.
154 Benjamin L. Read

Nonetheless, this open-ended approach convincingly results in insights that

can be taken as new hypotheses concerning politicians and voters’ behavior.
Though Fenno soft-pedals them somewhat, the hypotheses that he puts for-
ward include these: House members in the early phases of their careers are
most attentive to their districts; members focus most on constituents who are
most organized and thus accessible in groups; and constituents care about
personal attention, respect, and the assurance of access at least as much as
they care about the representative’s congruence with their own issue positions
(Fenno, 1978: 215, 235, 240–242). His book thus provides a good example of
the value of participant observation for inductively generating new theoretical
claims.
Some of Calvin Chen’s findings from his extensive ethnographic studies
of Zhejiang enterprises can also be understood as falling into this category.
As his chapter in this volume reports, among other things he discovered that
Communist Party institutions were being established in these firms. As Chen
indicates, without the benefit of the “worm’s-eye view,” it would be easy to
misinterpret this as the authorities imposing their organizational hegemony, or
else as the companies going through the motions of compliance in order to get
a political pass. Instead, a surprising third possibility emerges: that the firms
actively encourage this development in order to enlist the party’s assistance in
resolving social tensions among the workforce (C. Chen, 2006, 2008).
Second, site-intensive methods can also be used to test hypotheses by
gathering qualitative measures of their observable implications. This for-
mulation derives from King, Keohane, and Verba’s Designing Social Inquiry
(1994: 28–31). The notion that qualitative researchers can maximize the lever-
age they apply to the testing of hypotheses by identifying and evaluating as
many as possible of their “observable implications,” of which even a single
case may have many stands, as one of the more constructive and less contro-
versial messages of this book. Naturally, this echoes long-standing practices
of case-based researchers.
The fine-grained accounts of villagers’ narratives, rituals, insults, and
struggles in Scott’s Weapons of the Weak (1985) can be read as a series of tests
of the “false consciousness” hypothesis. The village ne’er-do-well, Razak, pro-
vides one: “As a beneficiary of local patronage and charity, however reluc-
tantly given, one might expect Razak to entertain a favorable opinion of his
‘social betters’ in the village. He did not.. . . ‘They call us to catch their (run-
away) water buffalo or to help move their houses, but they don’t call us for
their feasts . . . the rich are arrogant’” (Scott, 1985: 12). Others appear in per-
ceptions of local misers (Scott, 1985: 13–22), the celebration that results when
a job-displacing combine bogs down in the mud (Scott, 1985: 163), and the
conflict over the village gate that had protected villagers’ paddy-hauling jobs
(Scott, 1985: 212–220).
In my Beijing research, I used evidence gathered from the neighborhood
sites to test several hypotheses about relations between constituents and their
Residents’ Committees. One such hypothesis concerned clientelism. In China
More Than an Interview, Less Than Sedaka 155

and around the world, relationships between ordinary citizens and power-
holders can take the form of patron-client ties, typically involving exchanges
in which subordinates provide political support while superiors provide mate-
rial goods and opportunities such as jobs. But my RC office visits showed that
the committees’ activists and supporters obtained few or no tangible rewards
for helping to keep watch over the neighborhood and conveying information
to and from the committees. They often received only token gifts such as tow-
els and bars of soap at annual parties. The committees, closely managed by
the Street Offices, had little latitude to channel state benefits toward them, and
certainly not career opportunities. Instead, the activists’ motivations proved
to be similar to those of volunteers everywhere: a sense of pride and impor-
tance, and the pleasures of sociability. Private interviews with urban residents
and quantitative surveys helped to reinforce this conclusion, but participant
observation was crucial. Only by observing the activists day after day, by lis-
tening to them converse with the RC staff, by talking about these relationships
with committee members who were willing to speak frankly about their work,
and by acquiring a firsthand sense of just what material resources the com-
mittees control (and how limited these resources are) did I become convinced
that the clientelism hypothesis is not the key to understanding this aspect of
the state-society relationship.
Although the words, attitudes, and behavior of local interlocutors constitute
one form of evidence – essentially, data points – that can be gathered to test
hypotheses, this evidence can also take a different form, that of causal process
observations. Seawright and Collier define such an observation as “an insight
or piece of data that provides information about context, process, or mecha-
nism, and that contributes distinctive leverage in causal inference” (Brady and
Collier, 2004: 277). The search for such observations is essentially the same
as what is meant by “process tracing.” In political science, some of the most
prominent methodological discussions of process-related evidence have taken
place in the context of case-based research using historical sources.15 The logic
is the same for ethnography and participant observation. Indeed, site-intensive
methods may be even better suited for gathering causal process observations,
as they allow for active probing by the researcher rather than relying on the
passive analysis of extant sources.
Lily L. Tsai’s (2002, 2007a) multimethod research on conditions influenc-
ing public goods provision in Chinese villages provides outstanding examples
of causal process observations derived through participant observation. In this
project, much of the analytic power comes through quantitative evidence, in
the form of a survey of 316 villages in four provinces, as well as a survey of vil-
lagers. Regression models estimated on these data offer evidence that, ceteris
paribus, villages that possessed certain kinds of social institutions – notably,

15
Chapter 10 of George and Bennett (2004) gives an overview of process-tracing with an eye
toward research based on case studies drawing on historical documents. See also Sidney
Tarrow’s chapter in Brady and Collier (2004).
156 Benjamin L. Read

temple associations and villagewide lineage groups – also tended to provide

good roads and schools. Tsai further employs a series of focused qualitative
comparisons among at least nine villages to elucidate the causal link between
these social institutions, termed “solidary groups,” and the outcomes of inter-
est. Each village case study is derived from visits of between two and twenty
days and is presented in the space of several pages. Together they offer vivid
illustrations of the causal processes that are in play. In Li Settlement, for exam-
ple, leaders rallied their constituents on the basis of a common lineage and
community spirit in collecting donations to pave a road; in Pan Settlement,
by contrast, long-standing conflict among the sublineages impeded similar
efforts.
A fourth way in which site-intensive methods make contributions is by
creating, correcting, and refining theoretical concepts and the categories
within which they are understood, measured, and analyzed. As Henry E.
Brady points out in a defense of qualitative methods in political science more
generally:
Concept formation, measurement, and measurement validity are important in almost
all research and possibly of paramount importance in qualitative research. Certainly
notions such as “civil society,” “deterrence,” “democracy,” “nationalism,” “material
capacity,” “corporatism,” “group-think,” and “credibility” pose extraordinary con-
ceptual problems just as “heat,” “motion,” and “matter” did for the ancients. (Brady,
2004: 62)

Within the China field, the work of Kevin J. O’Brien and Lianjiang Li on
“rightful resistance” springs to mind as embodying the importance of pro-
tracted, up-close study for concept formation (O’Brien, 1996; O’Brien and Li,
2006). The authors note that their work to date leaves a number of avenues
open for later exploration, such as explaining when governments react nega-
tively rather than positively to these particular calls for justice and explor-
ing regional variations (O’Brien and Li, 2006: 114). The project’s major éclat
stems from its pioneering elaboration of an important category of conten-
tious collective action, with wide application both in China and elsewhere,
along with detailing the causes behind it and the dynamics through which
it plays out in interaction with successive levels of government. Although a
variety of sources are exploited, not all of which are ethnographic, the authors
make clear that a relatively small number of village sites in which extensive
access and deep background were available played a crucial role in launching
and guiding the study (O’Brien and Li, 2006: xi–xvii, 131–133, 139–141, and
­personal communication with O’Brien).

Trade-Offs: Depth versus Breadth

Would Weapons of the Weak (Scott, 1985) have been more compelling (or less)
if it had involved multiple village sites? What if Fenno (1978) had tracked just
four politicians, but spent months with each instead of days?
More Than an Interview, Less Than Sedaka 157

All research involves trade-offs. In ethnographic and participant-

­ bservation studies, by their very nature, the researcher is choosing to invest
o
time in observing and developing relationships with particular sites and infor-
mants. Given a finite amount of time and other resources for the project, there
is thus an inherent trade-off between depth and breadth: between working
more intensively at a smaller number of sites (or a single one) and developing
a larger number of sites but spending less time at each. How many sites, then,
should be developed in a given project?16

Site-Level Variation
The answer to this question depends in part on the number of site-level indepen-
dent variables and the nature of variation within them. Scott discusses reasons
he chose Sedaka: its apparent typicality as well as the fact that previous stud-
ies had established a baseline from which to assess change (Scott, 1985: 90).
He does not seem to justify in an explicit way his decision to focus on just
one village.17 Implicitly, the reasons seem to be that long-term immersion in
a single place was required to obtain access to “hidden transcripts,” and that
variation within the village itself (between elites and poor, and among differ-
ent informants) was a more important focus than, say, interregional variation
between Kedah and Johore. The overall framing of the research also seems to
discourage the exploring of different sites: the denizens of Sedaka are intended
to speak for peasants everywhere.
In the case of Home Style, Fenno (1978) adopts a casual tone toward the
choice of his eighteen congressional representatives (“I make no pretense at
having a group that can be called representative, much less a sample” [Fenno,
1978: xiv]). But it is a sample, of course, and he makes clear that he selected it
in such a way as to observe members and districts of different parties, regions,
races, ages, levels of seniority, and electoral competitiveness. Thus, he pursued
the familiar small-n strategy of obtaining variation on a number of potentially
important independent variables. The problem of small sample size remained,
but Fenno’s logic is clear.

Up-Front Costs of Site Development

Each site involves what might be called start-up or up-front costs, referring
to expenditures of time and effort as well as other costs. What is the mini-
mum, or optimal, degree of access required to obtain valid information for a

16
An essay by Maria Heimer (2006) contains a useful discussion of the choices she made in her
work on local cadres, traveling to twelve counties and pursuing a tightly focused set of ques-
tions in each. More generally, the volume in which this appears, Doing Fieldwork in China,
will be of interest to many readers of this book.
17
To be more precise: the study is hardly limited just to Sedaka; it brings to bear evidence
concerning the region as a whole, and Scott mentions excursions to nearby villages. Still, it is
framed as a single-village study.
158 Benjamin L. Read

particular research effort? In my encounters with the Residents’ Committees

of Beijing and other mainland cities, in most cases it appeared that at least
three or four visits were required to begin to cross a relaxation threshold.
During the initial visits, the atmosphere was at least a little tense and con-
versation stilted. Obtaining answers to straightforward factual questions –
what’s the population of the neighborhood? how many low-income residents
are there? and so forth – posed little problem, but open-ended questions typi-
cally resulted in guarded, bland responses. In most although not all of the
sites, this wariness gradually gave way to a much more accepting and, in some
cases, even welcoming atmosphere. Thus the early time investment paid divi-
dends by allowing me to observe something closer to the committees’ ordinary
day-to-day work and interaction with constituents. For present purposes, the
point is that the greater the costs and risks of establishing each site, the fewer
sites a researcher can afford to develop given his or her limited resources.

The Nature of the Data Sought

This is, of course, typical for ethnographic research, but what is particularly
important to point out is that the amount of time required for building trust
and rapport (in this and other settings) depends on the type of information
needed. If the point were merely to ascertain basic facts of the sort that the
RCs themselves regularly collect and often post on their office walls, such as
the number of women of child-bearing age within the jurisdiction, this could
probably have been accomplished in a single visit, depending on the creden-
tials one could present. To understand something partially hidden but limited
in scope – for instance, the actual and highly state-managed process through
which RC elections are run – would have required just a few visits.
On the other end of the scale, to aspire to write a comprehensive “neigh-
borhood study” on the model of the best urban anthropological work (e.g.,
Bestor, 1989) would have required drastically reducing the number of cases. It
would have meant applying much more comprehensive and sustained efforts
than I brought to bear at even my most intensive sites. It also would have
required a strategy of regularly going out within the same neighborhood and
talking to a wide sample of residents, in private and independent of their inter-
actions with the RC itself. This would have been possible in my project, but
might have threatened the premises of my tacit understandings with the com-
mittees. Instead, I chose to do my citizen interviews and questionnaire surveys
(designed to get at residents’ opinions in a private way, removed from the RC
members’ presence) in neighborhoods other than those where I was conduct-
ing participant observation.

Optimizing Validity
One question to consider in choosing a research strategy is whether the time
and effort involved in ethnographic work are justified relative to what can
More Than an Interview, Less Than Sedaka 159

be obtained through one-time interviews. For many purposes and in many

settings, the essential things that the researcher needs to understand can be
ascertained in a single session. By minimizing the time spent with each sub-
ject, the number of interviewees can be maximized, thus potentially obtaining
high degrees of variation and some or all of the benefits of large-n research.
Even to many qualitative researchers, the idea of returning to an informant for
a second or a tenth encounter may seem like a puzzling waste of time.
Some of the reasons one might want to return to a site multiple times have
already been made clear: because trust and familiarity are required to obtain
information. But there are other reasons as well. First, even the most coopera-
tive and forthcoming of informants may “change their stories” over the course
of a single interview and between interviews. It is well established in the world
of survey research that different question wordings, question orderings, and
contexts can lead to quite dissimilar responses. Similarly, interviewees can
reveal different and possibly conflicting sides to their beliefs and experiences
on different occasions. Moreover, if the research site is such that multiple
informants are present at any given time, their relationships with one another
may strongly condition what is revealed. If the point of the in-depth research is
to obtain the truest possible measure of an individual’s perspective, the extra
validity that can be gained through multiple sessions must be considered.
This is all the more evident when the project relies on critical informants,
particularly when their memories must be plumbed for recollections of events
that took place years or decades earlier. Peter Seybolt’s Throwing the Emperor
from His Horse (1996) serves as an example. Seybolt constructed a political
history of an impoverished Henan village through a series of long interviews
with Wang Fucheng, who served as its party secretary from 1954 to 1984. The
interviews were clustered in four visits, with the first in 1987 and the last in
1994. Several other villagers also provided information, but Wang was central.
As the author describes it, on each visit the relationship between Seybolt and
Wang (and between their families) deepened, and further layers of memory
and nuance were added to the account. This book simply could not have been
written on the basis of the author’s first visit alone.
A final circumstance requiring multiple or protracted encounters occurs
when the project relies significantly on watching the subjects interact with
their surroundings – for instance, by observing events that crop up in the
course of the visit. Fenno’s (1978) research drew heavily on this practice. He
points out that his project would not have been successful had it taken the
more conventional form of forty-five-minute audiences in the congressional
representatives’ Capitol Hill offices. He depended on being at his subjects’
side as they careened from one home-district event to the next, on noting
their actual behavior and demeanor at these events, and on catching them in
reflective moments as they unwound after meetings with one or another part
of their constituencies.
Residents’ Committee offices present broadly similar opportunities in
that they are sites of frequent and highly varied interactions among state
160 Benjamin L. Read

intermediaries (the committee staff), residents, and higher state officials. Some
days and locales are busier than others, but often a great deal of activity goes
on in these offices, and much of it is unpredictable from the RC members’ per-
spective and thus impossible for them to plan or script in advance. Residents
and transients drop in for an astonishingly vast assortment of reasons: com-
plaining about noisy neighbors; requesting help in applying for state benefits
or documents; paying small fees; negotiating the use of sheds, spare rooms, or
open space for commercial purposes; demanding that the committee resolve
whatever pressing problem they face, whether concerning housing mainte-
nance, part-time employment, match-making, or any number of other things.
At the same time, officials from the Street Office regularly stop by to explain
the latest tasks assigned to the RC and check on its performance. Police offi-
cers, particularly the “beat cop” (片儿警) assigned to the neighborhood, regu-
larly pull up a chair, sip tea, and have a cigarette while conveying or receiving
information about burglaries, parolees, recidivists, or dissidents.
Firsthand observation provides an opportunity to watch how residents and
committee members handle all these situations. Naturally, it is always possible
that the observer’s presence affects the behavior under observation, and thus
careful judgment and honest evaluation in one’s field notes are required. This
also underscores the importance of establishing trust and familiarity, as men-
tioned earlier. By taking note of body language (imperious or supplicating, for
instance) and listening not just to what is said but also to the tone of voice in
these interactions, one can infer a great deal about power relations between
citizens of many kinds and local agents of the state. Also, such observation
allows for follow-up questioning after the fact on the basis of details gleaned
from the conversation, which makes it possible to probe deeply into specific,
concrete cases rather than settling for general or impressionistic answers. For
instance, after a resident came in to demand the committee’s help in resolving
a dispute with a neighbor, I would inquire about the history of that dispute,
the parties involved, which government offices or courts had played a part in
dealing with it, and the like. If one were interviewing an RC member outside
the context of the office without such cues, she might well never happen to
mention that particular dispute and thus the information gained about media-
tion behavior could be much blurrier.

Conclusion
A great deal of political information that matters to researchers, from people’s
unvarnished opinions to the workings of closed institutions, is hidden. Also,
much in the political world is subtle, in that understanding or coding it is
no trivial matter, or in that its relationship to established social science con-
cepts is uncertain. When researchers are confronted by any of the above three
situations, or all at once, methods derived from ethnography and participant
observation, or “site-intensive methods,” may be called for. For the reasons
explained in this chapter, these conditions can obtain in any country, but they
More Than an Interview, Less Than Sedaka 161

are especially prevalent in politically restrictive and non-Western contexts like

China. The strategy discussed here is not the only way to approach hidden
and subtle politics, of course: careful analysis of public texts, unearthing and
interpreting written documents, and drawing inferences from published sta-
tistics can also bear fruit.
The purpose of site-intensive methods is to gain access to information that
would otherwise remain inaccessible and to obtain data of high validity. Space
constraints here have not permitted much discussion of the nuts and bolts of
how one goes about using these methods. But in essence, the researcher spends
time to form relationships with human subjects that are deeper than those
possible in a single interview. Rapport and trust are built up to the extent pos-
sible. Through prolonged interactions, the researcher aims to get below the
surface of things, the “party line,” and the vague or evasive or unconsidered
answer. Watching people in the context of their natural habitat makes avail-
able data that could never be captured through a questionnaire or an off-site
conversation, useful though those may be in other ways.
By definition, these methods involve an investment of time and effort that
can be greater than other methods on a per-informant basis. But these costs
may not be as steep as some think them to be. Full-blown immersion for
many months at a single site is not the only form that these methods can take.
Political scientists may have good reason to split their time among multiple
sites, or to use SIMS in limited ways to augment other methods.
This chapter has shown (or has provided a reminder) that contrary to the
beliefs of those who associate these methods with theory-free, barefoot empir-
icism, they are well suited for critical tasks in the building of theory: gener-
ating hypotheses, testing them in at least two ways, and refining concepts.
Classic and current examples of this work, both in the China field and in the
broader discipline, illustrate its value. To more fully capitalize on the potential
of this approach, it will be necessary to promote dialogue among practitioners
within political science about how to employ these methods well, efficiently,
and with maximum theoretical payoffs. It also means creating coherent, per-
suasive explanations of the utility of ethnography and participant observation
to our colleagues in the discipline so that it will be better understood and
appreciated.
9

Cases, Questions, and Comparison in Research

on Contemporary Chinese Politics

William Hurst

Not that long ago the opportunity to conduct fieldwork in the People’s
Republic was greeted with an uneasy combination of euphoric enthusiasm
and considered skepticism by political scientists. Well into the 1980s, scholars
seriously debated the merits of fieldwork on the mainland versus research con-
ducted exclusively in Hong Kong, Taiwan, or abroad (Thurston, 1983). From
the beginning, fieldwork on the ground in China was a touchy political subject
and researchers’ concerns continue to center on gaining access, ensuring the
safety of interviewees and collaborators, and the political impact of their find-
ings inside and outside China.
Frequently left aside, however, are questions of how to choose fieldwork
sites and what impacts one’s choice of locale or locales have on research
designs and outcomes. One recent exception is Maria Heimer’s thoughtful
essay in support of what she terms a “one-case multi-field-site approach” to
fieldwork research design, in which she argues that “authors can gain a deeper
knowledge of one phenomenon by probing for similarities, while downplaying
variations across place” and emphasizes that “this research design is different
from, say, going to four field sites and treating them as four different cases
of one phenomenon . . . and looking for variations between the four cases”
(Heimer, 2006: 62, 69).
In contrast to Heimer, I advocate for the advantages of selecting cases
with an eye toward explaining particular kinds of variation. What I term a
form of within-case comparison, one specifically based on selecting research
sites representative of larger subnational units, is offered as a way to at least
partially overcome the challenges of defining the scope of one’s findings and
guarding against an excessive focus on outliers or small subsets of the true
range of variation within China. Though the examples I use are from my own
study of Chinese laid-off workers, the arguments are meant to be generally
applicable to other research on China and in other branches of comparative
politics.

162
Cases, Questions, and Comparison in Research 163

What Is a Case? Why Study it?

The term “case” or “case study” is not always deployed or defined the same
way. John Gerring has proposed a clear definition of a case study as “an inten-
sive study of a single unit for the purpose of understanding a larger class of
(similar) units,” where a unit (case) is taken to be “a spatially bounded phe-
nomenon . . . observed at a single point in time or over some delimited period
of time” (Gerring, 2004: 342). The case study researcher must discern the rel-
evant “unit” as well as the larger class of units (universe of cases) from which
it is to be drawn.
We could designate significant phenomena in Chinese politics as our cases,
conducting field research in several localities and looking for commonalities.
This facilitates crafting descriptive or exploratory tales of how and why we
think these outcomes came to be. We could go even further and take “China
as a whole” as our case, conducting research on national-level phenomena
only. Alternatively, we can choose to view smaller units within China’s poli-
tics, society, or its administrative hierarchy as our cases for study. In practice,
this last approach is probably the most common (and most useful) in the study
of Chinese politics.
Adopting it raises several additional questions, however. How does one
decide which cases to study? Which research questions can be fruitfully
addressed with research on which cases? Is it best to research a single case or
might comparison of several cases provide for more reliable generalization?
These are just a few of the questions faced by researchers opting for the study
of one or several smaller cases in Chinese politics. I propose a solution to some
of these questions, which I refer to as systematic subnational comparison. This
framework is applicable to research questions about which some variation –
rather than Heimer’s sought-after commonality – across cases within China
can be expected.
Perhaps the most ambitious estimation of the general usefulness of case
studies is Stephen van Evera’s. He has claimed that case studies can be used
for “testing theories, creating theories, identifying antecedent conditions, test-
ing the importance of these antecedent conditions, and explaining cases of
intrinsic interest” (van Evera, 1997: 55). Theory testing can only be done with
case studies under special circumstances (e.g., the “crucial case study”) and a
few cases that are really of “intrinsic interest” beyond small subfields. These
two uses are thus less relevant for most researchers1. My method of systematic
subnational comparison can be used to create bounded theories, to specify the
“antecedent conditions” (i.e., background variables) required for these theo-
ries to operate, and to test the necessity of these background conditions and to
estimate the scope of the generalizabilty of the theories inferred.

1
For a strong indictment of these uses of case studies, particularly the “crucial case study,” see
King, Keohane, and Verba (1994:208–212). The earliest, and one of the best, pieces highlight-
ing the usefulness of “crucial case studies,” on the other hand, is Eckstein (1975).
164 William Hurst

Subnational Comparison and Case Selection

in Comparative Politics
Subnational comparative analysis has become fashionable in the broader
field of comparative politics (Linz and de Miguel, 1966; Mahoney and
Rueschmeyer, 2003: 14). Richard Snyder argues strongly that a focus on
units below the national level can help guard against two problems, which he
labels “mean-spirited analysis” and invalid part-to-whole mappings (Snyder,
2001b: 98–100). Mean-spirited analysis is the practice, common in cross-
­national research (particularly, though not exclusively, that which is quantita-
tive) of using aggregate national-level indicators as proxies for more detailed
knowledge of different parts of a particular country – that is, assuming that
all parts of a country resemble its national mean when in fact no localities may
closely match the country’s aggregate-level score.
Invalid part-to-whole mappings are often, though not exclusively, linked to
qualitative case study analysis. As Snyder puts it, “invalid part-to-whole map-
pings result when traits or processes specific to a well-studied region or other
subnational unit are improperly elevated to the status of national paradigm . . .
national cases are inappropriately coded as if the whole nation possessed the
set of attributes characterizing a specific region or set of localities” (Snyder,
2001b: 99). This problem has been especially common in research on Chinese
politics.
Much subnational analysis in the wider field of comparative politics has
sought to steer clear of the two pitfalls just described while concentrating on
various aspects of institution building, intergovernmental relations, or public
administration and policy implementation. From this angle, insightful research
on Latin America, Europe, North America, and parts of Asia has examined
the shape and functioning of various kinds of subnational governments (e.g.,
Anderson, 1992; Brace, 1993; Cornelius, Eisenstadt, and Hinley, 1999; Eaton,
2004; Gray, 1994; Heller, 1999; Herrigal, 1996; Kohli, 1987; Kooghe, 1996;
Michelmann and Soldatos, 1990; Putnam, 1993; Sinha, 2003; Snyder, 2001a;
Stoner-Weis, 1997; Tendler, 1997; Varshney, 2002). The China field has also
had its share of research on individual provinces or other subnational gov-
ernmental units in comparative perspective.2 When studying the shape and
workings of formal political institutions, one finds that the boundaries of sub-
national cases are often easily defined as congruent with the boundaries of
subnational political units – such as states, provinces, cities, counties, towns,
or villages.
But what about studying outcomes and processes that do not necessarily
fall within the bounds of subnational political units – like social movements,

2
On post-1949 military/administrative regions, see Solinger (1977). On Chinese provinces, see,
e.g., Fitzgerald (2002); Cheng, Chung, and Lin (1998); Hendrischke and Feng (1999); and the
journal Provincial China published by the Institute for International Studies of the University
of Technology, Sydney. For a more thorough discussion of recent work on provinces in China,
see Dittmer and Hurst (2002/2003: 18–20).
Cases, Questions, and Comparison in Research 165

poverty, economic growth, industrialization patterns, crime and terrorism,

and many other important forms of political activity? It can be artificial and
sometimes counterproductive to maintain rigid adherence to using subnational
political units as cases. Rather, socially, politically, or economically meaning-
ful divisions that are relevant to the questions at hand should be adopted as
ways of delineating subnational units for comparison.
Notable in the wider field are works that aim to explain macro-processes in
terms of subnational units that are purpose-specified especially for their rel-
evance to the research question. For example, Richard Bensel, in his analysis
of American industrialization in the late nineteenth century, argues that eco-
nomic development in the United States was profoundly uneven, with distinct
regional political economies discernible in the Northeast and Great Lakes,
South, and West (Bensel, 2000: ch. 2).3 In each region, popular claims against
wealth and capital accumulation were structured by this political economy
(Bensel, 2000: 12–15, ch. 4). Similar use of purpose-defined subnational units
of analysis can illuminate the study of any country’s politics.

Using Subnational Comparison to Study Chinese Politics

The best qualitative case study research in the China field draws only mea-
sured generalizations and places its findings in proper context. Still, much
work in this tradition suffers from a tendency to draw invalid part-to-whole
mappings. Innumerable studies of Shanghai or Beijing – China’s most atypical
cities – or of Guangdong or Jiangsu – China’s most atypical provinces – claim
to explain the dynamics of politics across the whole of China.
Since the mid-1990s, new research has explicitly sought to compare several
subnational units within China to draw more careful and accurate generaliza-
tions. Thomas Bernstein and Xiaobo Lü’s work on rural taxation and Kellee
Tsai’s work on informal finance are notable examples (Bernstein and Lü, 2003;
K. Tsai, 2002). These books represent a great leap forward in the quality and
systematic nature of subnational comparative analysis in the China field, but
their particular techniques may not travel well for the study of all other types
of research questions.
Work such as that by Bernstein and Lü begins with a deductive premise that
all of China can usefully be divided into “three great belts”: Coastal, Central,
and Western. Such research must start with the assumption that large, arbi-
trarily defined (usually divided by lines of longitude or in terms of “how many
provinces away from the sea”) swaths of China are both internally coherent
and distinct from one another in ways that are analytically useful. In the case
of the three great belts in China – any one of which may be more populous and
in many ways more diverse than the entire European Union, and may contain
a land area larger than the U.S. West Coast – this obviously does not hold

3
For more detail on how conflict among regions structured much of American politics for more
than 100 years after the Civil War, see Bensel (1984).
166 William Hurst

true for all research questions. A general disadvantage of this approach is that
the subnational units studied are too large and heterogeneous to permit the
intensive detailed analysis often needed to tease out causal mechanisms, trace
complex processes, or refine concepts.
Kellee Tsai’s work is more inductive. Rather than starting with a concep-
tion of Chinese regions, she begins with intensive studies of several localities
and then inductively suggests types or regions based on her micro-level find-
ings. The problem with this is that little justification is given a priori for the
selection of given research sites. Thus, the reader is left wondering whether
Tsai perhaps selected localities that are all outliers (and not truly representa-
tive of any broader types) or, alternatively, sites that all fall within a narrow
segment of the true range of variation across all of China (and are therefore
representative only of small subtypes). This issue also arises in other works,
such as Susan Whiting’s book on local institutions and economic develop-
ment. It is clear from the outset that her cases display only a small portion of
the full range of variation across all of China. This limits the generalizability
of the findings, even though Whiting emphasizes that she is mainly interested
in inductively suggesting types from a limited set of case studies (Whiting,
2001: 29–37).
This kind of work is useful for building what Alexander George once called
“typological theory” (George, 1979: 43–68) and helps advance van Evera’s
broader goal of theory development. It does not, however, allow researchers to
specify the scope of generalizations that can be drawn from their arguments.
It also does not promote inferences or arguments about antecedent conditions
or background variables necessary for theories to operate.
Scholars could ideally combine Bernstein and Lü’s deductive reasoning and
a priori justification of case selection with Kellee Tsai’s and Whiting’s fine-
grained, nuanced, and careful definitions of regions and categories. Selecting
cases in China is a complex task. Most often it is not feasible to conduct
intensive research (especially fieldwork) on the large subnational units (e.g.,
macro-regions, major branches of the administrative apparatus, etc.) whose
behavior we might want ultimately to explain. This means we are left to select
cases from among one or more of these units – treated as subpopulations –
and ask these to stand in for all cases in their respective populations. If we are
interested in regions, we may need to select provinces, counties, villages, or
cities for more intensive analysis and ask these to stand in for all provinces,
counties, villages, or cities within their regions.
This is a form of what has been called a “typical case study” approach
as applied to the comparative analysis of several subgroups within a larger
population (Gerring, 2001: 218–219). Such an approach is a “most similar
systems” design of subnational comparative analysis within a single country,
in which “X-variation” is largely confined to a clearly specified set of attri-
butes, but considerable “Y-variation” is observed (Przeworski and Teune,
1970: 32–34). The key is to select cases that are at least reasonably represen-
tative of the larger subgroups they are meant to stand in for. Some readers
Cases, Questions, and Comparison in Research 167

may object to the idea of setting out for fieldwork in China with a preestab-
lished list of cases. Such a research program would likely be impractical.
Some may even accuse those who attempt systematic subnational compara-
tive work of applying the rubric ex post facto to findings obtained by simply
following their connections (关系) and “going where their contacts are.” Such
critics assume that systematic subnational comparison requires all cases to be
chosen in advance of landing in China. This is not necessarily true, as I will
discuss below.
What is important upon setting out for the field is to know what sets of
cases might be suitable. This means being able to recognize a good case or
an inappropriate case when it presents itself. My framework does not ignore
the improvisational element – the “research as jazz” paradigm – necessitated
by the opacities, restrictions, and uncertainties of China research. Rather,
it introduces the idea of a standard on which an ensemble of cases can play
variations and solos.

Looking at Laid-off Workers in Chinese Cities

Regions and City Case Studies
State sector layoffs in China and their political and social impact can most
usefully be examined through subnational comparative analysis. Specifically,
the intensive study of micro-level politics and social change in individual cit-
ies, systematically chosen as representative of broader regions, can facilitate
the explanation of important outcomes. In addition, this sort of comparative
analysis allows the formulation of hypotheses about when and where these
explanations are valid. In this section I will outline my explanations for the
causes of layoffs and patterns of job losses experienced across several different
regions. Elsewhere in my research, I also apply this approach to an analysis of
policy responses to layoffs, workers’ coping strategies and informal paths to
reemployment, and patterns of workers’ contention (Hurst, 2009).
A clear understanding of the overall spatial scope of the phenomena is
needed to define meaningful regions for comparison (Hurst, 2009: ch. 1).
Layoffs were not a significant issue in all parts of China. Reliable informa-
tion on exactly where and when jobs were lost is surprisingly scant. Even
internal government reports and secret statistics were given little credence
by responsible officials.4 Without complete data on how many workers left
their jobs voluntarily, how many were reemployed, or how many had difficul-
ties finding ways to make ends meet, we must rely on previous observations,
available internal documents, and the opinions of the Chinese bureaucracy.
One internal report in 2001 listed twenty-one cities as having experienced
significant political and economic problems related to state-owned enterprise
(SOE) layoffs. The cities were grouped into what the authors considered to

4
Interviews with six State Council ministry officials, Beijing, 2000–2002.
168 William Hurst

be regional categories: the “Northeast,” the “lower and middle Changjiang

region,” “Northern China,” and “Western China” (DRC, 2001). In an inter-
view, a State Council ministry official with detailed knowledge of the issue
claimed that the central government has “come to the conclusion that we need
to focus policy solutions on thirteen provinces and special municipalities.”5
I maintain that there are four regional political economies in which significant
SOE layoffs have occurred: the Northeast (Liaoning, outside of Dalian, Jilin,
and Heilongjiang provinces), the Central Coast (Tianjin, Coastal Shandong,6
Jiangsu, Dalian City, and Shanghai), North-Central China (Shanxi, Shaanxi,
Inland Shandong, and Henan provinces, plus the cities of Lanzhou and Baotou),
and the Upper Changjiang (Hubei, Hunan, Chongqing, and Sichuan).
These regions exhibit variation on five key dimensions, including (1) local
state capacity, (2) the general business environment for SOEs, (3) working-
class society, (4) market opportunity, and (5) central-local relations. Local
state capacity in this context revolves around fiscal capacity – particularly the
ability of local governments to finance spending priorities rather than their
ability to collect particular proportions of the taxes they are owed. The SOEs’
business environment is determined by how much competition they face from
nonstate firms and by their general prospects for profitability. Working-class
society is a three-dimensional concept encompassing class identity (i.e., the
workers’ view of themselves as members of a working class), the structure of
workers’ social ties, and popular perceptions of the Maoist past. All three are
strongly rooted in distinct processes of working class formation. Physical lega-
cies of industrialization, especially housing patterns, also help structure work-
ers’ social ties such that they either cut across enterprise boundaries or are
concentrated within them. Market opportunity refers to both entrepreneurial
and other nonstate-sector employment opportunities. Central-local relations
refer to the degree of oversight, assistance, communication, and control from
the center over city governments.
This variation has its roots in the historical processes of industrialization
and development that were also differentiated by region. Though regional leg-
acies from even earlier periods still exert some influence, the industrialization
of these regions occurred primarily during the 100 years between 1880 and
1980. Each region had a distinctive pattern of sectoral distribution of SOEs,
timing and manner of industrialization, location of SOEs in particular types
of cities, relative presence of market activity and commercial centers, trans-
portation infrastructure, and historical relationship with the central govern-
ment. As reforms have intensified underlying divergent regional development
patterns, each region today thus has its own particular political economy and
faces its own particular employment situation.

5
Interview with State Council ministry official, Beijing, 2002.
6
I define Coastal Shandong as Weihai, Yantai, and Qingdao cities. The rest of the province
I define as Inland Shandong. The full argument is presented in much more detail in Hurst
(2009: chapter 2), upon which the discussion here is based.
Cases, Questions, and Comparison in Research 169

It is worth pointing out, as an aside, that regions of this sort are not the
most suitable level of disaggregation for every research question. For issues
surrounding many aspects of policy implementation, for example, key dimen-
sions of differentiation would be the urban-rural divide and levels in the
administrative hierarchy. Region in some form would have to be considered,
but it could certainly turn out to be that rural areas under the jurisdiction
of nonprovincial-capital prefectural-level cities implement this or that policy
in ways that are broadly similar throughout all of China. Conversely, if one
wanted to uncover the micro dynamics and complexities of specific types of
cross-national production networks, differentiating between townships or
counties within one city or across a couple of cities – like Kunshan, Dongguan,
or Wenzhou – could be more important than drawing distinctions between
regional models of political economy or development. For my study of laid-off
workers’ politics, however, region was the most salient axis of disaggrega-
tion, and the areas with significant layoffs could be divided into four major
regions.
In each region, I selected one or two representative cities for case study
analysis. But just how does one determine the “representativeness” of a city
within a regional political economy? We must start with the legacies of ear-
lier industrialization and development – sectoral distribution of SOEs, timing
and manner of industrialization, location of SOEs, relative presence of market
activity and commercial centers, transportation infrastructure, and historical
relationship with the central government – which produced differentiated con-
temporary regional political economies. Cities must then be selected whose
scores on these key criteria match those for the region at large. The cities’
scores on key dimensions of contemporary regional political economy, from
central-local relations to the business environment for SOEs, must also not be
regional outliers.
When selecting cities, I sought to include cases within the regions I had
identified that were broadly representative of their larger regional political
economies. Thus, for example, I turned down several chances to conduct exten-
sive fieldwork in Guangzhou since it was outside my regions of interest, even
though access there might have been better than it turned out to be in some
of the cities I did select. I also turned down chances for research in Dalian, as
this city was clearly not representative of the Northeast (I opted for more time
in Benxi and a second stint in Shenyang instead and, after much deliberation,
ended up categorizing Dalian as a Central Coast city). Finally, I tried with-
out success to arrange fieldwork in cities that would have been good fits for
my research design, including Ziyang, Zhuzhou, Jilin, Nanjing, and Wuhan.
However, these had to be left for future projects or other researchers.
As stated earlier, no one can reasonably go to the field with a list of per-
fect fieldwork sites etched in stone and proceed to simply run through an
established schedule. Contacts, institutional ties, and chance often influence
research access more than the best-laid plans. Those employing my brand of
subnational comparison therefore must be prepared to roll with the punches
170 William Hurst

of the field in a way that allows them to cobble together a set of research
locales that can serve the broader research design – the way a jazz musician
might improvise a path back to a standard theme when faced with surprise
harmonizing by her colleagues. This was the process through which I selected
and revised my selection of case study cities during my research.
As principal research sites, I selected cities that were as representative of
their regions as I could find among those places where I could get research
access. In the Northeast, the city of Benxi in Liaoning province was selected.
In the Central Coast and Upper Changjiang regions, Shanghai and Chongqing,
respectively, were studied. In North-Central China, I selected the two cities
of Datong and Luoyang, both representative of the region, so as to provide
a rough test of the internal coherence of the regions as I define them (Hurst,
2009: 32–36).

Causes of Layoffs
Deteriorating SOE business environments and divergent patterns of central-
local relations produced different patterns of layoffs across regions. Declining
profitability and spiraling costs led firms in the Northeast to begin laying off
workers in the 1980s. Sector-specific problems in the business environments of
some North-Central and Upper Changjiang SOEs led them to lay off workers
in the early and mid-1990s, even as many enterprises in other sectors in those
same regions added new personnel. The Central Coast and provincial capitals
were largely spared significant layoffs until after the Fifteenth Party Congress
in 1997, when the central state imposed a new cost-cutting policy on SOEs
in an effort to force them to adapt to what it perceived as new competition at
home and abroad.
Although other regions of China enjoyed intensive growth in agriculture
resulting from decollectivization, the Northeast’s efficiency advantage in sta-
ple grains eroded. Though rural production has always played second fiddle
to urban industry in the People’s Republic, the fate of SOEs was not entirely
divorced from that of the countryside, as rural surpluses provided investment
capital for industry. The Northeast’s relative (and occasionally absolute) losses
from decollectivization hampered local financing of new industrial develop-
ment or reform of ailing SOEs well into the 1980s (Luo, 1994: 118). Many
Northeastern SOEs faced a version of hardening budget constraints, as they
had difficulties balancing their books, fell into wage and pension arrears, and
were forced to lay off workers.
Layoffs across the Northeast reached a critical scale during the 1990s.
Though nearly all the layoffs remained informal, arrangements like “long
vacations” (长假) came to be accepted as a permanent state of affairs (DRC,
1999a). “I was officially laid off in 2000, but I had been on long vacation since
1991. In truth, it is the same thing,” said one former mine foreman in Benxi. A
retired Benxi miner angrily said, “The leaders told me to take a long vacation
in 1988. After that, I was just waiting to retire. I got nothing from the work
Cases, Questions, and Comparison in Research 171

unit. Then they finally allowed me to retire in 1995 – two years late – and I
only got one-fourth of my proper pension.”7 The trend remained broadly the
same as that in the 1980s, but it quickened and intensified.
In the cities of the Upper Changjiang, layoffs appeared for the first time.
Defense-linked companies and textile firms in particular faced mounting
competition from nonstate and foreign firms. Mismanagement under increas-
ingly lax supervision was also a factor. Furthermore, local governments and
higher level units were progressively less able to subsidize struggling firms.
The Upper Changjiang experienced proportionally more layoffs than North-
Central China during this period because the textile and military sectors there
employed particularly large numbers of workers. The Upper Changjiang had
an especially large concentration of small-size SOEs and urban collective
firms. Between 1987 and 1997, these firms experienced a particularly severe
worsening of their business environments, largely due to competition from
nonstate firms (often rural township and village enterprises). Official statistics
show a 15 percent drop in the number of collective-sector workers employed
in the Upper Changjiang between 1993 and 1997.8
North-Central firms experienced a net addition of workers during the first
part of the 1990s. Even so, some firms there were laying workers off. Many
of these SOEs faced competition from township and village enterprises in the
production of coal and textiles as well as declining demand for the region’s
agricultural machinery. Even though conditions were worsening for certain
firms and in certain sectors, profits for others outweighed negative trends
across the region as a whole.
Few workers seem to have lost their jobs involuntarily in Central Coastal
SOEs during the first half of the 1990s. Without clear fiscal imperatives, and
with significant political disincentives from a closely watchful central state,
most firms held their workforces at a constant size. Those that shed work-
ers generally lost them to better opportunities in the nonstate sector. As one
Shanghai city official explained, “before 1997, some workers left their work
units to take up higher paying private-sector jobs or to go into business for
themselves. In Shanghai at least, no one was forced out.”9
Written sources on Shanghai during this period tell a similar story. Case
studies of key Shanghai SOEs detail how most workers were retained prior
to 1997, and many of those who did leave did so voluntarily (DRC, 1999b,
1999c). Though Shanghai was at the forefront of national policy in admit-
ting to unemployment and instituting official mechanisms of assistance before
1997,10 the scale of the problem was held well in check through a combination
of relatively healthy SOE business environments and particularly generous
central government subsidies.
7
Interviews with a 47-year-old male laid-off foreman and a 58-year-old female retired miner,
Benxi, November 2001.
8
Zhongguo laodong tongji nianjian, various years.
9
Interview with Shanghai municipal official, July 2000.
10
On this, see D. Tang (2003).
172 William Hurst

In September 1997, the CCP convened its Fifteenth Party Congress in

Beijing. Key decisions were taken on many issues relating to SOE reform. It was
decided that SOEs were to evolve into profitable firms and that a chief means
to achieve this goal would be to lay off workers to cut costs. Ideally, enter-
prises were to become not just market firms but global competitors (Nolan,
2001), and excessive labor costs were an important obstacle that had to be
cleared from their path. With his speech opening the proceedings, General
Secretary Jiang Zemin threw down a gauntlet for SOEs, telling them they
must become profitable market actors and giving them a green light to do so
by any means necessary (with specific endorsement of cutting workers to trim
costs) (Renmin chubanshe, 1997: 23).
Within weeks of the party congress, SOEs felt pressure to trim what were
suddenly deemed bloated workforces. SOE managers and labor officials on
the Central Coast closely heeded the signals from Beijing. Over twelve months
in 1997 and 1998, Shanghai saw more than 26 percent of state- and urban
­collective-sector workers lose their jobs. Roughly another 10 percent went out
of work the following year (Yin, 2001: 101). Even though their business envi-
ronments had not obviously deteriorated, Central Coast SOEs came under
pressure from a closely observant central state to trim their labor forces. One
Shanghai labor cadre said he felt as though he had been told by the party
congress that “crying and complaining gets you nowhere in the market; [i.e.,
don’t ask for subsidies] forced exit [layoffs or firm closure] is the only way.”
Another Shanghai cadre derided the directives of the congress, chafed at
bureaucratic micro-management, and mocked the lack of genuine marketiza-
tion by quoting a joke that played on his city’s thick regional accent, saying,
“before, everything depended on the bureau chief, afterward, everyone had
to look to the market” (以前什么都靠司长，后来谁都要看市场)– the terms 司长,
bureau chief, and 市场, market, sound nearly identical when pronounced by
many Shanghaines).11
The CCP’s post-1997 policies did not so much harden SOE budget con-
straints as realign them. Instead of subsidies being tied to vertical connec-
tions, political behavior (表现),12 the attainment of production targets, or full
employment, they became linked to workforce reductions and the achieve-
ment of other “reform targets.” And the effects of this ideological shift in
incentives extended beyond the Central Coast to regions that had experienced
earlier problems. Before the end of summer 1998, firms in the Northeast were
instructed to “open up” previously hidden unemployment and to further
deepen workforce reductions. In August 1999, General Secretary Jiang Zemin
addressed a conference on SOE reform in Northeast and North China, artic-
ulating this explanation for the regions’ problems and endorsing layoffs as a
favored means of reducing “excessive costs” (Jiang, 1999). This gave managers
11
Interviews with Shanghai Municipal Party Committee cadre and trade union cadre, October
2000 and July 2000.
12
On the importance of 表现 within the firm, as well as in interfirm relations, see Walder (1986:
132–147, 160–162).
Cases, Questions, and Comparison in Research 173

an explicit opportunity to earn plaudits by cutting workers to reduce labor

costs, even when this failed to resuscitate their firms’ profitability.
North-Central and Upper Changjiang SOEs felt new central pressure to
openly reduce their workforces rather than to expand them or conceal cuts.
Many North-Central and Upper Changjiang SOEs in the late 1990s also con-
fronted a similar constellation of declining profitability and reduced subsidies
to what their Northeast counterparts endured in the 1980s. The difference
was that they were encouraged to respond with layoffs whereas firms in the
Northeast in the 1980s had been politically constrained from responding with
layoffs.
Overall, three distinct patterns of job losses can be observed across the
four different regions. Each of these, in turn, was the product of a specific
set of causal processes, rooted in the political economy of each region. The
Northeast faced deteriorating business environments and strained central-local
relations, causing its SOEs to struggle in the 1980s, when firms in most other
regions were doing well. North-Central and Upper Changjiang enterprises in
certain sectors grappled with increasingly competitive business environments
and declining government subsidies in the early 1990s, whereas firms in the
Northeast continued to struggle and Central Coast companies boomed.
Finally, after the Fifteenth Party Congress in 1997, the political environ-
ment for all SOEs shifted. Layoffs were no longer to be avoided at all costs. In
fact, they became highly politically desirable for many SOEs. Some firms in the
Central Coast even “stormed” to meet new targets for shedding workers whom
managers felt were still needed in the production process. Overall, Chinese
SOEs did not trim their workforces in the face of any relentless advance of the
market or true hardening of socialist budget constraints. Rather, a multiplicity
of factors, combining in different ways across space and time, shaped regional
patterns of unemployment and retrenchment.

The Utility of Subnational Comparison versus Single-

City Case Studies in Explaining the Causes of Layoffs
Single-city case studies, although exceedingly good at illuminating complex
causal mechanisms, would do little to explain the regional dynamics just dis-
cussed. If I had looked at only Chongqing, for example, I would have come
away with a story specific to the regional patterns of the Upper Changjiang
region, or even worse, to one or two automotive, defense, or textile factories.
Conversely, if I had examined only Shanghai, I would have found evidence for
an exaggerated role for top-down administrative guidance and command in
the genesis of SOE layoffs.
In fact, reliance on single-city or single-unit case studies can give rise to
spurious debates among scholars actually looking at apples and oranges.
The researcher who worked in Ziyang or Kaifeng would be just as correct in
his view of layoffs as sector-specific phenomena as his colleague working in
Jiamusi or Fushun would be in her claim that whole cities had been limping
174 William Hurst

along and sloughing off workers since the early days of reform. This is not
to say that either of these would not be able to offer superior nuance in
their descriptions of specific causal mechanisms. But if whole research ques-
tions are dominated only by such single-unit case studies, we would have
no way of knowing who is right or, as in this example, whether both sides
of an argument might be valid for regionally or otherwise bounded sets of
outcomes.

General Issues of Systematic Subnational Comparison

versus Intensive Studies of Single Cases and Large-N
Studies of Many Cases in China
Perhaps my brand of subnational comparison spreads precious field research
resources and time too thinly. Perhaps if I had spent all of my time in Shanghai,
for example, my study could have included many more interviews there, along
with extensive archival research and perhaps even analysis of detailed local
quantitative data. For comparativists, however, after a certain point it just
has to be “turtles all the way down” (Geertz, 1973: 29). Knowing everything
there is to know about Shanghai just for the sake of understanding Shanghai
may be valuable from some perspectives, but it does not, in itself, advance the
study of comparative politics.
No matter how much detail we can assemble about any one given case,
it does not help us to generalize from that case. If generalization is the aim,
we similarly cannot get very far using Heimer’s preferred one-case multi-site
method. Even if we study multiple places, if we treat them as forming one
unified whole, we cannot venture a decent guess as to how generalizable our
findings might be, even within China (unless we could somehow realistically
study some very large proportion of all possible field sites in the country). The
best we can hope for from the analysis of single units or cases is the genera-
tion of testable and plausible hypotheses that can later be evaluated in more
rigorously comparative work.
In fairness, the type of comparative analysis I advocate takes us only a small
part of the way along the journey from hypothesis generation to hypothesis
testing. With such small samples, so little real knowledge of the populations
from which they are drawn, and no way even to begin to estimate the proba-
bilities with which one’s conclusions might hold true, my brand of subnational
comparison falls far short of large-N quantitative work for testing hypotheses.
It can, however, give us a good idea of where and when certain explanations
might be valid. In other words, it lets us generate and evaluate hypotheses
about background variables and antecedent conditions. It can also help sort
out and reconcile competing claims across different subnational populations
of cases and generate much more nuanced, causally complex, and internally
consistent hypotheses. Finally, looking at several cases within China can help
us rule out invalid part-to-whole mappings drawn from explanations of phe-
nomena analyzed in single cases.
Cases, Questions, and Comparison in Research 175

This is not to argue against the utility of doing single case studies under any
circumstances. The intensive study of a single case, at a greater level of detail
than can be accomplished under most subnational comparative designs, helps
elucidate the fine-grained specifics of causal mechanisms and promotes refine-
ment and clarification of concepts perhaps better than any other method. But
what is gained in concept formation and the description of mechanisms is lost
in explanatory power and generalizability.
At the other end of the continuum are large-N quantitative studies of Chinese
localities. These can be undertaken using statistical information (either from
published sources or obtained by the researcher), data obtained from various
types of surveys, or a mixture of the two. Such an approach promises more
rigorous testing of hypotheses, a broader sweep than subnational comparative
analysis can offer, and the ability to quantify estimates of both the likelihood
any given explanation holds true and the level of generalizability of research
findings.
Despite all these apparent advantages, several factors detract from the
appeal of such methods in the study of Chinese politics. First, there is the
issue of practicability. It is not easy to obtain good quantitative data in China,
just as obtaining good qualitative or interview data is difficult as well. But
getting quantitative data is more costly in financial terms (something espe-
cially important for graduate students in the field) and the process of gather-
ing quantitative data is even more tightly controlled for foreign researchers
than the gathering of qualitative data. In order to get good quantitative data,
the foreign researcher needs to be relatively well funded and must work with
exceptionally talented and determined Chinese collaborators (who, fortu-
nately, have become much more numerous in recent years).
Second, there are some variables in China about which it is exceptionally
difficult to obtain or collect accurate quantitative data. An example from my
own research is the number of laid-off workers. Different sectors and locali-
ties record the number of laid-off workers differently from one another and
in ways that often diverge from national guidelines. These guidelines, in turn,
changed every couple of years during the 1990s and early 2000s, further com-
plicating the picture. Finally, the reporting of job losses is notoriously fraught
with deception, corruption, and political interference – some managers and
local officials deliberately overstate the totals, whereas many others deliber-
ately understate them; then higher level officials often “revise” the numbers
reported from below to suit their own purposes before passing them to the
next level, and so on. Given the political sensitivity of the issue, it would also
not be easy to conduct an original survey on unemployment. Even if possible,
sampling of respondents would present immense difficulties; it also would be
difficult to sort out the respondents’ answers because so many people have
different understandings of the concept.
Third, large-N work is good for some things but not for others. Just as
there are limitations to what subnational comparisons can tell us, so too
are large-N studies limited in their ability to examine causal mechanisms,
176 William Hurst

generate concepts, and probe the intricacies of political and social interactions
at the most micro level. Though I certainly support the use of quantitative
work in the study of many questions in Chinese politics, it is not always the
best method for the study of all questions. Perhaps more interesting, there is
likely a sequence in which the field as a whole can best benefit from each style
of research in a cycle over time.
I suggest that relatively new questions are likely best tackled with subna-
tional comparative analysis. This method is best at balancing fine-grained
work on mechanisms and concepts with hypothesis generation and assess-
ing the generalizability of explanations. Once we have hypotheses testable at
specified levels of generality along with new concepts and clear ideas about
causal mechanisms, we can move on to test the validity and generalizability of
competing claims with quantitative studies and to refine our understanding of
concepts and mechanisms with single-case intensive studies or ethnographic
research.

Conclusion: A Program for Taking Systematic

Subnational Comparison beyond the Study of
Chinese Workers
A method or analytical approach that is not applicable beyond the spe-
cific substantive questions it was developed to study is no method at all.
How, then, might one apply the basic framework of systematic subnational
comparison outlined earlier to contexts beyond the study of Chinese work-
ers, or even beyond China? The most basic answer is that researchers can
usefully divide national cases into subnational subpopulations of smaller
cases, select representative cases from each subpopulation, make their case
selection rubrics clear up front, and be careful about drawing generaliza-
tions that are too broad or not attempting to generalize enough from their
findings. Such broad programmatic statements are not entirely satisfactory,
however.
Starting from the decision of a researcher to engage in subnational com-
parative analysis, I suggest four concrete steps that I believe can make such
analysis more systematic and enhance the clarity of the results. First, research-
ers ought usefully to consider various ways of slicing up their national cases
into smaller subpopulations. Subnational political units do not always make
the best subnational cases or subpopulations. Likewise, political, economic,
topographical, or linguistic/ethnic regions are not necessarily natural or
appropriate. The specific definition of subpopulations, along with the degree
of disaggregation, must be driven by the research question and justified a
priori by the scholar in light of both the dynamics of the research question
and the relevant work of other researchers. To do a good job of systematic
subnational comparison, a researcher need not study every potentially rel-
evant subpopulation, but the inclusion of some and exclusion of others must
be explicitly justified.
Cases, Questions, and Comparison in Research 177

Second, before setting out for the field, researchers ideally ought to have a
clear sense of not only the relevant subpopulations of cases but also of what
representative cases in each subpopulation might look like. Although it would
be impractical, and from this writer’s experience usually unwise, to set out for
the field with a predetermined list of cases set in stone, one must be able to
recognize a good case when it reveals itself and to ration precious fieldwork
time and resources judiciously and accurately away from outliers and toward
cases that can illuminate the subpopulations of interest. Making a list of nec-
essary attributes of representative cases – or even of example representative
cases themselves – for each subpopulation to be studied before beginning one’s
research would be a very useful step.
Third, researchers could usefully pay more attention to just where the
boundaries of their theories lie – that is, to avoid invalid part-to-whole map-
pings but also to take care not to underestimate the generalizability of their
findings. They can accomplish both objectives by highlighting and specifying
the necessary antecedent conditions and the criteria by which they define the
relevant subpopulations. The goal should always be to make testable rather
than irrefutable or watered-down equivocal statements. By telling future
researchers what can properly be tested with which data under what condi-
tions, scholars help the general project of advancing knowledge and useful
debate in the field. Failure to do so both inhibits progress in the field and sells
short one’s own research findings.
Fourth and finally, it is important not to lose sight of the broader unit of
analysis: China as a whole. It is always useful for researchers pursuing sub-
national comparisons to at least speculate on how and why particular sets of
antecedent conditions came to be in particular countries as well as on whether
and why similar sets of conditions might or might not be important in sorting
out subpopulations of cases in other national contexts. By doing this, research-
ers can postulate “meta-hypotheses” that can travel at least as well as their
more grounded, but bounded, “real” hypotheses drawn from their fieldwork.
If a template used to define subpopulations and to select good cases in China
works or fails to work in research on Russia or Brazil, this finding often can be
at least as interesting to both China scholars and Russia or Brazil specialists
as the specific findings of the research.
By following these four steps and continuing open and clear discussions of
case selection and other issues, researchers not only can usefully develop a more
mature tradition of systematic subnational comparison in the study of Chinese
politics but also can help promote the integration of the China subfield into wider
debates and conversations in comparative politics. By making our research more
intelligible to non-China specialists, highlighting our more general findings and
hypotheses, and giving some hints as to just how generalizable we think our
results are and what kinds of data might be used to test our claims in China and
elsewhere, we could make great progress toward becoming a “normal subfield”
under the comparative tent, and perhaps even get more nonspecialists to pay
attention to important findings and arguments from the study of China.
Part III

Survey Methods
10

A Survey of Survey Research on Chinese Politics

What Have We Learned?

Melanie Manion

In a political environment that remains (at best) officially skeptical about the
enterprise, representative sample surveys on Chinese politics have nonetheless
grown substantially in number in the past two decades: political scientists
trained and based outside mainland China conducted a mere two such surveys
in the 1980s, but the number increased more than tenfold in the 1990s and
continues to rise steadily. By mid-2008, some sixty articles, books, and book
chapters drawing from original representative sample surveys on Chinese
politics had appeared – including many articles in top-tier journals of politi-
cal science and area studies alike. This chapter surveys the surveys and their
products. It briefly explains the focus here on probability sample surveys and
describes the changing regulatory context within which researchers conduct
their surveys. Most of all, it evaluates their achievements, with attention to
their cumulativeness, contributions to knowledge, and fit in Chinese area stud-
ies. The chapter is not a primer on the conduct of survey research on Chinese
politics in mainland China.1 It is instead a status report and reflection on this
research, aimed as much (or more) at its consumers (and nonconsumers) as at
survey researchers themselves.
From the corpus of English-language monographs and peer-reviewed jour-
nal articles authored by political scientists and published in nonmainland
sources through mid-2008, I identify studies that exploit original probability
sample survey data.2 In coauthored works, I include studies that meet these
criteria so long as at least one of the authors is a political scientist. I have
surely missed some relevant surveys in my search – but not, I think, any rep-
resented in studies published in major journals of political science or Chinese

I thank Kent Jennings and Kenneth Lieberthal for helpful comments on an earlier draft.
1
For good overviews, see Shi (1996) and Wenfang Tang (2003, 2005). On survey cooperation by
ethnic minorities, see Hoddie (2008). On nonresponse, see Zhu (1996). On political sensitiv-
ity, see Tsai, Chapter 14 of this volume. Among new monographs drawing from survey data,
a good discussion of methods can be found in Dickson (2008).
2
See Appendix 10.1 for a precise description of what is (and is not) reviewed for this chapter.

181
182 Melanie Manion

area studies or organized by any of the roughly half-dozen major players in

the enterprise of survey research on Chinese politics. I focus in this chapter
on original surveys, not the small literature by political scientists who analyze
datasets produced wholly or mainly by others. At the same time, the paucity
of studies analyzing existing datasets is a sign of the relative immaturity of
survey research on Chinese politics, a topic that merits discussion and is taken
up below.

Probability Sample Surveys

As mainland China is large and diverse, what we observe in our fieldwork is
necessarily a small subset of a range of players, beliefs, and actions. Further,
as China is changing rapidly, what we observe is a snapshot that may or may
not be relevant beyond a single point in time. More observations, across more
space and at more points in time, can improve confidence that our fieldwork
observations are not highly unusual, but the problem is inherent and remains
fundamentally unresolved. If the goal is to generalize from what we observe,
then the argument for probability sample survey research as a method is quite
strong: without a probability sample of a sufficiently large number of observa-
tions, analyzed with inferential statistics, we can say nothing about the gener-
alizability of our fieldwork observations.
The notion of a probability sample survey implies that survey research-
ers select localities and individual respondents into their samples probabilisti-
cally. This permits researchers to answer basic questions about the population
sampled. More to the point, although particular estimates about the popu-
lation will differ from statistics provided by a complete and perfectly accu-
rate census of the population (which is unattainable), probability sampling
allows researchers to associate the estimate with a specific degree of certainty
(conventionally, 95 percent or 99 percent) that the true population value lies
within a specific range of values. By contrast, estimates of population values
based on samples drawn in ways that violate probability sampling differ in
unknown ways from the population sampled.
It is not very difficult to select a probability sample of localities in mainland
China (although it is more difficult actually to conduct survey research in the
selected localities, as this requires the cooperation of the political authori-
ties). By contrast, increased mobility and outdated residence lists have made
it more difficult to select a representative probability sample of individual
respondents in any selected locality in mainland China.3 The survey of surveys
below reveals that relatively few surveys on Chinese politics select samples
that are nationally representative. Most of the surveys are local probability
surveys: a probability sample of individual respondents is selected within a
locality or localities selected for convenience. Descriptive statistics along any
single dimension from a local probability sample are generalizable to the local

3
See the discussions in Chapters 12 and 13 of this volume.
A Survey of Survey Research on Chinese Politics 183

population sampled, of course. These descriptive statistics (unlike those based

on a nonprobability sample of local respondents) are unbiased estimates about
the local population – but not beyond it. Yet, even local probability samples
can permit survey researchers to generalize beyond the locality on the sorts of
questions that interest social scientists most: questions about the relationship
between variables.4 Indeed, considering mainland China’s diversity and rapid
pace of change, these are exactly the sorts of questions that can potentially
contribute most to our knowledge. For good reason, then, this chapter focuses
exclusively on probability sample surveys.

Regulatory Regime for Survey Research

One of the earliest high-level official responses to collaborative survey research
on Chinese politics ominously signaled its vulnerability: in 1990 the Central
Committee instructed political scientists at Peking University to cease work
on a collaborative survey project with the University of Michigan; the State
Education Commission (SEC) confiscated the already collected data, officially
declared as “state secrets.”5 Not long after this incident, the SEC and the
Chinese Academy of Social Sciences (CASS) were asked to work out appropri-
ate procedures to guide surveys conducted in collaboration with nonmainland
researchers, but little was done. A 1996 Central Committee document appar-
ently indicates that proper approval is required for surveys by CASS and uni-
versities if the projects involve nonmainland scholars, but the document does
not set out procedures for obtaining approvals.6
Since 1999, the National Bureau of Statistics (NBS) has been charged with
the creation, refinement, and interpretation of a regulatory regime governing
all survey research in mainland China conducted with researchers based out-
side the mainland – in Chinese 涉外调查, translated here as foreign-affiliated
surveys. The term refers to research conducted by Chinese agencies jointly
with (委托) or funded by (资助) organizations or individuals based outside
the mainland. NBS regulations distinguish between business surveys (市场
调查) and social surveys (社会调查). Survey research on Chinese politics falls
under the latter category, which is more strictly regulated. It is governed by the
Measures for the Administration of Foreign-Affiliated Surveys (NBS, 2004a),

4
This depends on the analysis not being contaminated by local peculiarities in the theorized
relationships. See Manion (1994).
5
The survey drew official attention with a request to transport completed questionnaires to
the Institute for Social Research at the University of Michigan for coding and data input. The
National Science Foundation (NSF), a sponsor of the project, responded to confiscation of the
data with a ban on NSF funding for any collaborative research with mainland China. The ban
was lifted in mid-1993, when the data were returned to Peking University for coding and input
there. An exportable electronic dataset was apparently viewed quite differently from “docu-
ments” such as completed questionnaires. See Marshall (1993).
6
I have not seen this document. My account is based on discussions with mainland Chinese
survey researchers.
184 Melanie Manion

which replace significantly more restrictive interim measures issued in 1999.

The key principles of the regime governing social survey research are the sur-
vey permit and project approval systems.
Survey research involving foreign affiliates may be conducted only by the
several hundred mainland agencies that have obtained a permit (涉外调查许
可证) for such research.7 Only the NBS and the statistical bureaus of provin-
cial-level governments have the authority to issue such permits.8 No organiza-
tion or individual based outside the mainland is authorized to conduct survey
research on the mainland without the participation of an authorized mainland
agency.
Approval for foreign-affiliated social survey projects must be obtained
prior to implementation of the survey. In addition to a copy of the contractual
agreement between the Chinese agency and the foreign affiliate, the applica-
tion for project approval must include a description of the purpose, content,
scope, sample, methods, and time frame of the survey as well as a copy of the
questionnaire. If changes are made to the survey plan, including the question-
naire, prior to the survey, further approval must be obtained. The NBS and
provincial statistical bureaus are obliged to approve (or not approve) the proj-
ect within fourteen days of receipt of the application. In particular circum-
stances, the time for approval may be extended another ten days. A written
explanation must be provided if projects are not approved.
The existing project approval system is very significantly relaxed, com-
pared to that reflected in the earlier interim measures.9 Before 2004, Chinese
political scientists could not reliably assure their nonmainland colleagues
that they would be able to share survey data with them. The earlier mea-
sures required not only prior approval of projects, but also a postsurvey
second approval to share data with nonmainland affiliates. Chinese authori-
ties could reconsider the confidentiality of the survey data in what might be
a new political context. The earlier measures effectively ruled out an early
commitment (in grant applications, for example) to public accessibility of
survey data.
Foreign institutional review boards may appreciate that the new regula-
tory regime requires informed consent. Questionnaire cover sheets must con-
spicuously note: “The respondent has voluntarily agreed to participate in this
survey.” NSB measures also prohibit use of foreign-affiliated surveys to harm

7
The China Daily noted in 2000 that 29 agencies had received permits to engage in foreign-
affiliated survey research. The NBS Web site documents the particular agencies to which it has
issued permits over the years. I counted 204 in 2004, 83 in 2005, 58 in 2006, and 186 in 2007.
Permits are valid for a period of three years. http:///www.STATS.gov.cn/
8
Survey research that spans provincial boundaries requires NBS authorization.
9
Business surveys enjoy an even more relaxed regulatory environment: they no longer require
survey project approval as long as the survey is conducted by an authorized Chinese agency.
The relaxation for both business and social surveys reflected in the new measures is the result
of a State Council decision some six months earlier. That decision was prompted by passage of
the Law on Administrative Permits in 2003. See NBS (2004b).
A Survey of Survey Research on Chinese Politics 185

the interests of any individual and require that surveys be designed so as to

maintain the confidentiality of responses.
The regulatory regime presents some new difficulties for foreign-affiliated
survey research, however. An NBS clarification (2004c) specifies exactly what
information must be indicated conspicuously on the questionnaire cover
sheet: the term 涉外 (foreign-affiliated) appears five times in a mere four lines.
That is, the measures make all but certain that survey respondents and local
authorities facilitating the project know that the activity has some sort of non-
mainland connection. Such a disclosure may jeopardize official cooperation,
affect interview response (rate and quality), or both. This is particularly likely
if respondents are elites.
In sum, survey research by political scientists based outside mainland
China is less vulnerable today than in previous years. Progress on a regula-
tory regime has been made since the early 1990s and significant progress has
been made in the past few years. At the same time, survey research on Chinese
politics remains subject to the vagaries of Chinese politics – in particular,
to official sensibilities about what constitutes politically sensitive questions.10
Survey researchers who operate within (and outside) the regulatory regime
governing their research presumably take the political context into account.
Undoubtedly, what results is something of a compromise, usually reflected in
the questionnaire design. So far as I can tell, however, what has not been com-
promised in the surveys reviewed below is the integrity of the survey enter-
prise as social science.

A Survey of Surveys
Table 10.1 presents the original probability sample surveys on Chinese poli-
tics identified from English-language monographs and peer-reviewed journal
articles authored by political scientists and published in nonmainland sources
by July 2008.11 The growth of survey research is evident: the pace in the 1990s
is being matched in the first decade of this century and also reflects work by a
somewhat broader representation of scholars.
Only three of the surveys are nationally representative – although, as dis-
cussed below, two recent additions to nationally representative surveys are
not reflected in the table. A large number of the surveys focus exclusively on
Beijing, a convenient but highly atypical sample of the urban Chinese popu-
lation. At the same time, there is fairly good representation of a regionally
diverse rural China, which mainly reflects scholarly interest in village elec-
tions. Most of the surveys were conducted with face-to-face interviews, but

10
Wenfang Tang (2003) argues that economic interests also partly explain NSB reluctance to
relax standards further. If all survey organizations could conduct foreign-affiliated surveys,
then the market for high-priced NSB survey data would shrink.
11
The numbering suggests thirty-two surveys, but survey 15 combines six annual surveys, dis-
cussed at greater length below and in Chapter 13 of this volume.
186 Melanie Manion

Table 10.1. Probability sample surveys on Chinese politics resulting in pub-

lications analyzing original datasets. Based on a review of English-language
monographs and peer-reviewed journal articles by political scientists published
in nonmainland sources by mid-2008

Year Localities Surveyed Size Publications Associated

With Survey
1 1986–87 1 city: in Jilin 250 Manion (1991, 1993)
2 1988–89 Beijing 757 Shi (1997, 1999b)
3 1990 4 counties: in Anhui, 1,270 Manion (1996, 2006);
Hebei, Hunan, Tianjin Jennings (1997, 1998,
2003); Eldersveld and
Shen (2001)
4 1990–91 National probability 2,896 Nathan and Shi (1993,
sample 1996); Shi (1999c)
5 1992 44 cities 2,370 Tang and Parish (2000);
Tang (2001a, 2001b,
2005)
6 1993–94 National probability 3,287 Shi (1999a, 2000, 2001);
sample; also Taiwan Chen and Shi (2001); Chu
and Hong Kong and Chang (2001); Kuan
and Lau (2002)
7 1995 Beijing 916 Dowd, Carlson, and Shen
(1999)
8 1995 Beijing 658 Chen, Zhong, and Hillard
(1997); Chen et al. (1997);
Zhong, Chen, and Scheb
(1998); Chen and Zhong
(1999, 2000, 2002); Chen
(2004)
9 1996 Hubei, Shaanxi, 160† Oi and Rozelle (2000)
Shandong, Sichuan,
Zhejiang
10 1996 4 counties: in Anhui, 1,414 Eldersveld and Shen (2001);
Hebei, Hunan, Tianjin Jennings and Zhang
(2005); Manion (2006);
Jennings and Chen (2008)
11 1996 Beijing 895 Shi (1999b)
12 1997 Beijing 694 Zhong, Chen, and Scheb
(1998); Chen (1999, 2000,
2004); Chen and Zhong
(2000)
13 1997, 1999 8 counties: in Hebei, 754 Dickson (2002, 2003)
Hunan, Shandong,
Zhejiang
14 1998–99 Beijing, Chengdu, 1,543 Tong (2007)
Guangzhou, Lanzhou,
Shanghai, Shenyang
(continued)
A Survey of Survey Research on Chinese Politics 187

Table 10.1. (continued)

Year Localities Surveyed Size Publications Associated

With Survey
15 1998–2004 Beijing 551–757 Johnston (2004, 2006);
Johnston and Stockmann
(2007)
16 1999 87 village small 1,356 Li (2002)
groups: in 25
provinces
17 1999 1 county: in Jiangxi 400 Li (2003)
18 1999 Beijing 670 Chen (2001, 2004)
19 1999 Chongqing, Guangzhou, 1,820 Tang (2001a, 2005)
Shanghai, Shenyang,
Wuhan, Xi’an
20 1999 4 counties: in Anhui, 2,400 Zweig and Chung (2007)
Heilongjiang
21 1999–2001 4 counties: in Fujian, 1,600 Li (2004)
Jiangsu, Jiangxi
22 2000–2001 Rural Shaanxi 306 Kennedy (2002); Kennedy,
Rozelle, and Shi (2004)
23 2000 Rural Jiangsu 1,162 Zhong and Chen (2002);
Chen (2005a, 2005b)
24 2000 Jiangsu, Shanghai, 1,625 Wang, Rees, and
Zhejiang Andreosso-O’Callaghan
(2004)
25 2001 8 counties: in Fujian, 316† Lily Tsai (2007a, 2007b)
Hebei, Jiangxi, Shanxi
26 2001 2 counties: in Fujian 913 Rong (2005)
27 2002–2003 National probability 1,525 Kellee Tsai (2007)
sample
28 2002–2003 12 villages: in Anhui 1,503 Tan and Xin (2007)
29 2003–2005 2 counties: in Fujian, 800 Li (2008)
Zhejiang
30 2004 1 county: in Yunnan 700 Davis et al. (2007)
31 2004 Beijing 592 Chen, Lu, and Yang
(2007)
32 2005 8 counties: in Hebei, 1,337 Dickson (2007)
Hunan, Shandong,
Zhejiang
†
 Number of probabilistically selected villages in each of which surveyors interviewed a few
purposively selected informants.
Notes: National probability sample surveys exclude Tibet. Except for national probability sam-
ples, “localities surveyed” refers to nonprobabilistically selected localities within which survey-
ors used probability sampling selection methods. Except for informant surveys, “size” refers to
number of completed questionnaires.
188 Melanie Manion

ten were conducted with self-administered questionnaires, typically with a

survey team member present to answer queries.12
Six of the surveys have a longitudinal component, of considerable value to
our understanding of Chinese politics. One is a quasi-experimental design: sur-
vey 17 is actually a preelection and postelection survey of the same villagers.
For the most part, however, the longitudinal design reflects an effort to gauge
the impact of a rapidly changing social, political, and economic environment.
Surveys 3 and 10, discussed later in the chapter, are pairs in a panel study,
an effort to reinterview respondents in order to track change over time at
the individual level of analysis. The remaining four are longitudinal cross-
sections: they return to the same localities, with essentially the same question-
naire, but survey a new representative sample of the population. The Beijing
Area Study (BAS), which is discussed below and in Chapter 13, is the most
important of these if only because it tracks change annually. Survey 11 returns
to the Beijing mass public in 1996 to ask roughly the same questions about
political participation as asked in survey 2. Survey 32 returns to the same
eight counties six years later to survey local officials and private entrepreneurs
on the same issues. Surveys 8, 12, and 18 of the Beijing mass public contain a
common core of questions on popular political support for the regime, incum-
bents, and policies.
Also, several studies systematically survey and compare two different
populations: local government administrators and the mass public in Beijing
(Chen, 1999), villagers and local officials at various levels in the countryside
(Eldersveld and Shen, 2001; Jennings, 2003; Manion, 1996), private entre-
preneurs and local officials (Dickson, 2003, 2007, 2008), villagers and vil-
lage committee members (Chen, 2005b), villagers and village leaders (Tan and
Xin, 2007; Zweig and Chung, 2007), and urban industrial enterprise manag-
ers and local environmental sector bureaucrats (Tong, 2007).
Of the surveys presented in Table 10.1, five stand out as achievements
of particular note, milestones in survey research on Chinese politics. They
include surveys 2, 4, and 6, supervised by Tianjian Shi, each pathbreaking in
its own way.
Survey 2, conducted in Beijing and initially conceived as the pretest for
a nationally representative survey, is the first large-scale probability sample
survey of the Chinese mass public.13 The vision underpinning its design is
12
Self-administered surveys were used in three samples of elites: retired officials in survey 1,
local officials and private entrepreneurs in surveys 13 and 32, and local bureaucrats and
enterprise managers in survey 14. Self-administered questionnaires were also used to survey
the mass public in Beijing in surveys 8, 12, 18, and 31; in rural Jiangsu in survey 23; and in
rural Anhui in survey 28. Considering the relatively low effective literacy and general unfa-
miliarity with survey instruments in mainland China, self-administered questionnaires are
probably better suited to more literate samples.
13
The national survey was precluded by the events of June 1989. The mainland partner was
an independent survey research center, newly established under the Beijing Social Economic
Research Institute headed by Chen Ziming and Wang Juntao, both later charged and impris-
oned as the “black hands” behind the 1989 protests. Shi flew out of Beijing on June 10, 1989,
with the pilot study questionnaires.
A Survey of Survey Research on Chinese Politics 189

also distinctive and controversial for its time. It is a bold vision of “normal
science” that poses as an empirical and testable question the view of the pas-
sive citizen in authoritarian China. Its questionnaire borrows heavily from
a classic study of political participation in comparative politics (Verba, Nie,
and Kim, 1971) and investigates activities easily dismissed at the time as irrel-
evant (voting), apolitical (workplace cronyism), or too sensitive to discuss with
strangers (strikes, demonstrations, boycotts).
Survey 4, supported by the National Science Foundation, is the first nation-
ally representative survey on Chinese politics. Indeed, as Nathan and Shi
(1993) point out, it is not only “the first scientifically valid national sample
survey” explicitly focusing on political behavior and attitudes conducted on
the mainland but the first such survey ever conducted in a Communist coun-
try. Again, the design is prompted by a classic survey study in comparative
politics (Almond and Verba, 1963), this time with a focus on political cul-
ture. An underlying question is the relationship between Chinese culture and
democracy. In what sense does Chinese culture pose an obstacle to democra-
tization? In the event of multi-partism, what does the distribution of values
across Chinese society suggest about how preference aggregation might create
opportunities for political parties?
Survey 6 borrows the vision of cross-national comparative survey work and
adapts it to study political culture in three Chinese polities, asking essentially
the same questions of Chinese on the mainland, in Taiwan, and in Hong Kong.
This is a complex collaborative project, involving nine principal investigators
based in the United States, Taiwan, and Hong Kong. It also distinguishes itself
from surveys 2 and 4 by the greater attention, in the questionnaire itself, to the
possibility of Chinese exceptionalism. The concept of political culture investi-
gated in the survey prominently includes features of Confucianism, in particu-
lar a hierarchical orientation to the moral state. The mainland survey includes
a large enough urban subsample to permit statistically valid comparisons with
the Taiwan and Hong Kong samples.
A fourth pioneering effort is survey 3, the first probability sample survey on
politics in the Chinese countryside.14 Other new features of this study are its
survey of both local elites and the mass public, using some of the same survey
items to permit comparisons; a panel component that returned to the same
localities to reinterview the same respondents in 1996; and a survey team that
included several American political scientists for whom the study was their
first foray into the study of Chinese politics.15 This is very much an omnibus
survey, the product of negotiation among ten Chinese and American scholars
with a variety of research agendas.
14
Full disclosure: I participated in this survey project. Politics in the post-1989 conservative
interregnum intruded to create unusual setbacks for it. See note 5 above.
15
For example, Jennings authored (1997, 1998a, 2003) and coauthored with his graduate stu-
dents (Jennings and Zhang, 2005; Jennings and Chen, 2008) several articles drawing on the
survey data and published in top-tier journals in both political science and Chinese area stud-
ies. Eldersveld coauthored a study with Mingming Shen (Eldersveld and Shen, 2001), the key
force in the survey among mainland participants.
190 Melanie Manion

Finally, surveys 7 and 15 represent the BAS surveys, described in Chapter

13 of this volume. Begun in 1995, the BAS is an annual survey of the Beijing
mass public. It is modeled on the Detroit Area Study and the Chicago General
Social Survey and has similar aims: to gauge change and stability over time
across a broad range of social, economic, and political dimensions and to
understand their correlates. In addition to the basic questionnaire, a novel
feature of the BAS is the inclusion of modules of question batteries provided
by nonmainland scholars. The foreign policy questions explored by Johnston
(2004, 2006) and Johnston and Stockmann (2007) draw from such a module,
for example.
In addition to these five milestones, two recent surveys not listed in Table
10.1 are pathbreaking for their nationwide sampling of geographic space
using Global Positioning System (GPS) technology. The spatial sampling used
in these surveys produces representative samples, including representative
samples of the more than 100 million migrants living in Chinese cities, who
are typically missed with conventional methods based on household registra-
tion lists.16 First to use this spatial sampling technique for all China (including
Tibet) is a 2004 survey on the institutionalization of legal reforms, described
at greater length in Chapter 12 of this volume. Second is a 2008 omnibus
multidisciplinary survey organized by Texas A&M University and designed
to be a regular multiyear effort. In addition to its use of spatial sampling, this
survey is noteworthy because preparation included a broad invitation to the
scholarly community to contribute questions to the core questionnaire. This
feature is modeled on the American National Election Study and may be an
important step in building a larger survey research community.17 As neither of
the two surveys had yielded publications by mid-2008, I do not include them
in Table 10.1. Clearly, however, spatial sampling inaugurates a new generation
of survey research in mainland China, well adapted to ongoing demographic
changes.18
All of the surveys listed in Table 10.1 are the product of collaboration with
mainland partners, a relationship that ranges widely from full partnership to
subcontracting. As shown in Table 10.2, a few institutions dominate among
mainland partners. First is Peking University, especially its Research Center on
Contemporary China (RCCC), which has partnered with nonmainland politi-
cal scientists for the greatest number of surveys.19 This reflects (and builds) its
reputation as probably the most competent academic survey research agency

16
For a description of spatial sampling using GPS technology and its advantages over conven-
tional sampling in mainland China, see Chapter 12 in this volume and Landry and Shen
(2005).
17
Questions from eleven scholars outside the core team of survey researchers were added to the
survey instrument. Full disclosure: I am a member of the board for the Texas A&M China
Survey.
18
Articles analyzing data from the legal reform survey can be found in a special issue of China
Review, vol. 9, no. 1 (Spring 2009).
19
The RCCC also conducted the mainland component of the World Values Survey in 2000.
A Survey of Survey Research on Chinese Politics 191

Table 10.2. Mainland partner institutions for probability sample surveys on

Chinese politics

Mainland Partner Survey

Peking University, Department of Political Science 1, 11
Peking University, Research Center on Contemporary 3, 7, 10, 13, 14, 15, 19, 32
China
People’s University, Social Survey Research Center, 4, 6
Beijing
People’s University, Public Opinion Research Institute, 8, 12, 18
Beijing
Beijing Social Economic Research Institute, Opinion 2
Research Center of China
Economic System Reform Institute of China 5
Northwest University, Xi’an 22
East China University of Politics and Law, Shanghai 24
Chinese Academy of Social Sciences, Private Economy 27
Research Center
Chinese Academy of Social Sciences, Research Group, 31
Mass Participation in Community Residents
Committees
Unspecified 9, 16, 17, 20, 21, 23, 25,
26, 28, 29, 30
Note: Survey numbers refer to surveys listed in Table 10.1. “Unspecified” means either no main-
land partner institution is specified in publications associated with the survey or reference to the
mainland partner institution is nonspecific.

on the mainland. It also reflects substantial ties between former classmates at

Peking University and the University of Michigan that link survey research-
ers based on the mainland and in the United States.20 People’s University also
stands out as a frequent mainland partner.
For the most part, survey researchers have not worked with nonacademic
marketing firms, although such firms are plentiful and some have worked on
prestigious American surveys (e.g., Pew surveys by Horizon). Although com-
mercial firms may be subject to fewer regulatory requirements,21 they may be
wary of risking their livelihood with surveys on explicitly political topics. The
reliance on academic partners probably also reflects existing substantial ties
and common scholarly interests. Further, commercial firms can be less col-
legial (i.e., transparent) about sharing with clients details of sampling design
and survey implementation; this makes it difficult for survey researchers to
evaluate the quality of the data provided.
20
Mingming Shen, RCCC founder and director and Peking University graduate, earned his
doctorate in political science at the University of Michigan. Classmates at Peking University
include Tianjian Shi, Wenfang Tang, and Yanqi Tong. Classmates at the University of
Michigan include Dickson, Johnston, and Manion.
21
See note 9 above.
192 Melanie Manion

It is also worth noting that the mainland partners remain unspecified

for more than a third of the surveys listed in Table 10.1. This may reflect a
partnership that does not meet all of the official standards of the Chinese
regulatory regime for survey research. Where mainland partners remain
unspecified, notes in methodological appendices sometimes refer vaguely to a
“mutual agreement” or “political sensitivities.” Whatever the cause, the lack
of transparency is unfortunate. It can be difficult to gauge the quality of sur-
vey research based on the scant information provided in many of the journal
articles listed in Table 10.1.22 As anonymity provides protection, identification
provides accountability. Accountability for the quality of the work is particu-
larly important to the maturation of the relatively new field of survey research
on Chinese politics. Further, mainland partners protected from the vagaries
of politics are at the same time deprived of the opportunity to build their
reputations.
Arrangements with mainland partners may also partly account for the
dearth of publicly available datasets from the surveys listed in Table 10.1.
Only three have been publicly archived – from surveys 2, 3, and 6.23 The
dataset from survey 25 is available on the author’s Web site. Datasets from
surveys 16, 17, and 21 are available from the author upon request.24 Not sur-
prisingly in this context, analysis of existing datasets remains rare. In my
survey of the literature, I discovered only a few studies analyzing data from
either these datasets or the mainland data from the publicly archived World
Values Surveys and Asia Barometer.25 In short, the norm for survey research
on Chinese politics is for researchers to be involved in every part of the proj-
ect, from survey design to analysis and write-up. To the extent that there is a
division of labor, it is reflected in actual survey implementation by mainland
partners. As political scientists increasingly receive their survey research fund-
ing from agencies that expect a commitment to public availability of data after
a specified time, this state of affairs may change. The routine public archiving
22
Descriptions of methods in monographs are generally much better. Normal length guide-
lines for journal articles undoubtedly encourage sacrificing methodological for substantive
elaboration.
23
Data from survey 3 are archived at the Interuniversity Consortium for Political and Social
Research at the University of Michigan. Data from surveys 2 and 6 are archived at the China
Archive at Texas A&M University. Data from the Taiwan and Hong Kong surveys that com-
plement mainland survey 6 are archived at National Taiwan University.
24
Of course, even where researchers do not publicize the availability of their data, they may
make it available upon request. For example, Kent Jennings and I have made available data
from the 1990 survey of local officials that complement the rural mass public survey in survey
3. Wenfang Tang has made available data from surveys 5 and 19, as well as data from surveys
conducted by the Economic System Reform Institute of China in 33–44 cities in 1987–89 and
1991, analyzed in Tang and Parish (1996, 2000) and Tang (2001a, 2005), and from a survey
conducted by the All-China Federation of Trade Unions in 1997, analyzed in Tang (2005)
(personal communication, June 12, 2008).
25
Guo (2007) and Hoddie (2008) analyze data from survey 6. Tang (2005) and Wang (2007)
analyze data from the World Values Surveys. Chang, Chu, and Tsai (2005) and Zhengxu
Wang (2006, 2007) analyze data from Asia Barometer.
A Survey of Survey Research on Chinese Politics 193

of survey data will be one marker of the maturation of survey research on

Chinese politics.

Content and Cumulativeness

What are the main topics that consume the various products of survey research
on Chinese politics? What valuable knowledge about Chinese politics has
survey research contributed? When survey researchers examine the same or
similar topics, do they explicitly engage one another? To what extent are they
building a cumulative knowledge?
Of all the studies identified in Table 10.1, a conservative estimate would
place more than half in the category of democratization. The single most pop-
ular topic in survey research on Chinese politics is village elections, also popu-
lar in qualitative research. It figures in some of the earliest survey studies and
continues to thrive. It has garnered the attention of the widest range of survey
researchers. For these reasons, a brief review of survey research questions and
findings on village elections is a good point of departure.
A few studies (Oi and Rozelle, 2000; Shi, 1999a) seek to explain the sub-
stantial variation throughout the country in implementation of the 1987
Organic Law on Village Committees. With grassroots democratization as
the dependent variable, these studies measure village electoral democracy as
electoral contestation. A greater number of studies inquire into the impact of
village electoral democratization. These studies also develop more nuanced
measures of electoral quality, conceived as a bundle of variables, including
but not restricted to contestation. Although they focus attention on different
components of the relationship, their findings are all broadly consistent with
the proposition that electoral mechanisms strengthen elite-mass connections.
Specifically, village elections of higher democratic quality are associated with
greater congruence of views between villagers and village leaders (Jennings,
2003; Manion, 1996); a stronger likelihood of villager appeals through village
leaders (Li, 2002); villager satisfaction with the electoral process (Kennedy,
2002); higher levels of external efficacy (Li, 2003); perceptions that village
leaders are fair (Kennedy, Rozelle, and Shi, 2004) and trustworthy (Manion,
2006); and greater villager political interest, awareness, and optimism (Tan
and Xin, 2007). This is a fairly robust accumulation of evidence across sur-
veys conducted at different times in a variety of localities by a range of schol-
ars employing different measures. The overall conclusion is not particularly
surprising, however.
Other studies illustrate not only that survey researchers actively engage one
another in building a cumulative knowledge of Chinese politics but also that
not all survey research on Chinese politics is simply confirmatory. It can yield
surprises that contribute importantly to our basic repository of knowledge.
Three examples illustrate this fairly well.
Robust evidence attesting to regime support provides a first example.
Findings based on surveys of the Beijing mass public at three points in the
194 Melanie Manion

1990s produce the same surprising conclusion: despite only moderate sup-
port for government policy performance, the regime enjoys broad and strong
legitimacy, based on the elitism and authority orientation of ordinary Chinese
(J. Chen, 2004; Chen, Zhong, and Hillard, 1997; Zhong, Chen, and Scheb,
1998). Similarly, examining data from the first nationally representative survey
on Chinese politics, Nathan and Shi (1996) find that only a small number sup-
port an end to party leadership, even among Chinese who support democracy –
a concept that resonates more with traditional notions of leadership than with
pluralism. Shi (2001) finds a similar result in his comparative study of political
trust, analyzing data from somewhat later representative surveys of Chinese
on the mainland and Taiwan: political trust on the authoritarian mainland is
strong and based on traditional values, compared to trust in democratizing
Taiwan, where it is more contingent on government performance.26
A second example draws from survey research on the Chinese business-
state relationship. Dickson (2002, 2003) reveals that Chinese capitalists nei-
ther possess beliefs nor engage in activities that constitute a challenge to the
regime. Theories that identify them as likely activists confronting the ruling
Communist Party are not well founded. Moreover, “red capitalists” have not
become less embedded in the current political system over time (Dickson,
2007, 2008). Drawing from nationally representative survey data, Kellee Tsai
(2007) reaches the same conclusion: Chinese private entrepreneurs do not con-
stitute a politically assertive class (or share a common class identity at all) that
poses a challenge to the state – nor are they a likely source of contestation in
the near future.
A final example originates in the survey research on political participation.
As Shi (1997) demonstrates in his early study, political involvement by ordi-
nary Chinese is wide ranging and intensive, not simply formalistic. Further, its
intensity and range are increasing (Shi, 1999b). Jennings (1997) confirms that
this picture of frequent, varied, and autonomous acts of political participation
also extends to the Chinese countryside.

An Example of Competing Perspectives

The examples above show that survey researchers engage one another in
building a cumulative knowledge. One of the more lively examples of survey
researcher engagement unfolded in two Journal of Politics articles present-
ing competing theoretical frameworks, hypotheses, empirical tests, and find-
ings on the same question: why do Chinese vote in local congress elections?
The presentations by Chen and Zhong (2002) and Shi (1999c) are particularly
interesting because the debate is fundamental, clear, and explicit. As such, the
engagement merits particular discussion here.

26
Related but not exactly similar is the finding of Lianjiang Li (2004) in a comparison of politi-
cal trust in different levels of state power: there is substantially and significantly higher trust
in the center than in lower levels.
A Survey of Survey Research on Chinese Politics 195

The articles roughly agree in their observations and characterization of

partial electoral reform: local congress elections are semicompetitive, but in
an authoritarian political context. In both articles, Chinese voters are rational,
pursuing their interests as best they can under the circumstances. Both articles
present multivariate analyses of voter turnout in local congress elections and
include most of the same socioeconomic variables. Shi tests his hypotheses
with data from a 1990–91 nationally representative sample survey, Chen and
Zhong with data from a December 1995 survey of Beijing residents.
Most interestingly, the articles begin with fundamentally different per-
spectives on contemporary Chinese politics, reflected in completely different
hypotheses on voting behavior, given partial reform. The crux of the different
perspectives has to do with how much institutional change is required to pro-
duce behavioral change. For Shi, a little institutional change is expected to go
a long way: the relevant contrast for Chinese voters is with the noncompetitive
elections of the past. Electoral choice among candidates, although limited,
is the relevant institutional change. For Chen and Zhong, the key contrast is
with a nonexistent liberal democracy. Despite electoral choice among candi-
dates, current authoritarian institutions have not dislodged Communist Party
electoral management and ideological dominance; in this context, the only
important new choice for Chinese voters is the choice to abstain from voting.
In sum, Shi hypothesizes that the choice to vote will reflect the vote’s new rel-
evance in elections with choice, despite the authoritarian context; by contrast,
Chen and Zhong hypothesize that the choice to vote will reflect the vote’s
continued irrelevance in the authoritarian context.
Probably the hypotheses and findings of greatest interest have to do with
the impact on voting of democratic orientation and political efficacy – which
point in different directions in the two articles. Chen and Zhong hypothesize
and find that Chinese with stronger democratic values and sense of internal
political efficacy are less likely to vote in the (still authoritarian) local con-
gress elections, compared to those with weak democratic values and weak
internal efficacy. Shi hypothesizes and finds the opposite. The analyses also
yield different findings (each consistent with the different hypotheses) about
the impact on voting of anticorruption sentiment and socioeconomic variables
such as age and education.
These differences point to three issues for survey research on Chinese poli-
tics. First and most obviously, different scholars can begin at the same point
but formulate radically different hypotheses about the implications of what
they observe; these different hypotheses may both be plausible, even equally
plausible. Indeed, what is most interesting about the Chen and Zhong article
is their reframing of partial political reform: they present a different way of
seeing semicompetitive congress elections, which is completely at odds with
Shi’s framework, hypotheses, and findings. The second issue is less obvious
without a careful reading of the two articles: different frameworks are inter-
esting in the context of survey research because the rival hypotheses they yield
can be tested – but not unless they are set up in a true competition. In their
196 Melanie Manion

socioeconomic measures, the models are nearly the same (although Chen and
Zhong fail to include party membership), but the other variables are com-
pletely different in measurement and even conceptualization.27 Moreover, Shi
does not include a measure of democratic orientation (or regime support) in
his multivariate model: he includes it only in a bivariate analysis; as such, it
is not a strong enough empirical finding for a contrast. Of course, Chen and
Zhong do not pretend to replicate Shi’s model. At the same time, they miss
an opportunity for a persuasive contrast and greater cumulativeness by, for
example, presenting one model that mirrors Shi’s as closely as possible, given
the difference in measures.28
The third issue may be the most important. A true replication of Shi’s
model with the 1995 Beijing survey data might well produce very different
results from those presented in Shi’s article. These are very different popula-
tions: Shi’s nationwide sample is 70 percent rural, a far different population
from the more highly educated and politically savvy Beijing residents. Timing
is also important: for Shi’s respondents, the most recent congress elections
(in 1986–87 or 1988–89) occurred in a more liberal environment than those
recalled by Beijing residents in 1995. The more general point is the relevance of
context in thinking about relationships between variables and in interpreting
statistical results. The pace and unevenness of change in China requires us to
pay attention to geographic and temporal diversity. Some of the implications
of this have already been emphasized earlier: the value of longitudinal stud-
ies, for example. For large-scale surveys that span localities of different types,
it implies taking geographic diversity into account in our statistical models.
Finally, it serves as a caution against accepting findings as established facts
that require no further examination, even in different circumstances.

Survey Research and Area Studies

If the small community of survey researchers is indeed engaged in building
a valuable cumulative knowledge of Chinese politics, how broad is the com-
munity of consumers? Certainly, survey-based studies appear to have made
serious inroads in the discipline: roughly a dozen of the articles listed in Table
10.1 appear in top-tier journals of political science. Yet, how big is the divide
between survey-based studies and studies based on archival and qualita-
tive research methods, both with a longer tradition in Chinese area studies?
Twenty-six of the forty-six journal articles listed in Table 10.1 are published in

27
This is certainly true of internal efficacy and anticorruption support. For example, Chen and
Zhong use one question to measure internal efficacy: it focuses on workplace (not national)
politics; Shi uses both measures in an index of internal efficacy. Chen and Zhong ask for an
evaluation of anticorruption measures; Shi measures abuse of power by local leaders.
28
This would be simple: it would involve including party membership and leaving democratic
orientation and regime support out of the model; internal efficacy and anticorruption support
would still be slightly different measures, of course.
A Survey of Survey Research on Chinese Politics 197

journals that focus exclusively on China, East Asia, or Asia.29 This is hardly a
definitive answer to the questions posed above, but it does not suggest a chas-
mic divide. Nor does a close reading of the research products listed in Table
10.1 suggest a divide.
The overwhelming preponderance of survey-based studies explicitly situate
their research questions (and findings) within both the discipline of political
science and the field of Chinese area studies, regardless of publication out-
let. A few refer only to work in the qualitative tradition in Chinese studies,
ignoring other survey-based research on Chinese politics and also eschewing
a broader empirical and theoretical context. None ignore qualitative studies in
contextualizing their research, however. To the contrary, they typically draw
at some length from qualitative studies in setting out the research question,
justifying its importance, and interpreting statistical findings.
Engaging the qualitative research does not guarantee a readership among
qualitative researchers in Chinese politics. Ideally, in an area studies journal,
the account of the research and findings should be accessible beyond the com-
munity of scholars with sufficient training to assess the quality of the methods.
Most articles do not go the extraordinary lengths of Nathan and Shi (1993)
in attempting to educate readers about the relative merits of survey research,
but most do make an effort (perhaps with editorial prodding) to explain their
substantive findings clearly enough.
A more serious problem, to my view, is an overuse of descriptive statistics,
which have an inordinate impact because of their relative digestibility. Nearly
one-fourth of the research products listed in Table 10.1 are exclusively, essen-
tially, or mainly descriptive – without any multivariate model estimation or
inferential statistical analysis.30 Summary descriptive statistics are a useful
preliminary to analysis and also have inherent value when the survey data are
nationally representative or permit a valid comparison across time or popu-
lations. Many of the descriptive statistical presentations lack these features,
however. Even so, this does not reflect a restraint imposed (or self-imposed)
with an area studies readership in mind. Although a high proportion of these
presentations can be found in journals with an area studies focus, it is not the
case that these outlets shun more sophisticated work. Nineteen of the twenty-
six survey-based articles on Chinese politics appearing in area studies journals
present multivariate models and employ inferential statistics.
Finally, although survey-based studies engage the qualitative research, rela-
tively few integrate qualitative and survey research in a single project. That is,

29
Journal of Contemporary China dominates, with ten articles; China Quarterly is next, with
six articles.
30
Some of these present bivariate relationships, but most present only descriptive statistics along
a single dimension. More than half of these are fairly recent publications (i.e., 2005–2007),
so it is not that scholarship has simply been catching up in methodological sophistication. It
is worth noting that the standard (multivariate models or inferential statistics) I use here is by
no means a very high one.
198 Melanie Manion

if there is not a divide, there does seem to be a division of labor. Two ­welcome
recent exceptions are studies by Kellee Tsai (2007) and Lily Tsai (2007a,
2007b). Kellee Tsai integrates an original large-scale nationally representative
survey with some 300 in-depth qualitative interviews with local officials, pri-
vate entrepreneurs, and mainland researchers to describe and explain “infor-
mal adaptive institutions” in the relationship between Chinese capitalists and
the state. Lily Tsai works on a smaller scale in the Chinese countryside: from
a case study of a few villages in a single province, she develops a thesis on
the role of “solidary groups” in public goods provision, then collects admin-
istrative data through an informant survey of some 300 villages across four
provinces and subjects her proposition to statistical testing. Both works are
impressive examples of richly contextual research that successfully employs
multiple methods. Combining qualitative and quantitative research findings
is more difficult to realize in a journal article, given length restrictions, but
examples include studies by Manion (1991), Kennedy (2002), and especially
Lianjiang Li (2002, 2004).

Conclusion
What is the status of survey research on Chinese politics? This chapter points
to many reasons for optimism about the enterprise. Most important of all, the
products of survey research reflect a community that is actively engaged in
building a valuable cumulative knowledge of Chinese politics. Nor does this
community ignore the progress in our knowledge of Chinese politics gained
from qualitative fieldwork or nonsurvey quantitative work.
Survey research on Chinese politics is still a fairly young enterprise, how-
ever. Some important areas for improvement remain outside the range of
influence of American researchers. So long as the enterprise is subject to the
intrusiveness of a regulatory regime that reflects a wary authoritarian poli-
tics, it will be difficult to develop strong open working relationships in survey
research with Chinese colleagues at more institutions. This has several impli-
cations, none favorable. It obstructs the growth of accountability in survey
research and the emergence of reputations for excellence among more Chinese
institutions. Concerns to protect mainland partners are also a disincentive (if
not always an absolute barrier) to widespread and systematic sharing of data.
In turn, the dearth of publicly accessible datasets constrains the integration
of survey data more broadly into research on Chinese politics, limiting it to a
fairly small community of survey researchers.

Appendix 10.1: What Is Reviewed Here

For this chapter, I reviewed English-language monographs and peer-reviewed
journal articles authored by political scientists and published in nonmainland
sources through mid-2008 to identify studies that exploit original probability
sample survey data. This focus eliminates from consideration a substantial
A Survey of Survey Research on Chinese Politics 199

literature by sociologists and economists drawing from probability sample

survey data, including work that to varying degrees bears on Chinese politics
broadly defined. Nor do I review survey-based studies produced and published
on the mainland. Mainland Chinese political scientists have conducted a large
number of surveys, beginning in the 1980s. Nathan and Shi (1993) observe
that most such surveys on which they have information are methodologically
flawed. Surveys published in mainland sources appear to have improved in
quality, but descriptions of their methods are often too sparse to permit evalu-
ation. This is, of course, true of several survey-based studies published outside
the mainland too.
I focus in the chapter on original surveys. Researchers based outside main-
land China always work with mainland colleagues, but by “original survey”
I imply their participation in the decision-making process that shapes the
product in crucial ways. This includes choices about pretests, interviewers,
training, sampling, and data input. This decision rule was not always straight-
forward to implement. In particular, Wenfang Tang typically compares evi-
dence from original surveys and existing datasets in his work. For example,
Table 10.1 includes a 1992 forty-four-city survey and a 1999 six-city survey,
both of which he designed or played the main role in designing. The table does
not include three other surveys in which he did not participate in the decision-
making process as described earlier. At the same time, his access to these
datasets is a noteworthy event in the history of survey research on Chinese
politics. This is especially the case for the surveys conducted by the Economic
System Reform Institute of China, disbanded after 1989. A National Science
Foundation grant allowed Tang and Parish to recover the data and check
them against the original questionnaires. For work based on existing datasets
acquired by Tang, see Tang (1993, 2001a, 2005) and Tang and Parish (1996,
2000).
I invite readers to alert me to studies that do not appear in Table 10.1 but do
meet the criteria for inclusion noted above. Table 10.1 and its associated refer-
ences are updated regularly on my personal Web site at http://www.lafollette.
wisc.edu/facultystaff/manion/.
As discussed in footnote 24, few of the datasets are publicly archived, but
in my experience it is worthwhile to make a direct individual request to survey
researchers for access to their data. This can be an especially useful strat-
egy for graduate students to supplement qualitative fieldwork and archival
sources, for example.
11

Surveying Prospects for Political Change

Capturing Political and Economic Variation
in Empirical Research in China

Bruce J. Dickson

The potential political role of private entrepreneurs in China has been a salient
issue in recent scholarship on China. Some see private entrepreneurs as poten-
tial agents of political change; others see them as apolitical and even supporters
of the current political system. Although the political implications of China’s
rapid economic development have not yet been fully realized, many scholars,
politicians, and journalists anticipate not only that economic development is
leading toward democratization in China but also that private entrepreneurs
are likely to be key players in that process. They point to China’s growing
numbers of entrepreneurs and “middle class” as potential supporters or even
advocates of democratization (Parris, 1993; White, Howell, and Shang, 1996;
Zheng, 2004). Others see private entrepreneurs as the leading edge of an emerg-
ing civil society that will eventually transform China’s political system (Gold,
1998; He, 1997; Pei, 1998). These views were most prominent in the 1990s,
when economic privatization began in earnest. In contrast, empirically based
studies of the political interests and behavior of China’s capitalists reveal most
of them to be politically tied to the state or apolitical: very few exhibit strong
democratic beliefs, and few of them engage in political activities designed to
promote political reform (Chen, 2004; Chen and Dickson, 2010; Dickson,
2008; K. Tsai, 2007). These studies see China’s capitalists not as agents of
democratization but as key beneficiaries and supporters of the status quo.
Although much of the speculation about the potential role of privatization
in general and private entrepreneurs in particular in fostering democratization
in China is based on modernization theory, the comparative literature on the
role of capitalists shows that they are very ambivalent about political change.
They are rarely the leading edge of opposition to the authoritarian state, but
their support can tip the balance between whether the state can withstand
the challenge from democratic challengers or whether it will succumb to pres-
sure for regime change (Rueschemeyer, Stephens, and Stephens, 1992). In Eva
Bellin’s apt phrase, they are “contingent democrats”: to the extent that their
economic interests are dependent on the survival of the regime, they remain
supportive of the status quo (Bellin, 2000: 175–205).

200
Surveying Prospects for Political Change 201

These comparative insights are borne out by the most recent empirical
research in the Chinese context. On the one hand, the democratic movement
in China is weak and divided with seemingly little public support for the brave
efforts of individuals and fledgling groups that face unrelenting persecution
by the state. Capitalists are rarely first-movers in the democratization process;
but in China, they have virtually no one to support even if they were inclined
to do so. On the other hand, China’s capitalists have been the main beneficia-
ries of the Chinese Communist Party’s (CCP’s) economic reform agenda and
have little material reason to press for political change. As Kellee Tsai (2007)
has shown, the interests of China’s capitalists have been advanced by the CCP
without the need for sustained collective action on their part. In addition, they
are increasingly integrated into the political system, giving them easy access
to policy makers. Under these combined circumstances, we should not expect
that China’s capitalists would be acting as agents of change, and indeed they
are not.
This chapter will describe how the findings of comparative political research
can be applied and tested in the Chinese context. It will begin with a descrip-
tion of the design of my survey research project, which was meant to cap-
ture some of the most salient aspects of contemporary China: the relationship
between political and economic elites, but also divisions within each set of
elites; regional diversity; and the rapidly changing nature of the economy, soci-
ety, and political system. It will then provide some brief illustrations to show
how these types of variation are relevant for the research question involved. A
survey designed in this manner produces more reliable results than other stud-
ies that concentrate on a single locale or a single point in time.

Designing Survey Research

China watchers in the scholarly, journalistic, and policy-making communi-
ties often expect that China’s private entrepreneurs may be agents of politi-
cal change. These predictions make two assumptions: first, that the Chinese
Communist Party is passive in the face of social and economic changes; sec-
ond, that China’s private entrepreneurs have different policy preferences from
party and government officials and would press for greater reform, either from
within the state or from the outside. If private entrepreneurs are to serve as
agents of political change in China, then they should exhibit views on political,
economic, and social issues that are distinctly different from those of party
and government officials. Otherwise, their growing numbers and greater inte-
gration into the political system will be more likely to support the status quo
than to challenge it. Their increased integration into China’s political institu-
tions gives them the access they would need to press for change, but do they
desire change? If so, what kinds of change would they prefer?
These are substantive questions that require reliable empirical data to
confirm or challenge. According to most research done on these questions in
China, these assumptions do not stand up very well against empirical evidence.
202 Bruce J. Dickson

Rather than being the passive victim of economic and social change, the CCP
is actively integrating itself with the private sector to promote rapid growth,
thereby boosting its popular support and claim to legitimacy and preempting
a potential threat. Rather than being inherent supporters of democratization,
China’s capitalists exhibit views that are not always substantively different
from those of local officials, especially the ones with whom they interact the
most. They are generally not supportive of democratic activists to bring about
political change and they do not hold demonstrably democratic values (Hong,
2004; K. Tsai, 2007; Zweig, 1999). But these studies are not definitive. Most
are either based on a single city or industrial sector.1 Although insightful in
their findings and rich in their details, the findings from single site or single
sector studies cannot be generalized to the rest of the country with any degree
of confidence. The concern that their findings are idiosyncratic due to peculiar
features of their cases cannot be entirely dispelled, even if the findings seem
quite plausible. Moreover, all previous studies are based on a single point in
time and are therefore not able to assess how ongoing economic, political, and
social changes are affecting the relationship between the CCP and the private
sector, and the prospects for political change more generally.
An appropriately designed survey project can help address these potential
shortcomings. Over the last several years, I have been engaged in an ongoing
study of China’s capitalists, their basic political and social beliefs, their views
on economic policy, and their interactions with the state (Dickson, 2003,
2008). To arrive at more definitive answers, my survey was designed to cap-
ture the conceptual issues regarding the presumed link between economic and
political change and, equally important, to capture key aspects of the Chinese
context.

Regional Variation
Because China is a large and diverse country, the design of the survey had to
capture key aspects of regional diversity. The most important source of diver-
sity for my study was the local level of economic development. The effects
of economic development have been most strongly felt in the coastal prov-
inces; here the growth rates have been highest and the private sector devel-
oped early and in some areas became the main source of jobs, tax revenue,
and economic output. Conversely, inland provinces did not feel the effects
of economic reforms as early or as extensively as those on the coast. The
growth rates were lower, the standards of living lagged, and the private sector
was less developed. Accordingly, I selected three provinces (Hebei, Shandong,
and Zhejiang) and also selected one inland province (Hunan) as a control
group. Next, within each province I selected two counties or county-level

1
See Pearson (1994); Wank (1995); Unger (1996a); Nevitt (1996). The exception is Kellee Tsai’s
(2007) study, which is also based on a broad-scale survey that incorporates both regional and
sectoral variation.
Surveying Prospects for Political Change 203

cities according to their level of economic development in the late 1990s when
the first wave of the survey was conducted: one relatively prosperous and one
relatively poor county was chosen in each province. (In all, three counties and
five county-level cities were selected; one of the county-level cities had become
a district of a ­prefecture-level city by the time of the second wave of the survey
in 2005.) Within each of these counties, three townships were selected where
the private sector was relatively well developed for that particular county.
This purposive selection strategy has advantages over the more conven-
tional randomly selected sample. First, with only four provinces and eight
counties in the sample, a strictly random selection process might not capture
the economic diversity relevant to the research question at hand. Of course,
increasing the number of sampling units would alleviate this problem, but pri-
marily for budgetary reasons, this was not feasible. The selection of townships
where the private sector was relatively well developed was also done intention-
ally: because the research question dealt with the role of private entrepreneurs
in political change, I had to ensure that there was an adequate number of
private entrepreneurs in a given township. If a random sample would have
selected a township with a small or virtually nonexistent private sector, there
would have been little for me to study.

Individual Variation
The final stage of selecting the sample concerned what individuals would be
interviewed. The survey targeted two groups of people: the owners of rela-
tively large-scale private enterprises and the local party and government offi-
cials with responsibilities over the economy; in other words, the economic and
political elites of their local communities.
If capitalists are to be agents of change, a logical assumption is that it
would be the relatively large-scale enterprises and not the smaller-scale mom-
and-pop shops (individually owned enterprises, or 个体户) that would be most
relevant. In other countries, it was large firms that were most politically influ-
ential, both as individuals and as members of business associations. Smaller
firms in China, as in other countries, have less political clout and larger col-
lective action problems, and therefore are less likely to be drivers of democ-
ratization. With that assumption in mind, the criterion for including firms
in the sample was that their sales revenue during the previous year had been
at least 1 million yuan. In some counties, this threshold had to be relaxed
because there were not enough large firms to qualify. This was true not only
for relatively poor counties but also for one county in Hebei that had a large
private sector, though it was comprised primarily of small and medium-sized
firms. In the 2005 survey, the 1 million yuan threshold was relaxed and the
size of the sample was increased by about twofold to get a slightly wider range
of entrepreneurs.
The final selection of which entrepreneurs to include in the sample was based
on lists of registered private firms provided by the industrial and commercial
204 Bruce J. Dickson

bureaus of the county governments. I restricted the sample to only private firms
because the theoretical literature links privatization and democratization, and
because comparative studies of capitalists and democratization concentrate
on the role of non-state-owned enterprises. Given the changing economic
landscape in China, it may no longer be valid to exclude other types of firms
from research on the political impact of privatization. State-owned enterprises
(SOEs) increasingly have to operate in a market environment, and the ongoing
restructuring of SOEs has created firms of hybrid ownership, such as joint-
stock companies, limited liability corporations, and joint ventures. Many of
these firms are a mix of public and private ownership and domestic and foreign
investment. Most studies, like mine, do not investigate the political impact
of these types of firms, which may lead us to overlook an important ongoing
dynamic. I also restricted the sample to officially registered firms. Although
Kellee Tsai estimates that as many as 15.3 percent of private firms are unregis-
tered, they tend to be very small-scale operations, such as street vendors (K. Tsai,
2007: 109). In assessing capitalists’ willingness and likelihood of being agents
of change, excluding unregistered firms was not likely to skew the results.
To select the sample of private entrepreneurs to be surveyed, these officially
registered private firms were ranked according to their reported sales revenue
and selected using a random start, fixed interval process. For example, if there
were fifty firms in a township that met the sample requirements – registered
as a non-state enterprise and over 1 million yuan in sales – and we wanted ten
firms to be selected for the sample, the survey team would pick the first firm at
random and then select every fifth firm on the list. This guaranteed variation
in the level of firm size. Because most private firms are small and medium-
scale, selecting firms totally at random rather than from a ranked list ran the
risk of not including any large-scale firms in the sample. (On a more practi-
cal level, it also prevented local officials from handpicking firms they wanted
included [K. Tsai, 2007: 63–64].)
The second group of people included in the surveys was local party and
government officials. A purposive instead of random selection was again used.
The officials most relevant to the question of whether economic change and
privatization were creating pressure for political change were the top execu-
tives and those with direct responsibility over the private sector. On the CCP
side, this included party secretaries and heads of organization and united
front departments; on the government side, it included county magistrates,
heads of the industrial and commercial management bureaus, and the local
chapters of the official business associations (All-China Federation of Industry
and Commerce, Private Enterprise Association, and Self-Employed Laborers’
Association). In each county, the people holding these posts at the county level
were interviewed, and officials in townships and villages where the enterprises
in the sample were located were also interviewed. In each county, approxi-
mately thirty officials were selected.
These specific officials were selected because they had primary responsibil-
ity over the local economy and were most likely to have regular interactions
Surveying Prospects for Political Change 205

with local capitalists. They shared the common interest of promoting eco-
nomic development, so surveying these particular officials and capitalists
would indicate a degree of similarity in their views on economic, social, and
political issues. If China’s capitalists are to be agents of political change, then
those views should be quite different; conversely, if the views were quite simi-
lar, and if the degree of similarity was becoming stronger over time, then the
potential for China’s capitalists to promote political change would be small.
These two groups – capitalists and party and government officials – are
rather broad, and it would be wrong to assume that either group is uni-
form in its beliefs. For analytical purposes, I was interested in two types
of variation within the two groups. For capitalists, the key source of varia-
tion was their relationship to the CCP. The main distinction was between
“red capitalists” – capitalists who were also CCP members – and capitalists
who did not belong to the CCP. I further divided these two groups to get
a more nuanced sense of the relationship to the CCP: among the red capi-
talists, I distinguished between those who had been party members before
going into business (whom I dubbed xiahai red capitalists, because 下海 is the
Chinese term for going “into the sea” of the private sector) and those who
were co-opted into the CCP after being in business; among non-CCP capital-
ists, I distinguished those who wanted to join the CCP from those who were
not interested in joining. The result was a four-level variable that measured
the degree of embeddedness in the party-state: xiahai red capitalists had the
longest and presumably closest relationship with the CCP, followed by those
who had been co-opted into the CCP, those who were not yet CCP members
but wanted to be, and finally those who expressed no interest in joining the
party. The reason for these distinctions was straightforward: some scholars
have argued that capitalists will exert pressure for political change from out-
side the state, others that capitalists will work from within to try to bring
about change. Asking whether the capitalists as a whole would support polit-
ical change seemed too broad; it was more important to identify variations
within this group and their implications.
The category of “local officials” was also too broad for analytical pur-
poses.2 It was necessary to disaggregate county from township and village
officials for two reasons. First, bureaucratic level is a key determinant of the
organization of political power in China, as it is in most countries. There
is little reason to assume that officials at different levels will hold the same
views just because they are agents of the state. In a variation of the “where
you stand depends on where you sit” view of bureaucratic politics, the level
of appointment is likely to influence officials’ views on various economic and
political issues. The second reason to distinguish county from township and

2
In my book, Red Capitalists in China, I did not inquire into differences among officials, and
therefore missed an important distinction, as examples later in this chapter will demonstrate.
This oversight was duly noted by Jennings (2003). It became a prominent theme in Wealth into
Power (2008), which compares the results of the 1999 and 2005 surveys.
206 Bruce J. Dickson

village officials concerns their proximity to the capitalists in their communi-

ties. Township and village officials are likely to have more regular interactions
with the capitalists in their communities, and perhaps even ties of friendship
and kinship. The degree of similarity of views between officials and capitalists
may also be influenced by the degree of proximity: capitalists may have more
in common with township and village officials than they do with county offi-
cials. As will be shown later in this chapter, these distinctions among capital-
ists and officials not only allow for greater conceptual precision but they are
also empirically important.

Temporal Variation
In assessing the prospects for political change, China’s rapidly changing
economic, social, and political environments must also be considered. This
requires observations at more than one point in time. Although this project
was not originally conceived as a time-series study, the opportunity to do a
second wave of the survey is one of the factors that distinguishes it from oth-
ers on the same topic. Indeed, most survey research captures a point in time,
leaving open the question of whether the findings are still relevant by the time
the results are published. The first survey was completed in 1999, but the
book based on the survey data was not published until 2003. In the interven-
ing years, the private sector had experienced tremendous growth, the level of
political protests – in part driven by the strategy of relying on the private sec-
tor to create economic growth – also grew, and the top leadership of the party
and government had been replaced. How would these major changes affect
the relationship between the CCP and the private sector?
To answer these questions, a second wave of the survey was completed in
2005. The same eight counties made up the 2005 survey, regardless of their
current level of development, allowing me to observe trends in the same com-
munities over time. Although the local rates of economic development varied,
the rank ordering of the counties by level of development was nearly identical,
with the exception of the sixth and seventh (poorest) counties, which reversed
places. This was not a panel study. No attempt was made to identify the same
respondents who participated in the first wave of the survey. In part, there was
no theoretical need: I was concerned with changes in the relationship between
the CCP and the private sector over time, not whether individual capitalists
and officials had changed their minds on specific issues. In addition, the rapid
turnover of party and government officials and the opening and closing of
firms made a panel study infeasible.
As will be shown below, the second wave of the survey identified beliefs and
values that had changed in light of ongoing events and the changing policy
environment. At the same time, it also identified other areas of interest that
remained stable over time, giving greater confidence that the survey was tap-
ping into relatively stable beliefs and that the responses given were not ephem-
eral in nature.
Surveying Prospects for Political Change 207

Collecting Data
Once it was decided who would be interviewed and where, the next task was
to determine what would be asked. The questionnaire was written, revised,
and finalized over the summer of 1997. The timing was an important con-
sideration for several reasons. First, the CCP’s relationship with the private
sector was sensitive and controversial. At the time, a formal ban on recruiting
private entrepreneurs into the party remained in effect, even though it was
routinely violated at the local level. More generally, the propriety of the part-
nership between Communists and capitalists was repeatedly challenged by
party veterans, who believed this partnership was contrary to party traditions
and threatened to undermine the CCP’s hold on power. Getting the coopera-
tion of local officials and capitalists to research the nature of this relationship
therefore seemed risky. That risk was compounded by a second factor: the
political atmosphere concerning survey research in China at that time was
tense. The State Education Commission had recently issued a policy advi-
sory (or 精神) warning that all surveys by academic units should be approved
in advance and that foreign scholars (including Chinese scholars living and
working abroad) should not be directly involved in survey work.
The key to successful survey research in China is the local partner. In my
case, I was fortunate to work with the Research Center on Contemporary
China (RCCC) of Peking University. This is one of the premier survey research
teams in China; it is headed by Shen Mingming, who received his Ph.D. in
political science at the University of Michigan, widely recognized as one of the
top schools for survey research and methodology.3 It was Shen’s suggestion that
led me to do a survey of the CCP’s relationship with the private sector, and his
management of the actual implementation of the survey was crucial to its suc-
cess. He and the RCCC staff contacted local officials to get their cooperation,
visited each site in advance to do the groundwork, and then implemented the
actual survey in the eight counties. Given the recent warnings against foreign
scholars doing survey work in China, it would have been impossible to con-
duct this project without the RCCC as my partner. In later years, the political
atmosphere has become more hospitable to survey work, but the importance
of a local partner remains absolutely essential. Properly conducting a survey
involves interacting with multiple government offices, designing the sampling
frame, selecting individuals for inclusion in the sample, training numerous
people to implement the survey, and finally entering the data. Sloppy work at
any stage of the process will doom the project. Unlike interviews or archival
research, survey work cannot be done by individual scholars working on their
own, and mistakes and oversights can rarely be fixed with a return visit.
The questionnaire was finalized at the RCCC during the summer of 1997.4
Assembling the questionnaire involved multiple challenges, some common to
3
In the interest of full disclosure, I should note that Shen and I were classmates at Michigan.
4
The 1997 policy advisory against foreign scholars participating in survey research compli-
cated this process. In order to avoid implicating the center’s academic and administrative
208 Bruce J. Dickson

survey research, some specific to the Chinese context. First, the questions had
to reflect the theoretical questions involved but also had to be intelligible to
survey respondents in China. Second, the questions had to be worded care-
fully to avoid preprogrammed “politically correct” answers or, even worse,
no responses at all. There was no point in directly asking respondents if they
would prefer democracy to the current political system, even though that
was the ultimate research question. Posing the question so starkly would not
have elicited honest answers and likely would have led local governments to
refuse cooperation on the project at all. (In order to elicit their cooperation,
the project was titled “The Private Economy and Party Building.”) Third, the
questions had to make sense in China’s political climate of the time. Even the
phrase “political reform” presented a challenge. At the time the questionnaire
was written, the Chinese phrase for political reform, 政治改革, was politi-
cally sensitive. It was identified with former CCP General Secretary Zhao
Ziyang, who had been purged after the 1989 student demonstrations. Asking
respondents about their views on political reform ran the risk of cueing them
to think of Zhao Ziyang’s failed efforts to separate the party from government
in the late 1980s, which we did not intend. Instead, the more neutral phrase
“improve the political structure” (改善政治体制) was substituted. Similarly,
asking about official corruption was deemed too sensitive in the late 1990s,
although it became a common survey question in later years.
Separate questionnaires were created for the two groups of respondents –
capitalists and officials. Most of the questions were similar, or nearly so. For
example, capitalists were asked if foreign competition was a problem for them-
selves; officials were asked if foreign competition was a problem for entrepre-
neurs in their communities. Capitalists were asked whether they had made
charitable contributions; officials were asked if entrepreneurs in their com-
munities had made such contributions. A few questions were asked only of
each subgroup. For example, capitalists were asked about their sales revenue,
fixed assets, number of workers, and similar questions concerning their firms;
officials were asked about the relative importance of the state, collective, and
private sectors of the economy, priorities in party recruitment, and whether
they or their families operated a private firm.
When the second wave of the survey was done in 2005, the first question to
be decided was whether to use the original questionnaire or to revise it with
new questions. The advantage of keeping the original was that it would more
readily allow comparison over time, which was the purpose of the second
wave. The downside was that it would prevent gathering information about
new questions, either because I had not thought of them before or because

staff, I participated in the planning meetings not as the principal investigator but as a col-
league of the center’s director. It is not clear if anyone was fooled by this subterfuge, but the
uncertainty created by the new policy advisory made it seem necessary at the start of the
project. By the time the survey went into the field, it seemed safe to reveal my true role to
the survey staff, but not to the local officials who approved the project in their communities.
Surveying Prospects for Political Change 209

they had become more salient in subsequent years. In the end, the original
questionnaire was once again used, with only a minor revision: a set of ques-
tions was added regarding the entrepreneurs’ involvement in self-organized
business associations and their effectiveness relative to the officially sponsored
business associations.5
When the survey was actually administered, the questionnaires were self-
administered by the respondents under the supervision of the survey team
from the RCCC. The team also checked the identity of the respondents to be
sure they were the actual owners of the enterprise and not a family member or
manager. The respondents were promised anonymity, and their identities were
not recorded when the survey was conducted. Even the names of the counties
in the sample were not identified in the published works to protect the identi-
ties of the local officials who agreed to participate in the project. Because their
identities were anonymous, there was no need to encrypt or code the resulting
dataset. This guarantee of anonymity had a second and related benefit: it made
the study exempt from review by my university’s Internal Review Board (IRB)
for human subjects research. This is an often onerous and arbitrary process
that has bedeviled legitimate research, so the promise of anonymity not only
directly benefited my respondents but also indirectly benefited the research
project as a whole by saving the time and trouble of navigating the IRB.
At the time of the survey, the RCCC team also collected two other types
of data. The first came from meetings with the local party and government
officials. These were not structured interviews per se but group discussions
(座谈会) on questions regarding problems of party building in the private sec-
tor, the role of the official business associations, and related questions about
the party’s efforts to guide and manage the private sector at the local level.
The second type of additional data collected at the time of the survey was
aggregate county-level data about economic development and party member-
ship that were mostly not available in statistical yearbooks. These data were
necessary to get an objective measure of the importance of the private sector
to the local economy and the extent of the CCP’s relationship with the pri-
vate sector. This proved to be more troublesome than expected. Measuring
the importance of the private sector could be done by its share of GDP or
tax revenue, or the number of people employed in the private sector relative
to total employment, but not all counties had this information. Moreover,
many statistical yearbooks only report on firms “above a certain scale” and
therefore overrepresent SOEs and underrepresent private firms, which tend
to be small relative to SOEs. In the end, the only measure that all counties
could provide was the number of registered firms in the private, collective, and
state sectors. This is imperfect because it does not directly measure the size of
the private sector or its impact on the local economies, but it was the closest
measure available.6 Similarly, the CCP’s relationship with the private sector
5
This set of new questions was largely inspired by Kennedy (2005). See also Foster (2002).
6
One reason for the lack of systematic and consistent data on the private sector is terminology.
Chinese media and other sources often distinguish the state and non-state sectors, but the
210 Bruce J. Dickson

could be measured by the number of red capitalists (party members who were
also private entrepreneurs) or the number of firms with party organizations in
them. These questions were asked in the survey, but an objective measure for
the county as a whole would be useful as a reference point. However, although
some counties provided very detailed information, others did not. Even when
local officials provided what seemed to be very detailed information in the
group discussions, their information did not always match the aggregate data
provided by other local party or government offices. The consistency and
reliability of data is a problem for most empirical research in China, and it
certainly was for this project. On some variables, the information provided
by the local statistical office was different from the information in published
yearbooks or information provided by other offices. For a question as basic as
the size of the private sector, it is remarkable that systematic measures are not
readily available.7

Empirical Examples of Regional,

Individual, and Temporal Variations
In this section, I give the findings from the surveys to illustrate the importance
of capturing individual, regional, and temporal variations in survey design.
These examples focus on priorities among the goals of development, spe-
cifically the trade-off between promoting economic growth and maintaining
political stability, potential threats to social stability, and election to political
posts.

Competing Policy Priorities: Growth versus Equity

One of the most salient differences in development strategy is the trade-
off between economic growth and political stability. The need to preserve
stability amidst rapid economic development is a primary justification for
maintaining authoritarian rule, in China as well as in many other countries
(Huntington, 1987). This trade-off has become particularly salient in China
as the number of local protests has grown sharply in recent years. The num-
ber of public protests more than doubled during the period between the two
surveys, from 32,000 in 1999 to 87,000 in 2005 (McGregor, 2006). Many of
these protests have been a consequence of the strategy of rapid growth, such as
the conversion of agricultural land for industrial development, environmental
degradation, and official corruption. This growing threat to stability, and the

“non-state sector” not only includes private enterprises and getihu but also reformed SOEs
such as joint stock companies, limited liability corporations, and the like, where the state
continues to be a major investor. Precise and accurate definitions are essential for empirical
research, but the rapidly changing Chinese context often makes such precision and accuracy
hard to come by.
7
See Huang (2008: 13–24) for further discussion of the difficulty of accurately measuring the
size of the private sector.
Surveying Prospects for Political Change 211

Table 11.1. The trade-off between goals of development

among officials and entrepreneurs (percentages of those who
prefer growth over stability as the top goal)

1999 2005
All Entrepreneurs 41.7 44.6
Xiahai red capitalists 39.1 42.9
Co-opted red capitalists 29.9 47.3
Non-CCP, want to join 42.1 42.1
Non-CCP, don’t want to join 47.9 47.5
All Officials 60.6 49.1
County officials 76.2 59.3
Township and village officials 39.6 41.6

attention given to it by Beijing, should have helped change the views of local
officials, leading them to attenuate their support for growth at the expense of
stability.
The survey was designed to determine whether there were differences
between capitalists and local officials, as well as within both of these groups,
on this trade-off between growth and stability and whether these views
changed over time. Both officials and entrepreneurs were asked whether their
top priority was promoting growth or maintaining stability. One of the most
remarkable findings in the first survey was the apparent cleavage between offi-
cials and entrepreneurs on this trade-off, one of the few issues where the two
groups were “diametrically opposed” (Dickson, 2003: 132–134). However, as
shown in Table 11.1, disaggregating the officials by their bureaucratic level
reveals a more interesting story: the difference is not between officials and
entrepreneurs but between county officials and the rest of the respondents. In
the 1999 survey, county officials were most in favor of promoting growth as
their top priority: 76.2 percent favored growth over stability, almost double
the percentage of township and village officials and entrepreneurs. In the data
from the 2005 survey, a clear majority of county officials remained in favor
of growth over stability, even though the percentage dropped relative to that
in 1999. The numbers for township and village officials and entrepreneurs
showed a slight increase in terms of those who favored growth, but the major-
ity in these groups still had stability as their top priority. In both surveys,
differences across subgroups of entrepreneurs and between all entrepreneurs
and township and village officials are not statistically significant, but the dif-
ference between both groups and county officials is significant (at the 0.001
level). These findings are consistent with those of Kent Jennings and others
who have argued that the notion of the state needs to be disaggregated to be
most useful, in this case by dividing local officials according to their bureau-
cratic level (Jennings, 2003). In my previous study, I made much of the differ-
ence between entrepreneurs and officials on the trade-off between promoting
212 Bruce J. Dickson

growth and maintaining stability. By adding the question of bureaucratic

level, a more interesting and nuanced picture emerges. Although the differ-
ences between the groups narrowed during the time between the two surveys,
they remained substantively and statistically significant. The design of the
survey project made it possible to reveal these significant differences between
levels of the state, political and economic elites, and levels of development on
the goals of development.

Perceived Threats to Stability

The survey was also designed to elicit more specific information on what
respondents saw as potential threats to stability. All respondents were asked
whether increased competition and social pluralism were threats to stability.
Specifically, respondents were asked about their level of agreement with the
following statements:
●●
Competition between firms and individuals is harmful to social stability.
●●
If a country has multiple parties, it can lead to political chaos.
●●
If everybody does not share the same thinking, society can be chaotic.
●●
Locally, if there are many groups with different opinions, that can influence
local stability.
Responses are shown in Table 11.2, and several trends are worth highlighting.
First, on every question in both the 1999 and 2005 surveys, county-level offi-
cials have the lowest scores, indicating that on average they are less concerned
with threats to stability arising from economic competition and social plural-
ism than are township and village officials and all groups of entrepreneurs.
Second, on the three questions concerning increasing pluralism or diversifica-
tion, all groups of entrepreneurs saw less of a threat to stability in 2005 than
in 1999. In a few cases, the decline is slight, but it is mostly greater than five
percentage points. Third, the greatest difference is on the first question, focus-
ing on economic competition. Although the capitalists’ concern over social
diversity diminished between the first survey and the second, all groups of
entrepreneurs were more concerned about the threat of economic competition
to social stability than were officials, and that concern grew between 1999
and 2005, both in absolute terms and relative to officials. Finally, both entre-
preneurs and officials saw economic competition as less of a threat to stability
than political and social diversity. In both surveys, entrepreneurs and officials
agreed on the rank ordering of threats to stability: economic competition was
lowest, competition between parties was highest, and competition among indi-
viduals and groups fell in between. Although the absolute levels varied over
time and among subgroups, the rank ordering was the same. This similarity in
rank ordering reveals stability over time in the views of all groups. The survey
design allows us to see more systematically the differences between layers of
the state and changes over time, providing an advantage over studies of one
group, one locale, or one point in time.
Surveying Prospects for Political Change 213

Table 11.2. Perceived threats to stability among private entrepreneurs

and local officials (percentages who agree)

1999 2005 1999 2005

1. Competition between firms and individuals
is harmful to social stability.
All Entrepreneurs 24.5 26.4 All Officials 11.9 12.2
Xiahai red capitalists 18.7 22.5 County officials 9.9 10.1
Coopted red capitalists 22.7 27.7 Township and 14.7 13.8
village officials
Non-CCP, want to join 25.2 27.8
Non-CCP, don’t want 29.2 29.4
to join
2. If a country has multiple parties, it can
lead to political chaos.
All Entrepreneurs 48.0 45.7 All Officials 40.5 40.4
Xiahai red capitalists 50.8 49.9 County officials 36.9 33.9
Coopted red capitalists 47.7 47.3 Township and 45.3 45.3
village officials
Non-CCP, want to join 49.3 44.8
Non-CCP, don’t want 42.9 39.8
to join
3. If everybody does not share the same thinking,
society can be chaotic.
All Entrepreneurs 37.6 30.7 All Officials 22.0 21.6
Xiahai red capitalists 33.9 33.1 County offiials 16.7 18.7
Coopted red capitalists 43.1 26.1 Township and 29.5 23.8
village officials
Non-CCP, want to join 43.6 31.5
Non-CCP, don’t want 33.8 29.7
to join
4. Locally, if there are many groups with different
opinions that can influence local stability.
All Entrepreneurs 43.3 34.8 All Officials 33.8 28.8
Xiahai red capitalists 40.0 30.5 County officials 32.3 16.9
Coopted red capitalists 46.5 31.3 Township and 35.8 37.5
village officials
Non-CCP, want to join 41.3 38.9
Non-CCP, don’t want 44.7 38.6
to join
214 Bruce J. Dickson

Table 11.3. Distribution of private entrepreneurs in

local people’s congresses

1999 2005
All Entrepreneurs of which: 11.3 10.5
Xiahai red capitalists 40.7 58.6
Co-opted red capitalists 29.6 23.4
Non-CCP, want to join 13.0 7.2
Non-CCP, don’t want to join 16.7 10.8

Electing Capitalists to Political Posts

The final example concerns the participation of capitalists in China’s formal
political institutions, specifically local people’s congresses. If China’s capital-
ists are to serve as agents of change, one venue for their political activity would
be these formal institutions. Indeed, Li, Meng, and Zhang (2006: 318) argue
that private entrepreneurs sought representation in local people’s congresses
in order to assert and defend their property rights. The growing presence of
China’s capitalists in local legislatures has drawn considerable attention, not
only from scholars but from the Chinese media as well (People’s Daily, 2006).
More than 200 private entrepreneurs are delegates to the National People’s
Congress. At the local level, over 9,000 entrepreneurs have served as people’s
congress delegates (Zhang, 2004: 318).
Survey data allow us to go beyond these aggregate numbers to see which
capitalists are most likely to be politically active. Although delegates to local
people’s congresses are formally elected positions, the CCP has a determining
role in who can be nominated and elected. Accordingly, the relationship of the
capitalists to the CCP is the most important factor in determining this type of
political behavior. The survey was therefore designed to delineate the capital-
ists’ relationship to the party: for party members, whether they had joined the
party before going into business or were co-opted afterward; for nonmembers,
whether they were interested in joining the party. As shown in Table 11.3, the
vast majority of capitalists in local people’s congresses are red capitalists. In
particular, the largest single group in 1999 and the absolute majority in 2005
are the xiahai entrepreneurs, the ones who were already in the party before
going into business. In other words, the capitalists most likely to be people’s
congress delegates are the ones who were most embedded in the CCP before
going into business. Rather than being agents of change, these capitalists are
more likely to represent the status quo. In contrast, the group that might have
the greatest interest in political change – those who are not interested in join-
ing the CCP – has seen its numbers shrink over time. Expectations that capi-
talists may promote democracy by introducing new actors into the political
system are not borne out by these data: rather than introducing new actors,
the political system strongly favors those who are already part of it.
Surveying Prospects for Political Change 215

The real value of survey research is not simply in specifying the preferences
and behaviors of different groups as a whole, but more important, in ana-
lyzing the relationship between variables. This allows us to see what types
of individual and contextual factors determine which capitalists are more
likely to be local people’s congress delegates, not just which communities
have relatively large numbers of capitalist delegates or the percentages of
different types of capitalists among delegates. The interplay of individual
and contextual factors in determining which capitalists are more likely to be
local people’s congress delegates can be seen from the multivariate analysis
presented in Table 11.4.
Among those elected to local people’s congresses, red capitalists are more
likely to be delegates than are non-party members, even when controlling for
a variety of other factors. In addition, they are more likely to be older and to
operate larger firms than those who are not deputies. The impact of education
is curvilinear: high school graduates are more likely to be people’s congress
delegates than those with less education and those with college degrees. Since
the CCP is influential in deciding who is nominated, these results indicate
that the CCP favors its own members, older and better educated capitalists,
and those who operate the largest firms – in other words, the local political,
social, and economic elites. The coefficient for the size of the private sector
is negative, indicating that a given capitalist is less likely to be elected to the
people’s congress where the private sector is large. This may indicate a quota
for capitalists in the people’s congresses: if there is a ceiling on the number of
capitalists who can be in a people’s congress, the more capitalists there are in
the community, the less likely any one of them is to be chosen. Alternatively,
it could mean that in areas where the private sector is small, the capitalists
are more inclined to be politically active in order to defend and extend their
interests. More research is needed to determine a more definitive answer, but
given the CCP’s control over the nomination and election processes, a expla-
nation based on the CCP’s interests seems more plausible than one based on
the capitalists’ interests.
A final factor in the determination of which capitalists are people’s con-
gress delegates is the perceived threats to stability discussed in the section
above. Those elected to local people’s congresses are also more likely to see
competition and social diversity as threats to stability. Toleration for pluralist
viewpoints, interest groups, and competition among individuals, firms, and
parties is generally seen as representing liberal or “modern” values, but these
values are in short supply among this select but influential group of capital-
ists in China. If they were able to turn their political beliefs into policy, they
would be more likely to favor limits on competition and diversity, and such
steps would be antithetical to democratization.
In sum, on most of the questions examined here, the views of entrepre-
neurs cohere fairly well. Whereas Kellee Tsai (2007) finds the lack of a uni-
form viewpoint as indicating that China’s entrepreneurs do not constitute a
class, the data from my two surveys show they are nevertheless distinctive
216 Bruce J. Dickson

Table 11.4. Probit regression: Determinants of private

entrepreneurs in people’s congresses

Political Factors
Xiahai red capitalists .677***
(.157)
Co-opted red capitalists .788***
(.179)
Individual and Firm Characteristics
Age .036**
(.009)
Gender –.058
(.227)
Level of education 1.324**
(.538)
Level of education2 –.219*
(.091)
Sales revenue (log) .187***
(.039)
Years in business .027*
(.013)
Cultural Factors
Threats to stability .090***
(.026)
Contextual Factors
Per capita GDP, 1000 yuan –.004
(.009)
Size of private sector –.551*
(.267)
Constant –6.445***
(1.040)
N 978
Chi2 146.64***
Pseudo R2 .217
* p< 0.05, ** p < 0.01, *** p < 0.001
Robust standard errors in parentheses

when compared to local officials, and especially to county-level officials.8

Entrepreneurs may not have a singular voice, as indeed most groups do not,
but their views are similar on a range of issues. Entrepreneurs are more likely
to favor stability over growth, more likely to see the potential social conse-
quences of economic competition, and more likely to see the risks of emerging
8
This is the main theme of K. Tsai (2005); see also K. Tsai (2007). Besides our difference on
this point, our findings are otherwise remarkably similar. Replicated results like these provide
greater confidence in the reliability of our findings.
Surveying Prospects for Political Change 217

pluralism. More important, the views of xiahai, co-opted, and non-CCP

entrepreneurs have become more similar over time. If their concerns were
reflected in policy, it would most likely be to limit the degree of competition
and pluralism. Ironically, officials have more “progressive” views on these
matters. If entrepreneurs emerge as agents of change, it is hard to see them
advocating either economic or political liberalization. Where the difference
among subgroups of entrepreneurs matters most is their integration into the
political system: red capitalists are more likely to be nominated and elected as
people’s congress delegates. Since they are already part of the existing politi-
cal system, they have less incentive to promote regime change. Furthermore,
because they are more concerned about economic competition and political
diversity threatening social stability, they seem less likely to favor further­
liberalization. In contrast to assumptions that capitalists are inevitable sup-
porters of democracy, these findings suggest that they are more likely to sup-
port the status quo. The survey design – which captured individual, group,
and regional variation – was essential for uncovering these relationships.

Conclusion
These findings on the question of the political impact of privatization in China,
and the potential for private entrepreneurs to be proponents of democratiza-
tion, are the result of a survey research design that was sensitive to individual,
regional, and temporal variation in the current Chinese context. This design
provides greater confidence that the findings are reliable and not specific to one
place or one point in time. In particular, the survey design explicitly sought
to capture differences across political and economic elites, levels of the state,
and levels of development necessary to address the theoretical questions that
motivated the project in the first place: are China’s capitalists becoming sup-
porters of democratization, and if so, which ones are most likely to support
that type of political change? A different design might not have revealed the
divisions between officials and capitalists regarding the goals of development
and threats to stability, or the preponderance of red capitalists among those
who have been elected to formal posts.
Nevertheless, the survey design leaves room for improvement in later
research. The number of counties, and therefore the extent of variation in
the local context, was small. A single outlier could skew the results dramati-
cally. Moreover, the counties were not a random sample, but were purpo-
sively selected. This limits the generalizability of the results. More important,
the survey design lacks the kind of contextual details that would bring the
survey data to life. In looking at the relationships between variables, it is
easy to overlook the importance of local politics, personal relationships, and
individual agency, which are better uncovered with other research methods.
Most important, given this project’s main theme of whether China’s private
entrepreneurs are likely to be agents of change, the surveys cannot easily iden-
tify which individuals are most likely to be motivated to engage in political
action, individually or collectively, to bring about change. These two surveys
218 Bruce J. Dickson

did not directly ask respondents about types of political participation in which
they have engaged, types of political reforms they preferred, or ultimately
what they thought about democracy in China. At the time the questionnaire
was put together, these questions were too sensitive to be asked. They may
be more feasible in the current political environment, but that is not much
consolation.
In addition, there are questions I did not know to ask at the beginning of this
project. For example, whether a firm was a reformed SOE or a private enter-
prise from the start may be an important factor influencing its relationship to
the state and its owner’s degree of support for the status quo, but SOE reform
was just beginning in earnest at the start of this project. Similarly, whether
an entrepreneur formerly had been an SOE manager may be as important as
whether he or she joined the CCP before or after going into business, but this
also was not asked of the respondents. Unlike interviews and archival work,
survey research makes it nearly impossible to go back and ask a few more
questions to fill in the gaps. Instead, a new survey would be needed, with the
large investment of time and resources that it would require.9
These shortcomings were offset by the advantages of the survey methodol-
ogy. Individual case studies can provide a more in-depth look at an individual
or locality, but they run the risk of highlighting exceptions rather than the
norm. Research that simply extrapolates comparative findings to the Chinese
context runs the even greater risk of drawing invalid inferences. Social sci-
ence is based on probabilistic statements, not deterministic laws, and there is
no guarantee that findings from other countries and other points in time will
automatically apply to new cases. Empirical research is needed to test the gen-
eralizability of comparative claims and to adapt general concepts to China’s
rapidly changing political and economic environments. The design also took
into consideration how to properly adapt a general research question – the
role of capitalists in democratization – to the specific context of contemporary
China. The degree of political embeddedness is not unique to China, but the
specific way of measuring it – the relationship to the CCP – shows how general
concepts can be easily adapted for empirical research in a new setting.
Using the same questions in the same communities at two points in time
made it possible to show how multiple factors influenced the respondents’
views on the questions examined here: the importance of bureaucratic rank,
the distinctions between officials and entrepreneurs, the level of development,
and the size of the private sector. The consistency across time in the relation-
ships between the variables provides greater confidence that these results are
not ephemeral but tap into substantive matters. The survey research method-
ology and the specific design of this research project were essential to uncover
these relationships and to test comparative insights into the dynamic case of
China.

9
A new collaborative project with Jie Chen was designed to address these shortcomings, and
the survey was carried out in 2006–2007. The initial results can be seen in Chen and Dickson
(2010).
12

Using Clustered Spatial Data to Study Diffusion

The Case of Legal Institutions in China

Pierre F. Landry

As Manion details in Chapter 10 of this volume, social scientists with an

interest in China already have contributed a healthy stream of survey research
based on national probability samples, but the methodology behind these sur-
veys has historically been dependent on the household registration system, or
hukou (户口 ). Since the 1950s, Chinese households have (in principle) been
listed exhaustively by village or urban neighborhood; for decades, local offi-
cials took great care to maintain and update these lists. This system was an
efficient collection of population data, and household registration lists have
thus been widely used as the basis of the vast majority (if not all) of probability
samples drawn in China since 1978.
China’s rapid economic transformation has dramatically undermined the
key assumption that hukou lists are accurate and complete representations
of the population of their respective localities. Until the early 1990s, reason-
ably reliable lists of residents could still be obtained from villages and neigh-
borhoods, but China’s rapid industrialization and urbanization have severely
undermined the ability of local officials to track their populations effectively.
As a result, survey researchers who sample the countryside face the prob-
lem of drawing respondents who, according to official lists, live in a village
but in reality have become urban residents; yet, if researchers sample from
urban hukou lists, they will often discover that migrants are not properly
listed unless they happen to hold a formal temporary registration. Given the
magnitude of recent internal migration in China, the problem is severe. We
can no longer claim that samples drawn from hukou lists are representative
of the population in a given locality. Nor can we claim that such samples are
probability samples.
In this chapter, I demonstrate how the features of spatial probability sam-
pling can overcome the hurdle of population mobility and coverage errors
induced by “traditional” sampling methodologies. I also show how a spe-
cific feature of spatial sampling (geographical clustering prior to household
selection) can be leveraged to test whether diffusion effects take place at the
community level, using the specific example of the citizen’s willingness to go

219
220 Pierre F. Landry

to court. The chapter proceeds as follows. I first present the rationale and
key features of the spatial sample that was drawn by a team of researchers in
2003–2004 for a project on Institutionalization of Legal Reforms in China
(ILRC).1 I then discuss the process of institutional diffusion and how it can be
traced by linking individual data to the specific behavior of other respondents
interviewed in the same sampling clusters. In the last section of the chapter, I
test the argument that even small subpopulations (here, individuals going to
courts) can have a large impact on the propensity of other community mem-
bers to emulate their behavior, and presumably help consolidate these legal
innovations.

Untying Surveys from hukou Lists

The investigators of the survey on the “Institutionalization of Legal Reforms
in China” (ILCR) were concerned that increasing mobility in the population
seriously undermines the reliability of lists based on hukou and leads to the
systematic exclusion of subpopulations that are theoretically important to the
study of a rapidly changing society. Since the 1990s, the powerful combina-
tion of economic reforms and lax enforcement of household registration rules
have gravely weakened the institution of the hukou. Rural-to-urban migra-
tion is taking place on a massive scale: in 1995, there were about 80 million
migrants in Chinese cities, only half of whom had been formally registered by
the authorities (Congressional-Executive Commission on China, 2006). Only
seven years later, the population census of 2000 counted as many as 125 mil-
lion migrants, 79 million of whom had crossed provincial boundaries (Liang
and Ma, 2004).
As staggering as these numbers are, statistical authorities openly acknowl-
edge that the 2000 census exercise most likely resulted in the most severe
undercount of all recent Chinese censuses (by about 22 million people), pri-
marily due to the dependency of enumerators on inaccurate registration lists
(Zhang, Li, and Cui, 2005). The National Bureau of Statistics (NBS) estimates
that the number of “peasant workers” exploded to 225 million by the end
of 2008, 140.4 million of whom no longer work in their township of origin.
These 140 million do not include other socioeconomic groups and urban reg-
istration holders who also experienced some kind of migration or are within-
city movers, resulting in serious mismatches between official registration and
actual residence (NBS, 2009).
The emergence of a large migrant population has been widely noted
(Goldstein, 1987; Solinger, 1995, 1999; Yang, 1993; Zhang, 2001) but has

1
The ILRC project was undertaken by Professors Shen Mingming and Yang Ming of the
Research Center for Contemporary China at Peking University in collaboration with Wenfang
Tang (University of Pittsburgh), Yanqi Tong (University of Utah), and the author (Yale
University). Financial support by the Ford Foundation in China as well as our respective uni-
versities is gratefully acknowledged.
Using Clustered Spatial Data to Study Diffusion 221

only slowly led to a serious reevaluation of sampling techniques derived from

hukou lists. A number of studies have attempted to reach different types of
migrants. Goldstein et al. (1991) were able to analyze migrants who reside in
formally registered households, but they concede that this approach restricts
the analysis to a very specific subset of migrants. As Goodkind and West
(2002) discuss, migrants include a vast array of types, ranging from long-term
residents who are ultimately successful in obtaining a formal registration to
the ‘‘floating population’’ of informal, short(er)-term migrants who return to
their hometowns regularly.
It is clear that hukou-based survey methods result in biased samples and
that the problem will only worsen unless both residents and local officials
are given incentives to update their hukou in a timely manner. These biases
are not merely a statistical inconvenience but can profoundly impact theoriz-
ing about the extent and the nature of social change in contemporary China.
Migrants are after all one of the most dynamic and entrepreneurial segments
of the population. Using the ILRC sample, Tang and Yang (2008) have shown
that standard assumptions about the behavior of migrants regarding dispute
resolution do not hold once they are properly accounted for and included in a
sample with some probability of selection as nonmigrants. Clearly, excluding
migrants from surveys because they are admittedly hard to reach is bound to
skew and polarize findings based on groups that have historically benefited
from the “urban bias” of the regime (long-term urban residents) and on the
rural nonmigrants who tend to be economically and socially disadvantaged.
The ability to capture dynamic subpopulations (whether migrants or within-
city movers) is also closely related to any conceptualization of institutional
diffusion. People on the move bring norms, habits, and life experiences to
their new places of residence, where they are in turn likely to acquire new
ideas or norms or to adopt some of the behavioral characteristics of other
community members. Seemingly representative surveys based on hukou lists
make it more difficult to capture these changes when they reach only “perma-
nent” residents.
In addition to the phenomenon of rural-to-urban migration, dramatic
changes to housing policy in the mid-1990s ushered in an era of unprecedented
urban construction during which millions of citizens became ­homeowners
or rented housing in the private sector (Davis, 2000). The impact of these
changes on the reliability of registration lists is almost as powerful as the
phenomenon of rural-to-urban migration. Within-city mobility where movers
do not feel compelled to change their hukou status (because they are already
proper residents of the city in which they live) is a serious source of nonre-
sponse: based on the formal registration lists, these “movers” may be drawn,
but if interviewers are sent to their official addresses, they are unlikely to
make contact as these respondents actually live in a different part of the city.
During a proof-of-concept study of spatial probability sampling in Beijing
in 2002, we found that 20 percent of the sample consisted of such movers.
Adding the migrant population (25 percent of the sample), we concluded that
222 Pierre F. Landry

Figure 12.1. The Institutionalization of Legal Reforms survey of China (2003–2004).

(Dots represent sampled counties and lines denote provincial boundaries.)

only 45 percent of our respondents would have been reached via the tradi-
tional hukou-based method of selection (Landry and Shen, 2005).

Data: Survey on the Institutionalization

of Legal Reforms in China
The Institutionalization of Legal Reforms in China is the first national sample
of its kind. We collected data on the types and extent of civil, economic, and
administrative disputes on a national scale in order to examine in detail the
multiple mechanisms by which grievances evolve. The survey is based on a
multistage stratified sample in which each province, municipality, or autono-
mous region on the Chinese mainland is taken as a stratum. Within each
stratum, counties (or urban districts) were selected at random by probability
proportionate to size (PPS). Within each county, two townships or their jiedao
(街道) counterparts in urban areas were also selected at random. We used
2000 census data to develop measures of size at the township level. Overall,
respondents are thus clustered in 100 county units (Primary Sampling Units,
PSUs) and 200 township units (Secondary Sampling Units, SSUs). The PSUs
are mapped in Figure 12.1.

Spatial Sampling
The key methodological innovation in this multistage sampling design took
place below the township level. To bypass the household registration system
entirely, the spatial approach calls for the random selection of clearly defined
Using Clustered Spatial Data to Study Diffusion 223

sampling units that can be enumerated by the research team. Because this pro-
cess is time-consuming and costly, it is essential to keep these units as small as
possible so that trained enumerators equipped with global positioning system
(GPS) receivers can easily locate and survey all final spatial sampling units.
Prior experimentation in urban and rural areas suggested that grids of half
square minutes (HSM) of latitude and longitude should be developed for all
townships.
Whereas primary (counties) and secondary (townships and jiedao) sampling
units were selected using the traditional two-step sampling procedure based
on Probabilities Proportional to measures of Size (PPS) – with county and
township population data obtained from the 2000 census – the IRLC project
adopted a spatial sampling approach below the township level. In each town-
ship, town, or jiedao, two spatial units – specifically, HSM of latitude and
longitude – were selected by PPS from a geographical grid. The size of each
unit was defined in terms of the share of its surface that fell within a township.
For example, if only a quarter of a cell fell within the township map, its prob-
ability of selection would only be .25/N (where N is the number of HSMs that
are inside or straddle the township boundary). If a cell fell entirely within the
township, its probability of selection would be 1/N.
Figure 12.2 illustrates the results of this procedure in two townships in
western China. The three easternmost rectangles are the chosen HSMs for one
township in Huangyuan (H.Y.) county: one unit straddles the railroad near
the county boundary whereas the other two are located in more remote areas.
In this case, the northernmost block was randomly assigned as the backup
unit but in the end was not used.2 Figure 12.3 shows how the same procedure
worked in an urban setting, a district in the capital of a northeastern province.
In dense urban areas, units (here, jiedao) have rather small surfaces, which
leads to heavy clustering among the sampled HSMs.
The respondents were selected from micro-communities, that is, Final (spa-
tial) Sampling Units (FSUS) in which all households were visited for interviews
so as to ensure equal probability of selection throughout the entire sample
(Landry and Shen, 2005). Each FSU is 1/80th HSM, which is 11.25 square sec-
onds of latitude and longitude. Within each HSM, a varying number of FSUs
was drawn completely at random and enumerated systematically by trained
surveyors. To yield a consistent set of respondents in each township, the num-
ber of FSUs was inversely proportional to the expected population density
of the HSM. In the case of Figure 12.4, the population density is high. We
thus needed in expectation to draw only two FSUs, and enumerate them. By
contrast, Figure 12.3 shows how a township of H.Y. county (the easternmost
2
In each township, we drew a third HSM as a backup unit in the event that the first two draws
would not yield enough respondents. In the event that the number of respondents per town-
ship was still unacceptably low, the initial draw was discarded and two villages were selected
at random (by PPS) from the list of administrative villages in the township. These fresh HSMs
were centered on the village committee. Final Sampling Units were then drawn using the stan-
dard protocol.
224 Pierre F. Landry

Figure 12.2. Example of a county map displaying township boundaries and basic
infrastructure overlaid to the Google Earth model, with grids coded in Keyhole
Markup Language displaying three Tertiary Sampling Units (TSUs) in each sampled
township (one TSU per township was randomly assigned as a backup).

HSM in Figure 12.2) has such a low population density that a census of the
entire HSM was required to yield, on average, the same number of respon-
dents. The dots within each cell represent dwellings that were actually inhab-
ited and where interviews were successfully conducted.
In the end, over 16,000 FSUs were surveyed nationwide, and the HSMs
that were populated yielded a list of about 12,000 dwellings. Within a few
weeks of this enumeration, teams of interviewers selected and interviewed one
respondent per dwelling, using the Kish grid method. This procedure yielded
a sample of 7,714 valid respondents, drawn with equal probability from a
nationally representative sample of Chinese adults. Due to uneven response
rates by age and gender as well as the purposeful underrepresentation of
respondents drawn in very dense neighborhoods,3 we used the 2000 census
data to devise sampling weights. Finally, we specifically accounted for the
design effect of this complex multistage survey at the levels of PSUs (counties)
and SSUs (townships).4
3
To control operational costs, if the enumerated lists within a given HSM exceeded 60 dwell-
ings, only 60 dwellings were selected at random. In the final dataset, sampling weights incor-
porate this component to correct for this purposive but computable unequal probability of
selection among these respondents.
4
We do not account for design effects below the township level. The number of tertiary sam-
pling units per SSU is not only too small to do so (mostly two, sometimes three when the
backup unit was used), but numerous studies have also shown that in practice, most of the
design effect occurs at the PSU and to a lesser extent at the SSU level; variance estimates are
effectively unaffected. See Barron and Finch (1978); Dever et al. (2001).
Using Clustered Spatial Data to Study Diffusion 225

Figure 12.3. Example of a spatial sampling unit (half square minute) drawn in a low-
density rural area in western China.
(Squares represent FSUs and dots indicate that one (or more) interviews were success-
fully completed.)

Technical Advances since 2004

When the IRLC project was implemented, affordable satellite or remote sens-
ing information was neither available nor affordable to most social scientists.
Since then, powerful earth visualization software such as Google Earth allows
access to remarkable imagery that can be incorporated easily during the prep-
aration phase of a spatial sample.
The images allow samplers to specify more reliable priors on the data.
Without satellite imagery, it was not possible to model the distribution of the
226 Pierre F. Landry

Figure 12.4. Example of spatial sampling units in an urban area.

Half Square Minutes (HSMs [including backup units]) drawn in one jiedao of a large
city in northeast China. As the population density is large, drawing two Final Sampling
Units (FSUs) per HSM was sufficient to yield – in expectation – the requisite number of
respondents. Large squares are HSMs, and small squares are FSUs. Dots within FSUs
indicate that one (or more) interviews were successfully completed. The HSM without
interviews is the backup unit that was surveyed but not used.

population density precisely below the township level. Some areas were (very
conservatively) excluded from the sample a priori in remote regions where
the best cartographic evidence and common sense suggested that population
density was in effect nil, such as deserts, dense forests, or lakes. Without fur-
ther details about the specific distribution of dwellings, the number of FSUs –
drawn in inverse proportionality to the expected population density of the
nonexcluded zones – was large (16,000). This figure was based on the research
team’s prior experimentation to yield the requisite number of respondents per
SSU (townships), a costly and time-consuming process as enumerators had to
travel to verify personally whether each FSU was populated.
Using Clustered Spatial Data to Study Diffusion 227

With satellite imagery, a great deal of work can be accomplished in the

office, and in many areas, at a level of detail that allows excluding empty
squares a priori with a high degree of precision.5 If sparsely populated areas
happen to be sampled, researchers can literally zoom to an area to confirm
ex ante whether surveyors and enumerators must be dispatched to this loca-
tion.6 In 2009 high-resolution imagery for China was still not available for
the entire Chinese territory, but the data are improving constantly. Even
without access to high-resolution information, the default definition is suf-
ficient for coarser exclusion work, with the caveat that higher rates of Type
I and Type II sampling errors are likely. It is also advisable to overlay car-
tographic data or information external to the earth model, which greatly
improves the precision of the work (as we did ex-post with Figures 12.2,
12.3, and 12.4). If squares indicate populated points that are not clearly vis-
ible from the satellite image, these squares ought to be retained and verified
by the enumerators.
To sum up, the ILRC survey provides a solid foundation for testing the
validity of prior case-study findings and also allows making point predic-
tions and generalizable propositions about the behavior of ordinary citizens.
Furthermore, the sample is large enough that it captures rare events (such
as disputes), conforms to the principle of equal probability selection, and is
representative of China’s varied geographic, demographic, and social and eco-
nomic environments.

Institutional Diffusion
A great deal of the literature on institutional innovation stresses the role of
trustworthiness as a key determinant of successful institutional innovation. If,
under conditions that need to be specified, people trust an institution, they are
likely to rely on it should the need arise; otherwise, they are likely to turn to
reasonable alternatives. Trust is also a condition of institutional endurance in
the long run (Hetherington, 1998; Levi, 1999; Levi and Stoker, 2000; Ulbig,
2002).
Many scholars have demonstrated empirically that both interpersonal trust
and system-based trust are comparatively high in China (Inglehart, 1997; Shi,
2001; Tang and Parish, 2000). However, generalized trust may not be as reli-
able a predictor of success or failure of a specific institution. Just as trust
between individuals can be generalized or particularistic (Uslaner, 2002),
Jennings (1998b) has shown that individual trust in government institutions
can be highly differentiated. A more recent study by Lianjiang Li (2004)

5
Keyhole Markup Language (KML) is used to overlay data to the Google Earth model and
is relatively easy to program. The specifications of the language are available at http://code.
google.com/apis/kml/documentation/.
6
If the satellite image is slightly outdated, it is advisable to retain the seemingly empty squares
that are adjacent to populated ones in the sample frame to account for construction or urban
growth.
228 Pierre F. Landry

confirms that rural Chinese exhibit highly differentiated levels of trust regard-
ing central and local institutions.
The ILRC survey results demonstrate that these broad findings also hold with
respect to legal institutions. We find that trust is institution-specific: whereas
organizations that are frequently involved in dispute mediation (such as vil-
lage committees) fare especially poorly, the courts and the procuracy are held
in relatively high regard. Furthermore, most respondents trust institutions
that are closely associated with the state to a far greater extent than they trust
nonbureaucratic actors: legal professionals are less trusted than public secu-
rity organs, whereas village committees fare worst of all institutions listed on
the survey instrument.
Popular trust is overwhelmingly tilted in favor of central political and judi-
cial institutions. As a follow-up to a general measure of trust (for instance, in
courts), we asked respondents to contrast local institutions with central ones.
In the case of courts, we asked whether they trust the Supreme People’s Court,
the local court, or neither. We asked similar central/local comparisons for
people’s congresses, the Communist Party, and government agencies generally.
Although party members tend to be more trusting than nonparty members,
central institutions enjoy a considerable degree of support in both groups.
Theorists of trust attach great importance to its relational aspect, but survey
research has not been well equipped to test these arguments because respon-
dents chosen at random may or may not be connected to other respondents in
the sample. The spatial approach is a better alternative because of the inherent
properties of the final sampling units: all households that are located in a small
physical space (e.g., a spatial square second of latitude and longitude) are by
construction each other’s neighbors. Whereas in a typical survey, researchers
would ask a respondent whether she trusts other people in the neighborhood,
we would be better off knowing whether this trust is reciprocal. In a spatial
design, we fortunately interview her neighbors as well.
The spatial sampling design is also a powerful tool for measuring the
impact of unusual events within a community, such as going to court. This is a
rare event, but the important issue is whether other community members who
observe an event affecting a peer are learning anything from his experience.
The role of community networks is especially intriguing given the nature of
Chinese society and the structure of the post-Mao state. Considerable research
in anthropology (Kipnis, 1997; Ku, 2003; Yang, 1994), sociology (Bian, 1999;
Gold, Guthrie, and Wank, 2002; Guthrie, 1999), economics (Krug, 2004;
So and Walker, 2006), as well as political science and law (Lee, 1997; Oi,
1986) has been devoted to the extent and impact of social relations (guanxi or
关系) and its extent and its impact on individual behavior. If the importance
of guanxi has indeed endured in the contemporary period, Chinese society
should be more prone to the quick adoption (or rejection) of an innovation
once it becomes known to members of a tightly knit social network. Dense ties
facilitate information flows and rapid diffusion of the benefits and shortcom-
ings of innovations among connected individuals.
Using Clustered Spatial Data to Study Diffusion 229

Testing Diffusion Hypotheses with Spatially

Sampled Survey Data
The ILRC sample is uniquely suited for testing the impact of the small com-
munity networks due to its special design. This spatial clustering of the data
at the FSU level gives us greater leverage on the analysis of social processes
in which neighbors interact with one another. In such small communities, the
probability that respondents know each other is very high. If diffusion effects
are taking place within micro-communities, the ILCR survey is ideally suited
to detect their magnitude.
We can identify respondents in their specific communities who were directly
involved in legal disputes, went to court, and were therefore in the position to
influence their network based on their experiences with the courts. For each
dispute category covered by the instrument – civil, economic, and administra-
tive – we inquired whether the choice of going to court was decisive in the
resolution of the dispute. If so, we further asked whether they would be willing
to use the same method should a similar dispute occur in the future. We can
thus identify the specific communities with which adopters (defined here as past
disputants who would use the courts in the future) are present. For each com-
munity, we can compute the mean share of respondents (x) who are “decisive
adopters” in each kind of dispute. For additional precision, the computation of si
excludes the respondent herself, a necessary correction when clusters are small:

si
(x · N) xi
if N 1 and si 0 if N 1
N 1

where x is the variable of interest and N is the total of respondents sampled

in the community. These shares are theoretically specific to each community
member, although any pair of individuals who share the same behavior has
the same share. As the variable of interest occurs rarely, s = o in most cases
and has a theoretical maximum value of 1.
Consider the hypothesis that two parameters are jointly conducive to rapid
institutional diffusion: trustworthiness of the court and presence of indi-
viduals within small communities who have engaged courts successfully (see
Table 12.1). We cannot directly observe social networks, but it is reasonable
to assume that members of small communities interact with each other fre-
quently and learn quickly about unusual events. Given its rarity, victory in
court certainly qualifies as the kind of news that is likely to spread fast. Rapid
diffusion is likely to occur when a high proportion of satisfied and trusting
end-users propagate their behavior through dense social networks. If only one
of these factors is present, the process will be more gradual. If neither are pres-
ent, we should observe little or no diffusion.7
7
Note that the absence of diffusion does not imply that the number of court users will not rise;
it simply means that individuals who use the courts will not be emulated in their communities.
230 Pierre F. Landry

Table 12.1. Conditions for institutional diffusion

Trustworthiness of the institution

Low High

Density of adopters High Gradual diffusion Rapid diffusion

Low No/very slow diffusion Gradual diffusion

Community Experiences and Institutional Diffusion

Although diffusion theory assumes that “adopters” are local opinion mak-
ers, we must emphasize that these experienced individuals constitute only a
small fraction of a community. Using the township as the level of analysis, we
encountered very few localities where more than one respondent had actually
experienced a dispute that was decisively resolved in court. Indeed, we could
not identify any experienced users in most communities: we found 69 town-
ships (out of 200 surveyed) where civil disputes were resolved in court – 47
economic disputes and only 16 administrative disputes. However, the lack
of clustering among adopters does not mean that adopters have no measur-
able impact in their respective communities. As rare as they may be, diffusion
research suggests that they may impact the behavior of others if they happen
to have strong links with “ordinary” community members.
The survey data suggest that adopters outnumber nonadopters among
respondents who have successfully used formal legal institutions. Ninety
percent of citizens who have settled an economic dispute in court claimed
that they would do so again, against 78 percent in the case of civil disputes.
The odds are lower among administrative disputants, probably because the
scope of administrative litigation is still quite narrow and success rates remain
very low. Overall, these proportions are consistent with the diffusion hypoth-
esis: very few people ever go to court, but since those who do are willing to
use the institution in the future, they are likely to diffuse their behavior within
their social networks. This would explain the rising proportion of court users
among citizens who are engaged in a legal dispute for the first time.
We can gauge the potential impact of these adopters by comparing the pro-
pensity of inexperienced respondents to go to court across communities with
varying densities of actual adopters. Specifically, we asked respondents who
had not experienced a dispute whether they would be inclined to go to court
based on a hypothetical situation presented in a vignette of each dispute cat-
egory. If the diffusion hypothesis is correct, we should observe a greater pro-
pensity to go to court among respondents who happen to live in communities
where one (or more) of their neighbors has “adopted” the institution.

However, if individual-level variables that predict this behavior change over time, a greater
proportion of the population will still adopt the institution.
Using Clustered Spatial Data to Study Diffusion 231

Table 12.2. Impact of court adopters on the mean propensity to go to court

Civil Economic Administrative

N 200 200 200
F(1, 198) 5.86 0.08 3.85
Model Prob > F 0.02 0.78 0.05
Coefficient Coefficient Coefficient
Share of Court Adopters 1.662** 0.345 4.43**
in Township
Constant 0.419*** 0.492*** 0.30***
***, **, and * denote levels of significance at .01, .05, and 0.1 levels, respectively.

The preliminary evidence is again encouraging for the diffusion hypoth-

esis: using townships as the unit of analysis, two of the three simple bivari-
ate regressions show that in townships where residents have been to court in
civil and administrative disputes and are willing to use the institution again,
their neighbors who were never engaged in a dispute are more prone to go
to court than are residents of communities where no one has any experience
with courts (see Table 12.2). However, this does not seem to be the case for
economic disputants. To be certain of the net impact of these “adopters” on
institutional diffusion, we require a fully specified model that captures both
the impact of the individual characteristics of the respondents and the impact
of the institution adopters in their community. Such models allow proper mea-
sures of the magnitude of these diffusion effects.

Multivariate Analysis: Modeling the Propensity to Go to Court

The propensity to go to court is modeled as a probit equation that takes into
account the multistage stratified nature of the sample design and uses prob-
ability weights. Separate estimates were computed for each class of disputes
that was covered in the ILRC project. We asked all respondents whether they
had been involved in civil, economic, or governmental disputes in the past
twenty years and whether they chose to go to court to resolve the disputes.
The dependent variable is coded 1 if the respondent went to court, and zero
otherwise. Those who did not experience disputes were asked to react to a
simple vignette and describe the actions they would likely take under such
circumstances. This technique is more reliable than asking unstructured
questions, particularly since we do not have the problem of cross-cultural
comparisons in a single-country study (King et al., 2003). If the respondents
never encountered a civil (respectively, economic or administrative) dispute,
the dependent variable is also coded 1 if they asserted that they would use the
courts in their evaluations of hypothetical civil, economic, and administrative
cases presented as vignettes.
232 Pierre F. Landry

On the left-hand side, the model accounts for the disparity between respon-
dents who actually experienced disputes and those who responded to these
hypothetical situations.

prob(Court)|d=0 = Φ(X β) + ε for individuals who did not experience

a ­dispute of type d and,
prob(Court)|d=1 = Φ(X β + d) + ε for individuals who did.

The respondents who answered the hypothetical questions after a vignette

were more likely to state that they would go to court: their expressed prefer-
ences were costless, in contrast to the disputants who actually chose to go to
court and faced tangible transaction costs. I interpret the magnitude of the
coefficients associated with these actual dispute-specific dummy variables as
markers of the transaction costs of going to court.
Since the diffusion hypothesis rests on the impact of two variables (the
trustworthiness of courts and the density of adopters in the community),
we need to test whether adding these variables to a baseline model actually
improves the predictive power and the statistical significance of the model.
Since the likelihood-ratio test cannot be performed on probit regressions for a
complex survey design, we must instead rely on unweighted probits estimated
with the same set of independent variables. Because of a small number of miss-
ing observations when these two variables are added to the baseline model,
the likelihood-ratio test is restricted to a subset of observations that are com-
mon to the saturated and the nested model (Table 12.3).
As shown in Table 12.3, regardless of what kind of dispute – civil, eco-
nomic, or administrative – these likelihood-ratio tests are all consistent with
the diffusion hypothesis. The saturated models are always superior to the
nested ones.8 However, the specific significance of the variable that captures
the presence of community adopters varies by dispute category: it is consid-
erable for civil disputes, less so for economic disputes, and insignificant for
administrative disputes. The expected dynamics of diffusion hold very well
for civil disputes: the coefficients are positive (in the expected direction) for
both variables. The propensity to go to court varies across individuals and
communities as a function of these parameters. Varying these parameters and
holding all other variables at their sample mean yields a more intuitive picture
of the joint impact of these variables (as shown in Figure 12.5). The findings
of diffusion research through simulations that small networks can have large
behavioral consequences seem to apply here as well: even a small share of
adopters at the community level greatly increases the likelihood that commu-
nity members will also adopt the same behavior.

8
Due to space limitations, I do not discuss here a number of control variables measured at the
respondent level included under the headings “human capital and information,” “political and
social capital,” and “sampling units” in Table 12.3. Interested readers may consult Landry
(2008b) for a substantive elaboration of these variables.
Using Clustered Spatial Data to Study Diffusion 233

Table 12.3. Probit estimates of going to court in civil, economic, and

administrative cases

Model 1 Model 2 Model 3

Civil Economic Administrative

Number of strata 24 24 24
Number of PSUs (counties) 100 100 100
Number of observations 7160 7160 7160
Estimated population size 850 850 850
(millions)
Prob > χ 2 0.000 0.000 0.000
Control for Actual Disputes
Civil dispute –0.841*** – –
Economic dispute – -0.862*** –
Administrative dispute – – –1.062***
Diffusion Variables
Trustworthiness of courts 0.177*** 0.252*** 0.208***
Share of court adopters in
township
Civil cases 5.903** – –
Economic cases – 5.845 –
Administrative cases – – 3.717
Human Capital & Information
Formal education (years) 0.023** 0.030*** 0.020***
Legal knowledge score 0.061*** 0.044*** 0.056***
Television 0.119*** 0.052** 0.076**
Political and Social Capital
CYL member 0.172*** 0.130 0.005
CCP member 0.229*** 0.189** 0.102
Contact w/ Party or Gov. Cadre 0.055 0.091* 0.063
Contact w/ Legal or Public 0.225*** 0.174*** 0.162***
Security official
Contact w/ People’s Congress 0.084 0.050 –0.001
Contact w/ lawyer 0.131 0.191** 0.017
Contact w/ Legal Aid Bureau 0.038 –0.181* –0.176
Contact w/ Labor Union 0.030 0.129* 0.047
Demographic Variables
Age 0.006 0.008 –0.024**
Age-squared 0.000 0.000 0.000*
Female –0.019 –0.026 0.105**
Han nationality –0.016 0.008 –0.102
Urban registration 0.334*** 0.123 0.148*
Full-time farmer –0.040 –0.098 –0.041
(continued)
 234 Pierre F. Landry

Table 12.3. (continued)

Model 1 Model 2 Model 3

Civil Economic Administrative

Constant –1.548*** –1.356*** –1.133***
LR-test of full vs. nested model without diffusion variables
(unweighted probit with 7,160 observations)
LR χ 2(2) 102.61 131.33 112.15
Prob > χ 2 .000 .000 .000
Because only one PSU was drawn in the smallest provinces, the original strata are grouped in
24 post-estimation strata. Linearized variance estimates account for complex multistage survey
design effects, with stratification, first-stage selection of PSUs (counties) and second-stage selec-
tion of SSUs (townships). These calculations ignore design effects at and below the third stage.
***, **, and * denote levels of significance at .01, .05, and 0.1 levels, respectively.

0.8

0.7

0.6
Fitted probability

0.5

0.4
Trust Level
1- Not at all
0.3 2- Not much
3- Somewhat
4- A great deal
0.2
0 0.02 0.04 0.06 0.08 0.1
Share of community adopters
Figure 12.5. Impact of the combined presence of court adopters in the community
and the respondent’s level of trust in the courts on the probability of adopting courts
as a dispute resolution venue in a civil case. (All other right-hand side variables are set
at their sample mean.)
Using Clustered Spatial Data to Study Diffusion 235

Conclusion
The technology of survey research in China and elsewhere must adapt to rapid
social change: in China, migration to urban areas and general occupational
and residential mobility greatly complicate reaching respondents drawn at
random from official registration lists. This process is not unique to China,
but it is especially acute because the history of restricting population mobility
from the 1950s to the 1990s through the hukou system gave survey research-
ers extraordinarily reliable sample frames. Given the magnitude of the migra-
tion, early approaches have become inappropriate.
Spatial sampling is an effort to solve this problem, by incorporating
migrants into sample frames in a cost-effective manner, and to reduce bias
induced by coverage errors. This chapter also demonstrates an important sec-
ondary benefit of spatial sampling: the ability to investigate diffusion patterns
among small clusters selected at random into equal probability samples and to
explicitly test hypotheses about the process of institutional diffusion.
The efficiency and affordability of the technique was refined and improved
in other projects, including recent waves of the Beijing Area Study, a project
on Inequality and Distributive Justice, as well as the latest wave of the World
Values Survey (mainland China sample). Better technology combined with the
practical experience accumulated by the research teams has greatly improved
control over sampling and enumeration costs. There are nonetheless signifi-
cant challenges ahead. All surveys (spatial or otherwise) face the increasing
problem of access to gated communities. As the number of housing estates
grows in Chinese cities, even the best spatial sample cannot solve the problem
of convincing gatekeepers to allow enumerators into their estates. This risks
the loss of entire clusters from the survey, many of which are populated by
affluent citizens. Geographical Information Systems and satellite imagery can
help researchers understand and perhaps model this process of “community
nonresponse,” but they cannot solve the problem entirely. It seems advisable
to lower these risks by keeping clusters as small as is practical and to increase
sampling ratios at the lower stages of selection.
13

Measuring Change and Stability over a Decade

in the Beijing Area Study

Mingming Shen and Ming Yang, with Melanie Manion

Descriptive statistics from representative sample surveys conducted in main-

land China provide a static picture that is often soon overtaken by the impact
of rapid socioeconomic change. The importance of longitudinal data generally,
but especially in such a context, cannot be overstated. Survey researchers seem
to recognize this: as discussed in Chapter 10, a remarkable number of surveys
conducted in mainland China have a longitudinal component. By far the most
ambitious of these is the Beijing Area Study (BAS), an ongoing annual rep-
resentative sample survey of Beijing residents, designed and conducted since
1995 by the Research Center for Contemporary China (RCCC) at Peking
University. This chapter begins with an introduction to the underlying vision
and goals of the BAS, as conceived in the early 1990s. It then turns to specific
issues of questionnaire content, sampling design, and survey implementation.
We pay particular attention to the challenges and changes faced over the first
decade of the BAS; major changes in sampling were made in 2007, however,
and we review these here. For the most part, we do not present survey find-
ings, except to illustrate particular points in the discussion of methods.1

Vision and Goals

The BAS focuses mostly on socioeconomic rather than explicitly political
issues. Indeed, “politics” does not even appear in the full project title: Beijing
Annual Survey of Social and Economic Development (北京社会经济发展年度
调查). This is not simply because political topics are more sensitive than social
or economic topics, with implications for survey implementation, although
this is certainly a serious consideration for a project with a long projected life
span as opposed to a one-shot effort. It also has much to do with the context
of the early 1990s, when the vision of the BAS initially emerged. After 1992,
with a renewed policy emphasis promoted by Deng Xiaoping on the role of the

1
For a presentation of findings across issue areas for the first decade of the BAS, see Yang et al.
(2007).

236
Measuring Change and Stability over a Decade 237

private sector in the economy, the pace of economic and accompanying social
change in China increased rapidly. In the new historic period of “reform and
opening,” the broad goal of the BAS designers was to capture, with a continu-
ally updated dataset, the impact of the major ongoing reforms on the everyday
lives of ordinary Chinese.
An important influence on the BAS as it emerged was a particular American
survey experience. While earning his doctorate in political science at the
University of Michigan, Mingming Shen gained firsthand experience with
survey work as an interviewer for the Detroit Area Study. This study and
Chicago’s General Social Survey provided well-established models of longitu-
dinal single-city surveys. When designing the survey and questionnaire, BAS
organizers also sought the guidance of experienced experts, for example, con-
vening a forum in April 1995 to solicit the views of survey researchers from
the University of Michigan’s Institute for Social Research.
The influence of the Detroit Area Study and Chicago General Social Survey
is reflected in an early decision on questionnaire design. Specifically, in addi-
tion to demographic measures, which are repeated annually, the question-
naire is divided into two parts: (1) core items, repeated annually or at regular
but less frequent intervals, providing continuity of measures over time, and
(2) items designed to change from year to year to reflect changing research
agendas or new events. Three other early decisions also set the framework
for BAS planning. First, costs dictated the choice of a cross-sectional design
rather than a more ambitious panel study (i.e., where an attempt is made to
reinterview the same subjects year after year). Second, BAS organizers chose
face-to-face interviewing over other implementation methods because of the
greater control it offers over accuracy and quality of survey responses. Finally,
with few exceptions (described below), BAS questions are forced-choice items.
Again, this mainly reflects considerations of cost: before data input, open-
ended questions must be systematically coded, a difficult and time-consuming
process.
The BAS vision encompasses five goals, targeting different communities.
A primary goal is to gauge the impact of the reforms on the lives of ordinary
citizens, creating a valuable database by gradually, regularly, and frequently
accumulating substantial data on a standard set of indicators. This rich set of
empirical materials serves as a basis for systematic investigation into Chinese
social change by the scholarly community. Second, the BAS is designed to pro-
vide a reliable empirical foundation for policy recommendations to govern-
ment (and even to enterprises), as relevant. The Beijing Municipal Government
Policy Research Office offered highly valued sponsorship in the project’s early
years. Third, the BAS aims to promote by example the development of quanti-
tative empirical investigation and analysis in Chinese social science. A fourth
aim has to do with graduate education. As a major ongoing project of a univer-
sity research center, the BAS is an educational tool. Many cohorts of graduate
students at six universities in Beijing have participated in the project – imple-
menting the sampling design, interviewing respondents, reviewing fieldwork,
238 Mingming Shen and Ming Yang, with Melanie Manion

coding responses to open-ended questions, and entering data. Several cohorts

of graduate students associated with the RCCC have honed their statistical
skills by analyzing BAS data to produce research papers and graduate the-
ses. Finally, the BAS is a bridge between Chinese and nonmainland academic
communities, sharing data on Beijing’s social and economic development.
The RCCC works with nonmainland scholars to craft appropriate items that
reflect particular research interests. These items are “piggybacked” onto the
standard BAS questionnaire. Nonmainland scholars can obtain a unique
high-quality set of individual-level data that includes their own items as well
as a large assembly of standard social and demographic indicators. For exam-
ple, studies by Johnston (2004, 2006) and Johnston and Stockmann (2007)
draw largely from a time series of mainland Chinese views on foreign affairs
based on responses to questions designed by Johnston in collaboration with
the RCCC and included in the BAS from 1998 through 2004.
Certainly, the most interesting and important feature of the BAS is its
extended series of survey data representing the same population, the popula-
tion most easily accessible to the RCCC at Peking University. Beijing, at the
center of national politics, is obviously a distinct environment. Its residents
are by no means representative of all mainland Chinese or even of all urban
Chinese. At the same time, from everything we know, ordinary Chinese who
live in Beijing are probably also more highly attuned to policy changes than
are many other urban mainland Chinese. In this sense, then, the views of this
distinct population may be of intrinsic interest.

Questionnaire Content
As noted above, a key point of departure for the BAS was the policy con-
text: a time of major economic and social change in China. A simple work-
ing hypothesis emerged in the early stages of the questionnaire design: if a
majority in society have a supportive orientation toward the changes and
can accept disruptions brought about by them, then change can be carried
out smoothly. Further, to the extent that people see that change has actually
brought them benefits, their confidence in prospects for the future, their level
of understanding of change, and their ability to accept disruptions brought
about by change all increase – and the possibility of successful change is
thereby increased.

Core Questions
Core items on the BAS questionnaire include a wide variety of commonly
used individual-level social and demographic indicators: sex, age, education,
occupation, type of workplace, income, housing, marital status, and length
of residency in Beijing, for example. These items, included annually, are not
completely unproblematic to design well, but they pose relatively fewer chal-
lenges than other questionnaire items.
Measuring Change and Stability over a Decade 239

Table 13.1. Questions asked annually in the Beijing Area Study

Subjective ranking of household social class

Evaluation of household living conditions compared to last year
Evaluation of relative housing conditions (added in 1996)
Satisfaction with living in Beijing
Evaluation of achievements of “reform and opening”
Level of benefits to household from “reform and opening”
Evaluation of current national economic situation
Estimate of national economic situation one to two years in the future
Estimate of national economic situation five to ten years in the future
Evaluation of Beijing’s current economic situation
Estimate of Beijing’s economic situation one to two years in the future
Estimate of Beijing’s economic situation five to ten years in the future
Evaluation of seriousness of problems in Beijing: market management, health
­services, income distribution, education, prices, unemployment, social stability,
environmental protection, traffic management, communications, energy ­supply,
city government construction, housing, social welfare, city appearance and
­sanitation, “floating population”
Among the above problems, the three most serious problems in Beijing
Whether Beijing government has taken action on most serious problems
If action taken, effectiveness of action
Whether Beijing government has enough power to address most serious problems
Whether Beijing government has enough resources to address most serious problems

As respondent fatigue affects response, interviews cannot be too long;

this means that every questionnaire item carries an opportunity cost. Initial
decisions on core items to be included annually affect the future of all BAS
questionnaires as they take up space that might otherwise be used to pursue
changing research agendas or to explore the impact of new events. Some core
items that seem highly important in the mid-1990s may lose their importance
after a number of years. Should these items then be dropped, losing continu-
ity in measurement and gambling that they will never regain importance?
Question wording is also relevant here: whether expressions such as “reform
and opening” (改革开放) are easily understood by ordinary Chinese decades
after their introduction depends in part on policy changes. BAS core questions
mainly reflect caution (i.e., little or no change) on both these issues.
Core BAS items include questions asked annually since 1995 and questions
asked every few years. The former sorts of questions number about two dozen,
depending on how one counts items. These are listed in Table 13.1 and can be
grouped into five categories. A first set of questions has to do with household
living conditions, subjective social class, and satisfaction with living in Beijing.
The question on housing conditions, added in 1996, is an example of learning
from previous findings: housing emerged as one of the three “most serious
problems in Beijing” in another question. More generally, the BAS designers
reasoned that growing sales of commodity housing would produce significant
240 Mingming Shen and Ming Yang, with Melanie Manion

variation across urban Chinese residents on a basic livelihood issue. A second

short set of questions asks respondents to evaluate the achievements of the
“reform and opening.” A third set of questions asks for evaluations of the cur-
rent and future economic situation, both in Beijing and nationally.
The last two sets of questions are more politically sensitive. One focuses on
problems in Beijing in sixteen specific issue areas. This is an important and
fairly time-consuming set of questions. Respondents are asked whether there
is a problem in this issue area; if they respond “yes,” they are asked whether
the problem is serious. They are then asked to name the issue areas with the
three most serious problems in Beijing. In 1999 and 2000, the BAS added
open-ended questions that probed the exact nature of the problem for every
issue area where respondents indicated a problem existed. This fourth set of
questions sets up the most explicitly political set of core questions, which
focuses on the Beijing Municipal Government. These questions ask about pol-
icy measures taken by the municipal government to address each of the three
most serious problems noted by the respondent: whether the government took
action, the effectiveness of the action (if taken), whether the government has
sufficient power to address the problem, and whether the government has suf-
ficient resources to address the problem.
These core questions, asked annually, allow analysts to gauge change and
stability of views on basic issues of livelihood over time, something that is not
possible with other sorts of Chinese data. For example, focusing simply on
frequency distributions on a couple of policy-relevant and political items, we
find that the evaluations of benefits to the household from “reform and open-
ing” diverged greatly in the mid-1990s, with the preponderance of respon-
dents reporting few benefits; by the end of the 1990s, however, roughly half of
respondents reported few benefits and half reported many benefits. In another
area, we find the “floating population” and social stability dominating the
three most serious problems in Beijing for successive years; unemployment,
health services, and housing are also among the most serious problems for
high proportions of respondents. Finally, we find stability of responses on
the effectiveness of the Beijing Municipal Government at solving the city’s
most serious problems: except in 1998, when the proportions are roughly
equal, about 60 percent of respondents rate government response as effective.
Obviously, with these and all core items, the BAS demographic data also allow
us to analyze data (i.e., rather than simply report descriptive findings) and to
examine change over time across different subsets (e.g., by age, income, edu-
cation) of Beijing residents.
Core questions in a second category are now asked annually but were
added to the BAS after its initial inception. A large battery of questions on
views about foreign countries is an example of this. These questions were
added to the BAS in 1998 and asked annually thereafter. They include use of a
“feeling thermometer” to gauge feelings toward more than a dozen countries
(some added after 1998); evaluations of Japan, China, and the United States
as well as Japanese, Chinese, and Americans; and questions on the degree of
Measuring Change and Stability over a Decade 241

bellicosity of various powerful countries. Media consumption patterns are

another issue area in which questions were added to the BAS after 1995 but
then asked annually.
Core questions in a third category are asked regularly, but not annually.
Items measuring a wide range of values are a good example of this category of
questions. Values tend to be fairly stable, which suggests that they do not need
to be measured annually. At the same time, with major socioeconomic change
in China, it is important to investigate the degree of value change and stability
as an empirical question. The BAS measures specific values every three years.
These include equality of the sexes, income equality, equality of opportunity,
post-materialism, moral conduct, competitive conduct, traditional Chinese
values (e.g., collectivism), nationalism, and the relationship between the state
and the individual. Each value is measured with several questions.

Questions Asked Infrequently

In addition to core questions, the BAS asks some questions infrequently, usu-
ally only once. For example, in 1998, to commemorate the twentieth anni-
versary of reform, the BAS asked three open-ended questions that called on
respondents to reflect on the past two decades and report three important
events since 1978, three people with significant influence since 1978, and three
important changes in ways of thinking since 1978. Interestingly, the highest
proportion of respondents (39 percent) recalled June 4, 1989, as one of three
important events.
Most questions that appear in only one year have to do with current “hot
issues,” with several items used to measure orientations toward a single
issue. These include corruption (1995), social stability (1995 and 1996), envi-
ronmental protection (1997 and 2001), transfer of Hong Kong sovereignty
(1997), worker furloughs and a guaranteed minimum income (1997), the
Asian financial crisis (1998), China’s application to host the 2008 Olympics
(2000), Chinese entry into the World Trade Organization (2002), Severe Acute
Respiratory Syndrome (2003), and the “Three Represents.”2 These issues are
potentially politically sensitive. The BAS designers simply use their own judg-
ment in deciding what sorts of questions are too sensitive to include, but this
is always an imprecise art. For example, in 1995, when the first BAS was being
designed, corruption was by far the “hottest” issue for ordinary Chinese in
the capital. Beijing party secretary and Politburo member Chen Xitong was
under investigation for corruption, his longtime associate and Beijing deputy
mayor Wang Baosen had committed suicide in advance of questioning, and

2
The “Three Represents,” enshrined in the Chinese Communist Party constitution in 2002, are
the doctrinal contribution of Jiang Zemin, party leader at the time. They assert that the party
represents advanced social productive forces, advanced culture, and the interests of the major-
ity. By this doctrine, the party shifted its identity from a Leninist revolutionary vanguard; the
change also paved the way for admission of capitalists into the party.
242 Mingming Shen and Ming Yang, with Melanie Manion

the Central Discipline Inspection Commission had launched a nationwide

anticorruption campaign. Not surprisingly in this context, BAS designers
considered questions about corruption too politically sensitive to include in
the questionnaire – only to have the Beijing Municipal Government suggest
their addition.

Sampling
Sampling is perhaps the most challenging issue faced by BAS designers, elic-
iting regular minor adjustments and (in 2007) two major adjustments. The
BAS uses cross-sectional sampling to select a probability sample: every year, a
similar sample is newly drawn from the same population of Beijing residents.
Up through 2007, sampling procedures ensured generalizability of the sample
to the population of individuals aged 18 to 65 years, with a nonagricultural
Beijing residence permit, living in a fixed domicile in one of Beijing’s eight
districts.3 Based on previous experience with surveys, the RCCC expected a
response rate of 80 percent to 85 percent. Taking this into account, to reach a
95 percent confidence level that the sample is generalizable to the population,
BAS samplers drew a sample of 1,200 households (with a target of about 1,000
completed interviews) using probability proportionate to size (PPS) sampling,
with two-stage sampling selection.
Within each of the eight districts of Beijing are multiple street offices (街道办
事处), each of which manage dozens of neighborhood committees (居民委员会);
under each neighborhood committee are residential small groups (居民小组)
of varying numbers. After long consideration about appropriate primary sam-
pling units (PSUs), the BAS chose neighborhood committees as PSUs and indi-
vidual households registered with the neighborhood committees as secondary
sampling units (SSUs). BAS samplers initially selected a PPS sample of 65
(from a total of 3,500–4,800) neighborhood committees. BAS field supervisors
then confirmed that the committees had not disappeared in a major adminis-
trative reorganization within the city and that they did not include commit-
tees in which nongovernment surveys are prohibited (residences owned by the
military, for example). From the neighborhood committees that survived these
checks, PPS methods (with households as the scale measure) were used to
select 50. From the 50 selected neighborhood committees, a probability sam-
ple of 1,200 households was selected, 24 in each neighborhood committee.
BAS field supervisors recorded adult resident name-lists corresponding to the
selected households, and then randomly selected respondents from the lists.
The number of interviews completed and response rates are shown in Table
13.2. Actual response rates have varied from a high of 87 percent in 1995 to a
low of 66 percent in 2003, with generally lower rates in recent years.

3
In 1995 the age range was defined as 18 to 74 years; in 1996 and 1997 it was redefined as 18
to 70 years. Beginning in 1998, the BAS used 18 to 65 years as the age range for respondents.
Measuring Change and Stability over a Decade 243

Table 13.2. Overview of sampling and survey implementation in BAS first

decade

Year Sample Requirements Requirements Interview Response

size unmet met completed rate
1995 1,189 134 1,055 916 87%
1996 1,074 132 942 811 86%
1997 1,048 108 940 791 84%
1998 1,075 104 971 756 78%
1999 1,010 69 941 712 76%
2000 1,101 96 1,005 757 75%
2001 1,072 218 854 615 72%
2002 1,055 181 874 662 76%
2003 1,019 185 834 551 66%
2004 1,099 213 886 617 70%
Note: Response rate refers to completed interviews as a proportion of sampled individuals who,
at the time of the interview attempt, actually met survey requirements. Interviews with these
individuals may not be completed if the respondent is not at home on several interview attempts,
if the respondent’s health does not permit an interview, or if the respondent refuses to be inter-
viewed. See Table 13.3 on incomplete surveys due to unmet requirements.

Each year a proportion of selected respondents cannot be interviewed.

Some incomplete interviews are due to respondents who are not found at
home on several interview attempts; an increasing proportion is respondents
who refuse to be interviewed. These problems are endemic to survey research.
In 1995, BAS interviewers were instructed to make three attempts to interview
respondents; in 1998 this was changed to five attempts. In addition, however,
many incomplete interviews were due to reasons associated with reliance on
outdated household registration lists, as shown in Table 13.3.
Minor sampling adjustments in the BAS have included changes to the neigh-
borhood committee sampling frame on the basis of city administrative reorga-
nization; also, in 1999 and 2004, the BAS directly obtained from the Beijing
government the most up-to-date sampling frame. Neither of these adjustments
addressed the source of the noninterview problem: reliance on household reg-
istration lists.
On the one hand, relaxation of the household registration system (户口) has
produced greater population mobility in the past couple of decades. More than
100 million rural-to-urban migrants, the “floating population” (流动人口),
are an important part of this, with the result that a large proportion of the
Beijing population has official residence outside Beijing. Further, among those
with official residence in Beijing, a large number do not live at their official
household residence (人户分离). The traditional BAS sampling method did not
pick up these migrants and movers. This is a significant population: about
45 percent (25 percent migrants and 20 percent movers) of a Beijing sample
244 Mingming Shen and Ming Yang, with Melanie Manion

Table 13.3. Incomplete interviews due to unmet requirements in BAS first

decade

Year Requirements Incorrect Mismatch of Survey

unmet address household and status unmet
dwelling
1995 134 54% 43% 2%
1996 132 67% 29% 4%
1997 108 74% 19% 7%
1998 104 63% 24% 13%
1999 69 57% 43% 0
2000 96 57% 43% 0
2001 218 12% 82% 6%
2002 181 40% 55% 4%
2003 185 50% 42% 8%
2004 213 77% 15% 9%
Note: Incorrect address includes instances of relocation (搬迁); mismatch of household and
dwelling means the selected respondent does not reside at the dwelling listed for official resi-
dency (i.e., 空挂户 or 人户分离). Survey status unmet refers to a situation where the selected
respondent turns out to be older than 86 years old and there is no other individual in the house-
hold who meets survey status.
Percentages may not add up to 100 due to rounding.

selected by the global positioning system (GPS) in 2000 could not be reached
using the traditional BAS sampling method (Landry and Shen, 2005).4
Beginning in 2007, the BAS has used GPS sampling.
The increased pace of urbanization poses a second major sampling chal-
lenge. The original eight districts surveyed in 1995 no longer represent the
core city districts of Beijing. Beginning in 2007 the BAS added six new city
districts to the sample. The targeted sample size was increased to 1,500 to
accommodate addition of the districts.

Survey Implementation
The BAS is assigned high priority as an RCCC special project each year, with
a single project leader responsible for directing all its aspects. Survey imple-
mentation involves two teams: field supervisors and interviewers. BAS field
supervisors are RCCC research assistants or Peking University graduate stu-
dents; BAS interviewers are students recruited from six universities in Beijing.
Interviewer training includes an initial day and a half of training, followed by
a substantial survey pretest, followed by further training. The survey pretest
serves two functions: to pretest the questionnaire (which changes annually)
for readability and understandability, and to provide practical experience for

4
On GPS sampling, see Chapter 12 of this volume and Landry and Shen (2005).
Measuring Change and Stability over a Decade 245

interviewers. After the pretest, the questionnaire is adjusted, based on the

debriefing of interviewers and looking for signs of problems suggested by the
response distributions across questions (e.g., no variation, high “no response”
rates, etc.).
The actual survey is conducted in two stages. The first, intensive stage
involves the 8–17 field supervisors leading their teams of interviewers (50–83
altogether) into the neighborhoods to conduct interviews all at the same time.
The second stage focuses on recontacting households where identified respon-
dents could not be interviewed in the first stage (e.g., because they were not
at home); this stage is not intensive but takes cost into account and is often
preceded by arranging appointments with respondents.
To ensure interview quality, interviewers are compensated by the hour, not
by the interview. Also, completed questionnaires are checked in the field (by
interviewers and field supervisors) and by the project manager at the RCCC –
all of whom must “sign off” on the completed questionnaire itself. A further
check is conducted during data input: logical inconsistencies in responses and
suspicious questionnaires are tagged and checked again.
The BAS has faced challenges in survey implementation since 1995, some
noted above. Other challenges include difficulty in access to sampled house-
holds due to the rise of gated communities and continually rising costs asso-
ciated with survey work (e.g., printing of questionnaires, transportation of
interviewers). Over the years, the BAS has been funded from various sources –
including private foundations, the Beijing Municipal Government, and foreign
scholars who add research modules of their own questionnaire items to the
survey.
14

Quantitative Research and Issues of Political

Sensitivity in Rural China

Lily L. Tsai

Political sensitivity is always a challenge for the scholar doing fieldwork in

nondemocratic and transitional systems, especially when doing surveys and
quantitative research. Not only are more research topics likely to be politically
sensitive in these systems, but in trying to collect precise and unbiased data
to give us a quantitative description of a population, we are sometimes doing
exactly what the government – and sometimes certain members of that popu-
lation – would like to prevent. In this chapter, I discuss some of the method-
ological and ethical issues that face researchers working in these contexts and
describe strategies for dealing with these issues. I argue that in these contexts
a “socially embedded” approach to survey research that carefully attends to
the social relationships inherent in the survey research process can help allevi-
ate problems of political sensitivity, protect participants and researchers in the
survey research process, and maximize data quality.
For this chapter I draw on my experience conducting a village-level survey
on village conditions of officials in 316 villages in rural China in 2001 as part
of the twenty months of fieldwork I conducted for my doctoral dissertation
and book, Accountability without Democracy: Solidary Groups and Public
Goods Provision in Rural China (2007a). Unlike an individual-level opinion
survey of the mass public, this survey focused on village-level institutions and
outcomes and interviewed one or more village officials in each village as infor-
mants on their village’s economy, politics, and society.
After an overview of the project’s objectives and research design, I discuss
the difficulties I encountered in trying to conduct quantitative research in rural
China generally, and in studying politically sensitive questions more specifi-
cally. These difficulties will be familiar to anyone who tries to collect quan-
titative data, whether on individuals or communities, in rural China. I then
describe the methods I used to try to overcome these difficulties and evaluate
their strengths and shortcomings when used for a village-level informant sur-
vey. I conclude with a brief discussion of the appropriateness of these methods
for individual-level attitudinal surveys and how the trade-offs between bias and

246
Quantitative Research and Issues of Political Sensitivity 247

variability, interviewer and respondent effects, validity and generalizability,

and allocation of resources may differ.

Overview of the Research Project

For this project, I was interested in accounting for variation in local governmen-
tal performance and public goods provision. More specifically, I was interested
in evaluating the effects of formal bureaucratic and democratic institutions of
accountability and informal institutions of accountability provided by com-
munity religious and lineage groups. After conducting preliminary fieldwork
in seven provinces, I decided on a multistage, multimethod research strategy
combining ethnographic study of a single set of villages in Fujian province over
four months, a survey of 316 villages and 948 households in four provinces –
Shanxi, Hebei, Jiangxi, and Fujian – and a structured comparison of in-depth
village case studies selected from the same four provinces. The fieldwork took
place over twenty months from 1999 and 2002.
In the first stage of my fieldwork, I focused on developing a detailed under-
standing of local governance and political processes. During this period, I
was based in Xiamen and repeatedly visited four villages in the area almost
every week (and villages in other parts of China less frequently). Sometimes
I interviewed villagers and village officials. At other times, I simply observed
everyday interactions between villagers and daily administrative work by vil-
lage officials. I also periodically attended village meetings, participated in com-
munity festivals and social gatherings, and followed the informal politicking
behind the scenes of the village elections in the year 2000. Through these vis-
its, I discovered a variety of community groups and institutions that often
dominate village life and village politics but are sometimes hidden. Officially,
community groups were required to be registered with the state. Unofficially,
township and county governments often looked the other way. In one village,
for example, residents and village officials alternately referred to the unregis-
tered community council of villagers associated with the village’s temples as
the village council, the temple council, the senior citizens’ association (老年
人协会), the village elders, or the state-approved wedding and funeral council
(红白理事会).
My time in Xiamen gave me a basic understanding of political and social
interactions at the village level, but it was based on only four villages. The
next step was to collect data that would enable me to generalize about village
governmental performance and public goods provision for a broad range of
villages. The National Bureau of Statistics does not collect data on the provi-
sion of public services at a level as low as the village, and few studies of rural
governance had systematically addressed this topic. I thus designed an original
village survey to collect statistics on village-level provision of public services,
village public finance, township-village relations, village democratic institu-
tions, and community social institutions.
248 Lily L. Tsai

Since foreigners are not allowed to administer surveys in China, I needed to

find Chinese researchers who would be willing to act as guarantors and help
me administer my survey. I pursued discussions with three researchers who
had contacts in four different provinces: one, a researcher in the Ministry of
Agriculture, who had contacts in Shanxi and Hebei provinces; another, a pro-
fessor in Jiangxi, who could arrange for a survey in that province; and a third,
a professor in Fujian, who could potentially expand the sample for a survey
he was already conducting in that province to accommodate my survey and
sampling requirements.
These four provinces – Shanxi, Hebei, Jiangxi, and Fujian – varied along
two important macro-level dimensions. Coastal and inland regions dif-
fer significantly from each other in terms of economic development, and
north and south China vary greatly in their institutional history and social
organization. To make my findings as generalizable as possible, I sought
to conduct the survey in two provinces in north China and two provinces
in south China. Within each pair, one province was coastal and one was
inland. Pursuing leads for administering the survey in these four particu-
lar provinces also made sense in terms of backup plans because the survey
design would still make sense if one or two of the leads were to fall through.
Administering the survey in two northern provinces or two southern prov-
inces, for example, would allow me to hold geographical factors constant
while varying the level of development. (For more detailed information on
the research design, see the first chapter of my book Accountability without
Democracy, 2007a).
To my surprise, however, all three of these people eventually agreed to help
me administer my survey. Within each of the four provinces – Shanxi, Hebei,
Jiangxi, and Fujian – two counties were selected purposively according to a
combination of theoretical and practical requirements. Although selecting
counties within each province randomly would have increased the generaliz-
ability of my findings, my primary concern was maximizing the validity of
the data. To control the quality of the data, I wanted to supervise the survey
administration personally. The Chinese researchers who had agreed to assist
me thus had to draw on local government contacts who would be willing to
host a foreign graduate student, and the only contacts they had who were will-
ing to do so were in county governments. I thus decided to select two neigh-
boring counties in each province that were as similar as possible except that
one would be a model county for village elections and the other would not (see
book for details). With the help of these Chinese researchers, I was fortunate to
have complete control over questionnaire design and survey sampling within
each of the eight counties. Within each county, a random stratified sample of
forty villages was selected.
For the Shanxi and Hebei portion of the survey, I selected and trained a
team of eleven graduate and undergraduate students from Beijing. In Jiangxi,
I trained a team of twenty graduate students from the provincial party school,
and in Fujian, I trained a team of twelve or so undergraduate students from
Quantitative Research and Issues of Political Sensitivity 249

Xiamen University. The survey took approximately eighteen days per province
to administer in Shanxi, Hebei, and Fujian. In Jiangxi, we had twice as many
survey interviewers so administration time was cut in half. In each province, I
traveled with the interviewers to the countryside to supervise the administra-
tion of the survey directly. Every night or two, I met with the survey interview-
ers to discuss problems and issues arising in the field and to check through the
questionnaire. When I found errors or skipped questions, survey interviewers
called or revisited the respondents to make corrections and fill in the blanks.
To correct problems in the field, an assistant and I used laptop computers to
code and input the questionnaires each day.
In the final stage of my fieldwork, I wanted to evaluate whether community
social institutions really affect village governmental performance in the ways
that findings from the survey suggested. To check whether these findings really
made sense in a variety of different cases, I put together a set of in-depth vil-
lage case studies selected from the same four provinces in which the survey had
been conducted. These case studies allowed me to explore the causal processes
underlying the correlations identified by statistical analysis, make inferences
about interaction effects between different explanatory variables, and gather
more observations of the implications of the theories being tested. Gathering
data through case studies also helped to trace the evolution of a village’s politi-
cal and social institutions and understand how local historical and cultural
contexts shaped these institutions.

The Challenges of Quantitative Research

and Political Sensitivity in Rural China
First, the greatest challenge to doing quantitative research on nondemocratic
and transitional systems like China is to obtain high-quality data. In the case
of rural China, if we want village-level data, we have to collect them ourselves.
Official statistical yearbooks published by the National Bureau of Statistics
contain data aggregated at only the county level (with a little data aggregated
at the township level). Another major problem is that nondemocratic and tran-
sitional systems rarely collect – or if they collect them, rarely publish – data
on many of the political and social variables in which we are interested. In my
case, official data on voluntary associations, for example, which are available
for many Organisation for Economic Co-operation and Development (OECD)
countries, are not available for rural China for two reasons. First, the govern-
ment does not collect such data. Second, many groups that we might think of
as voluntary associations are unregistered and often deliberately try to avoid
notice by the official authorities.
These two reasons correspond to two different ways in which the collection
of quantitative data is politically sensitive. First, as in other nondemocratic
and transitional systems, survey research itself is politically sensitive in China.
Quantitative descriptions of a population can often be used as a measure of the
performance of particular officials or local governments. Not surprisingly, all
250 Lily L. Tsai

levels of the state thus seek to control the collection and flow of statistical data
(Cai, 2000; Huang, 1995). Local officials have strong incentives to manipulate
the reporting of statistics – grain output during the Great Leap Forward, income
per capita and industrial output during the reform period – so as to portray
their performance in the best light possible. China’s level-by-level reporting
of official statistics facilitates this manipulation. Because government officials
have so much at stake when it comes to statistical data, citizens also find survey
research a politically sensitive matter. As Belousov et al. (2007: 163) note for
Russia and other post-Soviet states, “there is still a general fear of answer-
ing questions per se.” This political context makes survey research by both
Chinese and foreign academics a politically sensitive matter.
Nevertheless, survey research and quantitative data collection by foreign-
ers is a particularly sensitive issue. As Melanie Manion discusses in more
detail in Chapter 10 of this volume, the Interim Measures for Administration
of Foreign-Related Social Survey Activities issued by the National Bureau of
Statistics in 1999 list numerous restrictions on survey research with foreign
participation. Any foreigner contemplating survey research in China should
look at the regulations in full.1
Second, the collection of quantitative data can also be politically sensitive
when the content of our research topics is politically sensitive. To complicate
matters, in an authoritarian or transitional system all sorts of topics may be
considered politically sensitive, and a researcher does not always know which
topics are politically sensitive. Topics that do not seem explicitly political may
be politically sensitive. What is considered politically sensitive may also vary
across regions and over time. Once, while chatting with a village official in
Fujian, I offhandedly observed that ancestral graves were interspersed with the
village fields, and the official suddenly fell silent and then changed the subject.
I later found out that local officials in that area had been struggling to enforce
a new regulation requiring cremation as well as trying to convince villagers
to cremate the remains of already buried ancestors in order to increase arable
land. The conflict had already led to more than one violent clash between vil-
lagers and officials. In another village, in Jiangsu province, I asked officials if
I could use the village’s public toilet. Again, the village officials fell silent. We
had been talking about public projects in the village, and the officials had high-
lighted their investment in the large new public toilet – centrally located in the
village, beautifully tiled, with a flushing system. They had, they emphasized,
gone above and beyond the targets set by the county’s recent sanitation cam-
paign for village public toilets. Instead of showing me to the toilet, however,
I was politely ushered out of the village. Confused, I asked the driver as we
left the village why I hadn’t been allowed to use the public toilet. The driver
explained that, in clear contravention of the spirit (if not the letter) of the
county’s campaign, village officials kept the toilet locked up so that villagers

1
These regulations are available online from the Supreme People’s Court of the PRC, at http://
en.chinacourt.org/public/detail.php?id=3897, accessed August 2009.
Quantitative Research and Issues of Political Sensitivity 251

would not dirty it. The man with the key was away that day so they were
unable to unlock it for me.
These examples highlight the distinction between topics that are politically
sensitive to government officials and topics that are politically sensitive to vil-
lagers. Residents of the Jiangsu village were happy to point out that village
officials had constructed the new public toilet only for show, whereas village
officials were understandably reluctant to highlight this fact. In the case of the
Fujian village, the new cremation and burial regulations were a sensitive topic
for both villagers and officials. If I had been surveying on the topic, I would
have had to frame my questions differently depending on whether my respon-
dents were villagers or officials. On the one hand, villagers might not have
responded well if I had asked them whether they were “complying” with the
new regulations, whereas officials would have been less likely to object to this
wording. Officials, on the other hand, might have responded poorly if I asked
how well they were enforcing the new regulations.
These examples also illustrate how questions can sometimes be politically
sensitive because disclosure of the truth can potentially harm the respondent
and sometimes because people feel uncomfortable talking about certain topics.
In the case of the public toilet, village officials did not want to admit to lock-
ing the toilet because their behavior was contrary to the policy objectives of
higher levels. The issue of digging up ancestral graves was not only politically
charged for policy-related reasons but for normative and historical ones as
well. Digging up ancestral graves violates deeply held moral and spiritual con-
victions and also reminds villagers of the state’s often violent efforts to stamp
out what it considered “feudal superstitions” during the Maoist period.
There is one last point worth making about political sensitivity and one’s
research questions: it is critical to know where the line is between subjects that
are politically sensitive and subjects that are taboo. Asking questions about
subjects that are taboo can destroy your ability to ask questions about subjects
that are politically sensitive. In his research on the guerrilla warfare of the Irish
Republican Army (IRA), Sluka (1990) found that he could ask people questions
about their support for and criticisms of the IRA but not questions about arms
or explosives, or who might actively be a guerrilla (Sluka, 1990: 114–126). In
the context of rural China, some questions related to the birth control policy,
for example, are politically sensitive; others are taboo. In one Hebei village
I visited, I talked with officials of the village’s branch of the state-mandated
women’s association about their responsibilities and activities. As they became
more comfortable with me, they described how one of the ways in which they
encouraged villagers to follow the birth control policy was to perform com-
edy skits (小品) that mocked “out-of-quota birthing guerrillas” (超生游击队),
villagers who go into hiding in order to give illegal out-of-quota births. They
were comfortable gossiping about frequent cases like these in their locality. If,
however, I had asked questions about illegally coerced abortions and steriliza-
tions, they might very well have stopped telling colorful stories and reported
me to the local authorities. Similarly, questions about underground Christian
252 Lily L. Tsai

churches were politically sensitive but questions about the Falungong were
taboo. In another area of Hebei, a township official felt comfortable telling me
that the greatest fear of the county government was the proliferation of under-
ground household churches in the area. Shortly after this discussion, however,
he volunteered that of course there were no Falungong activities in the area.
Since I had not asked about the subject, his comment gave me the impression
that probing into Falungong activities would be taboo.
In short, issues of political sensitivity complicate the collection of quan-
titative data in rural China in various ways. First, arranging for the admin-
istration of a survey is a challenge. Foreign researchers have to find Chinese
collaborators who are willing to take responsibility for conducting the sur-
vey, able to gain access to research sites and respondents in the sample, and
willing to vouch for your trustworthiness. Second, researchers have to worry
about getting respondents to give truthful and precise responses to politically
sensitive questions. We have to know what the political incentives and sanc-
tions are for giving certain answers. Sometimes there may be pressures on
respondents to decline answering a question or to give only a vague answer.
At other times, there may be pressure for respondents to avoid answering a
question and yet appear as if they are answering the question to the best of
their ability.
Third, researchers have to accommodate a suspicion of survey research in
general. Both local officials and villagers can be uncomfortable and wary of
being interviewed by people they do not know from outside their locality. For
some people, the basic format of a survey interview may be unfamiliar or remi-
niscent of unpleasant interrogations by state agents. In these contexts, innova-
tive question formats and questionnaire designs such as anchoring vignettes
(King, 2004: 197–207) or list experiments (Streb, 2008) may actually raise
suspicions and undermine data quality. When I tried a simplified version of the
political efficacy vignette described by King et al. (2003), I not only had trouble
securing the cooperation of respondents but also when they agreed to partici-
pate, they were extremely confused by the format. A few even reacted by walk-
ing away, leaving me alone in their house. List experiments raised suspicions
among respondents that they were being tricked in some way because they
found it hard to understand how the question worked. Finally, as with almost
everything about doing research in China, there can be tremendous variation
in political sensitivity issues across regions and individuals.

Strategies for Accommodating Political

Sensitivity Issues in Quantitative Research
Before I go on to discuss some of the methods I used to accommodate the dif-
ficulties associated with the various issues of political sensitivity in collecting
quantitative village-level data in rural China, I want to emphasize that one
should always be willing to change research topics due to issues of political
sensitivity. No academic project is more important than the safety and security
Quantitative Research and Issues of Political Sensitivity 253

of the people involved in the project.2 We have an ethical imperative to “do

no harm.” As Elisabeth Wood notes, “there are some settings where research
cannot be ethically conducted and should not be attempted or should be cur-
tailed” (Wood, 2006: 373–386). Wood provides a valuable discussion of the
research procedures that she followed in order to implement the “do no harm”
ethic during her fieldwork in El Salvador.
An important strategy for learning how to “do no harm” and when to
change research topics is to do qualitative research before attempting the col-
lection of quantitative data. The initial stages of my fieldwork – a preliminary
two-month trip, six months in Fujian, and several short trips to Shanxi and
Hebei – were invaluable in helping me to design and pretest survey questions
(Park, 2006:128). During these stages, I allocated much of my time simply to
chatting conversationally with villagers and local officials in the different prov-
inces about their lives and their communities in general. Like Wood, I found
that rural residents I interviewed had far more political expertise than I did and
a far better sense of what was politically risky. The more time and opportunity
I gave them to teach me about the specifics of political sensitivity, the safer and
more productive my subsequent research was (Wood, 2006: 380).
Although I always had my main research questions in the back of my mind,
I also just wanted to get to know people as much as possible. When I knew
I would have multiple chances to talk to someone, I often waited until later
meetings to ask the person the questions on my structured interview schedule
or on my draft surveys. The sociologist Ned Polsky’s first rule of field research
worked well: “Before you can ask questions, or even speak much at all other
than when spoken to, you should get the ‘feel’ of their world by extensive
and attentive listening – get some sense of what pleases them and what bugs
them, some sense of their frame of reference, and some sense of their sense of
language.” At the same time, it was also important to answer their questions
about my background. As Polsky also writes, “it is important that [the inter-
viewee] will be studying you, and to let him study you.. . . He has got to define you
satisfactorily to himself if you are to get anywhere” (Polsky, 2006: 128, 132).
This approach allowed me to find out which topics were easy to bring up
and which topics were off limits, and how the line between sensitive and taboo
varied from region to region. Often I would ask people to give me tours of
their neighborhoods, which often gave me the occasion to ask about something
that we saw – an abandoned road project, a Catholic church in the center of
town, a twenty-foot gully full of garbage, or a burned-out storefront. This
strategy resulted in interesting stories involving corruption, competition for
congregants between local Catholic and Protestant churches, conflict between
villagers and officials, and conflict among lineage groups. Chatting socially
with people also allowed them to bring up local current events, which some-
times touched on my research interests – a scandal in a neighboring county
where a local journalist reporting on local government investment in irrigation

2
See Howell (1990), Sluka (1990), and Barrett and Cason (1997).
254 Lily L. Tsai

reached down to show TV cameras a new irrigation pipe and it came out of the
ground, attached to nothing; or rumors about a contentious village election in
the area that had resulted in one of the candidates being lured to a karaoke bar
and stabbed. Although these kinds of rumors did not constitute reliable data,
they gave me a valuable sense of the political climate and a context for gauging
the topics people felt were politically sensitive and the ways in which they were
willing to talk about these topics.
A period of qualitative research also taught me about regional variation in
the political sensitivity of particular topics. The topic of underground Christian
house churches was very sensitive in Hebei but openly discussed in Fujian. In
some Fujian villages where conflict between lineages had erupted into physical
fights among villagers, local officials explicitly warned me not to ask about
it. By contrast, in Hebei, villagers freely recounted the long-standing feuds
among a village’s sublineages and found amusing the different ways in which
the sublineages tried to sabotage each other. There was also variation within
provinces. In eastern Fujian, I found that local officials felt they had to justify
the existence of unregistered village temple councils by talking about how they
contributed to village public goods provision and social stability. In western
Fujian I talked to local officials who simply stated that villagers did not trust
township and village cadres and informal villager councils were now running
the villages.
After this first stage of fieldwork, it became clear to me that I would have to
attend to data quality before pursuing generalizability. Getting survey respon-
dents to give accurate and truthful answers to politically sensitive survey ques-
tions was going to be a primary concern. I therefore made a conscious decision
to maximize the validity and reliability of the data rather than the generaliz-
ability of the findings. Drawing more valid conclusions about a smaller popu-
lation seemed like a more sensible way to build knowledge than drawing less
valid conclusions about a larger population. In theory I could have hired a
market research firm to administer the survey nationally or piggybacked on an
existing national survey by adding questions to an existing survey instrument.
However, after witnessing firsthand how much responses to politically sensi-
tive questions could vary depending on how comfortable respondents were
and how they perceived the person doing the asking, I decided that I needed as
much freedom, control, and participation in the actual administration of the
survey as possible. This decision guided my sampling for the survey. Within
each of the eight counties in the survey I used a multilevel stratified random
sampling strategy to select villages, but I selected both the provinces and coun-
ties purposively based on where the Chinese researchers assisting me had per-
sonal contacts who would allow me to conduct a large-scale survey freely.
Strictly speaking, this strategy limited the generalizability of the findings from
the survey data analysis to these eight counties, but it was crucial to maintain-
ing the quality of the data.
In the case of rural China, one is often forced to choose between obtaining
a nationally representative sample and controlling the local conditions under
Quantitative Research and Issues of Political Sensitivity 255

which the survey is administered so that the accuracy of the data is maximized.
Surveys based on nationally representative samples have become possible in
China, but once the terms of survey administration have been negotiated, for-
eign researchers participating in these surveys often have limited control and
leverage over the local conditions of the survey administration. The actual
administration of the survey is outsourced and often takes place quickly.
The extent to which one has to choose between data quality and generaliz-
ability, or internal validity and external validity, depends in part on the kinds
of questions one is trying to study. For research questions that are relatively
uncontroversial, controlling the local conditions of survey administration may
be less important. For these projects, the researcher may not have to choose
between obtaining a nationally representative sample and obtaining accurate
data. But for projects on potentially politically sensitive topics, researchers
have to worry not only about securing access to sampled research sites and
respondents but about creating an interview environment in which respon-
dents feel comfortable giving truthful responses to survey questions. In these
cases, obtaining a probability sample of a more limited population may be a
reasonable choice.

Conducting “Socially Embedded” Survey Research

To create this kind of environment and maximize the quality of quantitative
data collected on politically sensitive topics, I argue that researchers need to rec-
ognize that survey research is embedded in social relationships among research-
ers, official authorities, interviewers, and respondents. Moreover, researchers
need to invest in building and shaping these social relationships so that they
generate trust and mutual obligations.3 While this approach may sound obvi-
ous, many survey research projects in fact try to render the survey research
process as impersonal as possible. Researchers often pay firms or domestic
research institutions to conduct their surveys, treat official approvals as purely
bureaucratic hurdles, and seek to “standardize” interviewers and depersonal-
ize interviewer-respondent interactions to minimize interviewer error. Rather
than thinking about how to foster social relationships based on trust and recip-
rocal obligations with domestic collaborators and official authorities, survey
researchers often think in terms of principal-agent problems, incentives, and
monitoring (Fowler, 1993).
In the following sections, I discuss four types of social relationships that
influence the process of survey research and quantitative data collection, and
consider the ways in which these social relationships can generate trust and

3
The concept of social embeddedness I use here comes from Granovetter’s (1985) article on
economic exchange. Granovetter’s emphasis on the “role of concrete personal relations and
structures (or ‘networks’) of such relations in generating trust and discouraging malfeasance” is
the point I highlight here.
256 Lily L. Tsai

obligation to influence the quality of data on politically sensitive research

­topics (Granovetter, 1985: 482, 490).

1. Relationships between Foreign and Chinese Researchers

To collect quantitative data in China, foreign researchers must work with
Chinese researchers who are willing to take official responsibility for conduct-
ing the survey. In the case of surveys on potentially politically sensitive topics,
this responsibility is an especially serious one. It is not to be taken lightly by the
Chinese researcher or by the foreign researcher asking for the assistance of the
Chinese researcher. The fact that the Chinese researcher bears official respon-
sibility does not let foreign researchers off the hook. In asking for the help of a
Chinese researcher to collect data on potentially politically sensitive issues, the
foreign researcher is obligated to take responsibility for the security and well-
being of the Chinese researcher. Not only should we always listen to and defer
to our collaborators’ judgment on what is too politically sensitive, but if our
collaborators seem more daring than seems sensible, then it is also our respon-
sibility to rein them in. Although these collaborators may be right in judging
something to be perfectly safe, our responsibility for them requires us to listen
to our own judgment as well. It goes without saying that one should never do
something without the full knowledge and consent of one’s collaborator.
It is impossible to collect high-quality data without a skilled and reliable
collaborator. As Albert Park, an economist working on China, notes: “Nearly
all successful surveys in developing countries depend on the support of ener-
getic, capable research collaborators from the host country who know how
to get things done within the country’s institutional, political, and social envi-
ronment; are skilled at interacting with government officials and community
leaders; have developed reputations within the country that build trust, and
have valuable substantive insights into the research question. On the flip side,
collaborators pursuing agendas at cross-purposes with those of the researcher
can easily frustrate research plans” (Park, 2006: 122–123).
The more potentially politically sensitive one’s research topics are, the more
important it is to work with a collaborator whom one trusts. I grew to know
my collaborators and their families. We moved in the same professional circles
within China, we had mutual friends and acquaintances, which reinforced the
mutual trust and confidence necessary to collaborate on politically sensitive
survey research, and we continue to keep in touch.
Research collaborations always carry an ethical obligation to reciprocate
the other party’s time and efforts. As Park discusses, one can reciprocate by
providing intellectual benefits (providing them with useful ideas and tools for
their own research, acting as a guest lecturer, coauthoring papers, or facilitat-
ing a visit to one’s own research institution), material benefits (adequate com-
pensation for services), or personal benefits (developing personal relationships
and being friendly) (Park, 2006: 123). When, however, the research project
involves politically sensitive topics and political risk, relying solely on material
Quantitative Research and Issues of Political Sensitivity 257

compensation is unlikely to work – and if it seems to work, one should be

extremely cautious about proceeding.
One’s collaborator also has to have a finely honed sense of what political
concerns different types and levels of officials may have about the research
project. I usually found that researchers working for state ministries and gov-
ernment organs were better informed about the political sensitivity of particu-
lar issues and more experienced at negotiating people’s concerns. As a result,
they were generally more confident about tackling politically sensitive ques-
tions than researchers in universities or academic research institutes.

2. Relationships between Researchers and Official Authorities

After forging a research collaboration based on mutual trust and obliga-
tion, the next step is to build relationships with government officials whose
approval and support are needed. In many cases, the informal support is far
more important than the official approval. As Belousov et al. comment, “field-
work in difficult to access places often needs to be facilitated by key ‘gate-
keepers.’” Belousov et al. note that after their gatekeeper was murdered, his
“personal patronage” and “this informal status disappeared, even though the
formal agreements remained intact. While no-one now attempted to prevent
our research activity, in contrast with the earlier stage, nobody went out of
their way to help us either” (Belousov et al., 2007: 166).
Collecting valid and reliable data on potentially politically sensitive subjects
requires extremely careful attention to how the survey is administered and how
relationships between interviewers and respondents are structured. Control
over these aspects of survey administration in turn required a very high degree
of trust and confidence from local officials in the counties where I conducted
the survey. To achieve this level of comfort from local officials so that they
would not intervene in our survey of village officials, I had to work in places
where my Chinese collaborators had relationships with provincial, municipal,
and county officials. My collaborators generally went through contacts they
had made in their previous field research. When my sampling strategy required
us to work in a county where they did not have a contact, they would go
through a contact at the municipal or provincial level instead. Because my col-
laborators had already established relationships with local authorities that they
had worked at maintaining over time, local officials were willing to trust me
and to take more time to get to know us and our project, which also increased
their level of comfort with the survey administration.
As a result, we were given relatively free rein within each county, and county
and township officials did not attempt to intervene in the administration of the
survey. We were allowed to administer the survey in a village immediately
after it was sampled so that higher level officials had little opportunity to call
up sampled villages and debrief them on how they should respond to our ques-
tions. In most cases, I or one of my assistants accompanied the higher level
officials making arrangements for us and witnessed most of their telephone
258 Lily L. Tsai

interactions with lower levels. We were allowed to spend as much time as we

wanted in whatever area we wanted, which enabled us to probe the responses
of village officials to make sure they were giving us the most accurate answers
possible. We were allowed to talk with multiple village officials and, in some
cases, former village officials to corroborate information about village-level
conditions. Without this degree of freedom and the flexibility to adjust the
administration of the survey to local conditions, it would have been much
more difficult to ensure the accuracy of data on sensitive subjects such as the
existence of village religious activity or the collection of illegal local levies.

3. Relationships between Researchers and Interviewers

While forging relationships based on trust and reciprocity among researchers
and official authorities is essential for setting up the survey and setting up the
conditions for collecting high-quality quantitative data on politically sensitive
subjects, the people who are most important for ensuring the quality of the
data are the interviewers. They are the ones who are doing the actual col-
lection of the data. In the field, I realized that the efforts of the interviewers
depended heavily on my relationship with them. Even if my funds had not been
limited, it would have been hard to compensate them enough for undergoing
the hardship of administering a rural survey. Personally supervising adminis-
tration of the survey and traveling to research sites along with the interviewers
allowed me to build stronger relationships with them and to strengthen bonds
of mutual obligation and reciprocity. The more I was able to convince them of
the intellectual and social value of the project, the more effort they invested in
trying to obtain accurate data and the more they felt that they had a responsi-
bility to invest in this effort.
Leading by example and doing things to express my gratitude for their work
helped immeasurably to improve the quality of survey administration. These
things ranged from advising them on their theses to staying in the same accom-
modations to hand-washing their laundry when they were busy with survey
administration. I traveled with them to each locality and took all of the same
long-distance bus and train trips. When I spot-checked enumerators by drop-
ping in on them, I picked villages that were difficult to access as often as ones
that were easy to access. The more effort I showed, the more they realized how
important data quality was to me, and the more they realized how important
and valuable they were to the process, which in turn motivated them to put in
more effort.

4. Relationships between Interviewers and Respondents

The relationship between interviewer and respondent is the most immediate
and critical context for the generation of valid and accurate data. While in the
village, interviewers maximized the quality of the data from village officials by
spending a large amount of time visiting the village. Even though the survey
Quantitative Research and Issues of Political Sensitivity 259

focused on interviewing village officials as informants on village conditions,

interviewers also talked with villagers as well as current and former village offi-
cials. Interviewers typically spent half a day to a day in each village. Depending
on the time it took to travel to the village, interviewers sometimes stayed in a
village overnight.
We administered the survey as a genuine two-way conversation between
interviewers and village officials. When village officials gave responses that
seemed to conflict with their previous responses or with the personal impres-
sions enumerators had gained from walking around the village and talking with
villagers, interviewers would ask follow-up questions to probe their responses
more deeply and to reconcile contradictions. To corroborate the responses of
village officials, interviewers also asked for and were typically able to look
at supplementary village documents, including village account books, village
receipts, minutes from village government meetings, village election ballots,
and election records.
Conducting the survey as a conversation and spending a significant amount
of time visiting each village also helped interviewers and village officials get
to know each other as people. Not only did this process make village officials
more comfortable talking to the survey enumerators, but it also enabled us
to repay the village officials a little by providing them with information and
answering their questions about us, our research, and our backgrounds. The
more we were able to create a relationship based on reciprocity and trust, the
higher was the quality of the data we collected.
Several other factors also helped to build a relationship between interview-
ers and respondents and maximize the comfort of village officials with our
research on potentially politically sensitive topics. We were able to stress the
purely academic nature of the survey to village officials credibly. The survey
was in fact purely academic, and the data were not collected for policy-making
purposes. All of the interviewers and I looked like, and in fact were university
students. In the vast majority of cases, interviewers were not accompanied by
higher level officials when interviewing village officials. We also administered
the survey at the convenience of the village officials. We scheduled the survey
so that it did not coincide with peak times for agricultural work such as har-
vesting or administrative work such as tax collection or village elections. When
interviewers arrived in a village and village officials were busy, they waited
around until the village officials had sufficient time to sit down with them for
a lengthy conversation.
Perhaps most important, the high quality of the data was due in large part
to the skilled and diligent administration of the survey by the student inter-
viewers. All of the interviewers underwent two to five days of training in the
classroom and in practice administrations of the survey in the field. Most of the
students had grown up in villages themselves. A number of them had worked as
enumerators on previous rural surveys. Because of their personal backgrounds,
they were particularly adept at putting the village officials at ease by talking
about their own experiences growing up in a village and drawing on their
260 Lily L. Tsai

personal knowledge of rural life. Many of them applied to work on the survey
because they were writing theses on rural issues and could take advantage of
the time in the field to collect information for their own research projects.

Choosing a Mode of Interviewing: Conversational or Flexible

Interviewing versus Standardized Interviewing
One of the most important factors for data quality in this survey was the choice
to use conversational or flexible interviewing rather than standardized inter-
viewing. In standardized interviewing the ideal interviewer is a simple reader of
the questions as they are written in the survey instrument. Lavrakas (1993:132)
describes the standardized interviewer as an “intelligent automaton,” and as
Weisberg (2005:47) comments, “the emphasis often is more on the interviewer
as automaton than as intelligent.” The standardized approach to interviewing
characterizes the relationship between interviewer and respondent as a profes-
sional relationship in which the interviewer seeks to obtain high-quality data
from the respondent by providing incentives, appealing to the respondent’s
own values, and teaching respondents how to play their expected role in the
survey interview and what good answers should be like (Weisberg, 2005:48).
The standardized mode of interviewing has a number of advantages. If con-
ducted properly, it minimizes interviewer variance (Groves, 1987:164). Having
interviewers adhere strictly to a script reduces the demands on interviewer
skills and comprehension of the research project and the costs of training inter-
viewers (Weisberg, 2005:48). Standardized interviewing is also much faster
than conversational interviewing and decreases administration time (Biemer
and Lyberg, 2003:154).
There are few systematic studies providing data on the circumstances under
which standardized or conversational interviewing produces higher quality
data, and more such studies are sorely needed (Weisberg, 2005:62). Based on
my experience with my 2001 village survey, standardized interviewing pre-
sented a number of problems when conducting research in rural China on
potentially politically sensitive topics.
Some of these problems existed irrespective of the research topics and arose
because interviews and surveys were completely unfamiliar to most rural resi-
dents. Many villagers in China lack experience with multiple-choice questions,
standardized tests, interviews, or even informal conversations with strang-
ers from outside their locality. Even when dialect was not a problem, simple
misunderstandings were particularly common. Village officials would, for
example, confuse “preliminary” or “primary” village election candidates and
“final” or “formal” village election candidates, regardless of how clearly and
thoroughly we defined the concepts for them. During a question about pre-
liminary candidates, for example, it might become clear that the respondent
was thinking about final candidates because the number of candidates he had
in mind matched the final slate rather than the primary slate. In a standardized
interview the interviewer should simply record the respondent’s answer exactly
as given, even if he knows it does not represent the facts accurately (Groves,
Quantitative Research and Issues of Political Sensitivity 261

2004: 289). In conversational interviewing the interviewer can ask the respon-
dent whether he is definitely thinking about preliminary candidates or actually
thinking about final candidates and can clarify that the current question con-
cerns preliminary candidates.
One of the main arguments for using conversational interviewing rather
than standardized interviewing parallels the issue of translating cross-national
surveys into different languages: using the same words does not guarantee the
same meanings to different respondents (Iarossi, 2006: 85–86; Suchman and
Jordan, 1990: 233). Schober and Conrad (1997) demonstrate in a laboratory
experiment that while both standardized and conversational interviewing pro-
duce high levels of accuracy when respondents are certain about how concepts
in a question map onto their own circumstances, conversational interviewing
produces higher response accuracy when respondents are unsure about these
mappings and interviewers can provide additional assistance and explanation.
Unclear mappings are a particularly salient problem for China. Because of the
immense amount of regional variation, it is particularly difficult to anticipate
all the possible questions and definitional issues that might arise, regardless
of how thoroughly one pretests the survey instrument. In this context, con-
versational interviewing may offer significant advantages over standardized
interviewing. As Groves (1987) notes, “many of the normal mechanisms of
assuring clear communication, of correcting misimpressions, of addressing the
questions of the listener have been stripped away from the ‘standardized’ inter-
view.” Moreover, standardized interviewing may reduce interviewer-­related
error at the expense of increasing respondent-related bias: “The effects of
[standardized interviewing] may have been to minimize interviewer variance
but to increase bias, due to poor comprehension or minimal memory search for
relevant information” (Groves, 1987: S164).
Standardized interviewing in contexts where respondents lack experience
with surveys and strangers can also have a dramatic effect on response rates
and data accuracy. A stranger who appears on a villager’s doorstep and wants
him to provide answers to questions read mechanically from a prepared script
may elicit a number of reactions that are not conducive to the collection of
high-quality data. One reaction, as Suchman and Jordan observe, is simple
disinterest: “As respondents realize that their expectations for ordinary con-
versation are violated (and violated without recourse), they may react with
boredom (with consequent intellectual if not physical withdrawal) and impa-
tience (with answers designed to ‘get it over with’)” (Suchman and Jordan,
1990: 233). As a result, response rates go down, and missing data and “don’t
know” responses go up.
Another reaction that I experienced when I tried to conduct standardized
interviews in rural China was related to the political sensitivity of survey
research itself. Villagers and village officials often did a suspicious “double
take” if I refused to deviate from the prepared script. Even if the respondent’s
initial reaction to my request for an interview and explanation of the process
was good-natured willingness, as soon as I explained that I had to follow the
262 Lily L. Tsai

script in order to make sure that I had collected information in the same way as
all the other interviewers, the respondent would often ask, “Who did you say
you were again? What did you say this was for again?” or “Is this for a gov-
ernment office (政府部门)?” Even if I attempted at that point to reiterate reas-
surances that this research was purely academic and this practice was simply
to ensure that all the interviewers collected the same information, respondents
usually remained visibly disturbed or disengaged for the rest of the interview,
especially if I continued to refuse to deviate from the script.
Trying to get me to deviate from a standardized script was in fact a way
of equalizing the power dynamics in the interviewer-respondent relationship.
Deviating from the script was like agreeing to drink 白酒 (strong liquor) at
lunch – a concession that both symbolically and practically allowed the other
person to exercise power over my behavior, which in turn made him more
inclined to agree to my requests. Moreover, because formal interviews and the
collection of quantitative data themselves are politically sensitive matters in
China, the more formal and professional the process is, the more the experi-
ence smacks of political and governmental authority. Respondents assume that
the authority that the script has over the interviewer is because the study is
actually being commissioned by government authorities.
Conversational interviewing also had a number of other advantages over
standardized interviewing when it came to asking questions about politically
sensitive topics.4 Giving respondents the opportunity and conversational space
to explain and justify their behavior often made them feel better about giv-
ing truthful answers about politically incorrect behavior. In one Hebei vil-
lage, for example, village officials were willing to admit to using floating ballot
boxes instead of the officially mandated fixed polling stations. However, they
wanted to spend some time explaining to us that floating ballot boxes worked
much better in their village because many villagers worked on fishing boats
that went out to sea at different times of the day. Conversational interviewing
also allowed interviewers to cross-check responses and allowed respondents to
relate anecdotes that provided interviewers with information about the validity
of the data. More than once, respondents changed their minds later on in the
interview when they felt more comfortable and then indicated that an earlier
answer they had provided was false. In a standardized interview, this kind of
later admission of the truthful answer would be ignored.
Finally, the conversational mode of interviewing permits interviewers to ask
questions using terms that are not politically sensitive to respondents but may
be politically sensitive to the official authorities. Some Chinese researchers, for
example, are wary about putting questions on government corruption on the
written questionnaire. Their solution is to write a question about something
like “problems of public administration,” which will not raise the eyebrows of

4
Kish (1962) finds that interviewer effects are not necessarily greater for politically sensitive
questions.
Quantitative Research and Issues of Political Sensitivity 263

official authorities, and then have interviewers explicitly explain to the respon-
dents that the question actually asks about problems of corruption. Paluck
(2007) used a similar technique in Rwanda to collect survey data on opinions
about ethnicity. Since the Rwandan constitution bans speech about ethnicity,
Paluck had to replace the word “ethnicity” with “types of people” in her sur-
vey. She notes: “Researchers followed up these questions with an explanation
that implied the significance of this term. I am confident that these questions
were understood to implicate ethnicity, because Rwandans are accustomed to
using such ‘coded’ language to refer to ethnicity on a daily basis, and because
it was clear from our participants’ responses that they understood the ques-
tion, as many dismissed the coded language altogether and referred directly to
Hutus, Tutsis, and Twa” (Paluck, 2007: 54).

Reflections on Conducting Individual-Level

Attitudinal Surveys
Although these observations are drawn from my administration of a politi-
cally sensitive village-level informant survey, they may also be applicable to the
administration of public opinion surveys on politically sensitive topics. As with
village-level surveys, we need to attend to the problem of data quality before
pursuing generalizability.
Since individual-level public opinion surveys can be even more politically
sensitive to official authorities than village-level surveys, using a socially
embedded approach to survey administration is perhaps even more critical.
Creating relationships with Chinese researchers, official authorities, interview-
ers, and respondents that are based on mutual trust and obligation is doubly
important for creating an environment in which respondents feel less pressure
to give politically desirable responses.
In sum, the conclusion that I drew from my survey experience was that a
conversational mode of interviewing, which preserved the neutrality guide-
lines associated with but not inherently exclusive to standardized interviewing
(Dijkstra and van der Zouwen, 1987), created a mutually trusting relationship
between interviewer and respondent that maximized the quality of the quan-
titative data collected. It is not surprising that studies find that experienced
interviewers instructed to use standardized interviewing often use elements
of conversational interviewing anyway. Viterna and Maynard’s (2002) study
of twelve university survey centers that purported to use the standardized
approach found that only one consistently followed standardized interviewing
procedures. Houtkoop-Steenstra observes (2000: ch. 7) that interviewers have
a strong tendency to try to maintain rapport with respondents by breaking the
rules of standardized interviewing. In a survey on illiteracy, interviewers tried
to make respondents more comfortable by rephrasing questions frequently,
praising their achievements, and indicating that they sometimes shared some of
the respondents’ problems with reading.
264 Lily L. Tsai

Interviewers should be nonjudgmental of respondents, and probing should

be neutral. Depending on the circumstances, interviewers might act “bland
and naïve,” or matter-of-fact and knowledgeable about corruption and other
characteristic qualities of rural politics (Wood, 2006: 382). Conversational
interviewing enabled respondents to ask interviewers questions about what
survey questions meant and why we were interested in asking them. It enabled
respondents to obtain often-detailed information about us and our research
that helped to alleviate their concerns and suspicions and helped to build a
relationship governed by reciprocity of frankness. Dijkstra (1987: 312) also
finds that conversational interviewing in which the interviewer shows interest
and empathy helps to motivate the respondent to try harder to understand the
question, retrieve the information needed to answer the question, and to repeat
this process until an adequate response is provided.
Conversational interviewing gave respondents power and control not only
over the decision to participate in the survey but also during the survey inter-
view process itself. This sense of equality in the survey interviewing relation-
ship was critical to the willingness of the respondents to volunteer truthful
information on politically sensitive subjects.
Using conversational interviewing for attitudinal surveys may require even
more intensive training of interviewers, but I argue that this investment will
help us advance the study of public opinion in China more systematically.
Interviewers can be trained to answer factual questions about the survey and
definitional questions about terms and concepts in survey questions without
giving their opinions. They can be trained to answer, “I don’t know,” or “I’m
not sure” if respondents ask for their own opinions. Like standardized surveys
that include questions with statements like, “Some people do this, while others
do that,” interviewers can be trained to make conversation that refers matter-
of-factly to the existence of both politically desirable and undesirable behav-
iors and attitudes. It is true that conversational interviewing may sometimes
be less likely to introduce additional bias or variance when we are collecting
factual information than when we are collecting attitudinal data since there is a
definite “right” or “wrong” answer to factual questions. However, if the dan-
ger of political desirability bias is high, the decrease in political desirability bias
may be worth the risk of increased interviewer variance. Training interviewers
to use conversational interviewing for attitudinal surveys may also be more
costly than using standardized interviewing, but again, it may be important
to spend resources on improving data quality before we spend resources on
maximizing sample size.
Finally, starting with smaller scale attitudinal surveys can free up resources
for collecting systematic data on interviewers and interviewer-respondent
interactions. Instead of using standardized interviewing to allow us to ignore
or assume away interviewer variance, a better strategy is to collect data on the
interviews and interviewers and to study these effects explicitly.
Quantitative Research and Issues of Political Sensitivity 265

Additional Notes on Strategies for Dealing

with Politically Sensitive Issues
Several other nuts-and-bolts strategies were also very helpful when conducting
survey research in rural China on potentially politically sensitive subjects. One
important strategy was to send teams of two interviewers to conduct survey
interviews. One interviewer could try to draw away any higher level authori-
ties monitoring the interview by asking them for a tour of the environs, leaving
the other interviewer free to conduct the survey without interference. Another
strategy was to try to corroborate responses by collecting supplementary data
on things that could be seen. For example, in addition to asking about lineage
activities, I also asked about the existence of lineage hall buildings and, if they
existed, whether we could go and visit them. Similarly, survey interviewers also
collected various documents from the villages they surveyed, such as election
ballots and reports, villager tax receipts, and cadre responsibility contracts.

Conclusion
In the end, surveys are inherently compromises (Groves, 1987: S167). In mak-
ing decisions about survey design and administration, researchers make end-
less trade-offs – between data quality and generalizability, time and money,
interviewer effects and respondent effects, bias and variance, conducting pre-
tests and the main survey, and more. This chapter offers some thoughts on
how to make these trade-offs in nondemocratic and authoritarian contexts
where political sensitivity is a central issue. I argue that data quality is of both
paramount importance and concern in these contexts and that central to maxi-
mizing data quality is conducting survey research that is “socially embedded.”
Survey researchers in any context take on multiple roles and invest in different
interpersonal relationships: “The decision to conduct a survey is a decision to
become not just a scholar but also a project manager, a fundraiser, a survey
methodologist, and a motivator and supervisor of others” (Park, 2006: 128).
Prioritizing the investment of time, resources, and attention into constructing
these social relationships so that they are based on mutual obligation and trust
can be invaluable for overcoming the methodological and ethical challenges
associated with politically sensitive quantitative research.
Reflections on the Evolution of the
China Field in Political Science

Kenneth Lieberthal

The current volume highlights the range and vibrancy of current studies of
China by political scientists in the United States. This is a field that has become
relatively mature in terms of the number and types of institutions that produce
good China-related research, the array of generations of scholars engaged in
that research, the variety of sources available to understand developments in
China, and the methodological richness of the field overall. All of this repre-
sents a situation very different from and much better than that in the 1960s.
But the changes over the past four decades have also introduced problems that
require the ongoing attention of the field.

The Evolution of the Field

The world of the 1960s differed fundamentally from that of 2010 in terms of
how China is studied. China studies in the earlier period were just reviving
in the wake of the devastation wrought by the anti-Communist efforts most
memorably associated with Senator Joseph McCarthy, who, in February 1950,
asserted that he had a list of 205 Communists being protected in the State
Department. The senator – along with others asking “Who lost China?” – dec-
imated the ranks of China specialists in the State Department and questioned
the loyalty of scholars such as John K. Fairbank and Owen Lattimore, argu-
ing that they were at least Communist dupes and in some cases active secret
members of the Communist Party (Fairbank, 1982). The results were such
that Fairbank, generally regarded as the dean of the China field, addressed a
conference of China scholars in the early 1970s and advised the younger par-
ticipants to be sure to always keep a daily diary. He explained that this would
prove important when they are investigated by a congressional committee and
must explain what they were doing and thinking at any particular point in
their past.1

1
Author’s personal recollection from that meeting.

266
The Evolution of the China Field in Political Science 267

The senior faculty in the 1960s generally had lived in China before 1949.
Some were offspring of YMCA officials2 or missionaries,3 whereas oth-
ers became engaged in China via their service in World War II.4 Columbia
University’s A. Doak Barnett was not atypical. He had been raised in China
(his father directed the YMCA in Shanghai), attended Yale University in the
United States for a B.A. and later for an M.A. in International Relations, then
returned to Asia in basically reportorial positions in China in the late 1940s,
and in Hong Kong in the early 1950s. He then moved to the United States,
where by the 1960s he had become a key member of the Columbia University
faculty. This background gave him an intimate knowledge and “feel” for
China, but relatively modest formal training in political science.
Those who began their studies of China in the 1960s had better for-
mal training, virtually all studying for Ph.D.s in political science at major
American universities.5 Many came to the China field from having studied the
Soviet Union and were driven by abiding interests in communism, Marxism-
Leninism, and the dynamics of revolution. But the world of the China scholar
at that time in many ways differed vastly from that of today.
For these young scholars, China was an abstraction – Americans were not
permitted to travel to the PRC (then universally called “Communist China”).
Scholars learned about China completely via sources, not from firsthand expe-
rience. Those sources were quite limited.6
China research initially relied primarily on U.S. government translations,
along with analytical work, publications, and documents from Hong Kong
and Taiwan. The U.S. government provided voluminous translation series of
media broadcasts and articles in publications.7 But these were not indexed well.
For example, the most widely used source, the Foreign Broadcast Information
Service Daily Report, provided only single entries at the beginning of each

2
For example, A. Doak Barnett of Columbia University.
3
For example, Lucian Pye of MIT.
4
For example, Robert Scalapino of UC-Berkeley and Benjamin Schwartz of Harvard. John
Stewart Service was both a YMCA child and a U.S. government employee in China during
World War II. Harvard’s John K. Fairbank also served in the State Department in China dur-
ing World War II.
5
Steven Andors, Phyllis Andors, Richard Baum, Gordon Bennett, Thomas Bernstein, Parris
Chang, Edward Friedman, Steven Goldstein, Harry Harding, Ying-mao Kau, Steven Levine,
Andrew Nathan, Michel Oksenberg, Susan Shirk, Richard Solomon, Frederick Teiwes, James
Townsend, Lynn White, and the author, among others.
6
Oksenberg (1970) provides an excellent overview and analysis of the English-language sources
available to study China during this period.
7
Foreign Broadcast Information Service’s China Daily Report translated radio broadcasts and
newspaper articles, producing a daily “book” five times a week that often contained over eighty
single-spaced pages. Longer articles tended to be captured in the Survey of China Mainland
Press and Selections from China Mainland Magazines, also U.S. government translation series.
Items were selected for translation based on their potential value to U.S. government analyses.
Three other series also provided translations that many scholars used: the Joint Publications
Research Service (which included a far wider array of types of materials), the U.S. (Hong Kong)
Consulate General’s Current Background, and the BBC’s Summary of World Broadcasts.
268 Kenneth Lieberthal

daily “book” and quarterly single-entry compendia. Researchers often allo-

cated months in their research schedules to identifying articles that now can
be located literally in seconds via readily available search engines. These early
studies tended to focus on the analysis of documents, ideological framings,
and newspapers/media broadcasts.
China itself published some periodicals, such as China Pictorial, China
Reconstructs, Peking Review, and Hong Qi, but many of these stopped pub-
lication during the Cultural Revolution. The Cultural Revolution produced
a tide of Red Guard publications, which began (albeit in extremely polemi-
cal ways) to reveal the policy debates and elite conflicts that had taken place
in earlier years.8 The U.S. government acquired many of these publications
by purchase and, not surprisingly, Hong Kong-based counterfeiters quickly
sensed a gold mine and began to churn out fakes.
With the Cultural Revolution, as more refugees began to appear in Hong
Kong, refugee interviews became increasingly important as a source of infor-
mation. Refugees, by definition, are an unrepresentative lot, though. Out of
concerns about assuring personal safety, most scholars did not identify the
refugees they interviewed. This could present its own set of problems. Three
scholars who did important interviewing in Hong Kong one after the other,
and who developed relatively compatible views of how the Chinese system was
operating, only years later learned that they had been relying on the same key
refugee as a source.9
Communications were very poor and physical materials hard to obtain.
Copying technology other than microfilm and microfiche basically did not yet
exist, and electronic communications beyond telephone and telegraph were
still unavailable. Most young entrants to the China field went to Taiwan to
study language (and perhaps do some research in the few carefully guarded
rooms permitted to hold mainland “Communist bandit” materials), and then
on to Hong Kong, in many cases to the Universities Service Centre (USC) in
Kowloon, for their dissertation research. USC provided office space, a sense
of community, a network for finding refugees to interview about conditions
across the border, and good clippings files of mainland newspapers compiled
by the nearby Union Research Institute.10 Given the absence of copying facili-
ties, protecting the physical safety of one’s research notes from loss or inad-
vertent damage was a matter of serious concern.

8
 Many universities now have microfilm and microfiche collections of Red Guard papers and
other materials. These materials provided a major basis for such studies as Chang (1978).
The present author (1971) sought to evaluate the accuracy of some of these materials as they
pertained to past elite debates.
9
 A. Doak Barnett (1967), not one of the three scholars mentioned in this paragraph, wrote the
most detailed volume on the government system. Virtually the entire volume was based on
interviews with refugees who were ex-cadres. The interviews were conducted among refugees
who had left China before the Cultural Revolution.
10
The Union Research Institute also held extensive files of notes compiled from interviews of
refugees from the mainland.
The Evolution of the China Field in Political Science 269

Ideology and politics intruded deeply into scholarship. The Cultural

Revolution in China coincided with America’s escalation of the Vietnam
War and the extremely bitter, in 1968 bordering on revolutionary, politics
that ensued in the United States. These disputes deeply affected the China
field. A number of scholars in the Asia field formed their own progressive
association, called the Committee of Concerned Asian Scholars, which
held its own annual meeting and published a journal, the CCAS Bulletin,
and some books. Profound political and resulting personal disagreements
divided the scholarly community, with a great deal of pressure exerted by
some to take a stand against “American imperialism.” These political fis-
sures ran deep, and ideological differences ripped the field apart well into
the late 1970s.
Students almost without exception entered graduate school with no previ-
ous background in Chinese-language study. Acquiring the language, there-
fore, occupied a significant part of the graduate training program.
The study of China was concentrated at a few leading centers because of
a dearth of both scholars and materials. Harvard, Columbia, University of
California-Berkeley, and Stanford (especially because of the Hoover Institution
collection) played especially large roles in developing the field.
Each major university took a quintessentially area studies approach to
understanding China. Ph.D. students in political science who focused on
China often obtained an M.A. or certificate in China area studies along the
way. Their programs included courses in the history, sociology, and language
of modern China, in addition to dedicated courses on Chinese politics.11
Despite these limitations, a great deal of very careful work produced seri-
ous analyses of developments in the PRC. These tended to be richly contextual
studies of individual cases, locations, or policy developments,12 with insights
generated by careful consideration of the potential implications of the studies’
empirical findings.13 That reflected in part the way political science was taught
in the 1960s and in part the almost total lack of reliable statistical informa-
tion from China at the time.14 During the Cultural Revolution, of course,
even Chinese officials no longer had access to remotely reliable data.15 Earlier

11
Courses in economics became more important only after China moved well along its path of
reform.
12
For example, Barnett (1969), Baum and Teiwes (1968), Shirk (1982), and Vogel (1969).
13
The brief comments in this chapter do not seek to match the depth and richness of Oksenberg’s
(1970) essay.
14
The excellent series edited by Robert F. Dernberger for the Joint Economic Committee of the
U.S. Congress provided inadvertent testimony to how limited the concrete data were. Most
Chinese statistics consisted of statements concerning percentage increases over the previous
year in broad aggregates, where the base numbers for the series and concrete definitions of the
categories were never revealed.
15
At the height of the Cultural Revolution, the State Statistical Bureau had only fourteen people
left in its central office. On the rehabilitated statistical system, see the series of articles in FBIS
Daily Report: People’s Republic of China, February 17, 1984, pp. K17–K21.
270 Kenneth Lieberthal

periods, such as the Great Leap Forward, produced statistical black holes of
almost equally enormous scope.16
The field has subsequently evolved as a result of changes in virtually every
parameter noted here. First, access to China has been transformed. Very lim-
ited visits by scholars began to take place as early as 1971, and these increased
gradually during the 1970s. These afforded opportunities to meet with vari-
ous Chinese officials from local to central levels, but those officials generally
provided only carefully vetted information. Travel opportunities were so lim-
ited that in many cases pictures taken by recent visitors were of not only the
same cities but also of the same rooms in the same buildings as those taken by
visitors in earlier years. Visas were scarce, and the Chinese often paid all land
expenses and provided the guides and entertainment. The purpose was hardly
unfettered inquiry. But even these choreographed experiences began to lift the
veil on the realities behind the propaganda in China.
A personal anecdote illustrates this. I was in Shanghai in 1977 at the time
of the conclusion of the Eleventh Party Congress. Our minders had gathered
all foreigners into a large room at the Peace Hotel to watch on TV the cover-
age of Hua Guofeng’s Political Work Report to the congress.17 Many Chinese
hotel staffers were also with us. When Hua announced the formal conclu-
sion of the Great Proletarian Cultural Revolution, a spontaneous cheer went
up from the Chinese present. Hua then went on to say that there would be
another such movement every seven or eight years – which was met by dead
silence in the room.
Deng Xiaoping’s reemergence in a commanding position by the end of 1978
and normalization of diplomatic relations with the United States at the begin-
ning of 1979 opened up new vistas. Chinese, many of whom had spent roughly
twenty years in prison camps as Rightists, were now released and found them-
selves attending conferences at plush sites such as Airlie House in Virginia.18
By the early 1980s China had begun to admit American scholars to do
limited research and spend real time at Chinese institutions, and Chinese
scholars began to visit and study at American universities. These opportuni-
ties made scholars aware of the enormously difficult lives that their Chinese
counterparts led and the extent to which bureaucracy and political oppres-
sion weighed on virtually everything they did. One often heard Chinese col-
leagues explain patiently that “In China, little things are difficult and difficult
things are impossible,” as personal dependence on bureaucrats to accomplish
even the simplest things characterized every dimension of the system. In addi-
tion, during the 1980s, China began to open up to foreign businesses, and
an increasing range of people grappled with trying to get things done in the

16
See Becker (1998), which details how absurd the reported statistics became during the Great
Leap Forward.
17
Text carried by New China News Agency, August 22, 1977.
18
Many of these were people who had learned English before 1949 and were the most “present-
able” people China could produce for international conferences at the time.
The Evolution of the China Field in Political Science 271

Chinese context. Harry Harding captured the resulting change in perspectives

in an essay of that period (Harding, 1982).
The 1980s proved to be an extremely exciting period of reforms, and vari-
ous American scholars were sought out by reformers to provide advice and
insights. In political science, Americans advised on the development of the field
in China (political science had been disestablished as a discipline in the 1950s,
and in the 1980s individuals such as Yan Jiaqi, who had no previous training in
the discipline, were assigned to be political scientists and to develop the field).
Organizations such as the Committee on Scholarly Communications with the
PRC (of the National Academy of Sciences) and the Social Science Research
Council played significant roles in these efforts. America was then held in very
high repute in China, in part because it was seen as the quintessentially mod-
ern country and in part because it was viewed as an ally against the Soviet
Union. Reformers of all stripes often visited American scholars in search of
good counsel. American knowledge of Chinese politics and policy process
began to grow. In addition, the World Bank and other international orga-
nizations began to establish ties with China, and the World Bank especially
began to publish figures on the economy that previously were unavailable even
to most Chinese economists. At the same time, the World Bank and others
worked with China to improve the quality of economic reporting there.19
With some disruptions, most notably in the wake of June 4, 1989, access
to China has continued to grow. By 2010, many students entering Ph.D. pro-
grams in political science with a focus on China have already lived in the PRC
for a year or more and have developed a good personal feel for the country,
along with significant language skills. Most academics studying the country
have spent extensive time there in both academic institutions and various other
units. Chinese, both in China and in the United States, talk relatively freely
about their views and concerns and provide a wide variety of perspectives.
Second, changes in China and in sources have produced related changes in
the topics that are studied. The 1960s and 1970s saw many volumes devoted
primarily to analysis of elite politics and ideological battles.20 The 1980s
brought studies of the reforms and of bureaucratic organization,21 in addi-
tion to ongoing analyses of personal politics at the top of the Communist
Party. Toward the end of that decade, interviewing began to produce enough
of a basis to permit concrete explication of policy process.22 As access further
increased and the reforms produced major changes in the way the economy
functioned, attention increasingly focused on analysis of the country’s evolv-
ing political economy, along with a vast array of local studies based on inter-
views and participant observation.23 Most of these developments have been

19
Oksenberg and Jacobson (1990) provide an overview of this.
20
Two of many examples are MacFarquhar (1974, 1983, 1997) and Teiwes (1979).
21
See, for example, Harding (1981).
22
See, for example, Lieberthal and Oksenberg (1988) and Lieberthal and Lampton (1992).
23
See, for example, Blecher and Shue (1996), Oi (1998a), and Gallagher (2005).
272 Kenneth Lieberthal

additive, with perhaps only ideological studies largely disappearing from the
literature in the past decade. By 2010, moreover, studies of Chinese politics,
as illustrated by the contributions to this volume, have increasingly joined the
mainstream of political science literature in terms of methods and topics.
Third, sources of data have multiplied in every way. In the 1980s, former
top officials began to write memoirs that were sometimes very revealing. Over
the years, the volume and scope of memoir literature, both autobiographical
and through various types of party publications and reportage, have contin-
ued to mushroom.24 The Chinese media have diversified and multiplied, and
they have become enormously more informative. The statistical agencies have
become far more adept at collecting data (despite ongoing serious problems),
and far less of what they collect is considered secret. Publications abound for
all types of state units, including ministries, local governments, the Central
Party School, the Central Committee Party History Office, and others. Trade
associations and other groups publish specialized journals, as do foreign
NGOs, businesses, and news sources. And the various research units and
academic centers produce a veritable avalanche of published analytical work,
especially now that publications are considered a key metric of productivity.
As Allen Carlson and Hong Duan’s chapter in this volume explains in the
foreign policy realm, the Internet has introduced a phenomenal additional
array of sources, from personal blogs to Web sites for all types of publications
and bodies. A large percentage of government units, for example, now have
Web sites, from which it is possible to obtain data that in the early years of
study would have been difficult, if not impossible, to access.25
Search engines are making information in publications available in a way
that could not have been imagined in earlier years. The CNKI databases hosted
by EastView (中国知识资源总库 – – CNKI 系列数据库), for example, contain
full-text digital access to Chinese publications, including 7,200 journals start-
ing from 1915 (containing over 23 million articles), nearly 4,000 academic
journals dating back to 1887, and about 1,000 newspapers published since
2000. Other datasets focus on specialized areas such as laws and regulations.
The Internet has also enabled regular exchanges of information among large
groups of scholars of China organized through listservs.
Surveys are now feasible, and many are conducted. The authorities still
impose limits on what they deem to be sensitive inquiries, but these limits are
fundamentally looser than in earlier years.26
In-depth interviews provide far wider and deeper access to information
than in earlier times. Many more officials and knowledgeable outsiders are

24
See, for example, Jin (1989) and Zong (2008).
25
The Congressional-Executive Commission on China provides a useful list of links to govern-
ment Web sites in its PRC E-Government Directory, at http://www.cecc.gov/pages/prcEgov-
Dir/dirEgovPRC.php.
26
For details, see the contribution to this volume by Mingming Shen and Ming Yang, with
Melanie Manion.
The Evolution of the China Field in Political Science 273

prepared to talk with scholars, and in this author’s experience many are will-
ing to meet informally. Access is now available to leaders and staff in vastly
more units than previously, most are far more open in their discussions as
rules governing secrecy have narrowed in scope enormously, and social sci-
ence scholarship is now regarded with less a priori suspicion than was the case
in the early days of the reforms.
In short, China has gone from being a basically inaccessible, very low-
information society to being a relatively accessible, high-information society
since the 1960s. The major problems now are to gain control over the primary
and secondary sources. In the 1960s, a scholar could reasonably aspire to
read everything published in English on China – or at a minimum all serious
scholarly work – in addition to keeping up with the major Chinese-language
sources. Now it is no longer feasible to do either.
Fourth, technology has transformed the study of China. Scholars commu-
nicate with each other globally and instantaneously, and that includes many
scholars in China itself. Materials are now available, in many cases electroni-
cally, to far more institutions and scholars than was previously feasible. The
Web, scanning technologies, and other developments have changed the situa-
tion fundamentally. And computer programs now permit automated content
analysis and sophisticated data analysis that in earlier years were extremely
labor-intensive exercises.27
Even travel has changed dramatically, becoming far less expensive and more
rapid. That is true both between the United States and China and within China
itself. When this author first flew to Taiwan in 1969 from New York, for exam-
ple, it required two stops in the continental United States, a third in Hawaii,
and a fourth in Japan before landing in Taipei. When China began to open
up in the 1970s, internal flights were infrequent and equipment was primitive
(typically, old Aeroflot planes). Because there were no major highways, most
travel necessarily was by train. Transportation generally had to be booked
via the China Travel Service, which conducted operations only in person and
could take weeks to make even simple arrangements. Airplane tickets had to be
reconfirmed in person or they were canceled, and this often required waiting
in line for hours at the appropriate office. Getting into a city from an airport
could take hours if ground transportation had not been arranged ahead of
time. And major areas of every province were off limits to foreigners.
Fifth, changes in the discipline of political science have changed the schol-
arship on China. To put it in somewhat oversimplified terms, in the 1960s
“political science” was primarily an analysis of politics in order to generate
inductively insights of more general applicability – that is, it was basically the

27
For example, Yoshikoder, which can be downloaded for free from http://www.yoshikoder.
org/, can do frequency counts of terms, provide the context in which keywords appear, and do
simple evaluations of content (for example, ratio of positive-to-negative references to particu-
lar terms), among other functions. See also Daniela Stockmann’s contribution in the present
volume.
274 Kenneth Lieberthal

study of politics without science. By 2010, that situation has largely reversed
itself. Now the discipline privileges survey research, large-N studies, statistical
analyses, game theory, and formal modeling. Highly contextualized, granular
case studies do not easily lead to favorable tenure decisions in many of the
most highly ranked political science departments. And issues that inherently
are difficult to put into quantitative frameworks – such as cultural dimensions
of issue framing, policy making, and elite politics – receive less attention.
Finally, the content of graduate education for political scientists who want
to study China has changed significantly. The discipline now privileges meth-
odology, and courses in that subfield consume substantial graduate program
time. Combined with increasing pressure in many Ph.D. programs to shorten
the time from matriculation to degree, the opportunity costs of taking courses
in the history, sociology, economics, culture, and language of modern China
have risen to the point that relatively few students put these together as part of
their political science Ph.D. programs. Indeed, many graduate programs have
abolished foreign language requirements in favor of requirements on method-
ology. As a result, one or two courses on Chinese politics/foreign policy typi-
cally suffice, with much of the rest of the learning about China relegated to
dissertation proposal preparation and in-country dissertation research. Many
Ph.D. programs discourage students from pursuing an area-studies M.A. on
their way to obtaining a Ph.D.

Current Issues
Overall, the above-noted changes have moved forward the America-based
China field in political science enormously. Scholars generally have taken effec-
tive advantage of the facts that China itself is more open and accessible, the
available data are of higher quality and greater variety, methods of analysis
have become more rigorous and sophisticated, and the field itself has become
more “democratic” in that serious studies are no longer confined primarily
to a few leading universities and centers. Another change, that scholars who
grew up in China are now important members of the American political sci-
ence community studying China, has deepened the insights and broadened the
perspectives available in the U.S. academy. The chapters in the present volume
testify to the serious progress and types of results that have been achieved.
But all is not well. Some of the trends over the years have diminished
approaches that can provide rich insights and in the process threaten to reduce
the fruitful synergy between the study of China in particular and of politics
more generally. Four issues warrant particular attention.
First, the data standards demanded by the discipline often still cannot
be met in China. In some instances this reflects the unavailability of data
series of sufficient length or the simple lack of systematic data on various
issues. Scholars of the Americas or Europe who want to benefit from survey
research, for example, can often count on access to existing datasets, fully
documented, with which they can do their work. As Melanie Manion explains
The Evolution of the China Field in Political Science 275

in her contribution to this volume, the same is not true for such work on
China. This reflects in part the inherent difficulties of doing research in this
type of authoritarian system, where many types of data are considered sensi-
tive, the datasets produced cannot be accessed by others, and key information
is often missing concerning the sample and the Chinese partners involved in
the research effort. In part, this also reflects the rapid changes in China and
the lack of reliable time-series data. In addition, data quality frequently suffers
from many of the problems inherent in dealing with a country that is still in
transition from third world to first world institutions and capabilities.
Consequently, many graduate students who have completed courses in
methodology despair when they try to develop sufficiently “rigorous” research
projects on China. The overlap between available high-quality statistical data
and important, interesting questions to ask is still uncomfortably small in
developing countries. Where students of China must develop their data from
scratch, as is most frequently the case, they must spend enormous amounts of
time in questionnaire construction and pretests, gaining access to the relevant
populations, developing their sample frames, implementing their surveys, and
then analyzing and writing up the results. The same applies to many other
types of research that require in-country data collection. In this context, there
can be a lot of pressure to ask questions that are driven by data availability,
rather than asking different, challenging questions that can yield significant
results.
There is now tremendous focus on framing questions that can be pursued in
a methodologically rigorous fashion. But framing good questions is a necessary
first step in producing worthwhile outcomes. Thus, there needs to be serious
focus, too, on first understanding politics and deriving from that understand-
ing the key questions that need to be raised; then, within that universe, try-
ing to structure the questions so as to be most amenable to formal analytical
enhancements of the analysis. Otherwise, the rigor with which one can pursue
an issue tends to drive what issues are pursued. Since rigor itself is not directly
proportional to importance, its pursuit can weaken the field as a whole. As a
colleague of the author memorably commented during a heated discussion of
a tenure review case, “the most common form of ‘rigor’ is ‘mortis.’”
Second, ideas, culture, history, and social constructs can shape outcomes
in China profoundly. The ways issues are structured cognitively and how
they relate to other factors in the environment are influenced significantly by
culture and history. Even terminology affects intellectual constructs differ-
ently in different languages. As Lily Tsai’s chapter in this volume explains,
for example, there are advantages to conversational interviewing over stan-
dardized interviewing, as the former assures that survey questions are under-
stood correctly by respondents. But these dimensions in general are not readily
applicable to the types of rigorous inquiry and analysis increasingly demanded
by American political science departments. And graduate programs, as noted
above, train students less well to understand and analyze these types of factors
than was the case for their predecessors.
276 Kenneth Lieberthal

Third, students of Chinese politics who still utilize more traditional

approaches to understanding their topic often gravitate to think-tanks and
schools of public policy instead of leading political science departments.28
This is potentially a major loss to both the study of Chinese politics and to the
development of political science as a discipline. A more hospitable posture by
the discipline toward more traditional approaches to the study of China would
potentially make young scholars feel more comfortable in gradually adopting
more formal methods of analysis as the data from China warrant doing so.
In addition, the development of China studies in political science holds out
serious opportunities for the overall development of political science. Political
science developed from the study of Western historical experience, and many of
its most fundamental assumptions deeply reflect that background. But things
in China (and many non-Western areas) often do not fit into the conceptual
categories typically employed in the West. For example, Bruce Dickson’s work
in this volume and elsewhere (Dickson, 2003)29 has shown that entrepreneurs
in China do not, as was the case in modern Western history, seek to challenge
the regime. Rather, they tend to try to draw close to the state, viewing their
capacity to deal with the state as a competitive advantage in the Chinese econ-
omy. Others have found through surveys that political trust in the authoritar-
ian Chinese system is actually higher than that in democratic Taiwan (Shi,
2001). Thus, one of the major potential scholarly values of a more open and
accessible China is that it provides opportunities to test fundamental con-
clusions that have grown out of years of social science work based primar-
ily on Western developmental experience. Therefore, good studies of China
may contribute real insight into areas in which the conventional wisdom in
political science unknowingly reflects a more uniquely Western developmental
experience than universal laws concerning political systems.
In sum, as the China field matures, it has an enormous amount to offer to
the rest of the political science – and to broader social science – disciplines.
But those disciplines must be able to value the reality that different parts of the
world yield different types of data and pull things together in ways that may
differ substantially from those in the Western experience. Therefore, the value
of a maturing China field is in part that it can engage the broader discipline
in a serious analysis of fundamentals. This requires that the broader disci-
pline not impose too tight a boundary on defining the kinds of work that are
valued. Only in this context can the training programs and career incentives
nurture the full value of a mature scholarly community that is able to bring
China’s experience into the mainstream of political science.
Fourth, although things have changed enormously since John King Fairbank
issued the warning to younger China scholars in the early 1970s noted earlier,
there arguably is still an important need to have some students of Chinese

28
To name but a few: Erica Downs, Elizabeth Economy, David M. Lampton, James Mulvenon,
Jonathan Pollack, Anthony Saich, Michael Swaine, Murray Scot Tanner.
29
See also Kellee Tsai (2007).
The Evolution of the China Field in Political Science 277

politics who have a good grasp of overall developments in China and who
are able to articulate this to a broad public. Ironically, this is in part because
the American public is now deluged by presentations on China in the media
and by businesspeople, travelers, language teachers, and others. Too much
of this coverage of China succumbs to caricature and a focus on the colorful
and dramatic versus what is systematic. With the flood of coverage of things
Chinese, there is an acute need for informed judgments to create context and
perspective; these must be proffered in ways that reach and engage general
audiences.
The pressures, both from better accessibility and data and from the
demands of the discipline, however, move in the opposite direction – toward
developing a particular specialty that permits increasingly sophisticated anal-
ysis over time. This is valuable and certainly should be nurtured. However,
failure to develop some public intellectuals among each generation of students
of Chinese politics can diminish the quality of public discourse on China;
this, in turn, can reduce the resources available for ongoing development of
the field. This is also a problem for policy-making purposes. The more formal
the research methods used by political scientists are, the less likely it is that the
results of that work will inform in any serious way the deliberations of policy
makers. Public intellectuals who are able to translate such work into terms
readily accessible to the policy community, and to place their presentations in
outlets that command community attention, can play a vital role in making
academic work on China inform better public policy.
In sum, despite the reality that a volume of this scope and substance could
not possibly have been put together two decades ago, there are still troubling
questions that scholars of Chinese politics and those in other areas of political
science can and should address. These issues are, of course, not completely
unique to China, and in many ways they reflect the tremendous advances in
both the China field and the discipline of political science in the United States.
The maturity of the China field, and the enormous importance and visibility
of the country itself, now make the study of China a good vehicle for address-
ing issues that should engage the entire discipline.
Glossary

爱国主义 aiguo zhuyi patriotism

白酒 bui jiu strong liquor
搬迁 banqian relocation
北京社会经济发展年度调查 Beijing shehui jingji fazhan niandu diaocha
Beijing Annual Survey of Social and Economic Development
本地人 bendi ren locals
表现 biaoxian political behavior
比武 biwu “match of skill,” contest
布什 Bushi Bush
长假 changjia long vacation
超生游击队 chaosheng youjidui out-of-quota birthing guerrilla
朝鲜 Chaoxian North Korea
处 chu bureau
打工妹 dagongmei factory girl
党员 dangyuan party member
但是 danshi however
单位 danwei work unit
地 di below the departmental/prefecture level
调查 diaocha survey
地方志 difangzhi gazetteer
动态 dongtai trend
锻炼 duanlian training
妇联 Fulian Women’s Federation
复员军人 fuyuan junren veteran
改善政治体制 gaishan zhengzhi tizhi improve the political structure

279
280 Glossary

改革开放 gaige kaifang reform and opening

个体户 geti hu individually owned enterprise
关系 guanxi connections
国际问题研究 Guoji wenti yanjiu Journal of International Studies
国际资料信息 Guoji ziliao xinxi International Data Information
国情 guoqing national condition
和平崛起 heping jueqi peaceful rise
和平论坛 Heping luntan Peace Forum
红白理事会 hongbai lishihui state-approved wedding and funeral council
户口 hukou household registration system
检查 jiancha inspection
街道 jiedao neighborhood
街道办事处 jiedao banshichu street office
解放军报 Jiefangjun bao PLA Daily
精神 jingshen spirit
局 ju below the departmental/prefecture level
居民委员会 jumin weiyuanhui neighborhood committee
居民小组 jumin xiaozu residential small group
军属 junshu military dependent
开放档案 kaifang dang’an open archives
科 ke section
空挂户 konggua hu does not live at official household residence
空话 konghua empty verbiage
块 /政府 kuai/zhengfu horizontal coordinating governing bodies at various
levels
老户 laohu long-term petitioner
老年人协会 laonianren xiehui senior citizens association
邻长 linzhang block captain (Taiwan)
流动人口 liudong renkou floating population
利用处 liyong chu user services
里长 lizhang neighborhood head (Taiwan)
民族主义 minzu zhuyi nationalism
内部 neibu internal
片儿警 piarjing “beat cop”
强国 qiangguo strong nation
强国论坛 Qiangguo luntan Strong Country Forum
Glossary 281

企业 qiye enterprise
区 qu district
群众 qunzhong masses
群体 访 quntifang group petition
人户分离 renhu fenli does not live at official household residence
人民日报 Renmin ribao People’s Daily
人民网 Renmin wang People’s Net
认同 rentong identity
三乱 san luan “three disorders” (illegitimate fees)
社会调查 shehui diaocha social survey
社区居民委员会 shequ jumin weiyuanhui residents’ committee or community
residents’ committee (PRC)
涉外调查 she wai diaocha foreign-affiliated survey
涉外调查许可证 she wai diaocha xuke zheng permit for survey research
involving foreign affiliates
市场调查 shichang diaocha business survey
世界经济与政治 Shijie jingji yu zhengzhi World Economics and Politics
事业 shiye institute
事业单位 shiye danwei service unit
司 si departmental position
特护期 tehuqi specially protected period
天津信访 Tianjin xinfang Tianjin Petitioning
天益学术网 Tianyi xueshuwang TECN Academic Net
条/部门 tiao /bumen vertical bureaucracies of governance
厅 ting office
统计公报 tongji gongbao statistical report
突发 性 群体事件 tufaxing qunti shijian explosive mass incident
网民 wangmin netizen
委托 weituo jointly
维稳办 weiwenban Social Stability Maintenance Office
我感到担心 wo gandao danxin I feel worried
我感到害怕 wo gandao haipa I feel afraid
我感到骄傲 wo gandao jiao’ao I feel proud
毋忘国耻 wuwang guochi “Never Forget the National Humiliation”
下海 xiahai going “into the sea” of the private sector
现代国际关系 Xiandai guoji guanxi Contemporary International Relations
282 Glossary

小品 xiaopin comedy skit

信访 xinfang petitioning
信访信息 Xinfang xinxi Petitioning Information
新华网 Xinhua wang Xinhua Net
学科建设 xueke jianshe constructing the discipline
以前什么都靠司长, 后来谁都要看市场 yiqian shenmme doukao sizhang, hou-
lai shei douyao kan shichang “before, everything depended on the bureau
chief, afterward, everyone had to look to the market”
政法 委 zhengfawei Politics and Law Committee
政府部门 zhengfu bumen government office
正式体制改善 zhengshi tizhi gaishan improve the political structure
政治改革 zhengzhi gaige political reform
指南 zhinan guidebook
中国国关在线 Zhongguo guoguan zaixian IR China
中国国际关系研究网 Zhongguo guoji guanxi yanjiu wang Chinese IR Study
Network
中国军网 Zhongguo jun wang China Military Online
中国外交论坛 Zhongguo waijiao luntan Forum on China’s Foreign Relations
中国 网 Zhongguo wang China Net
中国知识资源总库系列数据库 Zhongguo zhizhi ziyuan zongku xilie shujuku
CNKI databases hosted by EastView
中国重要报纸全文数据库 Zhongguo zhongyao baozhi quanwen shujuku
China Core Newspapers Full-Text Database
中华人民共和国外交部 Zhonghua renmin gongheguo waijiao bu Ministry of
Foreign Affairs
资助 zizhu funded
总结报告 zongjie baogao work summary
座谈会 zuotanhui group discussion
References

Aberbach, Joel D., Robert D. Putnam, and Bert A. Rockman. 1981. Bureaucrats and
Politicians in Western Democracies. Cambridge, MA: Harvard University Press.
Adams, Laura. 2003. “Cultural Elites in Uzbekistan: Ideology Production and the
State,” in Pauline Jones Luong (ed.), The Transformation of Central Asia: States
and Societies from Soviet Rule to Independence. Ithaca, NY: Cornell University
Press: 93–119.
Adolph, Christopher. 2003. Paper Autonomy, Private Ambition: Theory and Evidence
Linking Central Bankers’ Careers and Economic Performance. Paper presented
at the annual APSA Conference, Philadelphia, PA.
Allina-Pisano, Jessica. 2004. “Sub Rosa Resistance and the Politics of Economic
Reform: Land Redistribution in Post-Soviet Ukraine.” World Politics, vol. 56,
no. 4: 554–581.
Almond, Gabriel A. and Sidney Verba. 1963. The Civic Culture. Princeton,
NJ: Princeton University Press.
Alonso, William and Paul Starr. 1983. The Politics of Numbers. New York: Russell
Sage Foundation.
Anderson, Elijah. 1990. Streetwise: Race, Class, and Change in an Urban Community.
Chicago: University of Chicago Press.
Anderson, Jeffrey J. 1992. The Territorial Imperative: Pluralism, Corporatism, and
Economic Crisis. Cambridge: Cambridge University Press.
Ash, Robert, David Shambaugh, and Seiichiro Takagi (eds.) 2007. China Watching:
Perspectives from Europe, Japan, and the United States. New York: Routledge.
Bachman, David M. 1991. Bureaucracy, Economy, and Leadership in China: The
Institutional Origins of the Great Leap Forward. Cambridge: Cambridge
University Press.
Banfield, Edward C. 1958. The Moral Basis of a Backward Society. Glencoe, IL: Free
Press.
Bar-Tal, Daniel. 1998. “The Rocky Road toward Peace: Beliefs on Conflict in Israeli
Textbooks.” Journal of Peace Research, vol. 35, no. 6: 723–742.
Barnard, Chester. 1968. The Functions of the Executive. 30th anniv. ed. Cambridge,
MA: Harvard University Press.
Barnett, A. Doak. 1985. The Making of Foreign Policy in China. Boulder, CO:
Westview.

283
284 References

(ed.). 1969. Chinese Communist Politics in Action. Seattle: University of Washington

Press.
Barnett, A. Doak, with a contribution by Ezra Vogel. 1967. Cadres, Bureaucracy and
Political Power in Communist China. New York: Columbia University Press.
Barrett, Christopher B. and Jeffrey W. Cason. 1997. Overseas Research: A Practical
Guide. Baltimore, MD: Johns Hopkins University Press.
Barron, Erma W. and Robert H. Finch, Jr. 1978. “Design Effects in a Complex
Multi­stage Sample: The Survey of Low Income Aged and Disabled (SLIAD).”
Proceedings of the Survey Research Methods Section, American Statistical
Association: 400–405.
Bartke, Wolfgang. 1997. Who Was Who in the People’s Republic of China: With More
than 3100 Portraits. 2 vols. München: K.G. Saur.
Bartke, Wolfgang and Institut für Asienkunde (Hamburg, Germany). 1991. Who’s
Who in the People’s Republic of China. 3rd ed. New York: K.G. Saur.
Bates, Robert, Avner Greif, Margaret Levi, Jean-Laurent Rosenthal, and Barry
Weingast. 1998. Analytic Narratives. Princeton, NJ: Princeton University Press.
Baum, Richard. 2007. “Studies of Chinese Politics in the United States,” in Robert
Ash, David Shambaugh, and Seiichiro Takagi, (eds.) 2007. China Watching:
Perspectives from Europe, Japan and the United States. New York: Routledge:
147–168.
1998. “The Fifteenth National Party Congress: Jiang Takes Command?” China
Quarterly, no. 153:141–156.
1994. Burying Mao: Chinese Politics in the Age of Deng Xiaoping. Princeton,
NJ: Princeton University Press.
Baum, Richard and Alexei Shevchenko. 1999. “The ‘State of the State,’” in Merle
Goldman and Roderick MacFarquhar (eds.), The Paradox of China’s Post-Mao
Reforms. Cambridge, MA: Harvard University Press: 333–362.
Baum, Richard and Frederick Teiwes. 1968. Ssu-Ch’ing: The Socialist Education
Movement of 1962–1966. Berkeley: Center for Chinese Studies, University of
California.
Bayard de Volo, Lorraine. 2001. Mothers of Heroes and Martyrs: Gender Identity
Politics in Nicaragua, 1979–1999. Baltimore, MD: Johns Hopkins University Press.
Bayard de Volo, Lorraine and Edward Schatz. 2004. “From the Inside Out: Ethnographic
Methods in Political Research.” PS: Political Science and Politics, vol. 37, no.
2: 417–422.
Becker, Jasper. 1998. Hungry Ghosts. New York: Macmillan.
Bellin, Eva. 2000. “Contingent Democrats: Industrialists, Labor, and Democratization
in Late-Developing Countries.” World Politics, vol. 52, no. 2: 175–205.
Belousov, Konstantin, et al. 2007. “Any Port in a Storm: Fieldwork Difficulties in
Dangerous and Crisis-Ridden Settings.” Qualitative Research, vol. 7, no.
2: 155–175.
Ben-Eliezer, Uri. 1995. “A Nation-in-Arms: State, Nation, and Militarism in
Israel’s First Years.” Comparative Studies in Society and History, vol. 37, no.
2: 264–285.
Bensel, Richard F. 2000. The Political Economy of American Industrialization,
1877–1900. Cambridge: Cambridge University Press.
1984. Sectionalism and American Political Development. Madison: University of
Wisconsin Press.
Benton, Gregor. 1992. Mountain Fires: The Red Army’s Three-Year War in South
China, 1934–1938. Berkeley: University of California Press.
References 285

Berns, Walter. 2001. Making Patriots. Chicago: University of Chicago Press.

Bernstein, Thomas P. and Xiaobo Lü. 2003. Taxation without Representation in
Contemporary Rural China. Cambridge: Cambridge University Press.
Bestor, Theodore C. 1989. Neighborhood Tokyo. Stanford, CA: Stanford University
Press.
Bewley, Truman F. 2004. “The Limits of Rationality,” in Ian Shapiro, Rogers Smith,
and Tarek Masoud (eds.), Problems and Methods in the Study of Politics.
Cambridge: Cambridge University Press: 381–5.
Bian, Yanjie. 1999. “Getting a Job through a Web of Guanxi in China,” in Barry
Wellman (ed.), Networks in the Global Village: Life in Contemporary
Communities. Boulder, CO: Westview: 255–278.
Biemer, Paul P. and Lars E. Lyberg. 2003. Introduction to Survey Quality. Malden,
MA: Wiley-Interscience.
Blecher, Marc and Vivienne Shue. 1996. The Tethered Deer: Government and
Economy in a Chinese County. Stanford, CA: Stanford University Press.
Bo, Zhiyue. 2004a. “The 16th Central Committee of the Chinese Communist
Party: Formal Institutions and Factional Groups.” Journal of Contemporary
China, vol. 39, no. 13: 223–256.
2004b. “The Institutionalization of Elite Management in China,” in Barry
Naughton and Dali L. Yang (eds.), Holding China Together: Diversity and
National Integration in the Post-Deng Era. Cambridge: Cambridge University
Press: 70–100.
2002. Chinese Provincial Leaders: Economic Performance and Political Mobility,
1949–1998. Armonk, NY: M. E. Sharpe.
Bobrow, Davis B. 2001. “Visions of (In)Security and American Strategic Style.”
International Studies Perspectives, vol. 2, no. 1: 1–12.
Bodnar, John (ed.). 1996. Bonds of Affection: Americans Define Their Patriotism.
Princeton, NJ: Princeton University Press.
Boylan, Delia M. 2001. Defusing Democracy: Central Bank Autonomy and the
Transition from Authoritarian Rule. Ann Arbor: University of Michigan Press.
Brace, Paul. 1993. State Government and Economic Performance. Baltimore,
MD: Johns Hopkins University Press.
Brady, Henry E. 2004. “Doing Good and Doing Better: How Far Does the Quantitative
Template Get Us?” in Henry E. Brady and David Collier (eds.), Rethinking
Social Inquiry: Diverse Tools, Shared Standards. Lanham, MD: Rowman and
Littlefield: 53–68.
Bruun, Ole. 1993. Business and Bureaucracy in a Chinese City: An Ethnography of
Private Business Households in Contemporary China. Berkeley: Institute of East
Asian Studies, University of California.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Silverson, and James D.
Morrow. 2003. The Logic of Political Survival. Cambridge, MA: MIT Press.
Burawoy, Michael. 2000. Global Ethnography: Forces, Connections, and Imaginations
in a Postmodern World. Berkeley: University of California Press.
1998. “The Extended Case Method.” Sociological Theory, vol. 16, no. 1: 4–33.
1985. The Politics of Production: Factory Regimes under Capitalism and Socialism.
London: Verso.
Burden, Barry C. and Joseph Neal Rice Sanberg. 2003. “Budget Rhetoric in Presidential
Campaigns from 1952–2000.” Political Behavior, vol. 25, no. 2: 97–118.
Cai Wenzhong. 1995. “Dubi jiangjun Cai Shufan” [One-Armed General: Cai Shufan].
Dangshi tiandi [Party History World], no. 3: 28–31.
286 References

Cai, Yongshun. 2000. “Between State and Peasant,” China Quarterly, no.
163: 783–805.
Calhoun, Craig. 1994. Neither Gods nor Emperors: Students and the Struggle for
Democracy in China. Berkeley: University of California Press.
Cambridge History of China, vols. 14, 15. 1987. 1991. Ed. John King Fairbank and
Roderick MacFarquhar. Cambridge: Cambridge University Press.
Cammett, Melani. 2007. Globalization and Business Politics in Arab North Africa.
Cambridge: Cambridge University Press.
2005. “Fat Cats and Self-made Men: Globalization and the Paradoxes of Collective
Action.” Comparative Politics, vol. 37, no. 4: 379–400.
Carlson, Allen. 2009. “A Flawed Perspective: The Limitations Inherent within the
Study of Chinese Nationalism.” Nations and Nationalism, vol. 15, no. 1: 20–35.
CASS. 2009. “Chinese Academy of Social Sciences,” at http://bic.cass.cn/English/
InfoShow/Arcitle_Show_Cass.asp?, accessed June 2009.
2007. “Surveying Internet Usage and Its Impact in Seven Chinese Cities.” Research
Center of Social Development, at http://www.wipchina.org/?p1=content&p2=
07122508155, accessed June 2009.
Central Organization Department and Party History Research Center of the CCP
CC. 2004. Zhongguo gongchandang lijie zhongyang weiyuan da cidian, 1921–
2003 [The Dictionary of Past and Present CCP Central Committee Members].
Beijing: Dangshi chubanshe.
Chan, Alfred L. and Paul Nesbitt-Larking. 1995. “Critical Citizenship and Civil
Society in Contemporary China.” Canadian Journal of Political Science, vol. 28,
no. 2: 293–309.
Chan, Anita, Richard Madsen, and Jonathan Unger. 2009. Chen Village: Revolution
to Globalization. 3rd ed. Berkeley: University of California Press.
1992. Chen Village under Mao and Deng. Berkeley: University of California
Press.
Chang, Parris S. 1978. Power and Policy in China. 2nd enlg. ed. University
Park: Pennsylvania State University Press.
Chang, Yu-Tzung, Yun-han Chu, and Frank Tsai. 2005. “Confucianism and Democratic
Values in Three Chinese Societies.” Issues and Studies, vol. 41, no. 4: 1–33.
Charmaz, Kathy. 2006. Constructing Grounded Theory: A Practical Guide through
Qualitative Analysis. London: Sage.
Chase, Michael and James Mulvenon. 2002. “You’ve Got Dissent.” Rand, at http://
www.rand.org/pubs/monograph_reports/MR1543/index.html, accessed June
2008.
Chen, An. 2002. “Capitalist Development, Entrepreneurial Class, and Democratization
in China.” Political Science Quarterly, vol. 117, no. 3: 401–422.
Chen, Calvin. 2008. Some Assembly Required: Work, Community, and Politics in
China’s Rural Enterprises. Cambridge, MA: Asia Center, Harvard University.
2006. “Work, Conformity, and Defiance: Strategies of Resistance and Control
in China’s Township and Village Enterprises,” in Jacob Eyferth (ed.), How
China Works: Perspectives on the Twentieth Century Workplace. New
York: Routledge: 124–139.
Chen, Jie. 2005a. “Popular Support for Village Self-government in China.” Asian
Survey, vol. 45, no. 6: 865–885.
2005b. “Sociopolitical Attitudes of the Masses and Leaders in the Chinese
Village: Attitude Congruence and Constraint.” Journal of Contemporary China,
vol. 14, no. 4: 445–464.
References 287

2004. Popular Political Support in Urban China. Washington, DC and Stanford,

CA: Woodrow Wilson Center Press and Stanford University Press.
2001. “Urban Chinese Perceptions of Threats from the United States and Japan.”
Public Opinion Quarterly, vol. 65, no. 2: 254–266.
2000. “Subjective Motivations for Mass Political Participation in Urban China.”
Social Science Quarterly, vol. 81, no. 2: 645–662.
1999. “Comparing Mass and Elite Subjective Orientations in Urban China.” Public
Opinion Quarterly, vol. 63, no. 2: 193–219.
Chen, Jie and Bruce J. Dickson. 2010. Allies of the State: Democratic Support and
Regime Support among China’s Capitalists. Cambridge, MA: Harvard University
Press.
Chen, Jie, Chunlong Lu, and Yiyin Yang. 2007. “Popular Support for Grassroots Self-
Government in Urban China.” Modern China, vol. 33, no. 4: 505–528.
Chen, Jie and Yang Zhong. 2002. “Why Do People Vote in Semicompetitive Elections
in China?” Journal of Politics, vol. 64, no. 1: 178–197.
2000. “Research Note. Valuation of Individual Liberty vs. Social Order among
Democratic Supporters: A Cross-validation.” Political Research Quarterly, vol.
53, no. 2: 427–439.
1999. “Mass Political Interest (or Apathy) in Urban China.” Communist and Post-
Communist Studies, vol. 32, no. 3: 281–303.
Chen, Jie, Yang Zhong, and Jan William Hillard. 1997. “The Level and Sources of
Political Support for China’s Current Regime.” Communist and Post-Communist
Studies, vol. 30, no. 1: 45–64.
Chen, Jie, Yang Zhong, Jan Hillard, and John Scheb. 1997. “Assessing Political Support
in China: Citizens’ Evaluations of Governmental Effectiveness and Legitimacy.”
Journal of Contemporary China, vol. 6, no. 16: 551–566.
Chen, Nancy N. 2003. Breathing Spaces: Qigong, Psychiatry, and Healing in China.
New York: Columbia University Press.
Chen, Shenghuo. 2004. The Events of September 11 and Chinese College Students’
Images of the United States. Paper presented at the Center for Strategic and
International Studies Conference on “Chinese Images of the United States.”
Chen, Xi. 2009. “The Power of ‘Troublemaking’: Chinese Petitioners’ Tactics and
Their Efficacy.” Comparative Politics, vol. 41, no. 4: 451–71.
2008. “Collective Petitioning and Institutional Conversion,” in Kevin O’Brien (ed.),
Popular Protest in China. Cambridge, MA: Harvard University Press: 54–70.
Chen, Xueyi and Tianjian Shi. 2001. “Media Effects on Political Confidence and Trust
in the People’s Republic of China in the Post-Tiananmen Period.” East Asia: An
International Quarterly, vol. 19, no. 3: 84–118.
Cheng, Peter T.Y., Jae Ho Chung, and Zhimin Lin (eds.). 1998. Provincial Strategies
of Economic Reform in Post-Mao China. Armonk, NY: M. E. Sharpe.
Christensen, Thomas J., Alastair Iain Johnston, and Robert S. Ross. 2006.
“Conclusions and Future Directions,” in Alastair Iain Johnston and Robert S.
Ross (eds.), New Directions in the Study of China’s Foreign Policy. Stanford,
CA: Stanford University Press: 379–420.
Chu, Yun-han and Yu-tzung Chang. 2001. “Culture Shift and Regime Legitimacy:
Comparing Mainland China, Taiwan, and Hong Kong,” in Shiping Hua (ed.),
Chinese Political Culture, 1989–2000. Armonk, NY: M. E. Sharpe: 320–347.
CNNIC. 2008. “The 21st Statistical Survey Report on the Internet Development in
China,” at http://www.cnnic.net.cn/index/0E/00/11/index.htm, accessed July
2008.
288 References

CNNIC. 2007. “Surveying China’s Blogging Market in 2007,” at http://www.cnnic.

net.cn/uploadfiles/pdf/2007/12/26/113902.pdf, accessed August 2008.
Congressional-Executive Commission on China. 2006. China’s Household
Registration System: Sustained Reform Needed to Protect China’s Rural
Migrants. Washington, DC.
Cornelius, Wayne A., Todd A. Eisenstadt, and Jane Hinley (eds.). 1999. Subnational
Politics and Democratization in Mexico. La Jolla: Center for U.S.-Mexican
Studies, University of California.
CPCR. 2005. Zhongguo shichanghua baokan quanguo toufang cankao 2005 (I)
[Market Intelligence Report on China’s Print Media Retail Distribution 2005
(I)]. Beijing: CPCR Kaiyuan Celue [Opening Strategy Consultation].
Cyert, Richard and James March. 1992. A Behavioral Theory of the Firm. 2nd ed.
Cambridge: Blackwell.
Damm, Jen and Simona Thomas. 2006. Chinese Cyberspaces: Technological and
Political Effects. New York: Routledge.
Davis, Deborah S. (ed.). 2000. The Consumer Revolution in Urban China.
Berkeley: University of California Press.
Davis, Deborah, Pierre Landry, Yusheng Peng, and Jin Xiao. 2007. “Gendered
Pathways to Rural Schooling: The Interplay of Wealth and Local Institutions.”
China Quarterly, no. 189: 60–82.
Dever, Jill A., Jun Liu, Vincent G. Iannacchione, and Douglas E. Kendrick. 2001. “An
Optimal Allocation Method for Two-Stage Sampling Designs with Stratification
at the Second Stage.” Proceedings of the Annual Meeting of the American
Statistical Association, August 5–9.
Diamant, Neil J. 2009. Embattled Glory: Veterans, Military Families, and the Politics
of Patriotism in China, 1949–2007. Lanham, MD: Rowman and Littlefield.
2000. Revolutionizing the Family: Politics, Love, and Divorce in Urban and Rural
China, 1949–1968. Berkeley: University of California Press.
Diao Jiecheng. 1996. Renmin xinfang shilue [A Brief History of People’s Letters and
Visits]. Beijing: Xueyuan chubanshe.
Dickson, Bruce J. 2008. Wealth into Power: The Communist Party’s Embrace of
China’s Private Sector. Cambridge: Cambridge University Press.
2007. “Integrating Wealth and Power in China: The Communist Party’s Embrace of
the Private Sector.” China Quarterly, no. 192: 827–854.
2003. Red Capitalists in China: The Party, Private Entrepreneurs, and Prospects
for Political Change. Cambridge: Cambridge University Press.
2002. “Do Good Businessmen Make Good Citizens? An Emerging Collective
Identity among China’s Private Entrepreneurs,” in Merle Goldman and Elizabeth
J. Perry (eds.), Changing Meanings of Citizenship in Modern China. Cambridge,
MA: Harvard University Press: 255–287.
Dijkstra, Wil. 1987. “Interviewing Style and Respondent Behavior: An Experimental Study
of the Survey-Interview.” Sociological Methods and Research, vol. 16, no. 2: 309–334.
Dijkstra, Wil and Johanese van der Zouwen. 1987. “Styles of Interviewing and the
Social Context of the Survey Interview,” in Hans–J. Hippler, Norbert Schwartz,
and Seymour Sudman (eds.), Social Information Processing and Survey
Methodology. New York: Springer.
DiMaggio, Paul, (Eszter Hargittai, W. Russell Neuman, and John P. Robinson.
2001. “Social Implications of the Internet.” Annual Review of Sociology, vol.
27: 307–336.
References 289

Dittmer, Lowell. 2001. “The Changing Nature of Elite Power Politics.” China Journal,
no. 45: 53–68.
1983. “The 12th Congress of the Communist Party of China.” China Quarterly,
no. 93: 108–214.
Dittmer, Lowell and William Hurst. 2002/2003. “Analysis in Limbo: Contemporary
Chinese Politics amid the Maturation of Reform.” Issues and Studies, vol. 38,
no. 4/vol. 39, no. 1: 11–48.
Dowd, Daniel V., Allen Carlson, and Mingming Shen. 1999. “The Prospects for
Democratization: Evidence from the 1995 Beijing Area Study.” Journal of
Contemporary China, vol. 8, no. 22: 365–380.
DRC [Guowuyuan fazhan yanjiu zhongxin]. 2001. “‘Dongbei xianxiang’ de xin
chulu” [A New Way Out of the “Northeast Phenomenon”]. Guowuyuan fazhan
yanjiu zhongxin diaocha yanjiu baogao, no. 60.
1999a. “Shenyang shuibengchang qiye gaige yu xiagang zhigong wenti diaocha”
[Investigation into the Shenyang Water Pump Factory’s Problems with Reform
and Xiagang Workers]. Guowuyuan fazhan yanjiu zhongxin diaocha yanjiu
baogao, no. 84.
1999b. “Shanghai dianhua shebei zhizaochang qiye gaige yu xiagang zhigong wenti
diaocha” [Investigation into the Shanghai Telephone Equipment Production
Plant’s Problems of Enterprise Reform and Xiagang Workers]. Guowuyuan
fazhan yanjiu zhongxin diaocha yanjiu baogao, no. 86.
1999c. “Baoshan gangtie (jituan) gongsi de qiye fazhan yu renyuan zhuangkuang
diaocha” [Investigation into the Enterprise Development and Employee Situation
of Baoshan Steel]. Guowuyuan fazhan yanjiu zhongxin diaocha yanjiu baogao,
no. 87.
Druckman, Daniel. 1995. “Social-Psychological Aspects of Nationalism,” in
John Comeroff and Paul Stern (eds.), Perspectives on Nationalism and War.
Amsterdam: Gordon and Breach: 47–98.
Druckman, James N., Donald P. Green, James H. Kuklinski, and Arthur Lupia. 2006.
“The Growth and Development of Experimental Research in Political Science.”
American Political Science Review, vol. 100, no. 4: 627–635.
Duriau, Vincent J., Rhonda K. Reger, and Michael D. Pfarrer. 2007. “A Content
Analysis of the Content Analysis Literature in Organization Studies: Research
Themes, Data Sources, and Methodological Refinements.” Organizational
Research Methods, vol. 10, no. 1: 5–34.
Earl, Jennifer, Andrew Martin, John D. McCarthy, and Sarah Soule. 2004. “The
Use of Newspaper Data in the Study of Collective Action.” Annual Review of
Sociology, vol. 30: 65–80.
Eaton, Kent. 2004. Politics beyond the Capital: The Design of Subnational Institutions
in Latin America. Stanford, CA: Stanford University Press.
Eckstein, Harry. 1975. “Case Study and Theory in Political Science,” in Fred Greenstein
and Nelson Polsby (eds.), Handbook of Political Science, Vol. 7: Strategies of
Inquiry. Reading, MA: Addison-Wesley: 79–138.
Edwards, Richard. 1977. Contested Terrain: The Transformation of the Workplace in
the Twentieth Century. New York: Basic Books.
Eldersveld, Samuel J. and Mingming Shen. 2001. Support for Political and
Economic Change in the Chinese Countryside: An Empirical Study of Cadres
and Villagers in Four Counties, 1990 and 1996. Lanham, MD: Lexington
Books.
290 References

Emerson, Robert, Rachel Fretz, and Linda Shaw. 1995. Writing Ethnographic
Fieldnotes. Chicago: University of Chicago Press.
Esarey, A. 2009. Who Controls the Message? Propaganda, Pluralism, and Press
Freedom in China. Paper presented at the Media in Contemporary Chinese
Politics workshop, Harvard University, April.
Fairbank, John King. 1982. China Bound: A Fifty-Year Memoir. New York: Harper
and Row.
Farrer, James. 2002. Opening Up: Youth Sex Culture and Market Reform in Shanghai.
Chicago: University of Chicago Press.
Fenno, Richard F., Jr. 1990. Watching Politicians: Essays on Participant Observation.
Berkeley: Institute of Governmental Studies, University of California.
1978. Home Style: House Members in their Districts. Boston and Toronto: Little,
Brown.
Fewsmith, Joseph. 1994. Dilemmas of Reform in China: Political Conflict and
Economic Debate. Armonk, NY: M. E. Sharpe.
Fitzgerald, John. 1996. Awakening China: Politics, Culture, and Class in the
Nationalist Revolution. Stanford, CA: Stanford University Press.
(ed.). 2002. Rethinking China’s Provinces. New York: Routledge.
Fletcher, George. 1993. Loyalty: An Essay on the Morality of Relationships. New
York: Oxford University Press.
Fong, Vanessa. 2004. Only Hope: Coming of Age under China’s One Child Policy.
Stanford, CA: Stanford University Press.
Forster, Keith. 1990. Rebellion and Factionalism in a Chinese Province. Armonk,
NY: M. E. Sharpe.
Foster, Kenneth C. 2002. “Embedded within State Agencies: Business Associations in
Yantai.” China Journal, no. 47: 41–65.
Fowler, Floyd. 1993. Survey Research Methods. Thousand Oaks, CA: Sage.
Fravel, M. Taylor. 2007. “Securing Borders: China’s Doctrine and Force Structure for
Frontier Defense.” Journal of Strategic Studies, vol. 30, no. 4/5: 705–737.
2000. “Online and On China: Research Sources in an Information Age.” China
Quarterly, no. 163: 821–842.
Frazier, Mark. 2003. The Making of the Chinese Industrial Workplace: State,
Revolution, and Labor Management. Cambridge: Cambridge University Press.
Friedman, Edward. 1975. “Some Political Constraints on a Political Science: Quantitative
Content Analysis and the Indo-Chinese Border Crisis of 1962.” China Quarterly,
no. 63: 528–538.
Friedman, Edward, Paul G. Pickowicz, and Mark Selden. 2005. Revolution, Resistance,
and Reform in Village China. New Haven, CT: Yale University Press.
Friedman, Sara. 2006. Intimate Politics: Marriage, the Market, and State Power in
Southeastern China. Cambridge, MA: Asia Center, Harvard University.
Gallagher, Mary. 2005. “‘Use the Law as Your Weapon!’ Institutional Change and
Legal Mobilization in China,” in Neil Diamant, Stanley Lubman, and Kevin
O’Brien (eds.), Engaging the Law in China. Stanford, CA: Stanford University
Press: 54–83.
Galvan, Dennis. 2004. The State Must Be Our Master of Fire: How Peasants Craft
Culturally Sustainable Development in Senegal. Berkeley: University of California
Press.
Geertz, Clifford. 1983. Local Knowledge: Further Essays in Interpretative
Anthropology. New York: Basic Books.
References 291

1973. The Interpretation of Cultures: Selected Essays. New York: Basic Books.
George, Alexander L. 1979. “Case Studies and Theory Development: The Method of
Structured, Focused Comparison,” in Paul Gorden Lauren (ed.), Diplomacy: New
Approaches in History, Theory, and Policy. New York: Free Press: 43–68.
George, Alexander L. and Andrew Bennett. 2004. Case Studies and Theory
Development in the Social Sciences. Cambridge, MA: MIT Press.
Gerring, John. 2004. “What Is a Case Study and What Is it Good For?” American
Political Science Review, vol. 98, no. 2: 341–354.
2001. Social Science Methodology: A Critical Framework. Cambridge: Cambridge
University Press.
Gladney, Dru C. 1996. Muslim Chinese: Ethnic Nationalism in the People’s Republic.
Cambridge, MA: Council on East Asian Studies, Harvard University.
Goffman, Erving. 1959. The Presentation of Self in Everyday Life. Garden City,
NY: Doubleday.
Gold, Thomas. 1998. “Bases for Civil Society in Reform China,” in Kjeld Erik
Brødsgaard and David Strand (eds.), Reconstructing Twentieth-Century
China: State Control, Civil Society, and National Identity. Oxford: Oxford
University Press: 163–188.
1989. “Guerrilla Interviewing among the Getihu,” in Perry Link, Richard Madsen,
Paul Pickowicz (eds.), Unofficial China: Popular Culture and Thought in the
People’s Republic. Boulder, CO: Westview: 175–192.
Gold, Thomas, Doug Guthrie, and David L. Wank. 2002. Social Connections
in China: Institutions, Culture, and the Changing Nature of Guanxi.
Cambridge: Cambridge University Press.
Goldstein, Alice, Sidney Goldstein, and Shenyang Guo. 1991. “Temporary Migrants
in Shanghai Households, 1984.” Demography, vol. 28, no. 2: 275–291.
Goldstein, Sidney. 1987. ‘‘Forms of Mobility and Their Policy Implications: Thailand
and China Compared.’’ Social Forces, vol. 65, no. 4: 915–942.
Goodkind, Daniel and Loraine A. West. 2002. “China’s Floating Population: Definitions,
Data and Recent Findings.” Urban Studies, vol. 39, no. 12: 2237–2251.
Goodman, Byrna. 2004. “Networks of News: Power, Language and the Transnational
Dimensions of the Chinese Press, 1850–1949.” China Review, vol. 4, no. 1: 1–10.
Granovetter, Mark. 1985. “Economic Action and Social Structure: The Problem of
Embeddedness.” American Journal of Sociology, vol. 91, no. 3: 481–510.
1973. “The Strength of Weak Ties.” American Journal of Sociology, vol. 78, no.
6: 1360–1380.
Gray, Clive. 1994. Government beyond the Centre: Subnational Politics in Britain.
London: MacMillan.
Gries, Peter Hays. 2005a. “Chinese Nationalism: Challenging the State.” Current
History, vol. 104, no. 683: 251–256.
2005b. “Nationalism, Indignation and China’s Japan Policy.” SAIS Review, vol. 25,
no. 2: 105–114.
2004. China’s New Nationalism: Pride, Politics, and Diplomacy. Berkeley: University
of California Press
Gries, Peter Hays, Kaiping Peng, and H. Michael Crowson. Under review.
“Determinants of Security and Insecurity in International Relations: Symbolic
and Material Gains and Losses: A Cross-national Experimental Analysis,” in
Vaughn Shannon (ed.), Ideational Allies: Constructivism, Political Psychology,
and IR Theory. Ann Arbor: University of Michigan Press.
292 References

Gries, Peter Hays, Quigmin Zhang, H. M. Michael Cronson, and Huajian Cai. Under
review. “Patriotism, Nationalism, and Perceptions of U.S. Threat: Structures and
Consequences of Chinese National Identity.”
Grodzins, Morton. 1956. The Loyal and the Disloyal: Social Boundaries of Patriotism
and Treason. Chicago: University of Chicago Press.
Groves, Robert M. 2004. Survey Errors and Survey Costs. Malden, MA: Wiley-
Interscience.
1987. “Research on Survey Data Quality.” Public Opinion Quarterly, vol. 51, part
2 (Supplement): S156-S172.
Gu, Edward X. 1996. “The Economics Weekly, the Public Space and the Voices of
Chinese Independent Intellectuals.” China Quarterly, no. 147: 860–888.
Guo, Gang. 2007. “Organizational Involvement and Political Participation in China.”
Comparative Political Studies, vol. 40, no. 4: 457–482.
Guo Liang. 2002. “‘Qiangguo luntan’: 9.11 kongbu xiji hou de 24 xiaoshi” [“Strong
Country Talk”: The First Twenty-four Hours after the Terrorist Attacks]. Xinwen
yu chuanbo yanjiu [Research in Journalism and Communications], no. 4: 2–13.
Guthrie, Douglas. 1999. Dragon in a Three-piece Suit: The Emergence of Capitalism
in China. Princeton, NJ: Princeton University Press.
Harding, Harry. 1994. “The Contemporary Study of Chinese Politics: An Introduction.”
China Quarterly, no. 139: 699–703.
1982. “From China, with Disdain: New Trends in the Study of China.” Asian
Survey, vol. 22, no. 10: 934–958.
1981. Organizing China: The Problem of Bureaucracy, 1949–1976. Stanford,
CA: Stanford University Press.
Hartford, Kathleen. 2000. “Cyberspace with Chinese Characteristics.” Current
History, vol. 99, no. 638: 255–262.
Hassid, Jonathan. 2007. Poor and Content Is Rich: A Computer-Assisted Content
Analysis of Twenty-Six Chinese Newspapers, 2004–2006, presented at the 9th
annual conference of the Hong Kong Sociological Association, Hong Kong.
He, Baogang. 1997. The Democratic Implications of Civil Society in China. New
York: St. Martin’s Press.
He, Zhou and Jian-hua Zhu. 2002. “The Ecology of Online Newspapers: The Case of
China.” Media, Culture and Society, vol. 24, no. 1: 121–137.
Heckman, James J., Hidehiko Ichimura, Jeffrey Smith, and Petra Todd. 1998.
“Characterizing Selection Bias Using Experimental Data.” Econometrica, vol.
66, no. 5: 1017–1098.
Heimer, Maria. 2006. “Field Sites, Research Design and Type of Findings,” in Maria
Heimer and Stig Thøgersen (eds.), Doing Fieldwork in China. Honolulu: University
of Hawaii Press: 58–77.
Heimer, Maria and Stig Thøgersen (eds.). 2006. Doing Fieldwork in China.
Honolulu: University of Hawaii Press.
Heller, Patrick. 1999. The Labor of Development: Workers and the Transformation
of Capitalism in Kerala, India. Ithaca, NY: Cornell University Press.
Hendrischke, Hans and Chongyi Feng (eds.). 1999. The Political Economy of China’s
Provinces: Comparative and Competitive Advantage. London: Routledge.
Herrigal, Gary. 1996. Industrial Constructions: The Sources of German Industrial
Power. Cambridge: Cambridge University Press.
Hertz, Ellen. 1998. The Trading Crowd: An Ethnography of the Shanghai Stock
Market. Cambridge: Cambridge University Press.
References 293

Hetherington, Marc J. 1998. “The Political Relevance of Political Trust.” American

Political Science Review, vol. 92, no. 4: 791–808
Hjellum, Torstein. 1998. “Is a Participant Culture Emerging in China?” in Kjeld
Erik Brødsgaard and David Strand (eds.), Reconstructing Twentieth-Century
China: State Control, Civil Society, and National Identity. Oxford: Clarendon
Press: 216–250.
Ho, David Yau-fai. 1976. “On the Concept of Face.” American Journal of Sociology,
vol. 81, no. 4: 867–884.
Hoddie, Matthew. 2008. “Ethnic Difference and Survey Cooperation in the People’s
Republic of China.” Asian Survey, vol. 48, no. 2: 303–322.
Hong, Zhaohui. 2004. “Mapping the Evolution and Transformation of the New
Private Entrepreneurs in China.” Journal of Chinese Political Science, vol. 9, no.
1: 23–42.
Honig, Emily. 2003. “Socialist Sex: The Cultural Revolution Revisited.” Modern
China, vol. 29, no. 2: 143–175.
Houtkoop-Steenstra, Hanneke. 2000. Interaction and the Standardized Survey
Interview: The Living Questionnaire. Cambridge: Cambridge University Press.
Howell, Nancy. 1990. Surviving Fieldwork: A Report of the Advisory Panel on
Health and Safety in Fieldwork. Arlington, VA: American Anthropological
Association.
Hu, Hsien-chin. 1944. “The Chinese Concepts of ‘Face.’” American Anthropologist,
vol. 46, no. 1, pt. 1: 45–64.
Huang, Yasheng. 2008. Capitalism with Chinese Characteristics: Entrepreneurship
and the State. Cambridge: Cambridge University Press.
1996. Inflation and Investment Controls in China: The Political Economy of
Central-Local Relations during the Reform Era. Cambridge: Cambridge
University Press.
1995. “Administrative Monitoring in China.” China Quarterly, no. 143: 828–843.
1994. “Information, Bureaucracy, and Economic Reforms in China and the Soviet
Union.” World Politics, vol. 47, no. 1: 102–134.
Huntington, Samuel P. 1987. “The Goals of Development,” in Myron Weiner and
Samuel P. Huntington (eds.), Understanding Political Development: An Analytic
Study. Boston: Little, Brown: 3–32.
Hurst, William. 2009. The Chinese Worker after Socialism. Cambridge: Cambridge
University Press.
Iarossi, Giuseppe. 2006. The Power of Survey Design: A User’s Guide for Managing
Surveys, Interpreting Results, and Influencing Respondents. Washington,
DC: World Bank.
Inglehart, Ronald. 1997. Modernization and Postmodernization: Cultural, Eco­
nomic, and Political Change in 43 Societies. Princeton, NJ: Princeton University
Press.
Jacka, Tamara. 2004. On the Move: Women and Rural-to-Urban Migration in
Contemporary China. New York: Columbia University Press.
Jankowiak, William R. 1993. Sex, Death and Hierarchy in a Chinese City: An
Anthropological Account. New York: Columbia University Press.
Jennings, M. Kent. 2003. “Local Problem Agendas in the Chinese Countryside as
Viewed by Cadres and Villagers.” Acta Politica, vol. 38, no. 4: 313–332.
1998a. “Gender and Political Participation in the Chinese Countryside.” Journal of
Politics, vol. 60, no. 4: 954–973.
294 References

1998b. “Political Trust and the Roots of Devolution,” in V. A. Braithwaite and M. Levi
(eds.), Trust and Governance. New York: Russell Sage Foundation: 218–244.
1997. “Political Participation in the Chinese Countryside.” American Political
Science Review, vol. 91, no. 2: 351–372.
Jennings, M. Kent and Kuang-Hui Chen. 2008. “Perceptions of Injustice in the Chinese
Countryside.” Journal of Contemporary China, vol. 17, no. 55: 319–337.
Jennings, M. Kent and Ning Zhang. 2005. “Generations, Political Status, and
Collective Memories in the Chinese Countryside.” Journal of Politics, vol. 67,
no. 4: 1164–1189.
Jervis, Robert. 1976. Perception and Misperception in International Politics.
Princeton, NJ: Princeton University Press.
Jiang Zemin. 1999. Jianding jinxin shenhua gaige, kaichuang guoyou qiye fazhan
de xin jumian: Zai dongbei he huabei diqu guoyou qiye gaige he fazhan zuotan
huishang de jianghua, 1999 nian 8 yue 12 ri [Keep the Faith, Deepen Reform,
Initiate a New Phase of SOE Development: Speech at the Conference on Reform
and Development of SOEs in the Northeast and North China Regions, August
12, 1999]. Beijing: Renmin chubanshe.
Jin Chongji (ed.). 1989. Zhou Enlai zhuan, 1898–1949 [A Biography of Zhou Enlai].
Beijing: Zhongyang wenxian chubanshe.
Jing, Jun. 1996. The Temple of Memories: History, Power, and Morality in a Chinese
Village. Stanford, CA: Stanford University Press.
Johnston, Alastair Iain. 2006. “The Correlates of Beijing Public Opinion toward the
United States, 1998–2004,” in Alastair Iain Johnston and Robert S. Ross (eds.),
New Directions in the Study of China’s Foreign Policy. Stanford, CA: Stanford
University Press: 340–377.
2004. “Chinese Middle Class Attitudes towards International Affairs: Nascent
Liberalization?” China Quarterly, no. 179: 603–628.
1996. “Cultural Realism and Strategy in Maoist China,” in Peter J. Katzenstein
(ed.), The Culture of National Security. New York: Columbia University
Press: 216–256.
Johnston, Alastair Iain and Daniela Stockmann. 2007. “Chinese Attitudes toward the
United States and Americans,” in Peter J. Katzenstein and Robert O. Keohane
(eds.), Anti-Americanism in World Politics. Ithaca, NY: Cornell University
Press: 157–195.
Jones, Benjamin and Benjamin Olken. 2005. “Do Leaders Matter? National Leadership
and Growth since World War II.” Quarterly Journal of Economics, vol. 120, no.
3: 835–864.
Judd, Ellen R. 1994. Gender and Power in Rural North China. Stanford, CA: Stanford
University Press.
Kang, David C. 2002. Crony Capitalism: Corruption and Development in South
Korea and the Philippines. Cambridge: Cambridge University Press.
Kateb, George. 2000. “Is Patriotism a Mistake?” Social Research, vol. 67, no.
4: 901–924.
Kennedy, John James. 2002. “The Face of ‘Grassroots Democracy’ in Rural China: Real
Versus Cosmetic Elections.” Asian Survey, vol. 42, no. 3: 456–482.
Kennedy, John James, Scott Rozelle, and Shi Yaojiang. 2004. “Elected Leaders and
Collective Land: Farmers’ Evaluation of Village Leaders’ Performance in Rural
China.” Journal of Chinese Political Science, vol. 9, no. 1: 1–22.
References 295

Kennedy, Scott. 2005. The Business of Lobbying in China. Cambridge, MA: Harvard
University Press.
Kertzer, David and Dominique Arel. 2006. “Population Composition as an Object
of Political Struggle,” in Robert E. Goodin and Charles Tilly (eds.), The Oxford
Handbook of Contextual Political Analysis. Oxford: Oxford University
Press: 664–680.
Kidder, Tracy. 2003. Mountains beyond Mountains: The Quest of Dr. Paul Farmer, a
Man Who Would Cure the World. New York: Random House.
Kilwein, John C. and Richard A. Brisbin, Jr. 1997. “Policy Convergence in a Federal
Judicial System: The Application of Intensified Scrutiny Doctrines by State Supreme
Courts.” American Journal of Political Science, vol. 41, no. 1: 122–148.
Kim, Samuel. 1979. China, the United Nations and World Order. Princeton,
NJ: Princeton University Press.
King, Gary. 2004. “Enhancing the Validity and Cross-Cultural Comparability of
Measurement in Survey Research.” American Political Science Review, vol. 98,
no. 1: 191–207.
King, Gary, Robert Keohane, and Sidney Verba. 1994. Designing Social
Inquiry: Scientific Inference in Qualitative Research. Princeton, NJ: Princeton
University Press.
King, Gary, Christopher J. L. Murray, Joshua A. Salomon, and Ajay Tandon. 2003.
“Enhancing the Validity and Cross-cultural Comparability of Measurement in
Survey Research.” American Political Science Review, vol. 97, no. 4: 567–583.
King, Neil Jr. 2005. “A Scholar Shapes Views of China.” Wall Street Journal,
September 8: 1.
Kipnis, Andrew B. 1997. Producing Guanxi: Sentiment, Self, and Subculture in a
North China Village. Durham, NC: Duke University Press.
Kish, Leslie. 1962. “Studies of Interviewer Variance for Attitudinal Variables.” Journal
of the American Statistical Association, vol. 57, no. 297: 92–115.
Kissinger Transcripts: The Top Secret Talks with Beijing and Moscow. 1998. Ed.
William Burr. New York: New Press.
Kline, Rex B. 2005. Principles and Practice of Structural Equation Modeling. 2nd ed.
New York: Guilford Press.
Kluver, Randy and Chen Yang. 2005. “The Internet in China: A Meta-review of
Research.” Information Society, vol. 21, no. 4: 301–308.
Kohli, Atul. 1987. The State and Poverty in India: The Politics of Reform.
Cambridge: Cambridge University Press.
Kooghe, Liesbet. 1996. Subnational Mobilization in the European Union.
Florence: European University Institute Working Paper RSC 95/6.
Kosterman, Rick and Seymour Feshbach. 1989. “Toward a Measure of Patriotic and
Nationalistic Attitudes.” Political Psychology, vol. 10, no. 2: 257–274.
Kou, Chien-wen. 2008. Chinese Political Elites Database. Taipei: National Chengchi
University.
Krebs, Ronald. 2006. Fighting for Rights: Military Service and the Politics of
Citizenship. Ithaca, NY: Cornell University Press.
Krug, Barbara. 2004. China’s Rational Entrepreneurs: The Development of the New
Private Business Sector. New York: RoutledgeCurzon.
Ku, Hok Bun. 2003. Moral Politics in a South Chinese Village: Responsibility,
Reciprocity, and Resistance. Lanham, MD: Rowman and Littlefield.
296 References

Ku, Lun-Wei, Tung-Ho Wu, Li-Ying Lee, and Hsin-Hsi Chen. 2005. Construction of
an Evaluation Corpus for Opinion Extraction. Paper presented at the NTCIR-5
workshop, December, Tokyo.
Kuan, Hsin-chi and Lau Siu-kai. 2002. “Traditional Orientations and Political
Participation in Three Chinese Societies.” Journal of Contemporary China, vol.
11, no. 31: 297–318.
Kuran, Timur. 1995. Private Truths, Public Lies: The Social Consequences of
Preference Falsification. Cambridge, MA: Harvard University Press.
Lacy, Stephen, Daniel Riffe, Staci Stoddard, Hugh Martin, and Kuang-Kuo Chang.
2001. “Sample Size for Newspaper Content Analysis in Multi-Year Studies.”
Journalism and Mass Communication Quarterly, vol. 78, no. 4: 836–845.
Lamb, Malcolm. 2003. Directory of Officials and Organizations in China. Armonk,
NY: M. E. Sharpe.
Landry, Pierre F. 2008a. Decentralized Authoritarianism in China: The Communist
Party’s Control of Local Elites in the Post-Mao Era. Cambridge: Cambridge
University Press.
2008b. “The Institutional Diffusion of Courts in China: Evidence from Survey Data,”
in Tom Ginsburg and Tamir Moustafa (eds.), Rule by Law: The Politics of Courts
in Authoritarian Regimes. Cambridge: Cambridge University Press: 207–234.
Landry, Pierre F. and Mingming Shen. 2005. “Reaching Migrants in Survey
Research: The Use of the Global Positioning System to Reduce Coverage Bias in
China.” Political Analysis, vol. 13, no. 1: 1–22.
Langman, Lauren. 2005. “From Virtual Public Spheres to Global Justice: A Critical
Theory of Internetworked Social Movements.” Sociological Theory, vol. 23, no.
1: 42–74.
Laver, Michael and John Garry. 2000. “Estimating Policy Positions from Political
Texts.” American Journal of Political Science, vol. 44, no. 3: 619–634.
Lavrakas, Paul J. 1993. Telephone Survey Methods: Sampling, Selection, and
Supervision. Thousand Oaks, CA: Sage.
Lee, Ching Kwan. 1998. Gender and the South China Miracle: Two Worlds of Factory
Women. Berkeley: University of California Press.
Lee, Hong Yung. 1991. From Revolutionary Cadres to Party Technocrats in Socialist
China. Berkeley: University of California Press.
Lee, Taeku. 2002. Mobilizing Public Opinion: Black Insurgency and Racial Attitudes
in the Civil Rights Era. Chicago: University of Chicago Press.
Lee, Tahirih V. (ed.) 1997. Contract, Guanxi, and Dispute Resolution in China. New
York: Garland.
Levi, Margaret. 2004. “An Analytic Narrative Approach to Puzzles and Problems,” in
Ian Shapiro, Rogers Smith, and Tarek Masoud (eds.), Problems and Methods in
the Study of Politics. Cambridge: Cambridge University Press: 201–226.
1999. “The Problem of Trust.” American Journal of Sociology, vol. 104, no.
4: 1245–1246.
Levi, Margaret and Laura Stoker. 2000. “Political Trust and Trustworthiness.”
Annual Review of Political Science, vol. 3: 475–507.
Li, Cheng. 2004. “Political Localism versus Institutional Restraints: Elite Recruitment
in the Jiang Era,” in Barry Naughton and Dali L. Yang (eds.), Holding
China Together: Diversity and National Integration in the Post-Deng Era.
Cambridge: Cambridge University Press: 29–69.
References 297

2001. China’s Leaders: The New Generation. Lanham, MD: Rowman and
Littlefield.
2000. “Jiang Zemin’s Successors: The Rise of the Fourth Generation of Leaders in
the PRC.” China Quarterly, no. 161: 1–40.
1994. “University Networks and the Rise of Qinghua Graduates in China’s
Leadership.” Australian Journal of Chinese Affairs, no. 32: 1–30.
Li, Cheng and Lynn White, III. 1991. “China’s Technocratic Movement and the World
Economic Herald.” Modern China, vol. 17, no. 3: 342–388.
Li, Hongbin, Lingsheng Meng, and Junsen Zhang. 2006. “Why Do Entrepreneurs Enter
Politics? Evidence from China.” Economic Inquiry, vol. 44, no. 3: 559–578.
Li, Hongbin and Li-An Zhou. 2005. “Political Turnover and Economic Performance: The
Incentive Role of Personnel Control in China.” Journal of Public Economics, vol.
89, no. 9–10: 1743–1762.
Li, Lianjiang. 2008. “Political Trust and Petitioning in the Chinese Countryside.”
Comparative Politics, vol. 39, no. 4: 209–226.
2004. “Political Trust in Rural China.” Modern China, vol. 30, no. 2: 228–258.
2003. “The Empowering Effect of Village Elections in China.” Asian Survey, vol.
43, no. 4: 648–662.
2002. “Elections and Popular Resistance in Rural China.” China Information, vol.
16, no. 1: 89–107.
2001. “Support for Anti-Corruption Campaigns in Rural China.” Journal of
Contemporary China, vol. 10, no. 29: 573–586.
Liang, Zai and Zhongdong Ma. 2004. “China’s Floating Population: New Evidence
from the 2000 Census.” Population and Development Review, vol. 30, no.
3: 467–490.
Liao, Kuang-sheng and Allen S. Whiting. 1973. “Chinese Press Perceptions of
Threat: The U.S. and India, 1962.” China Quarterly, no. 53: 80–97.
Lieberthal, Kenneth. 1995. Governing China. New York: W.W. Norton.
1971. “Mao Versus Liu? Policy Towards Industry and Commerce, 1946–1949.”
China Quarterly, no. 47: 494–520.
Lieberthal, Kenneth and David M. Lampton (eds.). 1992. Bureaucracy, Politics, and
Decision Making in Post-Mao China. Berkeley: University of California Press.
Lieberthal, Kenneth and Michel Oksenberg. 1988. Policy Making in China: Leaders,
Structures, and Processes. Princeton, NJ: Princeton University Press.
Linz, Juan J. and Armando de Miguel. 1966. “Within-Nation Differences and
Comparisons: The Eight Spains,” in Richard L. Merritt and Stein Rokkan (eds.),
Comparing Nations: The Use of Quantitative Data in Cross-National Research.
New Haven, CT: Yale University Press: 267–320.
Litzinger, Ralph. 2000. Other Chinas: The Yao and the Politics of National Belonging.
Durham, NC: Duke University Press.
Liu, Xin. 2000. In One’s Own Shadow: An Ethnographic Account of the Condition
of Post-Reform Rural China. Berkeley: University of California Press.
Luhtanen, Riia and Jennifer Crocker. 1992. “A Collective Self-Esteem Scale: Self-
Evaluation of One’s Social Identity.” Personality and Social Psychology Bulletin,
vol. 18, no. 3: 302–318.
Luo, Xiaopeng. 1994. “Rural Reform and the Rise of Localism,” in Jia Hao and Lin
Zhimin (eds.), Changing Central-Local Relations in China: Reform and State
Capacity. Boulder, CO: Westview: 115–134.
298 References

MacFarquhar, Roderick. 1997. The Origins of the Cultural Revolution: 3. The Coming
of the Cataclysm, 1961–1966. New York: Columbia University Press.
1983. The Origins of the Cultural Revolution: 2. The Great Leap Forward 1958–60.
New York: Columbia University Press.
1974. The Origins of the Cultural Revolution: 1. Contradictions among the People
1956–7. New York: Columbia University Press.
MacFarquhar, Roderick and Michael Schoenhals. 2006. Mao’s Last Revolution.
Cambridge, MA: Belknap Press of Harvard University Press.
MacLean, Lauren Morris. 2004. “Empire of the Young: The Legacies of State
Agricultural Policy on Local Capitalism and Social Support Networks in Ghana
and Cote d’Ivoire.” Comparative Studies in Society and History, vol. 46, no.
3: 469–496.
Mahoney, James and Dietrich Rueschmeyer (eds.). 2003. Comparative Historical
Analysis in the Social Sciences. Cambridge: Cambridge University Press.
Mainland China Research Center. 2006. Zhongguo jingying ziliao ku [Database on
Chinese Elites]. Taipei: National Chengchi University.
Malinowski, Bronislaw. 1922. Argonauts of the Western Pacific: An Account of Native
Enterprise and Adventure in the Archipelagoes of Melanesian New Guinea. New
York: E. P. Dutton.
Manion, Melanie. 2006. “Democracy, Community, Trust: The Impact of Chinese Village
Elections in Context.” Comparative Political Studies, vol. 39, no. 3: 301–324.
1996. “The Electoral Connection in the Chinese Countryside.” American Political
Science Review, vol. 90, no. 4: 736–748.
1994. “Survey Research in the Study of Contemporary China: Learning from Local
Samples.” China Quarterly, no. 139: 741–765.
1993. Retirement of Revolutionaries in China: Public Policies, Social Norms,
Private Interests. Princeton, NJ: Princeton University Press.
1991. “Policy Implementation in the People’s Republic of China: Authoritative
Decisions versus Individual Interests.” Journal of Asian Studies, vol. 50, no.
2: 253–279.
Marcus, George E. 1998. Ethnography through Thick and Thin. Princeton,
NJ: Princeton University Press.
Marshall, Eliot. 1993. “U.S. May Renew Collaboration after China Relents on Data.”
Science (August 6): 677.
Martin, Lanny W. 2004. “The Government Agenda in Parliamentary Democracies.”
American Journal of Political Science, vol. 48, no. 3: 445–461.
Maryland Study. 2002. “Perspectives Towards the United States in Selected Newspapers
of the People’s Republic of China.” Report for the U.S. China Security Report
Commission.
McDermott, Rose. 2006. “Editor’s Introduction.” Political Psychology, vol. 27, no.
3: 347–358.
McGregor, Richard. 2006. “Data Show Social Unrest on the Rise in China.”
Financial Times, at http://www.ft.com/cms/s/0/171fb682–88d6–11da-94a6–
0000779e2340.html, accessed January 2010.
McKinnon, Rebecca. 2008. “Flatter World and Thicker Walls: Blogs, Censorship and
Civic Discourse in China.” Public Choice, vol. 134, no. 1/2: 31–46.
Medeiros, Evan and M. Taylor Fravel. 2003. “China’s New Diplomacy.” Foreign
Affairs, vol. 82, no. 6: 22–35.
Mengin, Françoise. 2004. Cyber China: Reshaping National Identities in the Age of
Information. New York: Palgrave.
References 299

Michelmann, Hans J. and Pnayotis Soldatos (eds.). 1990. Federalism and International
Relations: The Role of Subnational Units. Oxford: Clarendon Press.
Mooney, Paul. 2005. “Internet Fans Flames of Chinese Nationalism.” Yale Global, at
http://yaleglobal.yale.edu/display.article?id=5516, accessed February 2009.
Murphy, Rachel. 2002. How Migrant Labor Is Changing Rural China.
Cambridge: Cambridge University Press.
Nathan, Andrew and Bruce Gilley. 2002. China’s New Rulers: The Secret Files. New
York: New York Review of Books.
Nathan, Andrew and Tianjian Shi. 1996. “Left and Right with Chinese
Characteristics: Issues and Alignments in Deng Xiaoping’s China.” World
Politics, vol. 48, no. 4: 522–550.
1993. “Cultural Requisites for Democracy in China: Some Findings from a
Nationwide Survey.” Daedalus, vol. 122, no. 2: 95–123.
National Bureau of Statistics of the PRC. 2009. “2008 nianmo quanguo nongmin-
gong zongliang wei 22542 wan ren,” at http://www.stats.gov.cn/tjfx/fxbg/
t20090325_402547406.htm, accessed August 2009. See a translation and discus-
sion of this document in Boxun News, “Chinese Migrant Workers Totaled 225.42
Million at the End of 2008,” at http://www.boxun.us/news/publish/chinanews/
Chinese_Peasant_Workers_Totaled_225_42_Million_at_the_End_of_2008.
shtml, accessed August 2009.
2004a. “She wai diaocha guanli banfa” [Measures for the Administration of
Foreign-Affiliated Surveys]. October 13, 2004, at http://www.stats.gov.cn/tjgl/
swdcglgg/xgfg/t20041118_402209105.htm, accessed June 2008.
2004b. “Tongjiju jiu gongbu ‘she wai diaocha guanli banfa’ dajizhe wen”
[National Bureau of Statistics Official Responds to Journalists’ Questions about
Measures for the Administration of Foreign-Affiliated Surveys]. October 19,
at http://www.chinacourt.org/public/detail.php?id=135541, accessed August
2009.
2004c. “Guanyu ‘she wai diaocha guanli banfa’ de ershiliu tiao” [On Article 26 of
the Measures for the Administration of Foreign-Affiliated Surveys]. November
18, 2004, at http://www.stats.gov.cn/tjgl/swdcglgg/spwb/t20041118_402209303.
htm, accessed June 2008.
Naughton, Barry. 1996. Growing Out of the Plan: Chinese Economic Reform 1978–
1993. Cambridge: Cambridge University Press.
Neuendorf, Kimberly A. 2002. The Content Analysis Guidebook. Thousand Oaks,
CA: Sage.
“Never Forget the National Humiliation” [Wuwang guochi]. 1992. History Book
Series. Beijing: Zhongguo huaqiao chubanshe.
Nevitt, Christopher. 1996. “Private Business Associations in China: Evidence of Civil
Society or Local State Power?” China Journal, no. 36: 25–45.
Nicholson-Crotty, Sean and Kenneth J. Meier. 2002. “Size Doesn’t Matter: In
Defense of Single-State Studies.” State Politics & Policy Quarterly, vol. 2, no.
4: 411–422.
Nisbett, Richard. 2003. The Geography of Thought: How Asians and Westerners
Think Differently, and Why. New York: Free Press.
Nohria, Nitin and Ranjay Gulati. 1994. “Firms and their Environments,” in Neil
Smelser and Richard Swedberg (eds.), The Handbook of Economic Sociology.
Princeton, NJ: Princeton University Press: 529–555.
Nolan, Peter. 2001. China and the Global Economy: National Champions, Industrial
Policy, and the Big Business Revolution. New York: Palgrave.
300 References

Notar, Beth E. 2006. Displacing Desire: Travel and Popular Culture in China.
Honolulu: University of Hawaii Press.
Nyiri, Pal and Joanna Briedenbach. 2005. China Inside Out: Contemporary Chinese
Nationalism and Transnationalism. New York: Central European University Press.
O’Brien, Kevin. 2006. “Discovery, Research (Re)design and Theory-Building,”
in Maria Heimer and Stig Thøgersen (eds.), Doing Fieldwork in China.
Honolulu: University of Hawaii Press: 27–41.
2003. “Neither Transgressive nor Contained: Boundary-Spanning Contention in
China.” Mobilization, vol. 8, no. 1: 51–64.
1996. “Rightful Resistance.” World Politics, vol. 49, no. 1: 31–55.
O’Brien, Kevin J. and Lianjiang Li. 2006. Rightful Resistance in Rural China.
Cambridge: Cambridge University Press.
Oi, Jean C. 1998a. “The Evolution of Local State Corporatism,” in Andrew Walder
(ed.), Zouping in Transition The Process of Reform in Rural North China.
Cambridge, MA: Harvard University Press: 35–61.
1998b. Rural China Takes Off: Institutional Foundations of Economic Reform.
Berkeley: University of California Press.
1986. “Commercializing China’s Rural Cadres.” Problems of Communism, vol. 35,
no. 5: 1–15.
Oi, Jean C. and Scott Rozelle. 2000. “Elections and Power: The Locus of Decision-
Making in Chinese Villages.” China Quarterly, no. 162: 513–539.
Oksenberg, Michel. 1970. A Bibliography of Secondary English Language Literature
on Contemporary Chinese Politics. New York: East Asian Institute, Columbia
University.
1964. “Sources and Methodological Problems in the Study of Contemporary
China,” in A. Doak Barnett (ed.), Chinese Communist Politics in Action.
Seattle: University of Washington Press: 577–606.
Oksenberg, Michel and Harold Jacobson. 1990. China’s Participation in the IMF,
the World Bank, and the GATT: Toward a Global Economic Order. Ann
Arbor: University of Michigan Press.
O’Leary, Cecilia. 1999. To Die For: The Paradox of American Patriotism. Princeton,
NJ: Princeton University Press.
Olzak, Susan. 1989. “Analysis of Events in the Study of Collective Action.” Annual
Review of Sociology, vol. 15: 119–141.
Orleans, Leo A. 1974. “Chinese Statistics: The Impossible Dream.” The American
Statistician, vol. 128, no. 2: 47–52.
Paluck, Elizabeth Levy. 2007. Reducing Intergroup Prejudice and Conflict with
the Mass Media: A Field Experiment in Rwanda. Ph.D. dissertation, Yale
University.
Park, Albert. 2006. “Using Survey Data in Social Science Research in Developing
Countries, ” in Ellen Perecman and Sara R. Curran (eds.), A Handbook for Social
Science Field Research: Essays and Bibliographic Sources on Research Design
and Methods. Thousand Oaks, CA: Sage: 117–134.
Parris, Kristin. 1993. “Local Initiative and National Reform: The Wenzhou Model of
Development.” China Quarterly, no. 134: 242–263.
Pearson, Margaret. 1994. “The Janus Face of Business Associations in China: Socialist
Corporatism in Foreign Enterprises.” Australian Journal of Chinese Affairs, no.
31: 25–46.
References 301

Pei, Minxin. 2006. China’s Trapped Transition: The Limits of Developmental

Autocracy. Cambridge, MA: Harvard University Press.
1998. “Is China Democratizing?” Foreign Affairs, vol. 77, no. 1: 68–82.
People ’ s Daily. 2006. “Private Business People Gain More Influence in China.” October
30, at http://english.peopledaily.com.cn/200610/30/eng20061030_316473.html,
accessed August 2009.
Perecman, Ellen and Sara R. Curran (eds.). 2006. A Handbook for Social Science
Research: Essays and Bibliographic Sources on Research Design and Methods.
Thousand Oaks, CA: Sage.
Perkins, Tamara. 2002. Village, Market and Well-Being in a Rural Chinese Township.
New York: Routledge.
Perreault, William D., Jr. and Laurence E. Leigh. 1989. “Reliability of Nominal Data
Based on Qualitative Judgments.” Journal of Marketing Research, vol. 26, no.
2: 135–148.
Perry, Elizabeth J. 2007. “Studying Chinese Politics: Farewell to Revolution?” China
Journal, no. 57: 1–22.
2006. Patrolling the Revolution: Worker Militias, Citizenship and the Modern
Chinese State. Lanham, MD: Rowman and Littlefield.
1994a. “Shanghai’s Strike Wave of 1957.” China Quarterly, no. 137: 1–27.
1994b. “Trends in the Study of Chinese Politics: State-Society Relations.” China
Quarterly, no. 139: 704–713.
Perry, Elizabeth and Xun Li. 1997. Proletarian Power: Shanghai in the Cultural
Revolution. Boulder, CO: Westview.
1993. “Revolutionary Rudeness: The Language of Red Guards and Rebel Workers
in China’s Cultural Revolution.” Indiana East Asian Working Paper Series on
Language and Politics in Modern China, no. 2.
Pierson, Paul. 2004. Politics in Time: History, Institutions, and Social Analysis.
Princeton, NJ: Princeton University Press.
Polsky, Ned. 2006. Hustlers, Beats, and Others. Piscataway, NJ: Transaction.
Popper, Karl R. 2001. All Life Is Problem Solving. New York: Routledge.
Porter, Bruce. 1994. War and the Rise of the State: The Military Foundations of
Modern Politics. New York: Free Press.
Posner, Daniel N. 2004. “The Political Salience of Cultural Difference: Why Chewas
and Tumbukas Are Allies in Zambia and Adversaries in Malawi.” American
Political Science Review, vol. 98, no. 4: 529–545.
Przeworski, Adam and Henry Teune. 1970. The Logic of Comparative Social Inquiry.
New York: Wiley-Interscience,
Pun, Ngai. 2005. Made in China: Women Factory Workers in a Global Marketplace.
Durham, NC: Duke University Press.
Putnam, Robert D. 1993. Making Democracy Work: Civic Traditions in Modern
Italy. Princeton, NJ: Princeton University Press.
Pye, Lucian W. 1968. The Spirit of Chinese Politics: A Psychocultural Study of the
Authority Crisis in Political Development. Cambridge, MA: M.I.T. Press.
Ragin, Charles. 1987. The Comparative Method: Moving beyond Qualitative and
Quantitative Strategies. Berkeley: University of California Press.
Ramseyer, J. Mark and Frances McCall Rosenbluth. 1998. The Politics of
Oligarchy: Institutional Choice in Imperial Japan. Cambridge: Cambridge
University Press.
302 References

Remick, Elizabeth. 2004. Building Local States: China during the Republican and
Post-Mao Eras. Cambridge, MA: Asia Center, Harvard University.
Renmin chubanshe. 1997. Zhongguo gongchandang dishiwuci quanguo daibiao dahui
wenjian huibian [Collection of Documents from the Fifteenth Party Congress of
the CCP]. Beijing.
Riffe, Daniel, Stephen Lacy, and Michael W. Drager. 1996. “Sample Size in Content
Analysis of Weekly News Magazines.” Journalism and Mass Communication
Quarterly, vol. 73, no. 3: 635–644.
Riffe, Daniel, Stephen Lacy, and Frederick G. Fico. 1998. Analyzing Media
Messages: Using Quantitative Content Analysis in Research. Mahwah,
NJ: Lawrence Erlbaum.
Roberts, Carl W. 2000. “A Conceptual Framework for Quantitative Text Analysis.”
Quality & Quantity, vol. 34, no. 3: 259–274.
1989. “Other than Counting Words: A Linguistic Approach to Content Analysis.”
Social Forces, vol. 68, no. 1: 147–177.
Robertson, Graeme B. 2007. “Strikes and Labor Organization in Hybrid Regimes.”
American Political Science Review, vol. 101, no. 4: 799–809.
Rofel, Lisa. 1999. Other Modernities: Gendered Yearnings in China after Socialism.
Berkeley: University of California Press.
Roitman, Janet. 2004. Fiscal Disobedience: An Anthropology of Economic Regulation
in Central Africa. Princeton, NJ: Princeton University Press.
Rong, Hu. 2005. “Economic Development and the Implementation of Village Elections
in Rural China.” Journal of Contemporary China, vol. 14, no. 44: 427–444.
Rosenau, Pauline. 1992. Post-modernism and the Social Sciences: Insights, Inroads
and Intrusions. Princeton, NJ: Princeton University Press.
Ross, Robert S. 2001. “Introduction,” in Robert S. Ross and Jiang Changbin (eds.),
Re-Examining the Cold War: U.S.-China Diplomacy, 1954–1973. Cambridge,
MA: Asia Center, Harvard University.
Rueschemeyer, Dietrich, Evelyne Huber Stephens, and John D. Stephens. 1992.
Capitalist Development and Democracy. Chicago: University of Chicago Press.
Saich, Tony. 1992. “The Fourteenth Party Congress: A Programme for Authoritarian
Rule.” China Quarterly, no. 132: 1136–1160.
Saunders, Robert A. and Sheng Ding. 2006. “Digital Dragons and Cybernetic Bears.”
Nationalism and Ethnic Politics, vol. 12, no. 2: 255–290.
Schatz, Edward. 2006. The Problem with the Toolbox Metaphor: Ethnography and
the Limits to Multiple-Methods Research. Paper prepared for the 2006 annual
meeting of the American Political Science Association, Philadelphia, PA.
2004. Modern Clan Politics: The Power of “Blood” in Kazakhstan and Beyond.
Seattle: University of Washington Press.
Schein, Louisa. 2000. Minority Rules: The Miao and the Feminine in China’s Cultural
Politics. Durham, NC: Duke University Press.
Schober, Michael F. and Frederick G. Conrad. 1997. “Does Conversational
Interviewing Reduce Survey Measurement Error?” Public Opinion Quarterly,
vol. 61, no. 4: 576–602.
Schram, Stuart R. 1989. The Thought of Mao Tse-Tung. Cambridge: Cambridge
University Press.
Schrank, Andrew. 2006. “Case-Based Research,” in Ellen Perecman and Sara R.
Curran (eds.), A Handbook for Social Science Research: Essays and Bibliographic
Sources on Research Design and Methods. Thousand Oaks, CA: Sage: 21–45.
References 303

Schumacker, Randall E. and Richard G. Lomax. 2004. A Beginner’s Guide to

Structural Equation Modeling. 2nd ed. Mahwah, NJ: Lawrence Erlbaum.
Schwartz, Benjamin. 1966. Chinese Communism and the Rise of Mao. Cambridge,
MA: Harvard University Press.
Scott, James C. 1998. Seeing like a State: How Certain Schemes to Improve the
Human Condition Have Failed. New Haven, CT: Yale University Press.
1990. Domination and the Arts of Resistance: Hidden Transcripts. New Haven,
CT: Yale University Press.
1985. Weapons of the Weak: Everyday Forms of Peasant Resistance. New Haven,
CT: Yale University Press.
Shambaugh, David (ed.). 1993. American Studies of Contemporary China. Armonk,
NY: M. E. Sharpe.
Shapiro, Gilbert and John Markoff. 1998. Revolutionary Demands: A Content
Analysis of the Cashier de doléances of 1789. Stanford, CA: Stanford University
Press.
Shatz, Robert T., Ervin Staub, and Howard Levine. 1999. “On the Varieties of National
Attachment: Blind Versus Constructive Patriotism.” Political Psychology, vol. 20,
no. 1: 151–174.
Shi, Tianjian. 2001. “Cultural Values and Political Trust: A Comparison of the
People’s Republic of China and Taiwan.” Comparative Politics, vol. 33, no. 4:
401–419.
2000. “Cultural Values and Democracy in the People’s Republic of China.” China
Quarterly, no. 162 (June): 540–559.
1999a. “Economic Development and Village Elections in Rural China.” Journal of
Contemporary China, vol. 8, no. 22: 425–442.
1999b. “Mass Political Behavior in Beijing,” in Merle Goldman and Roderick
MacFarquhar (eds.), The Paradox of China’s Post-Mao Reforms. Cambridge,
MA: Harvard University Press: 145–169.
1999c. “Voting and Nonvoting in China: Voting Behavior in Plebiscitary and
Limited-Choice Elections.” Journal of Politics, vol. 61, no. 4: 1115–1139.
1997. Political Participation in Beijing. Cambridge, MA: Harvard University
Press.
1996. “Survey Research in China,” in Michael X. Delli Carpini, Leonie Huddy, and
Robert Y. Shapiro (eds.), Research in Micropolitics: Rethinking Rationality. vol.
5. Greenwich, CT: JAI Press: 216–220.
Shie, Tamara Renee. 2004. “The Tangled Web.” Journal of Contemporary China,
vol. 13, no. 40: 523–540.
Shih, Victor. 2004. “Factions Matter: Personal Networks and the Distribution of
Bank Loans in China.” Journal of Contemporary China, vol. 13, no. 38: 3–19.
Shirk, Susan. 1993. The Political Logic of Economic Reform in China.
Berkeley: University of California Press.
1982. Competitive Comrades: Career Incentives and Student Strategies in China.
Berkeley: University of California Press.
Shue, Vivienne. 1988. The Reach of the State: Sketches of the Chinese Body Politic.
Stanford, CA: Stanford University Press.
Sil, Rudra. 2004. “Problems Chasing Methods or Methods Chasing Problems?
Research Communities, Constrained Pluralism, and the Role of Eclecticism,” in
Ian Shapiro, Rogers Smith, and Tarek Masoud (eds.), Problems and Methods in
the Study of Politics. Cambridge: Cambridge University Press: 307–331.
304 References

2000a. “Against Epistemological Absolutism: Toward a ‘Pragmatic’ Center?”

in Rudra Sil and Eileen M. Doherty (eds.), Beyond Boundaries? Disciplines,
Paradigms, and Theoretical Integration in International Studies. Albany: SUNY
Press: 145–175.
2000b. “The Foundations of Eclecticism: The Epistemological Status of Agency,
Culture, and Structure in Social Theory.” Journal of Theoretical Politics, vol. 12,
no. 3: 353–387.
Sinha, Aseema. 2003. “Rethinking the Developmental State Model: Divided Leviathan
and Subnational Comparisons in India.” Comparative Politics, vol. 35, no.
4: 459–476.
Sluka, Jeffrey. 1990. “Participant Observation in Violent Social Contexts.” Human
Organization, vol. 49, no. 2: 114–126.
SMA1. 1964. (Shanghai Civil Affairs Party Organization to Shanghai CCP) “Guanyu
baohu junren hunyin qingkuang de jiancha” [An Investigation Concerning
the Protection of Soldiers’ Marriages]. 1964. Shanghai Municipal Archives
B168–1–223.
SMA2. 1966. (Shanghai Civil Affairs Party Organization to Shanghai CCP) “Guanyu
junren hunyin caoshou pohuai qingkuang baogao” [An Investigation Report
Concerning the Situation of Soldiers’ Marriages that Have Been Violated].
Shanghai Municipal Archives B168–1–132.
SMA3. 1966. (Shanghai Civil Affairs Party Organization to Shanghai CCP) “Guanyu
guanche shixing zuigao fayuan guanyu junren hunyin caoshou pohuai qingkuang
baogao zuotanhui jiyao” [Notes on a Meeting Regarding the Implementation of
the Supreme People’s Court’s Investigation Concerning the Situation of Soldiers’
Marriages that Have Been Violated]. Shanghai Municipal Archives B168–2–132.
Smith, Arthur H. 1890. Chinese Characteristics. Shanghai: North China Herald.
Snyder, Richard. 2001a. Politics after Neoliberalism: Reregulation in Mexico.
Cambridge: Cambridge University Press.
2001b. “Scaling Down: The Subnational Comparative Method.” Studies in
Comparative International Development, vol. 36, no. 1: 93–110.
So, Ying Lun and Anthony Walker. 2006. Explaining Guanxi: The Chinese Business
Network. New York: Routledge.
Solinger, Dorothy J. 1999. Contesting Citizenship in Urban China: Peasant Migrants,
the State and the Logic of the Market. Berkeley: University of California Press.
1995. ‘‘China’s Urban Transients in the Transition from Socialism and the Collapse
of the Communist ‘Urban Public Goods Regime.’” Comparative Politics, vol. 27,
no. 2: 127–146.
1977. Regional Government and Political Integration in Southwest China, 1949–
1954. Berkeley: University of California Press.
Spencer-Rodgers, Julie, Kaiping Peng, and Lei Wang. 2004. “Dialectical Self-Esteem
and East-West Differences in Psychological Well-being.” Personality and Social
Psychology Bulletin, vol. 30, no. 11: 29–44.
Starr, John Bryan. 1976. “From the 10th Party Congress to the Premiership of Hua
Kuo-feng: The Significance of the Colour of the Cat.” China Quarterly, no.
67: 457–488.
Starr, Paul. 1992. “Social Categories and Claims in the Liberal State.” Social Research,
vol. 59, no. 2: 263–296.
1983. “The Sociology of Official Statistics,” in William Alonso and Paul Starr (eds.),
The Politics of Numbers. New York: Russell Sage Foundation: 7–58.
References 305

State Council. 2005. “Guanyu xiugai ‘Zhonghua renmin gongheguo tongji fa’ shishi
xize” [Revised Detailed Regulations on Implementation of the Statistics Law of
the People’s Republic of China]. December 16, at http://china.findlaw.cn/fagui/
jj/25/101390.html, accessed August 2009.
Steinfeld, Edward S. 1998. Forging Reform in China: The Fate of State-owned
Industry. Cambridge: Cambridge University Press.
Stempel, Guido H. 1952. “Sample Size for Classifying Subject Matter in Dailies.”
Journalism Quarterly, vol. 29, no. 3: 333–334.
Stockmann, Daniela. Forthcoming-a. “Who Believes Propaganda? Media Effects dur-
ing the Anti-Japanese Protests in Beijing.” China Quarterly.
Forthcoming-b. “What Kind of Information Does the Public Demand? Getting the
News during the 2005 Anti-Japanese Protests,” in Susan Shirk (ed.), Changing
Media, Changing China. Under Review. Oxford University Press.
2009. Political Voices in the Chinese Press: Does Media Commercialization
Change the Position of the Authoritarian State? Paper presented at the Media in
Contemporary Chinese Politics workshop, Harvard University, April.
2007. Propaganda for Sale: The Impact of Newspaper Commercialization on
News Content and Public Opinion in China. Ph.D. Dissertation, University of
Michigan.
Stockmann, Daniela and Mary E. Gallagher. 2007. Mass Media Mobilization as a
Means of Legal Reform in China. Paper presented at the annual meeting of the
American Political Science Association, Chicago.
Stone, Philip James. 1997. “Thematic Text Analysis: New Agendas for Analyzing Text
Content,” in Carl W. Roberts (ed.), Text Analysis for the Social Sciences: Methods
for Drawing Statistical Inferences from Texts and Transcripts. Mahwah.
NJ: Lawrence Erlbaum: 35–54.
Stone, Philip James, Dexter C. Dunphy, Marshall S. Smith, and Daniel M. Ogilvie.
1966. The General Inquirer: A Computer Approach to Content Analysis.
Cambridge, MA: MIT Press.
Stoner-Weis, Kathryn. 1997. Local Heroes: The Political Economy of Russian
Regional Governance. Princeton, NJ: Princeton University Press.
Strand, David. 1990. “Protest in Beijing: Civil Society and the Public Sphere in China.”
Problems of Communism, vol. 39, no. 3: 1–19.
Straus, Scott. 2006. The Order of Genocide: Race, Power, and War in Rwanda.
Ithaca, NY: Cornell University Press.
Strauss, Julia. 2002. “Paternalist Terror: The Campaign to Suppress
Counterrevolutionaries and Regime Consolidation in the People’s Republic of
China, 1950–1953.” Comparative Studies in Society and History, vol. 44, no.
1: 80–105.
Streb, Matthew. 2008. “Social Desirability Effects and Support for a Female American
President.” Public Opinion Quarterly, vol. 72, no. 1: 76–89.
Su, Fubing and Dali L. Yang. 2000. “Political Institutions, Provincial Interests, and
Resource Allocation in Reformist China.” Journal of Contemporary China, vol.
9, no. 24: 215–230.
Suchman, Lucy and Brigitte Jordan. 1990. “Interactional Troubles in Face-to-Face
Survey Interviews.” Journal of the American Statistical Association, vol. 85, no.
409: 232–241.
Svolik, Milan. 2006. A Theory of Leadership Dynamics in Authoritarian Regimes.
Paper presented at the Annual Midwest Political Science Association, Chicago, Il.
306 References

Swaine, Michael D. 1992. The Military and Political Succession in China: Leadership,
Institutions, Beliefs. Santa Monica, CA: Rand.
Tai, Zixue. 2006. The Internet in China: Cyberspace and Civil Society. New
York: Routledge.
Tamir, Yael. 1997. “Reflections on Patriotism,” in Daniel Bar-tal and Ervin Staub
(eds.), Patriotism in the Lives of Individuals and Nations. Chicago: Nelson
Hall: 23–41.
Tan, Qingshan and Xin Qiushui. 2007. “Village Elections and Governance: Do
Villagers Care?” Journal of Contemporary China, vol. 16, no. 53:
581–599.
Tang Diaodeng. 2003. Zhongguo chengshi pinkun yu fanpinkun baogao [Report on
Poverty and Anti-Poverty in Urban China]. Beijing: Huaxia chubanshe.
Tang, Wenfang. 2005. Public Opinion and Political Change in China. Stanford,
CA: Stanford University Press.
2003. “Research Guide: An Introduction to Survey Research in China.” Issues and
Studies, vol. 39, no. 1: 269–288.
2001a. “Political and Social Trends in the Post-Deng Urban China: Crisis or
Stability?” China Quarterly, no. 168: 890–909.
2001b. “Religion and Society in Taiwan and China: Evidence from Survey Data,”
in Shiping Hua (ed.), Politics and Political Culture in Contemporary China.
Armonk, NY: M. E. Sharpe: 298–319.
1993. “Workplace Participation in Chinese Local Industrial Enterprises.” American
Journal of Political Science, vol. 37, no. 3: 920–940.
Tang, Wenfang and William L. Parish. 2000. Chinese Urban Life under Market
Reform: The Changing Social Contract. Cambridge: Cambridge University
Press.
1996. “Social Reaction to Urban Reform in China.” Problems of Post-Communism,
vol. 43, no. 6: 35–47.
Tang, Wenfang, William L. Parish, and Tongqing Feng. 1996. “Chinese Labor
Relations in a Changing Work Environment.” Journal of Contemporary China,
vol. 5, no. 13: 367–389.
Tang, Wenfang and Qing Yang. 2008. “The Chinese Urban Caste System in Transition.”
China Quarterly, no. 196: 759–777.
Taubman, Geoffry. 1998. “A Not-So World Wide Web: The Internet, China, and the
Challenges to Non-Democratic Rule.” Political Communication, vol. 15, no.
2: 255–272.
Taylor, Charles. 1989. “Cross-Purposes: The Liberal Communitarian Debate,”
in Nancy L. Rosenblum (ed.), Liberalism and the Moral Life. Cambridge,
MA: Harvard University Press: 159–182.
Teiwes, Frederick. 1993. Politics and Purges in China: Rectification and the Decline
of Party Norms, 1950–1965. Armonk, NY: M. E. Sharpe.
1979. Politics and Purges in China. White Plains, NY: M. E. Sharpe.
Tendler, Judith. 1997. Good Governance in the Tropics. Baltimore, MD: Johns
Hopkins University Press.
Thelen, Kathleen. 2004. How Institutions Evolve: The Political Economy of Skills
in Germany, Britain, the United States, and Japan. Cambridge: Cambridge
University Press.
Thelwall, Mike and Alastair Smith. 2002. “Interlinking between Asia-Pacific Web
Sites.” Scientometrics, vol. 55, no. 3: 363–376.
References 307

Thøgersen, Stig and Soren Clausen. 1992. “New Reflections in the Mirror: Local
Chinese Gazetteers (difangzhi) in the 1980s.” Australian Journal of Chinese
Affairs, no. 27 (January): 161–184.
Thomas, William I. and Florian Znaniecki. 1918–1920. The Polish Peasant in Europe
and America. 2 vols. Chicago: University of Chicago Press.
Thornton, Patricia. 2007. Disciplining the State: Virtue, Violence and State-making
in Modern China. Cambridge, MA: Asia Center, Harvard University.
Thurston, Anne F. 1983. “The Social Sciences and Fieldwork in China: An Overview,”
in Anne F. Thurston and Burton Pasternak (eds.), The Social Sciences and
Fieldwork in China: Views from the Field. Boulder, CO: Westview: 3–36.
Tikhonov, Vladimir. 2007. “Masculinizing the Nation: Gender Ideologies in
Traditional Korea and in the 1890s-1900s Korean Enlightenment Discourse.”
Journal of Asian Studies, vol. 66, no. 4: 1029–1065.
Tilly, Charles. 2002. “Event Catalogs as Theories.” Sociological Theory, vol. 20, no.
2: 248–254.
1978. From Mobilization to Revolution. Reading, MA: Addison-Wesley.
Tilly, Charles, Louise Tilly, and Richard Tilly. 1975. The Rebellious Century, 1830–
1930. Cambridge, MA: Harvard University Press.
Tong, Yanqi. 2007. “Bureaucracy Meets the Environment: Elite Perceptions in Six
Chinese Cities.” China Quarterly, no. 189: 100–121.
Tsai, Kellee S. 2007. Capitalism without Democracy: The Private Sector in
Contemporary China. Ithaca, NY: Cornell University Press.
2005. “Capitalists without a Class: Political Diversity among Private Entrepreneurs
in China.” Comparative Political Studies, vol. 38, no. 9: 1130–1158.
2002. Back-Alley Banking: Private Entrepreneurs in China. Ithaca, NY: Cornell
University Press.
Tsai, Lily L. 2007a. Accountability without Democracy: How Solidary Groups
Provide Public Goods in Rural China. Cambridge: Cambridge University Press.
2007b. “Solidary Groups, Informal Accountability, and Local Public Goods
Provision in Rural China.” American Political Science Review, vol. 101, no.
2: 355–372.
2002. “Cadres, Temple and Lineage Institutions, and Governance in Rural China.”
China Journal, no. 48: 1–28.
Tsui, Lokman. 2005. “Introduction: The Sociopolitical Internet.” China Information,
vol. 19, no. 2: 181–188.
Tullock, Gordon. 1987. Autocracy. Boston: Kluwer Academic.
U, Eddy. 2003. “The Making of Zhishifenzi: The Critical Impact of the Registration of
Unemployed Intellectuals in the Early PRC.” China Quarterly, no. 173: 100–121.
Ulbig, Stacey G. 2002. “Policies, Procedures, and People: Sources of Support for
Government?” Social Science Quarterly, vol. 83, no. 3: 789–809.
Unger, Jonathan. 1996a. “‘Bridges’: Private Business, the Chinese Government and the
Rise of New Associations.” China Quarterly, no. 147: 795–819.
(ed.). 1996b. Chinese Nationalism. Armonk, NY: M. E. Sharpe.
1987. “The Struggle to Dictate China’s Administration: The Conflict of Branches
vs. Areas vs. Reform.” Australian Journal of Chinese Affairs, no. 18: 15–45.
Uslaner, Eric. 2002. The Moral Foundations of Trust. Cambridge: Cambridge
University Press.
van Evera, Stephen. 1997. Guide to Methods for Students of Political Science. Ithaca,
NY: Cornell University Press.
308 References

Varshney, Ashutosh. 2002. Ethnic Conflict and Civic Life: Hindus and Muslims in
India. New Haven, CT: Yale University Press.
Verba, Sidney, Norman H. Nie, and Jae-on Kim. 1971. The Modes of Democratic
Participation: A Cross-National Comparison. Beverly Hills, CA: Sage.
Vermeer, Edward. 1992. “New County Histories: A Research Note on Their
Compilation and Value.” Modern China, vol. 18, no. 4: 438–467.
Viroli, Maurizio. 1995. For Love of Country: An Essay on Patriotism and Nationalism.
Oxford: Clarendon.
Viterna, Jocelyn and D. W. Maynard. 2002. “How Uniform Is Standardization?
Variation Within and Across Survey Research Centers Regarding Protocols
for Interviewing,” in Douglas W. Maynard, et al. (eds.), Standardization and
Tacit Knowledge: Interaction and Practice in the Survey Interview. Malden,
MA: Wiley-Interscience: 365–397.
Vogel, Ezra. 1969. Canton under Communism. Cambridge, MA: Harvard University
Press.
Walder, Andrew, G. 1986. Communist Neo-traditionalism: Work and Authority in
Chinese Industry. Berkeley: University of California Press.
1979. “Press Accounts and the Study of Chinese Society.” China Quarterly, no.
79: 568–592.
Walder, Andrew and Yang Su. 2003. “The Cultural Revolution in the Countryside:
Scope, Timing and Human Impact.” China Quarterly, no. 173: 74–99.
Walzer, Michael. 1970. Obligations: Essays on Disobedience, War and Citizenship.
Cambridge, MA: Harvard University Press.
Wang, Hongying and Honggang Tan. 2008. Chinese Media and the Judicial System
under Soft Authoritarianism. Paper presented at the annual meeting of the
Association for Asian Studies, Atlanta.
Wang Jianying. 1995. Zhongguo gongchandang zuzhishi ziliao huibian [A Collection
of Material Related to CCP Organizational History]. Beijing: Zhongyang dang­
xiao chubanshe.
Wang, Shaoguang. 1995. “The Rise of the Regions: Fiscal Reform and the Decline
of State Capacity,” in Andrew Walder (ed.), The Waning of the Communist
State: Economic Origins of Political Decline in China and Hungary.
Berkeley: University of California Press: 87–114.
Wang, Xiaopeng. 2006. Exploring Sample Sizes for Content Analysis of Online News
Sites. Paper presented at the annual meeting of the Association for Education in
Journalism and Mass Communication.
Wang, Yanlai, Nicholas Rees, and Bernadette Andreosso-O’Callaghan. 2004.
“Economic Change and Political Development in China: Findings from a Public
Opinion Survey.” Journal of Contemporary China, vol. 13, no. 39: 203–222.
Wang, Zhengxu. 2007. “Public Support for Democracy in China.” Journal of
Contemporary China, vol. 16, no. 53: 561–579.
2006. “Explaining Regime Strength in China.” China: An International Journal,
vol. 4, no. 2: 217–237.
Wank, David L. 1999. Commodifying Communism: Business, Trust, and Politics in
a Chinese City. Cambridge: Cambridge University Press.
1998. “Political Sociology and Contemporary China: State-Society Images
in American China Studies.” Journal of Contemporary China, vol. 7, no.
18: 205–228.
References 309

1995. “Private Business, Bureaucracy, and Political Alliance in a Chinese City.”

Australian Journal of Chinese Affairs, no. 33: 55–71.
Watson, James L. and Rubie S. Watson. 2004. Village Life in Hong Kong: Politics,
Gender, and Ritual in the New Territories. Hong Kong: Chinese University
Press.
Wedeen, Lisa. 2004. “Concepts and Commitments in the Study of Democracy,” in Ian
Shapiro, Rogers Smith, and Tarek Masoud (eds.), Problems and Methods in the
Study of Politics. Cambridge: Cambridge University Press: 274–306.
2002. “Conceptualizing Culture: Possibilities for Political Science.” American
Political Science Review, vol. 94, no. 4: 713–728.
Weisberg, Herbert F. 2005. The Total Survey Error Approach: A Guide to the New
Science of Survey Research. Chicago: University of Chicago Press.
White, Gordon, Jude Howell, and Shang Xiaoyuan. 1996. In Search of Civil
Society: Market Reform and Social Change in Contemporary China.
Oxford: Oxford University Press.
Whiting, Susan H. 2001. Power and Wealth in Rural China: The Political Economy
of Institutional Change. Cambridge: Cambridge University Press.
Wich, Richard. 1974. “The Tenth Party Congress: The Power Structure and the
Succession Question.” China Quarterly, no. 58: 231–248.
Wood, Elisabeth Jean. 2006. “The Ethical Challenges of Field Research in Conflict
Zones.” Qualitative Sociology, vol. 29, no. 3: 373–386.
Wood, Linda A. and Rolf O. Kroger. 2000. Doing Discourse Analysis: Methods for
Studying Action in Talk and Text. Thousand Oaks, CA: Sage.
Wu, Xu. 2007. Chinese Cyber Nationalism: Evolution, Characteristics and
Implications. Lanham, MD: Lexington Books.
Xu Xiqiang. 2005. ‘“Zuolian’ ‘wulieshi’ anqing xintan” [New Exploration of the
Situation Surrounding the “Five Martyrs” of the “Left Association”]. Nanjing
shifan daxue wenxueyuan xuebao [Journal of the School of Humanities at
Nanjing University], no. 12: 61–68.
Yan, Yunxiang. 2003. Private Life under Socialism: Love, Intimacy, and Family
Change in a Chinese Village, 1949–1999. Stanford, CA: Stanford University Press.
1996. The Flow of Gifts: Reciprocity and Social Networks in a Chinese Village.
Stanford, CA: Stanford University Press.
Yang, Guobin. 2003. “The Co-Evolution of the Internet and Civil Society in China.”
Asian Survey, vol. 43, no. 3: 405–422.
Yang, Mayfair Mei-hui. 1994. Gifts, Favors, and Banquets: The Art of Social
Relationships in China. Ithaca, NY: Cornell University Press.
Yang, Ming, et al. 2007. 1995–2004 Beijing shehui jingji fazhan niandu diaocha
shuju baogao [Report on Annual Statistical Survey of the Social and Economic
Development of Beijing, 1995–2004]. Beijing: Beijing chubanshe.
Yang, Xiushi. 1993. ‘‘Household Registration, Economic Reform and Migration.’’
International Migration Review, vol. 27, no. 4: 796–818.
Yanow, Dvora. 2003. “Interpretive Political Science: What Makes This Not a Subfield
of Qualitative Methods.” Qualitative Methods (Newsletter of the American
Political Science Association Organized Section on Qualitative Methods), vol. 1,
no. 2: 9–13.
Ye, Wa and Joseph W. Esherick. 1996. Chinese Archives: An Introductory Guide.
Berkeley: Institute of East Asian Studies, University of California.
310 References

Yin Jizuo. 2001. Tizhi gaige shehui zhuanxing: 2001 nian Shanghai shehui fazhan
lanpi shu [System Reform, Social Transition: 2001 Blue Book of Shanghai’s Social
Development]. Shanghai: Shanghai shehui kexueyuan chubanshe.
Zhang Houyi. 2004. “Jinru xin shiqi de Zhongguo siying qiyezhu jiceng” [Chinese
Private Entrepreneurs Enter a New Era], in Ru Xin, et al. (eds.), 2004
nian: Zhongguo shehui xingshi fenxi yu yuce [Blue Book of China’s Society
2004: Analysis and Forecast of China’s Social Development]. Beijing: Shehui
kexue wenxian chubanshe.
Zhang, Li. 2001. Strangers in the City: Reconfigurations of Space, Power and Social
Networks within China’s Floating Population. Stanford, CA: Stanford University
Press.
Zhang, Weimin, Li Xiru, and Cui Hongyan. “China’s Inter-census Survey in 2005,” at
http://www.ancsdaap.org/cencon2005/Papers/China/China.Zhang.Weimin.etal.
pdf., accessed August 2009.
Zhejiang nianjian. 2002 [Zhejiang Yearbook 2002]. Hangzhou: Zhejiang nianjian
bianweihui, 2002.
Zheng, Yongnian. 2008. Technological Empowerment: The Internet, State, and
Society in China. Stanford, CA: Stanford University Press.
2004. Will China Become Democratic? Elite, Class, and Regime Transition.
Singapore: Eastern Universities Press.
Zhong, Yang and Jie Chen. 2002. “To Vote or Not to Vote: An Analysis of Peasants’
Participation in Chinese Village Elections.” Comparative Political Studies, vol.
35, no. 6: 686–712.
Zhong, Yang, Jie Chen, and John M. Scheb. 1998. “Mass Political Culture in
Beijing: Findings from Two Public Opinion Surveys.” Asian Survey, vol. 38, no.
8: 763–783.
Zhongguo laodong tongji nianjian [China Labor Statistical Yearbook]. Various Years.
Beijing: Zhongguo tongji chubanshe.
Zhou, Yongming. 2005a. “Living on the Cyber Border.” Current Anthropology, vol.
46, no. 5: 779–803.
2005b. “Informed Nationalism: Military Websites in Chinese Cyberspace.” Journal
of Contemporary China, vol. 14, no. 44: 543–562.
Zhu, Jian-Hua. 1996. “‘I Don’t Know’ in Public Opinion Surveys in China: Individual
and Contextual Causes of Item Non-Response.” Journal of Contemporary China,
vol. 5, no. 12: 223–244.
Zong Fengming. 2008. Zhao Ziyang: Ruanjin zhong de tanhua [Zhao Ziyang: Captive
Conversations]. Hong Kong: Kaifang chubanshe.
Zweig, David. 1999. “Undemocratic Capitalism: China and the Limits of Economism.”
National Interest, no. 56: 63–72.
Zweig, David and Siu Fung Chung. 2007. “Elections, Democratic Values, and
Economic Development in Rural China.” Journal of Contemporary China, vol.
16, no. 50: 25–45.
Index

archival research, 18, 33–6 capitalists. See also private entrepreneurs

gaining access, 15, 20–21, 30, 36–7, Carlson, Allen, 83
38 case selection, 168, 169, 202–3
intra-disciplinary debates, 37 case-study research, 162–77, 249
study of military dependents, single case studies, 173
45–49 causal mechanisms, 76, 86, 143,
study of patriotism and nationalism, 173–4, 249
41–5, 49–50 causal process observations, 155.
types, 38–41 See also causal mechanisms
Archives Law, 20–1 causality, 69, 76, 85, 130, 132, 151
area studies, 4, 269, 274 Central Committee
and survey research, 198 Central Committee Database, 53–57
authoritarian regimes changes in characteristics over time,
data quality, 15, 30, 249 60–65
elite studies, 51–3 factional influence, 65–7
information collection by Chen, Calvin, 154
government, 23–4 Chen, Jie, 194–6
necessity of approval from Chinese Chen, Yun, 53–4, 62, 65, 67
authorities, 130 China Institute of Contemporary
necessity of collaboration with International Relations, 94
Chinese partners, 30, 87, 152, China Institute of International
207, 248, 252, 255–7 Studies, 94
research in, 7–8, 152, 161, 250 China Military Online, 92
China National Knowledge
Barnett, A. Doak, 88, 267 Infrastructure (CNKI) database,
Beijing Area Study (BAS), 52, 71–3, 83, 103, 112, 272
135–8, 181, 190, 219–35, 236–45 China Net, 91, 99
Belousov, Konstantin, 250, 257 China studies and the political science
Bernstein, Thomas, 165 discipline, 273–7
blogs, 100, 104–05 “China threat”, 83–4
bulletin boards (BBs), 98–9 Chinese Academy of Social Sciences
bureaucratic politics, 17, 24–6, 205, (CASS), 93, 152, 183
210–12 researchers in Wenzhou, 135–8

311
312 Index

Chinese Communist Party (CCP), contrasted with site-intensive

201–2, 204–7, 209–10, 214–15, methods, 148
See also Central Committee and institution-building, 138–43
Chinese politics and labor politics, 134–8
and history as a discipline, 33–5 and multiple methods research, 131
and the comparative politics subfield, event analysis, 16, 18, 22
4–7 event catalogs, 18–19
Chinese university Web sites, 95 everyday interactions, 41, 45, 132, 237,
coding, 51, 54–60, 67, 86, 116, 121, 247
125 experimental methods, 69–70, 83–7
collective self-esteem (CSE) scale, 70, Chinese national identity, 76–83
77 “face” and foreign policy, 69–70
Computer aided text analysis (CATA), exploratory factor analysis, 78, 85
122, 125. See also Yoshikoder external validity. See generalizability
concepts, 49, 153, 156, 174, 218, 261
content analysis, 18, 102–3, 107–10, “face”, 86
116–17, 122–5 Chinese foreign policy, 70–6
drawing samples, 113–16 Fairbank, John K., 266
electronic sources, 110–13 Fenno, Richard F., Jr., 147, 149–50,
examples, 117–22 153–4, 156, 157, 159
source biases, 124 flexible interviewing.
contentious politics, 15–18, 20, 27, 30 See conversational interviewing
conversational interviewing, 16, 260–3 Foreign Broadcast Information Service
county gazetteers, 20, 31, 36 (FBIS), 107, 267
Cultural Revolution, 34, 48, 52, 270 foreign policy, 84
effects on research, 107, 268–9 Friedman, Edward, 123–4
leadership changes, 53, 61, 64
cybermedia, 90–92 Geertz, Clifford, 41, 129, 133
generalizability, 87, 174, 176, 177, 217,
data 218, 242
accessibility, 15, 151 as a goal, 37, 109, 125, 130, 146,
limitations, 274–5 173–4, 217, 248
multiple types, 17, 209 vs. data quality, 87, 248, 254–5,
quality, 6, 31, 151, 210, 248, 254–5 263–4
democratization, 193, 200, 201, 203, Global Positioning System (GPS), 190,
214, 217 223, 244
Deng, Xiaoping, 61–2, 65, 67, 236, 270 governmental materials. See also state-
generated data
EastView. See China National Great Leap Forward, 22, 52, 270
Knowledge Infrastructure (CNKI) guanxi. See personal connections
database
electronic aids. See Computer aided Heimer, Maria, 162
text analysis (CATA) Home Style, 147, 157
elites, focus on, 41, 43–4, 50, 132, 203 Hu, Yaobang, 51, 65–7
epistemology, 129, 130, 148 Hua, Guofeng, 51, 66–7, 270
ethnography, 18, 129–34, 143–4, 247, hukou, inaccuracy due to migration,
See also site-intensive methods 219, 220–22, 235, 243
contrasted with more deductive human subjects 73, 209.
approaches, 136–8, 143 See also respondents
Index 313

hypotheses local people’s congresses, 214–5, 217

generation, 153, 174 elections, 194–6
testing, 130–1, 137, 154, 174 Long March, 56, 60, 62–4
Lü, Xiaobo, 165
information
increasing availability, 2, 88, MacFarquhar, Roderick, 34
107–10, 123, 272–3 Mao’s Last Revolution, 34
information dependence, 24–5 Maryland Study, 110, 113–15
information distortion, 23, 24–6, May Fourth generation, 61–2
30 McCarthy, Joseph, 266
lack of, 2, 88, 107 measurement, 83, 85
usefulness of newly available Measures for the Administration of
materials, 89 Foreign-Affiliated Surveys, 183
informed consent, 184 mechanism of causation. See causal
institutional diffusion, 227–35 mechanisms
Institutionalization of Legal Reforms military dependents, 35, 46–49
in China (ILRC), 227–32 Ministry of Foreign Affairs (MFA), 90,
Interim Measures for Administration 92–4, 99, 101, 104
of Foreign-Related Social Survey Web site, 93
Activities, 184, 250 mixed methods. See multiple methods
Internal Review Board (IRB), 87, 209 modernization theory, 69, 200
internal validity, 87, 254–5 multiple methods, 18, 147, 155, 217,
internationalism, 77, 80–1 253–4
Internet, 88–9, 105–6, 272
foreign policy materials, 89–100 National Bureau of Statistics (NBS),
uses of Internet data, 100–5 183, 184, 220, 250
interviews, 16, 268, See also national identity, 69, 76–7, 78–80, 83
standardized interviewing, nationalism, 41, 43–4, 49, 69, 76–84
See also conversational “threshold problem”, 83–4
interviewing neighborhood organizations, 152–3,
elite interviews, 102, 159, 271–2 159
and ethnography, 135, 158 newspaper databases, 112–13
and multiple methods, 147, 155
political sensitivity, 261–2 O’Brien, Kevin J., 133, 156
response rate, 243 observable implications, 154, 249
suggestions, 239, 265 observations, 68, 130, 182
and survey research, 223, 245 Oksenberg, Michel, 123–4
invalid part-to-whole mappings, 164–5, organizational change, 138–43
174, 176 outliers, 131, 162, 165, 169, 176
IR China Web site, 96–7
Park, Albert, 256
Jiang, Zemin, 171–2 parsimony, 38, 130, 132, 137
Johnston, Alastair Iain, 85, 102, 190, participant observation. 138–9, 271.
238 See also site-intensive methods
contrasted with site-intensive
labor politics, 134–43 methods, 149
laid-off workers, 167–73 party-affiliated mass organizations,
Lattimore, Owen, 266 138–43
Likert scale, 72, 77, 83 path analysis, 81–3
314 Index

patriotism, 41–6, 49, 69, 70, 76–84 protecting safety of, 134, 252–3
blind patriotism, 76, 77, 80–1 Ross, Robert S., 85
Peking University, 76, 183, 190, 207, “ruining a solider marriage”, 46–9
236, 238
People’s Daily, 90–1, 112, 114 sample size, 114–16, 217
People’s Liberation Army (PLA), 35, 45, sampling, 132, 157, 203–6
49, 92. See also military dependents biased samples, 221
People’s Net, 90–1 constructed-week sampling, 113–16,
People’s University, 191 123
personal connections, 30, 36, 166, 228 efficient samples, 123
petitioning system, 16–17, 25–6, 27–32 final (spatial) sampling units (FSUs),
political culture, 69, 189 223, 229
political participation, 189, 194 multistage stratified samples, 222, 231
private entrepreneurs, 200–01, 210–18 nonrandom samples, 87, 113
data collection, 207–10 primary sampling units (PSUs), 242
survey design, 201–6 probability proportionate to size
process tracing, 130, 155 (PPS), 222–3, 242
provincial yearbooks, 20 probability samples, 181–3, 188–9,
psychological measures. 242–4, 255
See experimental methods purposive selection, 203–4, 217, 248,
Pye, Lucian, 70 254
random samples, 87, 109, 203–6,
qualitative methods, 18, 116, 124–5, 248, 254
147–8, 154. See also content representative samples, 112, 190, 236
analysis. See also case study secondary sampling units (SSUs), 242
research. See also interviews. spatial probability samples, 219–35
See also archival research. systematic random sampling, 113, 115
See also site-intensive methods. satellite imagery, 225–7
See also participant observation. Scott, James, 23, 131, 146–7, 149–50,
See also ethnography 154, 157
quantitative methods, 117, 273, security studies, materialist vs.
See also content analysis. symbolic, 71–2, 74
See also survey research. selection bias, 111–12, 132
See also statistical methods. sensitive research topics, 7–8, 15, 207,
questionnaires, 87, 207–9, 237–42 208, 218, 240, 241, 250–1
Chinese regulations, 184–5 framing, 36
information accessibility, 16, 20, 30,
red capitalists, 194, 205, 210, 215, 217 36–7
reform era, 62, 65, 270–1 protecting Chinese collaborators, 256
replicability, 87, 125, 130 regional variation, 253–4
Research Center for Contemporary socially embedded survey research,
China (RCCC), 190, 207, 236, 255–63
238, 244 vs. taboo research topics, 251
research design, relation to research Shanghai Institutes of International
goals, 87 Studies, 94
respondents, 73 Shi, Tianjian, 188–9, 194–6
building trust, 150–3, 158–61, 255, site-intensive methods, 129–34, 138–9,
258–63 143–4. See also participant
eliciting responses, 18 observation, See also ethnography
Index 315

contrasted with interviews, 159 face-to-face interviewing, 185, 237,

contributions to theory, 153 244–5
definition of term, 148 individual-level attitudinal surveys,
optimizing validity, 158–60 263–4
trade-offs between breadth and longitudinal, 236
depth, 146, 156–7 political sensitivity of surveys
when to use, 150–3 in non-democratic regimes,
Snyder, Richard, 164 246–65
“soaking and poking”, 149, 153 regulatory regime, 183–5
social networks, 142–3, 228–31, 232 socially embedded survey research,
standardized interviewing, 260–3 255–63
state-generated data. See also archival village-level, 247
research
accessibility, 15, 20 Technorati, 104–5
availability, 30 TECN Academic Net, 97
contrasted with newspaper data, 18, thick description, 10, 129, 131
19, 22 Tilly, Charles, 16, 18
information collection and time-series analysis, 67, 112, 206
processing, 28–30 Touch Graph, 104
published materials, 19–20 trust, 276. See also respondents,
quality of statistical information, building trust
29, 31 collaborators, 255–8
reliability, 15, 22–7, 30–1 inter-personal and system-based,
upstream sources, 31–2 227–29
state-owned enterprises (SOEs), 167–9, Tsai, Kellee, 165, 194, 198, 201, 204,
170–3 215
statistical methods Tsai, Lily L., 155, 198
challenges, 86
overemphasis on statistical University of Michigan, 191, 237
significance, 86
subnational comparative analysis. variation, 146
See also case-study research individual variation, 203–6
application of, 163, 176–7 regional variation, 6, 68, 202–3,
contrasted with large-N quantitative 261
analysis, 174–5 temporal variation, 206
contrasted with single-city case
studies, 173–4 Walder, Andrew, 123–4
identifying antecedent conditions, Weapons of the Weak, 147, 154, 156
163, 166, 174, 176 Whiting, Allen, 88, 123
survey research. See also Beijing Area Wood, Elisabeth, 253
Study (BAS)
access to gated communities, 235, xiahai red capitalists, 205, 214, 217
245 xinfang. See petitioning system
advantages and limitations, 192, 218 Xinhua Net, 91
contrasted with interviews and Xinhua News Agency, 91, 117–20
archival research, 218
contrasted with qualitative research, Yoshikoder, 4, 103, 116, 119–22, 125
196–8
conversational interviewing, 263–4 Zhong, Yang, 194–96