Political Science:

The State of the Discipline

Edz'ted by Ada W. Fz'nifter

1983

The American Political Science Association

1527 New Hampshire Avenue, N.W.
Washington, D.C. 20036
Copyright © 1983 by the American Political Science Association, 1527 New Hamp-
shire Avenue, N.W., Washington, D.C. 20036 under International and Pan-American
Copyright Conventions. All rights reserved. No part of this book may be reproduced in
any form or by any means without permission in writing from the publisher. The views
expressed are those of the authors and not those of the American Political Science
Association.

Library of Congress Cataloging in Publz'cation Data

Main entry under title:

Political science.
Includes bibliographies and index.
1. Political science-Methodology- Addresses, essays, lectures. I. Finifter,
Ada W., 1938- II . American Political Science Association.
JA73.P658 1983 320'.01'8 83-72628

ISBN 0-915 654-5 7-1

ISBN 0-915654-58-X (pbk.)
 Political Science:
The State of the Discipline
Edz'ted by Ada W. Fz'nifter

TABLE OF CONTENTS

Preface v

POLITICAL SCIENCE:
THE DISCIPLINE AND ITS SCOPE AND THEORY

1 Political Theory: The Evolution of a Sub-Field

john G. Gunnell 3
2 Political Theory and the Art of Heresthetics
William H. Riker 47
8 Toward Theories of Data: The State of Political Methodology
Christopher H. Achen 69
4 Self Portrait: Profile of Political Scientists
Naomi B. Lynn 95

AMERICAN POLITICAL PROCESSES

AND POLICYMAKING

5 The Scholarly Commitment to Parties

Leon D. Epstein 127
6 The Forest for the Trees: Blazing Trails for Congressional
Research
Leroy N. Rieselbach 155
7 Judicial Politics: Still a Distinctive Fiela
Lawrence Baum 189
8 Public Policy Analysis: Some Recent Developments and Current
Problems
Susan B. Hansen 217
9 Federalism: The Challenge of Conflicting Theories and
Contemporary Practice
David R. Beam, Timothy]. Conlan, and David B. Walker 247

COMPARATIVE POLITICAL PROCESSES

AND POLICYMAKING

10 Comparative Public Policy: An Assessment

M. Donald Hancock 283
 v

Preface
When I was appointed Program Chairperson of the 1982 Annual
Meeting of the American Political Science Association by APSA President-
Elect Seymour Martin Lipset, I chose as a guiding theme for the meetings,
"the state of the discipline." Nineteen eighty-two seemed a good year for
some stock-taking. The heady days of the growth of the social sciences in the
1960s were only a faint memory for many of the current members of the pro-
fession. Some of the attempts at applying social science theory to public policy
problems had produced unclear, and occasionally questionable , results. The
social sciences were under great pressure to prove their worth in the more
budget-conscious 1970s, and that pressure had increased under the current
administration in Washington. It seemed a good time to review where we
were and where we might be going. Nevertheless, despite my own conviction
that a review of the intellectual developments in the field over the past twenty
to twenty-five years was due, I had little inkling that many scholars were feel-
ing a similar need.
The then newly appointed Executive Director of the American Political
Science Association, Thomas E. Mann, responded enthusiastically to my
choice of a theme and suggested that we capitalize on the attention the sub-
ject would be receiving at the convention by collecting a series of articles on
the "state of the discipline" in an APSA publication. This idea was supported
by the APSA Council, and I agreed to organize and guide the effort. If the
volume is successful, a continuing series of this type may be authorized by the
Council.
The overall goal for this volume is to begin to fill the continuing need for
a frequently published overview of political science research. It is intended to
be used in tea'c hing and for research purposes, as well as by foundation and
government officials who need regularly updated information on current
research in the discipline. I hope that the book will be useful in introducing
graduate students to the problems about which political scientists are current-
ly concerned, and that it will help political scientists learn more about each
other's work. The volume should also be useful to foundation and government
officials, as they consider research opportunities in the discipline, and to
people working in government and other applied fields , who wonder what
political scientists can contribute to their daily concerns.
Since this book grew out of the theme of the 1982 Annual Meeting, it was
designed in conjunction with the program for that convention. In each sec-
tion of the program, a paper was commissioned to review the state of research
and future directions of that subfield. Since most of the subfields include a
variety of approaches to their subject matter, and cover enormous bodies of
literature, the authors were given latitude to define their scope of coverage in
a way that they felt was most meaningful to their own research interests. In
some cases, an entire subfield (as defined by the program section title) is
covered; in others, the author has defined a specific part of the subfield or
some significant intellectual problem for more comprehensive coverage.
Therefore, not every area of political science is included in this first review. If
vi

the series is continued over a period of years, any important omissions of par-
ticular subfields or within particular areas will certainly be addressed.
The authors were asked to define the area being reviewed; to analyze
conceptual frameworks employed by scholars working in the area, including
classical and older approaches if they were relevant for contrast with newer
frameworks; to treat the cumulation of knowledge in the field by discussing
the major research findings; to consider how important findings have been
changed or modified as knowledge has developed; and to suggest the issues
that need to be addressed in the future.
Each of the papers was presented at a "theme" panel at the 1982 Annual
Meeting. A large number of colleagues throughout the discipline contributed
by acting as discussants on these panels and commenting on the original ver-
sions of the papers. Most of the discussants also provided very helpful written
comments on the papers. Following the meetings, the papers were reviewed
by anonymous peer reviewers. Other colleagues provided informal consulta-
tion and advice. All of the papers benefited from this generous peer review.
The authors of the articles appearing in this volume are, of course, the
primary sources of any insight and wisdom contained in it. Many of them also
served as Section Chairs for the 1982 Annual Meeting of the APSA. In other
cases, the Section Chairs contributed to the volume by suggesting colleagues
especially well suited to prepare the "state of the discipline" papers, by
establishing panels that focused on important issues in the discipline, and by
their general encouragement and support.
The members of my own department at Michigan State University
proved to be an uncommonly congenial and helpful group. David W. Rohde,
the Chairman, was generous in his cooperation and support for this project.
Marty Lipset, with his broad view of the development of the social sciences,
supported the project from its inception and worked with me in its initial
stages. Tom Mann deserves special recognition for his knowledge of the pro-
fession and his ever helpful advice and enthusiastic support. Members of the
Association are forewarned should they receive a call from him: he's a man
who can get almost anyone to say "Yes" to almost anything.
I am also very grateful to the following colleagues and scholars for their
thoughtful and valuable contributions and would like to thank each of them
for their assistance:

Joel D. Aberbach , University of Michigan

Paul R . Abramson, Michigan State University
John H . Aldrich, University of Minnesota
Eugene J. Alpert, Texas Christian University
Charles W. Anderson , University of Wisconsin, Madison
Thomas J . Anton, University of Michigan
W. Lance Bennett, University of Washington
Richard A. Brody, Stanford University
Bruce Bueno de Mesquita, University of Rochester
Naomi Caiden, California State College, San Bemadino
James A. Caporaso, University of Denver
Jonathan D. Casper, University of Illinois
Aag-e Clausen, Ohio State University
 vu

Beverly B. Cook, University of Wisconsin, Milwaukee

Philip Coulter, University of Alabama
William J. Crotty, Northwestern University
Raymond Duvall, University of Minnesota
David Easton, University of Chicago
George C. Edwards III, Texas A&M University
Daniel Elazar, Temple University
Barbara Farah, University of Michigan
Bernard M. Finifter, Michigan State University
David D. Finley, The Colorado College
Morris P. Fiorina, California Institute of Technology
Alexander L. George, Stanford University
Fred I. Greenstein, Princeton University
Arnold]. Heidenheimer, Washington University
Raymond F. Hopkins, Swarthmore College
Samuel Huntington, Harvard University
Robert W . Jackman, Michigan State University
Norman Jacobson, University of California, Berkeley
Dorothy B. James, American University
M. Kent Jennings, University of California, Santa Barbara
Charles A. Johnson, Texas A&M University
John H. Kessel, Ohio State University
Mae C. King-Akesode, University of Benin
Jack Knott, Michigan State University
Gerald H. Kramer, California Institute of Technology
Everett C. Ladd, University of Connecticut
Joseph LaPalombara, Yale University
Richard H. Leach, Duke University
Arend Lijphart, University of California, San Diego
Catherine Lovell, University of California, Riverside
Gregory B. Markus, University of Michigan
David R. Mayhew, Yale University
Arthur Melzer, Michigan State University
Michael L. Mezey, DePaul University
Gary Miller, Michigan State University
Nicholas R. Miller, University of Maryland, Baltimore County
Warren E. Miller, Arizona State University
J. Donald Moon, Wesleyan University
Richard P . Nathan, Princeton University
Betty A. Nesvold, San Diego State University
Robert North, Stanford University
Charles W. Ostrom , Jr., Michigan State University
Elinor Ostrom, Indiana University
Samuel C. Patterson, University of Iowa
Charles Press, Michigan State University
Jewel Prestage, Southern University and A&M College
Michael Preston, University of Illinois
James W. Prothro, University of North Carolina , Chapel Hill
Robert Putnam, Harvard University
George Rabinowitz, University of North Carolina, Chapel Hill
Austin Ranney, American Enterprise Institute
Bert A. Rockman, University of Pittsburgh
Ronald F. Rogowski, University of California, Los Angeles
 Vlll

Richard Rosecrance, Cornell University

Virginia Sapiro, University of Wisconsin, Madison
Allen Schick, University of Maryland
Philip Schrodt, Northwestern University
Kenneth A. Shepsle, Washington University
John W. Sloan, University of Houston
Frank]. Sorauf, University of Minnesota
Harold]. Spaeth, Michigan State University
James Stimson, Florida State University
John L. Sullivan, University of Minnesota
Susette M. Talarico, University of Georgia
Norman C. Thomas, University of Cincinnati
Kathleen Toth, University of Texas, San Antonio
Sidney Verba, Harvard University
Jerry Weinberger, Michigan State University
Robert Weissberg, University of Illinois, Urbana
Aaron Wildavsky, University of California, Berkeley
Raymond E. Wolfinger, University of California, Berkeley
Deil S. Wright, University of North Carolina, Chapel Hill
Gerald C. Wright, Jr., Indiana University
Dina Zinnes, University of Illinois

The members of the profession indicated their interest in an intellectual

review of the discipline by attending the 1982 "state of the discipline" theme
panels in unusually large numbers. Many of the comments I received about
the Denver convention related to the special attraction and the utility of hav-
ing this kind of review take place in a forum that provided an opportunity for
discussion and feedback from a diverse representation of political scientists. I
hope that the publication of these papers will contribute to the continuation
of this dynamic process.

Ada W. Finifter
East Lansing, Michigan
June 19, 1983
 POLITICAL SCIENCE:
THE DISCIPLINE AND ITS
SCOPE AND THEORY
 I
Political Theory:
The Evolution of a Sub-Field
John G. Gunnell

It is necessary to make a distinction between Political Theory as a sub-

field of the discipline of political science (PT) and political theory as a more
general interdisciplinary body of literature, activity, and intellectual com-
munity (pt). It is also necessary to distinguish between those aspects of PT
that are closely tied to pt and those that are more directly related to various
research programs in political science. This is not to suggest that there are not
important relationships and points of overlap and intersection both between
PT and pt and between the elements within PT, but locating the boundaries
is important for an analysis of political theory as a whole and for an under-
standing of the controversies that have animated it. For example, according
to some political scientists, one of the problems has been to make Theory rele-
vant to the practice of the discipline of political science rather than an outpost
of history, ethics, and other fields that contribute to the constitution of pt. On
the other hand, some political theorists, while professionally and institu-
tionally situated within political science, identify themselves intellectually
much more closely with pt.
While there are recent works that attempt to characterize and survey the
present condition of political theory and speculate about what it might or
should be (e.g., Nelson, Ed., 1983), less detailed attention has been devoted
to the "odyssey" of political theory in American political science (Toth, 1982).
Most attempts at the latter have been quite general characterizations in the
service of some critique or apology. My basic approach will be archaeological.
I hesitate to say historical, since, in many respects, this is more an excavation,
or the presentation of the results of such an endeavor, than what many might
consider conventional history. I began at the surface and dug downward,
even though I am presenting my report in standard chronological order.
However much my personal concerns may be reflected in the course of sifting
through this dense material, I am more concerned with uncovering previous
sites of political theory than with evaluating their contents and achievements.
One characteristic of the subject matter is that the termini are poorly
defined. Rather than reaching the "beginning" of Political Theory as a sub-
field, I stopped when the traces seemed too amorphous to warrant further
dredging. On the other end, the deposits build up so rapidly that any definite

3
4 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD
conclusion is almost immediately obsolescent; I can do little more than in-
dicate what an initial survey of the surface yielded and what prompted my
particular angle of entry.
The state of both PT and pt in the 1980s is one of dispersion, and I will
risk the claim that this will remain its dominant characteristic for some time.
To some extent this is the result of developments within political science as a
whole, but it is also a reflection and consequence of trends within the sub-
field and its relationship to pt. PT has broken up and scattered in different
directions and adopted various mediums of expression, but the spectrum that
can be discerned is not the diffraction of any basic substantive issue or set of
concerns. Dispersion is less a symptom than the very condition of the field.
This condition grew out of the events of the 1960s, became clearly manifest in
the 1970s, and defined the state of the field by the beginning of the 1980s.
Some might suggest that this diffusion is basically the result of increased
specialization in political science as a whole as well as greater interdisciplinary
involvement. Others would argue that what is happening is, in the still
popular Kuhnian terminology, a paradigm breakdown or transformation.
Each of these claims may be correct in some respects, or at least convey
something about certain aspects of the situation, but rather than pursuing a
particular causal explanation, I will attempt to reconstruct the basic contours
of the evolution of the sub-field of Political Theory. But, first, it is appro-
priate to be a bit more specific about the character of this putative dispersion.
In recent years, there has been a great deal of enthusiasm about the
future of political theory. This is in sharp contrast to some of the attitudes
that characterized the 1950s and 1960s when, in some quarters, political
theory, at least in the form of political philosophy, was "thought to be on the
verge of extinction" (Richter, 1980, p. 6). For a long time thereafter, it was
considered to be an endangered species. While political theory, conceived as
part of the scientific study of politics, was vigorously propagated, other modes
of theory seemed to be on the decline. It now seems that, even among many of
its former advocates and supporters, the commitment to "scientific" theory is
somewhat muted, and many would argue that there is now, and should con-
tinue to be, greater complementarity between empirical and normative
theory. For example, the idea of political science as the study of public policy,
which has provided a self-image for the discipline in recent years, might imply
a closer relationship between PT and pt or between political science and
political philosophy. But it has also been claimed that "the great vitality in
the field of political theory during the last fifteen or twenty years" is the result
of a divergence between PT and pt and the fact that much of political theory
has "turned away from" and become "indifferent to much of academic
political science." Both "may prosper, but not interdependently" (Kateb,
1977, pp. 135, 136).
It will be necessary to consider such positions more carefully, but, for the
moment, it is enough to note that there is a continuing ambivalence about the
relationship between political theory and political science that has persisted
since the beginning of the Post-World War II period. There are sentiments
for secession and exile as well as for unification and integration. Today, it is
clear that at least in pt, and much of the literature of PT that is most closely
 john G. Gunnell I 5

allied with it, there is a sense of movement whose source is in some respects
quite easy to identify but whose significance and direction are more nebulous.
There is a widespread belief that during the 1970s there was an "upswell
of political and social theorizing" and that "political philosophy" now
"obviously flourishes, all over the English-speaking world and outside it too"
(Laslett & Fishkin, Eds., 1979, pp. 2, 5). This alleged upturn is usually linked
with the publication of John Rawls's A Theory ofjustice (1971) and related
works of a similar genre such as Robert Nozick's Anarchy, State and Utopia
(1974); the popularity of so-called Critical Theory and the work of individuals
such as Jurgen Habermas; and the variety of critiques, summaries, and com-
mentaries that grew up around this material and the pursuit of similar
themes. Various explanations were advanced to account for this "upsurge in
creative political theory" (Freeman & Robertson, 1980, p. 11) such as the
release of moral philosophy from the grip of positivism and the shock of
political events in the late 1960s, but it was generally acknowledged that
political theory was born again. This sense of having overcome the past was
quite evident by the end of the 1970s, but the present constitution and future
form of pt is far from evident. The situation in PT is no clearer. Since most of
the developments in pt are reflected there, its literature is equally diverse.
Dispersion does not mean that the universe of political theory cannot be
charted, but it is difficult to discern some overall form. Collections purport·
ing to represent the "frontiers" of political theory leave one in doubt both
about the territory that these margins circumscribe and the nature of the ter-
rain that is being penetrated. It is suggested that there is a "return to political
theory in the grand manner" but that it is now "pluralist" to an un-
precedented degree and can only be characterized by looking at "what
political theorists do" (Freeman & Robertson, 1980, pp. 11, 1). An attempt
by the journal Political Theory to project the "Prospects and Topics" of
"political theory in the 1980s," seemed to reach the same conclusion, or non-
conclusion. The editor decided simply to "invite a number of thoughtful
senior colleagues . . . to get on with what they were thinking about," and
claimed that the diverse contributions were "representative of both the scope
of the field and the pushing forward of its concerns and frontiers" (Barber,
1980, p. 291).

THE EARLY YEARS: BEFORE 1899

The idea of political theory as a distinct kind of activity, vocation, and
product is of relatively recent origin. The concept of political theory in its
contemporary sense or senses has not only been largely a creation of the sub-
field of Political Theory in political science but a convention that can be at·
tributed to the debates about the character and status of political theory that
began in the 1940s and 1950s. Notions of theory that emerged at that time
were read backward into the past and projected into the future. Even after
the establishment of the American Political Science Association and the con-
stitution of Political Theory as an official sub-field, political theory and/or
political philosophy was basically a category that referred to certain types of
 6 I POLITICAL THEORY : THE EVOLUTION OF A SUB-FIELD
claims and kinds of literature, various elements or functions in politics, and
some reflections on the study of politics.
Why, exactly, the discipline of political sdence was officially born in
1903 with a sub-field called Political Theory and, precisely, how it came to
receive this name is difficult to say. In part, it probably reflected the tradi-
tional theory/ practice distinction in nineteenth century philosophy and the
idea of social science as being concerned with the theory of the state. But to
some extent, political science at the time of the creation of the APSA was less
a distinct discipline than a holding company for a variety of endeavors that
were in various ways related, but no longer easily resided in other disciplines.
Political Theory was in part such a field.
Up through the late 1700s, the study of ethics and moral philosophy had
included politics and political philosophy. When Francis Lieber, "the begin-
ner in the United States of the systematic study of politics" (Haddow, 1939, p.
139), was appointed in 1857 as Professor of History and Political Economy at
Columbia, he indicated his intention to teach and lecture on Political
Philosophy which was concerned with the theory of the state and with
political ethics. After political science became a distinct discipline in many
universities after the Civil War, political theory began to find an even more
definite place, and by the early 1890s, a "History of Political Theories" course
appeared at Harvard (Haddow, 1939, p. 175). Political scientists, looking for
their ancestors, would find them in the classics of moral philosophy that were
concerned with politics, and thus began the idea that there was a tradition of
political thought to which political scientists would understand themselves to
belong.
In 1876, John Burgess succeeded Lieber at Columbia where he estab-
lished the graduate school of political science which opened in 1880 and by
1891 consisted of three departments including the department of History and
Political Philosophy. Courses dealing with the history of political theory and
the philosophy of the state appeared here and at other major universities. The
influence of Bluntschli's Theory of the State was evident as well as Yale's
Theodore D. Woolsey's widely read text on Political Science, or, the State
Theoretically and Practically Consi'dered (1878). W. W. Willoughby, one of
the most important figures in the early discipline of political science, par-
ticularly in Political Theory, was offering courses in both areas, first at Stan-
ford in 1894 and then at Johns Hopkins in 1895; he published An Examina-
tion of the Nature of the State; A Study in Poli'ti'cal Philosophy in 1896. With
the establishment of the APSA, the general context for the development of
PT took determinate form.

THE BEGINNINGS OF A DISCIPLINE: 1900-1919

During the early years of the discipline, political theory was still viewed
more as a subject matter than a mode of analysis. Munroe Smith had stated in
1886, in the first issue of Political Science Quarterly, that the "domain of
political science" was the historical and comparative study of the state, and
this included what people had thought about it. To write the history of
political theory was at once to write about the history of democratic institu-
 john G. Gunnell I 7

tions and about the development of political science which from the begin-
ning was already traced through the canon of classic texts from the Greeks to
modem political science. The influence of the evolutionary theories of Comte
and Spencer as well as the Hegelian analysis of the state, with its subjective
and objective division, was strong and provided the basic intellectual context
in which both political science and the study of the history of political theories
emerged.
The APSA was created for the purpose of "advancing the scientific study
of politics in the United States," and six sub-fields, including Political
Theory, were established with corresponding committees. The first Political
Theory committee consisted of Willoughby, Charles Merriam, and William
Dunning. In his presidential address to the APSA in 1904, Frank Goodnow
said that political science was the study of the state and the "realization of the
State will," and he viewed political theory as a special discipline concerned
with the authorities that express that will. Furthermore, he suggested,
"however contemptuous may be one's belief in the practical value of the study
of political theory, it is none the less true that every governmental system is
based on some more or less well-defined political theory" (1905 , pp. 37-38).
Political theory was understood in one way or another to be concerned with
ideas in and about politics. Willoughby's 1904 article on the newly formed
APSA claimed that the field consisted of three basic parts, and the first was
the "province of political theory and philosophy" which aimed at "the analysis
and exact definition of the concepts employed in political thinking" and a
consideration of the "nature of the state" (p. l18). Most of the material
published in Political Theory through 1920 was historical in the sense of Dun-
ning's paradigmatic treatises on the "history of political theories" (1902, 1905,
1920). Although there was some precedent and parallel in European
literature, the history of political theory was a distinctly American genre
(Gunnell, 1979).
For Dunning, the study of political theory was the study of transforma-
tions in political consciousness and a way of scientifically grasping the
dynamic character of politics that resulted from the interaction of institutions
and ideas. Not only were science and history understood as integrally related,
but the idea of political science as a practical science was a regulative assump-
tion. This was true both in the older sense of providing political or citizen
education and in the somewhat newer sense of a concern with participating in
social reform and modes of social control that characterized most of the social
sciences of this period. Between 1900-1920, the progressivist ideology con-
sistently found expression in social science which was viewed by many as the
link between knowledge and politics. Henry Jones Ford (1906) argued that
political science must be universal in its scope, put on "an objective basis, "
and "experience the reconstruction which the general body of science has
undergone at the hands of inductive philosophy, " but the purpose was to
"bring political science to a position of authority as regards practical politics"
(pp. 203, 206).
After 1910, political theory as a specific subject was increasingly
neglected as political scientists focused on practical issues in domestic and in-
ternational politics, but the views of the formative period persisted. R. G.
8 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD

Gettell, who wrote one of the most popular political science texts (1910) and
one of the major works in the history of political theory (1924), offered an ex-
tended analysis of the nature and scope of political theory which reflected the
dominant view in the discipline. Political theory in general was understood as
reflection on the institutional state or the "objective" phase of the state which
grew out of the need to cope with the environment. Thus political theory was
not ultimately true but "relative in nature" and both cause and effect in that
it simultaneously influenced and mirrored politics (1914, pp. 48, 50). At the
same time, he noted that a fundamental and revolutionary change was taking
place in political theory whereby it was being transformed from a deductive,
normative, and idealist enterprise to an inductive and realistic one concerned
with observation, classification, description, and generalization. Gettell
distinguished three distinct but related elements of political theory: historical,
analytical or descriptive, and applied (pp. 52, 54).

THE FIRST REVOLUTION IN POLITICAL THEORY:

1920-1929
In 1921, Merriam, in his presidential address to the APSA, announced
that while he had intended to survey the field over the past four decades, he
had decided, instead, to speak to the more pressing problem of the
"reconstruction of the methods of political study" (p. 174). This would re-
quire that the "theory of politics" be transformed so as to reflect the substan-
tive modem doctrine that "political ideas and systems ... are the by-products
of environment" and the methodological advances that had been made possi-
ble by "statistical observation" and the more accurate measurement of "facts
and forces" (p. 174). Merriam's particular concern was with developing a
theoretical "medium" (p. 175) for selecting and classifying the mass of facts
that social science was accumulating and with releasing political theory from
ideology or the "service of class and race and group" (p. 178). For Merriam,
the basic aim was not pure science but rather the "cross fertilization of politics
with science" and the more effective control and organization of practical
matters in domestic and international politics. The goal of political science
was to "interpret and explain and measurably control ... the forces of human
nature" (p. 183). But Merriam also believed that before the "processes of
social and political control" could be grasped, it was necessary to have a "bet-
ter organization of our political research." It would be impossible to con-
tribute to "political prudence" if there were "anarchy in social science, or
chaos in the theory of political order" (pp. 184-185).
Merriam's claims and concerns did not represent the disciplinary consen-
sus. He was the major figure of that period in political science, and he in-
fluenced both his contemporaries and later phases of the discipline. But his
ideas did not reflect the dominant research, publications, and curriculum of
his own time. On the other hand, it is easy to mistake the degree to which
Merriam's notions of theory and science diverged from his predecessors and
contemporaries. What is striking, if compared with the later tension between
the ideas of science and history, is that although Merriam attacked the
historical-comparative approach in political science, this did not entail a con-
 john G. Gunnell I 9

frontation with the history of political theory as conducted by individuals such

as Dunning, Gettell, and C. H. Mcillwain. All of these men agreed with
regard to their assumptions about the relativity of ideas and institutions, the
evolution of political thought toward science and democracy, and the prac-
tical mission of political science.
The APSA Committee on Political Research in 1923, under Merriam's
leadership, identified social methodology as the "recent history of political
thinking" (p. 275), and Merriam stressed the need in political inquiry to draw
upon the methods of economics, statistics, history, anthropology, geography,
and psychology as a basis for the "observation and description of actual pro-
cesses of government" and for eschewing the older "a priori speculation" and
the juristic and historical/comparative approaches (pp. 281-283). Here Mer-
riam first offered his famous historical typology of the development of
political inquiry that would appear in New Aspects of Poli'tics: a priori deduc-
tive up to 1850; historical and comparative between 1850-1900; a tendency
toward observation, survey, and measurement from 1900 to the present; and
the future pointed in the direction of the "psychological treatment of politics"
(a characteristic of the work of Graham Wallas and Walter Lippmann) (p.
286) . Like later attempts to transform the discipline and distinguish what was
conceived as innovation from the burden of the past, the image of the past
was a somewhat contrived one, but the rhetorical point was clear.
In the "Report of the National Conference on the Science of Politics" in
1924 (previous conferences had been held in .1922 and 1923), it was main-
tained that "the great need of the hour is the development of a scientific tech-
nique and methodology for political science" (p. 119). But this was still
understood as serving the practical concern of providing a scientific guide to
legislation and administration which government had neither the time nor
capacity to develop. What was required was intellectual authority if political
science was to play a role and, therefore, a "fact-finding technique that will
produce an adequate basis for sound generalization" and put "political
research upon a scientific, objective basis" (pp. 120-121). Merriam argued
that "the perfection of social science is indispensable to the very preservation
of this same civilization" that created modem science.

THEORETICAL CONTINUITY: 1930-1939

Substantively, as well as methodologically, the 1930s were years of affir-
mation: affirmation of science, democracy , and their complementarity.
Political theory was also still largely a functional or analytical category, and
as a classification of literature, it referred to subjective matters, both
cognitive and ideological. This tendency gained support, and maybe some
new meaning, with the psychological concerns of Merriam and others in the
1920s, but long before and after, dissertations in Political Theory in the
APSR were listed under " Political Theory and Psychology. "
It was not until 1930 that Political Theory was offered as a distinct sec-
tion in the annual meetings of the APSA , yet by the end of the 1930s, political
theory as a category disappeared from the annual program as increasing
pressures of domestic and international affairs drew attention away from
IO I POLITICAL THEORY : THE EVOLUTION OF A SUB-FIELD
issues about the scope and method in political science. Disciplinary concern in
political theory was concentrated in two areas.
It was during the 1930s that the history of political theory fully crystal-
lized as a genre of literature, and this literature largely embodied attempts to
demonstrate and celebrate the development of liberalism and its divergence
from fascism and communism. Many textbooks with similar versions of this
history were published during the 1930s; the genre culminated in George
Sabine's A History of Poli'tical Theory (1937) which, as David Easton (1953)
has noted, "exercised deeper influence over the study of political theory in the
United States ... than any other single work" (p. 249). Writers of this period
were untroubled about a simultaneous commitment to the study of the history
of political theory and to the development of a scientific study of politics. For
proponents of scientism, such as George Catlin (1939), the two endeavors
were mutually confirming.
With respect to the trends initiated by Merriam, no one in the 1930s con-
tributed more to their perpetuation than Harold Lasswell. The commitment
to the idea of a science that could play a role in the reformation of society was
not only sustained but in some respects radicalized. Lasswell (1939, 1941) put
less emphasis on education than on the development of theories and testable
hypotheses that would expose the reality behind politics and political
ideology- largely a psychological reality- and provide the basis of a
therapeutic policy science. Some of the work of Lasswell in the 1930s would
recede as a matter of disciplinary concern, but his behavioral realism, as well
as his emphasis on science and policy, would make a lasting impression on the
aspirations of political science as a science.
Sabine (1939) suggested that political theories characteristically con-
sisted of three logically distinct kinds of propositions: factual, causal, and
valuational. He stressed the need to distinguish these aspects in both
understanding the past and analyzing contemporary claims. The question of
the status, relationship between, and priority of these aspects would occupy
both PT and the discipline as a whole for some time, but even by the end of
the 1930s, many political scientists were unhappy with what the discipline had
achieved with regard to fulfilling the vision of science.
Some saw the problem as the failure of science, and others saw it as a
failure to be adequately scientific. For the latter, who would eventually
dominate the discussion, the problem seemed increasingly to be one centered
around the issue of theory. What they perceived as a theoretical core in other
disciplines seemed to be missing in political science. By the 1950s, many in-
dividuals would again be stressing that what was required in political science
was a theoretical revolution, but the seeds of that second revolution were
sowed in this much earlier period.

THE PRELUDE TO BEHAVIORALISM: 1940-1949

The claims about theory and science that are so familiar from the 1950s
did not appear on the scene as suddenly as we are often led to assume. Ben-
jamin Lippincott (1940) argued that, for the most part, political scientists still
equated empiricism with fact collection and that this had been the case since
 john G. Gunnell I 11

the turn of the century. Political scientists still looked upon "theories or ideas
about the facts (as] not only unnecessary but positively dangerous" (p. 130).
Lippincott pointed out, however, that theories are always implicitly involved
in the selection of facts and, consequently, the biases of the discipline are
often concealed. But if theory was required, it was to be theory of a particular
sort. William Foote Whyte (1943) represented an increasingly popular point
of view when he claimed that political scientists should "leave ethics to the
philosophers and concern themselves primarily with the description and
analysis of political behavior" (p. 692).
In the 1940s there was manifested a new and focused concern with
theory, but there was, from the start, a somewhat different emphasis on just
what it was and how it should be related to political science. The pressures
associated with the early years of the war had curtailed collegial discussion of
political science and political theory, but by 1943 the Political Theory panel
of the APSA Research Committee attempted to pick up and sort out the
issues. Although no uniquely new idea of theory emerged from these discus-
sions, there was an emerging sense of a "deep cleavage" on an "ultimate issue"
which even the participants had difficulty articulating (Wilson, 1944, p. 726).
This cleavage was variously understood in terms of "philosophical" reflection
on the study of politics as opposed to "logical analysis"; the "theological ap-
proach" versus the "empirical" or increasingly popular " 'positivistic,' scien-
tific, or liberal technique of social study"; the idea of a value-free science
seeking laws of political behavior as opposed to a more evaluative set of con-
cerns; and the conflict between the "philosophy of history" and an emphasis
on "ends-means relationship" (Wilson, 1944, p . 727 , emphasis added).
Although the behavioralist attack on the history of political theory in the
1950s is often characterized as an "offensive," there is reason to suggest that it
was a preemptive strike or a conservative reaction in defense of the traditional
"liberal technique of social study" that had dominated the discipline long
before the war. With the influence of emigre scholars such as Eric Voegelin
and Leo Strauss, new and somewhat "foreign" issues dealing with natural law,
relativism, and positivism were being introduced. There is no doubt that at
this time Political Theory did equal, in large measure, the history of political
theory, but the tradition of which Sabine was a part was not hostile to the
equally traditional commitment to science. Something new was happening in
Political Theory which was not easily defined. One can point to very little in
the way of concrete literature, but individuals such as Voegelin were becom-
ing active in the profession. Marcuse's work would not become a strong in-
fluence in political theory until the late 1960s and early 1970s, but Reason
and Revolution with its anti-scientific implications appeared in 1941.
Although the great tradition of political theory was deeply embedded in
American political science, the new generation of philosophical historians
would soon give that tradition new meaning, and the study would take on a
different significance that involved less the celebration of modern values and
social science than a critique not only of modernity but even of the liberal
vision of science and democracy.
There is little question but that the impending conflict between "scien-
tific" and "traditional" theory was also a consequence of the increasing shift
toward a vision of pure as opposed to applied science, and a perceived need,
12 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD
for several reasons, to place theory on an even firmer footing by
demonstrating its "scientificness." However, the discourse about political
theory through the 1940s increasingly revolved around questions of ethics,
relativism, and positivism which tended to produce a redefinition and
reconstruction of many issues in American political science (Hallowell, 1944;
Brecht, 1948).
Historically, these issues had not been a matter of great concern to
American political scientists. While in Europe fundamental value choices and
the grounds for them were practical as well as philosophical problems, the
American consensus and the belief in progress had made pragmatism and in-
strumentalism a reasonable position. The relativism of political values and
beliefs had been constantly stressed by historians of political theory from Dun-
ning to Sabine as well as by the more scientific school of individuals such as
Merriam and his successors. Although the atmosphere preceding and during
the war had raised questions about the ground of democratic values,
relativism had been basically viewed as in some way essential to liberalism.
For many of the Europeans and others touched by the transcendental urge,
the issue was somewhat different. Their philosophical background lent great
suspicion to the very enterprise of modern science which Americans had
trusted so basically, and, particularly, to its impact on politics and society.
Issues that concerned positivism, relativism, and historicism, were world-
historical matters of political as well as philosophical urgency that manifested
themselves in modern events.
The decade ended with the growing conviction of many political scien-
tists that there was a definite need for more work in the "field of scientific
method" that would yield "a body of testable propositions concerning
political nature and activities of man that are applicable throughout the
world" and "at all times" and that in the end would make possible a science of
"human political behavior" (Anderson, 1949, pp. 309, 314-315). This
summed up the governing motif of the forthcoming behavioral movement,
but it was also a reaffirmation of a basic faith. The rekindling of the scientific
mood was still tied to concerns about the realization of liberal values; this con-
nection would become, if not more attenuated, at least more submerged. The
growing sense that political science in the post-World War II period must be
more than a science of American politics and that liberalism was more than
an American mission moved the discipline further toward articulating a
vision of a universal conception of political science.

THE BEHAVIORAL REVOLUTION: 1950-1959

The behavioral revolution was a theoretical revolution in several senses.
First, it was a revolution in the theory of science. It introduced an un-
precedented metatheoretical consciousness about scientific theory and scien-
tific explanation. Second, much of the energy of behavioralists went into call-
ing for, creating, and applying, what they took to be theories. Third, there
was a distinct emphasis on pure or theoretical science and a turning away
from the idea of liberal reform and social control as the rationale of social
science. Finally, many of the individuals who were centrally involved in effect-
 john G. Gunnell I 13

ing the behavioral revolution were by trammg political theorists of the

historical and normative kind; they were what they sought to replace. This
group, who at least wrote dissertations in traditional political theory, in-
cluded David Easton, Robert Dahl, Heinz Eulau, John Wahlke, Karl
Deutsch, Herbert McClosky, Albert Somit, lthiel de Sola Pool, Alfred and
Sebastian de Grazia, and Austin Ranney, to name just a few.
The 1950s were the crucial decade in the development of both PT and
pt. Although there were echoes of the move toward scientism in the 1920s,
there were some important differences. First, the image of science embraced
by behavioralism had a significantly greater impact on the research programs
within the discipline. For the most part, historical, legal, and descriptive in-
stitutional analyses actually did dominate the field well into and beyond the
1940s; in the early 1950s, the belief that the practice of political science had
not lived up to its scientific promise was vehemently expressed. Samuel
Eldersveld (1951) noted that it was not necessary to change a syllable of Mer-
riam's 1925 statement that there are "signs of hope that genuine advance may
be made in the not distant future toward the discovery of scientific relations in
the domain of political phenomena" (p. 87). Second, although some still
linked the need for a "profound understanding of political behavior" to prac-
tical issues and the advance of democracy (White, 1950, p. 18), few explicitly
held fast to the old alliance. Lasswell (1950, 1951) continued to remind the
discipline that the basic purpose of political science was "to decrease the in-
determinacy of important political judgments" and enhance the chances for a
rational democratic society (1950, p. 425), but for most behavioralists the in-
ternal scientific conversion of the discipline was the first order of business.
Third, the discussions focused in a much more specific way on the concept of
theory and the place of theory in science. The view was that any change in the
direction of scientific progress would require a change in the nature of theory
in the discipline. Fourth, the claims of behavioralists about science and theory
fractured both the discipline and the sub-field of Political Theory. Fifth, the
arguments in defense of and opposition to behavioralism engaged political
theorists in a wide range of philosophical and metatheoretical arguments.
Finally, a distinct difference between PT and pt, and the problem of the rela-
tionship between them, became apparent.
Lippincott, in 1950, attempted a critical analysis of the field of Political
Theory. He took it to be the "most scientific branch" of political science but
an area in which political scientists had "produced little" if theory were
defined, as he believed it should be, as "the systematic analysis of political
relations" (p. 208). The methods in Political Theory had been basically his-
torical, and the "emphasis placed upon the history of political ideas has
meant very largely the abandonment of the aim of science" (p. 214). Some of
the shortcomings of Political Theory could also be attributed to the "inade-
quacy of empiricism, " or the crude inductivist misunderstanding of empiri-
cism which eschewed both evaluation and generalization (p. 218). Political
theorists had not contributed much to understanding the great political issues
and events of the age.
The call for scientific theory was often accompanied by an attack on cur-
rent practices in the discipline and particularly on Political Theory. Theory
14 I POLITICAL THEORY : THE EVOLUTION OF A SUB-FIELD
was characterized as teleological, moralistic, historical, ethical, and, in
general, in about the same state that Aristotle left it. Herbert Simon (1950)
was one of the first to suggest that much of the previous work in the field did
not deserve the name of theory and that it was "time that we maintained a
consistent distinction between political theory (i.e. , scientific statements
about the phenomena of politics) and the history of political thought (i.e.,
statements about what people have said about political theory and political
ethics)" (p. 411). He emphasized the need for interdisciplinary model con-
struction that would yield predictive generalizations and observably testable
propositions.
Exactly what was involved in this commitment, or recommitment, to
science and method was not very clearly defined. Most of the claims about
science were in terms of very abstract demands about the need for observation
and generalization and reflected versions of arguments about the logic and
epistemology of science that came, in a secondary or tertiary manner, from
the philosophy of science. The new scientific outlook was informed by-or
perhaps, more accurately, justified by- "a thorough-going empiricist
philosophy of the sciences" based on "logical positivism, operationalism, in-
strumentalism" (Lasswell & Kaplan , 1950, pp. xii , xiv). But no one was
directly coming to grips with the question of exactly how these philosophical
claims related to social scientific practice.
When viewed in historical perspective, Easton's now classic analysis of
the "decline of political theory" (1951; 1953) does not seem to startling. His
indictment of Political Theory for falling into historicism and failing to
engage in either creative and relevant evaluative theorizing or developing
causal scientific theory reflected, as well as gave impetus to, an increasingly
concretely articulated sentiment. Easton's initial essay appeared as part of a
symposium on the relationship of political theory to political research. All the
essays advocated a sharp break with past modes of theorizing; "the integration
of theory, methodology and research; interdisciplinary co-operation; and a
treatment of theory as much more than utopian goodness" (de Grazia, 1951,
p. 35). Although it is possible to recognize some attributes of the past litera-
ture in these critiques of the early 1950s, it is equally clear that these charac-
terizations did not accurately represent either the motives or practice of most
past political theorists. In part, the critique aimed at creating an image for
justifying a different direction in concrete research programs in the disci-
pline. But the problem also seems to have been what political theory was
feared to be becoming; it was charged with moralism and antiquarianism
more on this basis than on the basis of what it had been.
I have stressed this "conservative" aspect of the behavioral critique
because it has been neglected, but it is necessary to recognize that actual
research in political science had never corresponded to the scientific vision of
the discipline, and the critics were determined to change the practice. To do
this required an image of the past as well as the future. Although it would be
a mistake to attempt an explanation exclusively in terms of the sociology of
knowledge (response to funding sources, the need to clear political science of
ideological bias, etc.), it is useful to note that it was a matter of concern to
political scientists that "the National Roster of Scientific and Specialized Per-
sonnel, set up to find useful talent during World War II, classified Political
 John G. Gunnell I 15

Theory on the advice of the APSA as concerned with political ethics and the
history of political ideas" (Smithburg, 1951, pp. 61, 68). The notion that
political theory was at the time basically the history of ideas was probably true
in terms of the portion of the university curriculum that was designated as
Political Theory, and, apart from some scope and methods courses, this
would remain generally the case through the mid-1960s at most institutions.
Two developments in the 1950s are quite clear. First, there was the
establishment of a basic dispute between "new" and "old" ways of theoriz-
ing- based on less than accurate images albeit accepted by protagonists on
both sides. Second, PT was slowly but consistently differentiated from , but in-
volved with, the wider realm of pt that was in the process of formulation.
This attention to the concept of theory during the 1950s began to have
the effect of changing it from a category and a subject matter designation to a
term that was assumed to refer to a definite element in both science and
politics. Furthermore, the focus on theory created a whole realm of discourse
and controversy about the "theory of theory" that became increasingly dif-
ficult to separate from theory itself and that had a considerable impact on
theoretical practice. By the early 1950s, the debate about science and theory
was already being transformed into a debate about philosophical positivism.
In a symposium on "Recent American Political Theory," the comments by
Simon on an essay by Dwight Waldo had less to do with the substantive issues
than with Simon's perception of Waldo as a political theorist attached to a
pre-scientific conception of that role. Simon charged that the essay was
"characteristic of the writings of those who call themselves 'political theorists'
and who are ever ready to raise the battle cry against positivism and em-
piricism" as a threat to democracy but continue to write in a "loose, literary,
metaphysical style" (1952, pp. 494, 496).
One of the earliest attempts to specify the content of the behavioral
vision of political inquiry was the 1952 report of the Social Science Research
Council seminar on political behavior. Here the approach was "distinguished
by its attempt to describe government as a process made up of the actions and
interactions of men and groups of men" and to "discover the extent and
nature of uniformities" (p. 1004). These goals were to be accomplished by the
formulation of systematic concepts and hypotheses; the development of ex-
planatory generalizations that would raise inquiry beyond mere factual em-
piricism; interdisciplinary borrowing; empirical methods of research; direct
observation; and a distinct separation from concerns about "how men ought
to act" (p. 1004). This would all be codified in the behavioral credo by the
mid-1960s (Easton, 1965a, p. 9).
A significant factor in this behavioral affirmation was the problem posed
by the emerging field of comparative politics that seemed more than any
other aspect of the discipline to require a theoretical advance in order to deal
with new and complex data. An SSRC report in 1952 stated that "the prob-
lem of comparative method revolves around the discovery of uniformities"
and the "need for taking some kind of methodological position prior to or
along with the collection and descriptive enumeration of facts" (1953, p.
643). This sense of being overwhelmed by facts was hardly novel in the history
of the discipline. Although the answer to the problem that was given-politi-
16 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD
cal theory-was not novel either, the commitment to the answer in the form
of developing a "scheme of inquiry" or "analytical scheme" that would give
direction to inquiry and form to the data was unprecedented. The possession
of a model or conceptual framework would be the badge of the empirical
political scientists in the 1960s.
The poles of the argument within the discipline were increasingly repre-
sented in terms of normative concerns and the study of the history of political
theory versus theory as part of an empirical science of politics (Easton, 1953;
Hacker, 1954; Driscoll & Hyneman, 1955; Glaser, 1955). At the same time,
"theoretical" controversies were fast becoming more metatheoretical debates.
By the mid-1950s, political theory could be considered a subject of analysis
with a whole range of types and categories that would have been difficult to
conceive in the 1930s (e.g. , Jenkin, 1955). For behavioral political scientists,
the problem was to find out what theories were in science and what role they
played and to build or construct them . Thus, methodology involved not only
techniques of research but reading in the "fields of epistemology, logic,
philosophy of science" (Driscoll & Hyneman, 1955, pp. 192-193). To turn to
this literature, or more commonly to secondary accounts of it, was to turn to a
field almost entirely dominated by logical positivism and logical empiricism;
this then, was reflected in the discipline of political science and in the image
of science accepted by both the advocates and opponents of behavioralism.
Even a conference on political theory in the study of politics that was by
no means dominated by proponents of the behavioral persuasion was reported
as reaching a consensus on the view that "all types of inquiry involve the con-
struction of theory" and "that the title of 'political theory' has been un-
justifiably appropriated by historians of political thought" (Eckstein, 1956, p.
476). The spirit of the conference seemed to be in favor of demonstrating that
there was a "false distinction between 'behaviorists' and 'theorists,' " but the
conferees seemed unable to end the distinction. There was a major split be-
tween the "behaviorists" - they were not quite united on the term-who
wanted "to transform the field of political studies into a genuine scientific
disczpli"ne" and "anti-behaviorists" (or political philosophers) who believed
that "the end of the study of politics was something called political wisdom"
(pp. 476-477).
By the late 1950s, it was generally agreed that political theory had
"entered upon a time of troubles." On one side were those who saw theory as a
"historical, reflective, and 'literary' discipline more akin to moral philosophy"
than science (Smith, 1957, pp. 734, 743) , and on the other side were those
who saw it as a set of systematic generalizations for dealing with empirical
data. For some, the theoretical enterprise was grounded in a search for
political wisdom, and for others it was based on a "greater clarification of the
epistemological foundations of science" and "training in theory construction"
(Apter, 1957 , p . 761). V. 0 . Key, Jr.'s analysis of the state of the discipline in
1958 focused on this dilemma. Key suggested that most of the "worries about
the state of our discipline relate in one way or another to the place of political
theory in our studies" (p. 967). While theory had been understood largely as
the history of political thought and had possessed a relatively autonomous
place within the discipline, the behavioral unification raised the question of
 john G. Gunnell I 17
"what relevance has political theory for other branches of political science."
Many believed that a radical reconstruction of the sub-field was necessary.
Key claimed that it was clear that there was an "odd relation" between
"theoretical and empirical work" that tended to be "one.of antagonism, if not
hostility" and which had important implications for the future of the
discipline (pp. 967-968). For some, such as Norman Jacobson, the situation by
the late 1950s was such that the autonomy of political theory was threatened.
There was a danger of its absorption into the poles of scientism and moralism
as a consequence of the growing distance between "scientific" and "ethical
political theory" (1958).
Another view of the fate of Political Theory was offered by Robert Dahl.
In an extended review of Bertrand de Jouvenel's book on Soverei"gnty,
Dahl treated it as a "serious" but, in many ways, vestigial "effort to do
political theory in the grand style. In the English-speaking world, where so
many of the interesting political problems have been solved (at least super-
ficially) political theory is dead. In the Communist countries it is imprisoned.
Elsewhere it is moribund" (1958, p. 89). Dahl believed that political theory
had been reduced, in political science, to a kind of parasitic form of "textual
criticism and historic analysis," and that although attempts to return to
political theory in the grand manner in the modern age were to be applaud-
ed, they were faced with the inherent "impossibility of satisfying the scientific
function of political theory" (p. 95).
Many of those individuals who worked within the traditional sub-field of
Political Theory were, both through inclination and coercion, pushed further
away from the mainstream of the discipline. In the late 1950s, however,
political theorists who felt alienated were able to find little support in a wider
literature of political theory. Today we are accustomed to the notion that
political theory is a general field drawing from, and constituted by, work in a
number of disciplines including political science, history, and philosophy;
such a field simply did not exist or, at least, was seldom consciously perceived
as existing in the 1950s. To the extent that it was perceived, it was seen as
declining and it offered little support to political theorists in political science.
For individuals such as Strauss, the very notion of theory in modern
social science, as exemplified in the behavioral movement, signified the
"decline of political philosophy," while for Easton the identification of theory
with the history of political theory was symptomatic of the "decline." But the
argument about decline was also coming from other quarters. Arnold Brecht
(1959) elaborated his earlier views about relativism and the crisis of political
theory; Judith Shklar (1957) found decline to be part of the syndrome of post-
enlightenment thought; Alfred Cobban (1953; 1959) believed it was manifest
in the alienation of political theory from politics and its transformation into
an academic discipline as well as the tendencies of historicism. Finally, the
notion of the decline of political philosophy and/ or theory was beginning to
appear outside political science. It was a product of perceptions of the im-
plications of positivist ethical theory for the status of moral reasoning and no-
tions of rationality in normative discourse. Philosophers such as T .D. Weldon
(1953; 1956) in effect suggested that much of traditional political philosophy
rested on a mistaken belief that moral and political principles were, like em-
18 I POLITICAL THEORY : THE EVOLUTION OF A SUB-FIELD
pirical scientific claims, in some substantive sense demonstrable. In the early
1950s, philosophers did not really have much basis for questioning the
positivist image of either science or ethics, and to the extent that the possibil-
ity of political philosophy was tied to the belief in grounded value judgments,
Peter Laslett (1956) concluded that the "tradition has been broken" and
"political philosophy is dead."

RECOUPING AND REGROUPING: 1960-1969

Although the tension between the images of traditional and scientific
theory was a central motif in the literature of the 1960s, other things were
happening in Political Theory. As a result, any general characterization of
the state of the field is more difficult than it was in the 1950s.
The writings associated with the Straussian persuasion in political
philosophy (e.g . , Strauss, 1959; Jaffa, 1960; Storing, Ed., 1962; Strauss &
Cropsey, Eds., 1963) and to a lesser extent the work of Voegelin (1952; 1956)
and Hannah Arendt (1958; 1961) represented a distinct counterpoint to
behavioralism both in that they dominated work in the history of political
theory and in that they constituted an alternative to the behavioral approach
and its philosophical assumptions. For some, this indicated a "revival" of
political theory (Germino, 1963). Sheldon Wolin's Politi"cs and Vision,
published in 1960, was in many respects a significant departure from the type
of historical analysis associated with Sabine and at the same time represented
a position that was not altogether compatible with the direction of arguments
such as those of Strauss (e.g., Schaar & Wolin, 1963). Wolin indicated that
the great tradition was the basic object of the study of political theory and
stressed its relevance for understanding and dealing with the present. He of-
fered a general view about the decline or "sublimination" of political theory
in the modern age that reinforced the general mood in political philosophy.
More than any other work of the period, his book focused on the very idea of
theory, the theorist, and the activity of theorizing in a manner that con-
stituted, at least implicitly, a challenge to the behavioral notion of theory.
The presidential address to the APSA in 1961 suggested that
behavioralism and "the move toward scientism has made it more essential to
define the role of political philosophy in the study of political science" (Red-
ford, p. 758). But there were growing indications that the tensions would not
abate and that much of political theory would seek its reference point outside
the discipline of political science. There were pleas for reintegration, or at
least complementarity (e.g. , Thorson, 1961; Bluhm, 1965), but these often
pleased neither the behavioralists nor their opponents. It was, however, not
the case, as it is sometimes believed or suggested, that political theory, as
history and philosophy, was ostracized. The pages of professional journals
during this period contained more political theory of the non-behavioral
genre than ever. Although political scientists as a whole apparently did not
believe that, as a field, Political Theory was producing very significant work,
it was the only field between 1953 and 1961 that indicated any "startling" in-
crease in interest (Somit & Tanenhaus, 1980). It is probably accurate to say
 john G. Gunnell I 19

that what was taking place was less the rejection of traditional political theory
than its differentiation.
By the late 1960s, the profession had officially divided Political Theory
into three parts- "Political Theory and Philosophy: historical, normative,
and empirical" (APSA Biographical Directory, 1968). Dissertation listings in
the APSR changed to "Political Philosophy, Theory, and Methodology" and
the latter category (which included methods as well as metatheoretical mater-
ial in the philosophy of science and philosophy of social science) was also part
of the listing for Book Notes and Bibliography. Annual meetings of the APSA,
by the late 1960s, began to make a definite distinction between empirical
political theory and political philosophy (or "traditional political theory"). By
this time, the Foundations of Political Theory, the Caucus for a New Political
Science, and various other unaffiliated groups that represented the diverse
concerns of political theory in both PT and pt had begun to offer panels at
the annual meetings. But whether the basic distinctions within political
theory were bipartite or tripartite or even more complex, they did not ade-
quately capture the diversity that was becoming characteristic of the field.
The problem is to disentangle some of these elements and reconstruct the
relationship between them.
The 1960s were a decade of optimism about the advance of scientific
theory and the achievement of the behavioral goal of "a science of politics
modeled after the methodological assumptions of the natural sciences"
(Easton, 1965a, p . 8). Although, today, some of the more extravagant claims
about the development of predictive causal theory that would rival modern
physics (advanced at that time) seem less than credible, these claims were
not simply rhetorical. The vision was one of a "general theory" that would
consist of "a deductive system of thought so that a limited number of postu-
lates, as assumptions and axioms, a whole body of empirically valid generali-
zations might be deduced in descending order of specificity" and provide pre-
dictive causal explanations of political behavior (Easton, 1965a, p. 9).
Behavioralists believed that theories were arising, or would arise, inductively
from a wide variety of empirical studies and data gathering (e.g., Berelson &
Steiner, 1964); but more than anywhere else, the heartland of political theory
for political science in the 1960s was located in the development of "models,"
"orientations, " "approaches," "strategies" of inquiry, and various conceptual
and analytical frameworks. This was the product of theory "construction"
and theory "building" in the discipline. These activities were seldom accepted
as entirely fulfilling, in themselves, the scientific vision, but they were seen as
definite steps in that path or as prototypes of fully scientific theory. Whether
it was a form of systems theory, structural functionalism, decision theory,
game theory, or some other such construct, these "varieties" (Easton, Ed. ,
1966; Pool, Ed., 1967) were directed toward the same general goal of em-
pirical theory.
Paramount to all of these efforts was some notion of politics as a
"system." Karl Deutsch's Nerves of Government was published in 1963, and
Easton's A Framework for Political Analysis and A Systems Analysis of
Political Life in 1965. Sociologists such as Talcott Parsons, Marion Levy, and
Robert Merton had a significant impact on analytical theory in political
20 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD
science during this period. By 1965, David Truman, drawing upon Thomas
Kuhn's model of paradigmatic scientific change, revised Merriam's history
and cast the story of American political science in terms of a movement from
an early period (1880s-l 930s) of "non-theoretical empiricism" or "almost total
neglect of theory in any meaningful sense of that term" to "a new disciplinary
consensus" based on the "emergence of an explicit interest in the political
system" as well as a "revival of interest in theory" and a "self-conscious and
fruitful awareness of the necessary conjunction of theory and empirical in-
vestigation" (pp. 867, 870, 871).
Truman noted that "the theory chorus" was in some respects "less
polyphonus than cacophonus" and that "in practice" it might not be possible
in the "predictable future" to develop "general models" that would constitute
"hypothetico-deductive theories," but it was clear that contemporary political
science had made "a recommitment to the goal of science" and that this re-
commitment was manifest in its theoretical program (pp. 871, 872). Truman
allowed that political scientists still had a certain obligation to address nor-
mative issues and that the new consensus would, "for an indefinite future," in-
clude "a less or non-scientific component" which could best be filled by a
study of the "classics" of political thought which the earlier non-theoretical
consensus had "thrust ... out of the mainstream" but which could again con-
stitute part of a fruitful dialogue within the field (p. 874).
A survey of the field of political theory in the mid-1960s argued that
"theory" in general should be understood as a concern with "what 'is' or exists
in politics" and specifically as a "search for a coherent image of the political
system" (Deutsch & Rieselbach, 1965, p. 139). Concerns about "ought" ques-
tions should be considered as part of "political philosophy" or some such
category. Political scientists were still concerned that in Political Theory the
"preoccupation has been with history, exegesis, and methodological conserv-
atism" in "contrast to other fields in both natural and social science, where
theoretical physics and economic theory are clearly distinguished from the
history of past theories." The authors argued that a systems analysis of the
field of Political Theory itself in terms of structural-functional categories
might be useful. They concluded that, from this perspective, "at present
political theory is not a well-integrated field, nor does it seem well-oriented
toward a prominent goal, " but they did see a healthy tendency toward "adap-
tion" which they defined as becoming more scientific and less concerned with
"pattern maintenance" or traditional political theory (p. 162). In 1966,
Gabriel Almond, once more drawing upon Kuhn's scheme as a basis for a
revisionist history of the discipline, reaffirmed Truman's views and claimed
that "in the last decade or two the elements of a new, more surely scientific
paradigm seem to be manifesting themselves rapidly" and that it could be
summed up by the concept of the "political system" (p. 869).
By the m1d-1960s the "heterodoxy" of behavioralism had become the
"new orthodoxy" (Pool, 1967), and it was attacked as both politically and
methodologically conservative. Critics were concerned with the substantive
political messages both in what political scientists said and did not say. Many
believed that whether it was the commitment to pure science or an implicit
ideological bias, political science seldom spoke to, or about, significant cur-
 John G. Gunnell I 21

rent political issues. Despite the problems in Vietnam, the crises of the
American cities, the civil rights movement, international tensions, uprisings
on college campuses and the like, political issues were largely absent from the
literature of political science. In most instances, it was political theorists who
took up this critique.
As late as 1965 and 1966, individuals such as Robert Lane were speaking
about the "decline of politics in a knowledgeable society" and about how the
"age of affluence, " beginning in 1950, had produced "a growing state of con-
fidence between men and government" that in many respects was making
both ideology and conventional politics obsolete (1965, p. 895; 1966). The
notion of the "end of ideology" advanced by Daniel Bell (1960) and similar
themes in the work of individuals such as Seymour Martin Lipset (1960) and
the revision of democratic theory in the voting studies to accommodate em-
pirical findings (Campbell, et al., 1960) were giving support, or otherwise
linked, to various claims about the decline or death of political theory (see
Partridge, 1961; Rousseas & Farganis, 1963; LaPalombara, 1966; Waxman,
Ed., 1969). In 1968, the introduction to a symposium on the "advance of the
discipline," noted that "the discontinuity between classic Political Theory and
modem political theory is obvious." Political theory with a normative em-
phasis "has had much less appeal to the post-war generation of political scien-
tists," and this was a consequence of the fact that the "official ideologies of the
1950s became increasingly unsupportable in empirical theory and untenable
in normative philosophy" (Irish, 1968, pp. 298, 299).
During these years, the liberal-pluralist vision of political reality, which
was embedded in much of American political science and reflected in both its
methodology and substantive concerns, was criticized both as a descriptive
and prescriptive claim. Although behavioralism, mainstream political
science, and advocates of pluralist democracy were not entirely congruent
categories, they often coincided in the work of individuals such as V.O. Key,
Jr., David Easton, Robert Dahl, Gabriel Almond, and David Truman.
Dwight Waldo (1956) had already noted, in the middle of the previous
decade, that the basic structure and values of the American political order
had been accepted and endorsed by political science and that the very idea of
science in the discipline assumed a fundamental agreement on ends that
allowed political theory to be transformed into methodology. This theme was
more fully developed by Bernard Crick in 1959. Individuals such as C.W.
Mills (1956; Horowitz, Ed., 1963) and Barrington Moore, Jr. (1958) had
raised questions about both the conservative stance of social science and its ac-
count of the political world; arguments such as these became the property of
dissident political theorists who attacked the theory and practice of American
politics in both domestic and foreign affairs (e.g., Green & Levinson, Eds.,
1970).
From the beginning to the end of the decade, the health of pluralist
liberalism was questioned by individuals such as Henry Karie! (1961; 1969)
and Theodore Lowi (1969). Its philosophical assumptions were criticized
(Wolff, 1969), and the ideological biases of political science hidden behind
claims to scientific objectivity were examined (Connolly, 1967; Connolly, Ed.,
1969). There was, it was claimed, substantive political theory embedded in
 22 I POLITICAL THEORY: THE EVOLUTION OF A SUB -FIELD

the disciplinary matrix of political science and in its various methodologies

and conceptual frameworks, and political theorists turned to a critique of
those assumptions. Arguments such as those of Dahl (1961) and Nelson Polsby
(1963) about the character and structure of political power were countered by
individuals such as Bachrach and Baratz (1962a; 1967) and Jack L. Walker
(1966), and revisionist notions of democracy advanced by political sociologists
and political scientists were attacked also (Duncan & Lukes, 1963; Davis,
1964). It was argued that there was a "behavioral syndrome" consisting of
"conservatism, fear of popular democracy, and avoidance of vital polt"tical
issues" (McCoy & Playford, Eds. , 1967 , p. 10); by the end of the decade the
Caucus for a New Political Science claimed that there was, or should be, an
"end to political science" as it was practiced (Wolfe & Surkin, Eds., 1970).
For the most part, there was, however, little direct joining of issues. While
some voices, such as that of Morgenthau, pointed to the "complacency" of
Political Theory and the lack of social purpose in political science, the
discipline continued to emphasize methodology and technique
(Charlesworth, Ed., 1966).
Some of the material published in the late 1950s and 1960s that would
significantly contribute to shaping the concerns of political theory in the
1970s was still invisible, or barely visible, in both the literature and the cur-
riculum. This included such works as Marcuse's One Dimensi·onal Man (1964)
and Arendt's Human Conditi'on (1958). A category such as Critical Theory
was simply not part of the discourse of the 1960s. Although some of this
material, as well as Continental literature reflecting neo-Marxist thought, ex-
istentialism, and phenomenology, was discussed by political theorists in
courses and professional meetings, the discussions rarely surfaced in the pro-
fessional publications. For example, not one article in either the APSR or]OP
in the 1960s dealt in any direct way with such themes. The material that
would constitute the substance of pt in the 1970s was appearing, but the form
was not yet evident.
More prominent in the literature of the period was the influence of
British analytic philosophy and linguistic analysis. By the early 1960s, the
pessimism about the condition and future of ethics and political philosophy
that had marked the 1950s had given way to a guarded optimism based on the
appearance of a number of works in post-positivist philosophy and particular-
ly in the area of moral reasoning (e.g., Toulmin, 1950; Hare, 1952; 1963).
Certain thinkers were coming to the conclusion that social science and
political theory were not mutually exclusive but rather complementary enter-
prises. All empirical work involved value assumptions, and empirical evidence
was relevant in making and sustaining evaluative and prescriptive arguments
(Runciman, 1963; Taylor, 1967). Political philosophy entailed both concep-
tual and substantive claims and, like normative reason in general, constituted
a certain form of discourse that, while not scientific, was rationally
autonomous. To the extent that political philosophy was a type of normative
reasoning or involved producing knowledge in the form of conceptual
clarification that guided action as well as understanding, it seemed to many
that "political philosophy in the English-speaking world is alive again" and
that it would "not wholly perish from the earth" (Laslett, 1967; Berlin, 1962).
 John G. Gunnell I 23
The influence of this literature on PT was not very great in the 1960s, even
though there were some important examples (e.g. , Thorson, 1962; Flathman,
1966; Pitkin, 1967). What this approach promised was a way of actually "do-
ing" political theory-rather than merely talking about its history and condi-
tion - and of doing it in a manner that was intellectually secure from the
kinds of criticism leveled by positivists such as Simon in early years. It was
rigorous yet normative.
Despite all the things that political theorists had said about science, few ,
if any, of these individuals had any concrete knowledge of, or association
with, the practices of natural science. For the behavioralist as well as the anti-
behavioralist, science was an image, and it was an image gained by and large
from various philosophical sources. Both parties embraced basically the same
image. As the issues revolving around the possibility and desirability of the
scientific study of politiCs sharpened, the articulation of this image became a
matter of great importance; increasingly it was inseparable from concrete
theoretical claims and actual techniques of analysis. Before the 1960s, there
were relatively few books available, outside technical literature in the
philosophy of science, that spoke to the question of the nature of science and
scientific method, let alone its application to social science (Cohen & Nagel,
1934; Kaufman, 1944). Arnold Brecht's Polt"tz"cal Theory (1959) expanded the
scope of the discussion about science, but few political theorists, let alone
political scientists in general, were at that time interested in submerging
themselves in the intricacies of European thought (positivism •. natural law,
phenomenology, etc.). Brecht was not happy with the implications of what he
termed "scientific value relativism" for the future of normative political
theory, but he was forced to confirm the positivist notion of scientific ra-
tionality and its impact on Political Theory as well as the positivist image of
scientific explanation associated with the "hypothetico-deductive" model
which nearly everyone equated with "science. "
By the beginning of the 1960's, this genre of mediational literature
standing between philosophy and social science began to bloom. One of the
most inHuential works used in scope and methods courses of the period was
Abraham Kaplan's Conduct of lnquz"ry (1964). Although Kaplan stressed the
autonomy of inquiry, or the relativity of the logic of explanation with regard
to the subject matter, and distinguished between philosophical idealizations
of scientific explanation and the practice of science, he nevertheless presented
a synthetic but in the end largely Americanized logical empiricist account of
science that stressed the underlying unity of science and its philosophy. The
book also suggested that it could provide a methodological foundation for the
practice of inquiry in social science.
Much of the debate about science (e.g., Charlesworth, Ed., 1962),
through the mid-1960s, had been about the possibility and desirability of ap-
plying scientific methods to the study of politics with both sides generally ac-
cepting the idea of the unity of scientific method and the account of natural
science that had become standard in the literature of the philosophy of
science, that is, logical positivism and the emendations of logical empiricism
(Nagel, 1961 ; Hempel, 1965; Popper, 1961). However, by the late 1950s and
early 1960s, both the approach and claims of this school were being challeng-
24 I POLITICAL THEORY : THE EVOLUTION OF A SUB-FIELD
ed in the philosophy of science by authors such as Kuhn (1962) (see also , Han-
son, 1958). The implications for the image and demands of science in the
social sciences, as well as for the question of the relationship between science
and the philosophy of science, were considerable, but Kaplan, for example,
did not mention Kuhn.
Although Kuhn's impact on the literature of Political Theory was even-
tually of singular importance, there was a general growing awareness that
more was involved than merely the question of accepting or rejecting the
methods of the natural science. There were numerous instances of media-
tional works both in social science and the philosophy of social sciene (e.g.,
Gibson, 1960; Van Dyke, 1960; Brown, 1963; Taylor, 1964; Meehan, 1965;
Braybrooke, 1965; Rudner, 1966; Frohock, 1967; Brodbeck, 1968; Greer,
1969; Isaak, 1969). Although much of it may have tended to confirm, and
give aid to, the long standing assumption or claim of most behavioralists that
there was a basic identity among the philosophical reconstructions of science
produced in the literature of logical positivism and empiricism, the practice
of natural science, and the demands of scientific inquiry in social and
political science, the consciousness of political scientists regarding these mat-
ters was being raised to a significant degree.
Another closely related body of literature that appeared in the 1960s,
which to some degree overlapped the material discussed above, was a series of
works in the philosophy of social science that challenged philosophical and
social scientific assumptions about the applicability of scientific methods to
the study of social phenomena (Winch, 1958; Natanson, 1963; Schutz, 1967;
Appel, 1967; Louch, 1966; Macintyre, 1962). Much of this work was ground-
ed in post-Wittgensteinian philosophy and in Continental phenomenology,
with certain strong affinities with , or parallels to, notions of the explanation
of social action in the work of Max Weber. There was , however, a fundamen-
tal ambiguity running through much of this material with regard to whether
it was challenging the idea of the unity of science, and claiming that the sub-
ject matter of the social sciences demanded another or autonomous method of
inquiry and notion of explanation, or whether it was claiming that scientific
methods of explanation were inappropriate for understanding social
phenomena. Although this literature in the philosophy of social science was
beginning to enter the discourse of political theory, there were few specific at-
tempts in the 1960s to apply these ideas to issues in PT or to relate them to
challenges to positivism in the philosophy of science (Gunnell, 1968). By the
early 1970s, however, the concern with these arguments and their application
would constitute a relatively distinct body of literature in Political Theory.
In the 1968 symposium on the state ("advance") of the discipline, the
analysis of Political Theory made a point of the fact that political theory as a
field and political theory as an intellectual activity were far from coterminous.
The authors claimed that while the activity of theorizing within the discipline
had increased, dissatisfaction with the field was on the rise. In their judg-
ment, the field was broad, diverse, fragmented, and replete with paradoxes
and perplexities. Despite the fact that most major departments still treated
Political Theory as a basic sub-field, it had little to do with the discipline as a
whole. While most social scientists viewed theory as "a systematic and self-
 john G. Gunnell I 25

conscious attempt ... to explain diverse phenomena that have been or can be
observed," theory in political science had traditionally involved much more
than this "explanatory" type (McDonald & Rosenau, pp. 317, 318). It had in-
cluded, and continued to include, various dimensions of political philosophy,
or normative theory and ideology, that were inconsistent with the concerns of
the "behavioral revolution" and the conception of theory in contemporary
political science as a whole. The result was "sustained and unrestrained
argumentation" (pp. 320, 321).
In the authors' view, the future of this sub· field, with its current "ungain-
ly structure, " might proceed in three possible directions. First, empirical
political theory and theorizing might spread throughout the discipline, while
the "philosophical" component was retained as a separate sub -field. Second,
the behavioral approach to theory might take over the sub-field while
Political Philosophy became an additional sub-field. Third, political science
as a whole might become an entirely empirical science while the normative
and philosophical components would find a place as appendages or elements
of other departments.
The authors were relatively optimistic about what they saw as the two
principal directions of political theory as an intellectual activity. First,
although the many conceptual frameworks and methodological strategies
such as systems analysis that had appeared in recent years were, at best, pre-
theoretical and of doubtful durability, there were indications that "theory at
all levels in all fields will soon experience previously unimaginable
breakthroughs because of technological advances" (p. 334). Second, there
was a "growing appreciation of the theoretical relevance of the great works of
political thought" and, "perhaps as a reaction to the behavioral revolution,
the inclination to treat them as historical artifacts, philosophical formula-
tions, and ideological tracts has given way to a growing concern for probing
their theoretical content" (p. 337). Although none of these predictions would
prove to be precisely accurate and although the hopes for theory might not be
realized, the tendency toward dispersion was perceived, and the differentia-
tion between PT and pt was recognized.
Another development was subtle and incremental; it would be more ob-
vious in discussions of political theory in the 1970s but less consciously
recognized as a transformation. It has already been noted that in the begin-
ning of the century, "political theory" was understood more as a functional
category, a sub-field name, or a literature classification than a distinct entity
or activity. This gradually changed through the 1950s, but it was primarily
the debates about theory in the 1960s that gave rise to the contemporary no-
tion that theory is actually an element in the practice of science, and more
than an analytical specifiable element; that political theory is a concrete
activity with a past and future; and that theorizing is a distinct endeavor.
There is a definite sense in which the contemporary notion of political
theory was invented in the context of the debates about behavioralism
and the decline, and revival, of political theory. The concept became
reified to a degree that would have been unintelligible to earlier generations,
and the growth of pt lent credence to this idea. This can be concretely
illustrated.
 26 I POLITICAL THEORY : THE EVOLUTION OF A SUB -FIELD
In the 1937 edition of the International Encyclopedia of the Social
Sciences, not only was Political Theory not a separate heading, it was not even
treated as a distinct sub -division in the discipline of political science. In the
1968 edition, Political Theory was not only given prominence and separate
status in the section on Political Science, where it was treated as an activity,
product, and sub-field, but it became a separate and equal topic where it was
discussed almost as if it were an autonomous discipline with its various
divisions , problems, dimensions, and history. Although the views of theory, as
presented by Easton in a discussion of theory in "Political Science," and by
Wolin and Brecht, in their analysis of "Political Theory," were in many
respects quite antithetical, they both exemplified this reification of theory as
well as the tension between notions of theory in PT and the growing distinc-
tion between PT and pt.
Easton argued that political science, by mid-century, was still "a
discipline solving its "identity crisis" and emerging with a "systematic
theoretical structure of its own" that was largely the consequence of "the
reception and integration of the methods of science into the core of the
discipline" (p. 282). The basic thrust of this "theoretical revolution" was a
tum toward functional and systems analysis as opposed to a focus on institu-
tions such as the state. According to Easton, the core of the behavioral move-
ment was a "shift from an institutional and practical problem orientation" to
pure science, and this shift was most "sharply revealed" in "the sub-field of
political theory." The changes were distinctly reflected here, and at the same
time the work in theory was crucial in moving the discipline in an analytical
direction (p. 243). Easton claimed that behavioralism had dispensed with
"the last remnants of the classical heritage in political science." It produced
"a profound transformation in conceptions about the role that theory plays as
a tool in political research" - a role that was distinguished by a separation of
fact and value and an attempt "to drive political science away from a pre-
scriptive problem-directed discipline to one in which research depends in-
creasingly upon empirically oriented theoretical criteria" (p. 296).
Although Wolin took issue with Easton's notion of science and the role
that theory played in science, he disagreed less with Easton's description of
what theory had come to mean and to be in political science and with the
basic "traditional" /"scientific" distinction than with the implications of this
development. For Wolin, what was happening in political science was a
serious deviation from the historical role of theory and theorizing and its rela-
tionship to politics. According to Wolin, "the quest for a scientific theory of
politics has altered the character of theorizing in several significant ways" and
what it has produced on the whole has been the "sterilization of political
theory" (pp. 325, 328). In an important sense, theory for both sides was
something even bigger than political science and politics but something that
might be manifest or realized in either.
Those who valued the study of the history of political theory were seeking
to articulate exactly what it was all about in order to defend it against the
behavioral challenge. George Kateb (1968), for example, attempted to
specify the principal characteristics of "traditional political theory" and to
justify its uses in an age in which it had "fallen on hard times." For the
 john G. Gunnell I 27

behavioralist, the realization of theory was in the near future in political

science, and for the historian the future of theory depended on the recovery of
its past. Claims such as those of Easton and Wolin capture much of the flavor
of the debates about theory that characterized the 1960s, but the last year of
the decade indicated some new directions that would permeate discourse in
and about political theory during the 1970s.
First of all, in 1969 Easton announced, or called for, "a new revolution in
political science." This was to be a "post-behavioral revolution" which was not
"theoretical" or a change in the methods of inquiry but a change in orienta-
tion that grew out of a "deep discontent with the direction of contemporary
political research" and which advocated , at least in the short run, more atten-
tion to the public responsibilities of the discipline and to relevant research on
contempoary political problems and issues (p. 1052). The shift in distribution
of emphasis from Easton's stress on the priority of "pure science" and the
separation of fact and value in the "behavioral credo" (1965 , p. 17) as well as
his statements about political science and political theory the previous year
was, at least on its face, dramatic. This, in some ways, was the official birth
announcement of the public policy enterprise which would become the basis
of the self-image of orthodox political science in the 1970s, and in terms of
which it would attempt to establish its identity. With this shift would come a
distinct de-emphasis of the concern with general unified theory as the core of
the discipline as well as a retreat from any pointed confrontation with the
history of political theory.
In the same issue of APSR in which Easton announced the new revolu-
tion, Wolin presented his defense of "political theory as a vocation." This was
in a very real sense the quintessential statement, as well as, in many respects,
the effective terminus of the critique of behavioralism from the standpoint of
the history of political theory. Although Wolin's position might not spring
from the same intellectual and ideological source as that of Strauss and
others, he spoke for a generation of political theorists when he defended the
role of the historian of political theory as conservator and transmitter of
"political wisdom" against the "methodism" of political science and its
"historyless" posture that neglected the "tacit political knowledge" that
should inform both scientific and political judgment. He spoke also for the
preservation of that "vocation by which political theories are created" in the
midst of a world that is governed by "giant, routinized structures" and which
is "impervious to theory" and in which political science is marked by "com-
placency" in the face of political crisis and chaos (pp. 1070-1071; 1077; 1080-
1081). Although Wolin's arguments articulated the sentiments of many
political theorists , his statement was shouted in a desert, and it was clear that
he as well as other political theorists would have to seek a forum outside
political science. Behavioralism, which in the immediate sense at least, had
begun the war with the history of political theory, was disengaging. On the
other hand, there was a distinct sense in Wolin's essay that the history of
political theory no longer had a home in political science.
Although it would be several years before developments had progressed
sufficiently for a discussion of the "new history of political theory" as a positive
enterprise, Quentin Skinner- who would be prominently associated with this
 28 I POLITICAL THEORY : THE EVOLUTION OF A SUB-FIELD
position in the 1970s - offered a comprehensive and influential critique of
past research in the history of Political Theory in 1969. It was an indication of
the anomalous situation of much of this literature, caught between political
science and pt, that Skinner's methodological critique, in terms of under-
standing in the history of ideas, tended to neglect the philosophical, political,
and disciplinary contexts in which many of the arguments such as those of
Strauss developed. But the arguments of individuals such as Skinner were not
simply critical but legislative. In the 1970s much of the literature dealing with
the history of political theory would revolve around matters of the advocacy
and criticism of the "new history" . The shift was decidedly outside the context
of political science in both a methodological and substantive sense.
Finally, by the end of the 1960s, there were indications of a distinct
transformation in the theoretical or philosophical critique of political science.
It is easy to forget how little had been published in criticism of the behavioral
image of science by the late 1960s that went beyond general concerns about
treating politics scientifically and practicing science politically. Although
there were growing concerns about the integrity and validity of the behavioral
image of science, the critique of behavioralism had in many respects been an
external one that was relatively easy to defuse by drawing lines between those
who were committed to science and those committed to some other approach
to the study of politics. In the same issue of the APSR in which Wolin and
Easton's articles appeared, there was a critique of the philosophical assump-
tions of the basic model of scientific explanation that had been adopted by
political scientists and of the relationship between philosophical claims about
science and the practice of social scientific inquiry (Gunnell, 1969). This cri-
tique was published as part of a symposium that included two editorially
solicited rebuttals.
There had been a great educational, financial, intellectual, and emo-
tional investment in the behavioral image of science and theory, and despite
all the talk about policy, that image was at the core of disciplinary practice.
The battle over this issue was, in many respects, yet to come. The question of
the philosophical status of this image of science and related ideas would oc-
cupy a central place in the literature of political theory (both PT and pt) dur-
ing the next decade and would, as a matter of fact, contribute significantly to
the dispersion of political theory.

THE DIASPORA OF POLITICAL THEORY: 1970-1979

In the 1970s, there was the disappearance of an issue center within PT,
and the sub -field was progressively transformed into pt writ small as the
autonomy of the latter became more firmly established. Journals such as
Phi'losophy and Public Affafrs (1971) and Polit£cal Theory (1973) had begun
publication. Political theory was no longer centered in the discipline of
political science, and political science no longer defined the issues in the
literature of Political Theory. New influences, such as the translation of the
work of Habermas, came onto the scene the same year as the publication of
 john G. Gunnell I 29

Rawls' book; commentaries on, and tributaries of, this material occupied
much of the space in Political Theory during the decade. Throughout the late
1960s, graduate education in Political Theory had continued to be traditional
with an emphasis on the history of political theory. Although there was some
increased attention given to scope and method courses, and although a few
thinkers such as Camus, Sartre, Arendt or Marcuse were featured fairly
prominently in some limited contexts, few graduate students were exposed, in
any systematic or extensive manner, to much more than the traditional
canon. This was true even at institutions where theory was prominent. But
during the 1970s, graduate education changed significantly in response to
developments in pt. Names such as Rawls, Nozick, Habermas, Heidegger,
Foucault, Gadamer, Lukacs, Gramsci, Althusser, Kuch, and Popper, became
common currency, and students were asked to come to grips with such exotic
fields such as sociobiology, hermeneutics, and structuralism. It is easy to
forget just how radical the transformation was that began in the 1970s in both
education and the scholarly literature. Assessing its significance is a conten-
tious matter.
Many would see the developments of the 1970s as growth in the in-
dependence of political theory, and as a break from its less than happy home
in political science. But although pt became a recognizable body of literature,
and even a partially institutionalized interdisciplinary field with journals and
organizations that reflected its concerns, it was largely a collection of intellec-
tual enclaves with limited communication between them. However one views
the implications of the events of the 1970s, the result was the dispersion
of political theory both between and within PT and pt. To disentangle the
various themes and threads is not an easy task, but a natural starting point is
an examination of the remnant of political theory that was distinctly attached
to the discipline of political science.
It was clear from the beginning of the decade that, in practice at least,
the discipline's emphasis on universal theory was waning. The disjoining of
the debate between historians and scientists led to a more pluralistic view of
theory; behavioralism was quite willing to be tolerant of diversity once it had
captured the center; the "new revolution" also placed less emphasis on theory;
diversity and tolerance were forced on the discipline by the critics in the late
1960s; and specialization in research was leading to more emphasis on in-
tellectually localized research strategies. By the early 1970s, it was not easy to
find coherence among the trends or to bring the discipline together into a
common notion of theoretical endeavor.
In attempting to understand this situation, Karl Deutsch's 1970 presi-
dential address to the APSA is instructive as an attempt to confront the prob-
lem (1971). Several things were quite clear in Deutsch's eclectic view of
theory. First, it was a very sharp departure from the conscious attempt by
himself and others in the mid-1960s to define theory in a narrow manner and
to make a sharp distinction between empirical political theory and political
philosophy. Second, there was a good deal of at least implicit reproof as well
as conciliation directed toward both academic and political critics of the
discipline. And third, there was a distinct call for political relevance which
echoed the statement of Easton. Deutsch was determined to find a middle
30 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD

ground between scientism and political radicalism, between theory as an

instrument for knowing reality and as ideology, and to present various
analytically distinguishable aspects of theory as "an integrative process" in-
volving "stages in a single production cycle of political knowledge and
political action" (p. 17). All this required many different types of people and
diversity in distribution of emphasis. In the end, Deutsch wanted to give all
sides a place while to some degree maintaining the priority of the behavioral
vision of theory, but as with most attempts at intellectual synthesis after the
fact of differentiation, it did little to change the situation and involved such
an amorphous notion of theory that the very thing that was to be identified
became still more elusive.
The resonant note in Deutsch's essay was the idea of political science as a
policy science; this would also be sounded in Heinz Eulau's analysis of the
"Skill Revolution and the Consultative Commonwealth" (1973), Avery Leiser-
son's call (1975) for a synthesis of science and politics and a reaffirmation of
the traditional American faith in the complementarity of "scientific truth and
democratic decision making," and Austin Ranney's (1976) celebration of the
vision of a "divine science" that had traditionally informed political science in
the United States from the time of the Founders and that could provide a
basis for the contemporary need to recognize the possibility of "political
engineering in American culture." This was the tone of the textbooks and
other less official statements about the purpose and direction of the discipline
during the 1970s, and although "post-behavioralism" in political science was
a concept that meant many things (Graham, Ed., 1972), it certainly included
the policy tum in the mainstream of the discipline. But it also represented the
ideological critique of the field as it was in the late 1960s and the
philosophical critique of the behavioral image of science. "Post-
behavioralism" was more a condition or a name for less than coherent con-
geries of interests than an intelligible intellectual movement or even mood.
The old controversy about political science as a science was, by the early
1970s, largely focused on the critique and defense of the behavioral image of
science in terms of the philosophy of science and philosophy of social science.
A second symposium in the APSR, in which a battery of political scientists
and philosophers was enlisted by the editor to exorcize dissident philosophical
claims about scientific inquiry, appeared in 1972. In this case, the article
(Miller) at the center of the symposium was, ironically, far from fully suppor-
tive of, or concerned with, the new philosophy of science represented by in-
dividuals such as Kuhn. It was written from a Straussian perspective and
focused on matters such as the historicism implied in Kuhn's work as well as
the positivism of behavioralism. The critiques, which indicated where the
primary sensitivities of the discipline resided at this point, fastened, however,
on the alien images in the philosophy of science.
By the mid-1970s, there were philosophically sophisticated critiques of
behavioral assumptions about political theory (Spragens, 1973); books
defending the behavioral image of science in terms of both the basic logical
empiricist philosophy of science and the responses of this school to criticisms
mounted by individuals such as Kuhn and Feyerabend (Gregor, 1971); and
books which aimed at demonstrating, in light of the contemporary work in
 john G. Gunnell! 31

philosophy, both the dubious character of logical positivism and logical

empiricism and their impact on the conduct and understanding of inquiry in
political science (Gunnell, 1975). This kind of controversy was inevitable, but
it had the effect of drawing political theory into a realm of metatheoretical
debates that in many respects had decreasing relevance to both politics and
political inquiry. Somewhat the same problem is apparent in the literature
that attempted to apply new ideas in the philosophy of social science
(Dallmayr & McCarthy, Eds., 1977) to a critique of positivism in social and
political science and to a reconstruction of social scientific inquiry. Such syn-
thetic accounts drawn from linguistic analysis, phenomenology, and critical
and interpretative theory (e.g., Bernstein, 1976) were largely summaries and
restatements of material in philosophy, and, while useful and interesting in
many respects, this material transformed political theory into a less than
autonomous mode of discourse.
Compared to the 1960s, with the fixation on theory as conceptual frame-
works, it is difficult to discern exactly what orthodox political scientists under-
stood theory to be. The 1973 Supplement to the Biographical Directory for
the APSA separated out Methodology (which included epistemology and the
philosophy of science) and divided Political Theory into systems of ideas in
history, ideology systems, political philosophy (general), and methodological
and analytical systems. The (unofficial) Handbook of Polt'ti'cal Sdence
published in 1975, devoted Volume One to the Scope and Theory of Political
Sdence which included chapters on: the history of the discipline; an analysis
and synthesis of opposing philosophical views about the nature of social scien-
tific explanation; an essay on the contemporary relevance of the classics; a
neo-positivist philosophical analysis of the language of political inquiry and
political concepts; and a philosophical discussion of the language of political
evaluation. Volumes Two and Three were devoted respectively to what was
designated as Micro- and Macro-political Theory, but these were largely
categories for encompassing various subjects and research programs rather
than specifications of particular theoretical formulations (Greenstein &
Polsby, Eds.).
The official panels at the APSA meetings obviously had increasing dif-
ficulty deciding exactly what the structure of political theory was or should
be. Constantly changing configurations attempted both to capture the diver-
sity within the field and to reflect topical issues. "Formal" or "positive" theory
emerged as a distinct category in 1970, and diverse remaining panels were
listed under "Philosophical Analysis and Politics." In 1971, the panels were
divided between Formal Theory, Ethical Theory, and the Philosophical
Analysis of the Science of Politics. By 1972, the attempt to make any general
divisions in the sub-field was abandoned in favor of a potpourri including
various substantive, conceptual, and methodological issues. In 1973, there
was an attempt-without much apparent success since it included panels as
diverse as Statistical Problems and Research on Electoral College Reform - to
carve out an area designated as Methodological and Analytical Theory. The
1974 panels were divided between Macro Theory and Micro Analysis on the
one hand and Political Theory and Ideological Conflict on the other hand
with a large number of issues (post-behavioralism, epistemological alter-
32 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD
natives to behavioralism, etc.) relegated to the unaffiliated panels. An at-
tempt was made in 1975 and 1976 to hold on to a basic division between
Political Theory and Epistemology and Methodology, but it gave way in 1977
to Political Theory, Methods and Empirical Theory, and American Political
Thought.
From 1978 through 1983 there has been some basic division between
Analytical (and/or Empirical) Theory and Political Philosophy (and/ or
Political Theory or Thought) with the continued representation of many
issues in a wide variety of unaffiliated panels. But the criteria of demarcation
has not been very clear or consistent. By 1982, the APSA Directory and the
Guide to Graduate Study in Political Science divided the field into Political
Theory and Philosophy, Formal or Positive Theory, and Methodology, but
these categories are opt particularly descriptive of the field or the ten percent
of the members of the APSA who designate themselves as primarily political
theorists.
Although a general neo-positivist or logical empiricist view of science,
scientific explanation, and theory continued to be quite pervasive in the dis-
cipline at large, it usually found expression, in pro forma terms, in textbook
introductions and methodological prefaces. There were few criteria for con-
necting this persistent philosophical image of science with what political
scientists actually did. The criticism and defense of this image did not peak
until the mid-1970s; thereafter, even some of those who had done the most to
propagate it began to disavow it. The irony was, however, that this image, in
the face of heavy criticism , had only recently received its most thorough ex-
plication and defense. At the very point at which many dedicated younger
centrist scholars in political science found and accepted it at the core of their
education, the older generation began to deny it.
In 1977, Almond, for example, was ready to concede-or "discover,"
since he did not acknowledge or mention any of the criticism that had been
mounted within the discipline during the past few years-most of the points
that had been advanced against the philosophical foundations of
behavioralism. He suggested that the commitment to these methodological
doctrines had caused political science to lose touch with its "ontological base"
and mistakenly attempt, with the encouragement of neo-positivist philoso-
phers of science and their partisans, to treat political phenomena as natural
events when they should instead have developed approaches "appropriate to
human and social reality" (p. 522). He argued that such constructs as the
deductive model advanced by logical empiricists as well as the general "ex-
planatory strategy of hard science had only a limited application to the social
sciences" and that the "search for regularities and lawful relationships" and
the "enshrining of the notion of generalization" represented the wrong direc-
tion for the discipline to have taken (p. 493).
Few orthodox political scientists so openly rejected the essence of the
behavioral faith, but few attempted to defend it. Some, such as William
Riker, continued to describe and justify their work and define theory in terms
of notions of the unity of science based on "positivist or objectivist
philosophy, " as set forth by individuals such as Hempel and Nagel, and to
pass off the criticism that by this time had come close to a new orthodoxy in
 john G. Gunnell I 33

the philosophy of science as "idealist interpretations of science" (1977 , p. 16).

But such attempts to restrict the idea of theory in this manner were not
common.
For both the critics and defenders of behavioralism, the deep but
inevitable involvement of political theory in the issues of the philosophy of
science and epistemology had become a problem. Theorists were being drawn
off into a realm of metatheoretical concerns that alienated them from sub-
stantive political issues and a concern with political phenomenon in general
(Kress, 1979). Heinz Eulau noted that the cycle of justification, criticism, and
defense of behavioral claims in philosophical terms had produced an "exag-
gerated, almost pathological, concern" that culminated in "the curious
notion that the philosophy of science is the high road to scientific knowledge,
as if the business of the philosophy of science were science rather than
philosophy" (1977 , p. 6). This precise point had been made by critics of
behavioralism over a decade earlier. Although one might not agree with the
particular biopolitical concerns of John Wahlke, his critique of behavioralism
(1979) also indicated a certain basic agreement with many of the critics who
viewed behavioralism as an empty commitment to methodism. Wahlke
argued that despite all the talk about theory and despite the technical
methodological advances , political science was still characterized by a
"paucity of theoretical concerns."
By the end of the 1970s what had happened to the history of political
theory? It was alive and living at least a large portion of its life in pt. Although
some reverberations from the active discussions in pt might surface in the
literature of political science, most of the controversy that was generated was
not felt in the discipline in general. There was no question that the history of
political theory was accepted as part of the field of political science, but it
was not at the center of concerns about theory in the discipline. Its new found
autonomy guaranteed its existence, but it was increasingly divorced from the
issues that had dominated the field and the claims of both its advocates and
critics for so many years. Autonomy meant a new self-consciousness which
issued in the call for a "new history" (e.g. , Skinner, 1979; Pocock, 1971) that
would reflect "truly" historical methods and recover the actual meaning of
classic texts as well as trace "actual" historical traditions. In the view of the
new historians, this required turning away from the philosophical concerns
and assumptions which they believed had governed past scholarship. While a
considerable body of literature voicing and reflecting this position emerged
during this period and added significantly to the understanding of classic
authors such as Locke (e.g . , Ashcraft, 1980; Tarlton, 1979), the purpose of
this activity, its relationship to political science, and a range of similar issues
were not only unresolved but seldom directly discussed.
To adequately convey the structure of political theory, whether in PT or
pt, during the 1970s and into the early 1980s would require listing many dif-
ferent and discrete enterprises. Different interests and concerns would yield
different distributions of emphasis. The important point is that the field was
becoming too dispersed for there to be any core set of issues that would be
readily agreed upon as establishing its identity.
Significant work in conceptual analysis was apparent (e.g. , Pitkin, 1972,
 34 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD

1978; Flathman, 1972, 1976), and analytical political philosophy was

well represented in works such as the continuing series of Philosophy, Politics
and Sodety. Although the Straussian project remained active after the death
of Leo Strauss, it was dispersed within itself and increasingly insulated from
the discipline of political science. There was a legacy of arguments about
scientific method and the nature of explanation, but these reflected con-
tinually finer drawn points in the philosophy of science and philosophy of
social science. The large body of literature that surrounded the work of Rawls
continued to grow as well as other quasi-transcendental arguments in norma-
tive theory (e.g ., Dworkin, 1977). One of the trends that characterized the
1970s was a growing concern with political economy and renewed attention to
the "state" as an object of inquiry. Much of this material was not in political
science proper (O'Connor, 1973; Wolfe, 1977; Skocpol, 1979), but it in-
fluenced the discipline (Lindblom, 1977). There were continuing attempts
to bring the world of Continental philosophy closer to the concerns of Ameri-
can political theorists. The problem of summarizing Habermas and keeping
up with his dicta was supplemented by problems of understanding individuals
such as Foucault , Gadamer, and Althusser. The world seemed accessible only
through what somebody else said about it, and authors more than their sub-
jects became the object of analysis.
This somewhat.impressionistic account could be expanded considerably
and still not adequately encompass the varieties of political theory. The prob-
lem is less one of inclusiveness than of determining how to assess the state of
the field. The proliferation signified for many that there was vital energy in
political theory, but it was difficult to identify any basic coherence. One
might, at a certain level of abstraction, find underlying connections or
similarities in the literature, such as a concern with language and interpreta-
tion or the problem of achieving some transcendental ground for political
values. But such characterizations tell us less about the field per se than about
what someone might find interesting. Although one could say that the field
was vital, it could just as easily be said that it was without definite direction or
focus.

PROSPECTS: THE 1980s

Since it is only in retrospect that we can know exactly where we were,
what I have termed the dispersion of political theory may, in future years, ap-
pear to be something quite different . But an external vantage point-
temporal or spatial - often imposes a restricted perspective. Even now it is
interesting to note how the situation in political theory is described by an out-
side observer. The results of one such investigation, while not intuitively
counterfactual, would probably be unsatisfactory to most participants in the
field . A report on the state of political theory by the American Association for
the Advancement of the Humanities described it as a "discipline" that had
"died and been reborn several times." The conclusion was that in its present
"incarnation" it consisted principally of a debate between Straussians,
analytical philosophers, and contextualist historians and that these debates,
marked by "intellectual irreconcilability," structured both academic and pro-
 John G. Gunnell I 35

fessional life (Herbert, 1981 , p . 8). A much more extensive study of political
theory in the United States has been in progress in Germany by the late Peter
Lutz and now by the executors of his academic estate, but the difficulty facing
such anthropologists of political theory is in part a matter of where they hap-
pen to cut into the culture and the categories that they bring to their analysis.
They also probably fail to grasp the complexity of the society with which they
are dealing. One might not want to say that their results are wrong, but their
conclusions are too partial and narrowly focused. The principal difficulty of
all these analyses is that they are seeking the identity of political theory.
The dispersion of political theory that took place in the 1970s has
resulted in a loss of identity (both in PT and pt). The idea of political theory
advanced by behavioralists, the counter arguments of historians of political
theory, and the controversy in which they were joined may all be rightfully
interred. But these arguments gave a shape to the field that is lacking today.
The loss may not be at all lamentable, and there is no obvious reason why a
discussion of prospects should assume a need to reconstitute such a form. But
it is necessary to ask questions about the elements of the dispersed residue of
that form and the relationship between them as well as their relationship to
politics.
Although there are some individuals in political theory who believe that
the field would be best served by finalizing the divorce from mainstream
political science, such views must assume that there is a space for the activity
of political theory outside political science. Often, however, such ideas are
based upon abstract world-historical views of political theory when in fact the
very idea of political theory has been largely a product of the evolution of
political science. Its future outside of the discipline may be something of an
illusion. There is, for example, little reason to think that pt is likely to become
anything more than the loose collection of endeavors that now comprise it.
But the question of whether there is a life for political theory in political
science cannot be !!Voided.
Political theorists are often not only alienated from the concerns of most
research programs in political science, but not even engaged in a critical
assessment of these programs. Their principal reference point is the world of
pt. On the other hand, within political science, there are the views of those
who seek to narrow the meaning of political theory not only to orthodox
political science but to a particular position within the discipline. For exam-
ple, William Riker (1983) claims that "the main line of development during
the last generation" has been in the direction of social choice theory, and he
defines theory in terms of the logical empiricist deductive nomological model
which he in turn equates with the practice of natural science. These positions
have the unfortunate effect of artificially limiting critical and interesting dis-
cussion in or about political theory.
It is probably too optimistic to believe, however, that contemporary
trends reflect an end to the conflict between "humanists" and the "science
establishment" and signal a "paradigm synthesis" (Bluhm, 1982, p. 3). Nor is
it likely that an answer lies in a synthesis of philosophical perspectives about
the nature of social scientific explanation (Moon , 1975; 1982). Although it
would be nice to believe that the problem is basically one of "communication
36 I POLITICAL THEORY : THE EVOLUTION OF A SUB-FIELD
among all scholars committed to understanding and explaining political
phenomena" (Graham, 1982, p . 206), the problems run deep, are substantive
as well as methodological in nature, and have to do with the very idea of social
science and the relationship between social science and politics.
It is probably true, as the advocates of synthesis believe, that there are
"no necessary barriers that currently prevent discourse across the methodolo-
gies and paradigms of political science" (Graham, p . 222) . We are not in the
situation of the mid-l 960s when there were severe obstacles to discussion both
within political science and between political science and other disciplines
such as economics and philosophy. Clearly there are , in both theory and prac-
tice, synthetic exercises that have brought together philosophy, economics,
and political science in areas as diverse as public policy analysis and critical
theory. And the general ability of political scientists to move between these
areas is much more apparent than ever before. Furthermore, the general
mood of the early 1980s is not along the lines of the focused intransigent con-
frontational discourse that marked the 1960s and 1970s. The problems may
be more those of pure tolerance.
Brian Barry, whose own work is part of what many believe to be the
renaissance in political philosophy, sees, in comparison with the lack of
political philosophy in the early 1960s, a positive movement. But he also
admits to an occasional "nightmarish feeling that 'the literature' has taken off
on an independent life and now carries on like the broomstick bewitched by
the sorcerer's apprentice" (1980, p . 283). The question is not whether there is
a great deal of political theory and philosophy being written but whether we
have a wealth or a "glut" (1980 , p. 284) and whether there is any vital dia-
logue. Many believe that there is not only a need for more "organizational
boundary crossing" between political science and political philosophy (Barry,
1977, p. 299) but "a need to improve the quantity and quality of discussion
across the different schools that have gradually established enclaves in
political theory" (Nelson, 1983, p. 24).
One prediction that seems totally safe is that PT and pt will never again
be one, and it is unlikely that either will become a more concentrated field.
The latter is an interdisciplinary body of literature that shows no signs of con-
solidation, and most of the sentiment seems clearly in favor of pluralism.
Everybody on record is hoping for mutual interaction and cross fertilization ,
but few seem to care if it is actually effected. Since much of PT is now a
reflection of the structure, or lack of structure, in pt, a more central focus
there seems just as unlikely. If we look within PT for a core of theory directed
toward the empirical study of politics in the mode of the behavioral vision, it
seems to be very close to disappearing altogether. But there are some who still
hold out hope for a unified center.
Most commentators recognize that in one way or another post-
behavioralism also means post-positivism. The question is whether there is a
life for scientific theory "beyond positivism. " Elinor Ostrom suggests that "we
are coming to the end of an era in political science, a slow, whimpering end"
which indicates that "the hoped for cumulation of knowledge into a coherent
body of theory has not occurred" (pp. 11 , 13). Individuals such as Ostrom,
however, continue to stress "the need for the development of theory as the
 john G. Gunnell I 37

basis of our discipline." This, however, would be a more eclectic notion of

theory that would transcend both "the naive acceptance" of positivism and
some of the narrow approaches to empirical analysis that it was used to justify
and which are relevant to modem political issues. For Ostrom, as for a
number of others who have celebrated the policy tum in contemporary
political science, the demands of both science and relevance seem to be met
best in the application of economic models involving rational choice and col-
lective choice and the meta theoretical explication of such models. But, again,
there is little clear evidence at this point that a unifying vision of political
theory will or should emerge from this material (Moon, 1983).
Unity is still the message of most presidential addresses. Warren Miller
predicts unification based on the research methods developed since World
War II and claims that only problems of inadequate personnel and funding
now stand in the way of bringing "political science into a new age of intellec-
tual ferment and maturity as a discipline" which can at once "make a massive
contribution to the welfare of the nation while evolving into a conceptually
coherent scholarly enterprise" (1981, pp. 14, 15). A basic fact of intellectual
continuity in the field is the belief that we belong to an unfinished discipline
in which an imminent theoretical breakthrough will not only verify the dis-
cipline as a science but demonstrate its relevance to public policy. The ques-
tion is exactly what is to be the form of interaction between public and
academic discourse and the vehicle of that interaction. We can see what Mer-
riam, Wilson, Beard, Bentley, Lasswell and others gave for answers to these
questions, and we might well ask what , precisely, is the answer of the modem
political scientist?
A survey of the presidential addresses to the APSA , as well as various
other intellectual signals in the literature, seems to suggest a liaison between
political scientists and policy elites that would take place through consulta-
tion, the movement of political scientists in and out of the actual policy pro-
cess, and the influence of journals such as the Publ£c Interest. Although the
role is not perceived as necessarily uncritical, the principal thrust is more in
the direction of service than reform , let alone anything resembling a more
basic critique of political institutions. The mood, for the most part, is
conservative.
There are, however, some other directions in public policy analysis and
the relationship between political science and politics. First of all , it is not
merely in Critical Theory that a vision of a critical social science is present
(Connolly, 1981). There are even some obvious shifts within what might be
taken as the political science establishment. Charles Lindblom, in his 1981
Presidential Address to the APSA, recognized and endorsed the claims of the
dissenting academy. He claimed that the discipline as a whole and most of the
major figures in the behavioral movement had largely accepted and rein-
forced the initial premises, research programs, and methods of "a complacent
view of the liberal democratic process, government, and state" (1982, p. 9).
Even many erstwhile critics, he suggested, are actually "committed to the con-
ventional view and give little sustained analytical attention to the radical
model" which is manifest in contemporary neo-Marxist thought and other
elements of contemporary political theory such as "phenomenology,
 38 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD
hermeneutics, interpretative theory, and critical theory" (pp. 12, 20).
If the history of political science provides any clue, the argument will be
noted, but the direction of the discipline will not be fundamentally altered.
Whatever the impact of critical notions of political theory and political
science on the discipline , the very fact that they remain in the discipline will
probably ensure that such discourse remains ultimately academic. Through
the medium of education and osmosis, it may have some practical effect but
not the form of engagement often implied. There are some things happening
in the world of political theory that obviously suggest a more definite idea
about how to make political theory political or at least make contact between
political theory and politics. A venture such as that represented by Sheldon
Wolin's involvement with the journal Democracy deserves consideration.
There is an assumption that political theory must, or should, reach a
wider public audience and that there must be a mediating mode of discourse.
Exactly what this means, or can mean, is more difficult to say. It does not
seem that the basic purpose is to lead or support what might be termed move-
ment politics. Neither is it principally an attempt to enter directly the activity
of what might be termed conventional politics. It might be more accurately
understood as an attempt to evoke in a wide range of educated individuals, in
a variety of contexts , a sense of public consciousness and virtue that
transcends the mundanely political but which is not confined to the academic
world. Yet the project remains bound within the ambiguities of the relation-
ship between political and academic discourse.
The extent to which practical concerns, and the attempt to make such
concerns effective, will have an impact on the discipline of political science
and the direction of political theory, and, in tum on politics, is a definite
question for the 1980s. Archaeological analysis tends to produce skepticism,
since it demonstrates the inevitability of mortality and the demise of the
present. Digging into the past of American political science is no exception.
Furthermore, the history of the sub-field of Political Theory does not demon-
strate any great ability to transcend American political culture. As Norman
Jacobson has noted, the idea of a social science as the embodiment of a pro-
vocative vision has not been indigenous to American political science. Today
political theory is bringing many new ideas into the discipline, but a study of
the past indicates "that the genius of American social and political science has
been less the cultivation of novel ideas than their domesticator" (1982 , p. 6).
This has been the case in the sub-field of Political Theory.

REFERENCES
Almond , Gabriel. Political theory and political science. Ameri"can Political Sdence
R eview, 1966, 60, 869-879.
Almond , Gabriel. Clouds, clocks, and the study of politics. World Politics, 1977 , 29,
489-522 .
Anderson, William. Political science north and south. journal of Politics, 1949, 11,
298-317 .
Appel, K-0 . Analytic philosophy of language and the Geisteswissenschaften.
Dordrecht, Holland: D. Reidel, 1967.
 john G. Gunnell I 39

Apter, David E. Theory and the study of politics. American Political Review, 1957,
51 , 747-62.
Arendt, Hannah. The human condi.tion. New York: Doubleday, 1958.
Arendt, Hannah. Between past and future. New York: Viking, 1961.
Ashcraft, Richard. Revolutionary politics and Locke's two treatises of government.
Poli.ti"cal Theory, 1980, 8, 429-86.
Bachrach, Peter. The theory of democrati"c elitism. New York: Lieber-Atherton, 1967.
Bachrach, Peter & Baratz, Morton S. Two faces of power. Amen·can Political Science
Revi·ew, 1962, 58, 947-52.
Barber, Benjamin (Ed.). Political theory in the 1980s: Prospects and topics. Poli"tical
Theory, 1980, 9, 291 -424.
Barry, Brian. The strange death of political theory. Government and Opposition,
1980, 15, 276-78.
Barry, Brian. Do neighbors make good fences? Poli"tical Theory, 1981, 9,
293-301 .
Bell, Daniel. The end of ideology. New York: Free Press, 1960.
Berelson, Bernard & Steiner, Gary. Human behavi·or. New York: Harcourt, Brace,
1964.
Berlin, Isaiah. Does political theory still exist? In Peter Laslett & W . G. Runciman
(Eds.). Phi'losophy, society and poli"tics. New York: Barnes and Noble, 1962.
Bernstein, Richard. The restructuring of poli"tical and social theory. New York:
Harcourt, Brace, Jovanovich, 1976.
Bluhm, William T. Theon·es of the political system. Englewood Cliffs, N .J .: Prentice-
Hall, 1965.
Bluhm, William T. Introduction: a call for paradigm synthesis. In Bluhm (Ed.). The
paradigm problem in political science. Durham, N.C .: Carolina Academic Press,
1982.
Bluntschli, J. K. The theory of the state. Oxford: Clarendon Press, 1885.
Braybrooke, David (Ed.). Philosophical problems of the social sciences. New York:
Macmillan, 1965.
Brecht, Arnold. Beyond relativism in political theory. Amen·can Political Science
Review, 1947, 41, 470-88.
Brecht, Arnold. Political theory. Princeton: Princeton University Press, 1959.
Brodbeck, May (Ed.) . Readings in the philosophy of the social sciences. New York:
Macmillan, 1968.
Brown, Robert. F.xplanation in social science. Chicago: Aldine, 1963.
Campbell, Angus et al. The American voter. New York: Wiley, 1960.
Catlin, George. The story of the political philosophers. New York: McGraw-Hill, 1939.
Charlesworth, James C. (Ed.). The limits of behavi·oralism. Philadelphia: American
Academy of Political and Social Science, 1962.
Charlesworth, James C. A design for political science: Scope, objectives, and methods.
Philadelphia: American Academy of Political and Social Science, 1966.
Cobban, Alfred. The decline of political theory. Political Science Quarterly, 1953,
68, 321-37.
Cobban, Alfred. In search of humanity. New York: G. Braziller, 1959.
Cohen, Morris R. Reason and nature. New York: Harcourt , Brace, 1931.
Cohen, Morris & Nagel, Ernest. An introduction to logic and scientific method. New
York: Harcourt, Brace, 1934.
Connolly, William . Political science and ideology. New York: Atherton, 1967.
Connolly, William (Ed.). The bias of pluralism. New York: Atherton, 1969.
Connolly, William. Appearance and reality. Cambridge: Cambridge University Press,
1981.
Crick, Bernard. The Amen·can science of politics. Berkeley: University of California
Press, 1959.
 40 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD

Dahl, Robert. Political theory: truth and consequences. World Politics, 1958, 11,
89-102.
Dahl, Robert. Who governs? New Haven: Yale University Press, 1961.
Dallmayr, Fred & McCarthy, Thomas (Eds.). Understanding and socz"al inquiry.
Notre Dame, IN : University of Notre Dame Press, 1977 .
Davis, Lane. The cost of realism. Western Poli"tical Quarterly, 1964, 17, 37-46.
de Grazia , Alfred. Preface to four essays on the relation of political theory to research.
journal of Politics, 1951 , 13, 35.
Deutsch, Karl. The nerves of government. New York: Free Press, 1963.
Deutsch, Karl. On political theory and political action . Ameri·can Political Science
R eview, 1971, 65, 11-27 .
Deutsch, Karl & Rieselbach , Leroy. Recent trends in political theory and political
philosophy. Annals of the A meri·can Academy of Political and Social Science, 1965,
360, 139-62.
Driscoll, Jean M. & Hyneman, Charles S. Methods for political scientists. American
Political Science Review, 1955, 49, 192-217 .
Duncan, Graeme & Lukes, Steven. The new democracy. Political Studies, 1963, 11,
156-177.
Dunning, William A. A history of political theori·es (3 Vols .). New York: Macmillan,
1902, 1905 , 1920.
Dworkin, Ronald. Taking rights seri·ously. Cambridge: Harvard University Press,
1977.
Easton, David. The decline of modern political theory. journal of Politics, 1951, 13,
36-58.
Easton, David. The political system. New York: Knopf, 1953.
Easton, David. A framework for political analysis. Englewood Cliffs, N .J.: Prentice-
Hall , 1965(a).
Easton, David. A systems analysis of political life. New York: Wiley, 1965(b).
Easton, David (Ed.). Vari·eties of politi"cal theory. Englewood Cliffs, N.J.: Prentice-
Hall, 1966.
Easton, David. Political science. In David L. Sills (Ed .). Internati"onal encyclopedia of
the social scz"ences (Vol. 12). New York: Macmillan, 1968.
Easton, David. The new revolution in political science. Ameri·can Political Scz"ence
Review, 1969, 63, 1051-61.
Eckstein, Harry. Political theory and the study of politics. Ameri·can Political Scz"ence
Review, 1956, 50, 475-87.
Eldersveld, Samuel. Theory and method in voting behavior.journal of Politics, 1951,
13, 70-87.
Eulau, Heinz . The skill revolution and consultative commonwealth. Ameri·can
Political Science Review, 1973, 67, 169-91.
Eulau, Heinz. Drift of a discipline. Ameri·can Behavioral Scientist, 1977, 21, 3-10.
Flathman, Richard. The public interest. New York: Wiley, 1966.
Flathman, Richard. Political obligation. New York: Atheneum, 1972.
Flathman, Richard. The practice of rights. Cambridge: Cambridge University Press,
1977.
Ford, Henry Jones. The scope of political science. Proceedings of the American
Political Science Assocz"ation, 1906.
Freeman, Michael & Robertson, David (Eds.). The frontiers of political theory. New
York: St. Martin's Press, 1980.
Frohock, Fred M. The nature of politi"cal inquiry. Englewood Cliffs, N.J.: Prentice-
Hall, 1967.
Germino, Dante. The revival of political theory.journal of Politics, 1963, 25, 437-60.
Gettell, Raymond G. Introduction to political scz"ence. Boston: Ginn and Co., 1910.
 john G. Gunnell I 41

Gettell, Raymond G. Nature and scope of present political theory. Proceedings of the
American Polz'tical Science Association, 1914.
Gettell, Raymond G. History of political thought. New York: Century, 1924.
Gibson, Quentin. The logic of social inquiry. New York: Humanities, 1960.
Glaser, William A. The types and uses of political theory. Social Research, 1955, 22,
275-96.
Goodnow, Frank J. The work of the APSA. Proceedings of the Amen·can Political
Science Assodation, 1905.
Graham, George]. & Carey, George W . The post-behavioral era. New York: McKay,
1972.
Graham, George J. A critical overview: The present status of the synthesis question. In
William T . Bluhm (Ed.). The paradigm problem in political science. Durham,
N .C.: Carolina Academic Press, 1982.
Green, Philip & Levinson, Sanford (Eds.). Power and community. New York: Vintage
Books, 1970.
Greenstein, F. & Polsby, N . (Eds.). Handbook of political science. Reading,
MA: Addison-Wesley, 1975.
Greer, Scott. The logic of social inquiry. Chicago: Aldine, 1969.
Gregor, James A. An introduction to metapolitics. New York: Free Press, 1871.
Gunnell, John G. Social science and political reality: The problem of explanation.
Social Research, 1968, 34, 159-201.
Gunnell, John G. Deduction, explanation , and social scientific inquiry. Amen·can
Politi'cal Science Review, 1969, 63, 1233-46.
Gunnell, John G. Philosophy, science, and political inquiry. Morristown, N.J.:
General Leaming Press, 1975.
Gunnell, John G. Political theory: tradition and interpretation. Boston: Little-Brown,
1979.
Habermas, Jurgen . Knowledge and human interests. Boston: Beacon, 1971.
Hacker, Andrew. Capital and carbuncles: The great books reappraised. Amen'can
Political Science Review, 1954, 48, 775-86.
Haddow, Anna. Political science in Amen·can colleges and universities, 1636-1900.
New York: Appleton Century, 1939.
Hallowell, John H. Politics and ethics. Amen'can Political Science Review, 1944, 38,
639-55 .
Hanson, N . R. Patterns of discovery. Cambridge: Cambridge University Press, 1958.
Hare, R. M. The language of morals. Oxford: Oxford University Press, 1952.
Hare, R . M. Freedom and reason. Oxford: Oxford University Press, 1963.
Hempel, Carl. Aspects of scientific explanation. New York: Free Press, 1965.
Herbert, Wray. Political theory. Humanities Report, 1981, J, 4-9.
Horowitz , Irving (Ed.). Power, politics, and people: The collected essays of C. W.
Mills. New York: Ballantine Books, 1963.
Irish, Marian D. Advance of the discipline. journal of Politics, 1968, JO, 291-310.
Isaak, Alan. Scope and methods of political science. Homewood, Ill.: Dorsey Press,
1969.
Jacobson, Norman. The unity of political theory. In Roland Young (Ed.). Approaches
to the study of politics. Evanston, Ill.: Northwestern University Press, 1958.
Jacobson, Norman. Remarks. Presented at APSA annual meeting, Denver, Colo.,
1982.
Jaffa, Harry. The case against political theory.journal of Politics, 1960, 22, 259-75.
Jenkin, Thomas. The study of political theory. New York: Doubleday, 1955.
Kaplan, Abraham. The conduct of inquiry. San Francisco: Chandler, 1964.
Kariel, Henry. The decline of American liberalism. Stanford: Stanford University
Press, 1961.
 42 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD
Karie!, Henry. Open systems. Itasca, Ill .: Peacock, 1969.
Kateb , George. Political theory: Its nature and uses. New York: St. Martin's Press,
1968.
Kateb, George. The condition of political theory. American Behavioral Scientist,
1977, 21 , 135-59.
Kaufman, Felix. Methodology of the social sciences. New York: Oxford University
Press, 1944.
Key, V. 0 ., Jr. The state of the discipline. American Political Science Review, 1958,
52, 961-71.
Kress, Paul. Against epistemology.journal of Politics, 1979, 41, 526-42 .
Kuhn, Thomas. The structure of scientific revolutions. Chicago: University of Chicago
Press, 1962.
Lane, Robert. Politics of consensus in an age of affluence. American Political Science
Review, 1965 , 59, 874-95.
Lane, Robert. The decline of politics and ideology in a knowledgeable society.
American Sociological Review, 1966, 31, 649-62.
LaPalombara, Joseph. Decline of ideology: A dissent and interpretation. American
Political Science Review, 1966, 60, 5-16.
Laslett, Peter (Ed.). Philosophy, politics and society. New York: Barnes and Noble,
1956.
Laslett, Peter. Introduction. In Peter Laslett & W. G. Runcirnan (Eds.). Philosophy
politics and society. New York: Barnes and Noble, 1967.
Laslett, Peter & Fishkin, James (Eds.). Philosophy, politics and society. New Haven :
Yale University Press, 1979.
Lasswell, Harold. Discussion. American Politi"cal Science R eview, 1950, 44, 422-25.
Lasswell, Harold. Psychopathology and politics. Chicago: Chicago University Press,
1930.
Lasswell, Harold . Democracy through public opinion. Menasha, Wis.: George Banta,
1941.
Lasswell, Harold & Kaplan, Abraham . Power and society. New Haven: Yale Univer-
sity Press, 1950.
Leiserson, Avery. Charles Merriam, Max Weber, and the search for synthesis in
political science. Amerz"can Political Science R eview, 1975, 69, 175-85.
Lindblom, Charles E. Poli"tics and markets. New York: Basic Books , 1977 .
Lindblom, Charles E. Another state of mind. American Political Science Review,
1982, 76, 9-21.
Lippincott, Benjamin. The bias of American political science. journal of Politics,
1940, 2, 125-39.
Lippincott, Benjamin. Political theory in the United States. Contemporary Political
Science. Paris: UNESCO, 1950.
Lipset, Seymour M. Political man. Garden City: Doubleday, 1960.
Louch , A. R. &planation and human action. Berkeley: University of California Press,
1966.
Lowi, Theodore. The end of liberalism. New York: Norton, 1969.
Macintyre, Alasdair. A mistake about causality in social science. In Peter Laslett &
W . G. Runciman (Eds.). Philosophy, politics and society. Oxford: Basil Blackwell,
1962.
Marcuse, Herbert. Reason and revolution. New York: Oxford, 1941.
Marcuse, Herbert. Reason and revolution. New York: Oxford University Press, 1941.
McCoy, Charles & Playford, John (Eds.). Apolitical poli"tics. New York: Thomas Y.
Crowell, 196 7 _
McDonald, Neil A. & Rosenau , James N. Political theory as an academic field and
intellectual activity. journal of Politics, 1968, 30, 311-44.
 John G. Gunnell I 43

Meehan, Eugene. The theory and method of political analysis. Homewood , Ill. :
Dorsey Press, 1965 .
Merriam, Charles. The present state of the study of politics. American Political
Science Review, 1921, 15, 173-85.
Merriam, Charles. Progress report of the committee on political research. American
Political Science Review, 1923, 17, 274-312.
Merriam, Charles. Report of the national conference on the science of politics.
American Political Science Review, 1924, 18, 119-66.
Merriam, Charles. New aspects of politics. Chicago: Chicago University Press, 1925.
Miller, Eugene F. Positivism, historicism, and political inquiry. American Political
Science Review, 1972, 66, 796-817 .
Miller, Warren E. The role of research in the unification of a discipline. American
Political Science Review, 1981, 75, 9-16 .
Mills, C. Wright. The power elite. New York: Oxford University Press, 1956.
Moon, Donald]. The logic of political inquiry. In F. Greenstein & N . Polsby (Eds.).
Handbook of political science (Vol. I). Reading, Mass.: Addison-Wesley, 1975.
Moon, Donald J. Interpretation, theory and human emancipation. In Elinor Ostrom
(Ed.) . Strategies of political inquiry. Beverly Hills: Sage, 1982.
Moore, Barrington, Jr. Political power and social theory. Cambridge: Harvard Univer-
sity Press, 1958.
Nagel, Ernest. The structure of science. New York: Harcourt, Brace and World, 1961.
Natanson, Maurice (Ed.). Philosophy of the social sciences. New York: Random
House, 1963.
Nelson, John S. Natures and futures for political theory. In Nelson (Ed.). What should
politz"cal theory be now? Albany: State University of New York Press, 1983.
Nozick, Robert. Anarchy, state and utopia. New York: Basic Books, 1974.
O'Connor, James. The fiscal crisis of the state. New York: St. Martin's Press , 1973.
Ostrom, Elinor. Beyond positivism. In Ostrom (Ed.). Strategies of political inquiry.
Beverly Hills: Sage, 1982.
Partridge1 P.H. Politics, philosophy, ideology. Political Studies, 1961 , 9, 217-35.
Pitkin, Hanna. The concept of representation. Berkeley: University of California
Press, 1967.
Pitkin, Hanna. Wittgenstein and justice. Berkeley: University of California Press,
1972.
Pocock, J. G. A. Politics, language, and time. New York: Atheneum , 1971.
Polsby, Nelson. Community power and political theory. New Haven: Yale University
Press, 1963.
Pool, Ithiel de Sola. Foreword. In Pool (Ed.), Contemporary political science: Toward
empirical theory. New York: McGraw-Hill, 1967 .
Popper, Karl. The logic of scientific discovery. New York: Science Editions, 1961.
Ranney, Austin, 1976. The divine science of politics: Political engineering in Ameri-
can culture. American Political Science Review, 1976, 70, 140-48 .
Rawls, John. A theory of justice. Cambridge: Harvard University Press, 1971.
Redford, Emmett S. Reflections on a discipline. American Political Science Review,
1961 , 55, 755-62 .
Research in political behavior. American Political Science Review, 1952, 46, 1003-45 .
Research in comparative politics. American Poli"tical Science Review, 1953, 47,
641-679.
Richter, Melvin. Introduction. In Richter (Ed .), Political theory and political educa-
tion. Princeton: Princeton University Press, 1980.
Riker, William H. The future of a science of politics. American Behavioral Scientist,
1977 , 21, 11-38.
Riker, William H. Political theory and the an of heresthetics. (This volume) , 1983.
44 I POLITICAL THEORY: THE EVOLUTION OF A SUB-FIELD

Rousseas, Stephen & Farganis, James. American politics and the end of ideology.
British journal of Sociology, 1963, 14, 347-62.
Rudner, Richard. Philosophy of social science. Englewood Cliffs, N .J.: Prentice-Hall,
1966.
Runciman, W. G. Social science and political theory. Cambridge: Cambridge Univer-
sity Press, 1963.
Sabine, George. A history of political theory. New York: Holt, Rinehart, and Winston,
1937.
Sabine, George. What is a political theory? journal of Politics, 1939, 1, 1-16.
Schaar, John H. & Wolin, Sheldon S. Essays on the scientific study of politics: A
critique. American Political Sdence Review, 1963, 57, 125-50.
Schutz, Alfred. The phenomenology of the sodal world. Evanston, Ill.: Northwestern
University Press, 1967 .
Shklar, Judith. After utopia. Princeton: Princeton University Press, 1957.
Simon, Herbert. Discussion. Ameri"can Political Science Review, 1950, 44, 407-11.
Simon, Herbert. Replies and comments. American Political Science Review, 1952, 46,
494-96.
Skinner, Quentin. Meaning and understanding in the history of ideas. History and
Theory, 1969, 8, 3-53.
Skocpol, Theda. States and sodal revolutions. Cambridge: Cambridge University
Press, 1979.
Smith, David G. Political science and political theory. American Political Science
Review, 1957, 51, 734-46 .
Smith , Munroe. The domain of political science. Political Science Quarterly, 1886, 1,
1-8.
Smithburg, Donald W . Political theory and public administration.journal of Politics,
1951, 13, 59-69.
Somit, Albert & Tanenhaus, Joseph. The development of American political science.
New York: Irvington, 1982.
Spragens, Thomas A., Jr. The dilemma of contemporary political theory. New York:
Dunellen, 1973.
Storing, Herbert J. (Ed.). Essays on the scientific study of politics. New York: Free
Press, 1962.
Strauss, Leo. Natural right and history. Chicago: University of Chicago Press, 1953.
Strauss, Leo. What is political philosophy? Glencoe, Ill.: Free Press, 1959.
Strauss, Leo & Cropsey, Joseph (Eds.). History of political theory. Chicago: Rand
McNally, 1963.
Tarlton, Charles D. A rope of sand: Interpreting Locke's first treatise of government.
The Histori"caljournal, 1979, 21, 43-74.
Taylor, Charles. The explanation of behavior. New York: Humanities, 1964.
Taylor, Charles. Neutrality in political science. In Peter Laslett & W . G. Runciman
(Eds .). Phz"losophy, politics and society. New York: Barnes and Noble, 1967.
Thorson, Thomas L. Political values and analytic philosophy. journal of Politics,
1961, 23, 711-24.
Thorson, Thomas L. The logic of democracy. New York: Holt, Rinehart and
Winston, 1962.
Toth, Kathleen. Method and value: Lessons from the odyssey of political theory.
Presented at American Political Science Association annual meeting, Denver,
Colo. , 1982.
Toulmin, Stephen. An examination of the pface of reason in ethics. Cambridge:
Cambridge University Press, 1950.
Truman, David B. Disillusion and regeneration: The search for a discipline. American
Political Sdence R eview, 1965 , 59, 865-73.
 john G. Gunnell I 45

Van Dyke, Vernon. Politz"cal science: A philosophical analysis. Stanford, Calif.:

Stanford University Press, 1960.
Voegelin, Eric. The new science of politics. Chicago: University of Chicago Press,
1952.
Voegelin, Eric. Order and history (Vol. I). Baton Rouge : Louisiana State University
Press, 1956.
Wahlke, John C. Pre-behavioralism in political science. American Political Science
Review, 1979, 73, 68-77.
Waldo, Dwight. Political science in the United States: A trend report. Paris:
UNESCO, 1956.
Walker, Jack L. A critique of the elitist theory of democracy. American Political
Science Review, 1966, 60, 285-95.
Waxman, Chaim (Ed.) The end-of-ideology debate. New York: Simon and Schuster,
1969.
Weldon, T. D. The vocabulary of polz"tics. London: Penguin, 1953.
Weldon, T. D . Political principles. In Peter Laslett (Ed.). Philosophy, politics and
society. New York: Barnes and Noble, 1956.
White , Leonard D. Political science, mid-century.Journal of Politics, 1950, 12, 13-19.
Whyte, William Foote. A challenge to political scientists. American Political Science
Review, 1943, 37, 692-97.
Willoughby, W. W. An examination of the nature of the state; A study in polz"tical
philosophy. New York: Macmillan, 1896.
Willoughby, W. W. The political science association. Political Science Quarterly,
1904, 19, 107-11.
Wilson, Francis G. The work of the political theory panel. American Political Science
Review, 1944, 38, 726-31.
Winch, Peter. The idea of a social science. London: Routledge and Kegan Paul, 1958.
Wolfe, Alan. The limits of legitimacy. New York: Free Press, 1977.
Wolfe, Alan & Surkin, Marvin (Eds.). An end to political science. New York: Basic
Books, 1970.
Wolff, Robert Paul. The poverty of liberalism. Boston: Beacon, 1969.
Wolin, Sheldon S. Politics and vision. Boston: Little-Brown, 1960.
Wolin, Sheldon S. Political theory: Trends and goals. In David L. Sills (Ed.). Inter-
national encyclopedia of the social sciences (Vol. 12). New York: Macmillan, 1968.
Wolin, Sheldon S. Political theory as a vocation. American Political Science Review,
1969, 63, 1062-82.
Woolsey, Theodore D. Political science; Or, the state theoretically and practically
considered. New York: Scribner, Armstrong, 1878.
 2
Political Theory and the
Art of Heresthetics
Wz"llz"am H. Rz"ker
My assignment for this essay is to review the present state of the field of
political theory. This I will do by examining, first, the main line of its
development during the last generation. Then I will look to the future, not
only at the continuation of this main line of scholarship, but also at some
possibly fruitful new directions, as, for example, the art of heresthetics.
It is necessary in the beginning, because of a usage peculiar to political
science, to specify exactly what political theory is. In most· scientific disci-
plines, the word "theory" refers to a set of deductively related sentences that
together describe the portion of the world studied with a particular discipline.
Etymologically this is correct; "theory" weds observation and contemplation.
In political science, however, the word "theory" has come to refer primarily to
moral philosophy. While I recognize the necessity and importance of moral
philosophy for political studies, in this paper I intend to use the word "theory"
as it is used in other sciences. Given our tradition, this is rather difficult
because, until recently, there has been no set of sentences about politics to
which "theory" could be applied with etymological justification. Beginning,
however, with the publication of Duncan Black's essay, "On the Rationale of
Group Decision Making" Uournal of Politi"cal Economy, 1948, although the
main ideas were, as noted in Coase, 1981, worked out six years earlier), there
has existed a deductive and testable theory about political events. Political
theory developed gradually through the next two decades, and during the
1970s it came of age. It was the subject of books, essays, conferences, and
convention panels, and now even a summary essay in a book on the state of
the art.

I
The content of political theory is the authoritative allocation of values,
following Easton's (1954) definition of politics, in that it involves the
amalgamation of individual preferences into a social choice and the sub-
sequent enforcement of the result. At this quite general level, the goal of
political theory is to identify the conditions for an equilibrium of preferences.
Such an equilibrium is a social choice that the members of every sub-group in
the society that are capable of bringing about a social decision prefer to any

47
48 I POLITICAL THEORY AND THE ART OF HERESTHETICS
other alternative. This equilibrium is one the society will arrive at for certain,
regardless of its particular institutions; and, if by reason of some obstruction
the society is deflected from it or forced to abandon it, the society nevertheless
will return to it if the obstruction is removed. Equilibrium of this sort is there-
fore a generally accepted outcome which is, moreover, self-enforcing because
of the desire of individuals to arrive at it.
The goal of identifying the conditions for equilibrium is closely parallel
to the goals of other scientific disciplines, both social and physical, in which
the specification of conditions for equilibria is a central theme. The history of
political theory over the last forty years has consisted largely of increasingly
precise specification and increasingly deeper analysis of equilibrium
conditions. 1
As previously noted, Duncan Black (1948, 1958) began the construction
of theory by looking for an equilibrium of majority rule. He soon rediscovered
the paradox of voting (initially discovered by Condorcet, 1785, and then
subsequently overlooked), the significance of which is that, when it occurs in a
set of individual preference orders, it guarantees that there will be no major-
ity winner (i.e., disequilibrium) within that set. Looking for a way around the
paradox, Black devised the condition of the single-peakedness of a set of in-
dividual preference orders as a sufficient- but not necessary- condition for
the avoidance of the paradox and thus for the guarantee of an equilibrium.
This was an excellent beginning. Among its other virtues, the condition of
single-peakedness is highly practical because it allows for ready specification
of the substantive content of an equilibrium outcome as the median voter's
most preferred alternative. 2 Since it is often easy to identify the median voter,
it is also easy in many analyses of public policy and political conflict to
spell out the concrete details of the equilibrium social choice. As might be
expected, this is the feature of Black's work most widely used in practical
political and economic studies.
For the development of political theory, however, a more significant vir-
tue of the condition of single-peakedness is that it has an obvious political
interpretation. It reveals that an equilibrium among disputing voters rests on
an underlying agreement among them about the dimension of political con-
flict. The fact that equilibrium rests, in this case, on some kind of partial
agreement has led to the specification of a variety of other conditions of
equilibrium, all based on some kind of similarity of preferences, for example,
value restrictedness, extremal restriction , limited agreement, separability,
etc. (See Fishburn, 1970, for a method of classifying and generating all possi-
ble conditions for equilibrium.)
A third and, for theory-building, perhaps most valuable feature of the
condition of single-peakedness is that, once constructed out of the arrange-
ment of preferences on a dimension of judgment, it may appropriately be
used to inquire into the fragility of the condition. This Black and Newing
(1951) did by analyzing the effect of introducing a second standard of judg-
ment into a committee of three voters. In this case, they managed to find
some special equilibria by imposing particular agendas (e.g., sequential
voting on first one dimension, then another, then back to the first, et seq.).
But, in the general case, the similarity of tastes required for equilibrium
 William H. Riker I 49

almost entirely disappeared, even with as few as three voters. The net effect of
Black's work was, therefore, to identify conditions for equilibria in special
cases, such as single-peakedness, while at the same time generating con-
siderable doubt about the possibility of discovering a general equilibrium.
Arrow (1951) confirmed these doubts by demonstrating the possibility of
outcomes analogous to the paradox of voting in any (unspecified) method of
amalgamating individual preference. He proved that any such method, satis-
fying-as does majority rule-some elementary conditions of fairness , suffers
from the same kind of defect as majority rule, namely the admission of some
cases in which individually transitive preference orders amalgamate into an
intransitive social outcome. Many writers have modified or explicated Arrow's
theorem in order to soften its impact-for a survey, see Riker (1982a , pp.
123-136)- but his main point remains valid: There is an inherent conflict
between voter-dominated procedures and logically consistent outcomes. For
political applications, where it is necessary to choose a unique winner, it may
be that consistency does not greatly matter. But as Arrow (1963) himself
recognized and as Plott (1973) emphasized, failures of consistency mean that
the outcome is not independent of the procedure by which it is reached.
Hence random variations in procedure imply random variations in outcomes,
which is just exactly what is meant by a disequilibrium of tastes. Hence
Arrow's theorem may be regarded as a generalization-to all methods of
amalgamating tastes-of the fundamental disequilibrium that Black per-
ceived in majority rule. The various elaborations, refinements, and modifica-
tions of Arrow's theorem serve to emphasize the ineradicability of the poten-
tial for disequilibrium; Gibbard's (1968) theorem-that all methods of voting
to amalgamate tastes or values can be manipulated by sophisticated
voting-reveals one concrete way in which the disequilibrium may be
effected.
The fact that the potential for disequilibrium is ineradicable implies
nothing, however, about the frequency of disequilibrium. Though ineradica-
ble, it might-or might not-be so rare as to be insignificant. One line of in-
vestigation on this theme is to calculate the expected frequency of non-
equilibrium outcomes, given that voters pick preference orders at random
and equiprobably (Niemi & Weisberg, 1968; Garman & Kamien, 1968;
DeMeyer & Plott, 1970; Gehrlein & Fishburn, 1976). The conclusion is that,
as the numbers of alternatives and participants increase, the expectation of
disequilibrium very quickly approaches infinity, although just a bit of prior
agreement on values lowers the expectation of disequilibrium remarkably
(Niemi, 1969; Fishburn, 1973b). The usefulness of this kind of inquiry is
severely limited, however, because the assumption of equiprobability or of
any particular degree of agreement in picking preference orders is arbitrary
and unrealistic, especially when it may be worthwhile for participants to
manipulate the distribution of preference orders.
A more fruitful line of investigation about the frequency of dis-
equilibrium is to establish the conditions for equilibrium under a variety of
assumptions. In one sense this is an examination of the robustness of condi-
tions like single-peakedness, which seem initially to guarantee a considerable
amount of equilibrium in one dimension of choice. However, Kramer (1973)
 50 I POLITICAL THEORY AND THE ART OF HERESTHETICS
examined these conditions for cases of more than one dimension, when voters'
tastes can be represented by quasi-concave differentiable utility functions,
and showed that, with even a modest amount of heterogeneity of tastes,
majority rule breaks down completely. Kramer's work in effect demolished
various special guarantees of equilibrium.
Another possibility is to define what is required for equilibrium. This
direction of inquiry has led to an even more dramatic conclusion: equilibrium
is nearly impossible. Black and Newing (1951) began the inquiry, as already
noted above, by studying the condition for majority rule when three voters
used two standards of judgment. They showed that even in quite constrained
circumstances a sufficient condition for equilibrium was difficult to satisfy.
Plott (1967) elaborated on this result by defining a sufficient condition for an
equilibrium of majority rule when m( ;;;i:3) voters use n( ;;;i:2) standards of judg-
ment to select among continuous alternatives. Plott's condition (see also
Cohen, 1979) turns out to be so restrictive that even as few as five voters and
two standards render disequilibrium almost certain. Schofield (1978b), tak-
ing a superficially different tack, defined a condition for an alternative to be
undefeatable by any other alternative in its neighborhood. This condition is
also extraordinarily restrictive, revealing that equilibrium is highly unlikely,
3
unless tastes are extremely similar, and, indeed, is impossible when n ;;;i: m; •
The coup de grace for the expectation of equilibrium was supplied by
McKelvey (1976) when he looked into the possibility of a small set of alter-
natives which together might be in equilibrium. The motivation for this
research was the idea that, in a case with no majority-preferred alternative
(i.e., disequilibrium), a small set-call it T-of similar alternatives might,
even though themselves in a cycle, each beat all alternatives not in T. If so,
there would be a kind of equilibrium outcome, not unique, of course, but
nevertheless stable and predictable in the sense that it would always be a
member of the set T, the "top cycle." McKelvey showed, however, that, given
continuous alternatives and at least two standards of judgment, the set T is
either a Plott equilibrium (which is very rare) or all points in the space. This
means that any possible outcome is a feasible outcome, which is complete
disequilibrium. 3
This is where the matter now stands. (See Fiorina & Shepsle, 1982;
Aldrich & Rohde, 1982; and Schofield, 1982b.) In the abstract case, with no
institution assumed and the fewest possible restrictions on individuals' tastes
or values, there is no reason to expect equilibrium. The generality and
significance of this conclusion is sometimes misunderstood because it is
believed to apply only to situations in which outcomes are chosen by voting.
But on the contrary, the application is universal. If there is a majority pre-
ferred social policy, then, however it may be prescribed, it is welcomed and
supported by that majority. Suppose it is prescribed by a dictator. That fact
in itself does not make the majority that wants it unwilling to accept it,
although perhaps-and this is a matter oflocal custom - the members of that
majority might be even happier to have prescribed it for themselves. Further-
more, this contented majority is willing to retain the preferred policy, perhaps
even to defend it; this is precisely what is meant by equilibrium. Obviously
 Wz'LUam H. Riker I 51

equilibrium is based primarily on the actual distribution of tastes and only in-
cidentally, if at all, on the process of amalgamation. On the other hand, if
there is no majority preferred alternative, then any alternative chosen as the
socially enforceable outcome is, by definition, disliked by a majority. The dis-
like has nothing to do with the way the outcome is chosen- though commonly
the majority dissidents blame the process that has frustrated them. '.And this
irrelevance is revealed by the fact that, were another outcome to be chosen by
a "better" process, there would still be another dissident majority denouncing
the outcome and probably blaming the process. Thus, even if there is no in-
stitution (such as voting) by which participants can discover that a majority is
dissatisfied, there is still a motive for every one of the individual members of
that majority to search for a "better" outcome. This is the essence of dis-
equilibrium and it derives, not from particular methods of amalgamating
tastes (e.g. , monarchy, oligarchy, democracy) but rather from the distribu-
tion of tastes in society. Hence, the theory developed by Black, Arrow, Plott,
Schofield, and McKelvey is portentously relevant to all politics, regardless of
whether or not particular forms of government involve decision by voting
under majority rule.

II
However general the disequilibrium of tastes may be in theory, 1t IS
obvious that in the real world there do exist many local or partial equilibria
which take the form of recurrent and often reaffirmed outcomes of particular
categories of events. It is the task of political theory to explain the sources of
these local equilibria. I turn now to an assessment of how well theorists have
performed this task.
Inasmuch as something of the same situation (i.e. , a fragile or non-
existent general equilibrium, along with strong partial equilibria) prevails in
the science of economics, it is possible to begin the assessment by establishing
a standard of comparison for the accomplishments of political theory.
Economic theorists have long sought to identify a general equilibrium, quite
consciously ever since the speculations of Walras and Edgeworth and at least
implicitly since the days of Adam Smith. Although the economists' specified
general equilibrium is confined to one institutional setting (i.e ., the competi-
tive market) that, by its structure, contains restraints (e.g., concave indif-
ference curves) not found in the majority rule setting for a political
equilibrium, the goal of a fully identified general equilibrium, divorced from
institutions, has eluded them. All the equilibria defined and elaborated have
stipulated special institutions (such as UUonnement or coalition-formation of
any possible sort, etc.) which, because they involve stochastic variation in
costs of organization and information, fail to admit unqualified prediction or
description of equilibrium outcomes. Certain tentative equilibria can, how-
ever, be identified and theorists can infer certain features of markets from
them. Unfortunately the equilibria identified are extremely fragile, and
probably therefore meaningless, because of the presence of these stochas-
tically variable elements of preference (Weintraub, 1979; Fiorina & Shepsle,
1982). Consequently, the main bulk of economic science deals with partial
 52 I POLITICAL THEORY AND THE ART OF HERESTHETICS

equilibria, especially equilibrium prices for particular commodities as deter-

mined by quantities demanded and supplied. This kind of analysis has been
extremely enlightening because it has, for example, led to non-intuitive
revelations about the counter-productive effects of such policies as tariffs,
price controls, subsidies, and monopolies. In a sense, however, it has been an
easy kind of analysis because it has been based on highly constrained prefer-
ences (e.g., negatively sloping demand curves). The moral is that economics,
clearly the most scientifically successful of the social sciences, has benefitted
from some powerful simplifications not readily available in other social scien-
tific work.
Political scientists, lacking some principle as widely applicable as the law
of demand, have not been as successful as economists in identifying partial
equilibria. Nevertheless, they have identified some, even though it is
necessary for them to overcome more obstacles to theorizing than it is for
economists. To illustrate this success and to point out one extremely impor-
tant common feature of successful theorizing, I shall list three instances of
fairly well described partial equilibria, chosen because they have been the
subject of recent work.
1. Duverger's Law: This is the proposition that plurality voting is
associated with two-party systems, or, to phrase it in terms of partial
equilibria, the equilibrium number of political parties is, given the use of the
plurality rule for deciding on the winner, exactly two. This is a nineteenth
century observation that was eventually provided with a theoretical base in
the form of a rational choice theory of individual behavior, namely that voters
and political supporters choose the candidate with the highest expected value.
In the context of plurality voting, this kind of individual choice eliminates
third parties and thus stabilizes the two-party system. A number of recent
studies have offered convincing evidence of the existence of rational in-
dividual choice in precisely the kind of voting involved in Duverger's Law
(Black, 1978; Cain, 1978; Lemieux, 1977; Benzel and Sanders, 1979). I have
recently surveyed the history of this law to illustrate: (1) its reformulation into
progressively more precise and defensible sentences; and (2) its support with
increasingly persuasive evidence of the validity of the underlying theory of
rational choice (Riker, 1982b).
2. The initiation of war: One main intention of students of politics since
ancient times has been to identify the cause of war. To do so would be
equivalent to specifying a partial equilibrium, because a partial equilibrium
is simply the situation that occurs when a cause (i.e., a necessary and suffi-
cient condition) is present. Until recently, no one had been able to state and
support either a necessary or a sufficient condition and the study of the cause
of war had been largely anecdotal. Bueno de Mesquita (1981) , however, for-
mulates a precise statement of a necessary condition for warfare as well as
convincing evidence that this condition applies across cultures and through a
fairly wide space of time. The condition is that rulers who initiate war have a
positive expected value for the anticipated outcome. This condition goes far
beyond the conventional notion that rulers of more powerful nations start
wars against less powerful ones. Since it is expressed in terms of expected utili-
ty rather than power, it allows for the inclusion of ideology and thus admits
 William H. Riker I 53

application to situations in which, for example, weaker nations attack

stronger ones or a ruler attacks allies. The expected utility analysis of inter-
national politics is just beginning (Gilpin, 1981) and one can hope that Bueno
de Mesquita's necessary condition may ultimately be elaborated into a neces-
sary and sufficient one.
3. The size of parliamentary coalitfons for the formation of cabinets:
Some years ago I enunciated the size principle (Riker, 1963) that, in situations
with a constant sum character, participants form minimally winning coali-
tions. Although my derivation of this principle from the characteristic func-
tion of an n-person game has been (mistakenly) disputed (Butterworth, 1971 ;
Hardin, 1976), it has now also been derived from theories about the solution
of n-person games (Shepsle, 1974; Schofield, 1978a). It can, therefore, be
regarded as well established theoretically. Its empirical validity is somewhat
less certain, although it has been subjected to a large amount of testing in
analyses of coalition-formation in the real world and in the laboratory. Some
of this testing has been irrelevant because the situations studied have been
non-constant sum, but cabinet formation is one kind of situation that, at least
in the long run, should be constant sum. (That is, what is won or lost by
cabinet coalitions is the chance to run the country and this should have about
the same value over a period of years.) The evidence has, however, turned out
to be ambiguous. Many coalitions have been larger (or smaller) than minimal
and so a number of writers have sought to modify the principle with con-
siderations of ideological similarity (DeSwaan, 1973; Axelrod, 1970). Inter-
estingly enough, these modifications have turned out to be unnecessary when
cabinets have been analyzed on a long-run basis. While oversized and under-
sized coalitions are often formed, they have not lasted very long; minimal win-
ning coalitions last (Dodd, 1976). Recently Schofield (1982a) has gone over a
set of data siI,nilar to , but larger than, Dodd's in order to study the stability of
cabinets. He 'has found that the size principle predicts the duration of govern-
ments quite well. (This suggests that cabinet formation and support is indeed
a constant sum game, though there may be ideological variations of a non-
constant sum nature in the initial formation of cabinets.) In any event, the
work of Dodd and Schofield establishes a theory of equilibrium about
cabinet formation, provides good evidence of its credibility, and tends to
verify a broader political law, namely the size principle.

III
These three examples of partial equilibrium are not intended to be
representative of all branches of political science, although it does happen
that they come from different traditions of empirical work. Indeed, in order
to collect a reasonable sample, it is necessary to look into several such tradi-
tions; to date no one tradition could, in itself, yield as many as three exam-
ples. This is, of course, quite different from economic science, where many
sub-fields would yield several examples of theoretically justified and em-
pirically validated partial equilibria.
One appropriate question from this survey is, then: Why are the exam-
54 j POLITICAL THEORY AND THE ART OF HERESTHETICS

ples so few? I will undertake to offer some answers, in part as a guide to the
improvement of the productivity of political theorizing.
One answer is, as I have already indicated, the fact that political science
lacks a widely applicable, fundamental , and well-verified law of political
behavior comparable to the law of demand. As a result, we political scientists
have had no compass to guide us through the thickets of facts and we have,
therefore, been prone to build theories around regularities we have hap-
pened, more or less randomly, to observe.
But this situation need not continue. We still have no widely applicable
empirical law, but we do have in rational choice theory a widely applicable
method of analysis. The proposition that, in political affairs, people choose
the action (from among several) that yields the largest expected utility is
assumed in the justification of Duverger's law; it is the starting point for the
statement of Bueno de Mesquita's necessary condition for war; and, in Scho-
field's formulation at least, it is the basis for an equilibrium in the size of
cabinet coalitions. It is no accident, I believe, that these three more or less
validated theories of equilibrium share the rational choice assumption. As
Ordeshook and I have argued elsewhere (Riker & Ordeshook, 1973, ch. 2),
this assumption assures that actors in the political model behave regularly. It
plays, indeed, the same role in contemporary social science that the principle
of no action at a distance played in eighteenth-century physics: It banishes
witchery and other inexplicable variations from the model. If there are
regularities to be observed in the world - as indeed there are - then a theory
in which regularities are anticipated can conceivably describe them, while
approaches postulating actions based on emotional vagaries and variable
intentions have no place for regularities and cannot possibly be used to
explain them.
Although there are good reasons to believe that rational choice analysis is
the traditional paradigm for politics, just as it has been for economics,
political scientists have on the whole never quite adopted the rational choice
model as a generally accepted paradigm. Now that our understanding of the
paradigm has become conscious and explicit, we have begun to exploit the
technique of rational choice analysis and it seems reasonable to expect that,
as we learn how to use it better, we will be more successful, perhaps even as
successful as economists.
A tool is not enough, however. We also need some understanding of how
we should use it. It is the absence of this understanding that serves as a second
answer to the question of why examples of partial equilibria are so few in
political science. As many philosophers of science have pointed out, the selec-
tion of facts to be gathered in a science is largely dictated by the scientists'
conception of what they are doing. In the initial stages of any science the pro-
fessionals accept lay and common sensical categories for inquiry. Thus, for
example, in the beginnings of biology the main lines of endeavor were
anatomy, medicine, and taxonomy. But as the questions for research came to
be posed by biologists themselves, the main fields became genetics, evolution
and biochemistry, subjects of which the laity knows little and to which com-
mon sense has nothing to contribute. Similarly, in political science until now,
subjects of research have typically been determined by the interests of rulers
 William H. Riker I 55

or reformers who want to use scholars' discoveries to keep or get control of

society. This resulted in a science that consisted of the categorization of
institutions and the compilation of rules of thumb about techniques of
political management.
One great contribution of political theory, with its emphasis on the
amalgamation and manipulation of tastes, has been to redirect the questions
of political science away from the interests of rulers and reformers into the
description of what both kinds of manipulators do. The goal of a general
equilibrium theory is the identification of the consequences of distributions of
tastes. This leads to questions about the actions dictated by particular
distributions of taste, to questions about the origin, manipulation, and
modification of tastes, and to questions about the processes by which tastes
are transformed into social decisions. Unfortunately, when political theorists
pointed political science in these directions there was little data in the conven-
tional categories on which to base political analysis. Hence, along with the
tool of rational choice theory, political science needs the accumulation of
data appropriate for the application of that theory.
We need, I believe, a tradition of what I call herestheti·cs, which should
have been - but was not- invented by the Greeks as the counterpart in the
study of politics to rhetoric in the study of literature. It is perhaps odd that in
a survey of political theory one should advocate the accumulation of another
categorization of events. But theorizing must begin somewhere and it must
use appropriate data. It is to the initiation of that task that I dedicate the rest
of this paper.

IV
Heresthetics, in my coinage of the word, has to do with the manipulation
of the structure of tastes and alternatives within which decisions are made,
both the objective structure and the structure as it appears to partici-
pants. It is a study of the strategy of decision. Its source is the Greek
cxipeiaeerot , which has to do with choosing and electing.• I describe it with a
Greek-like word because it is a branch of knowledge that should have been,
but was not, invented by Athenian philosophers and sophists. The modern
impetus to heresthetics comes from game theory and social choice theory,
both of which are methods of describing social decisions abstractly. The con-
crete decisions themselves take many forms , however, and heresthetics is the
categorization of the circumstances of these decisions in a way that makes
them more amenable to abstract theoretical description. Concrete grammars
are necessary for the development of an abstract linguistics and heresthetics
serves in an analogous role in the development of the theory of social
decision. 5
Two examples of what would have properly been called heresthetics, if
the word had existed when they were written, are:
(1) Pliny the Younger's letter 14 of book 7, to Titus Aristo, as used by
Robin Farquharson for his running example in the Theory of Voting (1958
and 1969). It is not clear whether Pliny intended this letter as a commentary
on social decision or merely as a cry to Titus for help in defending the con-
56 I POLITICAL THEORY AND THE ART OF HERESTHETICS
stitutionality of Pliny's ruling from the chair of the Senate. About Farquhar-
son's intent, however, there is no doubt for he offered the letter as an example
of alternative methods of voting. Quite unbeknownst to him-for he had a
bad translation-the letter also serves as a commentary on manipulation of
the agenda (by Pliny) and on sophisticated voting (by Pliny's unnamed
opponent) (Pliny, Ed. by Radice, 1969). In the incident described in the letter
to Titus, Pliny ruled for plurality choice among three alternatives on a single
division, while the customary procedure would have been a sequence of pair-
wise decisions. His motive was to ensure the victory of his preferred alternative
over the Condorcet winner, which he placed second. In the voting, however,
one group of his opposition voted sophisticatedly so that the Condorcet win-
ner actually won, doubtless to Pliny's chagrin.
(2) Schelling, The Strategy of Conflict (1960). Several chapters of this
volume concern manipulative techniques. When originally published (in 1956
and 1957) as scientific papers, Schelling interpreted them as contributions to
a theory of bargaining, but later (in 1960) he believed they were contributions
to the theory of games. He was entirely justified in his uncertainty about the
classification of these chapters which concern threats, promises, and commit-
ments generally. They are not a theory of bargaining because they have
nothing to say about the compromise of tastes (and thus are quite unlike the
work of, say, Harsanyi, 1980). Nor are they about game theory because they
are not concerned with the way the interaction of players' choices determine
an outcome. Rather, they are about the strategems that participants use in
selecting their own actions and in limiting the choices of others. They are
about the strategies to create situations in which game theory and bargaining
theory apply. As such, they are squarely in the field of heresthetics.
Both Schelling's and Farquharson's work are regarded as path-breaking,
and properly so, because they deal with a kind of data no one had thought to
collect before and furthermore with data that is highly relevant to theories of
social decision. To think about the subject matter of these theories-which is
what both writers wanted to do-it helps to be able to refer to concrete detail
of the sort they collected. In that sense, the content of heresthetics can be said
to come out of the demands of science, not out of lay language and observa-
tion; it ought, therefore, to be regarded as a significant and new kind of data
in the study of society.
It is hard to define heresthetics other than by examples because the pur-
pose of the concept is to provide an open-ended set of categories for events
that have not heretofore been systematically categorized. 6 One more example
may be helpful, so I offer one that was brought to my attention by Richard A.
Smith who used it as an introductory example of agenda-setting in his superb
dissertation (1980) on agenda-setting by lobbyists. This example, as related in
Redman (1973), concerns Senator Warren Magnuson, who was trying to stop
the transport of nerve gas across his state of Washington, enroute from
Okinawa to Colorado. Because relatively few other Senators shared his pro-
vincial concern for impressing the voters of the state of Washington, he
expected to lose. But he won, mainly, so Redman reports, because of clever
heresthetical tactics. Redman tells the story well (p. 207):
 Wi"lliam H. Riker I 57

... during the Senate debate on his [Magnuson's] amendment to block ship·
ment, .. . I prepared another memorandum cataloguing the arguments he had
marshaled against the shipment and took it to him at his desk on the Senate
Floor. He surveyed the memo cursorily, then handed it back with an annoyed
"No, no, no!" Bewildered, I retreated to the staff couch and waited to hear what
argument he intended to use instead of the familiar ones of possible sabotage,
dangerous sections of track along the proposed route , and populations that
would have to be evacuated as a precaution against leakage. When the time
came, he took a wholly novel and ingenious approach. The issue, he told his col-
leagues, was not one of the people versus the Pentagon, as the news media
seemed to assume. Instead, it was another case of the President versus the
Senate. The Senator from West Virginia (Robert C. Byrd) had recently offered a
resolution, which the Senate had passed , stating that the Senate expected the
President to keep it informed throughout the treaty negotiations with the
Japanese government on the subject of Okinawa. The President's sudden deci-
sion to move the nerve gas off Okinawa must reflect some aspect of those treaty
negotiations, Magnuson insisted-and the Senate had not yet been informed of,
much less consented to, any such agreement . To allow the nerve-gas shipment
under such circumstances, he asserted, would be to abandon the Byrd Resolu-
tion and to abdicate the Senate's rightful role in treaty-making generally. The
President, Magnuson said, might get the idea that he could ignore the Senate
and its Constitutional prerogatives whenever he wished. Jolted by this reasoning,
the Senator from West Virginia and his Southern colleagues-friends of the
Pentagon almost to a man, but vigilant guardians of the Senate's Constitutional
responsibilities-voted down the line with Magnuson. The amendment, which
had been doomed a few minutes earlier, passed overwhelmingly.

Even though the collection of heresthetic detail is recent, the human

behavior involved is easy to observe in vastly different times and places. Heres-
thetics is in fact universal. One of the settings in which it is easy to observe the
manipulation of tastes is the parliamentary assembly, where there are many
groups, where individual loyalties are shifting, and where leaders win by
assembling ad hoc winning coalitions. Not surprisingly, the Roman Senate is
the locale for Farquharson's example (i.e., Pliny's letter) and the American
Senate is the locale for Smith's example (i.e., Redman's story about Mag-
nuson). One would expect, therefore, that examples for heresthetical
categories could be found in the assemblies of classic Greece, the earliest his-
torical assemblies. And so they can. W. Robert Connor (1971) describes the
Athenian assembly as "kaleidoscopic" or "polycentric," composed of several
relatively small groups of friends, almost clubs, which come together in sup-
port of particular motions and then regrouped on others. This is precisely the
kind of setting in which the open manipulation of the agenda and of the
salience of dimensions of judgment is likely to be most visible. And indeed,
tiny as the historical record is (mostly Thucydides' History and in the next
century Demosthenes' Orations), it contains quite a large number of descrip-
tions of heresthetical devices. Connor discusses, for example, Thucydides' ac-
count of the development of the coalition against Alcibiades. That coalition,
having failed to halt the disastrous Sicilian expedition, turned to an attack on
Alcibiades, its main sponsor. The leaders of the coalition raised the issue of
impiety against him-the charge was profaning the mysteries and defacing
the statues of Hermes-and, although they failed initially, they ultimately
 58 I POLITICAL THEORY AND THE ART OF HERESTHETICS

had Alcibiades recalled from his command, though the expedition itself was
not stopped, and he deserted to Sparta (Thucydides, VI , 28-29, 61). Connor
sees this as an example of the structuring of a coalition, but it is also an exam-
ple of the introduction of a new issue (Alcibiades' impiety) to transform the
coalition against the expedition from a minority to a majority. An even better
example of the heresthetical tactic of introducing a new dimension is found a
bit earlier (Thucydides, VI , 8-24), in Nicias' argument against the expedi-
tion. Initially he opposed it on the ground that it was too large an under-
taking and that it was intended only to gratify Alcibiades' tastes for the
magnificent. When Alcibiades carried the day by promising to make the
Athenians the rulers of all Hellas, Nicias responded in fine heresthetical
fashion by introducing a new dimension of judgment. Thucydides wrote (VI,
19-24):

Nicias, seeing that his old argument would no longer deter them, but that he
might possibly change their minds if he insisted on the magnitude of the force
which would be required, came forward and spoke as follows : ". . . I say,
therefore, that we must take with us as a large heavy armed force both of
Athenians and of allies .... Our naval superiority must be overwhelming . .. . "
These were the words of Nicias. He meant either to deter the Athenians by
bringing home to them the vastness of the undertaking or to provide as far as he
could for the safety of the expedition if he were compelled to proceed. The result
disappointed him. Far from losing their enthusiasm, . . . they [i.e., the
Athenians in assembly] were more determined than ever.

This reads just li~e an event in modem democratic politics: A politician who
loses on one platform or set of arguments tries another and, when he loses on
the second, his friends try still a third (i.e., the attack on Alcibiades' alleged
impiety) on which they finally win, at least in part. I find in Thucydides quite
a few other examples of heresthetics: false presentation of an issue (V, 45-47);
coalition building around a new issue (IV, 84-88); agenda control (II , 21-22);
and the addition of a dimension of judgment (I, 80ff.). Altogether, therefore,
it is clear that, given similar settings in different cultures, heresthetical
behavior is universal.
Furthermore, since it clearly concerns the dynamics of the approach to
important human decisions, one would expect that, even in classic Greek
times, men would have begun to collect and classify examples of heresthetic,
just as they did for rhetoric, which concerns another feature of the approach
to decision. And there was indeed a tiny start in that direction. Demosthenes
was a master of heresthetic as well as rhetoric and his orations have enough
detail for the modem reader to understand the full scope of his strategy. His
modem biographer, Werner Jaeger, commenting on Demosthenes' speech,
On the Symmorz"es, in which he advocated a larger navy based on a wider
assessment of taxes, ostensibly in support of a proposed Persian War, but
actually-and successfully-to dampen enthusiasm for that war by empha-
sizing its cost, observed that this was the same heresthetical technique Nicias
had used against Alcibiades. In the Art of Rhetoric the pseudo Dionysius of
Halicamassus classified Demosthenes' maneuver with a similar maneuver by
 William H. Rz"ker I 59

the Spartan king Archidamas (as reported by Thucydides, I, 80ff.) to intro-

duce a fiscal dimension to defeat an outcome expected to win if determined
only by a military dimension Qaeger, 1938, pp. 226-227). The false Dionysius,
putting the story of Archidamus and the text of Demosthenes together indeed
started up a heresthetical tradition. Unfortunately, it never developed
further.
I wonder, why not? One possible answer is that, while heresthetic and
rhetoric were initially entwined in classical thought, as the democratic
assembly-the one place where political strategy could be easily observed-
disappeared from classical institutions, only rhetoric in the sense of persua-
sion was left to be observed and classified. In the classical tradition it was
customary to categorize the settings for rhetoric into the deliberative (i.e., in
assemblies), the forensic (i.e., in courts), and the epideictic (i.e., in cere-
monies). Two historians of Greek rhetoric, Wilcox (1942) and Kennedy
(1963), assert that the early rhetoricians were concerned mainly with
deliberative persuasion. They argue in support, first, that rich young men of
prominent families would hardly have paid the sophists large fees simply to be
taught how to be lawyers, and second, that Plato's intense hostility (in Gorgias
and Protagoros) to rhetoricians must have been based on a fear that they
would teach people how to distort public policy in general (i.e., in the
assembly) rather than in simply private disputes (i.e., in the courts). These
arguments seem persuasive to me, especially since the rhetorical tradition
from the beginning emphasized persuasion by eloquence rather than persua-
sion by argument. Jacqueline de Romilly (1974) has argued that rhetoric
began in the association of poetry and magic (i.e., spells and charms) with
public speech and argument and was developed by the sophists into an art to
persuade about the correctness of any preferred alternative. "Plato resented,"
she wrote, "not the magic [in Gorgias' thought] but the offensive pretense of
turning it into science" (de Romilly, 1974, p. 38). Aristotle tried to save
rhetoric for philosophy by cleansing it of what Plato despised, by, in fact,
turning it into a policy science in which persuasion consists of showing that
sentences have a high probability of truth. In particular, he wished to banish
eloquence from style so that it "be transparent, not magical" (de Romilly,
1974, p. 73). If Aristotle had succeeded then rhetoric would have been con-
cerned primarily with political decision and would doubtless have continued
to subsume political strategy or heresthetics as well. But Aristotle failed and
rhetoric became almost entirely a matter of eloquence. It was Cicero who set-
tled the tradition for most of the next two thousand years. In the Orator (line
44) he says the rhetor "must consider three things, what to say, in what order,
and in what manner or style to say it," but then he goes on to remark that he
will treat the first two only briefly because, while important, they are a part of
"ordinary intelligence rather than eloquence." Plainly, therefore, eloquence
was in Cicero's mind the distinguishing feature of rhetoric and he banished all
else, including what might have become heresthetic.
If this historical speculation is correct that rhetoric became simply per-
suasion - and the least rational kind of persuasion at that- then rhetoric
could hardly sustain heresthetic, especially in the absence until modem times
of deliberative assemblies where political strategy is easy to observe. While it is
60 I POLITICAL THEORY AND THE ART OF HERESTHETICS
thus explicable that the subject of heresthetics has not been studied, It IS
nevertheless clear that the reason scholars have ignored it has nothing to do
with its significance, but rather with the accident that it was excluded from
the one tradition in which it might have survived.

v
In the previous section I sought to define heresthetics, to show its univer-
sality, and to explain why, despite its universality, it is unstudied. I turn now
to a further explication of heresthetics by distinguishing it from rhetoric with
which, given the relatedness of subject matter, it might be confused.
Rhetoric and heresthetic are both techniques of winning. But they are
different kinds of techniques. Rhetoric is persuasion. It involves confronting a
judge or jury in a courtroom, or voters on a committee or in a polity, or
buyers in a marketplace, or friends or philosophers in dispute with sentences
that may convince them that you are correct or believable. The sentences may
convince because they are beautiful or sonorously uttered, or because of their
irrefutable arguments, or because of their presentation of the situation in a
way the auditors are predisposed to accept. But the essential feature, qua
rhetoric, is that they convince. With heresthetic, on the other hand, convic-
tion is at best secondary and often not even involved at all. The point of an
heresthetical act is to structure the situation so that the actor wins, regardless
of whether or not the other participants are persuaded.
The contrast between rhetoric and heresthetic is nicely seen in the
dilemma, a form shared by the two fields. Rhetorically, the dilemma-maker
succeeds because he convinces the auditors that, if his opponent cannot
resolve the dilemma, then the opponent's position is intellectually weak.
Hence the dilemma is a device for persuasion. Heresthetically, the dilemma-
maker succeeds because he forces his opponent into a choice of alternatives
such that, whichever alternative is chosen, the opponent will alienate some of
his supporters.
Consider the best of the textbook examples of the rhetorical dilemma: It
is the pair of dilemmas posed by Tisias and Korax. Korax, whose name means
"crow," was the founder of the school of rhetoric at Syracuse, and was
Gorgias' teacher. Tisias, who was also Korax's student, had undertaken to pay
tuition when he won his first case; but he had neither practiced nor paid, so
Korax sued. Tisias responded with this dilemma: "If I win the case, then I
need not pay because I am freed by the judgment of the court. If I lose the
case, then I need not pay because the terms of the contract will not have been
satisfied. Since I must either win or lose, I need not pay." To this Korax
replied: "If I win the case, then I must be paid because of the judgment of the
court. If I lose the case, then I must be paid because the terms of the contract
will have been satisfied. Since I must either win or lose, I must be paid." The
judge dismissed the case, saying Tisias was a bad egg from a bad crow.
Although Korax lost, the advertisement was so good that people still repeat it
2500 years later. Not only did Korax show that he could teach a student to
present what appeared to be an absolutely convincing defense, he also showed
that he then could himself tear it all down with an exact riposte.
 William H. Riker I 61

Consider now an heresthetical dilemma , Lincoln's famous question to

Douglas in the Freeport debate. During the senatorial campaign of 1858
Lincoln met Douglas, the incumbent, in debates in several Illinois cities,
while both were soliciting support for candidates for the state legislature who
were in tum pledged to support them for the Senate. Lincoln's question was:
"Can the people of a United States Territory, in any lawful way, against the
wish of any citizen of the United States, exclude slavery from its limits prior to
the formation of a state Constitution?" (Lincoln, 1958, p . 108). To the con-
temporary reader this seems perhaps innocuous enough; but it is at the very
center of the partisan dispute of the era and expresses exactly the raison d'etre
of the Republican party.
The parties of the tradition of Federalist-National Republican-Whig-
Republican did not do well throughout the first two-thirds of the nineteenth
century. Indeed between 1797 and 1867 they elected only one President by a
majority, while the Jeffersonian-Jacksonian Democratic parties probably
elected ten by a majority. The main problem for the parties of the Federalist-
Whig-Republican sequence was that they usually espoused a platform of com-
mercial expansionism which appealed only to a minority; Democrats, on the
other hand, espoused agrarian expansionism which often appealed to a
majority. Naturally the Whig-Republican leaders searched constantly for a
better platform and they finally found one in the combination of commercial
expansionism with the limitation of slavery. This platform had the advantage
· of splitting the Democrats, who were well distributed in North and South,
more than it did the Whigs, who were weak in the South. Lincoln's question
exactly expressed this Republican strategy.
The Democratic defense was to cover up the slavery issue, to assert it was a
local concern that should be banished from national politics. In the 1850s
Douglas had been the main agent of this defense by sponsoring the Kansas-
Nebraska Act. This Act satisfied the South because it rendered slavery local
and it mollified those Northern Democrats who were tempted to oppose
slavery because it allowed the territories to eliminate slavery. It failed, how-
ever, to localize the issue, especially after the ruling in the Dred Scott case
(1857) placed slavery beyond the control of local legislators. But Douglas was
still trying in 1858 to patch up the national Democracy by localizing the issue
and Lincoln's question was a trap to encourage him to do so, thus:
If Douglas answered "yes" (that territories could exclude), then he would
alienate Southern Democrats for whom the Dred Scott decision was the new
status quo. Since he hoped and expected to be the Democratic candidate for
President in 1860, to answer "yes" would be to reduce his Southern support
and thus to jeopardize his chances in that election. At the same time a "yes"
answer would placate Illinois Democratic voters with free soil principles and
thus enhance his chances for reelection to the Senate.
If he answered "no" there would be converse results. He would alienate
Illinois free soil Democrats and possibly lose his Senate seat, but he would win
the loyalty of the South for the Presidential election of 1860.
Since he had to answer either "yes" or "no," he jeopardized either his
election in 1858 to the Senate or in 1860 to the Presidency.
Douglas answered a kind of "yes." Without disputing the Dred Scott
 62 I POLITICAL THEORY AND THE ART OF HERESTHETICS

decision, he said "slavery cannot exist a day or an hour anywhere, unless it is

supported by local police regulations," which, he argued, an anti-slavery ter-
ritorial legislature need not adopt (Lincoln, 1858, p . 113). This was the
answer Lincoln's advisors feared-they had opposed his asking the question-
for it probably helped Douglas to reelection. But, as things turned out,
Lincoln won in 1860, helped immeasurably by Douglas' shortsighted "yes."
An intriguing question: Did Lincoln in 1858 see ahead to his own candidacy
in 1860 or was he sacrificing himself for the greater Republican good?
Clearly the dilemma Lincoln posed was not intended to persuade the
audience that Douglas' position was intellectually weak. Indeed, it did not
depend for its effect on convincing anybody of anything. It was simply a
stratagem to force Douglas to reveal to one of his incompatible groups of sup-
porters that he was faithless to its cause. As such it was strictly a heresthetical
device that set up a situation for subsequent decisions in such a way that at
least one of the decisions would be to Lincoln's taste.
The essentially heresthetical character of Lincoln's act of posing the
dilemma is underlined by the fact that, viewed another way, it is very similar
to the heresthetical devices of Nicias or Demosthenes, which I described
earlier. They raised a new fiscal dimension of judgment in order that voters
would be less enthusiastic about the preparations for war so heartily approved
on a military dimension of judgment. Lincoln's stratagem also involved rais-
ing a new dimension, slavery, to drive some voters to oppose the agrarian
expansion they would otherwise approve. Douglas' response was to try to sup-
press the new dimension. Thus, although the Lincolnian device has the form
of a dilemma, it has the substance of the generation of a new issue.

VI
The purpose of this essay is to recommend the study of heresthetic, not
actually to study it. Nevertheless, in the hope of making my recommendation
more persuasive, I conclude with a brief statement of what I have so far
myself learned from the study of heresthetic.
In the study of politics and public policy we devote most of our attention
to the analysis and interpretation of the platforms and policies of the winners
of political disputes, elections, wars, and so forth. And this is quite proper
because the preferences of the winners are the values that are authoritatively
allocated. That is, the tastes of the winners are the actual content of social
decisions and thus the content of the immediately subsequent present time.
Conversely, we ignore the policies and platforms of the losers because these
are the junk heap of history, the might-have-beens that never were. But we
should not, I think, entirely overlook the losers and their goals for the losers
provide the values of the future. The dynamics of politics is in the hands of
the losers. It is they who decide when and how and whether to fight on. Win-
ners have won and do not immediately need to change things. But losers have
nothing and can gain nothing unless they continue to try to bring about new
political situations. This provides the motivation for change. And the con-
firmation of this fact comes from the study of heresthetics. Losers are the ones
 Wi"llz"am H. Rz'ker I 63

who search out new strategies and stratagems and it is their use of heresthetics
that provides the dynamic of politics. 7
In my recent work, Li'beralz'sm agaz'nst Populz'sm (1982) I have discussed,
mainly in chapters 6, 7, and 9, the way in which participants in an electoral
system can manipulate electoral procedures to their advantage. Strategic
manipulation is, of course, one important part of heresthetics. So the
categories of manipulation used in that work are also categories of heres-
thetics. The categories I used were:
(1) strategic voting, which involves voting for a less preferred alternative
(in lieu of a more preferred one) at some initial decision point in order that
the voter can achieve (or improve the chance of achieving) an outcome at the
final decision that the voter favors over the outcome expected from initially
voting for the more preferred alternative;
(2) manipulation of the agenda, which is structuring the set of alterna-
tives or the procedure of voting in such a way as to p~oduce an outcome more
favored by the manipulator than might otherwise have been achieved.
Within each of these two main categories are a number of kinds of exam-
ples, each of which has the property that the user of the particular heres-
thetical device must be a person who has lost a decision or reasonably antici-
pates losing one. Then the stratagem itself becomes a means either of winning
on a new issue or of transforming the anticipated loss into a victory. Thus,
strategic voters in each of the following examples are persons who anticipate
losing if they do not adopt the stratagem:
(1) Avoidance of "wasted" votes: In plurality voting over three or more
alternatives, a voter who most prefers some alternative other than the two
most popular may nevertheless vote for his or her favorite among those two so
that in the vote count the favorite (within the top pair) will have a better
chance of winning than if the voter had "thrown away" his or her vote on the
most preferred alternative. Clearly this heresthetic device is available only to
supporters of the third most popular (or, in the case of more than three, the
least popular) alternative;
(2) Creation of a voting cycle: When a motion as amended is so distaste-
ful to some of the supporters of the original motion that they reject the
amended motion, it is of course open to the opppnents of the original motion
to attach the amendment to the motion in the hope thereby of creating a cycle
in which
(a) the status quo beats the amended motion
(b) the amended motion beats the original motion
(c) and the original motion beats the status quo.
In such a cycle, some opponents of the original motion are, as line ( c) in-
dicates, losers; but they can perhaps become winners if they can attach the
amendment, as indicated by line (a);
(3) Vote-trading: Potential vote-traders are those who, on divisions on
two motions, can expect to win on the motion less valuable to them and to lose
on the more valuable one. If they can find enough persons oppositely situated
so that they can, by trading, together reverse the outcomes, then the traders
will win on their more valuable motions and lose on their less valuable ones.
Manifestly, traders must be losers in order to gain by trade.
64 I POLITICAL THEORY AND THE ART OF HERESTHETICS
The same features are exhibited by examples of manipulation of the
agenda:
(1) Arrangement of the sequence of decisions: Those directly in control
of the agenda can, if they foresee an unfavorable outcome from the normal or
customary agenda , rearrange the agenda so that alternatives more popular
than their favored one are eliminated prior to the decision on their favorite.
Hence they can improve its chances. They need to engage in this maneuver,
however, only when they expect the "natural" vote to go against them;
(2) Introduction of new alternatives: Those who expect their preferred
alternative to lose initially may introduce new alternatives, even as mere par-
ticipants not leaders. Assuming the new alternative is at least better for them
than the most popular alternative in the initial set and that some of the sup-
porters of the popular alternative will prefer the new alternative, then the
introducers improve their chances of defeating the initially most popular
alternative. Obviously only those who initially expect to lose are motivated to
use this strategy.
It seems clear, therefore, that, in the case of all the heresthetical devices I
have examined, the users are persons who are not now winning but hope to
become winners. Insofar as these devices involve a fairly wide range of
political life, it also seems clear that losers are the instigators of political
change and it is they who are thus motivated to exploit heresthetic and who
become the agents of political dynamics.
A generalization of the sort just uttered is exactly the kind of theoretical
payoff political scientists might expect to obtain from heresthetical categori-
zation. It is in the hope of further such generalizations that I recommend the
study of heresthetics.

NOTES
1. See Riker, 1977, 1980, 1982(a) for more detailed versions of this history.
2. This feature of Black's work is closely related to the Hotelling-Downs (1957) model
of spatial equilibrium . For a survey of the development of work on that model, see
Riker and Ordeshook (1973) and a forthcoming volume by Melvin Hinich and
James Enelow.
3. One of my critics believes the statement in the text is too strong because all sorts of
institutions impose constraints on outcomes and thereby preclude the possibility
that "anything can happen ." Of course, he is correct with respect to applications to
the real world , as I indicate in section II of this paper. But it is important to note
that, in the abstract without considering institutions, there is almost no likelihood
of an equilibrium.
4. "Heretics" might be a better Anglicization of the Greek source, but that word is
a form of the word "heresy," which has been co-opted by religion. Hence, I prefer
the term I have chosen.
5. One of my critics defines my call for a new kind of data as a plea for us "to devote
more systematic attention to the dynamics of collective decision making. A stan-
dard criticism leveled against existing political models is that they are static in
nature . Even when the more ambitious among us attempt to model dynamic
phenomena, the essentials of the model remain constant . Of course, we must learn
to walk before we can learn to run and all that, but at some point a progressive
political theory will begin to remove the basis of the criticism."
 William H. Riker I 65

6. One of my critics believes that my implicit definition of heresthetics is too broad. He

points out that I seem to be considering as heresthetical both the manipulation of
preferences and the manipulation of alternatives. He suggests that I ought to
restrict the reference to the latter manipulation (i.e., situations where preferences
are constant and only alternatives change) because we already have names for the
former manipulation . I hesitate to restrict reference in this way, however, because
one of the things that happens in the manipulation of alternatives is that as a conse-
quence the salience (and hence the content) of preferences are also changed .
7. One of my critics suggests that, by emphasizing the creative (perhaps artistic) ele-
ment of heresthetics, I am, unconsciously or slyly, removing it from the range of
science. I certainly do not intend to do so. While it is certainly true that human life
involves artistic creations, we would deny the possibility of all social science by
asserting that creativity cannot be scientifically described. Just as Von Neumann
and Morgenstern discovered how to discuss analytically creative acts of choice
among strategies, so I believe we can discover how to discuss analytically creative
acts of structuring strategies themselves.

REFERENCES
Aldrich, John, & Rohde, David. The limitations of equilibrium analysis in political
science. In Peter C. Ordeshook and Kenneth A. Shepsle (Eds.). Political equilibri-
um. Boston: Kluwer-Nijhoff, 1982, pp. 65-95.
Aristotle. Rhetoric. As interpreted in Larry Arnhart, Aristotle on political reasoning.
DeKalb , Ill .: Northern Illinois University Press, 1981.
Arrow, Kenneth. Social choice and individual values (2nd ed.). New Haven, Conn.:
Yale University Press, 1963. (Originally published, 1951)
Axelrod, Robert. Conflict of interest. Chicago: Markham , 1970.
Benzel, Richard F. , & Sanders, Elizabeth. The effect of electoral rules on voting
behavior: The electoral college and shift voting. Public Choice, 1979, 34, 609-38.
Black, Duncan. On the rationale of group decision making. Journal of Political
Economy, 1948, 56, 23-24. I
Black, Duncan. The theory of committees and elections. Cambridge: Cambridge
University Press, 1958.
Black, Duncan & Newing, R . A. Committee decisions with complementary valua-
tion. Edinburgh: William Hodge , 1951.
Black, Jerome. The multi-candidate calculus of voting: Application to Canadian
federal elections. American journal of Political Science, 1978, 22, 609-38.
Bueno de Mesquita, Bruce. The war trap. New Haven: Yale University Press, 1981.
Butterworth, Robert. A research note on the size of winning coalitions. American
Political Science Review, 1971 , 65, 741-45 .
Cain, Bruce E. Strategic voting in Britain. American journal of Political Science,
1978, 22, 639-55 .
Cicero. Orator. H. M. Turnbull (Ed.). London: Loeb Library, 1934.
Coase, R. H . Duncan Black: A biographical sketch. In Gordon Tullock (Ed.).
Toward a science of politics: Papers in honor of Duncan Black. Blacksburg, Va.:
Center for the Study of Public Choice, Virginia Polytechnic Institute and State
University, 1981 , pp . 1-10.
Cohen, Linda. Cyclic sets in multi-dimensional voting models. Journal of Economic
Theory, Feb. 1979, 22, 1-12 .
Condorcet, Marquis de. Essai sur l'application de l'analyse a la probabilite des
a
decisions rendues la pluralite des voix. Paris, 1785.
Connor, W . Robert. The new politicians offifth-century Athens. Princeton: Princeton
University Press, 1971.
66 I POLITICAL THEORY AND THE ART OF HERESTHETICS
DeMeyer, Frank, & Plott, Charles. The probability of a cyclical majority. Econo-
metrica, 1970, 38, 345-54.
de Romilly, Jacqueline. Magic and rhetoric in ancient Greece. Cambridge: Harvard
University Press, 1975 .
DeSwaan , Abram. Coalition t heories and cabinet formation. San Francisco: Jossey
Bass, 1973.
Dionysius of Halicarnassus (pseudo.). TEXNH PHEOPIKH. (Heinrich Schott, Ed.
and trans. into Latin) . Leipzig: Engelhard, 1804.
Dodd, Lawrence C. Coalitions in parlt"amentary government. Princeton: Princeton
University Press, 1976.
Downs, Anthony. An economic theory of democracy. New York: Harper, 1957.
Easton, David. The political system: An inquiry into the state of polt"tical sct'ence.
New York: Knopf, 1953.
Farquharson, Robin. Theory of voting. New Haven: Yale University Press, 1969.
Fiorina , Morris P., & Shepsle, Kenneth A. Equilibrium , disequilibrium, and the
general possibility of a science of politics. In Peter C. Ordeshook and Kenneth A.
Shepsle (Eds.), Polt'tical equi"lt"brium. Boston: Kluwer-Nijhoff, 1982, pp. 49-64.
Fishburn, Peter. The theory of soct'al choice. Princeton: Princeton University Press,
1973(a).
Fishburn, Peter. Voter concordance, simple majorities, and group decision methods.
Behavioral Sct'ence, 1973(b), 18, 364-73.
Garman, Mark, & Kamien, Morton. The paradox of voting: Probability calculations.
Behavioral Sct'ence, 1968, 13, 308-16.
Gehrlein , William, & Fishburn, Peter. The probability of the paradox of voting.
journal of Economic Theory, 1976, 13, 14-25.
Gibbard, Allan. Manipulation of voting schemes: A general result. Econometn·ca,
1973, 41 , 587-601.
Gilpin , Robert . War and change in world polt"tt"cs. New York: Cambridge University
Press, 1981.
Hardin, Russell . Hollow victory: The minimum winning coalition. American Polt"#cal
Sct'ence Review, 1976, 70, 1202-14.
Harsanyi , John C. Ra#onal behavior and bargai·m·n g equi"lt"brium in games and social
si'tuations. New York: Cambridge University Press, 1977.
Hempel, Carl G. The phiwsophy of natural sct'ence. Englewood Cliffs, N.J.: Prentice
Hall, 1966.
Jaeger, Werner. Demosthenes: The origin and growth of his polt"cy. Berkeley: Univer-
sity of California Press, 1938.
Kennedy, George. The art of persuasion in Greece. Princeton: Princeton University
Press, 1963.
Kramer, Gerald. On a class of equilibrium conditions for majority rule. Econometn·ca,
1973, 41, 285-297 .
Lemieux, Peter. The liberal party and Bn"tish political change: 1955-74 (Doctoral dis-
sertation, Massachusetts Institute of Technology, 1977).
Lincoln, Abraham. The Illinois politt"cal campaign of 1858: A facsimile of the printer's
copies of his debates with Stephen Arnold Douglas as edi"ted and prepared for press
by Abraham Li·ncoln. Washington, D.C. : Library of Congress, 1958.
McKelvey, Richard D. Intransitivities in multi-dimensional voting models and some
implications for agenda control.journal of Economic Theory, 1976, 12, 472-82.
Niemi, Richard. Majority decision making with partial unidimensionality. Amen·can
Polt'tical Sct'ence Revi·ew, 1969, 63, 489-97.
Niemi, Richard & Weisberg, Herbert. A mathematical solution for the probability
of the paradox of voting. Behavt·oral Sct'ence, 1968, 13, 317-23.
Pliny the Younger. Letters. Betty Radice (Ed.). Cambridge: Harvard University
Press, 1969.
 William H. Riker I 67

Plott, Charles. A notion of equilibrium and its possibility under majority rule. Amer£-
can Econom£c Rev£ew, 1967, 57, 787-806.
Plott, Charles. Path independence, rationality, and social choice. Econometr£ca,
1973, 41, 1075-91.
Popper, Karl. Conjectures and refutat£ons: The growth of sc£entif£c knowledge.
London: Routledge and Paul, 1963.
Redman, Eric. The dance of legislat£on. New York: Simon and Shuster, 1973.
Riker, William H. The theory of polz"tz"cal coalz"tz"ons. New Haven: Yale University
Press , 1963.
Riker, William H. , & Ordeshook, Peter C. An £ntroduct£on to posz"t£ve pol£t£cal
theory. Englewood Cliffs, N.J.: Prentice Hall , 1973.
Riker, William H. The future of the science of politics. The Amer£can Behav£oral
Sc£entist, 1977, 21, 11-38.
Riker, William H. Implications from the disequilibrium of majority rule for the study
of institutions. Amen·can Polz"t£cal Science Rev£ew, 1980, 74, 432-46.
Riker, William H . Liberalism against populism: A confrontat£on between the theory
of democracy and the theory of soc£al cho£ce. San Francisco: W. H. Freeman,
1982(a).
Riker, William H. The two-party system and Duverger's law: An essay on the history
of political science. Amen·can Polz"t£cal Sc£ence Review, 1982(b), 82, 753-66.
Schelling, Thomas C. The strategy of conjlz"ct. Cambridge: Harvard University Press,
1960.
Schelling, Thomas C. Arms and influence. New Haven: Yale University Press, 1966.
Schofield, Norman. Generalized bargaining set for cooperative games. Intemat£onal
journal of Game Theory, 1978(a), 7, 183-99.
Schofield, Norman. Instability of simple dynamic games. Revi·ew of Economic Stud£es,
1978(b), 45, 575-94.
Schofield, Norman. Instability and development in the political economy. In Peter C.
Ordeshook and Kenneth A. Shepsle (Eds.), Polz"tical equz'l£brium. Boston: Kluwer-
Nijhoff, 1982(a), pp. 96-106.
Schofield, Norman. Political fragmentation and the stability of coalition governments
in western Europe. Paper presented at the European Public Choice Society Meet-
ing, Poitiers, France, March 22-27, l 982(b ).
Shepsle, Kenneth. The size of winning coalitions. Amen·can Pol£tical Sc£ence Revi·ew,
1974, 68, 505-18.
Smith, Richard A. Lobbying influence in Congress: Processes and effects. (Doctoral
dissertation, University of Rochester, 1980.)
Thucydides. The Peloponnesian War. (Benjamin Jowett, trans.). In Francis Godol-
phin (Ed.), The Greek Historians. New York: Random House, 1942.
Weintraub, E. Roy. Microfoundations: The compatibilz"ty of m£croeconomics and
macroeconomics. Cambridge: Cambridge University Press, 1979.
Wilcox, Stanley. The scope of early rhetorical instruction in Greece. Harvard Studies
in Classical Philology, 1942, 53, 121-55.
 3
Toward Theories of Data:
The State of Political Methodology
Christopher H. Achen *
In one sense, political methodology in the early 1980s enjoys robust
health. Applications of powerful econometric techniques-simultaneous
equation estimation, time series methods, analysis of covariance structures-
appear in political science journals on a frequent, if not yet routine, basis.
Political Methodology, less than a decade old, prints a large fraction of the
best empirical research done in political science. Perhaps most tellingly, the
quantitative method has attained full legitimacy among serious scholars,
including those who do not use it. Just a decade ago, a graduate student inter-
viewing for an American politics position at a first-class liberal arts college
could be asked whether his interest in mathematical techniques was some
youthful fancy from which he might be expected to recover. Today, one
trusts, no department with an eye to its reputation would do the same.
Yet if these are the best of times, they are the worst of times as well.
Several decades after its beginning, political methodology has so far failed to
make serious theoretical progress on any of the major issues facing it.
Psychologists invented factor analysis and scaling methods to cope with men-
tal tests; economists created structural equation estimation to deal with their
models, especially the economy-wide macroeconomic theories; and even
sociologists have contributed latent class analysis to the corpus of social scien-
tific methodology. Political science has done nothing remotely comparable.
Political methodologists have largely occupied themselves with two ac-
tivities. First, they have continued to develop the major quantitative research

*I am indebted to Greg Markus, George Rabinowitz, and Jim Stimson, who served
as able discussants on the panel of the American Political Science Association 1982 An-
nual Meeting where an earlier version of this paper was presented. Doug Hibbs, Jerry
Kramer, Neal Beck, and especially an anonymous reviewer with an inordinate fond-
ness for aggregate data gave me trenchant comments and criticism. I would also like to
thank John Sullivan, chairman of the methodology panels at the 1982 meeting, for in-
viting me to organize my thoughts on this topic, and Henry Brady for many conversa-
tions over several years on these and other methodological issues.

69
70 I TOWARD THEORIES OF DATA : THE STATE OF POLITICAL METHODOLOGY
tool in the discipline, the opinion survey. Most issues of Polt"tz"cal Methodol-
ogy, for example, have at least one article exploring the reliability or validity
of alternate survey techniques. Political methodologists also contribute
regularly to the Amer£canjournal of Polt"tz"cal Sdence, which often contains
articles on survey research, and Publz"c Op£n£on Quarterly, which covers the
same topics almost exclusively. Unfortunately, nearly all this work has been
atheoretical. Question wordings, interview design, and techniques for reduc-
ing nonresponse are invented as needed, with no overarching framework, so
that every new topic must be tackled de novo. The result is a body of work
that is certainly "methodological" in the broad sense of the term, and without
it, empirical work in political science would be drastically impaired. But
survey research remains a purely applied science. If judged by the standards
prevailing elsewhere in the discipline, where good work is recognized for its
contribution to theoretical understanding, survey methodology has achieved
very little intellectual advance.
Second, political methodologists have expended much of their energies
teaching the rest of the discipline new statistical techniques invented in other
fields. In a typical publication of this kind, an intuitive exposition is com-
bined with an illustration or two from political science. The new method may
be compared favorably with an older, more common technique. The result is
an article that can be assigned to graduate students needing both motivation
and simplified mathematics to learn the material.
Intellectual middlemen have their uses, of course. Political science re-
mains a field with woefully little methodological capital, and any successful
investment scheme deserves praise. But remedial teaching is not scholarship,
and popularization does not a methodologist make. Here as in survey
research, there is no real intellectual achievement of our own to report.
Political methodologists as a class have largely avoided theoretical think-
ing. With few exceptions, we do not investigate carefully the properties of the
new methodological procedures constantly appearing in the discipline. Cer-
tainly we rarely invent a legitimate estimator, prove consistency theorems, or
derive confidence intervals. Instead we shop for hand-me-down techniques
invented by statisticians, psychologists, and economists-techniques often
meant for very different tasks. Empirical researchers then adopt the statistical
ideas and methodological standards we have propounded. ls it any wonder
that the pages of our journals frequently have the look of living rooms
decorated at garage sales?
The general lack of interest in fundamental questions is particularly dis-
quieting because it coexists with deep methodological conundrums specific to
political science. One thing political methodologists have done is to show that
measurement does matter. Apparently trivial differences in question wording
can lead to large changes in response patterns. (Among many possible exam-
ples, two particularly striking cases appear in Bishop, Tuchfarber, and
Oldendick, 1978; and Sullivan, Piereson, and Marcus, 1978.) Nor are the
peculiarities confined to survey items. The ghastly results all too common in
working with aggregate voting data are also well documented, and the
"ecological fallacy" has destroyed the credibility of many well-intentioned
projects. Indeed, skepticism about aggregate data is so widespread that the
 Christopher H. Achen I 71

quantitative historical research once so common in the discipline has very

nearly disappeared. Thus ignoring our methodological foundations because
"they probably don't matter much in practice" is simply naive. Whatever else
may be uncertain in methodology, there is ample evidence that the shaky
foundation it provides has weakened intellectual structures across the
discipline.
These puzzles in our data sets are not work-a-day procedural problems
with administrative solutions. Better question writing and more careful data
collection will make no more than a marginal difference. Only a better
understanding of our processes of measurement can truly help us. Nor can we
expect other disciplines to supply the necessary theory. In large part, the
statistical questions posed by political data are of relatively little interest out-
side political science. No one is likely to solve them but ourselves. So far , we
have not been much interested.
If political methodologists were to take up their main agenda , the first
item of business would be the formulation of our troubles clearly enough that
they could be diagnosed and treated. The remainder of this essay is meant to
begin this formulation . Of course, remarks of this length cannot hope to cover
the entire field. Political scientists are interested in simultaneous equation
estimation, time series, discrete choice analysis, and a host of other interesting
and important statistical methodologies not reviewed here in detail. None of
these statistical fields , however, raises difficulties so much our own as the two
long-standing research areas that are the focus of this article. The first is the
issue of measurement error, the second the aggregation problem. Each il-
lustrates in striking fashion the dimensions of the task before us.

MEASUREMENT ERROR
Quantitatively-trained political scientists are very much aware that
measurement error can bias their findings . Most have at least a vague recol-
lection from introductory econometrics that under some conditions, it can
attenuate regression coefficients. Frequently, several other features of
measurement error are "remembered" as well, for example:

(A) If some of the independent variables in a multiple regression are mea-

sured with error, all coefficients will be attenuated. Thus if one can
establish the existence of an effect in spite of measurement error, the true
effect must be even larger.
(B) Measurement error can be a serious problem in the bivariate regression
case, but the more independent variables without error in the regression
equation, the smaller the biases will be.
(C) If the measurement error variance in the independent variables is a small
fraction of their total variance (high reliability), the bias in the coeffi-
cients will be a small fraction of their true size.
(D) In any case, discussions of all these questions are contained in any stan-
dard introductory text on econometrics.
72 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY
All four statements are false. To take the last first, the surprising fact is
that some of the best multiple regression texts ignore the subject of measure-
ment error in the independent variables entirely (Kelejian & Oates, 1981;
Dhrymes, 1970). Nearly all the rest discuss the bias only in very general terms
Qohnston, 1972; Hanushek & Jackson, 1977; Pindyck & Rubinfeld, 1976;
Rao & Miller, 1971; and Theil, 1971). The direction of bias for the coeffi-
cients is given solely for random error in the bivariate case-a single indepen-
dent variable-where, of course, the effect is to drive the estimated coefficient
toward zero. (Rao and Miller also reference , but do not discuss, an antique
article by Theil that gives a rough approximation for the asymptotic bias
when there are exactly two independent variables, one of which is observed
with random error.) Only Maddala (1977) discusses non-random error (in the
bivariate case), or random error in multiple regression (with just two indepen-
dent variables, each observed with error).
None of these texts gives any explicit results when there are more than
two independent variables. In fact , as late as 1973, Econometrica was still ex-
ploring the case of multiple regression with a single variable observed with
random error, and finding the direction of the asymptotic bias only for the
variable with error. (It is biased toward zero.) A complicated procedure was
given for learning the direction of the biases in the other coefficients, but
nothing could be said about their sizes (Levi, 1973). More recently, as part of
a sophisticated and fairly lengthy treatment of random measurement error in
regression, Dhrymes (1978 , pp. 242-266) has shown that R 2 is driven down-
ward in errors-in-variables regression, which implies that the F-test for the
overall significance of the regression will also be depressed. But on the ques-
tion of the direction of the bias in coefficients or t-ratios, apart from the usual
bivariate regression result, he throws up his hands.
Apparently the only result in the literature more general than these was
published in Politz"cal Methodology (Greene, 1978). In the ordinary multiple
regression setup with k independent variables, each possibly observed with
random error, let g1 be the fractional asymptotic bias (inconsistency) in the
jth coefficient. For example, gj = - 0.5 means 50% attenuation. Then in
effect, Greene's main result is that a certain weighted average of these biases
must be negative. In particular, let (3j be the truejth coefficient and w} the er-
ror variance of the corresponding independent variable. Then:
k
E f3}w} g1 < 0. (1)
j=I

Thus in some sense, the "average gj' is negative, and measurement error
drives coefficient estimates toward zero.
Note, however, that this result is limited in several respects. First, as
Greene notes, it gives only an average direction of bias that may not apply to
most of the coefficients in question. In fact, the theorem guarantees only that
at least one of the coefficients will be driven down. The rest might actually be
increased in absolute size. Moreover, no limits on the size of the biases are
given. Nothing in the result guarantees that coefficients will be attenuated in
the usual sense that they will fall somewhere between the truth and zero. We
 Christopher H . Acken I 73

know that they will be driven toward zero, but nothing prevents their being
driven right on through it to the other side. Finally and more importantly, in-
dependent variables with no error (wj = 0) are weighted zero in (1). That is,
the inequality tells us nothing about the direction or size of the biases in
variables observed without error. Thus the result covers only the variables
with error. There appears to be no literature at all on the size of the biases in
the other coefficients.
Returning to the other three "facts" researchers remember from econo-
metric theory, one can see that the available literature neither supports nor
contradicts them. Answers to them simply are not known, and in their most
general form , the complexity of the mathematics will probably prevent much
progreas. However, a great deal can be done in certain simple cases.
Consider, then , the case of multiple regression with random measure-
ment error in a single independent variable, the other independent variables
being measured exactly. Then denote the variable with error by x 1, its error
variance by a;,and its true coefficient by /j 1. Finally, let the residual variance
when the tru~ first variable is regressed on the other independent variables be
s2 • Then if /j 1 is the estimated coefficient for the variable with error, its
"asymptotic value" (i.e. , probability limit) is as follows (see the appendix):
s2
ptim <Si> = S2 + 2 /j1. (2)
<Te

S
For any other estimated coefficient, say 1, its asymptotic value is:
~ ~
plim({j)
J
= fj1. + s2 + a.2 /jl b·J ' (3)

where b1 is the coefficient on thejth independent variable when the true first
independent variable is regressed on all the others.
With this machinery, the questions (A) , (B), and (C) raised above can be
addressed. Note first that the coefficient on the variable with error is indeed
attenuated. Equation (2) shows that asymptotically, the estimated value of
the coefficient will fall between zero and the truth. Nothing of the kind is true
for the other coefficients, however. As seen in (3) , other coefficients will "pick
up" part of the explanatory power of the variable with error, with the size of
the acquisition varying according to their correlation with it (b) . Thus
variables measured without error can have their effects increased or
decreased; there is no net tendency for them to be attenuated.
Next, observe that the size of the bias depends inversely on s2 , the
residual variance when the true first independent variable is regressed on the
others. Now s2 can be taken as a measure of collinearity: the more inter-
correlated the independent variables, the smaller is s2 , until at perfect col-
linearity, s2 = 0. It follows that high collinearity will induce relatively large
asymptotic biases, no matter how small the measurement error variance. The
coefficient for the variable with error, for example, can be driven arbitrarily
closer to zero just by increasing collinearity, even if the reliability is 99%.
Small error variances do not necessarily give much protection against large in-
consistencies. And of course, this bias is particularly pernicious because it
74 j TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY

depends on the unexplained variance in the regression of the true first in-
dependent variable on the others, a quantity that cannot be computed due to
the measurement error in the first variable.
Finally, note that s2 is quite likely to become smaller and smaller as addi-
tional variables are added to the equation, simply because the more variables
there are , the better that x 1 can be forecast and the smaller its residual
variance s2 will be. Thus additional independent variables will tend to reduce
s2 and raise the inconsistencies in (2) and (3), making the biases worse rather
than better. Therefore, as promised, we have shown that all four statements
(A)-(D) are false in general, since they are false for the simple case of one in-
dependent variable with error. Measurement error, then, is more dangerous
than commonly believed.
So far only purely random measurement error has been considered.
Unfortunately, political scientists can rarely be certain that their errors are
uncorrelated with the true values and with other independent variables. In
fact, in many applications just the reverse is suspected. For example, Ross and
Duvall ( 1982) have reminded us once again of the treacherous nature of cross-
national political information. They note that even the size of standing armies
in modern nations is subject to gross errors. In 1965, for instance, one major
source reports 132,000 men under arms in Norway, while another gives
36 ,000 as the correct figure (p. 31). Using sources like the International
Institute for Strategic Studies and the U.S. Arms Control and Disarmament
Agency, Ross and Duvall commonly find discrepancies of 50% or more in the
estimates for the same country in the same year throughout the modern era,
and the mean is almost 25 3. Some sources correlate rather well with each
other, others quite poorly. It seems unlikely that this kind of error is purely
random.
Unfortunately, nonrandom error presents even more serious threats to
inference than the random kind. Suppose that just one variable in a multiple
regression is observed with error, and that the error may be systematic in the
sense of being correlated with any of the independent variables. (Correlation
with the disturbance term, which can only make matters worse , will be
excluded.) Then we have this mathematically trivial but substantively sig-
nificant result: no matter what the true regression coefficients, if there is
systematic error in a single independent variable, ordinary regression can
converge to any pattern of coefficient estimates.

Proposition. Let {3* be any given vector of dimension k. Then in the multiple
regression setup with k independent variables, there exists a pattern of
systematic measurement error in the first independent variable such that the
OLS coefficients will tend asymptotically to {3*.

Proof Let the true model be:

(4)

where y is then-dimensional column vector of observations on the dependent

variable, the column vector x 1 contains the n observations on the first
 Christopher H. Achen I 75

independent variable and the n X (k - I) matrix X 2 the observations on the

other k - I independent variables, u is a disturbance term , and {3 1 and {32 are
a constant and a column vector, respectively, of coefficients to be estimated.
This relationship is not observed. Instead, we have:

(5)

where e is the vector of systematic error in x 1• Now partition {3* '

[{3 1* {32* '], where {3 1* is the first element of {3* (corresponding to x 1) , and {32*
is the vector of remaining elements (corresponding to X 2). Suppose that e has
the following form:

(6)

Now e is just a linear combination of the independent variables x 1 and X 2 .

Hence from standard regression theory, its only effect on the asymptotic
values (probability limits) of the coefficients is to transform them accordingly.
That is, suppose that the original relationship is:

y = X{3 + u.

Now suppose that the matrix of independent variables is linearly trans-

formed , which we can represent as postmultiplication of X by the nonsingular
square matrix A . Then regression theory tells us that the new coefficients will
be given by A - I {3. The new setup can be written as:

The obvious fact here is that (XA) (A- 1 {3) = X{3 . Given the transformed
data matrix XA, then, we can find the new coefficients simply by solving for
constants which, when multiplied by XA, will equal X{3.
In the systematic error problem, the estimated coefficients will be those
values of {3 1 and {3 2 which, when substituted in the transformed regression (5),
make (5) the same as the original regression (4). Elementary algebra will show
that when e is defined as above, (4) and (5) are equivalent when {3 1 and {32 are
replaced by {3 1* and {32 *, respectively. Thus, for any arbitrary coefficient vec-
tor {3*, there exists a systematic error structure (6) that will generate it. This
ends the proof

These developments make it clear just how dangerous to inference

measurement error is. Tossing raw attitudinal measures of survey responses
into regression equations or cross-tabulations is likely to lead to sensible con-
clusions only by chance. Correcting for measurement error in independent
variables alone guarantees useful results.
A number of correction procedures for measurement error have been
proposed. The first, and perhaps the most useful generally, is the set of tech-
niques known as factor analysis or scaling. These methods all assume that the
 76 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY

researcher has available several different measures of the same true under-
lying variables. Strong distributional assumptions are then added (e.g., all
measures are normally distributed, the underlying true scores are normally
distributed , the measurement errors are normally distributed, or even, quite
commonly, all of the above). Estimates of the true variables are then
produced.
A wide variety of models are available, suitable for continuous or discrete
variables, one underlying dimension or many, and several different theories
of how the scores were produced. (The most complete source remains Lord &
Novick, 1968. See also Torgerson, 1958; and Rasch, 1980.) These theories
constitute an impressive armamentarium, and they are far too little used and
understood in political science. But we stand a great distance off from making
a theoretically-informed choice among them. Do survey respondents obey the
assumptions of any of these models, even approximately? The answer is surely
no. If they do not, how much difference does it make? Brady (1981) has
begun to explore factor analytic models with no assumptions about either the
distributional form of the errors or about the functional form that relates fac-
tors to observed variates. A great deal more work of that kind will be
necessary before one can have full faith in scaling and factor analytic
methods. Intensive, theoretically-informed research on these techniques
should have a prominent place among methodologists' priorities.
Scaling methods are not the sole route to correcting regressions for
measurement error. Attractive and relatively simple correction procedures
also exist, for example, when the error variances are known or when they are
estimable (Johnston, 1972, pp. 281-291). Procedures for computing the stan-
dard errors of the adjusted coefficients have also been derived (Warren,
White, & Fuller, 1974; for a cruder but simpler approach, see Achen, 1978).
Sometimes, as with well-known scholastic tests like the SAT or GRE batteries,
reliabilities have been published. In other cases, more elaborate estimation
procedures are required.
In one case, political scientists have expended considerable effort to
learn a set of error variances. Converse (1964) argued brilliantly that most in-
dividuals have little conception of political issues. As part of his evidence, he
showed that a set of standard political attitude questions from the Michigan
Survey Research Center's National Election Studies (NES) correlated quite
poorly with themselves over a two-year interval, with Pearson's r's often no
more than 0.4. Implicitly assuming that the reliability of the questions was
1.0, he said that responses with so little stability indicated a lack of real
understanding. By contrast, individuals showed considerably more stable
party identification, demonstrating that party loyalties were genuine in a way
that issue preferences were not.
Converse added other evidence as well, such as the higher inter-item cor-
relations for elites than for mass samples, but his study (and virtually all such
elite-mass comparisons since) examined groups who were asked quite dif-
ferent questions. The elite questions were more precise, the mass items dif-
fuse. Differences in the correlations would be expected on that basis alone.
Similar objections apply to the comparison of elite over-time correlations with
the corresponding mass correlations, as in Kinder (1983). For this reason, the
 Christopher H. Achen I 77

stronger part of Converse's argument has always been the mass over-time cor-
relations considered on their own.
Since Converse's time, a series of scholars have tried to learn the relia-
bilities of the items to correct his correlations for measurement error. Con-
structing careful measurement models and exploiting the three measurements
in the NES 1965-58-60 panel study, they have derived reliabilities for the data
Converse considered. The uniform conclusion of them all (Asher, 1974;
Achen, 1975; Erikson, 1975; Dean & Moran, 1977; Jackson, 1979) has been
that, apart from the party identification question, reliabilities were quite low,
on the order of 0.5. Correcting for them typically brought the over-time at-
titude stabilities to 0.90 or more, meaning that most respondents were quite
stable in their true opinions. Moreover, respondents' measurement error
variances did not diminish much as they became better educated or more
interested in politics, indicating that measurement error was not primarily a
matter of misunderstanding the questions.
All of these revisionist studies require that the response uncertainties con-
stitute "measurement error" in the statistical sense. That is, instability must
be due to the questions, not to the respondents. Confidence on this point is
necessary if the corrected correlations are to be meaningful.
Several arguments blaming the respondents rather than the questions
have been developed. For example, it is sometimes said that if the measure-
ment error corrections are accurate, then the populace is very sophisticated
about politics-too sophisticated to be believable. Hence the measurement
models must be false. This argument misses the point of the revisionist litera-
ture. Nothing in the high over-time correlations implies that the citizenry has
a good grasp of political life. As noted repeatedly by the revisionists, opinions
may be stable for all sorts of unimpressive reasons. H. L. Mencken could
accept the measurement models without changing his estimate of American
intelligence.
In a similar fashion, some scholars have wondered whether the questions
could have been as badly worded as the measurement models seem to imply
(Converse & Markus, 1979; Kinder, 1983). But again, nothing in the
measurement literature implies anything of the kind. Rather, the low
reliabilities may simply reflect the difficulty of eliciting certain kinds of
attitudes; difficulties that no opinion researcher can escape. Even in an inter-
view with a foreign policy expert, for example, learning his general orienta-
tion toward American interventionism will be no simple matter. Inquiring
after his party identification will be far easier. There is no reason to believe
that all opinions are equally complex, even for individuals who understand
them thoroughly. Thus the fact that most opinion reliabilities are low, while
the party identification reliability is not, carries no necessary implication that
the opinion questions are badly worded or that one item is better written than
another. Objections based on those assumptions attribute a theoretical con-
tent to the models which they do not contain.
Other objections to the measurement literature have been based on col-
lateral evidence. Thus Converse and Markus (1979) have argued that in two-
wave panel studies, attitudes toward a political candidate correlate over time
in proportion to the candidate's visibility. Feelings about Edward Kennedy
78 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY

are more stable than those about Henry Jackson. Since the question wordings
are essentially identical, differences in stability must be due to differences in
comprehension. Thus low correlations show low knowledge. Converse and
Markus conclude that something similar must be going on with the original
Converse items, so that the measurement models were mistaken in finding
that true policy opinions were relatively stable. In their view, "measurement
error" consists primarily of misunderstanding.
Converse and Markus's argument is inventive but not convincing. Its
logic is this:

1. Low over-time correlations in candidate preference items are due to a lack

of understanding on the part of respondents.
2. The revisionists' models imply that candidate preference items correlate
poorly over time solely because of random measurement error.
3. Therefore the measurement models are mistaken.

The conclusion here certainly follows if the two premises are correct, and
Converse and Markus make a good case for the first premise. Unfortunately,
they present no evidence at all for the second one. They apparently assume
that the measurement models would give the same result for candidate prefer-
ences as for policy views, so that the low correlations would be due to random
error in both cases. But it is by no means clear that this supposition is true.
Why should the rather precise notion of candidate preference have the same
measurement characteristics as the necessarily nebulous policy attitudes?
Without the second premise, the issue is not joined. It does no harm to the
case for stable policy attitudes to point out that voters sometimes fail to learn
much about unsuccessful Presidential candidates.
A serious indictment of the measurement models using the Converse-
Markus data must support the second premise above: that the candidate
preferences behave as policy preferences do. Two pieces of evidence are need-
ed. First, the measurement models would have to be shown to yield high over-
time stabilities for both visible and obscure candidates. Secondly, the
measurement variances would have to be demonstrated to vary little with dif-
ferences in education, political interest, or other relevant proxies for sophis-
tication. Absent either one of these results, the measurement models would
imply precisely what Converse and Markus say is true, namely that the dif-
ferences in stabilities for candidates reflect genuine differences in under-
standing. Without these two findings , then, Converse and his critics agree on
the analysis of candidate preferences, and those data are simply irrelevant.
Since Converse and Markus provide no evidence on either point, the measure-
ment models escape unscathed.
One is left, then, with the conclusion that the revisionists' case has yet to
be seriously challenged. No one has been able to formulate a measurement
model for Converse's data which supports his analysis. This consensus might
seem to give hope to political scientists facing error-laden data. Unfortunate-
ly, matters are not so simple. All the available measurement models are close-
ly related to a model of D. Wiley and J. Wiley (1970), in which attitude
change over time is supposed to follow a first-order autoregressive law with
 Christopher H. Achen I 79

purely random disturbances. In the Wiley-Wiley (1970) model, let observed

survey responses (standardized to mean zero) at times 1, 2, and 3 be denoted
by Xi. x 2 , and x 3 , and let u 1 be the true attitude at time 1. Let v1 and v 2 be
random disturbances in true attitude, and e 1 , e2 , and e3 be measurement
errors at times 1, 2, and 3, both sets of which are assumed to have mean zero
and to be distributed independently of the u's. In addition, the errors are
assumed to have constant variance over time. Finally, let r 1 and r2 be con-
stants. Then the equations for the three time periods in the panel study are:

X1 = U1 + e1
X2 = T1 U1 + V1 + e2
Xg = r2(T1U1 + V1) + V2 + eg (7)

The six variances and covariances may be used to solve for the six parameters,
namely, Ti. r 2 , and the variances of u., vi. and v2 , plus the error variance.
Other models can be imagined. For example, one might want to make
the measurement errors slightly systematic, so that they were correlated over
time. This corresponds to the notion that people may answer the same ques-
tion the same way on two different occasions at least partly because the same
extraneous forces impinge on them at both periods. For instance, the same in-
terviewer may have come around again, the same election-time propaganda
may have appeared on the doorstep, and so on. In short, individuals may be
stable from one election to the next for reasons having little to do with the
persistence of attitudes in their own minds.
A model of this kind has been proposed by J. Wiley and M. Wiley (1974).
It differs from the D. Wiley-J. Wiley model in just two respects: first, the
autoregressive parameter is fixed equal over the second and third time periods
(r1 = r 2); and second, the errors of measurement are assumed to obey a first-
order autoregressive scheme with lag coefficient s. Thus the equations are as
follows:

X1 = U1 + e1

X2 = T1 U1 + V1 + se1 + e2
Xg = T1 (r1 U1-+ V1) + V2 + se2 + eg. (8)

The six parameters here are the same as before, except thats replaces r 2 , and
they may be estimated in the same way.
The results of applying these two models to two of the questions used by
Converse are given in Table 1. Both models fit the data exactly, in the sense of
reproducing the variance matrix with perfect accuracy. Yet their conclusions
are astonishingly different. If the first model, which is standard in the litera-
ture, is correct, the over-time stabilities are very nearly 1.0, more than twice
as large as the raw correlations, making Converse dead wrong. On the other
hand, if the second model is the truth, most of the over-time correlation is due
to correlations between error terms, and the true stabilities are about 0.2, just
half the size of the raw coefficients. In that case, attitudes at one time period
statistically explain almost nothing in attitudes two years later (in no case
 80 J TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY

more than 73 of the variance), so that Converse was even more right than he
imagined.
With present knowledge, no grounds exist for choosing between these
competing models. The rather dreary conclusion follows that even in the most
heavily studied case, no confidence about the error variances is possible. In
less trodden fields, the situation is inevitably more desperate. Bluntly put,
political scientists do not know what survey responses are measuring.
The fundamental problem here is lack of a mathematical theory of the
survey response. As a recent report of the National Academy of Science
(Turner & Martin, 1981) has emphasized, we have little or no rigorous under-
standing of the effects of question wording, question order, and response
error. And if these difficulties are severe in the case of ordinary surveys, they
become crippling in the case of important international studies like Almond
and Verba (1965), as was emphasized by Scheuch (1968). Real progress waits
on basic theoretical research by political methodologists interested in surveys.
Even without a theory, a great deal could be learned by extending our
current techniques to lengthier panel studies. For example, with just four
waves of interviews, the two Wiley-Wiley models could be combined and
tested. The new model would subsume (7) and (8) and extend them to a
fourth wave in the obvious way, allowing for first-order autoregressive struc-

TABLE 1
Overtime Correlations of Political Attitudes Corrected for
Measurement Error Under Two Different Models
(Original Items Are from Converse, 1964)

Guaranteed Jobs Housing and Power

56-58 58-60 56-60 56-58 58-60 56-60

D. Wiley-J. Wiley Model

(1970)
rawr .45 .47 .43 .37 .36 .37
corrected r .95 .99 .94 .99 .99 .98

56 58 60 56 58 60
reliability .46 .49 .45 .38 .36 .38

56-58 58-60 56-60 56-58 58-60 56-60

J . Wiley-M. Wiley Model
(1974)
rawr .45 .47 .43 .37 .36 .37
corrected r .25 .27 .07 .18 .18 .03

56 58 60 56 58 60
reliability .84 .85 .84 .88 .87 .88
 Christopher H. Acken I 81

tures in both true opinions and measurement errors, with the autoregressive
parameters fixed equal over time for the errors. Thus the lag coefficients for
true opinions would be ri. r2 , and rg, and the lag parameters for the errors
would be s. Along with the measurement error variance and the variances of
ui. v 1 , v2 , and Vg, there would be nine quantities to estimate, but ten variances
and covariances in the data. Thus one could distinguish these two models
(and many others) statistically and evaluate how well they fit. A better
understanding of the Converse problem would result , and more importantly,
clues would be provided as to how mathematical modeling of survey responses
might proceed.
The urgency of theoretical research on the survey response bears empha-
sizing. Correcting for measurement error is the heart of any use of survey
data. As yet, we simply do not know how to do that.

AGGREGATION BIAS
When political scientists doing empirical work are not using survey data,
they are most often working with aggregate data of one sort or another: cen-
sus information, voting returns by county or electoral district, crime statistics
by state, and so on. These data are often cheaper to collect than survey infor-
mation, and they extend back considerably further in time than do surveys. If
one wants to know who voted for the Nazis, for instance, there is little alter-
native to working with aggregate elections returns. Moreover, even when
surveys are feasible and affordable, aggregate information may be preferable.
If one wants to study the effect of capital punishment on the murder rate,
murder statistics are likely to yield more trustworthy conclusions than any
possible survey. The answers to questions like "Have you ever murdered any-
one?" or "Have you ever wanted to murder someone but were deterred by the
thought of the electric chair?" would not be of much value.
Since Robinson (1951), social scientists have been painfully aware that
ecological data were full of pitfalls. If one has a proposition cast at the
individual level (e.g., Catholics were more likely to vote for the Nazis), voting
returns from constituencies allow no direct test of it. Catholic constituencies
might be more likely to vote Nazi even though individual Catholics are not.
Robinson spoke of correlations, but his point is perfectly general. Goodman's
(1953) "ecological regression" has the same problems.

Aggregation Bias as Contextual Effect

Two reactions are possible. In the first, one simply ignores the individual
level of analysis. All explanations are cast at the constituency level, and the
ecological inference problem disappears. In this model , single citizens do not
vote; constituencies do. The statistical demand of this approach is that there
be a unitary actor at the constituency level. Interactions among individuals do
not suffice to create group-level explanations; processes of that kind can be
formulated mathematically at the individual level (Erbring & Young, 1979).
A legitimate "group mind" is needed to make the group-level explanation
coherent.
 82 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY
Theories of group mind have always depended for their credibility on
their imprecision. Stated clearly, they lack persuasive power. If the voting
decision is truly made at the group level, one must come very near saying that
the constituency rises sleepily on election day, rubs its eyes, takes thought, and
says, "I imagine I will vote 373 Nazi." Researchers who find this a veridical
theory of voting will be able to use logistic transformations of vote percentages
-which destroy all chances of talking about the voters themselves (see Han-
nan, 1971, pp . 23-30)-and every other technique of quantitative analysis
without qualm. No statistz"cal difficulties occur.
For political scientists who cling to the conventional wisdom that
individual citizens do the voting, more complex responses are needed. First,
one may derive the conditions under which the ecological inferences will be
the same as the individual-level ones. Theil (1954) considered the case of a
single aggregate unit observed over time. His conclusions were extremely
pessimistic. Roughly, every individual within the aggregate had to obey iden-
tical statistical laws or had to respond to changes in aggregate statistics in the
same way. In practice, of course, neither is true.
Although no similar results are available for the cross-sectional case more
common in political science, practical examples have been just as depressing
as Theil's theoretical results. In Goodman's ecological regression model, for
example, one might want to find the proportion of Democratic voters at time
1 who continued to vote Democratic at time 2, and the proportion of Repub -
lican voters at time 1 who switched to the Democrats at time 2. Call these pro-
portions P and Q, respectively. Then if the fraction of Democratic votes at
time 1 is denoted by D 1 and the same fraction at time 2 by D 2 , and if everyone
votes either Democratic or Republican, we have the simple accounting
relationship:

(9)

If P and Qare constant across constituencies, or at least vary in ways that can
be absorbed into a random disturbance term u, then (9) can be applied to a
set of constituencies. Thus if i denotes the constituency number, we have from
(9) in an obvious notation:

D2; = Q + (P - Q)D1; + U;, (10)

which has the form of a bivariate regression with intercept Q and slope P- Q.
Thus one simply regresses the proportion Democratic at time 2 on the same
proportion at time 1, and solves for P and Q from the resulting slope and in-
tercept. It is quite easy to extend the model to multiple parties, abstention,
controls for demographics, and so on, just by adding more terms to the
regression and estimating additional regressions.
Computational experience with this model has been most unhappy.
Commonly, the intercept is too small or even negative, and the slope is so
large that the estimate of Pis too high or even above one. Thus meaningless
probabilities below zero and above one result. In fact, a substantial literature
has developed to force the usual implausible estimates into the meaningful
 Christopher H. Achen I 83
range from zero to one (e.g. , Irwin & Meeter, 1969; Crewe & Payne, 1975). So
far, the corrective treatments merely palliate the biases.
The standard interpretation of aggregation errors is to attribute them to
"contextual effects." Ecological results are said to differ from individual find-
ings because people influence each other. For example, heavily Democratic
districts will produce Democrats who vote even more heavily Democratic than
they otherwise would, simply because the environment drives them in that
direction. Thus an individual-level-though, of course, not individualistic-
explanation is used to account for the aggregate findings .
Much of this literature is not closely related to data , with predictable
results. For example, literally dozens of articles have been written arguing
that ecological regressions like Goodman's need a quadratic term. The argu-
ment goes that P and Q are not really constant across constituencies but in-
stead depend on Dli· Making P and Q linear functions of D 1;. and substituting
in the original equation (10) yields a quadratic equation in Dli. Unfortunately
for this line of thinking, almost all ecological regressions exhibit strikingly
linear relationships. The key to Goodman's woes must lie elsewhere.
The most specific and best grounded contextual argument is Butler and
Stokes (1969, pp. 303-312). First, they give examples to show that the well-
documented phenomenon of "uniform swing" in Britain is incompatible with
Goodman's model. Their point holds generally. Formally, define uniform
swing by the condition that for all i and some constant k, D 2; = Dli + k. Tak-
ing expectations in (10) and rearranging:

k = Q + (P - Q - l)Dli. (11)

Since k and Q are fixed, if (11) is to hold for all Dli. we must have
P - Q - I = 0, or:

P=Q+l. (12)

But then for any Q > 0, we have P > 1, which is impossible. Hence Good-
man's model cannot be correct for any electoral system capable of uniform
swing.
Butler and Stokes propose to modify Goodman by distinguishing two
groups of voters. The first group attend to national media and thus respond
to national forces, obeying Goodman's model. The second group are locally-
oriented and take their cues from the partisan context of their local constitu-
ency. Butler and Stokes show that combinations of these two forces can pro-
duce uniform swing, and they give some evidence that in the 1964-1966 elec-
tions, the two groups exhibited the approximate qualitative behavior
expected from the model.
Sad-to-say, the great difficulty of producing adequate models of con-
textual effects suitable to aggregate data can be illustrated even in this case.
Formalize Butler and Stokes in the following way. Suppose that, in addition
to the usual Goodman effects, a contextual force operates in proportion to the
difference in the party strength. Following Butler-Stokes, let us measure this
difference at the previous election. Then the contextual force is proportional
 84 j TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY

to Dli - (1 - Dli), or 2Dli - 1. Letting its coefficient be G and adding this

effect to the original Goodman equation (10), we have 1 :

D2; = Q+ (P - Q)D1; + G(2Dli - 1) + u;,

or:

D2i 5 (Q - G) + (P - Q + 2G)D1, + U;. (13)

At first glance, Butler and Stokes appear to have broken through to high
ground. Goodman's model tends to produce intercept terms too small and
slopes too large, and equation (13) seems to explain why. The intercept is
depressed by the amount G; the slope is increased by 2G. The equation is
perfectly linear, just as Goodman and the data had assured us, but the coeffi-
cients must be interpreted differently than Goodman had supposed.
Unfortunately, the model does not hold up upon closer examination.
Return again to an electoral system with uniform swing. Under the Butler-
Stokes model, two statements can then be made. First, in a constituency in
which the two parties each received 503 of the vote at time 1, no contextual
effect operates at time 2. That is, if Dli = 0.50, 2D 1; - 1 = 0, then the
original Goodman model must fit. Making use of uniform swing gives (11)
with D 1 = 0.5:

k = Q + (P - Q - 1)0.5. (14)

Next, we note that (13) must also hold when averaged over constituencies.
Letting D 1 be the mean vote at time 1, exploiting the fact of uniform swing,
and taking expectations:

k = (Q-G) + (P-Q+2G-l)D 1 . (15)

The problem now becomes apparent. In any given election, P, Q, and G

are parameters determined by the contests at times 1 and 2, plus the nature of
the contextual effect. In addition, i'i; is fixed. Thus these four factors can be
taken as given. Among them, they determine the size of the uniform swing.
Unfortunately, they do so in two distinct ways, (14) and (15). In general, the
two equations are not consistent.
As an example, consider two elections with P = 0.8, Q = 0.3, G = 0.1,
and D 1 = 0.4. (Incidentally, though it is beside the point, these are all quite
reasonable values that might be expected to occur in an actual election.)
Then (14) gives a swing of 53 , while (15) implies an 8% swing. Thus the
model contradicts itself. Butler and Stokes cannot cope with uniform swing
any more than Goodman can.
The dilemma cannot be escaped by making contextual effects dependent
on the vote at the second time period. Though this assumption makes more
sense substantively, essentially the same problem occurs. In this case, the
Goodman model must fit constituencies where the vote at time 2 is 50 3, plus
 Christopher H. Acken I 85

fit a modified version of equation (15). Consistency again occurs only by

chance.
As this example makes clear, even our most sophisticated aggregate-level
contextual models have important lacunae. Contextual effects simply are not
well understood. In spite of an enormous literature extolling their impor-
tance, almost no one has suggested explicit models by which they might
operate. The key exception is Erbring and Young (1979), an article that sum-
marizes and eviscerates an enormous amount of careless thinking on this
topic. Although entirely focused on individual-level data , in principle this
article proposes for the first time the explicit statistical models of social inter-
action that contextual theories assume. Whether its very heavy data demands
will be met in the near future is questionable. But success at the individual
level might make possible an inference to the aggregate-level model that is so
badly needed.
Whether a contextual effect is there to be found remains an open ques-
tion. As Prysby (1976) notes, even the best arguments in its favor are deeply
flawed, and their evidence is certainly consistent with its absence. Weather-
ford (1982) finds that most people do not talk politics with their neighbors in
any case, which eliminates the usual mechanism for communicating con-
textual effects. Finally, in one of the most direct tests of the contextual
hypothesis, Tate (1974) showed that a large number of contextual variables
proved to be useless in predicting individual-level British voting behavior. In
short, the form of contextual effects, their internal logic, and even whether
they exist at all remain unknown. Serious theoretical research on deriving an
aggregate-level contextual model from empirically-verified assumptions
about individual-level interactions should have a high priority for students of
social context.

Aggregate Bias as Specift:catz"on Error

Since Hanushek, Jackson, and Kain (1974), many researchers have taken
up the view that aggregation error occurs because of specification errors. On
this view, contextual effects may or may not exist, but aggregation bias would
be expected in any case, simply because erroneous equations neve~ yield the
right answer. Hence the task is to properly specify the regression equations so
that bias disappears. Hanushek et al. show that Robinson's original example
is very nearly corrected by adding variables to the equation to specify it more
accurately.
Needless-to-say, better specifications are always welcome. But one can
question whether this formulation resolves the theoretical question. "Specifi-
cation error" is often taken to mean a deviation from the one true causal law.
On this logic, aggregation bias occurs because some of the causally important
variables are omitted from the equation. Adding them, it is said, eliminates
the bias.
A moment's thought will show that in this form, eliminating specifica-
tion error is a hopeless task. For we are not very likely ever to have perfect
knowledge. Causally important variables are omitted from every social
science regression, simply because not everything is measurable. If good
 86 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY
estimates from aggregate data wait on specifications with all causal factors
included, no progress can be hoped for in our lifetimes. More importantly,
adding a variable here and there may make marginal improvements, but no
guarantee exists that the bias is thereby reduced. Statistical theory demon-
strates that the one true model is better than any other, but not much more.
An equation missing several key causal factors is not necessarily improved by
adding one of them; the biases in fact may get worse.
If on the other hand, one takes "proper specification" of a regression
equation to mean that its independent variables are uncorrelated with the
disturbance term (the conventional statistical definition), the notion that the
absence of specification error implies the absence of aggregation bias is simply
false. Take the case of Goodman's model. It is purely descriptive and hence
perfectly well specified at the individual level. With individual level data, its
independent variables are necessarily uncorrelated with its disturbance, and
accurate estimates result. However, with aggregate data, it often gives mean-
ingless results, even with large amounts of data. Thus proper specification in
this sense is not the answer either.
We are left then with just one possibility. "Proper specification" might
mean that at the aggregate level, the disturbances are uncorrelated with the
independent variables. Now of course this does guarantee asymptotically
accurate results. But what are we to make of it? What sort of modification of
our specifications is called for? Our theoretical knowledge is entirely at the in-
dividual level, and we rarely understand much of the grouping procedure
that created the geographic districts. The advice to seek proper specification,
interpreted in this fashion, is either tautological (Avoid bias!) or else quite dif-
ficult to interpret. What characteristics of the indz'tridual-level specifz'cations
lead to uncorrelated disturbances at the aggregate level? That is the central
question for this point of view, and its answer is unknown.
What makes the aggregation problem so severe, then, is this dilemma.
Aggregate data are often superior to individual-level information, even when
no contextual effect or group mind is at work. But they commonly lead to
severe biases, of whose solution we understand almost nothing. For serial cor-
relation, selection bias, underidentification, or the other difficulties routinely
encountered in non-experimental work, every textbook carries straightfor-
ward counsel on eliminating the biases. One may not always be willing or able
to take the advice, but there is never any doubt about what should be done.
One looks in vain for similar instruction on aggregation. At this stage in our
knowledge, the aggregation problem simply is in a class by itself. Clearly, in-
tensive theoretical research on this problem deserves high priority in
methodologists' research programs.

CONCLUSION
The state of our knowledge in political methodology gives little cause for
self-congratulation. The two principal topics that have concerned us,
measurement error and aggregation bias, remain both poorly formulated and
even more poorly understood. Enough data has been collected that one could
surely recognize a successful solution to either one if it were presented. But the
 Christopher H. Achen I 87

anemic mathematics that has been applied to both tasks is not likely to pro-
duce one.
More than anything else, what would help are formal theories with
measurement models built into them. These have been common in psychol-
ogy (see, for example, Atkinson, Bower, & Crothers, 1965) and sociology (see
Berger, Cohen, Snell, & Zelditch, 1962) for two decades now. Political science
remains dominated by fact -free theory of the verbal or rational-choice
variety. In a charming exception, Enelow and Hinich (1983) have recently
attempted to link the spatial theory of voting to factor analytic models of
public opinion. Work of that kind remains all too rare.
LISREL-type models (Joreskog & Sorbom, 1979) also constitute a partial
exception. These sophisticated structures offer the opportunity to mesh sub-
stantive specifications with measurement models. In a very general way, they
represent the direction in which empirical work must move, and political
scientists have not been slow to exploit them. But here again our theoretical
limitations are painfully obvious. First of all, we often have no knowledge of
either the substantive or the measurement model, and so assume linearity
without further argument. Doing so may be better than ignoring measure-
ment altogther, but it creates additional doubt in what is usually a compli-
cated specification with enough credibility problems to go around. No
LISREL model abolishes the theoretical questions discussed earlier.
Even ignoring these first difficulties, LISREL models raise many inferen-
tial questions. They assume normal distributions for every variable in the
equations, including both measured and unmeasured variables, plus dis-
turbances. Thus one dummy exogenous variable will violate the postulates.
No doubt the consistency of the estimates will survive such minor transgres-
sions (though this has not been proved, to my knowledge), but the standard
errors of the coefficients will not. Thus even if all the other assumptions held,
it would be a safe bet that every LISREL standard error published in a politi-
cal science journal is erroneous. Perhaps jackknifing the estimates would
help; again, no such result is known.
Other limitations of the standard LISREL model have become clear in
applications. All equations are assumed linear. No interaction effects can be
employed, no quadratic terms, no transformations of variables already used
untransformed elsewhere in the model. Moreover, variables must be con-
tinuous. Dichotomous or polychotomous endogenous variables are beyond the
capacity of the system (except in rather specialized cases, e.g. , all exogenous
variables normally distributed). Brady (1982) has investigated the extension
of LISREL models to discrete data , but very little else has been accomplished
in political science. Here as everywhere else, there is much theoretical work
remaining.
The progress that might be expected from more explicit attention to
methodological theory became clear in the recent debate on how economic
conditions affect the vote (Kramer, 1971; Kinder & Kiewiet, 1979; and
others). Initially, one side used over-time aggregate data which implied
strong economic effects on the vote, and the other studied survey responses
which showed none. Since neither side had a measurement model, contradic-
tion and confusion were the inevitable result.
88 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY
Happily, however, Kramer (1983) has produced formal statistical models
of the theories on each side. He assumes that citizens respond politically to
government-induced changes in their income, while researchers can observe
(with either survey or aggregate data) only the total income change. Thus a
measurement error is generated. As his main finding, Kramer shows that, as
one would expect under his assumptions, aggregate-level data largely wash
out the errors. Though not unbiased, they are greatly superior to the
individual-level survey responses. In this case, then, aggregate data are to be
preferred, an iconoclastic finding reminiscent of Grunfeld and Griliches
(1960) . Similarly, Kramer shows that if voters respond to the true overall state
of the economy rather than their own personal situations ("sociotropic
voting"), an interpretation seemingly supported by some of the cross-sectional
survey data, then in fact. those data are hopelessly biased. Because there is no
variance in the state of the economy at a single time period, a cross-section
contains no statistical information about sociotropic voting. Once again,
aggregate data are superior.
Kramer's powerful argument poses a severe challenge to the conventional
wisdom about aggregate data. At a minimum, it dramatically raises the level
of the debate, suggesting that our lack of understanding of aggregation may
not be so debilitating after all. Yet the force of his conclusions can be ques-
tioned. As Kramer himself notes, aggregate data are inadequate in principle
to disentangle self-interested from sociotropic voting. Regardless of whether
citizens respond to their own income changes or to changes in national in-
come, the effect at the aggregate level is the same: national income changes
predict the vote. What Kramer shows is that if we must choose between
individual-level cross-section data with uncorrectable errors and over-time
aggregate data with the same weaknesses, the latter are to be preferred in
studying economic effects on the vote. But since the aggregate data cannot
answer the central question, and probably have biases beyond those Kramer
mentions2 , his result shows only that in the land of the blind, the one-eyed
man is king. If depth perception is needed, it may be best to find a new king.
Kramer's findings do leave an alternative open to the devotees of
individual-level data: the estimation of a substantively plausible model taking
account of measurement error in the survey responses (both Kramer's kind of
error and ordinary response error). In theory, truly adequate estimates could
be produced from a single cross-section, at least if we relax Kramer's assump-
tion that in sociotropic voting, everyone acts on the same view of the national
economy. In practice, however, panel studies seem more likely to be helpful,
not only because they generate real variation in the national state of the
economy and thus escape Kramer's critique, but also because they contain
repeated measures and lagged instrumental variables to control measurement
error. A panel study with individual-level survey responses and adjustments
for measurement error would be greatly superior to aggregate data, even
under Kramer's postulates. Most importantly, it would distinguish sociotropic
from individualistic voting, which aggregate data cannot hope to do. Of
course, as noted earlier, correcting for measurement error in the survey
responses will be no trivial affair. The plausible substantive assumptions and
analytic techniques that would guarantee consistent estimates are not yet
 Christopher H. Achen I 89

evident. Certainly, intriguing methodological challenges on both sides of this

topic remain to be investigated. Kramer's work demonstrates just how great
the rewards of progress on our principal agenda might be.
Mounting a sustained attack on such questions across political science
may demand some changes in our work habits. Fundamental research must
come to take priority, at least some of the time, over applied work. It should
be possible to get funding, publish respectably, and make a career studying
the principal agenda of political methodology. And to provide for the future,
much better mathematical training will have to be provided at both the
graduate and undergraduate level.
Basic methodological research will not be of general interest to the pro-
fession. Almost by definition, any work simple enough to be understood
widely will not dent the problems. Politi.cal Methodology, if it is to keep its
leadership position in the field, will become more specialized and technical.
To some degree this has already occurred, just as it did in economics and
psychology in an earlier era. So long as it does not lead to sterile theorizing,
the growing separation is a sign of health and progress.
In sum, the tenor of this essay should be taken as optimistic rather than
pessimistic. Certainly it provides no excuse to abandon quantitative work.
One does not escape logical gaps in an argument by becoming less rigorous,
or by abandoning logic entirely. We have no choice but to find out how peo-
ple answer survey questions, how national and international agencies produce
cross-national data , and what aggregation bias amounts to.
A beginning has been made. Political methodologists may still be
rewarded more for "substantive" (i.e., merely applied) work than for "tech-
nical" (i.e., theoretical) contributions, but their numbers grow and the
product improves. Poli'ti'cal Methodology may look like samizdat, but its
sophistication steadily distances it from the more prestigious journals. And
even the fact that our theoretical failings have Jed to so many anomalies gives
cause for hope. For if we can come to face our weaknesses, then in the manner
of the alcoholic standing up for the first time to confess his drunkenness, a
major step toward improvement may have been taken.

APPENDIX
This appendix gives a proof of the inconsistency result for multiple
regression when a single independent variable is observed with error. The true
relationship is assumed to be

y = X*{3 + u, (Al)

where y is an n-dimensional vector of observations on the dependent variable,

X * is an n X k matrix of observations on the fixed true values of the indepen-
dent variables, (3 is a coefficient vector to be estimated, and u is a disturbance
term whose elements are mutually independent. The usual regression
assumptions are made: the disturbances are assumed to have mean zero, con-
stant variance, and zero correlation with the true independent variables.
That is, E(u) = 0, E(uu ') = a!I, and E(X* 'u) = 0.
 90 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY
All but one of the independent variables are observed without error. The
first true independent variable, x 1*, is not observed but is related to an
observed variable, X1> as follows:

(A2)

where x 1, x 1*, and e1 are each n -dimensional vectors, and where it is assumed
that the elements of e 1 are mutually independent, and that they have mean
zero, constant variance, and zero correlations with the true independent
variables and the disturbances. ThusE(e 1) = 0, E(e 1 e1 ') = u!I,E(X* 'e 1) =
0, and E(u' e 1) = 0. Finally, to guarantee that certain probability limits exist,
it will also be postulated that lim X * 'X *In = 0, a constant positive definite
matrix, and that the elements of u and e1 have uniformly bounded absolute
fourth moments.
Letting E = [e1 0 ... O] , it follows that plim E 'Eln = E , a matrix which
is zero everywhere except for its upper left comer element, which is u~. Now
let {3 1 be the first element of {3 and let X be the matrix of observed indepen-
dent variables, so that X = X* + E. Then write (1) as:

y = X{3 + u - E{3
= X{3 + u - f31e1. (A3)

Then the OLS estimate is:

S= (X'X) - 1 (X'y)
= {3 + (X'X/n) - 1(X* + E)' (u - f31e1)/n. (A4)

Thus:

plim (~) {3 + (0 + E) - 1 (A5)

Now partition the matrix of observed independent variables as X = [x 1 X 2]

and similarly {3' = [{3 1 {3 2 '].Thus x 1 is the first independent variable and {3 1 is
its coefficient, while X 2 is the matrix of all the other independent variables
and {3 2 is its coefficient vector. Set b = (X 2 'X2 ) - 1 X 2 'x ~ and s2 = [x 1 'x 1 -
x~ 'X 2(X 2 'X 2)- 1 X 2 'x1] / n. Hence b is just the coefficient vector when x1 is
regressed on X 2 , and s2 is the corresponding residual variance. Now using
standard results on partitioned inverses (e.g., Theil, 1971, p . 18), we find
from (A5):
s2
plim (~ 1 ) = 52
+ a; {3 1 • (A6)
 Christopher H. Acken J 91

and

(A7)

The last equation is a relationship between vectors. Taking any single

line from it, say the jth, gives the result (3) in the text.

NOTES
1. To avoid unnecessary complexity, the division of the population into locally and
nationally oriented groups has been suppressed. If these groups have constant pro-
portions across constituencies, for example, (13) can easily be rewritten to reflect
the fact that the contextual effect applies only to the former group and the Good-
man model only to the latter. As written, (13) absorbs the proportionality weights
for the two groups into the parameters P, Q. and G.
2. Another of Kramer's assumptions is that in expectation, all citizens respond iden-
tically to income changes (in dollars), meaning that their regression coefficients are
all equal. This postulate allows him to escape aggregation bias (Theil, 1954).
However, systematic variation in political responses to income change seems quite
likely, even within the rational-choice framework Kramer adopts. Diminishing
marginal utility of income would generate it, for example. Thus his aggregate level
estimates probably have additional biases beyond those he considers. Some random
coefficient models also generate biases in cross-sectional or even panel data, of
course, so that the relative magnitudes of the biases need investigation. The
advantage of individual-level data is that more powerful statistical techniques can
eliminate those biases. With aggregate data, the parameters of coefficient variation
are hopelessly underidentified.

REFERENCES
Achen, Christopher H. Mass political attitudes and the survey response. American
Political Science Review, 1975, 69, 1218-23.
Achen, Christopher. Measuring representation. American journal of Political Science,
1978, 22, 475 -510.
Almond, Gabriel A., & Verba, Sidney. The civic culture. Boston: Little, Brown,
1965 .
Asher, Herbert B. Some consequences of measurement error in survey data. Ameri-
can journal of Political Science, 1974, 18, 469-85.
Atkinson, Richard C., Bower, Gordon H. , & Crothers, Edwardj. An introduction to
mathematical learning theory. New York: Wiley, 1965.
Berger, Joseph, Cohen, Bernard P. , Snell, J. Lauri , & Zelditch, Morris, Jr. Types of
formalization in small-group research. Boston: Houghton Mifflin, 1962 .
Bishop, George F., Tuchfarber Alfred J. , & Oldendick, Robert W . Change in the
structure of American political attitudes: The nagging question of question word-
ing. American journal of Political Science, 1978, 22, 250-69 .
Brady, Henry E. The factor analysis of ordinal preference data. Working Paper 106.
Graduate School of Public Policy, University of California , Berkeley, 1981.
Brady, Henry E. Random utility models of primary preferences with latent variables
and undecided voters. Paper presented at the meetings of the Public Choice
Society, San Antonio, Texas, March 1982 .
92 I TOWARD THEORIES OF DATA: THE STATE OF POLITICAL METHODOLOGY
Butler, David, & Stokes, Donald. Political change in Britai·n. New York: St. Martin's,
1969.
Converse, Philip. The nature of belief systems in mass publics. In David E. Apter,
Ed. , Ideology and discontent. New York: Free Press, 1964, pp. 206-61.
Converse, Philip, & Markus, G. B. Plus ca change ... The new CPS election study
panel. American Political Science Review, 1979, 73, 32-49.
Crewe, Ivor, & Payne, Clive. Another game with nature: An ecological regression
model of the British two-party vote ratio in 1970. Bn:tish journal of Poli"tical
Science, 1975 , 6, 43-81.
Dean, Gillian, & Moran, Thomas W . Measuring mass political attitudes: Change and
unreliability. Pol£tical Methodology, 1977, 4, 383-414.
Dhrymes, Phoebus. Econometrics. New York: Harper and Row, 1970.
Dhrymes, Phoebus. Introductory econometrics. New York: Springer-Verlag, 1978.
Enelow, James, & Hinich, Melvin J. The spatial theory of voti"ng. Cambridge:
Cambridge University Press, forthcoming.
Erbring, Lutz, & Young, Alice A. Individuals and social structure. Sodological
Methods and Research, 1979 , 7, 396-430.
Erikson, Robert S. The SRC panel data and mass political attitudes. Paper delivered
to the 1975 annual meeting of the American Political Science Association, San
Francisco, September 1-5, 1975.
Goodman, Leo. Ecological regression and the behavior of individuals. American
journal of Sociology, 1953, 64, 610-25 .
Greene, Vernon L. Aggregate bias effects of random error in multivariate OLS
regression. Political Methodology, 1978, 5, 461-67.
Grunfeld, Y. , & Griliches, Z. Is aggregation necessarily bad? Review of Economi·cs
and Statistics, 1960, 42, 1-13.
Hannan, Michael T . Aggregation and disaggregation in sociology. Lexington, Mass.:
D.C. Heath, 1971.
Hanushek, Eric A., & Jackson , John E. Statistical methods for social scientists. New
York: Academic Press, 1977.
Hanushek, Eric A . . Jackson, John E., & Kain, John F. Model specification, use of
aggregate data, and the ecological correlation fallacy. Pol£tical Methodology, 1974,
1, 89-107.
Irwin, Galen, & Meeter, Duane A. Building voter transition models from aggregate
data . Midwest journal of Political Science, 1969, JJ, 545-66.
Jackson, John E. Statistical estimation of possible response bias in close-ended issue
questions. Political Methodology, 1979, 6, 393-423.
Johnston, J. Econometric methods (2nd ed.). New York: McGraw-Hill, 1972.
Joreskog, Karl G. , & Sorbom, Dag. Advances in factor analysis and structural equation
models. Cambridge. Mass. : Abt, 1979.
Kelejian, Harry H., & Oates, Wallace E. Introduction to econometrics (2nd ed.).
New York: Harper and Row, 1981.
Kinder, Donald R . Diversity and complexity in American public opinion. This
volume, 1983.
Kinder, Donald R ., & Kiewiet, D. Roderick. Economic discontent and political
behavior: The role of personal grievances and collective economic judgments in
congressional voting. Americanjoumal of Poli"tical Science, 1979, 23, 495-527.
Kramer, Gerald H . Short-term fluctuations in U.S. voting behavior, 1896-1964.
American Political Sdence Review, 1971 , 65, 131-43.
Kramer, Gerald H. The ecological fallacy revisited: Aggregate- versus individual-level
findings on economics and elections, and sociotropic voting. Amen·can Poli"tical
Sdence R eview, forthcoming.
 Christopher Ii. Achen I 93

Levi, Maurice D. Errors in the variables bias in the presence of correctly measured
variables. Econometn'ca, 1973, 41, 985·86 .
Lord, Frederic M., & Novick, Melvin R. Statistical theon·es of mental test scores.
Reading, Mass. : Addison-Wesley, 1968.
Maddala , G. S. Econometrics. New York: McGraw-Hill, 1977.
Pindyck, Robert S. , & Rubinfeld, Daniel L. Econometn'c models and economic
forecasts. New York: McGraw-Hill, 1976.
Prysby, Charles L. Community partisanship and individual voting behavior: Methodo-
logical problems of contextual analysis. Political Methodology, 1976, 3, 183-98.
Rao, Potluri, & Miller, Roger LeRoy. Applied econometn'cs. Belmont, Calif.:
Wadsworth, 1971.
Rasch, Georg. Probabilistic models for some intelligence and attainment tests.
Chicago: University of Chicago Press, 1980.
Robinson, William S. Ecological correlations and the behavior of individuals. Ameri-
can Sociological Review, 1950, 15, 351-57.
Ross, Keith, & Duvall, Raymond. The size of armed forces: Dispelling some myths
about cross-national data. journal of Conflict Resolution, forthcoming.
Scheuch, Erwin K. Cross-cultural use of sample surveys: Problems of comparability. In
Stein Rokkan, Ed. Comparative research across cultures and nations. Paris:
Mouton, 1968.
Sullivan, John L. , Piereson, James E., & Marcus, George E. Ideological constraint in
the mass public: A methodological critique and some new findings. Amen·can]our-
nal of Political Science, 1978, 22, 233-49.
Tate, C. Neal. Individual and contextual variables in British voting behavior: An
explanatory note. American Political Science Review, 1974, 68, 1656-62.
Theil, Henri. Linear aggregation in economic relations. Amsterdam: North Holland,
1954.
Theil, Henri. Principles of econometn'cs. New York: Wiley, 1971.
Torgerson, Warren S. Theory and methods of scaling. New York: Wiley, 1958.
Turner, Charles F. , & Martin, Elizabeth, Eds. Surveys of subjective phenomena: Sum-
mary report. Washington, D.C.: National Academy Press, 1981.
Warren, Richard D. , White, Joan Keller, & Fuller, Wayne A. An errors-in-variables
analysis of managerial role performance. journal of the Amen'can Statistical
Association 69, 1974, 348, 886-93.
Weatherford, Stephen. Measurement problems in contextual analysis: On statistical
assumptions and social processes. Political Methodology, 1982, 8, 61 -69.
Wiley, David E. , & Wiley, James A. The estimation of measurement error in panel
data. Amen'can Sociological Review, 1970, 35, 112-16.
Wiley, James A., & Wiley, Mary Glenn. A note on correlated errors in repeated
measurements. Sociological Methods and Research, 1974, 3, 172-88.
 4
Self-Portrait:
Profile of Political Scientists*
Naomi" B. Lynn

This survey offers a current profile of the political science profession in

the United States, drawing on a substantial number of studies that have been
made over the years. What emerges can be termed a self-portrait. It is,
however, no simple self-portrait of an individual against a neutral
background. Rather it is a group picture which shows a profession resting on
a somewhat fragmented foundation , grappling with current challenges--both
individually and collectively--in order to build a viable future. Thus, the por-
trait is complex, more in the tradition of Hieronymus Bosch than
Rembrandt.
At the 1978 meeting of the American Political Science Association
Everett Carll Ladd, Jr. and Seymour Martin Lipset characterized political
science as "a discipline in decline" (Ladd & Lipset, 1978, p. 21). This sum-
mary statement is an appropriate and proper backdrop for the self-portrait of
the discipline. Decline, however, for political science is a new development;
the profession has not suffered from any gradual, long-term loss of ideas or
energy. On the contrary, the profession experienced recent periods of ex-
plosive growth. But the excitement of the intellectual currents and debates
led us to ignore the impending demographic and economic forces that have
shaped our current problems. The past half dozen years have been difficult
ones for political science.

The Discipline's First Century Foundations:

Some Broad Currents
Political science, as an organized academic discipline in American
universities and colleges, was one of the products of the flowering of higher

*The author is grateful to Seymour Martin Lipset for his criticisms and sugges-
tions and to Walter B. Roettger, Margaret Conway, Eugene J. Alpert, David Finley,
Thomas Mann and Neale Pearsons for their review comments. She also thanks Charles
0 . Jones and Austin Ranney for some perspicacious observations.

95
 96 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

education in this country in the decades following the Civil War. In June of
1880 the School of Political Science was founded at Columbia University,
under the leadership of John W. Burgess (Somit & Tannenhaus, 1978, p . 17).
Interest in the subject grew at numerous other institutions; by December of
1903 the numbers of and the interactions between political scientists had
reached the point that it was useful to establish the American Political
Science Association. Thus our discipline is barely into its second century as an
established field, while as an organized professional group we are an even
younger phenomenon.

A Capsule Look at Some Major Intellectual Issues. The century following

1880 had many turbulent aspects as the field sought to define itself and to
select its objectives and methods of reaching them. Lipset (1969 , p. vii) has
pointed out that the term political science, as originally used, meant what
would now be called policy science; that is, it initially emphasized the struc-
ture and function of government and the concerns of authorities. Only later
did political science also stress systematic generalizations based on empirical
data and statistical analyses of the sort now familiar in all social sciences. This
fluctuation of emphasis between the study of the polity on the one hand and
the relationship of the polity to the society in which it operates on the other
has been a main feature of the discipline. Such fluctuation can be frustrating
for those who feel a need for an integrating model of the field. Greenstein and
Polsby (1975) reflect on this when they observe that "Early in his career, the
fledgling political scientist learns that his discipline is ill-defined, amorphous,
and heterogeneous" (p. v).
The discipline of political science has problems: political scientists per se
are highly differentiated from political philosophers on the one hand and
pure mathematicians who may handle political data on the other; the
discipline lacks a central paradigm and is unlikely to get one soon; political
science relies mainly on conceptual schemes from such disciplines as
economics, sociology, and psychology; there is a notable current lack of
methodological concerns; political cleavages of the late 1960's and 1970's
have left scars on the discipline; and finally, economic and demographic fac-
tors have adversely affected the environment within which political scientists
work. We may draw, however, some comfort from a brief review of some of
our field's past challenges.
One of the pioneer scholars, Charles E. Merriam of the University of
Chicago, sought in the early 1920's to move toward a "science of politics"
(Somit & Tannenhaus, 1967, p. 87). He wanted more scientific rigor, and his
efforts led to the establishment in the American Political Science Association
of a Committee on Political Research. (Dwight Waldo, 1975, p. 48 traces to
Merriam's "school" many leaders in the behavioral movement of later
decades.) In contrast, Thomas H . Reed of Harvard argued that the primary
responsibility of political science was to educate citizens and to prepare
students for careers in government. As Somit and Tannenhaus (1967) note
"Though neither Merriam nor Reed captured the Holy Land, or came very
close to it, both crusades left their mark on the discipline" (p. 88).
After the second World War, with the advent of computers,
 Naomi· B. Lynn I 97

"behavioralism" became the dominant intellectual force in political science.

Behavioralists argued for a more scientific approach to the study of politics.
They borrowed many of their models from the physical and natural sciences.
By the l 960's the behavioral movement made extensive use of the sort of
scientific approaches and systematic tools that Merriam had urged. It ap-
peared that if behavioralism delivered on its major promises, it might become
the "predominant paradigm" of political science (Somit & Tannenhaus,
1967, p. 210). This did not occur.
A survey of members of the American Political Science Association in-
dicates that while behavioralism made important contributions, it did not
establish hegemony; indeed, it appears that there is no consensus of commit-
ment to it, nor great certainty about the meaning of the term (Roettger,
1978, p. 10). In part this reflects the deep doubt of most political scientists
that the field has the capacity to become "scientific" in the sense of the
natural sciences (Roettger, 1978, p. 14). More seriously behavioralism has
been accused of developing "a number of blind and unquestioned ideological
biases . .. that followed almost directly from the dominant belief that facts
could and should be separated entirely from values, and that followed from
the criterion that rigor always comes before relevance in questions of good
political science" (Lowi, 1972, p. 14).
By the early 1970's the term behavioralism was often used in contra-
distinction to post-behavioralism (see Graham & Carey, 1972). The latter em-
bodied the value orientations that were associated with the counterculture
and the New Left (Waldo, 1975, pp. 113, 114). Post-behavioralism also in-
corporated other points, as David Easton observed in his 1969 presidential ad-
dress to the APSA: Substance must precede technique; behavioral research
must not lose touch with reality; knowledge must be implemented (Easton,
1969, p. 1052). Iri summary, post-behavioralism "supports and extends
behavioral methods and techniques by seeking to make their substantive im-
plications more cogent for the problems of our times" (Easton, 1969, p.
1061). Like the counterculture post-behavioralism left a noticeable mark, but
it did not produce enduring basic changes in political science's intellectual
and methodological underpinnings.
As political science ended its first century, it became increasingly clear
that the search for a single paradigm was futile . Rather, each sub-specialty
searches for its own unifying model. As it does so, it tries to incorporate ap-
propriate scientific methodology, but it also seeks to discover meanings that
"science" alone may miss (Baum, Griffiths, Mathews, & Scherruble, 1976,
pp. 915-917). Baum et al. points out the need for the latter with a quotation
from Charles Darwin's Autobiography:

But for many years I cannot endure to read a line of poetry. My mind seems to
have become a kind of machine for grinding laws out of a large collection of
facts, but why this should have caused the atrophy of that part of the brain on
which tastes depend, I cannot conceive. The loss of these tastes is a loss of hap-
piness, and may possibly be injurious to the intellect, and more probably to the
moral character, by enfeebling the emotional part of our nature (p. 917).

Other intellectual issues are also unresolved in the discipline. The

98 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

dispute between rationalism and anti-rationalism is a notable example.

Essentially rationalism contends that people are rational and seek to max-
imize stated goals in a calculating manner; this is the approach of economics.
According to this formulation the bureaucrat seeks to maximize the budget,
the voter tries to maximize expected utility, and the analyst attempts to for-
mulate theories of collective choice (Barry, 1970; Frolich & Oppenheimer,
1978; and Goodin, 1976). The anti-rationalist view is more social-
psychological in nature. It points out that irrationality is an inescapable
feature of social experience and that feelings and needs defy rational ex-
amination (see Glass, 1978, pp. 63-92 in Frolich and Oppenheimer).
A parallel dispute arises over whether political scientists should seek to
build deductive or inductive theories, or both. Similarly there is disagreement
between some who downplay the role of the study of classical political
philosophy, and others who consider that these roots are our rich foundation
that should not be ignored.
Two decades of intellectual ferment have led to productive and exciting
work. They have also exacted a price as David Finley (1982) points out:

The hiatus of the methodological debates, squabbles and cabals of the 1960's,
championing or reviling symbols of behavioralism and then post·
behavioralism-essential as the underlying issues have been and remain - did
forestall a lot of the substantive progress that the outside world (and govern-
ments and private foundations that want a product) had been encouraged to ex-
pect from the discipline. That generally undermined the popular respect for
political science among important external constituencies on whom we still de-
pend. The historical fact is a part of diagnosing the condition of the profession
today. ... We do suffer from the absence of an agreed paradigm, even if that
absence is inevitable. (Comment made as panel member, APSA convention,
Denver, 1982)

Indz'cators of Excellence. Political science is taught and learned at many hun-

dreds of American colleges and universities. Political science research and
writing have produced a voluminous literature. As in all academic fields,
some schools, journals and individuals have achieved special prominence.
The work that carries the mark of special excellence forms part of our com-
mon foundation.
Over the years there have been many ratings of the "top departments" in
virtually all areas. Such ratings are subject to inevitable criticisms, chief
among them that the underlying validity of the reputational perceptions on
which most are based are doubtful. All measures are necessarily symptomatic,
and they should not be taken too seriously. They seek signs which may be
reasonable surrogates for excellence, but which are not excellence per se.
Somit and Tannenhaus (1967), in their monumental study of our field,
recognize that, "Full many a flower is born to blush unseen." They go on,
however, to publish a comparative ranking shown here as part of an in-
teresting 40 -year comparison study in Table 1. These data show great con-
sistency, even though some schools have gained (e.g. Yale) and some lost (e.g.
Illinois) standing over the years. While a few position shifts have occurred, the
striking fact that emerges is the stability in these ratings over a 57-year period.
 Naomi B. Lynn I 99
This is especially significant when.one observes that no two studies used exact-
ly the same criteria and questions.
John S. Robey (1979) made a "productivity" study of departments that
considered, not reputations, but publications by departments' faculty. This is
shown in Table 2. Michigan is first in productivity-above its rating in the
other studies. Ranked second through fifth are Kentucky, Florida State,
Michigan State and Georgia, none of whom ranked in any of the reputational
studies. While one would not want to claim that counting journal articles in
the American Poli"tical Science Review and six regional journals is an ultimate
criterion, it does suggest that excellence is more widely distributed than some
suppose. It must, of course, be noted that many leading scholars publish most
of their articles in books because they are under pressure to do so by people
with project and/or conference money. (Robey did not count books, even
highly important ones; he also did not control his numbers for department
size.)
In a 1982 study McCormick and Bernick found results generally consis-
tent with the earlier studies cited above. They controlled, however, for two
things: the size of the authors' current departments and the number of recent
graduates produced at the Ph.D. granting institutions (see Table 3).
The identification of prestigious journals is shown in Table 4. Data come
from surveys from Somit and Tannenhaus with a later update by Roettger.
The main change shown is the emergence of the American journal of Poli"tical
Science to "second tier" status, as it moved from its former regional (Midwest)
identification (Roettger, 1978, p. 19).
The identification of the most outstanding contributors to the field of
political science can be done both by reputation and by a count of others'
citations to their work. Table 5 shows the results of Roettger's 1978 reputa-
tional study. Table 6 was compiled by John S. Robey; it is based on 1970-1979
citations. Also shown in this table are the schools where these scholars re-
ceived their Ph.D.s and where they teach. It is noteworthy that the "elite"
universities identified by the early reputational studies again show up pro-
minently.

THE PROFESSION'S JOB MARKET

A full picture of the job market would have to include the observation
that there are many political scientists among the tenured faculty at soundly
financed universities. But, just as it is unemployment more than employment
that reflects the economy's health, so our profession's economic situation is
heavily influenced by the problem-ridden segments of the job market.

The Stagnation of Demand. The demand for academic political scientists is

related significantly to the number of potential students who are able to pur-
sue higher education. From the middle l 940's to the early l 960's a "baby
boom" occurred. This produced a large number of college students from the
1960's to the early 1980's. There was general prosperity (with notable excep-
tion periods such as 1973-1975), and government-sponsored financial aid pro-
 100 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS
TABLE 1
Longitudinal Ranking of Graduate Departments of Political Science:
1925, 1957, 1963, 1964,* 1975-1976, 1982

1925 1957 1963

{Hughes) (Keniston) {Somit-Tannenhaus)

1 Harvard 1 Harvard 1 Harvard

2 Chicago 2 Chicago 2 Yale
3 Columbia 3 California 3 California (Berkeley)
4 Wisconsin (Berkeley) 4 Chicago
5 Illinois 4 Columbia 5 Princeton
6 Michigan 5 Princeton 6 Columbia
7 Princeton 6 Michigan 7 Michigan
8 Johns Hopkins 7 Yale 8.5 Stanford
9.5 Iowa 8 Wisconsin 8.5 Wisconsin
9.5 Pennsylvania 9 Minnesota 10.5 California
11 California 10 Cornell {Los Angeles)
(Berkeley) 11 Illinois 10.5 Cornell
12 California 12 Johns Hopkins
(Los Angeles) 13 Northwestern
13 Stanford 14 Indiana
14 Johns Hopkins 15 Illinois
15 Duke 16 Minnesota
17 North Carolina
18.5 Duke
18.5 Syracuse
20 Pennsylvania

grams multiplied'; thus the market demand was good. The birth rate peaked
in 1957, so by 1975 the number of potential freshmen started downward. In
1973-1974 there were 291 new positons for political scientists; by 1974-1975
there were only 239 (Mann, 1976, p. 412).
In the 1980's the smaller number of college-age people and the lessened
amount of financial aid create a market that can optimistically be called stag-
nant. A moderate recovery of the birth rate points toward some improvement
by the year 2000, especially if prosperity then prevails.
The demand for academic political scientists affects the whole field , in-
cluding the best paid tenured professors. In the l 960's the strong demand
helped attract large numbers of highly talented students into undergraduate
and graduate programs. Those with intellectual interest and high ability in
political science could afford to pursue their interest, confident that with a
Ph.D . they would be eagerly sought in the job market. We will never know ex-
actly how many of the potential top contributors to our field will never enter it
because of the present low level of demand, but surveys (e.g. Ladd & Lipset,
1978, pp. 9-12) suggest that it is a substantial number.

Doctoral Output. In most years of the mid-1970's, there were about 1000 firm
candidates for academic positions (Mann, 1976, p. 413). This reflected in
large part the dramatic increase in doctoral output that took place after 1960.
 Naomi B. Lynn I 101

TABLE 1 (continued)

1982
1964 1975-76 Uones, Lindzey and
(Cartter) (Roettger) Coggeshall)
1 Yale 1 Yale 1 Yale
2 Harvard 2 Harvard 2.5 California (Berkeley)
3 California 3 California 2.5 Harvard
(Berkeley) (Berkeley) 4.5 Chicago
4 Chicago 4 Chicago 4.5 Michigan
5 Columbia 5 Michigan 6 M.I.T.
6 Princeton 6 Stanford 7.5 Stanford
7.5 M.I.T.* 7 Princeton 7.5 Wisconsin
7.5 Wisconsin 8 Wisconsin 9 Princeton
9 Stanford 9 North Carolina 10.5 Minnesota
IO Michigan 10 Minnesota 10.5 Cornell
11 Cornell 11.5 California 12 Rochester
12 Northwestern (Los Angeles) 13.33 Columbia
13 California 11.5 Johns Hopkins 13.33 North Carolina
(Los Angeles) 13 Northwestern 13.33 Northwestern
14 Indiana 14 Columbia 16.5 Indiana
15 North Carolina 15 Cornell 16.5 California
16 Minnesota (Los Angeles)
17 Illinois 18 Duke
18 Johns Hopkins 19.33 Illinois
19 Duke 19.33 Ohio State
20 Syracuse 19.33 Washington
(St. Louis)

Source: 1925-1964 data taken directly from Albert Somit and Joseph
Tannenhaus (1967) , The Development of Political Science: From Burgess to
Behavioralism, Boston: Allyn and Bacon, p. 164. 1975-76 data are from Walter
B. Roettger (1978), "The Discipline: What's Wrong, and Who Cares?" Paper
presented at the Annual Meeting of the American Political Science Association,
New York City. 1982 data are from Lyle V. Jones, Gardner Lindzey and Porter
E. Coggeshall (Eds.), An Assessment of Research-Doctorate Programs in the
United States: Social and Behavioral Sciences, Washington, D.C.: National
Academy Press, 1982.
*M.I.T. was not included in the 1925, 1957 and 1963 studies.

Table 7 shows the number of American doctorates produced in political

science. The 1960-1970 production was absorbed fairly easily by the "baby-
boom" induced demand, but the burgeoning output of the 1970's swamped
the market. The post-behavioralists of the 1970's urged relevance and future
orientation in their writings; perhaps too little attention was paid to these
points in our graduate advising. By the last half of the 1970's the production
of Ph.D.s started to slow down. Especially important was the dramatic reduc-
tion in the number of new Ph.D. students (see Table 8). Even with these
reductions, there were still more new Ph.D.s than the academic market could
absorb.
 102 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS
TABLE 2
Political Science Departments:
Top 20 Departments Ranked by Productivity

L Michigan 1L Minnesota
2. Kentucky 12. Texas
3. Florida State 13. Arizona
4. Michigan State 14. Harvard
5. Georgia 15. California (Berkeley)
6. Iowa 16. Rochester
7. Wisconsin 17. Houston
8. Massachusetts 18. North Carolina
9. Ohio State 19. California (Los Angeles)
10. Indiana 20. Yale

Source: John S. Robey. Political Science Departments: Reputations Versus

Productivity, PS, 1979, 12, p. 205.

Law schools apparently attracted many of the stronger students of the

late 1970's who might have sought Ph.D.s in political science a few years
earlier (Ladd & Lipset, 1978, p. 9). The growth in Masters in Public Ad-
ministration programs also drained off some of the bright career-oriented
students. About a third of political science departments have reported a
decline in the quality of new Ph.D.s (Sheilah Mann, 1982, p. 91). Among
sub-fields there have been moderate gains in the number of students in public
administration and public policy as areas of concentration, and a decline in
the students choosing comparative politics and political theory (Sheilah
Mann, 1982, p. 91).

The Pay Record and Outlook. Academic folklore has long accepted as an en-
during truth that there was low faculty pay in the field of political science.
Thus it was surprising when Ladd and Lipset, using 1969 survey data, found
that the pay of political scientists was relatively high; only law and medicine
did better in the elz"te schools (Ladd & Lipset, 1974, pp. 21, 22). Salaries in
political science relative to other fields dropped from 1969 to 1977, although
similar drops occurred in other social sciences (Ladd & Lipset, 1978, p. 6).
The most recent expectation has been that incomes, adjusted for infla-
tion, will drop in the future (Walker, 1978, p. 484). Given the imbalance be-
tween the number of openings and the number of people seeking academic
positions, it could hardly be otherwise. The effects of low starting pay extend
to all faculty ranks, as hard-pressed institutions are forced by state budget
crises to economize. The outlook, even for those who have tenured jobs, is
economically bleak. And since the median age of political science faculty was
37 in 1971 (Baker, 1971, p. 34), and it only rose to 42by1980 (Lane, 1982, p.
52), many of us could face decades of economic hardship.
In addition to the economic toll there are side effects which drastically
alter professional and interpersonal relations in faculty departments.
 TABLE 3
Comparative Rankings of Political Science Departments by Reputation
and Alternate Standardized Measures of Productivity

Reputational Graduate-Training Graduate Training Affiliation Affiliation Rankings

Rankings3 Rankings Jb Rankings uc Rankingsd (1974-1978 data)e

1. Yale 1. Iowa 1. Rochester 1. Florida Atlantic 1. Carnegie-Mellon

2. Harvard 2. North Carolina 2. Washington- 2. Carnegie-Mellon 2. Michigan State
3. Berkeley 3. Vanderbilt St. Louis 3. Kentucky 3. Kentucky
4. Chicago 4. Michigan State 3. Kentucky 4. Emory 4. Iowa
5. Michigan 5. Syracuse 4. Stanford 5. Rochester 5. Virginia
6.5. MIT 6. Yale 5. Vanderbilt 6. Florida 6. Rochester
6.5. Stanford 7. Rochester 6. Brown 7. Iowa 7. Ohio State
8. Wisconsin 8. Kentucky 7. Michigan State 8. California-Riverside 8. Houston
9. Princeton 9. Minnesota · 8. Boston College 9. ~Iichigan State 9. use
10. North Carolina 10. Duke 9. Yale 10. Georgia 10. Wisconsin-Milwaukee
11. Columbia 11. Stanford 10. North Carolina 11. Cal. Tech. 11. Florida
12.5. UCLA 12. Illinois 11. Duke 12. Ohio State 12. California-Riverside
12.5. Minnesota 13. Princeton 12. Tulane 13. Stanford 13. Wisconsin
14.25. Cornell 14. Wisconsin 13. Georgetown 14. Minnesota 14. Michigan
14.25. Indiana 15. Tulane 14. Minnesota 15. California-Irvine 15. Minnesota
14.25. Northwestern 16. Berkeley 15. Michigan 16.5. Arizona 16. Texas Tech
14.25. Rochester 17. Michigan 16. New School 16.5. Cincinnati 17. Duke ~
Q
18.5. Iowa
18.5. Oregon
18.
19.
Chicago
Harvard
17. Case Western
18. Houston
18. Vanderbilt
19. Michigan
18.
19.
Rice
Tulane ...
~
20.33. Illinois 20. Florida 19. Pennsylvania 20. Massachusetts 20. Arizona ~
20.33. Johns Hopkins 20. Ohio State
20.33. Washington- ~~
St. Louis
-.....
0
(,IQ
TABLE 3 (continued) .......
0
~

Vl
M
Source: Joseph M. McCormick and E. Lee Bernick. Graduate Training and Productivity: A Look at Who Publishes, Journal of t;;
Politics, 1982, 24, 212·227. ..;,
3
0
The reputational rankings are drawn from David R. Morgan and Michael R. Fitzgerald, Recognition and Production Among
American Political Science Departments. Western Political Quarterly, September 1977, 30: 348. The numbering for tied ranks
~
has been changed slightly to conform with the convention and in other tied rankings. ~
g
bTo obtain these graduate training ranks, the weighted department scores were standardized by the number of recent "ti
graduates. The figure used for each department was the average number reported in the Guide to Graduate Study in Political ~
0
Science for the 1977 through the 1979 editions. Since the figure reported in each edition of the Guide is in itself averaged over ::l
the past three years, the figure ultimately employed in the analysis tends to cover the years of our study. [After controlling for t""
M
number of graduates in Ph.D. programs, the authors weighted the journal articles on the basis of the estimated quality of the 0..,.,
journal: APSR as 1.0, JP as .957, AJPS as .943, Polity as .843 and the WPQ as .829.]
"ti
cTo obtain these graduate rankings, the weighted department scores were divided by the number of political scientists in the 0
t""
profession who received their graduate training from that institution (as determined by our systematic sample of the discipline). ::i
dThese standardized rankings come from Robey, "Political Science Department." (=j
>
eTo obtain these rankings, the weighted present affiliation scores were divided by the number of faculty members in a t""
Vl
department. The figure used was the average of the number reported for 1976 and 1977 in the Guide to Graduate Study in Ci
Political Science and the 1978 figures from Robey, "Political Science Departments."
~
::J
Vl
;;J
 Naomi B. Lynn j 105

TABLE 4
Journal Ranking 1963 and 1976

1976 1963
Journal Rank Index Rank Index

American Political Science Review 1 2.75 1 2.78

journal of Politics 2 2.42 3 2.31
World Politics 3 2.40 2 2.32
American journal of Political Science 4 2.25 9 1.89
Public Administration Review 5 1.96 7 1.99
Political Science Quarterly 6 1.94 4 2.07
Public Opinion Quarterly 7 1.90 8 1.93
Administrative Science Quarterly 8 1.82 5 2.01
American Behavioral Scientist 9 1.84 10 1.73
Western Political Quarterly 10 1.81 6 2.00

Source: Table compiled by Walter B. Roettger. 1976 data were from a random
sample drawn from APSA's Directory of Members; 1963 data from Somit and
Tannenhaus, American Political Science: A Profile of a Discipline. See Walter B.
Roettger, "The Discipline: What's Right, What's Wrong, and Who Cares?" Paper
presented at the 1978 Annual Meeting of the American Political Science
Association.

Caplow's The Academic Marketplace (1958) notes schisms prevailing in

higher education in the 1950s: young turks vs. elderstatesmen; teachers vs.
researchers, generalists vs. specialists; conservatives vs. liberals; pro-
administration vs. anti-administration; humanists vs. scientists; and inbred
vs. outbred. To these we would have to add-perhaps at the top of the
list-tenured vs. non-tenured. It is not unfair to say that there are three dis-
tint classes of political scientists; tenured, those with hope of getting tenured,
and those with little or no hope of getting tenure. This situation introduces a
new stress on hierarchy within departments. Organizational hierarchy in-
volves at least three modes of unequal allocation of resources: inequality in
security, inequality in perceived punishment for deviance, and inequality of
authority. In all cases non-tenured persons are at a disadvantage and those
"without hope" are uniquely so. The result is an atmosphere scarcely con-
ducive to stimulating the free exchange of ideas which traditionally has
characterized the academic arena.
Academic and Non-Academic Prospects. Academics have always been the
majority among political scientists. A 1970 National Science Foundation
study showed the following occupational breakdown of political science
Ph.D.s (quoted in Waldo, 1975, p. 120):

Academics 76.9
Business 1.8
Federal(}overnment 5.4
Military 1.4
State and Local (}overnment 3.5
 TABLE 5 ......
0
Ranking of Significant Contributors: A Longitudinal Perspectivea en

Pre-1945b 1945-1960 1960-1970 1970-1976

"'l:'1t'"'
71
"C
Rank Name Rank Name RankName Rank Name 0

1 Merriam 1 Key (353) 1 Dahl (403 ) 1 Lowi (183 )

~
2 Lasswell 2 Lasswell (323) 2 Easton (193 ) 2 Wildavsky (103 ) ~
3 White 3 Dahl (203 ) 3 SRC Groupe (183) 3 Dye (93 ) ~
"C
4 Beard 4 Easton (183) 4 Deutsch (173 ) 4 Dahl (83 ) ::i::i
5 Corwin 5 Morgenthau (18%) 5 Almond (163 ) 5 Huntington (73 ) 0...,
6 Bentley 6 Truman (163 ) 6 Wildavsky (7 3 ) 7 SRC Groupe (63 ) t=:
t'1
7 Wilson 7 Strauss (83) 7 Lowi (43 ) 7 Verba (63 )
0...,
8 Herring 8.5 Deutsch (63) 9 Lipset (43 ) 7 Sharkansky (63 )
9 Wright 8.5 Simon (63) "C
9 Wolin (43) 10.5 Barber, Deutsch, 0
10 Ogg 10.5 Friedrich (53) 9 Huntington (43) Left Radicalsd, Riker t'"'
10.5 Schattschneider (53) ::i
(=)
>
t'"'
Source: Compiled by Walter B. Roettger, Strata and Stability: Reputations of American Political Scientists. PS, 1978, 11: 9.
aFigures in parentheses represent the percentages of respondents designating the contributor. Sample size for 1945-1960 was "'C"l;;
181; for 1960-1970, 179; and for 1970-1976, 113. The variation between periods (and the departure from the overall response z
....,
level) is due to the failure of all respondents to designate significant contributors in each period. u;
bTaken from Somit and Tannenhaus, American Political Science: A Profile of a Discipline, New York: Atherton Press, 1964, ;;l
p. 66.
eThe "SRC Group" consists of Angus Campbell, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. Mention of one
or more of these persons was coded as "SRC Group."
dThe "Left Radicals" include: Ira Katznelson, Herbert Marcuse, Ralph Miliband, C. Wright Mills, James O'Connor, and
Bertell Oilman. Mention of one or more of these persons was coded as "Left Radicals."
 Naomi B. Lynn I 107

TABLE 6
Rank Order of 20 "Most Significant Political Scientist Contributors"
by Number of Citations 1970-79,
Where They Received Degree and Where They Teach

Number
of
Name Citations Ph.D. Teaching

1. Seymour Martin Lipset 3425 Columbia Stanford

2. Herbert Simon 3425 Chicago Carnegie-Mellon
3. Robert Dahl 2235 Yale Yale
4. Angus Campbell 2184 Stanford Michigan
5. Karl Deutsch 1870 Harvard Harvard
6. Gabriel Almond 1799 Chicago Stanford
7. David Easton 1644 Harvard Chicago
8. Samuel Huntington 1511 Harvard Harvard
9. Harold Lasswell 1410 Chicago Yale
10. Philip Converse 1282 Michigan Michigan
11. V. 0. Key 1110 Chicago Harvard
12. Theodore Lowi 913 Yale Cornell
13. Charles Lindblom 958 Chicago Harvard
14. Robert Lane 782 Harvard Yale
15. Aaron Wildavsky 766 Yale California, Berkeley
16. W. H. Riker 759 Harvard Rochester
17. Thomas R. Dye 709 Pennsylvania Florida State
18. Carl J. Friedrich 701 Heidelberg Harvard
19. Sidney Verba 645 Princeton Harvard
20. Ira Sharkansky 589 Wisconsin Wisconsin

Source: Compiled from John S. Robey, Reputation vs. Citations: Who Are the
Top Scholars in Political Science. PS, 1982, 15: 200. Biographical data from
Who's Who in America. Herbert Marcuse and C. Wright Mills were dropped from
Robey's list because they were not political scientists.

TABLE 7
Production of Ph.D.s in Political Science

Number of Doctorates
Awarded in
Year Political Science

1880-1960 3700
1960-1970 3836
1970-1980 8519

Source: 1880-1970 compiled from Walter B. Roettger. "I Never Promised You a
Rose Garden: Career Satisfaction in an Age of Uncertainty." Paper presented at
the Iowa Conference of Political Science, 1977. 1970-1980 data compiled from
Sheilah Mann. Placement of Political Scientists, 1980-1981. PS, 1982, 15: 85.
 108 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

TABLE 8
Supply of Political Scientists, ,1969-1981

New Students Beginning Graduate Student Enrollments

Ph.D. Study in in Ph.D. Programs in Ph.D.s
Political Science Political Science Awarded
Fall, 1981 1,042 1981-82 5,491 679
Fall, 1980 1,068 1980-81 5,756 729
Fall, 1979 1,1 00 1979-80 5,888 766
Fall, 1978 1,051 1978-79 5,742 851
Fall, 1977 1,182 1977-78 5,737 881
Fall, 1976 1,064 1976-77 5,462 885
Fall, 1975 1,174 197 5-76 6,150 862
Fall, 1974 1,443 1974-75 6,150 907
Fall, 1973 1,414 1973-74 6,450 906
Fall, 1972 1,576 1972-73 * 811
Fall, 1971 1,695 1971-72 * 821
Fall, 1970 2,138 1970-71 * 634
Fall, 1969 2,487 1969-70 * 559

Source: Data obtained from "Political Science Degrees Awarded and Graduate
Students Enrolled: 1982 Update." PS, 1982, 15: 459-460.
*Not available.

The remainder were presumably retired, unemployed, or not in the labor

force. This suggests that, given a good job market, such as that of 1970, the
great majority of political scientists prefer an academic position.
Even in recent job markets most of those who get positions find them in
universities and colleges. In 1981 only 18 percent of Ph.D. placements were
non-academic (less than in 1979 and 1980). The proportion of 1981
placements to temporary positions was 28 percent, down from all earlier
surveys. The placement success percentage- -81 percent--was the highest since
1974. The reason for this may well lie in the relatively low number of can-
didates, 697 (Sheilah Mann, 1982, p . 86).
Success in placement varies by sub-fields. Public administration, public
policy and American government Ph.D.s fare better than those in com-
parative politics, international relations, and political philosophy (Sheilah
Mann, 1982, p. 89). A shift in placement opportunity has resulted in the need
for the Ph.D. to be completed; ABDs found many placements in the early
1970s, but later in the decade the completion of the degree was required for
the majority of new positions (Mann, 1978, p. 27).
Some have suggested that the skills of political scientists fit them well for
many non-academic positions. About one fifth of Ph.D . programs -
particularly those of smaller departments - have sought to adapt themselves
and prepare students for non-academic positions (Sheilah Mann, 1982, p.
85). Also, about 150 to 175 Ph.D.s (2 percent of the total) leave academia an-
nually for outside jobs. The 1981 placement data suggest that non-academic
 Naomi" B. Lynn J 109

placements are not becoming the solution to our profession's market problem.
A few want such positions; others, such as those who do not gain tenure, must
seek them. In some fields , such as government and lobbying, the political
science degree is, of course, a distinct asset. For most people starting post·
baccalaureate study, however, the alternative to being a political science
Ph.D. teaching in a university is not being a political science Ph.D. at all. It is
setting out for another educational and career objective in the beginning.

VIEWPOINTS
The views of political scientists have been noticeably different on the
average from those of other academics. Certainly we are not a representative
sample of the nation's population. We are drawn disproportionately from the
middle class and almost half of us come from families where the father holds a
professional or managerial position. Even when we compare ourselves to
others in the academic community we are more advantaged. Only the
students and faculty in the medical profession can claim a higher socio-
economic background than political scientists (Ladd & Lipset, 1974, pp. 5-6).
In terms of ethnocultural background it is worth noting that even though
Catholics outnumber Jews by roughly nine to one in the general population,
twenty-two percent of political science faculty is Jewish compared to 10 per-
cent Catholic. Jews are also disproportionately represented at elite schools
(Ladd & Lipset, 1974, pp. 7-9). In a 1982 survey of 637 political science
departments it was found that women made up 11.17 percent of full-time
tenure-track faculty. Blacks were 2.89 percent and 1.07 were Spanish sur·
named Americans (American Political Science Association, 1982, p . 1). The
political science professoriate is indisputably male, white and non-hispanic.

Politz"cal Views. Political views are taken with special seriousness by our pro-
fession. The major currents of these views have been consistent over a substan-
tial period. A 1977 faculty study conducted by Ladd and Lipset indicates that
58 percent of social scientists identify themselves on the liberal side of the
political spectrum (Lipset, 1981, p . 3). Among social scientists, economists
are the most conservative; sociologists and anthropologists are the most
liberal. Political scientists take a central position in this group , but are still
liberal when compared to the average American (Lipset, 1981 , p. 5). The
Christian Science Monitor surveyed political scientists attending the 1981
American Political Science Association Meeting and 73 percent of those
responding characterized themselves as "moderately liberal" or "very liberal"
(Christz"an Sdence Moni"tor Survey, 1981). 1 This liberal stance is partially ex-
plained in terms of self-selection. Students with more liberal orientations are
attracted to fields with a heavy focus on social problems and to fields where
they feel ideologically comfortable (Ladd & Lipset, 1975, pp. 152-157).
Ideology and partisan preference are determined more by adult socialization
than by family background. Political scientists credit knowledge gained in the
profession and the influence of colleagues as major determinants of their par-
tisan preferences and ideologies (Turner & Hetrick, 1972, p . 365 ; Ladd &
Lipset, 1975, pp. 157-159).
 110 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

Political scientists are strongly Democratic, although there has been an

increase in the number of Independents. This is evident in a comparison of
the 1970 study of members of the American Political Science Association and
those surveyed by the Christian Science Monitor in 1981 (See Table 9). It is
obvious that the Democratic party's loss is not necessarily the Republican
party's gain. Political scientists may become disillusioned and frustrated by
the failure of the system to respond adequately to the problems confronting
it- 32 percent agreed that the American political system was not sound and
needed many improvements or fundamental overhauling- but apparently
they are not willing to join the Republican party to seek a solution. In the
1981 APSA convention survey respondents reportedly saw a conservative
trend in the country, but 67 percent saw it as a brief phenomenon (Christian
Sci"ence Monitor Survey, 1981).
Seymour Martin Lipset contends that the leadership of American
political science is neoconservative. These neoconservatives support govern-
ment action to curb social injustice and the welfare state, but they have
serious misgivings about what they consider to be misguided efforts to achieve
equal opportunity, and they are concerned about policies which fail to meet
what they view as communist expansionism (Lipset, 1981, p. 5). It is likely
that some variance does exist between these leadership views and those of the
membership in general.
Franklin Delano Roosevelt was by far our favorite president. When asked
to rank the best overall President from Roosevelt to Carter, 69.3 percent
chose Roosevelt. The race for the worst president was won easily by Nixon
who received 59.5 percent of the vote. His nearest competitor was Carter with
22 percent.
John F. Kennedy, who is one of the most popular presidents with the
general public (Gallup, 1980, p. 27), does not have high standing among

TABLE 9
Party Preference of Political Scientists

Party Preferences 1959 1970 1976 1981

Republican 16.4 11.8 12 10.04
Democrat 73.7 73.4 74 61.19
Independent 8.0 12.5 12 22.39
Other 1.9 2.3 2 6.37

Source: 1959 data from Henry A. Turner, Charles G. McClintock and Charles B.
Spaulding. The Political Party Affiliation of American Political Scientists, The
Western Political Quarterly, 1963, 16: 652. 1970 data from Henry A. Turner
and Carl C. Hetrick. Political Activities and Party Affiliations of American
Political Scientists. The Western Political Quarterly, 1972, 25: 363. 1976 data
furnished by Walter Roettger, Drake University, from unpublished Ph.D.
dissertation, University of Colorado, 1977. 1982 data computed from Christian
Science Monitor/American Political Science Association Survey, data obtained
from the Roper Center, The University of Connecticut.
 Naomi" B. Lynn I 111
political scientists. Only 7 .6 rank him as the best overall, and he is tied with
Eisenhower in that rating. This is consistent with the earlier Ladd-Lipset
finding that although Kennedy made a conscious effort to gain their support,
he was never popular with intellectuals. Adlai Stevenson and Hubert Hum-
phrey were much preferred for the 1960 Democratic nomination. Intellec-
tuals disdained Kennedy's mediocre congressional record , and his failure to
take a strong stand against McCarthy (Ladd & Lipset, 1975, pp. 22 , 23). Dur-
ing Kennedy's term of office, James Reston of the New York Ti"mes discussed
Kennedy's lack of support among intellectuals and said that they were
describing his presidency as "the third Eisenhower administration" (quoted in
Ladd & Lipset, 1975, p. 24). Some twenty years later Kennedy is still equated
with Eisenhower among political scientists; for example, as best on foreign
policy, 12 percent chose Eisenhower and 11.3 Kennedy. (Eisenhower and
Kennedy rank on foreign policy behind Roosevelt, Truman and Nixon, in
that order.)
Jimmy Carter's ranking among political scientists is low. Sixty-three per-
cent rate him as the president who was least able to get things done. This is a
consistent evaluation regardless of ideology, party preference, academic rank,
gender or age. His predecessor, Gerald Ford, comes in second, but far behind
with 20.6 percent.
When asked to compare Reagan with the past eight presidents 66 per-
cent rate him as either not as good as most or worse than most. Political scien-
tists do give him a grade of A, however, for his relations with Congress. From
then on his grade point average begins to slip. He gets a majority of Ds and Fs
on economic policy (58.4 percent), foreign policy (63 .1 percent), social issues
(73.8 percent), and over-all performance (51.1 percent). On all the scores
mentioned above, women political scientists consistently rate Reagan lower
than their male counterparts. Reagan's lack of relative popularity with
women was manifested in the presidential election, which showed a signifi-
cant sex difference in preference for Reagan.
As anticipated, political scientists exercise considerable "constraint" in
their political ideologies. That is, their political positions and practices
demonstrate a logical or systemic order (Ladd & Lipset, 1975, pp. 37-40).
This constraint is shown by the fairly consistent positions taken by self-
identified liberals and conservatives on major social issues in the Christian
Sct'ence Monz.t or Survey. Self-identified liberalism/conservatism was the
single most accurate predictor of policy position. It also confirms that
political scientists are comfortably clustered at the liberal end of the political
spectrum (see Table 10).
Educati'onal Vt"ews. Clark Kerr has observed that "Few institutions are so con-
servative as the universities about their own affairs while their members are so
liberal about the affairs of others" (quoted in Ladd & Lipset, 1975, p. 33).
This has not been true of political scientists. The same consistency of ideas
discussed above applied when ·political scientists were questioned on issues of
student activism such as a broadened student role and the demands of blacks
in the 1960's. Liberals were significantly more supportive of student positions
(Ladd & Lipset, 1971, p. 138).
A major and controversial issue directly affecting higher education has
112 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS
TABLE 10
Views on Selected Policy Issues of Political Scientists Attending the
1981 Meeting of the American Political Science Association

Poli tic al Sci en tis ts

(n = 526) Sample U.S. Population
Agree/ Disagree/ Agree/ Disagree/
Issue Approve Disapprove Approve Disapprove

Producing and readying

the neutron bomb 34.5 65.4 48 44a
Reagan's handling of
economic conditions 28.0 72.0 52 26b
Reagan's New
Federalism 36.0 63.9 67 18a
Amendment to the
Constitution that
would permit prayers
to be said in the
public schools 7.9 92.03 76 15C
Equal Rights Amendment 83.04 16.9 63 32d

Source: Political Science data computed from data obtained from the Roper Center,
University of Connecticut, Christian Science Monitor/ American Political Science
Association Survey. General population sample from:
acallup R eport No. 193, October, 1981.
bcallup R eport No. 191, August, 1981.
c Gallup R eport No. 177, April-May, 1980.
dcallup Report No. 190,July, 1981.

been affirmative action. Among political scientists there has been a loss of
support for affirmative action from 1969 to 1975. When asked about relaxing
standards for the admission of minority students, 69 percent supported such
measures in 1969 while only 43 percent did so in 1975; when asked similarly
about minority faculty, support dropped from 40 percent in 1969 to 19 per-
cent in 1975 (Ladd & Lipset, 1978, graph 3). The shift may be the result of a
re-confirmation of the university as a meritocracy, but the shift in attitude
away from minority faculty may also reflect a protective response to the tight
academic market.
Political scientists are not , on the whole, well-satisfied with the product
of their educational efforts. Data from 1969 show that 45 percent of the
political science faculty at top schools disagreed with the proposition that
graduate education in their field did a good job. The only field with a higher
percentage of dissatisfaction was sociology (47 percent); by way of contrast,
for business administration and chemistry the figures were 18 and 13 percent
respectively (Ladd & Lipset, 1974, p. 35).
 Naomi' B. Lynn I 113

In a later study the attitudes remained about the same. Only half of the
political science faculty said that graduate education did a good job, com-
pared to three-fourths of all academics surveyed (Ladd & Lipset, 1978, p.
14).
This dissatisfaction may have several roots. One is that students must
take courses from widely disparate fields, such as political philosophy and
statistical applications. The sense of a unified whole may be apparent to only
a few. The varieties of intellectual turnings may have caused some graduate
students who need structure to lose confidence. Dissatisfaction is also nur-
tured now by the bleak job market outlook for many political science
graduates. Perhaps we should not expect nor desire a high level of satisfaction
with graduate education in political science at this time. Indeed, Charles 0.
Jones contends that, "We should probably be suspicious of a high level of
satisfaction, since it could indicate that we have not kept up with the turns,
twists and reversals which have characterized our discipline as it continues to
undergo painful development" Qones, 1982).

Vz'ews on Professz'onal Satz'sfactz'on. This section can be summarized with

Walter Roettger's (1978) conclusion that political science is not a very happy
profession (p. 48). In all fields a minority of academics have always expressed
regret concerning their choice of careers. Ladd and Lipset in 1973 found that
10 percent of political science faculty would not want to be professors, if they
could start over again. This was about the same figure that applied to
business,. engineering, and the medical professionals (quoted in Roettger,
1978, p . 31). More relevant is the finding of statistically significant deteriora-
tion in career satisfaction from 1963 to 1976 among academic political scien-
tists. Asked in a 1976 survey, "If you were able to start over and pick your pro-
fession again, would you still pick a career in political science?," academics
gave these answers among others (Roettger, 1977, p . 12):

Definitely yes Definitely No

1963 41% 1%
1976 29% 73

The main "pockets of optimism" were found in two seemingly disparate

groups: public administration and political philosophy. The one is a growing
field, and the other, perhaps, is a field that attracts those who care most for
the joys of pure scholarship (Roettger, 1977, p . 33).

THE ASSOCIATION IN AN ERA OF STRESS

The membership of the American Political Science Association enjoyed
the sort of post-World War II growth that characterized many fields of
114 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

endeavor during that boom period. Its numbers went from 3300 in 1945
(Waldo, 1975, p. 54) to 15,758 in 1974 (Mann, 1982a). By 1982 the number
had dropped to 11,597. This reflects economic conditions, but it also reflects
a breakdown of shared consensus; Roettger (1978) reported many "can't say"
answers in a 1978 survey that covered many educational, intellectual and pro-
fessional issues that might inspire controversy, but hardly lack of opinion. A
former executive officer of the American Sociological Association has stated
that calling the ASA an "association" may be an overstatement, since not
everyone is associating with everyone else (Demerath III, 1981, p. 87). The
same description may be appropriate for the APSA.
The Association's stresses have been compounded by the controversies
arising from behavioralism and post-behavioralism, by the emergence of
policy oriented associations, and the growth of cognate organizations such as
the American Society for Public Administration and the International Studies
Association (the latter accompanied by a growing conviction that public ad-
ministration and international relations ought to break off from political
science), by the Viet Nam War, by the rise of the Caucus for a New Political
Science, and by the emergence of women and minorities as groups with claims
on the Association.
Preliminary results from a 1982 APSA survey of political scientists show
some of the expressed reasons for discontent with the Association:

Those who let their membership lapse did so because they thought the dues were
too high, they didn't like the APSR, and they felt the APSA didn't serve their
interests. Lapsed members most frequently agree with the statements that "The
dues are unreasonably high" and that there is " too much emphasis on quan-
titative research." Moreover, lapsed members appear to be slightly less profes-
sionally active (in terms of memberships, journal subscriptions and publishing)
and earn somewhat lower salaries than current members. (Mann, 1982c)

The Problem of Membership Decline. APSA membership peaked in 1974; by

1982 it was down 25 percent from that peak (Mann, 1982a). Annual meeting
registration peaked before membership in 1969 at a level of about 4200 at the
New York City meeting. Low points in registration of under 2500 were hit in
1973, 1976, and 1978. Both 1979 and 1980, however, showed growth
(American Political Science Association, 1981, p. 610). The 1981 meeting
totaled 2,887 which was the highest registration since the 1972 meeting
(French, 1981, p. 786). Perhaps the meeting locations (Washington, D.C.
and New York City) helped.
One reason for the membership and attendance decline is the emergence
of specialized groups. This is related to the lack of a single paradigm discussed
earlier. As political scientists have become more specialized, some members
have concluded that their interests are better served by other organizations. A
comparative government area specialist, for example, may find that he/she
has more in common with economists, sociologists and anthropologists work-
ing in the area than with other political scientists. This may also decrease the
value of the American Political Science Review, the most prestigious journal
in the field. The journal may not be providing an adequate vehicle for ex-
changing ideas and concepts among those with similar interests. Scholars may
 Naomi B . Lynn I 115

find it more rewarding to publish in journals read by their "significant others"

where they are more likely to be cited and where they may establish their na-
tional reputation more quickly. Specialization has devalued the two principal
reasons for joining the APSA- the annual meeting and the journal.
Another problem is that of communication among members of the
discipline, not only between areas of specialization, but also between scholars
who have training such that they can comprehend research using quantitative
methods or arguments presented in terms of formal logic and those who
cannot.
The APSA has been generous in its approach to unaffiliated groups
which meet at its annual meeting. The number of unaffiliated groups that got
courtesy listings in APSA programs rose from six (with two and a half pages)
in 1972 to 31 groups with 30 pages in 1982. Some have contended in the
APSA Administrative Committee that unaffiliated groups were prospering
at the expense of the APSA itself (Minutes, 1982). It seems unlikely, however,
that non-cooperation with sub-groups would be an effective long-run
strategy. In 1982 the Association developed guidelines for the establishment
of sections. Sections will make it possible for groups of APSA members who
share a common interest to meet , coordinate communications and receive
help from the APSA national office in collecting dues and maintaining
membership lists (American Political Science Association, Fall 1982b, p .
627). It is hoped that the establishment of sections will lead to restored
integration of the discipline.

The Caucus for a N ew Political Sdence. The emergence and activities of the
Caucus for a New Political Science have been among the most noteworthy
features of the Association's history over the past two decades. The New
Caucus was an organizational manifestation of the intellectual forces that led
to the post-behavioral movement, with which it has significant ties. The
Caucus for a New Political Science was convinced of the inseparability of
politics and intellectual work. Beyond this the New Caucus was what one of its
founders calls "an organized insurgency" (Lowi, 1972, p . 12) that believed
that the APSA should have a strong stand against the Viet Nam War and
racial discrimination and that it should actively support the "war on poverty"
programs of the Johnson Administration (Lowi, 1972, p . 13). There were im-
mediate negative reactions among those with the strong conviction that pro-
fessional academic associations should not take policy positions on issues not
directly related to the main function of the organization: promoting the in-
tellectual and professional interests of the discipline and its members. The
position of others can be summarized in a statement of a former president of
The American Sociological Association reacting to the issue of taking policy
positions within his organization, "The most that can be accomplished is to
announce a policy position on an issue and thereby provide a catharsis for
members who need it" (Hawley, 1981 , p . 108).
In 1966 it was discovered that some APSA officials had worked with the
CIA; this had a catalytic effect on the New Caucus' emergence. Lowi describ-
ed the New Caucus' leaders efforts in the 1967 APSA Business Meeting as
 116 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

precipitating "probably the ugliest confrontation in the history of the pro-

fession" (Lowi, 1972, p. 13).
The groundwork was thus laid for a major schism in the Association. In
the l 970's the New Caucus continued to attempt to take control of the APSA.
Probably the high water mark of these efforts occurred in 1973. The New
Caucus had a full slate of officers to compete with the slate offered by the
APSA Nominating Committee. The final vote (which was taken by mail
ballot) gave the New Caucus candidate for President, Peter Bachrach, 3191
votes; Austin Ranney, the choice of the Nominating Committee got 3803
votes (American Political Science Association, 1974, p. 36).
The New Caucus continues to function in the 1980's, although it does not
issue the sort of challenges that it did in the l 970's. The appeal of the New
Caucus has diminished with the end of the Viet Nam War, the worsening of
the job market, and the declining role of the American Political Science
Association resulting from the emergence of competing organizations. Partly
in reaction to their recent lack of significant success in sponsoring candidates,
and because the APSA Nominating Committee, in an effort to be more
representative, has regularly nominated some New Caucus members, the New
Caucus has stopped nominating its own slate. In 1978 the New Caucus pro-
claimed itsef a socialist organization; in 1979 it amended this to include
socialist feminist. Its avowed objective is to create "a socialist center of gravity
in the profession" (Caucus For A New Political Science, 1981). It has
developed local chapters and is working with other groups, such as the
Marxist Literary Group in sponsoring conferences. Its journal, New Political
Science, deals with such topics as "The Left and Civil Liberties" and "The
Socialist Academic: Between Theory and Practice." It meets annually as an
unaffiliated group at the APSA convention. In 1982 it sponsored 25 panels. A
major attraction of the New Caucus has been that of providing a forum and a
critical means of support for those doing analysis of policy issues relevant to
creating the social changes necessary to transcend capitalist society (Caucus
For A New Political Science, 1981; Lankowski, 1982).

The Status of Women. Another important and related aspect of the challenge
raised by the non-establishment groups in the late 1960's was the situation of
women political scientists. In 1969, about 5 percent of APSA's members were
women. About 8 percent of the faculty members in the field were women
(Gruberg & Sapiro, 1978, p. 318); the lesser amount of membership may have
reflected low pay, and temporary appointments.
In 1969 the Association adopted a resolution supporting equal treatment
for women (American Political Science Association, 1969a, p. 671) and the
Committee on the Status of Women was established (American Political
Science Association, 1971, pp. 3, 10). In the same year women's concerns
were organized through the establishment of the Women's Caucus for
Political Science. In the 1970's the Women's Caucus ran candidates for APSA
offices. Efforts to combine slates with the New Caucus were rarely successful
because many women who were concerned with feminist issues did not share
the New-Left ideology of the New Caucus.
The 1970's saw a dramatic percentage increase in the proportion of
 Naomi B. Lynn I 117

women in the field , except at the rank of full professor:

% Women
1972 1981

Bachelor's degrees in Pol. Science 19% 37%

Master's degrees in Pol. Science 19% 28%
Ph.D. in Pol. Science 10% 20%
Assistant Professors 10% 20%
Associate Professors 8% 12%
Full Professors 4% 5%

Source: Data from Thomas Mann, Executive Director, American Political

Science Association.

The APSA has never had a woman president. In 1979 the Women's
Caucus nominated Betty Nesvold for president. The New Caucus also ran a
candidate against the nominating committee's choice, Charles Lindblom.
Although the Women's Caucus ran a vigorous campaign their candidate
came in third and far behind the official nominee (813 to 2,335) (American
Political Science Association, 1980, p. 42). Over the past decade Women's
Caucus candidates who were also official Nominating Committee choices
gained election to office. This has provided them with representation on the
Executive Council of the Association.
In 1978 the Women's Caucus and other supportive groups, in an intense-
ly debated motion before the Association's business meeting, passed a resolu-
tion calling on the Association not to hold its annual meeting in Chicago
because the State of Illinois had not passed the Equal Rights Amendment to
the U.S. Constitution. If the resolution had received less than a two-thirds
majority it would have had to be ratified by the total membership by mail
ballot. Those opposing the action argued that the ERA position violated the
Al>SA constitution's ban on committing its members on questions of public
policy. Others believed that the APSA should honor its hotel contract which
had been made before the extension of the time period for ratification of
ERA. They feared the law suit which ultimately occurred. Those supporting
the business meeting action responded that in 1972 APSA had passed a
resolution in favor of ERA and the Chicago boycott was simply an implemen-
tation of present policy. They also contended that one of the Association's
main raisons d'etre is to promote and protect the interests of its members. As
long as women members of the Association are victims of professional and
employment discrimination, the Association has a responsibility and an
obligation to take any and all steps necessary to help alleviate the situation.
The reaction of the Association's leadership was similar to its response in
1969 when it feared the New Caucus would take over the Association at
business meetings. The 1969 response was to suggest a change in the Associa-
tion's constitution so that candidates who were opposed had to be elected by a
mailed secret ballot rather than at the business meeting which had been the
earlier practice. The argument was that "temporary majorities" should not be
 118 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

making decisions for the vast majority of members. This was countered by the
contention that members with sufficient interest to attend the annual meeting
can adequately represent the membership (American Political Science
Association, 1969b, pp. 269-302). There was thus a sense of deja vu in 1979 as
the establishment moved to limit the power of the business meeting to take
policy positions. A detached observer of the 1969 meeting made some obser-
vations which were equally applicable a decade later. Anthony King of the
University of Essex, England commenting on the struggle between the New
Caucus and the establishment wrote:

One is bound to feel sorry for ... (New Caucus spokesperson). The poor man
claims to value democracy and participation. But he knows very well that the
more democracy and participation there is in the Association, the worse the
prospects for the Caucus, so he is reduced to extolling the virtues of the Annual
Business Meeting which he knows can, with any luck, be controlled by a tiny
unrepresentative minority .. . Meanwhile, the defenders of the status quo profess
to be anxious to embody democratic principles in the Association's constitution,
whereas they are really worried about the possibility of a Caucus takeover. It is
all an act on both sides . The Establishment's act is marginally more enjoyable, if
only because it is being played with such a straight face . (American Political
Science Association , 1969b, p. 294)

In 1980 the Constitution was amended by a vote of 2400 to 887 so that

controversial policy issues receiving one-third vote at the annual meeting must
be submitted to the entire membership by mailed secret ballot (American
Political Science Association, 1980, p . 42).
The influence of the Women's Caucus, however, has been felt, and
women have influenced the Association in many significant ways. In 1981,
women held 11 percent of all political science full-time positions (American
Political Science Association, 1982a, p . 1). This represents progress, if not
great strides. The outlook is that the percentage of women is likely to do little
more than hold steady because of the small number of openings, lower Ph.D.
output, and the lower attractiveness of academic careers.

Minority Members. The American Political Science Association had a Black

President, Ralph Bunche, in 1954. Despite this milestone the inclusion of
racial minorities in the discipline remained extremely low, relative to the
population at large. In 1969 there were only 65 black American Ph.D.s in
Political Science, well under one percent of the total (Prestage, et al., 1977,
p. 1). In 1969, however, two developments improved the situation. One was
the establishment of a Committee on the Status of Blacks in the Profession;
the other was the establishment in the APSA of a fellowship program for
black graduate students. Evron Kirkpatrick, then Executive Director of APSA
was prominent among those who worked to improve the status of blacks. By
1977 the number of black political science Ph.D.s reached 200 (Woodard,
1982).
In 1977 the Committee on the Status of Chicanos was established for
similar reasons. This committee also encourages members of its group to
enter the field, seeks aid for graduate students, and stimulates research.
 Naomi· B. Lynn I 119
Chun-tu Hsueh argues that Asian and Asian-Americans are under-
represented and discriminated against in political science. To support this
contention he observes that no Asian has ever served on the APSA Council
and that the Council denied a 1976 request to establish a Committee on the
Status of Asians. The Council, however, agreed to publish professional
notices of special concern to Asians in PS and to make space available at
annual meetings for meetings of Asian political scientists. The status of Asians
in political science contrasts with their prestige in other disciplines such as
physics. More Asian-American political scientists could serve as helpful role
models for Asian and Asian American students; they could also offer useful
insights into many issues (Hsueh, 1976-77, pp. 11-15).
Had the favorable market conditions of the l 960's persisted, the efforts
of the committees established to improve the status of underrepresented
groups in the profession would probably have yielded more positive results. As
things have evolved, the profession was opened noticeably for minority groups
at the time of the market decline. The placement success of blacks was 90 per-
cent in 1978 and 96 percent in 1981; for Spanish surnamed political scientists
it was 100 percent in 1980 and 67 percent in 1981. The reader is reminded
that we are dealing with small numbers. Blacks were 4 percent of 198l's
placement class of 696; Spanish surnamed candidates comprised only 2 per-
cent. By contrast the placement of women in 1981 was 75 of 101 candidates
(including women of all races); this was the same percentage as that of men of
all races (Sheilah Mann, 1982, p. 88).

Association Responses. The establishment of the committees on women,

blacks and Chicanos was in itself a significant response on the part of the
APSA and demonstrated the willingness of the Association to open itself more
to the full participation of all members. All the recent Executive Councils
have had women and minority officers. There are those who believe that the
comparatively small number of women and minorities involved in leadership
positions represents "tokenism" and object to the tendency to re-appoint the
same small group instead of seeking to identify new talent.
One available measure of tangible progress is participation in the annual
meetings. Progress by women has been dramatic in this respect. In 1970, for
example, 5.6 percent of the papers presented at APSA meetings were by
women; by 1980 the figure was 21.9 percent and in 1981 it was 18.8 percent
(Gruberg, 1981, p. 725). Women are thus represented more as paper givers
than as full-time faculty.
Blacks have had more uneven progress, due perhaps in part to their
relatively small numbers. Black program participation rose from 14 items in
1969 to 23 in 1972, but it dropped back to 16 in 1976 (Prestage, 1977, p. 16).
Unfortunately full data on minority representation on panels are not
available.
120 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

STRIVING FOR QUALITY AND VITALITY IN A

DECADE OF AlJSTERITY: APPROACHES TO
MEETING THE CHALLENGES
Political scientists in the 1980's must understand and accept the realities
of their situation, but they also must be future-oriented. There are discourag-
ing aspects to the stagnant job market and the low turnover of personnel. We
should remember, however, that most academic fields share a similar en-
vironment. Non-academic fields-many in the private sector and some in the
public - often look at mere stagnation with envy, as they fare worse.
This period of increasing scarcity may not be all negative. Some seren-
dipitous results may come out of the experience. Let me suggest two: First,
the cut in research funding may minimize what many consider to be a prosti-
tution of academia. Some observers have seen instances in which new ideas in
the area of pure theory were shunted aside because researchers found it more
convenient to work on projects for which funds were more readily available.
Perhaps more people will be making decisions on the basis of academic value
rather than fund availability. Second, we may see a resurgence of an emphasis
on teaching. Areas such as political theory which have been devalued because
of limited outside funding sources may experience a renaissance. Third, we
are witnessing what Thomas Mann has described as the "political mobiliza-
tion of the social science community" (Mann, 1982a, p. 416). In response to
the Reagan Administration's proposed cuts for social science research, the
Consortium of Social Science Associations was established. This new
organization has the potential for providing a unified and effective voice in
representing social science interests in Congress and with the federal
bureaucracy.
The prime responsibility for maintaining academic vitality must rest at
the individual department level. We must accept the fact that we have
relatively young faculties who will not be able to move away. Faculty develop-
ment must attain a very high priority. Support for attendance at professional
meetings and seminars must be augmented, perhaps with outside gifts. As the
pattern of student interests inevitably shifts, our nearly "tenured-in" faculties
will have to re-tool. We may need to encourage a network of faculty ex-
changes. We will need to make more productive use than ever of sabbaticals.
Department heads and chairpersons must assume an appropriate leadership
role in seeking adequate support from their administrations; then they must
work together nationally - in large part through APSA and the regional
associations - to share their thoughts and to learn what methods work most
effectively. More attention needs to be given at national meetings to profes-
sional development, and it also merits more research attention.
The Association itself had demonstrated that it is capable of responding
to changing conditions and membership concerns. It now must move beyond
response toward future-oriented actions.
 Naomi B. Lyrtn I 121

NOTES

1. The respondents are sufficientiy close in overall profile to the membership of the
American Political Science Association to warrant the use of the data shown. The
results are also generally consistent with other sample surveys of members.

REFERENCES
American Political Science Association . Association notes. PS, 1969, 2: 671. (a)
American Political Science Association. Special symposium: The governance of the
association. PS, 1969, 2 : 269-302. (b)
American Political Science Association. Women in politt"cal science: Studies and
reports of the APSA Committee on the Status of Women in the Profession.
Washington, D.C .: American Political Science Association, 1971.
American Political Science Association. Reports of APSA Committees: 1973 Elections
Committee. PS, 1974, 12: 36.
American Political Science Association. Reports of APSA Committees: 1979 Elections
Committee. PS, 1980, 13: 42.
American Political Science Association. Annual meeting registration - 1967-1980. PS,
1981, 14: 601.
American Political Science Association . APSA departmental seroices program: 1981-
82 suroey of departments. Washington, D.C.: American Political Science Associa-
tion, 1982. (a)
American Political Science Association . Association news . PS, 1982, 15: 627 -628. (b)
Baker, Earl M. The political science profession in 1970: Basic characteristics. PS,
1971, 4: 33-39.
Barry, Brian M. Sociologists, economists and democracy. London: Collier-MacMillan,
1970.
Baum , William C. , Griffiths , G.N., Mathews, Robert & Scherruble, Daniel. American
political science before the mirror: What our journals reveal about the profession.
The journal of Politics, 1976, 38: 894-917.
Caplow, Theodore. The academic market place. New York: Basic Books, 1958.
Caucus for a New Political Science. Convention program, statement of objectives. New
York, 1981.
The Christian Science Monitor/ American Political Science Association survey. Data
obtained from the Roper Center, University of Connecticut, Storrs, CT 06268, 1981.
Demerath , N. J. Assaying the future : The profession vs. the discipline? The American
Sociologist, 1981, 16: 87 -90.
Easton, David. The new revolution in political science. The American Political Science
Review, 1969, 63: 1051-1061.
Finley, David . Draft of Comments made as panel discussant, Annual Meeting of the
American Political Science Association, Denver, Colorado, 1982.
French, Eloise. Registration rises at annual meeting. PS, 1981 , 14, 786 .
Frolich, Norman & Oppenheimer, Joe A. Modem political economy. Englewood
Cliffs, N.J.: Prentice Hall, 1978.
Gallup Opinion Index. Report No . 182, 27, 1980.
Glass, James M. In Goodin, Robert E. The politics of rational man. New York: John
Wiley, 1976.
Goodin, Robert E. The politics of rational man. New York: John Wiley, 1976 .
Graham, George J. Jr. & Carey, George W. The post-behavioral era: Perspectives on
122 I SELF-PORTRAIT: PROFILE OF POLITICAL SCIENTISTS

Greenstein, Fred I. & Polsby, Nelson W. Preface. In Fred I. Greenstein & Nelson W .
Polsby (Eds.) Political science: scope and theory (Vol. 1) Handbook of Political
Science. Reading, Massachusetts: Addison-Wesley, 1975.
Gruberg, Martin. Letters. PS, 1981 , 14: 725-726.
Gruberg, Martin & Sapiro, Virginia. Participation by Women in Annual Meetings.
PS, 1979, 12: 318-324.
Hargrove, Erwin C. Can political science develop alternative careers for its graduates?
PS, 1979, 12: 446-450.
Hawley, Amos H . Whither the ASA? The American Sociologist, 1981 , 16: 108-110.
Hsueh , Chun-tu. Asian political scientists in America. Hong Kong University, Political
Sdence]oumal, 1976-77: 11 -15 .
Jones, Charles 0. Correspondence with the author, 1982.
Ladd, Everett Carll , Jr. & Lipset, Seymour Martin. The politics of American political
scientists. PS, 1971, 4, 135-144.
Ladd, Everett Carll, Jr. Portrait of a discipline: The American political science com-
munity, part I. Teaching Political Science, 1974, 2, 3-39. (a)
Ladd, Everett Carll, Jr. Portrait of a discipline: The American political science com-
munity, part 2. Teaching Political Sdence, 1974, 2, 144-171. (b)
Ladd, Everett Carl, Jr. The divided academy. New York: McGraw Hill, 1975.
Ladd, Everett Carll, Jr. Us revisited . Presented at the Annual Meeting of the American
Political Science Association, New York, 1978.
Lane, John C. The slow graying of our profession. PS, 1982, 15: 50-52.
Lankowski, Carl. Telephone interview with the 1980-81 President of the Caucus for a
New Political Science. June 23, 1982.
Lipset, Seymour Martin. Politics and the social sciences: Introduction. In Seymour
Martin Lipset (Ed.) Politics and the social sciences. New York: Oxford University
Press, 1969.
Lipset, Seymour Martin. The limits of social science. Public Opinion, 1981, 4: 2-9.
Lowi, Theodore]. The politics of higher education: Political science as a case study. In
George J. Graham, Jr. and George W. Carey (Eds.), The post-behavz'oral era:
Perspectives on political science. New York: David McKay, 1972.
Mann, Sheilah. Placement of political scientists. PS, 1982 , 15: 84-91.
Mann, Thomas. Placement of political scientists. PS, 1976, 9: 412-414.
Mann, Thomas. Placement of political scientists in 1977. PS, 1978, 11: 26-29.
Mann, Thomas. Executive director's quarterly column. PS, 1981, 15: 582-583.
Mann, Thomas. From the Executive Director: First year report. PS, 1982, 15: 415-420.
(a)
Mann, Thomas. Letter to the Council of the American Political Science Association,
1982. (b)
Mann, Thomas. Memo to the Council of the American Political Science Association,
1982. (c)
Minutes of the Administrative Committee of the Executive Council of the APSA, April
5, 1982.
Prestage, Jewel, Adams, Russell, Jones, Mack, Martin, Robert, Moreland, Lois &
Willingham, Alex. Report of the conference on political science curriculum at
predominantly black institutions. In Maurice C. Woodard (Ed.), Blacks and
political science. Washington, D.C.: The American Political Science Association,
1977.
Prestage, Jewel, Adams, Russell, Jones, Mack, Martin, Robert, Moreland, Lois &
Willingham, Alex. Quelling the mythical revolution in higher education: Retreat
from the affirmative action concept.journal of Politz'cs, 1979, 41: 763-783.
·Robey, John S. Political science departments: Reputations versus productivity. PS,
1979, 12: 202-209 .
 Naomi B. Lynn I 123

Robey, John S. Reputations vs citations: Who are the top scholars in political science?
PS, 1982, 15: 199-200.
Roettger, Walter B. I never promised you a rose garden: Career satisfaction in an age
of uncertainty. Paper presented at the Iowa Conference of Political Science. Iowa,
1977.
Roettger, Walter B. The discipline: What's right, what's wrong, and who cares? Paper
presented at the annual meeting of the American Political Science Association. New
York, N.Y ., 1978.
Roettger, Walter B. Strata and stability: Reputations of American political scientists.
PS, 1978, 11: 6-12.
Roettger, Walter B. & Winebrenner, Hugh. Professional knowledge and electoral
behavior: The case of political scientists. Paper presented at the American Political
Science Association convention. Washington, D.C., 1981.
Roose, Kenneth D. & Anderson, Charles J. A rating of graduate programs.
Washington, D.C.: American Council on Education, 1970.
Rossi, Peter H. The ASA: A portrait of organizational success and intellectual
paralysis. The Amen·can Sociologist, 1981 , 16: 113-116.
Sornit, Alben & Tannenhaus, Joseph. The development of political science: From
burgess to behavioralism. Boston: Allyn and Bacon, 1967.
Turner, Henry A. & Hetrick, Carl C., 1972, Political activities and party affiliations
of american political scientists. The Western Political Quarterly, 1972, 25: 361-374.
Turner, Henry A, McClintock, Charles G., Spaulding, Charles B., 1963, The political
party affiliations of American political scientists. The Western Political Quarterly,
1963, 16: 650-665.
Waldo, Dwight. Political science: Tradition, discipline, profession, science. In Fred I.
Greenstein and Nelson Polsby (Eds.), Political science: Scope and theory.
Reading, MA: Addison-Wesley Publishing Company, 1975.
Walker, Jack L. Challenges of professional development for political science in the
next decade and beyond. PS, 1978, 11: 484-490.
Woodard, Maurice C. Interview. Washington, D.C., July 12, 1982.