Exemplification in Newspapers: A Content Analysis and Case Studies

A thesis presented to

the faculty of

the Scripps College of Communication of Ohio University

In partial fulfillment

of the requirements for the degree

Master of Science

Dustin A. Weaver

August 2009

© 2009 Dustin A. Weaver. All Rights Reserved.

 This thesis titled

Exemplification in Newspapers: A Content Analysis and Case Studies

by

DUSTIN A. WEAVER

has been approved for

the E. W. Scripps School of Journalism

and the Scripps College of Communication by

Bill Reader

Assistant Professor of Journalism

Gregory J. Shepherd

Dean, Scripps College of Communication

ii
Abstract

WEAVER, DUSTIN A., M.S., August 2009, Journalism

Exemplification in Newspapers: A Content Analysis and Case Studies

(129 pp.)

Director of Thesis: Bill Reader

The exemplification theory of communication suggests that the human

mind is acutely receptive to concrete information from its surroundings. In a

series of experimental studies, communication researchers have consistently

found that people tend to extrapolate from exemplars they see in the media—

even when given conspicuous statistics that suggest the given exemplars are

in the minority of cases. Despite the findings of those experimental studies,

only a handful of researchers have attempted to catalog the use of exemplars

in media reports. This study was undertaken as a first step toward correcting

that research imbalance.

This study employed a content analysis to measure the use of

exemplars in six metropolitan American newspapers in 2003. The content

analysis used a sample of two constructed weeks to search for exemplified

narratives on the front page of each newspaper and on the front of the local

news/metro section. The content analysis did not uncover a significant

number of exemplified narratives in a constructed two-week sample of

papers, which suggests that the experimental model of exemplification may

be at odds with real-world journalism practices.

iii
 This study also examined four of the exemplified narratives found by

the content analysis in qualitative case studies. Those case studies employed

Hall’s model of encoding/decoding and Barthes’ conception of myth to

reconsider the underlying theoretical assumptions of exemplification theory.

The results of the exercise suggest that the interpretation of exemplars by an

audience is a much more complex process than researchers may have

previously acknowledged in the literature on exemplification.

Approved: _____________________________________________________________

Bill Reader

Assistant Professor of Journalism

iv
Acknowledgments

I want to pay special thanks to Dr. Daniel Riffe for all his

encouragement and guidance during the long course of this project. It was a

pleasure to have had him as an instructor and mentor during my short time

at Ohio University.

I owe an enormous amount of gratitude to Professor Bill Reader for

assuming the chairmanship of my thesis committee late into the process. His

suggestions and edits expanded the scope of the study far beyond what I had

originally envisioned, and I couldn’t be more pleased with the results. I’d also

like to thank Professor Cary Frith for her service and contributions as a

member of my committee.

Finally, I want to thank my parents, Allen and Cheryl, for lovingly

insisting that I get an education. I hope it’s obvious that I got the message.

v
For my wife, Milada

vi
Table of Contents

Page
Abstract...............................................................................................................iii

Acknowledgments................................................................................................v

List of Tables.......................................................................................................ix

Chapter 1. Purpose...............................................................................................1

Chapter 2. Introduction........................................................................................5

Exemplification Concepts and Terminology ................................................... 5

The Ethical Implications of Exemplars .......................................................... 9

Chapter 3. Literature Review............................................................................13

Measuring the Exemplification Effect .......................................................... 13

Exemplification: The Foundational Studies ................................................. 14

Exemplar Distortion and Base-Rate Precision ............................................ 21

Exemplification and Prior Opinion ............................................................... 22

Emotion in Exemplars................................................................................... 23

Quotes and Paraphrases ............................................................................... 24

Social Similarity ............................................................................................ 25

Measuring Exemplars in Content Analysis ................................................. 26

Chapter 4. Research Questions.........................................................................30

Chapter 5. Method..............................................................................................34

Protocols Reliability Test .............................................................................. 42

Chapter 6. Results..............................................................................................43

Discussion of Results of Content Analysis ................................................... 52

vii
 Exemplification and Isolated Incidents ........................................................ 55

Exemplified Narratives ................................................................................. 57

Chapter 7. Case Studies: Purpose.....................................................................58

Chapter 8. Case Studies: Perspectives..............................................................60

Encoding and Decoding Exemplars .............................................................. 60

Exemplars as Myth ....................................................................................... 64

The Study of Signs......................................................................................... 65

Myth in Society .............................................................................................. 67

Myth in the News .......................................................................................... 69

Summary ....................................................................................................... 72

Chapter 9. Case Studies.....................................................................................74

Case #1: Exemplification by the Numbers ................................................... 74

Case #2: Where Did the Crabs Go? ............................................................... 76

Case #3: Suicides in Japan ........................................................................... 81

Case #4: Young Offenders ............................................................................. 88

Case Studies: Review .................................................................................... 93

Chapter 10. Suggestions for Future Research..................................................97

Works Cited ………………………………………………………………………….100

Appendix A. Codebook......................................................................................100

Appendix B. Coding Protocol...........................................................................110

Appendix C. Case Study Articles ....................................................................112

viii
List of Tables
Page

Table 6.1: Number of Articles (n) Included in Coding Process .....................43

Table 6.2: Frequency of Exemplar Totals in Articles ....................................45

Table 6.3: Frequency of Exemplified Topics ..................................................46

Table 6.4: Frequency of Exemplar Sources ....................................................47

Table 6.5: Frequency of Exemplar Sources in Each Newspaper...................48

Table 6.6: Average Number of Sentences Per Exemplar Source Type .........49

Table 6.7: Frequency of Exemplar Distribution. ..........................................50

Table 6.8: Frequency of Exemplar Foci ..........................................................50

Table 6.9: Precision of Statistics ....................................................................51

ix
Chapter 1. Purpose

The recently conceptualized1 exemplification effect of communication

suggests that the human mind is acutely receptive to concrete information

from its surroundings.2 The findings of exemplification research, which have

accumulated over the past two decades, are consistent and provocative.

Researchers have found that people unknowingly make judgments from

exemplars they see in the media—even when given conspicuous statistics

that suggest the rarity of the example. To borrow a term from Noelle-

Neumann,3 it seems that people possess a quasi-statistical sense for scanning

and assessing the social environment in which they live. Unfortunately, that

type of cognitive scanning can foster perceptions out of line with empirical

measurements of reality.

To date, the findings of exemplification research have been replicated

enough to warrant confidence in the concept’s validity. Having established

that the exemplification effect occurs (under admittedly precise conditions), it

would seem prudent to compare the findings to samples of media content.

1 For an historical overview of exemplification research that begins with the first media
study in 1992, see Hans-Bernd Brosius, “Exemplars in the News: A Theory of the Effects of
Political Communication,” in Communication and Emotion: Essays in Honor of Dolf
Zillmann, eds. J. Bryant, D. Roskos-Ewoldsen and J. Cantor (Mahwah, New Jersey:
Lawrence Erlbaum, 2003).
2 Rhonda Gibson and Dolf Zillmann, “Exaggerated Versus Representative Exemplification in

News Reports: Perceptions of Issues and Personal Consequences,” Communication Research

21 (1994).
3 Elisabeth Noelle-Neumann, “The Spiral of Silence: A Theory of Public Opinion,” Journal of

Communication 24 (1974).
Many researchers, including Zillmann and Brosius, have commented on the

troublesome ethical implications of the exemplification effect,4 and yet little

has been done to track the use of exemplars in the news. Without such data,

it is difficult to assess the role of exemplars in day-to-day media presentation

and reception.

Furthermore, there is an assumption in the writings of Zillmann and

Brosius that deserves closer scrutiny. The authors claimed, “In the media of

the United States, exemplification abounds,”5 and that the media generate

“an exemplar flood.”6 According to those researchers, exemplification is

ubiquitous in mass communication, as common as the weather report or the

daily box score. “The news media thrive on exemplification,” they wrote. “The

case report, or more accurately, the aggregation of intriguing case reports, as

it appears to hold considerable public interest, may be considered the

lifeblood of journalism.”7 (emphasis added).

Unfortunately, there is a scarcity of published data to justify or expand

upon those claims. The only known content analyses of exemplars performed

on American media were never published in complete form, and the coding

protocol used to compile them is no longer available, according to Gibson and

4 Dolf Zillmann and Hans-Bernd Brosius, Exemplification in Communication: The Influence

of Case Reports on the Perception of Issues (Mahwah, New Jersey: Lawrence Erlbaum, 2000):
127-132.
5 Ibid., 19.
6 Ibid., 3.
7 Ibid., viii.

2
Zillmann.8 The data from those studies, which were outlined in the book on

the subject by Zillmann and Brosius, also seem to be of limited scope due to

the study’s design. While those analyses did find prevalent exemplar usage,

they focused on news magazines and network television news shows, which

have a very different style of reporting than their daily print counterparts.

Thus, laying the groundwork for exemplification measurement in daily

newspapers is the explicit aim of this research project.

This study will employ a content analysis to measure the use of

exemplars in six major American newspapers. It will (a) begin with an

overview of exemplification theory, terminology, and research, (b) review the

literature to chart the theory’s origins and development, (c) form research

questions from the literature review, (d) outline a content analysis method

and coding protocol to answer the research questions, (e) present the findings

of the content analysis, (f) introduce a framework for conducting case studies

of selected coded articles, and (g) relate the coded articles to the findings of

exemplification research to propose protocols for qualitative textual analysis

of exemplars in the media.

The findings of the content analysis will call into question the idea that

exemplification is common in newspapers. Chapter 7 will review possible

explanations for the scarcity of cases found by this study and suggest

8 E-mail correspondence with Dolf Zillmann, August 2004.

3
methodological tweaks for future attempts to study the concept via content

analysis.

Because this study provides an opportunity to examine real-world

instances of exemplification, Chapter 8 will present a framework for case

studies of selected coded articles. That framework will incorporate theoretical

perspectives from outside traditional exemplification research to show the

potential for a broader approach to studying the phenomenon. Chapter 9 will

present the case studies, and Chapter 10 will suggest further ideas for

exemplification research that are worthy of attention.

4
Chapter 2. Introduction

This section will define and explain the terminology of exemplification

theory to build a foundation for this study. The media’s ability to shape public

perception will be linked to exemplars, and the practical and ethical issues

raised by the exemplification effect will be introduced.

Exemplification Concepts and Terminology

Media exemplification theory came about largely due to the research

efforts of Zillmann and his collaborators.9 Besides producing important

studies in the field, Zillmann co-wrote the most comprehensive summary of

the theory and its ramifications, titled Exemplification and Communication.

That text largely informs the summary that follows.

Understanding the theory of exemplification requires familiarity with

a few key terms. The first, exemplar, is best understood in relation to other

concepts, as an exemplar is simply a representation of a category. A man who

killed his wife in New Jersey, for instance, is a person who exemplifies the

concept of “murderer.” Depending on the details of the event, the Jersey

homicide may also be used to exemplify extreme reactions to spousal

disagreements, domestic abuse, heartbreak, infidelity, and so on. The

possibilities for the use of exemplars in media messages are essentially

limitless, as even singular events can take on exemplified properties. For

9 Credit should be given to researchers in psychology, whose findings laid the groundwork for

a media-focused approach. Zillmann and Brosius cited studies from that field extensively in
Exemplification and Communication.
5
example, Zillmann observed that the first manned moon landing could be

employed as an exemplar of a successful spacecraft landing despite its unique

place in history.10

Certain features of human cognition, studied extensively by

Kahneman, Slovic and Tversky,11 explain the psychological rationale behind

exemplification studies. The researchers discovered people use cognitive

shortcuts when collecting information about reality, unconsciously employing

techniques called heuristics. The representative heuristic leads people to

extrapolate from known examples when making judgments, essentially

treating mental recollections as an objective population sample.12 “As a rule,

information is not being evaluated in a detached, objective, and sciencelike

fashion,” Zillmann and Brosius argued. “Statistical information and stated

probabilities tend to have less influence on judgment than do concrete case

histories because, concerning everyday situations, people usually do not have

access to statistical information and thus are unfamiliar with incorporating

such information when rendering judgment.”13

10 Dolf Zillmann, “Exemplification Theory: Judging the Whole by Some of Its Parts,” Media

Psychology 1 (1999): 72.

11 Judgment Under Uncertainty: Heuristics and Biases, Eds. Daniel Kahneman, Paul Slovic

and Amos Tversky (New York: Cambridge University Press, 1982).

12 Dolf Zillmann, “The Exemplification Theory of Media Influence,” in Media Effects:

Advances in Theory and Research, 4th edition, eds. D. Zillmann and J. Bryant (White Plains,
N.Y.: Longman, 1997): 27.
13 Zillmann and Brosius, 42.

6
 The availability heuristic holds that there is a hierarchy in the mind

that makes some thoughts and memories easier to recall than others.14 That

heuristic exposes rationality’s nagging blind spot: The easier a thought is to

remember, the more influence it has, which means even the most

disinterested of introspections are biased by the mind’s all-encompassing

drive toward parsimony. This puts pallid, scientific information at a great

disadvantage heuristically, since exemplified human drama tends to be very

memorable.15

Some exemplars are frequently repeated throughout the media, which

can lead to chronic accessibility.16 As every propagandist knows, repetition

can make a message more salient over time—a deception made possible by

the heuristic machinery of the mind. Journalism’s well-documented culture of

habits and routines,17 complemented by the tendency to succumb to agenda

setting,18 seems to indicate that, at any given time, there will be a widely

understood collection of exemplars dominating media coverage. As Zillmann

explained, “Most significant media effects are, after all, thought to be built on

14 Rick W. Busselle and L.J. Shrum, “Media Exposure and Exemplar Accessibility,” Media
Psychology 5 (2003): 257.
15 Zillmann, “The Exemplification Theory of Media Influence,” 28.
16 Zillmann and Brosius, 46-47.
17 See L. Erwin Atwood and Gerald L. Grotta, “Socialization of News Values in Beginning

Reporters,” Journalism Quarterly 50 (Winter 1973); Warren Breed, “Social Control in the
News Room: A Functional Analysis,” Social Forces 33 (1955); Dan Berkowitz, “Non-Routine
News and Newswork: Exploring a What-a-Story,” Journal of Communication 42 (1992); Gaye
Tuchman, “Making News by Doing Work: Routinizing the Unexpected,” American Journal of
Sociology 79 (July 1971).
18 Maxwell E. McCombs and Donald L. Shaw, “The Agenda-Setting Function of Mass Media,”

Public Opinion Quarterly 36 (Summer 1972).

7
frequent and consistent exposure to largely redundant concepts, most of

which fall well within the conceptual range of exemplification.”19

Researchers have argued that human beings habitually engage in

heuristic thinking to make sense of the world. To illustrate, imagine a man

who takes the same highway route every day to work. On two separate

occasions, he witnesses horrific accidents at an intersection along the way.

Rather than viewing those accidents as isolated incidents, he assumes that

the intersection is incredibly dangerous and drives slowly past it every

morning. Statistically speaking, it is entirely possible that the two accidents

the man witnessed were anomalies, and that the intersection is actually the

safest on the route. The man relies on exemplars, and the representative

heuristic, to understand the world around him.

That kind of intuitive inductive reasoning, based primarily on personal

observation, is equally applicable to media content; the man could have easily

read about the accidents in the newspaper and then decided upon the same

cautious driving pattern. As Shapiro noted, “The mental processes engaged in

answering a social-reality question are probably not very different from the

process a person uses to decide how safe it is to walk to the corner store at

night.”20

19Zillmann, “The Exemplification Theory of Media Influence,” 28.

20Michael A. Shapiro, “Memory and Decision Processes and the Construction of Social
Reality,” Communication Research 18 (1991): 5.
8
The Ethical Implications of Exemplars

For most of human history an individual’s knowledge was limited to

people, objects, and events in the immediate realm of physical sensory

experience. As Lippmann explained, “The only feeling that anyone can have

about an event he does not experience is the feeling aroused by his mental

image of that event.”21 Media technology eliminates physical barriers,

bringing a wealth of second hand information about all manner of people,

places and events. That information, selected and filtered by news

organizations and journalists, helps create what Lippmann called the pseudo-

environment, or the limited environment of a person’s perceptions.22

Because it is impossible for journalists to hold a mirror up to show all

of society, media professionals compromise by modestly aiming to reveal

superficial truths about the world through vigilant effort and disinterested

analysis. Professional standards such as objectivity implicitly acknowledge

human limitations, allowing journalists to construct a superficial summary of

“the truth” without agonizing over its ultimate accuracy or

comprehensiveness. As Schudson observed, “[Journalistic] objectivity is an

ideology of the distrust of the self.”23 Detached, impartial weighing of

evidence is widely viewed as a morally responsible way for journalists to

21 Walter Lippmann, “The World Outside and the Pictures in Our Heads,” in The Process and
Effects of Mass Communication Revised Edition, eds. W. Schramm and D. Roberts (Urbana:
University of Illinois Press, 1974): 273.
22 Ibid., 274.
23 Michael Schudson, Discovering the News (USA: Basic Books, 1978): 71.

9
construct the news—a formula to be used, in good faith, to navigate the

inherent limitations of the profession and inform the public about the world.

Exemplification research indicates that basic ethical framework is

incomplete because it fails to account for the persuasiveness of anecdotes and

quotes in news coverage. Exemplification research suggests the media has

the potential to create unfounded fears and beliefs through the cumulative

presentation of exemplars. Noted Craig, “Anecdotes are morally a double-

edged sword. While they can bring attention to the plight of individuals in

riveting fashion, and even place that plight in institutional, professional, and

social contexts, the attention remains on a representative, or perhaps

unrepresentative, individual or family.”24

The potential for deceit is especially acute because newsworthy

information is, by the common definition, novel and out of the ordinary,

which makes it more likely exemplars will be drawn from extreme cases. As

Sundar explained, exemplars “more often than not represent the fringes

instead of the center.”25 The tendency of journalists to select dramatic

exemplars can be seen in a New York Times article about unemployment

from Nov. 7, 2008.26 The article told the story of Ken Stelma, a New Jersey

24 David A. Craig, “The Promise and Peril of Anecdotes in News Coverage: An Ethical

Analysis,” Journalism & Mass Communication Quarterly 80 (2003): 804-805.

25 S. Shyam Sundar, “News Features and Learning,” in Communication and Emotion: Essays

in Honor of Dolf Zillmann, eds. J. Bryant, D. Roskos-Ewoldsen & J. Cantor (Mahwah, New
Jersey: Lawrence Erlbaum, 2003): 276.
26 Peter S. Goodman, “Jobless Rate at 14-Year High After October Losses,” The New York

Times, 7 Nov. 2008 <http://www.nytimes.com/2008/11/08/business/economy/08econ.html>

(Accessed March 8, 2009).
10
man who was on the verge of losing his jobless benefits after failing to find

work for a year. Stelma was being supported financially by his girlfriend and

was at risk of losing his health insurance, which was an especially difficult

problem, given he had a tumor in his foot. He was also unable to pay his car

insurance, and the lack of transportation made it nearly impossible for him to

find a new job. Stelma is the only unemployment exemplar in the Times

article; viewed through the lens of exemplification theory, that privileged

position effectively turned his tale of extreme misfortune into a typical case of

joblessness from a reader’s point of view.

Journalists seem to consider exemplars mainly in terms of helping

readers follow the narrative of a complicated story. One recent journalism

text instructed, “Examples are especially important in stories about abstract

issues. Sometimes numbers help put those issues into perspective ... In

addition to reporting the general trends, a good writer would illustrate the

story by describing ... specific examples of the trend.”27 Exemplars instruct,

inform and entertain media consumers by turning abstractions into

compelling human dramas that are easier to follow and understand.

Given the current research on the exemplification effect, it is

imperative that journalists consider the persuasive power of anecdotes in

news coverage and re-evaluate what meets standards of professional ethics

and accuracy. The “inability on the part of news consumers to factor in the

27Fred Felder, John R. Bender, Lucinda D. Davenport and Michael W. Drager, Reporting for
the Media, 7th ed. (New York: Oxford, 2001): 198.
11
potential atypicality of exemplars in news reports puts an enormous onus on

journalists,” explained Sundar.28 Zillmann and Brosius further argued, “The

problem is that the journalistic format of apprising the public of relevant

occurrences is fraught with manipulatory means of which neither journalists

nor citizens are cognizant.”29 So while exemplars may help journalists make

meaning more clear to the public, their use may actually create a misleading

message that distorts the true extent or importance of an issue.

Exemplification research suggests journalists should reexamine their

professional practices and consider how exemplars can undermine good-faith

efforts at balanced reporting. Based on the data collected so far, it seems that

exemplars should be used judiciously and sparingly in the news, and only

after determining whether the mix of exemplars is representative of the issue

being covered. Those kinds of ethical considerations are absent in journalism

today, which is why it is critical to determine whether exemplification is

common in the media. Until that issue is studied thoroughly, the implications

of the research for professional practice will remain unclear.

28 Sundar, “News Features and Learning,” 286.

29 Zillmann and Brosius, 39.
12
Chapter 3. Literature Review

The first section of this literature review will cover the foundational

studies of exemplification theory. The discussion will then turn to the

theoretical distinctions made by other studies, which will be presented in

separate subsections for the sake of clarity. A section also will be devoted to

an in-depth discussion of one research project—a content analysis of

exemplars—that is of particular interest to this study.

Measuring the Exemplification Effect

The exemplification effect has been tested within a number of media,

including radio, television and print. In a typical exemplification experiment,

researchers manipulate a news story that contains statistical information

and exemplars. Respondents are usually divided into three conditions, with

each receiving a slightly altered version of the news story. That alteration

comes in the form of different exemplar ratios.

The three exemplar conditions have been given various names in the

literature, the most common being selective, representative and blended. In

selective exemplification, all of the exemplars in a story agree with the

“take,” or focus, of the narrative. Representative exemplification is a more

balanced approach, as the exemplar ratio is proportionally similar to the

given statistic. In blended exemplification, a mixture of supportive and

contradictory exemplars is presented.

13
 When respondents in the three conditions are asked to estimate the

prevalence of the phenomenon in the media content, they consistently fail to

remember the statistics in the story, instead offering estimates that are

proportionally similar to the exemplar ratio they are given. If the majority of

exemplars in the story support the idea that a phenomenon is rare,

respondents tend to rate it as statistically rare. If the majority of exemplars

support it as common, respondents tend to rate it as common. As Zillmann

explained, “The point is that our dispositions reflect a quantitative

assessment of exemplar experiences, not necessarily in precise numerical

estimates but in proportional terms.”30

The method used in exemplification experiments has not changed

much over time, despite the innovative contributions of some researchers.31

The precision of the methodology has produced a strong set of findings, but it

is open to criticism, as Chapter 7 will show. More research will ultimately

prove whether the exemplification effect exists outside the confines of tightly

controlled experiments.

Exemplification: The Foundational Studies

The starting point for a review of the exemplification literature is an

early study by Hamill, Nisbett and Wilson that investigated the appearance

30Zillmann, “Exemplification Theory: Judging the Whole by Some of Its Parts,” 88.
31Busselle and Shrum, “Media Exposure and Exemplar Accessibility,” employed a
particularly innovative approach.
14
and power of heuristic thinking in human thought.32 The term

exemplification was not used at that time, but the research can be accurately

labeled under the category; indeed, the format of the study was very similar

to the one later employed in much of Zillmann’s research.

Respondents were given a booklet that contained an abridged story

about a woman on welfare. According to the researchers, “The article

provided a detailed description of the history and current life situation of a

43-year-old, obese, friendly, irresponsible, ne’er-do-well woman who had lived

in New York City for 16 years, the last 13 of which had been on welfare.”33 In

addition to that extreme exemplar, the booklet contained statistical

information. In the typical condition, the statistic confirmed that the

exemplar was representative of the larger population of welfare recipients. In

the atypical condition, the statistic firmly contradicted the validity of the

exemplar, stating that “the average length of time on welfare ... is 2 years.”

After reading the booklet, respondents were given a survey designed to

ascertain their attitudes toward welfare recipients.

There was not a significant difference between the conditions, as both

groups conveyed negative attitudes toward welfare recipients. The atypical

condition—which contained a contradictory statistic showing lengthy stints

on welfare are rare—failed to influence respondents’ judgments. However,

32 Ruth Hamill, Richard E. Nisbett and Timothy DeCamp Wilson, “Insensitivity to Sample

Bias: Generalizing from Atypical Cases,” Journal of Personality and Social Psychology 39
(1980).
33 Ibid., 580.

15
researchers acknowledged that respondents might have regarded the

statistical information as irrelevant when making character judgments about

welfare recipients.34 To correct for that possibility, researchers conducted a

second study.

In the second experiment, four groups of respondents watched a

videotaped interview with a prison guard. Two of the groups saw an

interview with a gentle, humane guard who espoused liberal views on

rehabilitation; in the other interview, also seen by two groups, a bitter, angry

guard railed against the prisoners and announced his extremely low opinion

of their characters. As in the previous study, the groups were divided into two

conditions and told the humane/inhumane worker was either representative

or unrepresentative of prison guards at large. When respondents’ attitudes

about prison guards were surveyed after watching the videos, comparison

with control groups showed that all of the respondents had their beliefs

altered. The “humane” groups rated guards more favorably, and the

“inhumane” groups less favorably, than the control groups, which suggests

the sampling information had little effect on their judgments.35

Hamill, Nisbett and Wilson argued that the results of the studies

showed people tend to draw too much information from atypical examples.

“Subjects in the atypical sample condition would seem to have little

justification for their unfavorable generalizations,” they wrote. “We are

34 Hamill, Nisbett and Wilson, 583.

35 Ibid., 586.
16
proposing that subjects may engage in an unconscious, memory-mediated

generalization from sample to population that remains unaffected by any

conscious processing of information about sample typicality.”36

Zillmann, Perkins and Sundar performed the first media

exemplification study in 1992.37 Researchers manipulated an article about

post-diet weight gain, creating three distinct exemplar conditions—selective,

blended and representative. The story contained statistical information (often

called “base-rate”) that placed the overall weight gain in the given population

at 32 percent. This study was the first to explicitly discern the power of

exemplar distribution; respondents’ estimates of post-diet weight gain clearly

reflected the exemplar emphasis they were given. The average estimate of

post-diet weight gain in the selective condition was 75 percent of the original

weight loss; in blended, 62.3 percent; and in representative, 58.5 percent.

In addition, the study found that respondents in a delayed condition—

asked to make estimates 14 days after reading the articles—seemed to escape

the influence of the exemplar distributions, instead reverting to prior beliefs.

Estimates of weight gain given by people who had not read the article were

statistically consonant with the delayed response, giving credence to a prior

belief explanation.

36Hamill, Nisbett and Wilson, 582, 587.

37Results are summarized in Zillmann and Brosius, Exemplification in Communication. The
original study was written in German.
17
 Brosius and Bathelt conducted a diverse battery of exemplification

experiments that were published together in 1994.38 One of the most

important elements of that research was the use of differing media—namely,

radio and print. A radio news package that was professionally recorded

contained the following stories, all framed in terms local to the student

subjects: changes to a community pay phone system, the quality of cafeteria

food, the quality of the region’s famed hard cider, and the introduction of

mandatory computer courses at a university. All of those stories contained

statistical data. In the radio stage of the experiment, the exemplars were

presented in the form of direct interviews. The print version of that package,

used in the final leg of the experiment, presented the same stories with the

exemplars in paraphrased form.

In the first experiment, limited to radio, Brosius and Bathelt

manipulated exemplar distribution, exemplar vividness and exemplar form.

Regarding distribution, respondents heard a news package in which a

majority of exemplars were consistent or inconsistent with the statistics. As

expected, “Perceived majority and minority opinions were judged in almost

perfect accord with the exemplar distribution.”39 The vividness of the

exemplars was manipulated such that the interviewees in one package spoke

in emotional tones, while the other featured dry, monotone deliveries. No

38 Hans-Bernd Brosius and Anke Bathelt, “The Utility of Exemplars in Persuasive

Communications,” Communication Research 21 (1994).

39 Ibid., 59.

18
effect was found for the vividness manipulation. Similarly, exemplar form—

all paraphrases in one condition versus all quotations in another—did not

produce a difference in the data. A second run-through of the experiment

produced the same results.

In a third experiment, again using the same materials, Brosius and

Bathelt once again looked for an effect of exemplar distribution. The results

were conclusive. “In the extreme condition, which illustrated only one opinion

with exemplar interviews, subjects attributed this opinion almost exclusively

to the population.”40 Similarly, when the exemplars were presented in an

equal split, respondents estimated about half of the population shared one

perspective.

In the fourth experiment, radio and print were varied between

conditions. Respondents in both conditions were influenced by the exemplar

distribution, although the effect was slightly weaker with the print-media

condition. In a final print-only test, the researchers manipulated the

persuasiveness of the exemplar arguments, while also repeating the

statistical information at the end of the text. As Brosius and Bathelt

explained, “If the effects of exemplar distribution are considerably weaker in

versions with weak arguments and repetitions of the base-rate information,

one cannot attribute the effects to the specific quality of exemplars but rather

40 Brosius and Bathelt, 63.

19
to information imbalance or strength of arguments or recency.”41 However,

no effect was found for persuasiveness or recency, leaving exemplar

distribution as the most likely explanation for the effect.

In another study, Gan, Hill, Pschernig and Zillmann studied the

exemplification of a massacre in Israel using an actual news package from

CNN.42 The package was professionally edited to create four exemplar

conditions that the authors labeled “justification,” “condemnation,”

“balanced” and “without exemplification.” In the “justification” condition, all

of the interviewees in the news package were Jews who condoned and

glorified violence against Islamic worshippers. In the “condemnation”

condition, all of the Jewish interviewees strongly condemned the brutality.

The “balanced” condition featured a mixture of the two. The without-

exemplification condition lacked interviews altogether. After watching one of

the four news packages, subjects were given a survey instrument; in it, they

were asked to assign fault for the violent state of Israel. The “justification”

condition produced significant differences in the respondents’ assignments of

blame, effects that failed to appear in the “condemnation” condition.

Researchers speculated that the “condemnation” condition portrayed the

Jewish population as being beyond the control of the country’s leaders, and

therefore as a driving force for violence. Most importantly, the balanced

41Brosius and Bathelt, 69-70.

42Sun-lin Gan, John R. Hill, Elke Pschernig and Dolf Zillmann, “The Hebron Massacre:
Selective Reports of Jewish Reactions, and Perceptions of Volatility in Israel,” Journal of
Broadcasting and Electronic Media 40 (1996).
20
condition produced judgments similar to those in the “without”

exemplification condition. “Balanced exemplification thus recommends itself

once again as a method of informing about social phenomena that is more

appropriate than selective alternatives.”43

With the basic contours of the theory settled, Zillmann, Brosius and

others continued to test for the exemplification effect, but turned much of

their attention to other factors that may mitigate or enhance the

persuasiveness of exemplars in message reception.

Exemplar Distortion and Base-Rate Precision

Gibson and Zillmann built on earlier findings using manipulated

magazine coverage of carjackings.44 Three exemplar conditions were offered

to respondents, and once again, the exemplification effect produced

statistically distinct estimates that reflected the given manipulation.

Gibson and Zillmann also found that the level of exemplar distortion

played a powerful role in message reception. “Respondents exposed to

extremely distorted exemplars, when compared to respondents who had

received less distorted exemplification, were more likely to consider the

carjacking issue a serious national problem.”45 The study found that

exemplars of carjacking deaths were more influential than ones in which

43 Gan et. al, 130.

44 Gibson and Zillmann, “Exaggerated Versus Representative Exemplification in News
Reports.”
45 Ibid., 619.

21
victims escaped unharmed, lending support to a vividness effect in

exemplification.

Contrary to expectations, the precision of the base-rate information did

not affect respondents’ answers; a vague estimate (e.g., “more and more” or

“some”) was ultimately as unpersuasive as a numeric statistic. Researchers

also found that the exemplification effect increased after a one-week delay,

but only in the two extreme exemplar conditions. They hypothesized that “the

dominant influence of sensational accounts on judgments is bound to grow

over time.”46

Exemplification and Prior Opinion

In 1996, Zillmann, Gibson, Sundar and Perkins Jr. clarified the results

of the first media exemplification study from 1992.47 An article about family

farming was manipulated in the study because researchers believed it was an

issue for which the respondents would lack a strong prior opinion. This

hypothesis was supported by the data, as the exemplification effect persisted

unchanged after a two-week delay. Similar to Brosius and Bathelt, the

researchers did not find an effect for the precision of the statistical

information. The post-test survey also found that the selective

overrepresentation of suffering farmers created distress in readers.

46 Gibson and Zillmann,“Exaggerated Versus Representative Exemplification in News

Reports,” 620.
47 Dolf Zillmann, Rhonda Gibson, S. Shyam Sundar and Joseph W. Perkins Jr., “Effects of

Exemplification in News Reports on the Perception of Social Issues,” Journalism & Mass
Communication Quarterly 73 (1996).
22
 Perry and Gonzenbach further tested the exemplification effect with

television coverage of a controversial topic in the United States—institution-

sanctioned prayer in public schools. As they hypothesized, the exemplification

effect was found even for that emotional political issue. More importantly, the

data showed a positive correlation between exemplar condition and change of

opinion.48 Significantly, researchers found that exemplification alters not

only the perception of public opinion, but also overwhelms prior beliefs—even

for an issue as heated as the role of religion in public education in the United

States, where there is a strong but contested tradition of church and state

separation.

Emotion in Exemplars

Aust and Zillmann conducted a television-specific study to see if the

emotional content of exemplars has any effect on persuasiveness.49 They

utilized three manipulated conditions in stories about handgun violence and

food poisoning: “highly emotional” exemplars, “unemotional” exemplars and

“without-victim” exemplars. In the “highly emotional” condition, paid actors

“spoke in a choked fashion, paused to control their apparent emotional

disturbance, fought back tears, cried, and wiped off their tears.”50 In the

“unemotional” condition the actors spoke in a calm, collected manner. The

48 Stephen Perry and William Gonzenbach, “Effects of News Exemplification Extended:

Considerations of Controversiality and Perceived Future Opinion,” Journal of Broadcasting

and Electronic Media 41 (1997).
49 Charles F. Aust and Dolf Zillmann, “Effects of Victim Exemplification in Television News

on Viewer Perception of Social Issues,” Journalism & Mass Communication Quarterly 73

(1996).
50 Aust and Zillmann, 793.

23
“without-victim” condition lacked personalized exemplification altogether.

The emotional content produced a significant difference in the subjects’

judgments. Those in the highly emotional condition rated the threat of food

poisoning and handgun violence the highest, followed in succession by the

unemotional and without-exemplar conditions. In addition, respondents in

the highly emotional condition rated their own personal risk as being higher,

an effect that did not appear in the other two conditions. The results suggest

sensational exemplars are more persuasive and memorable to news

consumers.

Quotes and Paraphrases

Gibson and Zillmann sought to ascertain the power of direct quotes in

print journalism.51 Researchers once again manipulated an article about

family farming in the United States. That time, the story placed rich and

poor farmers in ideological opposition. In the “impoverished” exemplars,

farmers blamed banks for their financial woes; in contrast, the rich farmers

praised the financial community for its reliable support. In one condition, the

plight of the poor farmers was presented in direct quotations, while the

testimony of the rich was paraphrased. In a second condition this

arrangement was reversed. The control condition article contained statistics

and other information about family farms, but lacked exemplars.

51Rhonda Gibson and Dolf Zillmann, “Effects of Citation in Exemplifying Testimony on Issue
Perception,” Journalism & Mass Communication Quarterly 75 (1998).
24
 Subjects who read the poor-farm testimony in direct quotes attributed

more fault to the nation’s banking system and offered the highest estimate of

failing farms. The effect was not as strong in the “rich-farm” quotation

condition; researchers suspected that outcome was caused by a failure to

rotate exemplars to different positions within the manipulated stories.

Nonetheless, the study suggested direct quotes are a powerful form of print

exemplification. Contrary to expectations, ratings of story quality, fairness,

and balance did not differ between conditions. “The findings lend no support

to the contention that readers are capable of detecting bias in reporting,”

Gibson and Zillmann observed.52

Social Similarity

Another consideration in exemplification theory is the degree of socio-

economic similarity between audience members and exemplars. Brosius

tested the role of social similarity in exemplification in a study in the late

1990s in Germany.53 Respondents were given four newspaper articles dealing

with a proposed abolition of public phones, a possible traffic ban, tele-

shopping and the quality of the local hard cider. Brosius hypothesized that

exemplars that resembled the respondents—in this case, college-aged

students—would have a stronger effect on judgment. To test that hypothesis,

52 Gibson and Zillmann, “Effects of Citation in Exemplifying Testimony on Issue Perception,”

174.
53 Hans-Bernd Brosius, “Research Note: The Influence of Exemplars on Recipients’

Judgments,” European Journal of Communication 14 (1999).

25
two versions of each article were created, one describing the speakers in the

articles as “students,” and another describing them as “pensioners.”

Brosius found that group membership did not cause differences in the

data. Social similarity had no effect on respondents’ judgments. The data

suggested that “for issues that are mostly a matter of taste, exemplars appear

to posses the power to change opinions; for issues that touch existing

political, economic or social attitudes, the effect is weaker and appears only

as a tendency.”54 Brosius based that claim on the finding that exemplar

distribution influenced subjects’ personal opinions about cider quality, but

not about the three social issues.

Measuring Exemplars in Content Analysis

So far in this literature review, the focus has been on experimental

media-effects studies of exemplification. There appears to be little published

research of exemplification using other methods of analysis. In particular,

there appears to be a large gap in the literature regarding the actual use of

exemplars in media messages, both in terms of frequency (as would be

determined by quantitative content analysis of news content) and in terms of

possible cultural influence (via some type of cultural/critical inquiry).

Analyzing exemplar use in the media is a critical step toward understanding

the full implications of the experimental research, but the theory has rarely

been examined outside the confines of experiments.

54 Brosius, “Research Note,” 220.

26
 The only published content analysis of exemplars found by this author

was a study by Daschmann and Brosius of television news magazines aired in

Germany.55 That study did find prevalent exemplar usage in the media,

although the application of those findings to American media is uncertain.

The study was an invaluable guide and helped clarify some of the difficult

methodological issues encountered in this thesis.

Daschmann and Brosius studied news magazine shows televised in

Germany to try to gauge the use of exemplars by television journalists. The

analysis considered 63.5 hours of footage drawn from five consecutive weeks

of coverage. From that footage, 806 individual stories were identified for

coding. The stories were then separated into two categories: generalizing

portrayals and isolated incidents. Of the 806 stories recorded for coding, a

total of 474, or 58.8 percent, were classified as generalizing portrayals.

Because those stories made a connection between the subject matter and

broader trends in society, they were considered pertinent to the theory of

exemplification and suitable for an analysis of exemplars.56

All of the stories recorded for analysis were coded by topic so

researchers could determine which kinds of stories were most likely to be

presented as generalizing portrayals. The topic categories were: politics,

business, cultural, sports, society, crime, human interest, accidents/disasters,

55 Gregor Daschmann and Hans-Bernd Brosius, “Can a Single Incident Create an Issue?

Exemplars in German Television Magazine Shows,” Journalism & Mass Communication

Quarterly 1 (1999).
56 Daschmann and Brosius 38, 40.

27
health, and environment. Researchers found a significant difference between

the categories when it came to exemplification. Human interest, sports and

cultural stories were the least likely to be generalized, at 22.1 percent, 25

percent and 27.6 percent, respectively. For all of the other topics, a

generalized approach was the norm.57

Researchers took a close look at the specificity and validity of the base-

rate information cited in the generalized portrayals. They found that less

than half (45 percent) of the reports presented concrete base-rate information

and only 10 percent cited an external source when making generalized

statements.58 That finding suggests journalists may sometimes fall prey to

dubious inductive reasoning in their reporting. Furthermore, more than one-

third of the generalized portrayals presented statistics without giving the

source of that information.

Daschmann and Brosius catalogued the stylistic devices used in the

generalized reports. Those devices included exemplars, expert interviews,

inserted film and documents, and on-screen appearances by the reporter.

Only four of the 474 reports were without one of those stylistic devices. The

average story contained more than seven exemplars, and 94.5 percent of the

stories contained at least one.59 Expert interviews appeared in 57 percent of

57 Ibid., 38-41.
58 Ibid., 42-43.
59 Daschmann and Brosius, 46.

28
the stories; inserted film and documents were seen in 55 percent; and on-

screen commentaries by reporters were found in 15 percent of the stories.

The study found that German television journalists overwhelmingly

used exemplars to bolster the chosen story focus, as more than 96 percent of

the exemplars analyzed supported the storyline. Only 9.4 percent of the

stories contained exemplars that were contradictory to the generalized

account. “[The exemplars] are obviously not intended to document the

plurality of opinion,” Daschmann and Brosius observed.60

The study supported the notion that exemplification is a common

practice on television news programs in Germany. Daschmann and Brosius

also concluded more generally that exemplars are “a popular journalistic

device,” and are used “as a service to the viewers” to facilitate easier

comprehension.61

Given its unique place in the literature, the format of Brosius and

Daschmann’s study was used as a general roadmap for the content analysis

formulated in this study. Although the goal of analyzing print newspapers,

rather than television, required slightly different coding definitions, the

overall structure of the method translated easily. The study was also used to

help devise research questions, which will be outlined in Chapter 4.

60 Ibid., 46-47.
61 Ibid., 47.
29
Chapter 4. Research Questions

This study was undertaken to measure the use of exemplars in six

major American newspapers. Because there were no comparable studies to

build upon, this study used basic, exploratory questions intended to yield

descriptive data upon which future studies could be advanced.

As previously stated, the relationship between exemplification

research and everyday journalism has not been fully explored. Brosius and

Daschmann showed that exemplars were common in German television news

magazines, but other media formats have not been studied, according to a

review of databases of published research.62 Without even descriptive data

about the use of exemplars in the news, it is impossible to fully assess the

implications of the theory for journalists and media consumers alike. Thus,

the first research question posed by this study is:

R1 How often are exemplars employed in major American

newspapers?

That question reflects the dominance of large, metropolitan

newspapers in establishing professional standards and practices in American

journalism. It does not suggest, however, that the practices of those

62The author performed a full-text search for the terms “exemplar” and “exemplification” in
the following research databases: Communication and Mass Media Complete,
Communication Abstracts, ComAbstracts, Eric, and LexisNexis Academic. The citations from
all the studies in the literature section were also reviewed.
30
newspapers can be generalized to all U.S. newspapers, the vast majority of

which are small-circulation.

Lauterer analyzed newspaper circulation figures from Editor &

Publisher in 2004 and found that 97.7 percent of the listed American

newspapers had a circulation below 50,000, giving so-called “small papers” a

total audience of 108.9 million people. In contrast, Lauterer found “large

papers” had a total audience of 38.2 million.63 A recent study by Reader

suggested there may be significant differences in the way journalism is

conceptualized and practiced at large and small newspapers, specifically

when it comes to matters of professional ethics.64 Given the relative

dominance of small newspapers in America, it is possible a sample of

metropolitan papers, such as the one proposed here, captures only one subset

of journalistic practice.

Following the lead of Brosius and Daschmann, this study will look at

exemplification from a topical point of view, guided by the second research

question:

R2 Which story topics covered in major metropolitan American

newspapers most often use exemplification?

63 Jock Lauterer, Community Journalism: Relentlessly Local (Chapel Hill, North Carolina:

University of North Carolina Press, 2006): 5-7.

64 Bill Reader, “Distinctions That Matter: Ethical Differences at Large and Small

Newspapers,” Journalism & Mass Communication Quarterly 4 (2006).

31
 That question will test whether exemplification is a universal practice

within the sample or whether it is employed only for certain topics (the

specific topic categories will be outlined in the next chapter).

To gauge whether exemplar use in the sample may also be related to

the types of information sources journalists use, this study poses the

following question:

R3 What sources do large U.S. newspapers exemplify most often?

The results from experimental studies suggest exemplars’ impact on

human perception is increased when similar exemplars are grouped together

within a news story. That finding is what led researchers to create three

categories—selective, contradictory and blended—to describe an exemplar’s

relationship to the focus of a story. If a news story presents a hypothetical

“Statement A” as true, but then presents a litany of exemplars that are in

opposition to “Statement A,” exemplification research predicts the news

audience will largely discount “Statement A” as well. Therefore it is critically

important that exemplars not be viewed in isolation, but rather in relation to

the dominant viewpoint established within a story. That is why the scarcity

of contradictory exemplars found in the content analysis by Brosius and

Daschmann is noteworthy, for it suggests exemplars are usually selected to

bolster specific assertions about reality. Therefore, the fourth question for

this study is:

32
 R4 To what degree does story exemplification in major U.S.

newspapers follow the story focus?

Despite evidence from Zillmann, Brosius, and others that the precision

of base-rate information does not affect the persuasiveness of exemplars, this

study will look at the use of statistical evidence in six major U.S. newspapers

based on the following question:

R5 Do major U.S. newspapers typically use high- or low-precision

statistics in newspaper stories?

The second half of this study employed a cultural/critical studies

approach to examine selected articles from the content analysis sample.

Those case studies explored questions such as: What are the possible

interpretations of exemplars from a socio-cultural perspective? Is it possible

to interpret an exemplar incorrectly? How do exemplars reflect and shape

social reality? And what role do cultural codes play in the interpretation of

exemplars?

33
Chapter 5. Method

This study used a constructed-week sample of stories published in

2003 by six major U.S. newspapers. The stories were drawn from the

following major U.S. newspapers: Los Angeles Times (L.A. edition), The

Washington Post, Chicago Tribune, The (Baltimore) Sun, The New York

Times and the Washington Times. With the exception of the Washington

Times, all of those papers have a reputation for professional, impartial

journalism that gives them tremendous influence within the news industry.

The Washington Times is not as large or as influential as the other sample

papers, but it produces original content in a large metropolitan region, which

makes it similar enough to warrant examination.

A sample of two constructed weeks was employed in the content

analysis, with two newspapers selected for each of the Monday through

Sunday editions.65 A computer script was used to pick the random dates for

the sample, based on a spreadsheet that included every calendar date in

2003, organized into seven columns by days of the week. All weekly editions

published in 2003 had an equal chance of being pulled for the sample.

Coding was limited to the front page of each newspaper and to the

front of the local news/metro section. That limited the sample to what was

65 Two constructed weeks was found to be an adequate sampling method for a year’s worth of

newspapers by Daniel Riffe, Charles F. Aust and Stephen Lacy. See “The Effectiveness of
Random, Consecutive Day and Constructed Week Samples in Newspaper Content Analysis,”
Journalism Quarterly 1 (1993).
34
presumably the paper’s strongest news content on that particular day (as

opposed to features, sports, or opinions, all of which deviate from the norm of

objectivity and “straight” reporting that has been the traditional concern of

exemplification research). In the case of front-page articles that jump to later

pages, coders followed the article to the indicated page and continued coding

to the end of each article. Two coding protocols, included in Appendix B, were

used to analyze articles and exemplars in the analysis. The first protocol was

used to collect data at the article level, while the other was employed at the

level of single exemplars found within individual articles.

The first criterion of the coding protocol was the use of statistical

information in the articles. News stories that did not include references to

statistical data were excluded entirely from the coding process. Furthermore,

to warrant inclusion in the sample, the statistical data in the story had to be

cited to an external source. Thus, articles that offered generalized statements

such as “Statistics seem to show...” or “Recent surveys support the claim...”

were included, while articles that contained unattributed opinions were not.

Similarly, statistical information that was presented in quotes from a human

source did not qualify the story for coding unless the person in turn offered a

source for his or her statistical claims. That rule equally applied to experts,

public officials and scholars. The study was purposely limited by that

criterion in the hopes of a obtaining a data set based on articles that closely

35
resembled the narratives used in exemplification experiments, in which

empirical claims are a prominent research consideration.

The second coding criterion was a trend-story orientation. To warrant

inclusion in the sample, the story had to focus on the appearance or

prevalence of a continuing phenomenon or social issue (i.e., crime rates,

economic trends, policy debates, etc.). Stories that focused on an isolated

incident were uniformly excluded from coding; in their content analysis of

exemplars, Daschmann and Brosius classified such stories as generalized

portrayals, and their definition of that category is instructive here. In such

stories, they stated, “the reporter is suggesting or claiming that the story

deals with a topic that occurs repeatedly in everyday life (and is in some form

quantifiable) and that can serve as an example of a general trend (no matter

how strong this trend may be).”66 The quantification of the frequency of

“exemplar” news articles thus provided data in response to Research

Question 1.67

If an article met the standard of being “trend” articles that use

attributed statistics, it was ready for the next stage of coding. Coders were

instructed to record the date and source of the article. Next, they indicated

the story’s subject matter from a list of 14 general topics. Those categories,

along with the coding explanations, were adapted from Stempel’s

66 Daschmann and Brosius, 38.

67 “How often are exemplars employed in major American newspapers?”
36
foundational “Gatekeeping” study of 1985.68 (A full explanation of the

category definitions can be found in the protocol, which is included in

Appendix A.) In cases in which an article seemed to match more than one

category, coders were permitted to make multiple selections, although they

were to indicate the best-fitting category with an asterisk. Those data were

collected to answer Research Question 2.69

Coders then recorded the exact statistic(s) in the story, along with a

brief explanation of the implied meaning of the statistic(s), if necessary. If an

article contained competing statistics from different sources, each was

included on a separate line on the coding sheet.

After the summary of statistics, coders indicated the precision of the

given statistical information. Two precision categories, high and low, were

employed based on the use of similar categories in numerous exemplification

studies.70 Statistics that contained specific counts, percentages, ratios and

estimates were coded as high precision. In contrast, low-precision statistics

were vague, generalized statements such as “more and more” or “a strong

majority.” Those low-precision cases cited statistics without offering specific,

quantitative proof, relying instead on comparative language for emphasis. In

cases in which both high and low precision statements were present, coders

68 Guido H. Stempel III, “Gatekeeping: The Mix of Topics and the Selection of Stories,”
Journalism Quarterly 62 (1985).
69 “Which story topics covered in major metropolitan American newspapers most often use

exemplification?”
70 See Gibson and Zillmann, “Exaggerated Versus Representative Exemplification in News

Reports,” and Zillmann et. al, “Effects of Exemplification in News Reports on the Perception
of Social Issues.”
37
selected a third category, “mixed.” Those data were collected to answer

Research Question 5.71

The next task in the article-level coding process was the counting of

exemplars. That task was guided by the simple idea that exemplars are

merely examples—illustrations, aides—of a larger phenomenon. Zillmann

and his colleagues defined exemplars as “case descriptions or specifications of

singular incidents that fall within the realm of a particular social

phenomenon and that exhibit the pertinent properties of that phenomenon to

some degree.”72 However, exemplars need not be a direct representation of a

category. Quotes and paraphrases are a form of exemplification because they

relate concrete information from the social world and therefore serve an

exemplifying function. Similarly, reporters often include descriptions that

exemplify the topic at hand (e.g., a description of a worn-down school building

in an article about education costs) from a first-person perspective. Those

common forms of exemplification—quotations, summaries and descriptions—

were all considered exemplars in this analysis.

The article-level coding protocol specified that each counted exemplar

be “distinct.” However, the source or the content of the exemplar need not be

unique. Rather, distinct exemplars were sometimes the same person or

example, found as separate entries in the text of the article. An expert, for

71 “Do major U.S. newspapers typically use high or low precision statistics in newspaper

stories?”
72 Zillmann et. al, “Effects of Exemplification in News Reports on the Perception of Social

Issues,” 427.
38
example, who was quoted and discussed for three straight paragraphs was

counted as one distinct exemplar. If that same expert was quoted again later

in the article, the second quote was counted as a separate exemplar.

Essentially, so long as an exemplar remained uninterrupted in the narrative,

it was counted as one. The counting for every article was executed in that

fashion, with one exemplar added to the count each time a distinct one was

encountered. These data were collected also to answer Research Question 1.73

It could be reasonably argued that the length of exemplars should be

given weight or influence in the counting procedure. Under the first coding

system, a 1,500-word article with one long, unbroken exemplar would be

counted the same as a 200-word story with one short exemplar. That is

problematic because one would expect longer exemplars to exert more

influence than shorter ones; thus it could be argued that a coding protocol

that did not take exemplar length into account would be an inadequate

measurement system.

However, the influence of length in exemplification is an open question

in the literature. In one recent exemplification experiment, Brosius found

that the number of sources did not influence recipients’ judgments.74 An

article with one quoted source was as convincing as an article with five. The

amount of information (length) was the same between conditions, which

points to a possible length effect—but that is merely a tentative inference.

73 “How often are exemplars employed in major American newspapers?”

74 Brosius, “Exemplars in the News: A Theory of the Effects of Political Communication,” 184.
39
Lacking a definite empirical answer, the present research considered

exemplars as both distinct units and as strings of text.

For purposes of comparison and discussion, exemplar ratios were coded

two separate ways in the article-level protocol. In one section, the coding was

based on article focus; coders simply counted the number of exemplars that

supported the discernible “take” of the article. By support, it is simply meant

that the exemplar was an example of the phenomenon/risk/social

problem/opinion described. For example, if an article stated, “1 percent of the

population is susceptible to rapid, premature aging,” all exemplars of people

suffering from the ailment were counted as supportive exemplars.

Consequently, all exemplars of doctors or experts talking about the ailment

were also counted as supportive exemplars. Only exemplars that downplayed

the risk, in this instance, were not counted as supportive. That approach was

apparently used in the content analysis cited by Zillmann and Brosius, in

which “story focus” was the criterion for exemplar coding.75 The number of

exemplars that were contradictory and neutral to the article focus also was

recorded. A more detailed explanation of those categories can be found in the

codebook in Appendix A. Those data were used, in part, to answer Research

Question 4.76

75Zillmann and Brosius, 20.

76“To what degree does story exemplification in major U.S. newspapers follow the story
focus?”
40
 Next, coders indicated whether the exemplification in the article was

selective or blended. (The definitions of those categories can be found in a

more detailed form in Appendix A.) Generally stated, any article that

contained an exemplar contradictory to the article focus was categorized as

blended. These data were also collected in response to Research Question 4.77

After coding at the article-level, the individual articles were analyzed a

second time, with data collected regarding each individual exemplar. As

before, exemplars were counted as distinct, unbroken units in the text. First,

coders recorded the date and publication in which the exemplar was

presented. They then counted and recorded the number of sentences in the

article in question. Finally, to answer Research Question 3,78 coders recorded

the information sources that served as the exemplars. The possible

information-source categories were as follows: quote of expert/academic;

paraphrase of expert/academic; quote of public official; paraphrase of public

official; quote of resident; paraphrase of resident; anecdote; hypothetical; and

“other.” (A detailed definition of the categories can be found in Appendix A.)

Specifying the use of direct quotation is important in light of the study

by Gibson and Zillmann, discussed earlier in the literature review, which

77 “To what degree does story exemplification in major U.S. newspapers follow the story

focus?”
78 “What sources do large U.S. newspapers exemplify most often?”

41
suggested quotations can under some conditions make exemplars more

persuasive.79

Protocols Reliability Test

An intercoder reliability test of the coding protocols was performed

using 10 articles culled from the five newspapers sampled for the content

analysis. Those 10 articles, which all met the basic trend-with-statistics

criteria of the coding protocol, came from newspapers published in 2003. The

reliability test with those articles produced a simple agreement of 81.4

percent between two coders, reflecting agreement in 202 out of 248 total

cases. Riffe, Lacy and Fico noted 80 percent is usually considered the baseline

level for a simple agreement test, which puts the reliability test for this study

barely above the acceptable threshold.80 However, Riffe, Lacy and Fico also

said new, untested coding methods—such as the exemplification protocols

devised for this study—should be given more leeway during reliability tests,

so the simple agreement results generated by the exemplar protocols were

deemed sufficient for moving forward.81

79Gibson and Zillmann, “Effects of Citation.”

80 Daniel Riffe, Stephen Lacy and Frederick Fico, Analyzing Media Messages: Using
Quantitative Content Analysis in Research (Mahwah, New Jersey.: Lawrence Erlbaum,
1998): 128.
81 Riffe, Lacy and Fico, 131.

42
Chapter 6. Results

This section will outline the findings of the content analysis, which

suggest exemplification was not common in the sample of six major American

newspapers in 2003. A discussion section following the results will look for

methodological limitations that may have affected the results. The issue of

isolated-incident coverage also will be explored.

R1 How often are exemplars employed in major American

newspapers?

A random sample of two constructed weeks from 2003 yielded 20 trend

stories for coding. Exemplified content was most prevalent in the Los Angeles

Times and The New York Times.

Table 6.1: Number of Articles (n) Included in Coding Process.

Newspaper Total n

Los Angeles Times 8

The New York Times 5

Washington Times 3

The Sun 2

Washington Post 1

Chicago Tribune 1

43
 Twelve of the exemplified articles, or 60 percent, came from the front

page of one of the newspapers. The other eight came from the front of a metro

section. The 20 coded articles produced 182 exemplars for coding, for an

average of 9.1 exemplars per story. No story contained fewer than three

exemplars, and 14 of the articles contained seven or more.

Those results seem to suggest, contrary to the assertions of Zillmann

and Brosius,82 that the type of exemplification presented in experimental

research was rare in the sample of major U.S. newspapers analyzed in this

study. To put that figure in perspective: Given that the front and metro

sections of each edition typically had at least five articles, exemplified content

can be estimated at about 0.05 percent of lead-section content in the front

and local sections of those six metropolitan dailies. Given such a small data

set, the descriptive data generated by this analysis cannot be used to draw

any conclusions about exemplification in major U.S. newspapers. However,

given the novelty of the study’s subject matter, the findings may be

instructive for future studies, so a brief summary of the data will be

presented here. A discussion about the relative lack of exemplification in the

sample will follow.

82 Zillmann and Brosius, Exemplification in Communication.

44
Table 6.2: Frequency of Exemplar Totals in Articles.

Total # of
exemplars Frequency

3 3

4 1

5 1

6 1

7 2

8 3

9 1

10 2

11 1

13 2

15 1

19 1

20 1

45
 R2 Which story topics covered in major metropolitan American

newspapers most often use exemplification?

Half of the coded articles dealt with issues of public health and

welfare. Of the 14 possible categories on the coding protocol, only six

appeared in the sample.

Table 6.3: Frequency of Exemplified Topics.

Topic Frequency

Public Health and 10

Welfare

Economic Activity 3

Public Moral Problems 3

Crime 2

Science and 1
Technology

Politics and 1
Government

Total 20

46
 R3 What sources do large U.S. newspapers exemplify most often?

The most widely used exemplar type in the sample was the anecdote,

defined here as the description of an experience from a first or second-hand

perspective, with a total of 69 cases. Anecdotes represented 38.3% percent of

total exemplars, followed by quote of expert/academic at 22.2% percent and

quote of resident at 19.5% percent.

Table 6.4: Frequency of Exemplar Sources.

Source Frequency Percent of total

Anecdote 69 38.3%

Quote of expert/academic 40 22.2%

Quote of resident 35 19.5%

Quote of public official 24 13.3%

Paraphrase of 5 2.8%
expert/academic

Paraphrase of public 4 2.2%

official

Paraphrase of resident 3 1.7%

Total 180*

* Source data was accidentally not recorded for two of the exemplars.

The anecdote was the most common exemplar source in the Los

Angeles Times, The New York Times and The Washington Post. The

47
newspapers with the most exemplified stories, The Los Angeles Times and

The New York Times, represented the majority of total exemplars from

academics and experts in the sample.

Table 6.5: Frequency of Exemplar Sources in Each Newspaper.

Exemplar The Chicago Los The Washington Washington

Source Sun Tribune Angeles New Post Times
Times York
Times

Anecdote - - 37 17 8 7

Quote of 7 2 16 14 1 -
expert/academic

Quote of 3 - 16 3 3 10
resident

Quote of public - - 11 11 1 1
official

Paraphrase of - 2 2 1 - -
expert/academic

Paraphrase of - - 2 2 - -
public official

Paraphrase of - - 1 1 - 1
resident

Total 10 4 85 49 13 19

48
 Table 6.6: Average Number of Sentences Per Exemplar Source Type.

Exemplar Source Average length

Quote of resident 4.11

Anecdote 3.88

Paraphrase of 3.4
expert/academic

Quote of public official 3.2

Quote of expert/academic 2.92

Paraphrase of public 2
official

Paraphrase of resident 2

Direct quotes represented 55 percent of the exemplar total, as

compared to 6.6 percent for paraphrases. The prevalence of direct quotes not

surprising, but noteworthy in light of Gibson and Zillmann’s finding that

“selective direct quotation greatly increases the persuasive power of

exemplars.”83

R4 To what degree does story exemplification in major U.S.

newspapers follow the story focus?

83 Gibson and Zillmann, “Effects of Citation,” 173-174.

49
 The majority of the articles in the sample, 65 percent, contained

blended exemplar distribution. Seven of the blended articles, but only one of

the selective, came from the Los Angeles Times.

Table 6.7: Frequency of Exemplar Distribution.

Distribution Frequency Percentage

of total

Selective 7 35%

Blended 13 65%

Exemplars were most likely to agree with a story’s focus.

Table 6.8: Frequency of Exemplar Foci.

Exemplar Type Total # Percentage

Supportive 98 53.8%

Contradictory 30 16.5%

Neutral 54 29.7%

Total 182

Only 35 percent of the coded articles contained selective

exemplification, which indicates the newspapers often attempted to construct

multi-faceted narratives. However, it is clear that article focus could be a

50
substantial influence on exemplar selection at large newspapers. The

overwhelming majority of exemplars—83.5 percent—were supportive of or

neutral to the story’s focus, which suggests the newspapers did not stray far

from a central organizing principle and rarely seek to undermine it. The data

complement Fico and Soffin’s exploration of story balance, which found that

48 percent of newspaper articles were one-sided in their coverage.84 The data

also supports Zillmann’s musings about focal points in news coverage85 and

validates the findings of the exemplification content analysis performed by

Brosius and Daschmann.

R5 Do major U.S. newspapers typically use high or low precision

statistics in newspaper stories?

The majority of statistics cited by the newspapers were high in

precision.

Table 6.9: Precision of Statistics.

Precision Total # Percentage

High 15 75%

Low 3 15%

Mixed 2 10%

84 Frederick Fico and Stan Soffin, “Fairness and Balance of Selected Newspaper Coverage of

Controversial National, State, and Local Issues,” Journalism & Mass Communication
Quarterly 3 (1995): 626.
85 “As focal story points are usually defined by cases drawn from a minority of newsworthy

cases, all illustrations tend to be consistent with the minority, not the majority, of cases of
interest.” From Zillmann, “Exemplification Theory,” 434.
51
Discussion of Results of Content Analysis

The small number of articles found to be suitable for analysis begs the

question: Why did the coding protocol fail to produce a substantially larger n?

Was the study’s design faulty or is exemplification in trend stories simply

rare in print journalism?

A study by Berger provides some corroborating evidence for the

findings.86 Berger’s research, which focused on the use of quantitative data in

newspaper and television reports, dealt with theoretical and methodological

issues similar to the ones encountered in the present study. Berger’s analysis

also encountered difficulty producing a large (n).

Berger limited his data to accident, crime and health stories that

presented statistical trends. After failing to produce a robust (n) in a pilot

study, Berger settled on a large population for the sample—four weeks of

coverage from four newspapers and six television networks. After considering

2,075 stories, the coding produced 84 stories for analysis—a mere 4 percent of

the total population. That result led Berger to conclude that “news reports

containing quantitative data about trends are relatively rare in both

newspapers and television news.”87 Berger also found that health was the

86 Charles Berger, “Making It Worse Than It Is: Quantitative Depictions of Threatening

Trends in the News,” Journal of Communication 51 (2001).

87 Ibid., 664.

52
most-covered trend story topic, which is consistent with the relative

dominance of public health and welfare stories in this analysis.88

Despite the corroborative nature of the findings, the (n) found by

Berger was significantly larger than the one found by the present study.89

While Berger’s research does lend support to the finding that trend stories

are infrequent in daily news coverage, it does not eliminate the possibility of

methodological flaws.

First, consider the sample. The data for the present analysis came

from six urban newspapers; mid-size and small publications were not

included. A more comprehensive sample, inclusive of varying geographic

regions and circulation numbers, may have produced a larger (n) for analysis.

It is possible that the use of exemplification varies greatly within the

industry. Perhaps small and mid-size papers are more frequent users of the

technique? On the other hand, the sample that was employed may have

simply been two small to produce a robust data set; a content analysis of four

constructed weeks (or perhaps more) would provide a larger population of

stories for analysis, and perhaps more exemplified narratives as well.

Second, because coding was limited to the front pages of the

newspapers and the fronts of the local news/metro section, it is possible that

more exemplified content appeared on the inside pages of the newspapers

88 The study also found that newspaper stories with trend statistics were twice as long as

those without.
89 Though it is worth noting that only 65 of the 84 trend stories he found came from

newspapers.
53
and thus escaped analysis. Perhaps exemplified content was viewed as less

important than the just-breaking stories of the day and ended up in less

prominent placements inside the newspapers? A 1989 study by Bridges

suggested editors rely heavily on timeliness when choosing stories for a

paper’s front page, so there could be some validity to that idea.90 At the very

least, a content analysis should be performed on entire newspapers in a

sample, from beginning to end, before trend exemplification can be

considered an anomaly.

Third, it is possible that the sample was greatly reduced by the

statistical data criterion; a number of articles, otherwise appropriate for

analysis, were excluded from the sample due to a lack of statistical

information. Given the basic tenets of exemplification theory, there is no

reason for trend stories without statistics to be excluded completely from

analysis; in the absence of base-rate information, the exemplifying effect

should remain. However, including trend stories without statistics would not

have produced a dramatic difference in the total (n) in this particular study.

Trend stories were simply not that prevalent in the population.

Finally, the most debatable aspect of the study’s design is the trend

story criterion itself. If that requirement had not been part of the coding

protocol, any news story with sourced statistics would have been included in

the sample and the total number of coded stories would have been

Janet A. Bridges, “News Use on the Front Pages of the American Daily,” Journalism
90

Quarterly 66 (1989).
54
significantly larger. Was the trend-story criterion a critical design flaw? Or

was it a necessary and justifiable limitation?

Exemplification and Isolated Incidents

The trend-story criterion was included in the content analysis so that

the data could be interpreted in the framework of exemplification theory.

Without that sampling limitation, any purported link between the results of

the study and the findings of exemplification research would have been

tenuous at best.

Is an example serving an exemplifying function if it is not explicitly

presented as the representation of a statistical trend? The small (n) produced

by the analysis can be attributed partly to the fact that most of the stories in

the newspapers were about isolated incidents in which there was no text to

exemplify the event at hand. Stories such as: “Man sentenced to jail time”;

“Politicians bicker over budget”; “Police hire 40 new recruits.” In stories such

as those, connections between one incident and a larger population are

usually not made.

The above raises an important theoretical issue. Zillmann has made it

clear that samples from a population are the basic building blocks of

exemplification. “An exemplar is, of course, nothing other than an event

subsumed in a population or subpopulation.”91 In the same essay, however,

Zillmann claimed that exemplification “is essentially independent of the

91 Zillmann, “Exemplification Theory: Judging the Whole by Some of Its Parts,” 78.
55
competition with base-rate information.”92 If the exemplification effect is

distinct from population samples, nearly anything that appears in the news

media could be considered an exemplar—symbolic interactions between story,

population and heuristic processing would be essentially context-free.

In fact, nearly all of the empirical evidence for the exemplification

effect has come from experiments that used trend stories, many of them

hypothetical stories created simply for the experiments. None of the studies

cited in the literature review utilized the isolated-incident format that

pervades news coverage. That rather narrow aspect of the research method

has not been fully explored in exemplification studies. More research is

needed to determine whether the exemplification effect persists when a news

event is presented as an isolated incident.

If trend stories are rare in the media, as the present analysis suggests

for daily newspapers, the theory of exemplification will need a major

theoretical expansion to retain its relevance for print media. In the second

section of this study, some of the possibilities for theoretical expansion—

explored through textual analysis of selected content analysis articles—will

be presented in the form of case studies.

Until the question of population sampling and isolated incidents faces

empirical testing, the trend-story criterion recommends itself as the only

method compatible with the theory’s design. If methodological flaws

92 Zillmann, “Exemplification Theory: Judging the Whole by Some of Its Parts,” 71.
56
hampered the coding protocol for the analysis, the trend-story requirement

was not to blame.

Exemplified Narratives

Finally, it must be considered that exemplification as defined by

Zillmann and colleagues simply might not that common in the news. If

further study of the phenomenon, undertaken with larger sample sizes and

refined coding protocols, fails to turn up meaningful samples of exemplified

narratives in newspapers, exemplification researchers may need to consider

whether it is worthwhile or even appropriate to use such articles as a basis

for experiments.

Furthermore, if studies of other media (e.g. television news programs,

Web site postings, magazine articles) also fail to turn up significant instances

of exemplification, researchers may have to confront the ironic possibility

that the theory is itself wholly founded on a misleading exemplar—a form of

exemplified narrative that rarely appears in the news.

57
Chapter 7. Case Studies: Purpose

The most noteworthy finding of the content analysis was the rarity of

exemplification found in news articles. Having exhausted the discussion of

that particular result, an opportunity presented itself. Although the sample

was small and not generalizable, the exemplified articles collected for the

content analysis were, in turn, suitable for qualitative case study.

Unanswered qualitative questions about exemplification remained, such as:

What does exemplification look like? What forms does it take? How is it

written? How does it organize and reflect the social, political and cultural

landscape in which it was created? The articles gathered for the quantitative

study could be analyzed using a cultural/critical approach to shed some light

on those questions.

Up to this point, the study of exemplification theory per se has

remained firmly rooted in the social-scientific paradigm. Considering the

theory via an approach more reserved for the humanities might expand the

utility of the theory. The following case studies suggest that such an

approach is worth exploration.

Ahead of the case studies, a brief literature review will introduce

theoretical concepts not traditionally associated with exemplification. Those

concepts were incorporated into the case-study discussions, along with the

58
exemplification research already described in Chapter 3. Copies of the

articles studied via case study can be found in Appendix C.

For the case studies, exemplars were analyzed from a cultural

perspective to reveal the wide range of meanings they could suggest to a

likely reader, beyond the simple comparison with a statistical benchmark.

The case studies also employed concepts from semiology to show how

exemplars can naturalize constructed cultural meanings and function as

widely understood myths in society.

59
Chapter 8. Case Studies: Perspectives

When it comes to the form exemplars take in the news, it is important

to recognize the possibility for different styles of presentation. Exemplars can

be packaged into a baseline format unaccompanied by narrative elements,

and they can also be linked, to varying degrees, to larger cultural meanings.

Analyzing the context surrounding an exemplar is critical to discerning the

possible meanings they may suggest to a reader.

Important differences can also exist among individual exemplars.

Because this study concerns itself with newspapers, the pertinent differences

are grounded in the use of language, as the meanings of exemplars are

inevitably shaped by the rhetoric used to construct them. That is not to say,

however, that there is only one “correct” interpretation of an exemplar, or

that the meanings they suggest are the same for every individual. Viewed

through the encoding/decoding model of mass communications devised by

Stuart Hall, it is instead suggested that the meanings of exemplars are

complex and subject to “determinate moments” of active sense-making within

a circuit of discourse.93

Encoding and Decoding Exemplars

On a basic level, exemplification theory is structured around the

premise that there is a measurable causal relationship between media

Stuart Hall, “Encoding/Decoding,” in Culture, Media, Language, eds. Stuart Hall, Dorothy
93

Hobson, Andrew Lowe and Paul Willis (London: Hutchinson, 1980).

60
production and media reception. The theoretical model essentially postulates

the following: When people are exposed to certain stimuli (exemplars), they

will respond in predictable ways (by making mental estimates that

correspond in proportional terms to the exemplars). That effects-driven,

reductive approach has yielded valuable data on the psychological responses

to media patterns, suggesting exemplars alter an individual’s perceptions of

reality in ways that can be observed and measured.

However, that reductive approach is limiting because it assumes a

linear relationship between media content and media reception and fails to

consider the active interpretation of an audience, or of individuals within

that audience. Exemplification’s effects-driven orientation toward the media,

grounded in the transmission model of communication, simplifies both the

meanings suggested by exemplars and the process by which they are

produced and interpreted.

Hall’s model offers a different view of mass communication, grounded

in the idea that “there is no intelligible discourse without the operation of a

code.”94 McQuail defined codes as “systems of meaning whose rules and

conventions are shared by members of a culture,” and the idea is central to

the approach Hall developed.95

94Hall, “Encoding/Decoding,” 170.

95Denis McQuail, McQuail’s Mass Communication Theory, 4th ed. (London: Sage
Publications, 2000): 350.
61
 In Hall’s model, media messages pass through “distinctive moments”

that are linked together but ultimately subject to different systems of

meaning. In the first stage of the model an identifiable media producer, such

as a newspaper, turns a “raw” event from the world into meaningful

discourse (e.g., a written narrative) by encoding it. At the point of encoding

the sense-making structures of the media institution are in “dominance,”

producing a pattern of “preferred meanings” for an audience.96

In the second stage of Hall’s model the meaningful discourse

constructed by the media institution is placed into circulation by being

published or broadcast. The meaningful discourse (i.e., media message) is

subsequently decoded in the third stage. Hall conceptualized decoding as an

active process; rather than simply “receiving” a message, audience members

construct meaning from within the context of their own sense-making

structures and cultural understandings.

Hall argued the moments of encoding and decoding are not always

symmetrical, because audience members decode messages from their own

cultural perspectives. The “preferred meanings” encoded in a message can be

misinterpreted, distorted or missed entirely (in the view of the encoder) due

to a “lack of equivalence” between the sense-making systems of media

producers and audience members.97 Nonetheless, there most be some degree

of equivalence between the sense-making systems of encoders and decoders

96 Hall, “Encoding/Decoding.”
97 Ibid., 169.
62
for an act of communication to take place. Hall argued that encoding has “the

effect of constructing some of the limits and parameters within which

decodings will operate,” and thus confines, but does not determine, the

interpretive acts of an audience.98

Hall identified three positions from which an audience member can

construct meaning during the decoding of a message. In the first position,

which he termed the “dominant-hegemonic position,” audience members

construct meaning from within the same dominant code as the media

producers, thus creating a crude symmetry of understanding between the two

sides. From the dominant-hegemonic position, the constructed cultural

meanings in a message are viewed as natural, transparent and self-evident.

From the second, “negotiated” viewpoint, audience members also

accept the dominant code and construct meaning within it, but reject or

modify some of the codified meanings to bring them in line with their own

“local conditions” (i.e., personal experiences). Finally, from the “oppositional”

position, audience members recognize the dominant code employed in the

discourse and consciously reject it, instead employing an “alternative

framework of reference” for decoding.99

This study will argue Hall’s encoding/decoding model of mass

communication can be applied to qualitative studies of exemplars in the

news. The model’s emphasis on the active interpretation of an audience

98 Hall, “Encoding/Decoding,” 173.

99 Ibid., 175.
63
provided a useful counterpoint to the transmission-model orientation that

dominates exemplification literature, and its system of decoding positions

offered a framework for re-conceptualizing the possible interpretations of

exemplars by an audience. The decoding systems highlighted the capacity of

media consumers to consciously accept or reject, to varying degrees, the

exemplifying function of an exemplar in a media message.

The encoding/decoding model provided a structure for thinking about

the production, transmission and interpretation of media messages, but it

offered little insight into the ways those messages are individually

structured. For help on that front, this study turned to the study of

semiology, a research field that is chiefly concerned with the way social

meanings are made in a culture.

Exemplars as Myth

Perhaps the most applicable translation of exemplification theory from

media-effects research to cultural theory is to consider how single incidents

and oversimplified examples can take on lives of their own and become seen

as more widespread cultural phenomena, akin to what Roland Barthes called

modern “mythologies.”100 Indeed, Barthes alluded to a media connection in

Mythologies when he observed, “the press undertakes every day to

demonstrate that the store of mythical signifiers is inexhaustible.”101

100 Roland Barthes, “Myth Today,” in Mythologies, Annette Lavers, trans. (New York:

Noonday Press, 1972): 109-159.

101 Ibid., 127.

64
The Study of Signs

Before delving into the specifics of Barthes’ work, it is necessary to

introduce a few concepts from the study of semiology. Barthes’ definition of

myth was based, in part, on the work of the Swiss linguist Ferdinand de

Saussure, who proposed studying language and communication as signs.

Saussure defined the sign as the recognizable association of a signifier, or

sound pattern, and a signified, or mental concept. The sign “smile,” for

instance, is composed of the sound pattern produced when the word “smile” is

spoken (the signifier) and the mental concept of a happy, upturned face (the

signified). Saussure defined the interplay between a signifier and signified as

a signification, and argued that the two entities are in practice inseparable,

as there can be no thought (i.e., a signified) without sound (i.e., a signifier).102

Nonetheless, Saussure’s model did not assume a fixed, exclusive relationship

in signs, as a signifier can be associated with more than one signified, and

vice-versa.

Subsequent scholars generalized Saussure’s dyadic model, expanding

the definition of a sign to include other forms of communication. “Signs take

the form of words, images, sounds, odours, flavours, actors or objects,”

Chandler noted.103 A “stop” sign for traffic in the United States, for example,

is a red octagon on a pole (signifier) that is interpreted as an injunction to

bring your vehicle to a complete stop (signified). The “sign” is the associative

102 Daniel Chandler, Semiotics: The Basics, 2nd ed. (New York: Routledge, 2007) 14-28.
103 Chandler, 13.
65
totality of the signifier and the signified—a unified, inseparable and widely

recognized meaning. Within the context of Saussere’s arguments, the stop

sign is a purely human creation, arbitrarily endowed with meaning; a purple

circle on a pole, to take just one hypothetical, could just as easily be a

signifier for the “stop” injunction, if so constructed. The ultimate

arbitrariness of the sign suggests that language does not reflect reality, but

rather constructs and organizes it.

It is important to note that, in semiology, signs are viewed as complex

and typically contain multiple levels of meaning. In semiology, the literal or

“common-sense” meaning of a sign is called the “denotation.” That literal

meaning is the foundation for broader “connotations,” which are meanings

that are more idiosyncratic and socio-cultural in nature.104 Connotations are

created when a sign becomes a signifier, and thus part of a second-order

system of meaning. Champagne, for example, on a denotative level is a sign

composed of the word “champagne” (the signifier) and the concept of a bubbly

white wine (the signified). On a connotative level of meaning, champagne (a

self-contained sign) is reconstituted as a signifier, thus allowing for an

association with new signifieds; possible connotations include, but are not

limited to, concepts such as celebration, glamour, happiness, or wealth.

Through such layering of signs, numerous connotative meanings can be

chained together within the confines of a single sign.

104 Ibid., 138.

66
 On a connotative level, signs are polysemic, or open to multiple

interpretations. A reader of signs constructs meaning from within the context

of available social codes, which are broadly, but not universally, shared in a

culture. In contrast, the denotation of a sign is typically defined as a meaning

that nears cultural consensus.105 That does not mean that the interpreted

connotations of a sign are arbitrary—as they are often widely shared—but

rather that they can never be catalogued in full. The polysemic nature of

signs is very much in line with Hall’s conception of decoding, which also

leaves room for idiosyncratic interpretations.

Myth in Society

Barthes took Saussere’s model of the sign and conceived a system for

deciphering the dominant ideologies of a society. Myth, Barthes argued, is a

peculiar type of trope that naturalizes historical meanings through the

appropriation of existing signs. Barthes called myth a “second-order”

semiological system because it is structured on two levels of meaning. On one

level is the “language-object,” which is a linguistic system (i.e., a sum of

signs) that contains a rich, self-contained history of connotations. Myth is a

second-level “metalanguage” that speaks about a language-object and

subsumes it within a larger, cultural signification.106

In myth, a fully formed linguistic system becomes a signifier associated

with a new signified. On the first level of the semiological order, a linguistic

105 Chandler, 139.

106 Barthes, “Myth Today,” 115.
67
system is a sign replete with denotative and connotative meanings; on the

second, myth level, a linguistic system is a “form” for a mythical signified.

The distinction between “meaning” and “form” is critical to Barthes’ model.

When a linguistic system becomes form in myth, its original meaning is

suppressed and held at a distance. “When [meaning] becomes form, [it] leaves

its contingency behind,” Barthes explained. “It empties itself, it becomes

impoverished, history evaporates, only the letter remains.”107

By putting the original meanings of a linguistic system at a distance,

myth makes room for an association with a second, higher-order concept: the

mythical signified. The mythical signified fills the form, implanting the

linguistic system with a “whole new history.”108 Thus the tri-dimensional

pattern of the sign reconstructs itself on the second level, with a linguistic

system existing as meaning and form, sign and signifier, in the service of a

global signification, which is the myth.

Barthes observed that myth is a unique semiological system in that it

distorts the meaning of a signifier. Recall that on a simple, denotative level, a

signifier is wholly arbitrary; the word “champagne,” to reuse the previous

example, has no natural, intrinsic relationship to a bubbly, alcoholic beverage

(the word actually refers to a particular wine-making region of France), and

thus it is not distorted in the sign. Myth, in contrast, takes a linguistic

signifier (i.e., a sign) rich with connotative meanings and distorts its history

107 Ibid., 117.

108 Ibid., 119.
68
in the service of a mythical concept. That distortion of the signifier, Barthes

argued, is motivated, intentional, and never arbitrary. It is a purposeful

construction for the creation and maintenance of ideology in a society.

Barthes devoted attention to mythical speech because he argued it

naturalizes culturally constructed meanings, thereby “transforming history

into nature.”109 To a reader of myth, “things appear to mean something by

themselves,” and the associative link between a signifier and mythical

signified is viewed as a self-evident fact, rather than as a constructed

reality.110 It is through myth that man-made values, attitudes and beliefs

become widely recognized and accepted in a culture as “the way things are.”

What is at once historical and intentional—the mythical signified—is

transformed into common sense from the perspective of a myth-reader.

Barthes viewed myth as a vehicle for naturalizing the ideology of the

bourgeoisie, although one need not accept that formulation to recognize the

naturalization of meaning that the semiological system represents.

Ultimately, Barthes conceived myth as a vehicle for the dominant ideologies

of a culture, regardless of who may benefit from them.

Myth in the News

Applying Barthes’ concept of myth to Zillmann’s concept of

exemplification seems like a natural marriage of two theories from very

different paradigms, as both concepts deal with the construction of meaning

109 Barthes, “Myth Today,” 129

110 Ibid., 143.
69
based on assumptions derived from limited bits of information. Exemplars in

the news that are widely publicized have the potential to become myths in a

culture. Consider this example: During discussions of high-profile lawsuits in

America, someone (e.g., a pundit on a television news show) will often

mention the infamous “hot coffee case” against McDonald’s.111 The case

involved Stelle Liebeck, a 79-year-old woman who spilled a cup of McDonald’s

coffee on herself and “suffered third degree burns over 6 percent of her body,”

leaving her “partially disabled for up to two years.”112 After the incident,

Liebeck asked McDonald’s for $20,000 in compensation to cover her medical

costs; the company returned with an offer of $800, so Liebeck took

McDonald’s to court seeking millions of dollars.113 Few people remember the

actual facts of the case, which have been overshadowed by the constructed,

mythical meaning that greedy, litigious Americans will do anything to win

large financial settlements from corporations.114

A Barthesian analysis of the McDonald’s hot-coffee exemplar could go

something like this: The case itself (the “sign”) is reported in the news media

with comments that suggest it is a possible “frivolous lawsuit.” Over time, the

“frivolous” comments become codified into a new, second-order mythical

signification. The hot-coffee case, a sign, becomes a signifier associated with a

111 Michael Schudson, “Runaway Jury Awards ... And Other Made-for-Media Myths,”
Columbia Journalism Review 2 (2005).
112 Ibid.
113 Schudson.
114 Ibid.

70
second-order signified, which is the concept of the “frivolous lawsuit” in the

United States. Thus the constructed myth of the McDonald’s hot-coffee

lawsuit becomes a widely understood example of “frivolous lawsuits” in

reports about various issues, such as tort reform, corporate liability reforms,

safety labels on common products, etc.

The mythical construction of the hot-coffee exemplar illustrates one of

Barthes’ key points: that myth impoverishes a meaning (i.e., a sign) and

obscures its history. The “frivolous” construction of the hot-coffee lawsuit

persists as a “self-evident” interpretation of the experience of Stelle Liebeck,

despite the complexity of the case. As a meaning, the hot-coffee case is a self-

contained linguistic system with a complex history. As a mythical form, the

complex hot-coffee case has been distorted into a simple caricature, such that

only the basic outlines of the incident (i.e., woman spills coffee on self, then

sues the coffee-seller for millions) are brought to the fore. As Barthes

observed, “The meaning will be for the form like an instantaneous reserve of

history, a tamed richness, which it is possible to call and dismiss in a sort of

rapid alternation.”115 Thus Liebeck’s personal tale of pain and suffering is

obscured and appropriated in the service of a constructed, ideological

signification: that Americans are too quick to turn to the legal system to solve

their problems, and often have spurious reasons for doing so.

115 Barthes, “Myth Today,” 118.

71
 Barthes observed that the concept of a myth (i.e., the signified) could

be associated with a potentially unlimited number of signifiers. It is through

such repetition that myth draws its power. The construction of the hot-coffee

case, for example, could represent but one example of a lawsuit that is

routinely constructed as “frivolous” in the news. All of that suggests that the

journalists who collect, arrange and propagate the news may be at the

forefront of mythmaking in culture.

Summary

Hall’s encoding/decoding model and Barthes’ system of analyzing

mythical speech are useful tools with which to analyze exemplars in

newspapers. Hall’s model suggests exemplars are open to multiple

interpretations, and that audience members actively construct their meaning

from varying social contexts. Hall’s model also suggests that news consumers

can decode narratives from an oppositional viewpoint, and thus reject,

outright, the exemplifying function of an exemplar. The following case studies

employed Hall’s analytical framework to consider some of the possible

interpretations of the exemplars in various news articles that were

scrutinized for the content analysis portion of this study. The case studies

also contrasted those possible interpretations with the transmission-model

assumptions about exemplars found in the experimental literature.

Saussure’s system of semiotic analysis, coupled with Barthes’ writings

on myth, provided an additional theoretical framework with which to analyze

72
the way individual exemplars may be structured to create meaning that is

“decoded” by the recipients of a message. Thus, the following case studies also

deconstructed the exemplars as possible instances of myth-construction by

the press.

73
Chapter 9. Case Studies

Case #1: Exemplification by the Numbers

The first case study is based on a report about child poverty from the

Sept. 23, 2003, edition of The Sun.116 The article could be categorized as a

“baseline” example of exemplification in a newspaper—to borrow a term from

the literature—because it covered complex statistical evidence without the

aid of explicitly illustrative anecdotes. The first paragraph established the

report’s by-the-numbers approach:

The number of children living in severely distressed neighborhoods

in Maryland and nationwide increased substantially in the 1990s
despite the decade’s booming economy, according to a study
released yesterday.117

The report examined the statistical indicators used to tally the number

of “distressed neighborhoods” and contrasted the study’s findings with other

poverty studies from the same time period. The article focused on statistics

from Maryland, and particularly Baltimore, in the second half, and reviewed

the ethnic disparities that were found for distressed neighborhood residency.

The article also compared the local statistics on distressed neighborhoods to

the rates found in other states and cities.

Under the coding guidelines employed in the content analysis, the

distressed neighborhoods article was found to contain three exemplars. Those

exemplars all came in the form of direct quotations related to the study’s
116 Eric Siegel, “Groups’ Study on Kids is Grim,” The Sun, 23 Sept. 2003.
117 Ibid.
74
findings. For example, one of the sources (an author of the study mentioned

in the article) was quoted trying to make sense of the trend:

“We were surprised to see that kids in distressed neighborhoods

increased at all,” said Mark Mather, a policy analyst with the
Population Reference Bureau and a co-author of the study. “I think
it says there’s some persistent problems, particularly in inner-city
areas. A lot of families are isolated from what goes on in the
economy outside their neighborhood.”118

Although there are cultural meanings suggested by the phrasing of

Mather’s statement, the exemplar itself is mostly explanatory, and not

illustrative, in nature. The other two exemplars in the article were of a

similar construction.

Within the context of exemplification research, the child-poverty

article’s lack of illustrative exemplars can be viewed as a key characteristic.

The article was not localized with exemplars of “distressed” Baltimore

neighborhoods, nor were children in difficult circumstances profiled as “living

proof” of the statistical findings. The article demonstrates that information

about social trends can be presented in newspapers without the exemplified

narrative elements that Zillmann and others have found so troubling.

Also of note is the article’s thorough presentation of statistical

evidence. Research has shown that high-precision base-rate information

alone does not mitigate the exemplification effect,119 but Baesler and Burgoon

have demonstrated that heavily emphasized base-rate information can, in

Siegel.
118

Gibson and Zillmann, “Exaggerated Versus Representative Exemplification in News

119

Reports.”
75
some instances, lessen the persuasive power of exemplars.120 With

manipulated, illustrative exemplars added, the distressed neighborhoods

story could serve as an interesting experimental test case of the limits of the

exemplification effect.

The “baseline” example of the distressed neighborhoods report stood in

stark contrast to the three other case study articles, which all used

illustrative exemplars open to broad interpretation.

Case #2: Where Did the Crabs Go?

This case study considers an article from the July 22, 2003, edition of

the Washington Times that explored the sudden disappearance of blue crabs

from Maryland’s Eastern Shore. The statistical basis for the story was, in this

case, complementary to the overall story focus: according to the Maryland

Department of Natural Resources, the area’s blue crab harvest had dropped

40 percent from the average harvest.121 In contrast to the distressed-

neighborhoods report, the crabs story did not examine the relevant statistical

evidence in detail, instead making it a secondary element in the larger

cultural narrative.

After a lead paragraph that introduced the subject matter, the article

related an anecdote of one crab-picking house that was forced to shut its

doors after 53 years in business:

120 E. James Baesler and Judee K. Burgoon, “The Temporal Effects of Story and Statistical
Evidence on Belief Change,” Communication Research 5 (1994).
121 Gretchen Parker, “Eastern Shore Crabs Hard to Come By,” Washington Times, 22 July

2003.
76
 James Dodson, owner of the waterfront Original Captain’s Galley
restaurant, this year closed his crab-picking house. For 53 years,
women gathered at the plant to harvest meat from hundreds of
bushels of Maryland blue crabs a day.

This year it wasn’t worth Mr. Dodson’s time.

“There is just a scarcity of crabs. It’s hard to get domestic crabmeat

now,” said Mr. Dodson. .... “Whether it’s weather or overfishing or
the regulations, it’s a combination of everything.”122

In a traditional exemplification study, that exemplar would typically

be viewed as “supportive” of the story’s contention that Maryland’s blue crabs

were disappearing. Researchers would treat the exemplar as a

straightforward, unambiguous illustration of the crab’s decline.

Applying Hall’s model of encoding/decoding provides a different

perspective on the failed business exemplar. When readers are viewed as

active decoders of media messages, the “supportive” construction of an

exemplar cannot be taken for granted as a universally shared interpretation.

A contemporary reader could have adopted a negotiated decoding position, for

example, and interpreted the exemplar as a factual anecdote that

exaggerated the reported decline in blue crabs. Or a reader could have taken

an oppositional decoding position and rejected the construction of the

exemplar outright, perhaps viewing the crab house’s failure as a simple case

of mismanagement by the owner. In any case, readers did not have to accept

the encoded message of the exemplar, as they were fully capable of

constructing a limitless number of different interpretations that accepted and

122 Parker.
77
rejected, to various degrees, the purported exemplarity of the crab house

anecdote.

One of the exemplars in the crab house article can also be seen as a

possible instance of myth in the news. Near the middle of the report, the

pricing effects of the blue crab’s decline were considered:

The dearth of crabs quickly translates into higher prices for

customers.

People are looking at us like, ‘Are you dumb?’ for putting such an
expensive price on them,” said Bob Shorb, 25-year manager of Sea
Pride Crab House in Baltimore. He called prices for crabs this year
“ridiculous.”123

At first glance, that exemplar would seem to be another instance of an

anecdote constructed as “supportive” of the purported decline in blue crabs.

However, it can be argued that there was an additional level of meaning

encoded into the exemplar that was mythical in nature. The exemplar seems

to have been built upon a widely understood ideological construct in the

United States: the theory of supply-and-demand. The first sentence made

that association explicit, by stating a causal relationship between crab

scarcity and high prices. The supply-and-demand construct was, in effect,

validated by the crab house manager, who attributed his high prices to

supply forces beyond his control.

That’s not to say the exemplar is a perfect fit for Barthes’ mythical

model; the crab house manager’s story was not given “a whole new history” in

123 Parker.
78
the service of a second-order signified, nor was its meaning distorted within a

global signification. Nonetheless, the exemplar displayed one of the defining

qualities of myth. Through the association with the exemplar (the signifier),

the supply-and-demand theory (the signified) was arguably presented as a

self-evident, “natural” way of understanding the world. Thus, through its

reinforcement of an historical, ideological point of view, the exemplar can be

viewed as a sign that supported and contributed to a larger mythmaking

process in the United States.

The supply-and-demand exemplar also highlights the importance of

what Hall termed “equivalence” in decoding. A hypothetical reader from

another country—particularly one without a capitalist background—could

have missed the mythical meaning the exemplar seemed to suggest, or simply

found it puzzling. When there is a general lack of equivalence between the

knowledge and social understandings of an encoder and decoder, the

dominant/hegemonic meanings suggested by a media message can be lost, as

the naturalized meanings of mythical speech are always rooted in a distinct

time and place.

The article also explored the causes of the blue crab’s population

decline:

No one can agree on what’s causing the downslide. Crabbers blame

farmers for putting too much nutrient runoff into Chesapeake Bay.
Farmers blame watermen for overharvesting.

79
 “You can ask 20 different people. They’ll have 20 different
opinions,” said Bill Cox of Fresh Catch in Crisfield. “But it’s
overfishing if you ask me.”124

Bill Cox seems to have exemplified the town’s confusion in the report;

his statement implied that the disappearance of the blue crab was not only

unprecedented, but also that its cause was perhaps unknowable. At no point

did the article present the opinion of a scientist or researcher who had

studied the crabs’ disappearance, so the statement that “no one could agree”

on the cause of the trend went unchallenged in the text. The newspaper’s

readers, however, may have interpreted the possible meaning of the exemplar

quite differently.

Given that The Sun is located in Baltimore—a major U.S. seaport—it

is likely that some of the people who read the report brought previous

knowledge of the fishing industry to bear on their interpretation of the

exemplar. That knowledge might have led some readers to adopt negotiated

or oppositional readings of the encoded message. A reader with second-hand

knowledge of the crab industry, for example, may have rejected the

“mysterious” construction of the exemplar in favor of a strongly held prior

opinion about the cause of the decline. When considering the meaning of an

exemplar, it is important to recognize the idiosyncratic nature of

interpretation, and how individual knowledge invariably factors into its

outcome.

124 Parker.
80
 In a traditional exemplification study the blue-crabs article would

likely be categorized as “selective” in its presentation, as there were no

exemplars in the report that seemed to contradict or call into question the

crabs’ population decline. However, the “selective” category is debatable, as it

assumes the encoded exemplarity was the “correct” interpretation for all of

the exemplars. It is possible, for instance, that one of the exemplars in the

blue crabs report that seemed “supportive” of the story’s focus was actually

decoded as “contradictory” to the trend by the majority of readers. If that

were the case, it could be argued that the story was not an example of

“selective” exemplification at all, since it contained what was widely viewed

by the audience as a contradictory exemplar. When the possible

interpretations of readers are given primacy in this way, it becomes much

more difficult—if not untenable—to categorize a report as an example of

“selective” exemplification.

Case #3: Suicides in Japan

The third case study will consider an article from the August 24, 2003,

edition of the Los Angeles Times that was headlined, “Internet Suicides

Plague Japan: Young People Make Death Pacts With Strangers.”125 As in the

previous case, this report examined a distressing trend in a community. The

story began with the exemplar of Murata, a 28-year-old sushi deliveryman in

125 Anthony Faiola, “Internet Suicides Plague Japan,” Los Angeles Times, 24 August 2003.
81
Japan who became ensnared in the “recesses of the Internet.”126 Murata

spent his days “glued” to the Web until he met online two anonymous

individuals who shared his “dark interest” in committing suicide.127 The three

individuals died together on a mountain pass after following suicide

instructions they apparently found on the Internet.

The introductory anecdote of Murata’s death was presented as part of

a larger trend in Japanese society:

The deaths of the three men marked only one incident in an

extraordinary string of Internet suicides to hit Japan. Over the past
six months, police investigators say at least 32 people—mostly in
their teens and twenties—have killed themselves nationwide after
meeting strangers online.128

It is worth noting that contemporary readers of the report could have

adopted an oppositional decoding position and rejected the existence of the

purported Japanese suicide trend altogether. Given that 32 is (arguably) not

a large number of suicide cases, that oppositional interpretation seems

particularly plausible in this instance. A reader who rejected the trend also

would have been likely to take an oppositional approach when interpreting

the rest of the suicide exemplars in the report.

Murata’s brother was quoted in the article trying to make sense of the

suicide:

“Maybe he didn’t have high hopes for the future, but it is still so
hard to understand how he could have done it,” said Murata’s 35-
year-old brother, who shared their apartment. He spoke on

126 Ibid.
127 Ibid.
128 Ibid.

82
 condition that both their names be withheld. “I’ve disconnected the
Internet at home, at least while our family comes to terms with
this.”129

The exemplar of Murata’s brother humanized the anecdotal suicide by

showing the distress it caused family members. Given the research that

found vivid, dramatic exemplars were more persuasive to readers than pallid

ones,130 such emotional flourishes are worth noting, as they may have a

measurable impact on the interpretations of an audience.

One of the things that differentiates the L.A. Times report from the

previous two cases is the fact that the exemplified trend (i.e., an outbreak of

Internet-assisted suicides) was explicitly linked to another, broader trend in

Japanese society:

Psychiatrists and suicide experts are linking the phenomenon to a

profound national identity crisis during Japan’s 13-year economic
funk. Indeed, the Internet deaths come at a time when Japan is
undergoing an alarming surge in its overall suicide rate—with
financial reasons cited as the fastest growing reason for despair.

That statement added considerable complexity to the narrative, as the

exemplars that followed in the report could have been viewed on two

mutually inclusive levels of meaning: as exemplified cases of the Web suicide

trend, and as exemplified cases of Japan’s national identity crisis. The

complicated interpretive situation—which could be termed “multi-

Faiola.
129

Aust and Zillmann, “Effects of Victim Exemplification in Television News on Viewer

130

Perception of Social Issues.”

83
exemplarity”—does not seem to have a precedent in the research literature

on media exemplification.131

The report described young people in Japan as a “lost generation”

unmoored from the traditional “pillars of society.”132 The article cited rising

homelessness, unemployment and job insecurity as evidence of the country’s

growing social instability. Near the middle of the report a psychiatrist was

introduced as a “supportive” exemplar for the purported identity crisis

phenomenon:

“[Japan’s youth] are lost and confused. The long-held direction and
goals of Japanese society are collapsing around them,” said Rika
Kayama, a Tokyo psychiatrist who has studied the Internet suicide
phenomenon.133

The report claimed Japan’s identity crisis was manifesting itself in

other ways besides suicide. One sign of the trend, the report claimed, was an

increase in so-called “shut-in” lifestyles among young people. The article also

cited an increase in kreru—known in Japan as episodes in which people

respond violently to minor provocations—as more proof of the trend. Those

two exemplars (shut-ins and kreru) seem to have been encoded as

“supportive” of the national identity crisis, a meaning that may not have been

similarly constructed by the contemporary readership of the article.

131 None of the exemplification studies presented in the literature review of this project used
news reports with multi-exemplarity for an experimental test.
132 Faiola.
133 Ibid.

84
 After the exploration of the identity crisis trend, the article presented

another exemplar of Internet-assisted suicide. That case, which the report

cited as the “beginning” of the Web-death trend, involved a man and two

women who killed themselves in a vacant apartment after meeting on a

suicide site:

A 17-year-old girl who had originally agreed to join them but

backed out at the last minute, told police that the three were jobless
and worried about the future. They had met after Michio Sakai, a
26-year-old unemployed magazine salesman from just north of
Tokyo, posted a “death ad” on an underground suicide site: “I am
looking for suicide partners,” it said. “If you join me, I will give you
sleeping pills ... It is lonely to die alone.”134

Note that this second suicide exemplar seems to have been organized

to reinforce both of the trends that were discussed in the report. On one level,

the exemplar was clearly an illustration of the Web-suicide “trend.” But the

exemplar also mentioned that all three of the suicide victims were “jobless

and worried about the future,” apparently in order to suggest their deaths

were also a symptom of Japan’s social instability. Thus it can be argued that

the three-death anecdote was an instance of multi-exemplarity in action, at

least within the context of the report.

The proposed multi-exemplarity of the three-death anecdote made for

an interesting interpretive puzzle. Contemporary readers of the report could

have adopted the dominant/hegemonic construction of the exemplar

wholesale and interpreted it as factual evidence of both the suicide and

134 Faiola.
85
identity “trends.” Readers also could have constructed the meaning of the

exemplar from more than one decoding position, perhaps accepting the

exemplar as evidence of the Web death trend (a dominant/hegemonic

decoding) but rejecting the purported link to a crisis in national identity (an

oppositional decoding)—and that’s just one example of the possible meanings.

When an exemplar illustrates more than one trend, it becomes that much

more difficult to justify a “supportive,” “neutral” or “contradictory” label, as

the possible interpretations by an audience could display the kind of

fragmentation that is beyond the scope of a simple descriptor.

The multi-exemplarity of the suicide cases also can be viewed, from the

Barthesian perspective, as a possible instance of myth-construction by the

press. On one level of meaning, each suicide exemplar was a fully formed

linguistic system (i.e., a sum of signs) that relayed a rich history. But each

exemplar also could have been interpreted as a mere signifier for Japan’s

national identity crisis, a mythical signified.

Nonetheless, the L.A. Times report cannot be considered a pure

example of mythical speech, because the mythical association was presented

in a distant, hazy manner that preserved the independence of the exemplars

as self-contained meanings. It is perhaps more accurate to think of the

suicide exemplars as half-finished or in-progress myth constructions.

Consider, for example, the suicide of Murata that was introduced at the

beginning of the L.A. Times report. That exemplar could have been

86
presented, from the outset, as an example of Japan’s national identity crisis;

in that hypothetical revision, the details of the suicide (i.e., his emotional and

employment problems, the Internet’s role) likely would have been omitted or

mentioned only in passing. In the form of simple caricature, the Murata

exemplar could have been appropriated as a signifier for Japan’s national

identity crisis. In that scenario, the exemplar would have existed as both

meaning and form under the totality of the global, mythical construct of

societal instability.

Of course, Murata’s death wasn’t presented in the manner of that

hypothetical, and neither were the other suicide exemplars in the article.

This thought exercise is valuable nonetheless because it shows the materials

of myth were present in the article; the potential for myth was there, even if

it wasn’t fully constructed in this instance. The exemplars were structured

around ideological associations that could have solidified, over time, into the

“self-evident” meanings that Barthes defined as myth. Barring that

development, the exemplars nonetheless were structured to suggest a

cultural, ideological meaning (i.e., cultural dissolution in Japan) that was

mythical in nature.

The suicides story also illustrates the importance of cultural

knowledge in the interpretation of exemplars. It seems likely that most L.A.

Times readers lacked substantive knowledge of Japanese culture or history

and were dependent on the interpretations encoded in the story to make

87
sense of it. Compare that situation to the blue crabs report, which covered a

hyper-local issue known to many readers on a first-hand basis. Readers of the

blue crabs report could have drawn on knowledge of the Eastern Shore’s

culture and social history to construct idiosyncratic interpretations of the

report. Readers of the suicides story, in contrast, would have been hard-

pressed to adopt a negotiated or oppositional interpretation of a phenomenon

occurring in a culture foreign to them.

Finally, similar to the other case study articles explored thus far, the

suicides article did not undercut the story focus with explicitly contradictory

exemplars. All of the statistics and exemplars in the story were used to

highlight the Internet-assisted suicides and social instability trends, and no

attempt at questioning the prevalence of the alleged phenomena was made.

Looking at it purely from the encoding side of Hall’s process, it can be argued

that the story was an example of selective exemplification in the press.

Case #4: Young Offenders

This case study will review a report about juvenile crime from the

December 18, 2003, edition of the Los Angeles Times.135 The article, which

was headlined “Jobs, Not Juvenile Hall,” examined a program in Santa Cruz,

California, that placed young criminal offenders under house arrest as part of

a comprehensive rehabilitation program. Participants in the program were

closely supervised and given jobs, tutoring and counseling as part of the

135 Eric Bailey, “Jobs, Not Juvenile Hall,” Los Angeles Times, 18 Dec. 2003.
88
government-led effort to cut recidivism, according to the report. One of the

statistics presented in the article suggested the program was having its

intended effect, as only 2 percent of the juvenile participants “committed new

offenses while awaiting resolution of their cases.”136 The report singled out

Santa Cruz as a notable exception to a shift toward increased incarceration of

juvenile offenders in California and in the United States.

The article examined the Santa Cruz reform effort within the context

of a broader cultural and political debate about the proper way to deal with

young criminals. The lead paragraph of the report noted that a “get-tough”

approach to juvenile crime was sweeping the nation, a mindset that had been

rejected by “liberal” Santa Cruz.137 That competition of ideas between

increased incarcerations for juvenile offenders, one on hand, and

rehabilitation, on the other, appeared often throughout the article, and may

have influenced the possible interpretations of the exemplars that were

presented.

The article introduced the exemplar of Nick Jackson, 17, who was

trying to turn around his life with the help of the Santa Cruz reform

program:

On one recent evening, the 17-year-old aspiring chef stood rapt

amid glistening pots and pans, learning to whip up wok-seared
salmon. Jackson, who started shoplifting at 8 and had graduated to
grand-theft auto by his teens, said he has traded crime for a
commercial kitchen.

136 Ibid.
137 Ibid.
89
 “It was getting to the point in life where I didn’t know what to do,”
he admitted, gingerly flipping a fish filet under the watchful eye of
his tutor, a local executive chef. “I fell in love with this.”138

That anecdote seems to have been encoded as “supportive” of the idea

that the Santa Cruz effort was working. By extension, the exemplar also

seems to have been encoded with a preferred meaning that was “supportive”

of the idea that juvenile criminals should be given a second chance before

being institutionalized.

Unlike the other case study articles examined thus far, the juvenile

crime report also explicitly acknowledged the possibility of oppositional

readings of the exemplar. In the paragraph immediately following the

Jackson anecdote, the article noted that skeptics viewed the Santa Cruz

program as a “soft-headed experiment with potentially dangerous results.”139

The proximity of that statement to the exemplar is interesting, because it

could have prompted a reader who interpreted the exemplar from a

dominant/hegemonic position to question or possibly abandon that original

interpretation in favor of a negotiated or oppositional reading (e.g., a reader

could decide Jackson actually should be in a juvenile prison because he’s

likely to commit another crime). At the very least, the report’s

acknowledgment of a different viewpoint would seem to have made the article

138 Bailey.
139 Ibid.
90
an instance of what researchers have termed “blended” exemplification in the

news.

After a section that reviewed pertinent national statistics on juvenile

incarceration, the report presented an exemplar that seems to have been

encoded as “contradictory” to the purported success of the rehabilitation

program in Santa Cruz:

Stalwarts in the juvenile justice system contend the pendulum

shouldn’t swing too far. Jail, they say, is the best spot for many
teenage offenders.

Larry Price, president of Chief Probation Officers of California,

admits to an “uncomfortable feeling” that reformers in places such
as Santa Cruz, despite commendable intentions, sometimes fail to
make safety a high enough priority. Price said “they need to bring it
back to a middle point.

Harsher critics say reformer want to coddle natural-born criminals.

The nation’s get-tough stand, they contend, has played a role in
stemming youth crime, proving that we should stay the course.140

That oppositional exemplar was given more than just a passing

mention in the report, but the opinion it expressed was marginalized to some

extent. At no point did the report present evidence, statistical or otherwise, to

bolster the oppositional view. Within the context of the report, the preferred,

encoded meaning of the exemplar seems to have been that the successful

program in Santa Cruz disproved the “contradictory,” dissenting viewpoint.

Therefore, it can be argued that only readers who adopted a negotiated or

oppositional decoding of the exemplar would have interpreted the exemplar

140 Bailey.
91
in a manner “contradictory” to the report’s claims about rehabilitating

juvenile offenders.

The counterintuitive idea that seemingly “contradictory” exemplars

can be encoded as “supportive” of a story’s focus had no precedent in the

exemplification studies that were reviewed for this project. The phenomenon

could help explain why respondents in experimental studies typically

discounted contradictory exemplars when they were in the minority of cases;

perhaps some of those contradictory exemplars were not encoded in a way

that actually challenged the dominant message of the manipulated articles.

That idea deserves to be explored formally in future research, as it calls into

question the consensus that the experimental results uniformly reflect a

stable bias in human perception.

After the oppositional exemplar, the report explained that officials in

Santa Cruz decided to focus on juvenile rehabilitation partially out of a desire

to ease overcrowding in their lockup facilities:

Even in Santa Cruz, where a progressive spirit seems part of the

civic constitution, bureaucratic inertia had to be overcome. Some
police and prosecutors needed to set aside the notion that juvenile
hall is the best way to handle errant kids.141

That statement was an example of the subtle phrasing used in the

report that appears to marginalize opponents of juvenile rehabilitation. The

description of law enforcement officials in Santa Cruz was arguably

dismissive of their concerns and suggested they were wrong to have

141 Bailey.
92
reservations about the new approach. Readers nonetheless may have

disregarded the preferred meaning and constructed the “contradictory”

exemplars as “correct” and “legitimate” points of view.

Case Studies: Review

The case studies in this project employed Hall’s model of

encoding/decoding and Barthes’ mythical system to explore the possible

meanings suggested by exemplars. The results of the exercise suggest that

the interpretation of exemplars is a much more complex process than

researchers may have previously acknowledged.

The articles that were analyzed in the four case studies illustrate the

untidy realities of exemplification in newspaper journalism. The case studies

suggest the exemplifying function of a quote or anecdote is complex and open

to multiple interpretations by an audience. Nonetheless, all of the stories

analyzed here fit the theoretical mold used in a traditional exemplification

study. One can easily imagine any of the case study articles being used in an

experiment by Zillmann, Brosius or other exemplification scholars, so the

issues raised by the case studies are deserving of serious research

consideration in the future.

The case study articles validated the research finding that exemplars

tend to follow the focus of a report—at least from an encoder’s point of view.

The debatable professional standard of “telling both sides” seems to fall by

the wayside in many exemplified narratives, which again raises concerns

93
about the coverage of dramatic-but-rare events, illnesses and crimes, since

the potential for exaggeration is so great.

The first case study article, about distressed neighborhoods, was

notably different from the other three reports, as it lacked explicitly

illustrative exemplars. Rather than using the data on poor neighborhoods as

a jumping-off point for the examination of a “distressed” community in

Baltimore, the article stuck to the statistics at hand and avoided cultural and

social debates about poverty. The article’s thorough presentation of research

data would likely meet the approval of Zillmann, who has long advocated for

a more empirical approach to news coverage in his writings on

exemplification.142

The second article, about the disappearance of blue crabs on

Maryland’s Eastern shore, was a fairly conventional instance of an

exemplified narrative. However, the case study showed how Hall’s model of

encoding/decoding can be used to challenge the notion that there is a single,

“correct” interpretation of an exemplar. When readers are viewed as having

an active role in the communication process, the possible meanings of an

exemplar become nearly limitless, as one can no longer assume the

“dominant” construction of an exemplar will be shared by audience members.

The case study also showed how an exemplar can potentially naturalize

For an example, see Zillmann, “Exemplification Theory: Judging the Whole by Some of Its
142

Parts.”
94
culturally constructed ideologies and thus be viewed as an example of myth-

making by the press.

The third report, about suicides in Japan, was arguably the most

intriguing of the four case study articles. The report purported the existence

of two trends in Japan—Internet-assisted suicide and social instability—that

were related to each other, yet fundamentally distinct. The presentation of

two trends created what could be called “multi-exemplarity,” as the

exemplars in the report could be viewed as constructed representations of

more than one observed phenomenon. The report’s dual structure could also

be viewed as a likely precursor to myth-construction, as one of the purported

trends in the report (social instability in Japan) was a culturally constructed

meaning that could have solidified, over time, into a naturalized, mythical

signified.

The suicides story also highlights the need to take reader familiarity

into account when considering the possible interpretations of exemplars.

Reports about cultures foreign to an audience are fundamentally different

from stories about local issues, as readers are much more dependent on the

encoded perspectives of a reporter to create meaning.

Finally, the fourth case study article, about juvenile rehabilitation,

showed that the concept of “blended” exemplification in the news is

problematic, as the exemplars in a report that seem “contradictory,” at first

glance, may not actually challenge the focus or “take” of an article when

95
viewed in context. Furthermore, when readers are viewed as active decoders

of exemplars, it would seem problematic, if not impossible, to classify the mix

of exemplars in an article as “supportive,” “contradictory” or “blended,” unless

the categories are defined from the point of view of an encoder.

96
Chapter 10. Suggestions for Future Research

Exemplification is a relatively new theoretical framework with room

for expansion and refinement. Further analysis using various research

methods holds great promise for the creation of a theoretical viewpoint with

substantial explanatory power.

The content-analysis format used in this study is a logical starting

point for additional inquiry. The study needs to be replicated on a larger

scale, so that the real-world implications of exemplification theory can be

better understood. For newspapers, the key refinement should come in

sample size; a robust content analysis would examine more stories from a

diverse mix of publications. The analysis would also code for exemplification

in stories without a trend orientation, so that all aspects of exemplification

can be better understood. The coding protocol used in this study seemed

suitable for the task of recording exemplification, and no major revisions to it

would be necessary if it were to be adapted to future studies.

Second, it is imperative that the type of analysis performed in this

study be expanded to a variety of media. Based on the content analyses

performed by Brosius, Daschmann and Zillmann, there would seem to be a

strong possibility that exemplification is far more prevalent in some media

than in others. If that is found to be the case, then the focus of future

exemplification research should shift accordingly.

97
 Third, it would be worthwhile to perform longitudinal case studies on

exemplars from different eras in history. For example, one could review

stories about AIDS from the 1980s to the present to see how the exemplars

used to illustrate a problem have changed over time. That analysis would be

more than a simple framing study, as focusing on exemplars alone could be a

revealing exercise; charting the patterns, recurring motifs and phrases in the

exemplars would be a powerful addition to the historical record on almost any

topic, and the possible connections to contemporary public opinion data would

open new areas of exploration.

Fourth, more exemplars—ideally sampled from a variety of media—

should be subjected to rigorous case study analyses. Such case studies would

not necessarily need to utilize the same critical studies framework that was

employed in this project, as more insight may well be gained from an entirely

different approach. Regardless of form or method, such exercises have the

potential to enrich the literature on exemplification by uncovering possible

refinements and improvements to the model that could then be incorporated

into the design of experimental studies.

Finally, the exemplification model itself is deserving of further

scrutiny from researchers. The analyses of case study articles presented in

this project suggest many of the operational definitions that have been used

in exemplification studies (e.g., “supportive,” “contradictory”) are overly rigid

and fail to take into account the possibility of multiple interpretations. The

98
transmission-model orientation of exemplification research will not be easy to

change, as it is intrinsic to most of the quantitative findings about exemplars.

Nonetheless, it is worth considering whether a more nuanced model of

exemplification can be created that recognizes the active, idiosyncratic

interpretations of an audience.

The results of the content analysis performed for this study call into

question the idea that exemplified narratives are common in the news. It is

imperative that researchers undertake additional quantitative studies of

media content to determine whether the exemplification theory of media

effects correlates to real-world journalism practices. If further study does not

uncover robust samples of exemplified narratives in the news, researchers

may be forced to consider the possibility that the exemplification theory is a

mythical research construct with limited relevance to contemporary media

practices.

99
Works Cited

Anthony Faiola, “Internet Suicides Plague Japan,” Los Angeles Times, 24 August 2003.

Bill Reader, “Distinctions That Matter: Ethical Differences at Large and Small
Newspapers,” Journalism & Mass Communication Quarterly 4 (2006).

Charles Berger, “Making It Worse Than It Is: Quantitative Depictions of Threatening

Trends in the News,” Journal of Communication 51 (2001).

Charles F. Aust and Dolf Zillmann, “Effects of Victim Exemplification in Television

News on Viewer Perception of Social Issues,” Journalism & Mass
Communication Quarterly 73 (1996).

Dan Berkowitz, “Non-Routine News and Newswork: Exploring a What-a-Story,”

Journal of Communication 42 (1992).

Daniel Chandler, Semiotics: The Basics, 2nd ed. (New York: Routledge, 2007).

Daniel Riffe, Charles F. Aust and Stephen Lacy, “The Effectiveness of Random,
Consecutive Day and Constructed Week Samples in Newspaper Content
Analysis,” Journalism Quarterly 1 (1993).

Daniel Riffe, Stephen Lacy and Frederick Fico, Analyzing Media Messages: Using
Quantitative Content Analysis in Research (Mahwah, New Jersey: Lawrence
Erlbaum, 1998).

David A. Craig, “The Promises and Peril of Anecdotes in News Coverage: An Ethical
Analysis,” Journalism & Mass Communication Quarterly 80 (2003).

Denis McQuail, McQuail’s Mass Communication Theory, 4th ed. (London: Sage
Publications, 2000).

Dolf Zillmann, “Exemplification Theory: Judging the Whole by Some of Its Parts,”
Media Psychology 1 (1999).

Dolf Zillmann, “The Exemplification Theory of Media Influence,” in Media Effects:

Advances in Theory and Research, 4th edition, eds. D. Zillmann and J. Bryant
(White Plains, N.Y.: Longman, 1997).

Dolf Zillmann and Hans-Bernd Brosius, Exemplification in Communication: The

Influence of Case Reports on the Perception of Issues (Mahwah, New Jersey:
Lawrence Erlbaum, 2000).

100
Dolf Zillmann, Rhonda Gibson, S. Shyam Sundar and Joseph W. Perkins Jr., “Effects of
Exemplification in News Reports on the Perception of Social Issues,” Journalism
& Mass Communication Quarterly 73 (1996).

Elisabeth Noelle-Neumann, “The Spiral of Silence: A Theory of Public Opinion,”

Journal of Communication 24 (1974).

Eric Bailey, “Jobs, Not Juvenile Hall,” Los Angeles Times, 18 Dec. 2003.

E. James Baesler and Judee K. Burgoon, “The Temporal Effects of Story and Statistical
Evidence on Belief Changes,” Communication Research 5 (1994).

Eric Siegel, “Groups’ Study on Kids is Grim,” The Sun, 23 Sept. 2003.

Fred Felder, John R. Bender, Lucinda D. Davenport and Michael W. Drager, Reporting
for the Media, 7th ed. (New York: Oxford, 2001).

Frederick Fico and Stan Soffin, “Fairness and Balance of Selected Newspaper Coverage
of Controversial National, State, and Local Issues,” Journalism & Mass
Communication Quarterly 3 (1995).

Gaye Tuchman, “Making News by Doing Work: Routinizing the Unexpected,” American
Journal of Sociology 79 (July 1971).

Gregor Daschmann and Hans-Bernd Brosius, “Can a Single Incident Create an Issue?
Exemplars in German Television Magazine Shows,” Journalism & Mass
Communication Quarterly 1 (1999).

Gretchen Parker, “Eastern Shore Crabs Hard to Come By,” Washington Times, 22 July
2003.

Guido H. Stempel III, “Gatekeeping: The Mix of Topics and the Selection of Stories,”
Journalism Quarterly 62 (1985).

Hans-Bernd Brosius, “Exemplars in the News: A Theory of the Effects of Political

Communication,” in Communication and Emotion: Essays in Honor of Dolf
Zillmann, eds. J. Bryant, D. Roskos-Ewoldsen and J. Cantor (Mahwah, New
Jersey: Lawrence Erlbaum, 2003).

Hans-Bernd Brosius, “Research Note: The Influence of Exemplars on Recipients’

Judgments,” European Journal of Communication 14 (1999).

Hans-Bernd Brosius and Anke Bathelt, “The Utility of Exemplars in Persuasive

Communications,” Communication Research 21 (1994).

101
Janet A. Bridges, “News Use on the Front Page of the American Daily,” Journalism
Quarterly 66 (1989).

Jock Lauterer, Community Journalism: Relentlessly Local (Chapel Hill, North Carolina.:
University of North Carolina Press, 2006).

Judgment Under Uncertainty: Heuristics and Biases, Eds. Daniel Kahneman, Paul Slovic
and Amos Tversky (New York: Cambridge University Press, 1982).

L. Erwin Atwood and Gerald L. Grotta, “Socialization of News Values in Beginning

Reporters,” Journalism Quarterly 50 (Winter 1973).

Maxwell E. McCombs and Donald L. Shaw, “The Agenda-Setting Function of Mass

Media,” Public Opinion Quarterly 36 (Summer 1972).

Michael A. Shapiro, “Memory and Decision Process and the Construction of Social
Reality,” Communication Research 18 (1991).

Michael Schudson, Discovering the News (USA: Basic Books, 1978).

Michael Schudson, “Runaway Jury Awards … And Other Made-for-Media Myths,”

Columbia Journalism Review 2 (2005).

Peter S. Goodman, “Jobless Rate at 14-Year High After October Losses,” The New York
Times, 7 Nov. 2008.

Rhonda Gibson and Dolf Zillmann, “Effects of Citation in Exemplifying Testimony on

Issue Perception,” Journalism & Mass Communication Quarterly 75 (1998).

Rhonda Gibson and Dolf Zillmann, “Exaggerated Versus Representative Exemplification

in News Reports: Perceptions of Issues and Personal Consequences,”
Communication Research 21 (1994).

Rick W. Busselle and L.J. Shrum, “Media Exposure and Exemplar Accessibility,” Media
Psychology 5 (2003).

Roland Barthes, “Myth Today,” in Mythologies, Annette Lavers, trans. (New York:
Noonday Press, 1972).

Ruth Hamill, Richard E. Nisbett and Timothy DeCamp Wilson, “Insensitivity to Sample
Bias: Generalizing from Atypical Cases,” Journal of Personality and Social
Psychology 39 (1980).

102
S. Shyam Sundar, “News Features and Learning,” in Communication and Emotion:
Essays in Honor of Dolf Zillmann, eds. J. Bryant, D. Roskos-Ewoldsen and J.
Cantor (Mahwah, New Jersey: Lawrence Erlbaum, 2003).

Stephen Perry and William Gonzenbach, “Effects of News Exemplification Extended:

Considerations of Controversiality and Perceived Future Opinion,” Journal of
Broadcasting and Electronic Media 41 (1997).

Stuart Hall, “Encoding/Decoding,” in Culture, Media, Language, eds. Stuart Hall,

Dorothy Hobson, Andrew Lowe and Paul Willis (London: Hutchinson, 1980).

Sun-lin Gan, John R. Hill, Elke Pschernig and Dolf Zillmann, “The Hebron Massace:
Selective Reports of Jewish Reactions, and Perceptions of Volatility in Israel,”
Journal of Broadcasting and Electronic Media 40 (1996).

Walter Lippmann, Public Opinion (New York: Free Press, 1972).

Walter Lippmann, “The World Outside and the Pictures in Our Heads,” in The Process
and Effects of Mass Communication Revised Edition, eds. W. Schramm and D.
Roberts (Urbana: University of Illinois Press, 1974).

Warren Breed, “Social Control in the News Room: A Functional Analysis,” Social
Forces 33 (1955).

103
Appendix A. Codebook

For purposes of this study we will only be considering stories on the front
page of the newspaper and the front page of the local news/metro section. The
only time you should go inside the paper is if an article jumps to another
page.

*** In order for a news article to be included in the coding process, the story
must contain statistical information. News stories that do not include
statistical data are to be excluded from the coding process.

To warrant inclusion in the sample, the quantitative data in the story has to
come from an identified external source. This rule applies equally to experts,
public officials and scholars. If empirical data is mentioned, and it is linked to
a source (however vague), the article should be included for coding in the
sample.

There are two key exceptions to the method detailed above. One: Articles
focused on the budgetary matters of governments, businesses and other
agencies are not to be included in this analysis. Financial statistics are
typically not generalizations about the world and thus fall outside the
exemplification paradigm. Two: in some cases, a statistic in a story has no
relationship to the exemplars given and is completely tangential to the focus.
Looking through the story reveals that all of the exemplars are neutral
because none of them relate to the statistic. These articles should be
excluded.

Finally, news columns (opinion pieces) are not fit for inclusion.

104
Article-Level Protocol

Once you have decided that an article is fit for coding, you should read the
entire article from start to finish, taking careful note of the story’s subject
matter. Then you are to indicate the story’s topic in the category list. In
ambiguous cases, refer to the following list of definitions for guidance:

• Politics and Government: Government acts and politics at the

local, state and national level.
• War and Defense: War, defense, rebellion, military use of space.
Includes both foreign and domestic stories.
• Diplomacy and Foreign Relations: Both foreign and domestic
items dealing with diplomacy and foreign relations. Includes
United Nations.
• Economic Activity: General economic activity, prices, money,
labor, wages and natural resources.
• Agriculture: Farming, farm prices and economic aspects of
agriculture.
• Transportation and Travel: Transportation and travel,
including economic aspects.
• Crime: All crime stories including criminal proceedings in court.
• Public Moral Problems: Human relations and moral problems
including alcohol, divorce, sex, race relations and civil court
proceedings.
• Accidents and Disasters: Both man-made accidents and
natural disasters.
• Science and Technology: Science other than defense related
and other than health and medicine.
• Public Health and Welfare: Health, public welfare, social and
safety measures, welfare of children and marriage and marriage
relations.
• Education and Classic Arts: Education, classic arts, religion
and philanthropy.
• Popular Culture: Entertainment and celebrities, newspage
sports, TV, movies, radio and other media.
• General Human Interest: Human interest, weather, obits,
animals, cute children and juvenile interest.

After recording the story’s statistical information you are asked to

indicate its precision. The three categories are defined as follows:

105
 High precision: Statistics that contain specific counts, percentages,
ratios and estimates. Data that is quantitative in nature; specific instead of
vague; numeric and not generalized.
Low precision: Statistics than use comparative language, such as
“some,” “more and more” and “a majority.” Specific numbers are not given.
Mixed: Statistics that are both high and low in precision.

The next task in the protocol is the counting of exemplars. Consider

this textbook definition of the term exemplar:
Exemplars are “case descriptions of specifications of singular incidents
that fall within the realm of a particular social phenomenon and that exhibit
the pertinent properties of that phenomenon to a certain degree.”143

In non-technical terms, exemplars are merely examples—illustrations,

aides—of a larger phenomenon. However, exemplars need not be a direct
representation of a category. Quotes are a form of exemplification because
they relate concrete information from the social world. Oftentimes,
quotations are offered in summary form, but they nonetheless retain their
exemplifying function. Similarly, reporters often include descriptions that
exemplify the topic at hand (e.g., a description of a worn-down school building
in an article about education costs) from a first-person perspective. These
common forms of exemplification—quotations, summaries and descriptions—
should all be considered exemplars in the analysis.

In order to count as a unit, an exemplar must be distinct in the text.

However, the source or the content of the exemplar need not be unique.
Rather, distinct exemplars may sometimes be the same person/example,
found as separate entities in the layout of the article. An expert, for example,
who is quoted and discussed for 3 straight paragraphs should be counted as
one distinct exemplar. If this same expert is quoted again later in the article,
the second quote should be counted as a separate exemplar. Essentially, so
long as an exemplar remains uninterrupted in the narrative, it should be
counted as one. The counting for every article should proceed in this manner,
with one exemplar added to the count each time a fresh (distinct) one is
encountered.

After recording the exemplar count, you need to record how many of
the exemplars support the story focus. To determine story focus, ask
yourself: What is the main thesis of this article? What perspective is the
writer trying to convey? If someone asked you what the article you’re reading
is about, what would you say?

143
Zillmann et. al, “Effects of Exemplification in News Reports on the Perception of Social
Issues,” 427.
106
 By support, it is simply meant that the exemplar is an example of the
phenomenon/risk/social problem/opinion described. For example, if a story
states, “1 percent of the population is susceptible to rapid, premature aging,”
all exemplars of people suffering from the ailment should be counted as
supportive exemplars, since that is the story’s focus.
Consequently, all exemplars of doctors or experts talking about the
ailment should also be counted as supportive exemplars. Only exemplars that
downplay the risk, in this instance, would escape the supportive category—
instead falling under the contradictory label. If an exemplar does not support
or contradict the story focus, it should be counted as neutral.

Next, you need to indicate the story’s exemplar distribution. The

two categories are defined as follows:

Selective: All of the exemplars in a story follow the given focus. No

contradictory examples are given. The focus of the story does not shift, and no
attempt at balance is made.
Blended: The exemplar distribution is not selective.

107
Exemplar-Level Protocol

In this portion of the coding you will be looking at each exemplar

individually. Exemplars must be distinct in the text, using the same
guidelines of the earlier article-level protocol.
After recording the date and source of the exemplar, count and record
the total number of sentences in the exemplar.
Finally, record the exemplar’s source. The division between quotes and
paraphrases is not rigid: if an exemplar contains a full quote, it is not a
paraphrase. However, quote marks around a phrase or word do not
make the exemplar a quote. The quotation must be a full sentence.

• Quote of expert or academic: The source is described as an

expert, doctor, researcher, professor, etc. The journalist
ascribes authority to the source, either by title or by explicit
statement.
• Paraphrase of expert or academic: The source is described
as an expert, doctor, researcher, professor, etc. The journalist
ascribes authority to the source, either by title or by explicit
statement. The source is never offered in direct quotation.
• Quote of public official: The source is someone in government
(local or national) or in a government agency.
• Paraphrase of public official: The source is someone in
government (local or national) or in a government agency. The
source is never offered in direct quotation.
• Quote of resident: Anyone who is not a public official or an
expert is almost certainly a resident/community member. The
person need not reside in a particular community, since some
stories have a national focus. Resident simple means that a
person is not in the government or the academy.
• Paraphrase of resident: Anyone who is not a public official or
an expert is almost certainly a resident/community member. The
person need not reside in a particular community, since some
stories have a national focus. Resident simple means that a
person is not in the government or the academy. The source is
never offered in direct quotation.
• Anecdote: In an anecdote the journalist is the speaker. The
writer is not relaying what a person said; instead, they are
providing a description of an experience from a first or second-
hand perspective. Sometimes journalists paraphrase a person’s
experience in anecdote form; so long as the person does not
speak (key word: said), it is an anecdote. Some exemplars may
be a combination of an anecdote and a quote or paraphrase—in

108
 which case it should be filed under one of the above categories,
and not anecdote.
• Hypothetical: The writer offers a purely hypothetical situation
or instance to illustrate a point.

109
Appendix B. Coding Protocol

110
111
Appendix C. Case Study Articles

Eric Siegel, “Groups’ Study on Kids is Grim,” The Sun, 23 Sept. 2003.

112
Eric Siegel, “Groups’ Study on Kids is Grim,” The Sun, 23 Sept. 2003.

113
Gretchen Parker, “Eastern Shore Crabs Hard to Come By,” Washington Times, 22 July 2003.

114
Gretchen Parker, “Eastern Shore Crabs Hard to Come By,” Washington Times, 22 July 2003.

115
Anthony Faiola, “Internet Suicides Plague Japan,” Los Angeles Times, 24 August 2003.

116
Anthony Faiola, “Internet Suicides Plague Japan,” Los Angeles Times, 24 August 2003.

117
Eric Bailey, “Jobs, Not Juvenile Hall,” Los Angeles Times, 18 Dec. 2003.

118
Eric Bailey, “Jobs, Not Juvenile Hall,” Los Angeles Times, 18 Dec. 2003.

119
Eric Bailey, “Jobs, Not Juvenile Hall,” Los Angeles Times, 18 Dec. 2003.

120