Citation
Explaining the "Bandwagon" and "underdog" effects

Material Information

Title:
Explaining the "Bandwagon" and "underdog" effects
Creator:
Hollander, Barry A
Copyright Date:
1991
Language:
English

Subjects

Subjects / Keywords:
Capital gains taxes ( jstor )
College students ( jstor )
Control groups ( jstor )
Motivation ( jstor )
Personality psychology ( jstor )
Persuasion ( jstor )
Polls ( jstor )
Public opinion ( jstor )
Social psychology ( jstor )
Taxes ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
25604963 ( OCLC )
AJC6967 ( LTUF )
0027013247 ( ALEPH )

Downloads

This item has the following downloads:


Full Text















EXPLAINING THE "BANDWAGON" AND "UNDERDOG" EFFECTS:
A STUDY OF PERSONAL RELEVANCE AND UNCERTAINTY ORIENTATION
AS FACTORS IN PUBLIC OPINION POLL INFLUENCE
















By

BARRY A. HOLLANDER


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1991















ACKNOWLEDGEMENTS


There are many people who deserve recognition for their assistance in my completing my graduate studies, which culminate in this dissertation.

First, I would like to thank my committee: Drs. Mary Ann Ferguson, Michael Martinez, Leonard Tipton, Michael Weigold, and John Wright. It was a Leonard Tipton research seminar that first sparked my interest in public opinion and perceptions of opinion. No doubt this fascination will continue for some time.

I would also like to thank Kim Walsh-Childers for acting as a substitute on my final dissertation. Thanks also to Michael Weigold and Julie Dodd for use of their students in the dissertation experiments.

Mary Ann Ferguson deserves special note. She hired

me as her research assistant at the Communication Research center when I was a lowly and hungry master's student barely able to find a computer's on-off switch. My threeyear apprenticeship with her at the Center is the reason I continued my graduate studies, and her research seminars provided me a theoretical direction. Her enthusiasm for research and teaching was contagious. If ever a graduate student bears the stamp of his major professor, I suspect












I do. With luck and hard work, I may live up to this ideal.

A number of fellow graduate students deserve

acknowledgement, more than I have the time or space to mention. Geetu Melwani from the period of my work on a master's degree and fellow doctoral students Erika Engstrom, Art Emig, and Alan Fried come immediately to mind. There are many others.

Finally, and most important, I thank my wife Edith for her patience, love, and support. She worked as I continued my graduate studies, making it possible for me to write this dissertation. I look forward to spending time with her again.


iii

















TABLE OF CONTENTS


page

ii vi



1

2 4 5 6 7 8 16 19

21 21 25 28 28 33
34 35 36 39 39
43 44 46


ACKNOWLEDGEMENTS .

ABSTRACT .

CHAPTERS

1 INTRODUCTION .

A Different Perspective .
Research Hypotheses .
Summary of Dissertation Contents .

2 PUBLIC OPINION AND POLLS .

Polls and Mass Communication .
Bandwagon Research .
Conformity Research .
The Poll Bandwagon--A Summary .

3 PERSUASION RESEARCH .

Cognitive Response Analysis .
Persuasion and Individual Differences .

4 UNCERTAINTY ORIENTATION .

Theoretical Discussion .
Findings in uncertainty Orientation .
Authoritarianism .
Other Personality Theories .
On the Study of Personality, Politics,
and Communication .

5 SYNTHESIS AND HYPOTHESES .

Bandwagon Research and Persuasion .
Consensus as a Heuristic Cue .
Summary and Hypotheses .
Overview of the Studies .













6 STUDY 1 METHODOLOGY AND RESULTS . 47

Methodology . 47 Results . 52 Summary . 60 Notes . 64

7 STUDY 2 METHODOLOGY . 66

The News Articles . 66 Measurement of Variables . 71 Uncertainty Orientation . 72

8 STUDY 2 RESULTS . 77

Assumptions . 77 Manipulation Checks . 78 Tests of Hypotheses . 80 Other Significant Findings . 94 Notes . 95

9 DISCUSSION . 96

Summary of Study Findings . 96 Limitations of the Studies . 100 Conclusions . 105 Implications for Bandwagon Research . 107 Future Research . 110

APPENDICES

A NEWS ARTICLE EXAMPLES . 113

B MEANS AND STANDARD DEVIATIONS OF ALL
UNCERTAINTY ORIENTATION QUESTIONS . 126

C FACTOR LOADINGS OF ALL UNCERTAINTY
ORIENTATION QUESTIONS . 130

REFERENCES . 132

BIOGRAPHICAL SKETCH . 147
















Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy EXPLAINING THE "BANDWAGON" AND "UNDERDOG" EFFECTS:
A STUDY OF PERSONAL RELEVANCE AND UNCERTAINTY ORIENTATION
AS FACTORS IN PUBLIC OPINION POLL INFLUENCE

By

Barry A. Hollander

December 1991

Chairman: Leonard Tipton
Major Department: Mass Communication

Concerns about the influence of public opinion polls have existed almost as long as polls themselves. "Bandwagon" research results have been inconsistent, however, even with an underdog effect being found. The conceptual framework of persuasion research from social psychology was used to review the bandwagon literature and to offer hypotheses. The individual difference variable of uncertainty orientation was presented as one explanation as to why some people are influenced in the direction of a poll while others are influenced in the opposite direction.

Three hypotheses were presented. First, it was

hypothesized that subjects would be more susceptible to poll influence in a low relevance situation as opposed to one of high relevance. Second, it was hypothesized that












certainty-oriented persons would be more likely to move in a "bandwagon" direction, or with the poll influence, while uncertainty-oriented persons would be more likely to move in an "underdog" direction, or opposite the poll influence. Finally, it was hypothesized that when relevance was high, certainty-oriented persons would be susceptible to poll influence.

A 2 (relevance) X 2 (uncertainty orientation) X 3

(type of poll) factorial design experiment was conducted using news articles with bogus poll results. Mixed support was found for all three hypotheses, primarily in situations where relevance was manipulated rather than between issues thought to differ in relevance. It was suggested that future research into public opinion poll influence take into consideration the relevance of the campaign or issue under consideration as well as the personality traits of those hearing such poll results.


vii















CHAPTER 1
INTRODUCTION


In 1935, U.S. Rep. Walter Pierce of Oregon introduced legislation that would have prohibited use of the mail for gathering political straw ballots. If enacted, pollsters convicted under the law faced a maximum penalty of one year in prison or a $1,000 fine or both. Such polls are "powerful, subsidized propaganda," Pierce warned in a later Public Opinion Quarterly article, and should be recognized as "a potent, if not the most powerful, agency now used to influence public opinion" (Pierce, 1940, p. 241 and p. 243, respectively).

While Pierce's bill failed, the potential influence of public opinion polls has been a focus of debate for more than 50 years. Pollsters assert that their efforts reflect rather than affect public opinion, arguing no "bandwagon" effect exists (Gallup & Rae, 1940; Robinson, 1937). Other early commentators such as Pierce (1940) and Lewis (1940), however, doubted that polls were so innocent and incapable of persuasion.

Despite arguments, evidence from survey and

experimental studies remains inconclusive. Reviews by Marsh (1983) and Merkle (1991) found that while some studies do favor a bandwagon interpretation, many do not.












Laboratory studies of artificial elections coupled with bogus poll results produce an "underdog" effect, of people shifting to the minority rather than majority view. In general, there appears little support for the bandwagon thesis, particularly in controlled settings, which seem to favor an underdog effect. Possible reasons for such inconsistent conclusions range from the kinds of issues or candidates studied to how the "majority" and "minority" are identified. Few attempts have been made, however, to explain and systematically test why people may choose the bandwagon or underdog route or the process that takes place. Almost exclusively devoted to elections at the expense of issues, the bandwagon literature typically assumes an undifferentiated public, though some scholars have begun to investigate the influence of predispositions on likelihood to move with or against a poll (e.g., Lavrakas, Holley, & Miller, 1991).

A Different Perspective

Some public opinion scholars have hinted at the

possibility that polls are influential in situations of less importance to the public. McBride (1991) suggests such influence is more likely in political races below the presidential level, while Hickman (1991) points to presidential primaries rather than elections as better opportunities to find the bandwagon effect.












Another perspective is offered, this one from social psychology. Theorists such as Petty and Cacioppo (1986a, 1986b) outline two distinct modes of processing which bear directly on how and why persuasion takes place. The key to these persuasion models is whether people have the motivation and ability to process a message. Where motivation is low, people are thought to be more likely to use shortcuts in making decisions. An issue or political campaign of little personal relevance, then, is unlikely to provide the motivation necessary to carefully scrutinize all the questions involved. In such a case, a credible source or the presence of some other source characteristic can be of greater influence. In situations of high relevance, such source characteristics are thought to have little influence. With this conceptual framework in mind, the research in poll influence is reviewed and an attempt is made to organize the results along the lines seen in these persuasive models.

As noted above, one criticism of bandwagon research is its focus on an undifferentiated public. One explanation for both the bandwagon and underdog effects may also be in people themselves. That is, some people may be more persuaded by polls, while others may be more open to the arguments presented by a minority. The individual difference variable uncertainty orientation is












offered as one possible explanation (Sorrentino, Short, & Raynor, 1984). Briefly, Sorrentino and hiscolleagues posit that some people -- certainty-oriented persons -seek to maintain clarity about themselves and their environment. That is, they will avoid self-diagnostic situations if they feel that information is threatening, and, in fact, when facing a situation of personal importance they will rely on simple decision-making rules rather than consider all the information possible. Their polar opposites, uncertainty-oriented persons, are open to such information and in fact seek it out.

Research Hypotheses

Three research hypotheses can be generated from this discussion, all of which are discussed in more detail in later chapters. First, it is expected that issues of little personal relevance will be more influenced by poll results than issues of high personal relevance. When considering uncertainty orientation, it is hypothesized that certainty-oriented persons will be more likely to be persuaded by a poll than uncertainty-oriented persons.

Finally, building on other work in uncertainty

orientation, it is expected this relationship between uncertainty orientation and poll results will be more pronounced in situations of high relevance, which certainty-oriented persons would find threatening. That












is, certainty-oriented persons would be even more persuaded by a poll in personally relevant situations rather than in low-relevant situations.

Summary of Dissertation Contents

Chapter 2 reviews the research on the influence of

polls, most of which has been conducted by public opinion and mass communication scholars, while Chapter 3 discusses the persuasion literature from social psychology. Chapter 4 reviews the work on uncertainty orientation, and Chapter

5 provides a synthesis and the resulting hypotheses.

Study 1 (Chapter 6) is designed mainly to select issues for use in Study 2. The second study tests the research hypotheses discussed above. Chapter 7 presents the methodology of Study 2 and Chapter 8 the results. The final chapter provides an overall discussion of the findings.















CHAPTER 2
PUBLIC OPINION AND POLLS


Separating the poll from public opinion is difficult, and public opinion sometimes defined as that which a public opinion poll measures. Public opinion is often viewed as the aggregation and distribution of individual opinions, simply "any collection of individual opinions" (Childs, 1939, p. 331). Others regard public opinion as a communication process (e.g., Cooley, 1909; Price 1989). Price and Roberts (1988) distinguish between the pollster's view of Public in public opinion as a noun and their conceptualization of Public as an adjective describing the opinion process. An important feature of this conceptualization is F. D. Allport's (1937) view of public opinion formation occurring through an individual's observation of his social environment, a theoretical orientation extended by others, notably Noelle-Neumann's "spiral of silence" theory (Noelle-Neumann, 1974, 1977).

Noelle-Neumann (1974, p. 51) suggests individuals possess a "quasi-statistical organ" that allows them to assess the opinion environment, largely through information provided by the mass media. However, it appears from research such as the false consensus effect (Ross, Greene, & House, 1977; Marks & Miller, 1987) and












pluralistic ignorance (Fields & Schuman, 1976; O'Gorman & Garry, 1976) that if such an "organ" exists, it is indeed a faulty one. False consensus research has demonstrated that people tend to believe others think as they do, even when those individuals are obviously in a minority. Pluralistic ignorance has to do with misunderstanding of true opinion distribution.

Polls and Mass Communication

Key to Noelle-Neumann's spiral of silence theory is how people learn about the distribution of opinion. A likely source is public opinion polls, a popular staple of the news media since early in the century. More than onethird of newspapers surveyed reported the use of polls to gather information (Rippey, 1980), and about 15 percent of newspaper election stories focused on polls (Stovall & Solomon, 1984). While there has been some improvement in how the media conduct or report on polls (Salwen, 1985), others charge news organizations have fail to provide necessary explanatory information on polls and, when they do so, the reliance on such polls places a "horse-race metaphor" emphasis on elections (Broh, 1980). Content analyses of poll stories found newspapers often fail to provide the information deemed necessary by polling experts (Miller & Hurd, 1982; Sanders, Rollberg, & Buffalo, 1989), despite findings that inclusion of this













information may improve polls' credibility (Paletz, et al., 1980; Salwen, 1987).

People are aware of public opinion and polls. Gollin (1980) reported that 84 percent of a 1976 national sample had heard of polls. Kohut (1986) and Roper (1986) both report results suggesting the public regards polls and pollsters as generally credible.

While polls are a popular means of people learning about the distribution of opinion on public issues, many question whether, rather than reflecting public opinion, polls can actually influence public opinion (see, for example, Bogart, 1972; Gollin, 1980; Roll & Cantril, 1980).

Bandwagon Research

Early scholars and pundits termed the influence of actual or perceived majority opinion on election results as the "bandwagon" effect. A bandwagon effect occurs when the majority receives additional support from the publication of a poll; an underdog effect is just the opposite, in which people respond to an underdog's status and go against the majority or with the minority. This notion of a bandwagon effect has existed almost as long as public opinion polling, although much of the literature is simply speculative.













Though a number of elegant theoretical models have been published arguing for the effect (Gartner, 1976; Henshel & Johnston, 1987; H. A. Simon, 1954; Straff in, 1977; Zech, 1975), the literature is sprinkled generously with reports of null findings (Beniger, 1976; Dizney & Roskens, 1962; Fleitas, 1971; Lang & Lang, 1984). Adding to the confusion is the discovery of an "underdog" effect (Ceci & Kain, 1982; Laponce, 1966).

As discussed in the previous chapter, early in this century congressional efforts to control straw ballots failed. Even the news media, a major user of polling today, argued for stricter regulation. A 1936 New York Times editorial, for example, cautioned that polls often "develop a bandwagon rush" and "in cases where public opinion is less definitely rooted, and the allegiance of large groups of voters is less definitely won, a 'bandwagon' tendency may play a decisive part in determining the result of an election" ("Straw Ballots," p. 22). Similar arguments were made by others (F. D. Allport, 1940; Perry, 1968; Stoler, 1986). Early pollsters brought forth more systematic though problematic evidence. Directly following Pierce's 1940 condemnation of polls in Public Opinion Quarterly, for example, Gallup and Rae (1940) presented sequential survey evidence finding no bandwagon effect.












Research in public opinion poll influence typically takes one of two methodological directions. The first is the use of survey, panel, or trend survey data to uncover opinion movement; the second is the use of experiments or quasi-experimental methods in laboratory or field settings. First we review the survey evidence, then that from experiments.

Survey Evidence

One popular survey method is to ask respondents what influenced their decisions and whether, more specifically, polls had some influence (Hastings & Hastings, 1990; Roshwalb & Resnicoff, 197l)--a self-report method not without criticism as to whether individuals are aware or can accurately describe what influenced their opinions (Nisbett and Ross, 1977). Berelson, Lazarsfeld and McPhee (1954) found only about 10 percent mentioned polls as a source of information about changes in who they expected to win the presidential campaign. In asking directly about poll influence, Hastings and Hastings (1990) found between 3 and 11 percent of those polled willing to admit such influence. Roshwald and Resnicoff's (1971) survey of 1,000 respondents found no voluntary mentions of polls as being influential in a vote decision and, when asked, 66 percent of the respondents said polls were not at all important in their decision-making.













Other survey research has attempted to track effects of polls through panel or trend studies. one of the earliest discussions of the effect on voting by Lazarsfeld, Berelson, & Gaudet (1948) found many voters attempting to sense the direction of public opinion and the outcome in order to vote "with the winner." Follow-up work (Berelson, Lazarsfeld, and McPhee, 1954) revealed that a bandwagon effect (perception guiding preference) and a projection explanation (preference guiding perception) carried about equal weight in presidential election voting.

In another use of survey methodology, an extensive study by Beniger (1976) analyzed data from the date of gathering and publication of Gallup candidate preference polls and state primary elections from 1936 to 1972. Essentially, Beniger's sequential method was looking for whether a poll influenced the primary, or vice versa. In 83 of 183 preference polls followed by a subsequent poll, the ratings of the leading candidate increased from the first to the second poll, while in 81 cases it fell. Like many other sequential poll studies (Kavanagh, 1981; Roll & Cantril, 1980; McBride, 1991), Beniger found no consistent bandwagon effect and even hints of an equally inconsistent underdog effect, though others have successfully demonstrated more indirect poll effects such













as those on campaign contributions, volunteer work, or endorsements by key political actors (Henshel & Johnston, 1987).

Non-experimental cross-sectional evidence cannot

resolve the crucial question of causality. Trend studies provide a slightly clearer picture, but the results do not support the intuitive appeal and theoretical justification of a bandwagon effect of poll influence. Other methodologies have not improved matters. Experimental and Quasi-Experimental Evidence

Given the weaknesses noted above, Campbell (1958) asserted "the question of a bandwagon effect resulting from publication of poll results can only be settled by experimentation" (p. 252). Even controlled settings have not been kind to the bandwagon thesis, however, with findings as mixed as those from survey research.

To investigate the question, Campbell suggested a national manipulation of poll results, a difficult and costly task in the 1950s and one far too expensive--if even possible -- today. A similar but less ambitious study reported by Gallup (1972) compared cities with and without newspapers carrying the Gallup Poll and found no differences in candidate preferences. No further details were provided.












An early experiment by Cook and Welch (1940) used college students and as a stimulus the infamous 1936 Literary Digest poll showing Alf Landon leading Roosevelt in the presidential election. Students informed about the Digest poll were more likely to favor Landon than those in the control group not told about the poll. A second experiment conducted over a longer time period produced results in the direction of a bandwagon prediction, though the results were not statistically significant. These results led Cook and Welch to suggest any bandwagon effect may be a small one, reminiscent of Robinson's (1937) cautioning that the bandwagon thesis contained "more an element of fancy than fact" (p. 49).

Katz (1972) argued that it was not sheer numbers from polls that resulted in influence, "but the character of the majority in relation to the individual's psychological group membership" (p. 24). That is, a poll of similar others was thought to have more influence than a poll of a more general public. Atkin (1969) found some support by using "student preferences" on one issue, though no effect was found on two other issues.

Two British studies favored an underdog effect on likelihood to turn out to vote, indirectly affecting results. The first, cited in Teer and Spence (1973), varied information provided to three treatment groups of












400 voters before an election. There was no effect on preference, though supporters of the party they were told was behind at the polls said they were more likely to turn out to vote. In the second study, Gaskell (1974) varied information given about who was ahead in the polls, but rather than a no-information control group, he told respondents the polls were close. Fewer supporters said they would vote if a clear-cut victory was predicted, but this finding was stronger for supporters of the party predicted to win. However, deBock's (1972) investigation of the one-sided 1972 presidential election with a field experiment suggested the trailing candidate may suffer loss of support intensity and turnout motivation after supporters have been exposed to election poll results.

Navazio (1977) divided a sample of 500 persons into two groups, one a control and the other an experimental group that was told of bogus national poll results. The mail survey found no poll effect. In a similar vein, Roper, in a study reported by Cantril (1980), conducted a split-ballot trial with presidential approval ratings by introducing half of the questions with "As you know, all the polls have been showing support for Carter going down. We'd like to get your opinion about him." Few technical details were provided by the author, though it was












reported that the experimental manipulation resulted in only a 1 percent movement from Carter.

Ceci and Kain (1982) asked students their preferences for Carter and Reagan in the 1980 presidential election during morning classes. Before being asked, some sections of the class were told a survey of "college-educated persons" showed support for Carter (first section of the class), Reagan (second section), or no result (third section). Later that night, confederates posing as national pollsters telephoned students and requested a presidential preference. Prior to this, students were again provided bogus poll data. Despite the methodological problems of this study's use of intact groups, the results point to movement toward the underdog.

A quasi-experimental panel study by Cloutier, Nadeau, and Guay (1989) found 75 percent of their 1,005 subjects had the same attitude over a one-month period. However, 14 percent changed in the direction of the poll manipulation, while 11 percent went with the underdog or minority view.

Fewer studies combine experimental control with a

representative sample. A two-wave national panel survey conducted during the 1988 Bush/Dukakis presidential campaign by Lavrakas, Holley and Miller (1991) included an experimental manipulation of informing respondents about












actual public opinion poll results. Some interesting results were obtained. Persons who did not-graduate from high school and who were told that Bush was leading in the polls showed a significant increase in uncertainty about which candidate they would vote for compared to those with similar education who were not told Bush was leading. No effect was seen on those with higher education. The authors conclude better-educated persons were more resistant to the external source of influence.

Summarizing the experimental research, Marsh (1983) concluded no experimental evidence exists for a bandwagon hypothesis, though there is some support that a perception of growing strength may aid in the bandwagon process.

Conformity Research

In a related field of study, research on conformity has long documented the influence of the majority on the minority in small-group settings (Allen, 1965; Kiesler & Kiesler, 1969). The majority's superior size, power, and status provides a firm base for establishing social reality and a position in which to reward those who agree (Festinger, 1950). Recent findings have led many to question whether the minority must be willing and helpless recipients or whether a minority cam instead bring about modification and change (Moscovici & Personnaz, 1980), speculation not unlike the underdog thesis. It has even












been suggested that majority and minority influence processes differ (Moscovici, 1980; Nemeth, 1987), though many others maintain a single process is in effect (Doms, 1983; Latane & Wolf, 1981; Wolf, 1987). Latane's (1981) social impact theory is one example of single process argument, proposing that influence by either a majority or minority is a multiplicative function of the strength, immediacy, and number of its members.

Typically these studies rely on a single confederate who disagrees with the majority on some perception task, the results showing a persistent minority can influence both the public and private responses of a majority. In one study, Moscovici & Personnaz (1980) presented subjects a series of blue slides consistently labeled as green by the confederate. on each trial, subjects were required to indicate the color of the slide and the color of the afterimage perceived on a screen following removal of the slide. The confederate was presented as either a member of the majority or minority. The minority viewpoint was found to influence subjects' judgments of the afterimage. The authors conclude that the persistent minority causes the majority to start a validation process by considering the deviant response and its reasoning. "In other words, as a result of trying to see or understand what the minority saw or understood, the majority begins to see and












understand as the minority would" (Moscovici & Personnaz, 1980, p. 272).

It appears the minority, at least in these

experimental settings, can indeed influence the majority, confirming Asch's original suggestion (Asch, 1951) that some persons distort their perceptions of an unambiguous stimuli as a function of influence of others. In Moscovici's (Moscovici, 1980) theory, a consistently stated minority opinion can lead to careful scrutiny by the group, a detailed processing of the minority position similar to that of systematic or central processing. This influence is held to be latent, indirect, or covert, an internalized change of opinion not necessarily revealed publicly. Influence by the majority is held to bring about manifest, direct, or overt influence, what researchers call public compliance.

One study in this paradigm stands out as closest to the bandwagon question (Kaplowitz, Fink, D'Alessio, & Armstrong, 1983). Subjects were presented six issue statements with different bogus poll results about those issues. Subjects were either told they would either have to discuss their attitudes publicly or were given a private response option. No conformity effect was found for those who reported high commitment to their attitudes about the issues, even those responding privately.












However, the authors did find the bogus-poll technique could produce a conformity effect on those with little commitment to their attitudes. A previous study (Tyson & Kaplowitz, 1977) concluded that if people respond privately, results of public opinion polls will not directly influence responses to issues on which they have strong commitment to their views.

The Poll Bandwagon--A Summary

Throughout the accumulation of these assorted

findings on poll influence, a number of factors have been suggested for opinion movements, including perception of majority trends (Marsh, 1984), the strength or unexpectedness of these perceptions (Aronson, Turner, & Carlsmith, 1963), and the commitment to an opinion (Kaplowitz, Fink, D'Alessio, & Armstrong, 1983). In some research a bandwagon effect is found, in others an underdog effect. Some studies find no results, while others find both effects occurring.

Why the difficulties? Merkle (1991) suggests a

number of methods of uncovering the underlying process: more powerful experimental designs, care in looking for both bandwagon and underdog effects, assessing or controlling for previous perceptions, analysis of the relevance of reference groups, the kinds of topics used as a stimulus, investigation of personality characteristics,









20


and consideration of how subjects interpret the poll results.

Another way to approach the question is from a single theoretical perspective that attempts to consolidate findings in persuasion research in general and to apply that to poll influence studies in particular. The next chapter focuses on the work by cognitive response analysis theorists and two specific and complementary models of persuasion.















CHAPTER 3
PERSUASION RESEARCH


Soon after Gordon Allport declared attitudes to be "the most distinctive and indispensable concept in contemporary American social psychology" (Allport, 1935, p. 798), the construct blossomed as a research focus. Motivated by World War II, early scholars in social psychology investigated how persuasion and propaganda influenced attitudes (Hovland, Lumsdainef & Sheffield, 1949; Hovland, Janis, & Kelley, 1953).

The framework described by Lasswell (1948a) and

elaborated by Hovland et al. (1953) most closely resembles the work of cognitive response theorists and is the theoretical direction taken in this dissertation. It has enjoyed remarkable growth in the past 10 years. The early approach focused on "who says what to whom with what effect," the investigation of the communicator, the message, the audience, and the effects of attitude change.

Cognitive Response Analysis

cognitive response analysis delves further into the audience, arguing that attitude change is primarily a function of an individual's personal elaboration of external stimuli (Fiske & Taylor, 1984; Petty, Ostrom, & Brock, 1981). A stimulus causes a cognitive effect, which












results in a response. Variables in this research include audience involvement, message repetition, message comprehension, and communicator credibility.

Critical to the cognitive response perspective is

audience involvement or personal relevance, the variable most mentioned as a determinant of processing styles. Research in the social judgment perspective shows high involvement to be associated with less attitude change, while the cognitive response to persuasion model shows just the opposite, a contrast noted by Petty and Cacioppo (1979a) and explained by differences in conceptions of involvement (O'Keefe, 1990). Without involvement, social cognitive theorists argue, an individual will not use thoughtful, central, systematic information-processing strategies and instead will rely on peripheral, heuristic, or shortcut strategies that depend heavily on such irrelevant cues as communicator characteristics (Chaiken, 1980; Petty & Cacioppo, 1979b). Simply put, less involved persons are more likely to operate "on automatic," while highly involved persons will operate in a more "controlled" mode (Fiske & Taylor, 1984, p. 354).

This research into the two distinct styles of making social judgments is best exemplified by the elaboration likelihood model (ELM) of persuasion and the heuristicsystematic model (HSM) of persuasion (see Petty &













Cacioppo, 1986a or 1986b, for more complete descriptions of the ELM. For a similar view on HSM, see- Chaiken, Liberman, & Eagly, 1989).

Elaboration in the ELM is generally described as "the extent to which a person thinks about the issue-relevant argument contained in a message" (Petty & Cacioppo, 1986b, p. 128). In their conceptualization, when elaboration likelihood is high, more cognitive resources should be used by the receiver. Use of issue-relevant elaboration typically results in the new arguments, or a personal translation of them, being integrated into the underlying belief structure or schema for the attitude object. Current social psychological research provides support for two general modes of processing discussed above, termed by these theorists as central and peripheral routes to persuasion.

In this view, the likelihood of elaboration depends on a person's motivation and ability to evaluate the communication. motivation is influenced by such variables as personal relevance or personal responsibility, while ability to process can be influenced either by personal characteristics such as prior knowledge or by such variables not related to the person as message comprehension or distraction in the communication situation. With both motivation and ability present,













elaboration likelihood is high. With only one or neither present, elaboration likelihood is low.

This model presents a continuum from no thought about the issue-relevant information to complete elaboration on the issues. On this theoretical continuum's elaboration side is the central route to persuasion, on the opposite end the peripheral route. The former has persuasion achieved through close scrutiny to message arguments and consideration of other issue-relevant material, the latter is conducted through use of some simple decision rule or heuristic principle to evaluate an advocated position. Persuasion through the central route is posited to be long lasting, while persuasion through the peripheral route is relatively temporary. Examples of heuristics are number of arguments presented, use of expert sources, communicator credibility, likability, and attractiveness. The two routes are not mutually exclusive; rather they represent the extremes of the elaboration likelihood continuum. For example, a mixture of central and peripheral processes are theorized to occur at moderate levels of elaboration (Petty & Cacioppo, 1986a).

A series of experiments by Petty, Cacioppo, and their colleagues show that high relevance increases the likelihood of issue-relevant thinking and invoking the













central route to persuasion (Petty & Cacioppo, 1979a, 1979b, 1984; Petty, Cacioppo, & Goldman, 1981).

The HSM (see Chaiken, 1980, 1986, 1987; Chaiken & Eagly, 1983) similarly views heuristic processing as involving more limited cognitive effort and capacity. It goes further, however, by describing heuristics as learned knowledge structures that may be used by social perceivers. Such rules can range from "Experts' statements can be trusted" to "If most people think so, it must be right." In general, both models agree that two modes of processing exist, and that heuristics such as source expertise or consensus can lead to less systematic processing of arguments.

Persuasion and Individual Differences

The Petty, Cacioppo, and Goldman (1981) study of the effects of argument strength and communicator expertise on persuasive effectiveness provides a clear example of the cognitive response perspective. Subjects were presented a message with an issue that was either relatively involving personally (designed to prompt high elaboration) or not involving (presumably to cause less elaboration). Personal relevance or involvement was manipulated by making receivers believe the recommendations in the message for comprehensive examinations for senior undergraduates were being considered for adoption at their












university the following year (high involvement) or 10 years later (low involvement). Also manipulated were argument quality (strong versus weak) and the expertise of the communicator (high versus low). For those in the high-relevance condition, strong arguments produced more attitude change than weak arguments, but source expertise did not affect persuasion. In the low-relevance condition, subjects were more persuaded by an expert than by a non-expert source, but the quality of the arguments in the message did not affect subjects' attitudes. It appears that when issue relevance is high, people are more persuaded by message content (the central route); but when relevance is low, they are more persuaded by attributes or cues (peripheral route).

But is this always the case? Little work has been conducted on individual differences in the effects of personal relevance, though Cacioppo, Petty, and Morris (1983) demonstrated that high need for cognition, defined as the tendency for people to engage in and enjoy thinking, resulted in more central processing than for those low in the need for cognition. Need for cognition was not found to interact with the manipulations and only enhanced the likelihood for central processing. Others questioned whether personal relevance might actually reduce the likelihood for elaboration for some people










27


(Sorrentino, Bobocel, Gitta, Olson, & Hewitt, 1988). Uncertainty orientation, identified by Sorrentino, Short, and Raynor (1984), is one such possible dimension.















CHAPTER 4
UNCERTAINTY ORIENTATION


Uncertainty orientation is an individual difference occupying a niche called the "warm look," an interaction and interdependence of motivation and cognition to explain behavior (Sorrentino & Higgins, 1986; Sorrentino & Short, 1986). The construct remains in an early state of development and researchers have concentrated less on a definition than on describing those who are certainty and uncertainty-oriented, conceding "we are still learning about them ourselves" (Sorrentino & Roney, 1990, p. 242).

This chapter describes the theoretical and

operational rationale behind the construct and sketches the research to date.

Theoretical Discussion

Definitions proposed for uncertainty orientation (and as we will see later, the measurement itself) have changed as well, though Sorrentino maintains the variable is a cognitive rather than motivational one (Sorrentino & Short, 1986). For example, need for cognition (Cacioppo & Petty, 1982) has been described as a variable measuring motivation to think, while uncertainty orientation is a measure of when to think (Sorrentino, et al., 1988).












uncertainty orientation has been described as the

degree to which situations of certainty versus uncertainty are cognitively relevant (Sorrentino, Short, & Raynor, 1984); as involving the degree of uncertainty surrounding the outcome of one's activity (Sorrentino & Short, 1986); as how one orients him or herself toward the uncertainty concerning the self or the environment (Sorrentino & Hancock, 1987); as people's relative interest in either maximizing information gain or maintaining clarity (Roney & Sorrentino, 1987); and as an individual's approach to uncertainty or ambiguity (Sorrentino, et al., 1988). In general, uncertainty orientation is the concern or interest in finding out new things about the self and the environment and the tolerance of ambiguity about the self or environment.

What should interest mass communication scholars is Sorrentino's contention that uncertainty orientation is primarily concerned with information value and the maintenance and preservation of current knowledge. Interest in the variable arose from Atkinson and Raynor's (1974) theory of achievement motivation and replies from cognitive informational theorists (Trope, 1975; Weiner, 1972). In short, it was argued that situations that attain clarity about the self result in the greatest arousal of achievement-related motives for uncertainty-












oriented people, and situations that maintain clarity lead to the greatest arousal of these motives for certaintyoriented people. one series of experiments showed the theory of achievement motivation only held for uncertainty-oriented persons (Sorrentino, Short, & Raynor, 1984), that is, certainty-oriented people demonstrated no interest in using a diagnostic achievement situation to learn about themselves.

Generally, it is posited that the uncertaintyoriented person seeks to attain clarity about the self and the environment, while the certainty-orientated individual seeks to maintain present clarity about the self or the environment. To attain clarity, one is willing or likely to deal with information that may even threaten the selfperception or challenge beliefs. Maintaining clarity is just the opposite, the tendency to hold on to beliefs and to avoid situations that are self-diagnostic. Attaining or maintaining clarity is done through use or failure to use or process information that is available. Historical Perspective

The prototype for uncertainty orientation is traced to Rokeach's The Open and Closed Mind (1960), which presents a continuum of "gestalt types" to "psychoanalytic types." Rokeach characterized the former as having a "need for a cognitive framework to know and understand"












and the latter as having the "need to ward off threatening aspects of reality" (Rokeach, 1960, p. 67)_ Thus the open-minded person possesses a belief system oriented toward the new, the closed-minded a belief system oriented toward the familiar, a dichotomy similar to the Freudian notion of basic trust and mistrust of the world. The more closed the system, according to this argument, the more difficult it is to distinguish between information received about the world and information about the source--a perspective not unlike that of persuasion theorists discussed in the previous chapter that emphasizes elaboration of issue matter versus influence by source characteristics.

Rokeach's work included the famous "Joe Doodlebug" experiments to explore differences between those Rokeach called "closed" and "open" persons. The experiments showed that persons with "closed" systems had more difficulty with integrating new beliefs and often showed early closure. These results were not associated with intelligence or ideology.

Sorrentino also notes the influence of Kagan (1972), who described uncertainty resolution as one of four primary determinants of behavior. The motive to resolve uncertainty, he writes, is the "wish to know" (p. 54). While not discussing individual differences, Kagan












suggested that people in a state of uncertainty will either assimilate, remove, or escape from an event that is discrepant from an established schema. Certainty-oriented persons would attempt to remove or escape such discrepant information, while uncertainty-oriented persons would attempt to assimilate such information. cognitive Perspective

The term schema shows up in Sorrentino's work as

defense for uncertainty orientation as a largely cognitive variable with motivational roots. Building on the work of construct assessability (Bargh 1984; Higgins & King, 1981), King and Sorrentino (1988) found that uncertaintyoriented and certainty-oriented persons differ in the chronic accessibility of positive and negative constructs reflecting certainty (e.g., "cautious" and "stubborn") and uncertainty (e.g., "adventurous" and reckless").

In perhaps the most complete explanation of the construct, the authors write that their model assumes

that uncertainty-oriented persons are those
who have been rewarded for autonomous
exploratory behavior. Such persons develop
general schemes for situations that allow
resolution of uncertainty about the self and
the environment. When placed in these
situations, they are effectively charged to
resolve this uncertainty. Certainty-oriented
persons, on the other hand, have not been
rewarded for autonomous, exploratory behavior and in fact might have been punished for such
behavior. Consequently, these persons
gravitate toward and develop schemes for safe
and familiar situations or schemes that do












not require dealing with uncertain
situations. Hence, they are effectively
charged by situations involving certainty and
the self and the environment (Sorrentino, et
al., 1988, p. 358)

An "affective charge" closely resembles the language of motivation scholars who describe motivation as the drive that energizes, orients, and guides behavior. Uncertainty-oriented individuals will be motivated in situations that involve uncertainty about the self or the environment, while certainty-oriented persons should be more motivated in situations that involve certainty about the self or the environment, setting into motion respective cognitive responses.

Findings in Uncertainty Orientation Research

The initial test of this hypothesis by Sorrentino et al. (1984) found that for uncertainty-oriented persons, relevant sources of motivation were most aroused in performance situations that contained uncertainty about the self or the environment. The reverse was found for certainty-oriented persons. The authors argued their findings integrate existing cognitive and motivational interpretations of achievement behavior.

This first study was aimed at explaining problems

arising from motivational research. In work more closely related to social influence, Sorrentino & Hewitt (1984) tested whether there would be differences in how












uncertainty-oriented and certainty-oriented people would approach information that was personally re-levant. As predicted, uncertainty-oriented persons chose activities aimed at resolving uncertainty about a new and potentially important ability, whereas certainty-oriented persons actually chose an alternative activity that would tell them nothing new about this ability. This finding held even when the information was likely to be positive.

Again studying personal relevance, Sorrentino and

Roney (1986) demonstrated that motivational arousal would occur in situations relevant to a person's uncertainty orientation. Achievement-related motives of uncertaintyoriented persons were most aroused in a diagnostic task about a new and possibly important ability (mental flexibility), but those of certainty-oriented persons were most aroused in a non diagnostic task.

Authoritarianism

An important aspect of uncertainty orientation is use of authoritarianism, a construct elaborated by Adorno, Frenkel-Brunswik, Levinson, & Sanford (1950) and ably criticized by Shils (1954) as dealing only with authoritarianism of the "right." The preoccupation is understandable, however, given the political climate of












the 1940s. Still, most of the early questions measuring the construct were directed at the political right.

One of the earliest and most influential studies linking psychological and political concepts, The Authoritarian Personality, identified a group of individuals who possessed anti-Semitic attitudes and then expanded this view to generally ethnocentric aversions for all people. Theoretical development of the F-scale (for fascism) was drawn largely from Freudian concepts of superego, ego, and id, and was traced to childhood development (Sanford, 1971). Authoritarians were found to display considerable cognitive rigidity and intolerance for ambiguity (Block & Block, 1951; Frenkel-Brunswick, 1949; Jones, 1954; Rokeach, 1948; Steiner & Johnson, 1963), to believe others thought the same way they do (Granberg, 1972; Scodel & Mussen, 1953; Simons, 1966), to favor conservative political attitudes (Adorno, et al., 1950), and to vote for conservative or authoritarian candidates (Higgins, 1965; Milton, 1952; Poley, 1974; Wrightsman, 1965).

Other Personality Theories

There is a long history of attempts to provide an

exhaustive taxonomy of personality (Cattell, 1943; Digman, 1990; Fiske, 1949; Norman, 1963). Some consensus for a "Big Five" model appears to be emerging (Digman, 1990).












Of these five factors, one dimension appears closely related to uncertainty orientation. This dimension goes by a number of labels, but perhaps can best be described as an openness to experience (Costa & McCrae, 1985; Coan, 1974).

Another similar personality dimension is repressionsensitization (Byrne, 1964), where persons are thought to deal with threatening stimuli in one of two ways--a person who approaches threats is a sensitizer, a person who avoids threats is a repressor. Repressors focus on supportative information and avoid information that is not supportative (Olson & Zanna, 1979). On the Study of Personality. Politics, and Communication "To talk about politics without reference to human

beings," Lippmann wrote, "is just the deepest error in our political thinking" (1913, p. 2). Early work by Lasswell emphasized the need to bring psychology into political analysis (Lasswell, 1930, 1948b, 1951). Yet personality variables, which indicate an organized, stable internal predisposition a person brings to a situation, have not enjoyed a great deal of success in political communication research. Even supporters of the theoretical perspective concede personality variables are unlikely to account for much of the variance when looking at such behaviors as joining a major political party or in voting, though












evidence exists that such variables are "selectively felt at a number of points in the political process" (Knutson, 1973, p. 44).

Much of this "vast and uneven work" falls under the externalization hypothesis, which treats political opinions as externalized versions of the mind's inner conflicts (Kinder & Sears, 1985, p. 676). Sorrentino argues for extending the work in individual differences from the achievement area to that of social influence and persuasion. Reviewing the work of persuasion theorists, Sorrentino and others questioned whether uncertainty orientation may interact with personal relevance in a persuasive context. As discussed earlier, research has typically shown that perceivers process information more thoughtfully when the subject has high rather than low personal relevance (Petty & Cacioppo, 1981; Fiske & Taylor, 1984; Nisbett & Ross, 1980). The authors replicated the work of persuasion theorists, primarily that of Petty & Cacioppo (1984) and Petty et al. (1981), to study the effects of personal relevance on attitude change as a function of one's uncertainty orientation. Sorrentino et al. (1988) predicted that, unlike uncertainty-oriented persons, high personal relevance would make certainty-oriented persons less careful or systematic in their processing of message arguments and












more dependent on heuristics, or persuasion cues, than would low personal relevance. The investigators found that personal relevance does not appear to increase systematic processing for all persons--that certaintyoriented persons are more motivated to consider a matter further when personal relevance decreases. These individuals do not prefer to reason matters out when facing a situation with uncertainty about themselves or their environment, the authors conclude. "When forced into these situations, they rely on heuristics rather than use their own judgment" (p. 368). This reversal of a principle "thought to be universal," they add, highlights the value of an individual-differences approach in social influence (p. 369).

The question is not one of merely varying the

strength of a persuasive effect, then, but varying the direction of the effect. "In other words, variables assumed to enhance systematic forms of information processing may work for some, but may actually dampen such tendencies in others" (Sorrentino & Hancock, 1987, p. 265).
















CHAPTER 5
SYNTHESIS AND HYPOTHESES


Bandwagon Research and Persuasion

Reviewing the bandwagon research from the perspective of persuasion theory offers some clues as to why the bandwagon effect occurs in some cases but not in others.

A key variable in this paradigm is personal

relevance. A look at many bandwagon and poll studies finds some influence in what we might suspect are low relevance issues, and little or no persuasive (or bandwagon) effects on what we might suspect are high relevance issues. For example, Atkin's (1969) use of a "student preference" poll found effects only on an income tax surcharge issue but not on the issues of marijuana or reopening the Warren Commission investigation (controversial matters at the time). While no measures of relevance are provided in the study, it seems likely that the tax surcharge issue was less important to his college student subject pool that the use of marijuana or the Warren commission issue. Atkin used a three-week period between measures, thus it may be that the poll had more influence on the less relevant issue but no effect on the higher relevance issues.












A look at Table 5-1 provides a look at some selected poll persuasion studies. Though drawn largely from Merkle (1991), the table organizes the list chronologically with bandwagon or bandwagon/underdog findings listed first, underdog-only effects listed second, and then null studies listed third.

Most null findings occurred in the 1970s, a time when belief in persuasive effects of mass communication were in their decline. Studies conducted in the 1980s have uncovered greater support for the bandwagon/underdog thesis, perhaps due to a reliance on experiments and quasi-experiments.

Looking closely at the table of the thirteen studies finding a bandwagon or bandwagon/underdog effect, only five were elections or candidate evaluations. Both of the studies finding only an underdog effect were elections, while seven of eight studies finding no poll influence were of an election or candidate evaluation. It appears elections offer less an opportunity to demonstrate a poll's influence. This hardly seems surprising if, as is suggested by the ELM and other persuasion models, people are more highly involved in presidential or other elections or hold more closely to candidate evaluations than they would an abstract issue, therefore less likely to be persuaded by a heuristic such as a poll result.














Table 5-1

Chronological Listing of Selected Bandwagon/Underdog and Consensus Persuasion Studies


Authors Subjects Topic Conclusion


Cook & Welch (1940)

Allard (1941)


Lazarsfeld, et al., (1944) and Berelson, et al., (1954) Atkin (1969) Roshwalb & Resnicoff (1971) de Bock (1972)


Meyers, et al. (1977)

Kaplowitz, et al. (1983) Marsh (1984)


Mackie (1987) Cloutier, et al. (1989) Lavrakas, et al. (1990)


College students

College students

General public



College students

General public

College students

Church members

College students

General public

College students

College students

General public


Pres. election Issues


Pres. election



Issues


State election Pres. election Issues


Issues


Abortion
issue Issues Issue


Pres. election


Some bandwagon support

Support for bandwagon effect

Support for bandwagon and underdog effect


Some support for bandwagon effect

Indirect support for bandwagon

Support for indirect bandwagon

Support for bandwagon effect

Bandwagon support on low-commitment

Trend produced bandwagon effect

Support for bandwagon/underdog

Support for bandwagon/ underdog

Support for bandwagon/underdog














Table 5-1--continued


Authors Subjects Topic Conclusion


Laponce (1966) Ceci & Kain (1982)


College students


Hypoth. election


College Pres. students election


Support for underdog effect

Some support for underdog effect


Dizney & Roskens (1962)

Sales Research Services Ltd. (1964) Cited in Teer & Spence (1973)

Fleitas (1971)


Gaskell (1974) Beniger (1976)


Tyson & Kaplowitz (1977)

Navazio (1977) Roper (1980) cited in Cantril (1980)


College students

General public




College students


General public

Gallup polls

College students

General public

General public


Pres. election

British election




Hypoth. election


British election

Primary elections


Issues


Attitudes on Nixon

Attitudes on Carter


No support for bandwagon/underdog

No support for bandwagon/underdog




No support for bandwagon/underdog


No support for bandwagon/underdog

No support for bandwagon/underdog

No support for bandwagon effect

No support for bandwagon/underdog

No support for bandwagon/underdog












Looking at issues, Kaplowitz, et al. (1983) found no conformity effect on what subjects described as highcommittment issues, but some conformity on low commitment issues, a finding that would be anticipated in the persuasion perspective. Atkin (1969) also found support for the bandwagon thesis, but only on one of three issues (one less likely to be relevant to respondents).

Consensus as a Heuristic cue

A number of different heuristic principles have been suggested in persuasion studies, but three have received the most research attention: the credibility, liking, and consensus heuristics (O'Keefe, 1990). The consensus heuristic can be broadly defined as "if other people believe it, then it's probably true" (O'Keefe, 1990, p. 107). Two recent studies using the cognitive response conceptual framework to investigate consensus effects arrive at different, though not contradictory, conclusions.

Axsom, Yates, & Chaiken (1987) studied audience

response as a consensus cue in order to clarify a history of mixed results, finding these reactions influenced opinions only under low involvement conditions; in high involvement conditions only argument quality influenced












opinions, a result similar to that of Kaplowitz, et al. (1983).

In a series of experiments more closely aligned to

this dissertation, Mackie (1987) investigated the kind of processing induced by consensus information. Mackie found subjects simultaneously exposed to a majority poll result with which they disagreed and a minority with which they agreed showed more issue-relevant processing of the majority message. She concluded that the majority provides validity to the arguments presented, directing attention to them and resulting in considerable processing. That is, disagreeing with a majority results in more attention being paid to the issue-relevant facts themselves, leading to systematic processing. majority status, then, can in some cases provide the motivation for recipients to systematically process communications in certain circumstances.

Summary and Hypotheses

Polls influence people--sometimes. The conceptual framework of the ELM and HSM provides one hope of unraveling these tangled findings. As noted by Marsh (1983), research on the bandwagon effect or poll influence in general has tended to focus on elections at the expense of issues. This dissertation will explore issues and poll influence using the persuasion paradigm, which typically












uses dependent variables other than a "yes" versus "no" or one candidate versus the other, instead relying on an attitude measure.

If an issue is personally relevant, then a consensus heuristic is predicted to have less influence. If an issue is not relevant, presentation of consensus should have more chance to sway attitudes. Following this, it is hypothesized that an issue poll will be more persuasive in low- rather than high-relevance issues.

Second is the matter of uncertainty orientation. In a series of speculations about the possible interactive effects of uncertainty orientation, Sorrentino and Hancock (1987) argue that uncertainty-oriented persons, exposed to consensus information, would try to discover why a minority insisted on its side of an issue in the face of a majority. Certainty-oriented persons would be more likely to ignore the inconsistency and "go along with the crowd" (Sorrentino & Hancock, 1987, p. 256). Here we come to the bandwagon versus underdog effect. If this theorizing holds, uncertainty-oriented persons should be less likely to be influenced by a poll or move in the "underdog" direction, while certainty-oriented persons should be more persuaded by poll.

Finally, based on the work of Sorrentino, et al. (1988), it is hypothesized that in situations of high












relevance, certainty-oriented persons threatened by such relevance would turn to a heuristic cue such as a poll and be more influenced by a poll than would uncertaintyoriented persons.

Overview of the Studies

This dissertation is organized around two studies. Study 1 largely focuses on the testing of issues for use in Study 2, though it also includes analysis of two experimental manipulations helpful in deciding how best to proceed with the second phase.
















CHAPTER 6
STUDY 1 METHODOLOGY AND RESULTS


Methodology

Choosing the Issues

Two steps were taken to decide what issues to use. First, students from four mass communication writing classes, of about 20 students each, were asked in fall 1990 and spring 1991 to list the three most personally important social or political issues.

In the first issue-listing phase, the Persian Gulf crisis received the most mentions (26), followed by abortion (15), the environment (14), AIDS (11), the economy (11), education (9), censorship, drugs, elderly and child abuse, and energy (each with 6). Next, from this list, issue statements were presented to two sections of a large introductory mass communications class for attitude and personal relevance measures.

Two questions accompanied each issue statement, the first asking for the subjects' attitude toward the issue, the second for the personal relevance of that issue. Both were measured on 1-to-5 scales as provided by the machineread answer sheets (bubble sheets). An issue attitude could range from a 1 for "disagree" to a 5 for "agree."












Relevance measures could range from 1-to-5 as well, a 1 representing "not relevant" and a 5 "very relevant."

There were two reasons for this phase of the research. First, two issues rated as low and high relevance by students themselves would be presented to the subjects in the later study. Second, those issues had to be relatively unrelated to uncertainty orientation. An association between the issues and uncertainty orientation might have occurred due to a number of factors, the most likely being ideological differences between certaintyoriented and uncertainty-oriented persons. Uncertainty Orientation

Measures of uncertainty orientation resembles those used to assess achievement-related motives (Atkinson, 1958; Atkinson & Feather, 1966). Frederick and Sorrentino (1977) developed a projective measure of n-uncertainty from which to infer to relevance of uncertain situations.

Early attempts to unconfound authoritarianism and

acquiescence response sets (Christie, Havel, & Seidenberg, 1958; Couch & Keniston, 1960) have led many to question the usefulness of the construct. Sorrentino and his colleagues adopted the measure of authoritarianism of Byrne and Lamberth (1971) to infer the relevance of certain situations. Previous research has shown that people high in authoritarianism are more intolerant of












ambiguity (Kirscht & Dillehay, 1967) and have less experience with uncertain situations (Kelman & Barclay, 1963). Interrater reliabilities for the n-uncertainty projective measure have been found to be above .90 by Frederick and Sorrentino (1977) and Sorrentino (1977) had similar results for the authoritarianism measure.

Generally, sentence leads are used for the projective test, with the open-ended responses coded for certainty and uncertainty imagery (see Fredrick & Sorrentino, 1977, for detailed scoring instructions).

The acquiescence-free measure of authoritarianism results was used with the n-uncertainty measures as opposing elements of uncertainty orientation. This resultant measure is used, according to the authors, because persons scoring low on n-uncertainty may not necessarily be high in certainty orientation, and vice versa. The method is not unusual in the motivational literature, for example, where two uncorrelated measures are used to measure achievement motivation (Atkinson & Feather, 1966). Briefly, z-scores are computed on the two measures for a resultant measure and a tertile split of the scores are used. The middle third is typically excluded from further analysis because of inconsistent results with this middle group.












Projective measures are time consuming and in some research settings, such as surveys, not feasible. A 35item questionnaire, still being validated with the projective measures, has been produced by Sorrentino and his colleagues and is used in the research reported here. This instrument contains two sets of questions. The first set is designed to measure need for uncertainty, using questions rather than the previous projective measure. The second set of questions is the same as the original instrument and is designed to measure authoritarianism.

Sorrentino and his colleagues believe the new

instrument essentially taps the same construct and initial testing is positive, though it is also thought that the pen-and-pencil measure does not have the explanatory power of the projective method (Sorrentino, R. M., personal communication, 1989, 1991).1 Independent and Dependent Variables

Following the issues on the Study 1 instrument were measures of media reliance, attitudes about poll credibility, importance of opinions of others, demographics, political ideology, and uncertainty orientation. Finally, two of the issue statements were presented again for attitude measures (comprehensive exams and agricultural pesticide reforms). Bogus poll results












were built into the issue statements and these versions randomly assigned to subjects.

one manipulation involved the poll sample. Some subjects received a version that said the poll was of college students, others received a version that stated the poll was of Americans. A second manipulation had subjects either being told a single poll result or a version that added that a trend was seen in the direction of the poll result. A third factor was dropped due to question wording problems, reducing the number of subjects available for analysis.2 To summarize, subjects received both issues (senior comprehensive exams and agricultural reforms in pesticide use) and were randomly assigned to one of four conditions: (1) Poll of Students; (2) Poll of Americans; (3) Trend-Poll of Students; and (4) Trend-Poll of Americans. Uncertainty orientation was added to this, making it a 2 (uncertainty- or certainty-oriented) X 2 (poll or trend and poll) X 2 (students or Americans as sample) analysis design. A control group was also included, these subjects receiving only the issue statements a second time with no poll results.

An example of the dependent variable question is one from the poll-student condition, which stated: "A public opinion poll shows college students favor university senior comprehensive exams to graduate, with 76 percent of













college students in favor. What do you think?" Subjects could disagree or agree on the same 1-to-5 scale as above. The second question in this condition stated: "The same poll shows college students in favor of agricultural reforms in pesticide use, with 78 percent of college students in favor. What do you think?" The slight difference in percentages of the bogus poll results was used to decrease suspicion. These issues differed significantly in relevance as reported by the subjects, with comprehensive exams being slightly higher in relevance (M = 3.9) than pesticide reform 3.6, t(190)

2.6, p<.01).

Results

One hundred and twenty-eight students in the two

sections of an introduction to mass communication class took part in Study 1 on March 5, 1991. Nearly 63 percent were female. A plurality listed themselves as advertising majors (45, or 36.3 percent). Issue Statements

Students were first provided the issue statements

(Table 6-1 contains the issue statements and Table 6-2 the attitude and relevance mean scores and the association of these scores with uncertainty orientation). Subjects were presented 17 issue statements followed by the attitude and relevance questions after each issue. A bogus













agricultural act received the lowest relevance score (M

2.4), while worries about the recession and-job availability produced the highest score (M = 4.6). Relevance scores differing by about one-tenth of a point are significantly different at the .05 level.

The recession/jobs issue met the criteria of both high relevance and apparently no relationship to uncertainty orientation. Of the "low relevance" issues, the best choice appeared to be the capital gains tax (M

2.9), also not significantly related with uncertainty Table 6-1

Issue Questions and mean Relevance and Attitude Scores



Relev. Attit.


1. Congress should pass the recentlyproposed Agricultural Act of 1991. 2.4 3.9

2. The United States can help its own
economy and that of poor nations with
increased use of the International
Monetary Fund. 2.9 4.4

3. The capital gains tax should
be changed. 2.9 4.4

4. Mandatory AIDS testing should be
required for some people. 3.1 3.1

5. Weak federal regulations are to
blame for the U.S. savings and
loan crisis. 3.1 4.1













Table 6-1--continued


Relev. Attit.


6. Reforms are needed in the use
of pesticides in agriculture. 3.7 3.0

7. Taxes are so high that they must
be cut even if it means a reduction
of public services. 3.7 4.4

8. Senior comprehensive exams for
graduating seniors is a good idea. 3.9 3.3

9. Women should have the right to an
abortion if they wish. 4.0 3.2

10. We should support the allied
efforts in the Persian Gulf War. 4.1 3.3

11. U.S. universities should toughen
education requirements to be
competitive in the world. 4.2 3.8

12. President Bush is doing a good job. 4.2 2.4

13. Americans need to sacrifice more
in order to cut our energy use. 4.2 3.0

14. We should do more to protect the
environment, even at the cost
of some businesses. 4.3 4.2

15. Jobs should go to the best person
regardless of race or sex. 4.6 2.5

16. The proposed University of Florida
tuition increase is not necessary. 4.6 4.0

17. The recession means good jobs vill be
hard to come by for the next fev years. 4.6 4.6



Note - Issues in boldface are those selected for use in
Study 2. Mean scores of approximately 0.1
difference are significantly different at the .05
level.













Table 6-2

Attitude and Relevance Scores by Uncertainty Orientation



Attitude Relevance


Issue C-0 U-0 t C-0 U-0 t


1.

2.

3.

4.

5.

6.

7.

8.

9.

10. 11.

12. 13.

14. 15. 16. 17.


4.1 3.0 3.2

4.2 3.3

4.0 2.6

2.4 4.2 4.6 3.3

4.2 4.1 4.2 4.6 3.4 4.1


4.1 2.7 3.2 3.6 3.3

4.0 2.0 2.8

4.4 4.3 3.0 3.7

4.1 4.5 4.4 3.1 4.1


-0.2

1.6



2. 5 0.1

-0.4

2. 4b

-1.4

-2.0

1.2 1.0

2. 4b

0.2

-1.4

1. 1 1.3

-0.2


2.5 2.7

2.8

2.8 3.0

3.4 3.4 3.7 3.5

4.1 3.9

4.0 4.0 4.0 4.3 4.4 4.5


2.3 3.1 3.0 3.3 3.1 3.9

4.0 4.2 4.4 4.1 4.4 4.3 4.4 4.3 4.7 4.6 4.7


1.0

-2.a

-1. 1

-1.9

-0. 5



-2.5

-2.a

-3 2

-0.2

-2. 0

-1.2

-2.a

-1.4

-1. 8

-0. 8

-0.9


Note - C-0 is
oriente


certainty-oriented, U-0 d. a p<.05; b p<.01; c


is uncertaintyP<.001












orientation. These two issues were significantly different in personal relevance (t(191) = 20.1, R<.001) and were selected for further use in the later study. Uncertainty Orientation

A subset of the questions provided by Sorrentino and his colleagues was used after factor analyses in bothStudy

1 and Study 2 were inconclusive. The subset of six need for uncertainty and six authoritarianism items were used (a more detailed discussion of the factor analyses is provided in Chapter 7).

The need for uncertainty scale had a mean of 4.0 (SD

0.7) and the authoritarianism scale a mean of 2.8 (SD

0.7). As expected, the two had a significant negative association (.r = -.22, p<.001). The need for uncertainty measure demonstrated fair internal consistency (standardized Cronbach's Alpha of .77). The authoritarianism scale was less internally consistent, with a Cronbach's Alpha of .59.

To create a resultant uncertainty orientation measure, scores on the authoritarianism scale were subtracted from the need for uncertainty scale, resulting in a scale that ranged from -1.5 to 3.5 (M = 1.3, SD 1.1). Individuals high in this resultant measure are considered uncertainty-oriented, while those low are considered more certainty-oriented. As noted in earlier












discussion of the construct, the middle textile is typically dropped from further analysis, leaving the extremes of uncertainty-oriented and certainty-oriented persons but reducing the number of subjects available. This procedure was not done in this research. Instead, the middle textile is included to keep these subjects in the analysis.

Uncertainty-oriented subjects were more liberal (M

3.7) than certainty-oriented subjects (2.7, t(82) = - 3.9, p<.001) However, uncertainty-oriented persons did not differ from certainty-oriented persons on newspaper or television reliance, the credibility they placed on polls reported in newspapers, or in the importance they place in other people's opinions.

Assumptions

Analysis of variance was used to test for significant interactions. There are three main assumptions in the use of analysis of variance: that there be normally distributed treatment populations, homogeneity of error variance, and independence of error components (Keppell, 1982). The F-distribution is robust against violations of the first two conditions, though such violations are more serious in situations with unequal cell sizes. In general, Monte Carlo studies have shown large sample sizes will offset any moderate violation of normality, with












significance usually misinterpreted by the F-test by only a small amount (for example, a reported significance of .05 when actually .06).

Various statistical tests are available to compare variances (e.g., Hartley, Cochran, and Bartlett), though as Keppell (1982) notes, many share in common a sensitivity to departures from normality as well as homogeneity of variance. That is, though experimental groups may demonstrate homogeneity of variance, a lack of normality will result in these homogeneity statistical tests reporting a lack of homogeneity among groups. Use of the SPSS-X statistical package confines the tests to the Bartlett-Box F, Cochran's C, and Hartley's F max. (SPSS-X, 1988). Examination of Study 1 results revealed no threats to the assumptions of homogeneity of variance or normality. This allows us to confidently examine the unequal cell sizes found in this study. The third assumption, independence of error components, means that treatment groups are in no way related to any other observation in the experiment. Random assignment, used in the experiment reported here, achieves independence.

Tests of statistical significance are at the

traditional probability level of .05, though some results below the .10 level are reported when they provide insight into relationships.












Attitudes on Two Issues

As noted earlier, a pair of issues were selected to help study the effects of two poll story manipulations. The first manipulation had subjects receiving either a single-shot poll result or a poll with additional information that the trend of opinion was also in that direction. The second manipulation was of the poll sample, with the poll either being of Americans in general or of students. A control group with no manipulations was asked for an opinion.

No significant interactions were found on the

comprehensive exams issue (see Table 6-3 for the F-table, Table 6-4 for the cell means). Only one near-significant two-way interaction was found on the agricultural pesticides issue, that between uncertainty orientation and the student versus American poll manipulation (.E(2,81)

2.5, p<.09). This result offers little in terms of interest since it is apparently the "middle textile" group causing the relationship.

No main effects were found between the types of poll manipulations. However, those in the control group showed less agreement with pesticide reforms (M = 3.6) than those receiving any of the poll conditions (4 = 4.1, t(116) = 2.3, R<.02). However, when looking at the more relevant comprehensive exams issue, no significant difference was













found between persons in the control group and those receiving some kind of poll condition. In general, comparisons with the control group showed poll influence only on the less relevant issue (see Table 6-7).

Summary

Two issues -- jobs during a recession and the capital gains tax -- were selected for Study 2 based on the requirements that they differ in relevance and that the


Table 6-3

F-Table of Analysis of Variance of Poll Manipulations on the Agricultural Pesticides Issue



Sum Of Mean Sig.
Source of Variation Squares DF Square F Of F


Main Effects 3.5 4 0.9 1.0 .39
1 Poll vs. Trend 0.4 1 0.4 0.5 .50
2 Students vs. Americans 0.1 1 0.1 0.1 .75 3 Uncertainty Orientation 3.0 2 1.5 1.8 .18 2-Way Interactions 5.5 5 1.1 1.3 .27
1 By 2 0.4 1 0.4 0.4 .51
1 By 3 0.8 2 0.4 0.5 .62
2 By 3 4.2 2 2.1 2.5 .09

3-Way Interactions 0.8 2 0.4 0.4 .62
1 By 2 By 3 0.8 2 0.4 0.4 .62

Explained 9.9 11 0.9 1.1 .40

Residual 68.3 81 0.8

Total 78.1 92 0.8













attitude and relevance measures not be associated with uncertainty orientation. Study 1 included a pilot experiment investigating whether using students as a sample in the bogus poll would be more influential than Americans in general, and whether a report of a single poll or poii with reports of a trend would be more influential. Few differences were expected given the mild manipulations.


Table 6-4

Attitude about Agricultural Pesticide Reforms by Poll Manipulations and Uncertainty Orientation.


Poll Trend


Cert. -Oriented 4.0 4.2
Students Polled N = 10 N = 9

Moderate U-0 3.8 3.2
Students Polled N = 4 N = 5

Uncert.-Oriented 4.4 4.5
Students Polled N = 10 N =8


Cert.-Oriented 4.8 4.1
Americans Polled N = 4 N = 7

Moderate U-0 4.2 3.8
Americans Polled N = 9 N = 10

Uncert.-Oriented 4.0 4.0
Americans Polled N = 10 N = 7


Note. Interaction F(2,81) = 0.4,ns


ns.













Table-6-5

F-Table of Analysis of Variance of Poll Manipulations on the Comprehensive Exams Issue




Sum Of Mean Sig.
Source of Variation Squares DF Square F Of F


Main Effects
1 Poll vs. Trend
2 Students vs. Americans
3 Uncertainty orientation

2-Way Interactions
1 By 2 1 By 3 2 By 3


3-Way Interactions
1 By 2 By 3


9.2
4.3 1.4 4.2

5.2 0.9 1.1 3.0


1.4 2 1.4 2


2.3
4.3 1.4 2.1

1.0 0.9 0.5 1.5

0.7 0.7


1.1
2.0 0.7 1.0

0.5
0.4 0.3 0.7


.38 .16
.42 .38

.79 .52
.78
.49


0.3 .71 0.3 .71


Explained Residual


15.9 11 1.4


173.1 81


0.7


.76


2.1


Tal189.0 92 2.1


Total












Table 6-6

Attitude about Senior Comprehensive Exams by Poll Manipulation and Uncertainty Orientation.



Poll Poll/Trend


Cert.-Oriented 2.1 2.2
Students Polled N = 10 N = 9

Middle Tertile 2.3 3.0
Students Polled N = 4 N = 5

Uncert.-Oriented 2.6 2.6
Students Polled N = 10 N = 8


Cert.-Oriented 2.0 3.3
Americans Polled N = 4 N = 7

Middle Tertile 2.0 2.6
Americans Polled N = 9 N = 10

Uncert.-Oriented 2.9 3.1
Americans Polled N = 10 N = 7


Note. No significant interaction.












Table 6-7

Mean Attitudes (SD) Toward Two Issues by Podl/Trend and Student/American Manipulations



Type of Issue


Comp. Exams Pesticides



Poll 2.4 (1.4) 4.2 (0.9)

Trend 2.7 (1.5) 4.0 (1.0)

Students 2.4 (1.4) 4.1 (1-0)

Americans 2.7 (1.4) 4.1 (0.9)

Poll By Students 2.3a (1.4) 4.1 (0.9)

Poll By Americans 2.4 (1.4) 4.2 (0.9)

Trend By Students 2.5 (1.4) 4.1 (1.0)

Trend By Americans 3.0ab(1.5) 4.Oc (0.9)

Control 2.4b (1.4) 3.6c (1.1)



Note. Column entries sharing a superscript significantly
different at the .05 level.


NOTES


Because a particular type of computer-readable answer sheet was used, the scales had to be a 1-to-5 measure instead of the usual six-point scale provided for uncertainty orientation. Both scales were presented in terms of statements where respondents could either answer a 1 for "disagree" across to a 5 for "agree." Responses of 2 or 4 were for "somewhat disagree" or "somewhat agree," while a 3 was for "neither disagree or agree." The typical z-score method of previous uncertainty orientation research was not used since









65


both scales making up the variable have the same range (1-to-5) and the reporting of the means is perhaps more
meaningful than that of z-scores.
2 A third factor manipulated the direction of the poll
result. Half of the subjects were to be told the poll
supported senior comprehensive exams and reforms in
agricultural pesticide use, while half were to be told
the poll shows most people against the issues. A
problem with question wording on the "against"
manipulation, making it unclear what was being asked,
was discovered after a first session with students.
These 35 cases were dropped from analysis and a second
session with another set of students did not include
versions with this manipulation, leaving a total number
of subjects at 128.















CHAPTER 7
STUDY 2 METHODOLOGY


The News Articles

News Article Preparation

Two short news articles about the capital gains tax

and jobs were prepared with a desktop publishing system to resemble genuine newspaper articles (see Appendix A for versions of each article). Both articles contained twoline headlines and bogus "jumps" informing the reader to continue reading on another page. In addition, a breakout quote or piece of information was presented in larger format to make the bogus poll results more accessible.

The articles were written in a similar fashion as to structure, the placement and length of quotations, the use of sources, and placement of manipulated information. The Manipulation of Relevance

Issues, no matter how careful the presentation, may differ in ways other than those intended or manipulated. Despite the relevance differences inherent in the issues as reported by Study 1 subjects, the first stimulus has to do with personal relevance.

It is apparent that the capital gains tax and the

possibility of a tight job market are very different, not only in the relevance people may place in them, but also












in the kind of issues they represent. The capital gains tax is a largely political matter. The jobs issue can be viewed politically, but it is likely subjects would consider it on other, more personal grounds. While this may seem to fit the "relevance" definition, it is possible subjects might fit the jobs issue into a different kind of category.

Persuasion research in the cognitive response

paradigm typically manipulates relevance on the same issue (see Petty & Cacioppo, 1986a, 1986b, for examples; O'Keefe, 1990, for a discussion). By manipulating relevance through telling subjects that the issue is or is not important to them at the present, researchers are able to avoid the pitfalls that come with differences between issues other than relevance itself. The question is one of theoretical and methodological control versus a more "real-world" test. This dissertation attempts both by providing subjects two issues apparently differing in personal relevance and then manipulating relevance within each issue. The manipulation occurred in the second paragraph of each version of the article. In the "low relevance" condition of the news article on the capital gains tax, for example, readers were told that "experts predict the final decision will not affect college students." In the jobs and recession article with a "high












relevance" manipulation, readers were told that "analysts predict most college students will be greatly affected-" The Manipulation of Consensus

The second manipulation had to do with perceptions of consensus through use of poll results. Subjects could receive one of three kinds of stories: a poll result in the direction of the article's chief arguments, a poll in the opposite direction, and a control group with no poll results.

For example, in the capital gains tax article,

subjects in the supporting poll condition--where poll results were in the direction of the majority of arguments presented--the fourth paragraph informed readers that a "recent nationwide public opinion poll showed 81 percent of Americans support the reduction of the capital gains tax, with 15 percent against and 4 percent undecided." In addition, the "blurb" or "pull out" added that "polls show most Americans are for a reduction in the capital gains tax." Subjects could also receive a version with the polls results contradicting the article's argument. The same sentence on poll results was provided, but instead the majority argued either against a capital gains tax or against the likelihood that jobs would be difficult to find for college students. A control group received a slightly shorter article (minus the paragraph on the poll












results) and with a "blurb" pulled from the article itself.

No hypotheses were presented concerning possible differences between how people process a supporting or contradictory poll or differences in the persuasive abilities of either type of poll. On the one hand, we might expect an contradictory poll to promote central processing, perhaps making the poll less influential (similar to the argument of Mackie, 1987). On the other hand, when the issue is of little relevance, we might expect the contradictory poll to persuade people as a consensus heuristic. This dissertation does not directly probe the different processes, but does use this research as a guide in explaining results. In general, it is expected that the contradictory poll will act in a similar fashion as the supporting poll. Summary of Manipulations

Volunteers from an introductory advertising class

participated in Study 2 over a two-week period for extra credit. Subjects could participate on March 19 or 20, 1991, for what will be called from this point T1. During this time subjects were measured on attitudes about the two issues selected for further study, uncertainty orientation, demographics, and other variables.












The design was a 3 (uncertainty versus middle versus certainty oriented) X 2 (low and high relevance) X 3 (poll result for or against or control) factorial for both issues.

Subjects therefore could be assigned to any of six

conditions for each article (three poll manipulations and two relevance manipulations). This meant a 6 X 6 or 36 different questionnaire versions. In addition, the order of article presentation was randomized to ensure no order effects would be present, resulting in 72 different questionnaire versions. Based on their categorization on uncertainty orientation, subjects were assigned to any of these 72 versions. Each subject's computer-readable bubble sheet from T1 was clipped to the assigned T2 version and subjects picked up these instruments at the T2 stage. To simplify presentation of the results, the condition where the poll agrees with most of the news articles will be called the "supporting poll" condition; the condition in which the poll results are the opposite of the arguments will be called "contradictory."

Before returning one week later for T2, uncertainty orientation scores were determined and subjects randomly assigned to experimental conditions. The T2 instrument opened with additional media use questions then presented the first news article followed by a thought-listing task












and attitude questions. A set of filler questions followed, and then the second article was presented along with its questions. The order of the two articles was randomized to ensure no order effects occurred.

The chief independent variable measured at T1 was uncertainty orientation, handled in the same way as in Study 1 except for differences in scaling. Rather than a 1-to-5 scale, a different computer-readable answer sheet version was used, this version ranging from 0 to 9. This scale was adopted because it allows more movement on the attitude measures, a greater range for uncertainty orientation responses, and the possibility of answers beyond the 1-to-5 range constraints such as age or days a week subjects read a newspaper.

Measurement of Variables

This study views attitudes as the evaluation of an object and as representing a residue of experience; different from moods, which are more temporary, attitudes are relatively enduring though susceptible to change (O'Keefe, 1990).

Attitude was measured by the same question at T1 and T2 on both issues. The results of the T1 attitude measure were used as a covariate and the results at T2 were used as the dependent variable on each issue in an analysis of












covariance. This allowed the statistical control of attitudes before the experimental manipulation.

Subjects were asked, on a 0-to-9 scale, to strongly disagree or strongly agree to the following statements: "The recession means jobs for college graduates will be hard to come by for the next several years" and "The federal government should cut the capital gains tax." The questions at T1 were listed with a number of other attitude measures to ease suspicion.

A second approach to this dependent variable, and one perhaps more closely resembling traditional bandwagon research, is to look for differences between T1 and T2. To investigate this possibility, a change score was also computed by subtracting the T1 attitude from the one at T2. Therefore, a negative score represents a decrease in support for the issue statement and a positive score an increase in support.

Uncertainty Orientation

As noted briefly in the previous chapter, factor

analysis of the full 35-item index provided by Sorrentino and his colleagues was problematic. In both studies, a principal components and principal factor analysis using varimax and oblique rotations arrived at a 10-factor solution, some factors consisting of only a single item. This analysis, however, could be interpreted as a three-












or perhaps two-factor solution (Appendix B contains all questions with means and standard deviations, Appendix C the 10-factor solution). The first factor appears to tap the general construct of need for uncertainty, although one authoritarianism question also loaded on this factor. The second factor was made up of only three questions asking about respect for authority, parents, and friends. The third factor are all authoritarianism questions generally aimed as beliefs and values.

Forcing a two-factor solution from all these

questions resulted in some authoritarianism items loading on the need for uncertainty scale, though no uncertainty items loaded on the authoritarianism factor.

After scrutiny of the factor analysis results and using the larger Study 2 subject pool, a subset of 12 items was selected and a second set of factor analyses conducted (factor analysis of the Study 1 pool resulted in roughly the same solution). The six need for uncertainty questions asked whether subjects enjoyed spending time discovering new things or finding out why things happen. The six selected authoritarianism measures focused on the need for discipline, obedience, and respect for authority. A principal factors analysis with a varimax rotation showed the items explaining 36.5 percent of the variance with a correlation between the factors of -.24. Both












scales demonstrated moderate internal consistency, with Cronbach's Alpha of .81 for the need for uncertainty scale and at .73 for the authoritarianism scale. The mean score for need for uncertainty was 6.3 (SD = 1.4) and for authoritarianism was 4.5 (U = 1.6). As in Study 1, the authoritarianism score for each subject is subtracted from the need for uncertainty orientation score and an overall index constructed. A textile split was conducted with the middle third included in analysis (see Table 1 for means and factor loadings of the 12 questions used in this study).

The shortened instrument resulted in little movement of subjects from being categorized as either uncertaintyoriented or certainty-oriented under the original format (X 2(4) = 238.5, p<.0001). Only six persons found themselves classified differently, three in each direction. That is, three originally classified as certainty-oriented were later classified as uncertaintyoriented, and vice versa.













Table 7-1

Factor Analvsis Results of Study 2 Uncertainty Orientation Questions



Question Mean SD F1 F2


I enjoy spending time discovering new things.

I like to find out why things happen.

I like to fool around with new ideas, even if they turn out later to be a total waste of time.

I enjoy thinking about ideas that challenge my views of the world.

I usually try to learn about something I don't understand, even if I might end up in trouble because of it.

I often put myself in situations in which I could learn something new.

What the youth needs most is strict discipline, rugged determination, and the will to work and fight for family and country.


6.5 1.8 6.9 1.8 6.1 2.0 6.9 1.9 5.6 2.2 6.5 1.7 4.6 2.3


.71 .67 .66 .61 .60 .59


-.16


-.05




-.13



-.24





-.23



-.17


-.17 .62
















Question Mean SD F1 F2


Note - Loadings above are from the rotated structure
matrix of a principal factors analysis with oblique rotation. The correlation between factors was -.24.


Table 7-1--continued


There is hardly anything lower than a person who does not feel a great love, gratitude, and respect for his or her parents.

Young people sometimes get rebellious ideas, but as they grow up they ought to get over them and settle down.

Obedience and respect for authority are the most important virtues children should learn.

No sane, normal, decent person could ever think of hurting a close friend or relative.

Books and movies ought not to deal so much with the unpleasant and seamy side of life; they ought to concentrate on themes that are entertaining or uplifting.


Eigen Value

Percent Variance Explained


4.9 2.5 4.9 2.3 4.4 2.5 4.8 2.7 3.6 2.5


-.09





-.17




-.12




-.15







-.12


.60 .57 .55 .53







.45


1.6 13.2


2.8 23.3















CHAPTER 8
STUDY 2 RESULTS


The experiment in Study 2 was conducted over a twoweek period.1 Two-hundred and eighty-nine subjects participated in the first phase of the experiment (M). Two hundred and sixty-three returned for T2, a loss of approximately 9 percent. The average age of all subjects was 20.6 (SD = 2.7). Fifty-nine percent of the subjects were female and 41 percent male. Nearly half reported their academic standing as being juniors (137, or 47.4 percent).

Assumptions

Analysis of covariance were the primary statistical tests used in analysis.2 As in Study 1, examination for assumptions found no violations of normality or homogeneity of variance.

Study 1 was used to select issues of differing relevance and with no relationship to uncertainty orientation. Association of simple issues statements with uncertainty orientation may differ in news articles as compared to simple issue statements. T-tests of pretest attitudes (before the manipulations) revealed no significant difference in attitudes toward either issue and uncertainty orientation.












ManiDulation Checks

In a manipulation check of the poll condition,

subjects were asked to agree or disagree with a statement that most Americans agree with the stance argued in the article. If the manipulation was successful, those receiving the condition where poll results favored the article should provide a higher score (measured on a 0-to9 scale) than those in the control group, and those receiving a poll in the opposite direction (a contradictory poll) should demonstrate less agreement as compared to the control group. T-tests show significant differences in all cell means in the expected directions (see Table 8-1).

Manipulations of relevance were mixed, however. A question after each article asked subjects to agree or disagree (on a 0-to-9 scale) as to whether "the article was very relevant to me personally." As Table 8-2 shows, the manipulation appears to have worked on the capital gains tax article, with those receiving the low relevance paragraph (N = 3.1) reporting less personal relevance than those receiving the high relevance paragraph (M = 3.7, :t(261) = -2.5, R<-01). However, there was no significant difference in relevance reported for the highly relevant jobs issue manipulation, perhaps due to a ceiling effect of the already high relevance associated with that













Table 8-1

Means (SD) of Manipulation Check Asking HQie Likely Most Americans Agree by Poll Type and Issue



Poll Manipulation Received


Issue Contradictory Poll Control Supporting Poll Jobs For 4.8 (3.2) 6.4 (2.0) 7.6 (1.6)
Graduates

Capital 3.7 (2.9) 4.6 (1.7) 7.1 (1.8)
Gains Tax



Note. All row entries are significantly different from
each other at the .01 level by t-tests. Table 8-2

Means (SD) of Manipulation check of Effects of Experimental Manipulation of Personal Relevance on Relevance Scores by Issue



Relevance Manipulation Received Issue Low High


Jobs For
Graduates 7.1 (2.0) 7.4 (2.1)

Capital
Gains Tax 3.1a (2.1) 3.7a (2.3)



Note. Entries sharing a superscript are significantly
different at the .01 level by t-tests.












issue.3 Relevance scores associated with the jobs issue are significantly higher than those associated with the capital gains issue regardless of relevance manipulation.

Tests of Hypotheses

This study has three chief hypotheses. The first, based on cognitive response models of persuasion (e.g., the elaboration likelihood model of Petty & Cacioppo, 1986a, 1986b), predicted that poll results would be more influential on attitudes about less relevant issues as compared to more relevant issues. The second hypothesis had to do with explaining why the bandwagon or underdog effects are seen in different studies. Following the reasoning of Sorrentino (Sorrentino & Hancock, 1987), it was hypothesized that uncertainty-oriented persons are more likely to wonder why a minority thinks the way it does, to "go with the underdog," to react against a poll result as compared to certainty-oriented persons, while certainty-oriented persons would be more likely to "jump on the bandwagon," to be influenced by a poll. Third, following the work of Sorrentino, et al. (1988), it was hypothesized that in high-relevant situations, certaintyoriented persons would be even more likely to be influenced by a poll as compared to uncertainty-oriented













persons. The results section is organized around these three hypotheses.

In the following results, T2 attitudes are used as the dependent variable and the TI attitudes treated as covariates. For the capital gains tax issue, the T1 attitude had a raw regression coefficient of 0.18. For the jobs and recession issue, the raw regression coefficient was 0.30. Where change scores are reported in tables (usually in parentheses), these are the result of a subtraction of T2 attitude from T1 attitude. Statistical tests are for the attitude measures. Change scores are provided as a guide for change between T1 and T2.

Finally, before moving to the results, recall that if the poll is influential, those in the contradictory poll condition should demonstrate lower attitude scores as compared to the control group. On the other hand, an influential supporting poll should result in attitude scores greater than those seen among those in the control group.

Hypothesis 1 - Poll Effects and Relevance

The first hypothesis predicts that polls are more influential in situations of low relevance (Petty & Cacioppi, 1986a, 1986b). However, poll influence was found only on the more relevant jobs issue (see Table 83). Those receiving the supporting poll showed greater













approval of the article's arguments (NI = 6.8) than those in the control group (K = 6.2, t(175) = -2.0, R<.05) or those receiving the contradictory poll (K = 6.0, t(170)=

-2.5, p<.01). No significant differences were found on the capital gains tax issue (F-tables for both issues are found in Table 8-4 and 7-5).

capital gains tax issue

Looking within issues, a two-way interaction between poll manipulation and personal relevance was found on the capital gains tax issue (F(2,242) = 3.2, p<.05). First, looking at the contradictory poll, no influence was seen Table 8-3

Means of T2 Attitudes (Change Scores) by Issue and Poll Manipulation Received



Poll Manipulation Received


Issue Contradictory Poll Control Supporting Poll



Capital 4.8 (+0.3) 5.2 (+0.3) 5.1 (+0.5)
Tax N =88 N =87 N =85


Jobs For 6.0 (-0.6)a 6.2 (.2)b 6.8 (+O.6)ab
Graduates N =85 N =90 N =86



Note. Row entries sharing superscript are significantly
different at the .05 level by t-tests. Cell means
for the jobs issue are significantly higher than
those on the capital gains tax issue.













Table 8-4

F-Table of Analysis of Covariance of Capital Gains Tax Issue by Poll and Relevance Manipulations and Uncertainty Orientation



Sum Of Mean Sig.
Source of Variation Squares DF Square F Of F


Covariates

Ti Attitude Main Effects

1 Poll Type 2 Relevance
3 Uncertainty Orientation 2-Way Interact ions


1 By 2 1 By 3 2 By 3


31.4 31.4


9.4 4.6 0.0
4.2 70.3 22.5 27.0 17. 0


1 31.4 1 31.4 5 1.9


2.3 0.0
2.1


6 8.8


11.3 6.8 0.9


9.0 9.0


0.5 0.7 0.0 0.6


2.5 3.2 1.9 0.3


.00 .00


.74 .51 .91 .55 .01

. 04 . 10 .90


3-Way Interactions


3.8 3.8


1 By 2 By 3


Explained Residual


4 0.9 4 0.9


114.9 18 6.4 840.0 242 3.5 954.9 260 3.7


Total


0.3 .90


0.3 1.8


.90


. 02


Covariate T1 Attitude


Raw Regression Coefficient


.17










84


Table 8-5

F-Table of Analysis of Covariance of Jobs I-ssue by Poll and Relevance Manipulations and Uncertainty Orientation



Sum Of Mean Sig.
Source of Variation Squares DF Square F Of F


Covariates

T1 Attitude


Main Effects

1 Poll Type 2 Relevance
3 Uncertainty Orientation


2-Way Interactions


1 By 2 1 By 3 2 By 3


114 .4 114 .4


178. 5

54 .4
65.0 63 .0


35.2 6.9
27.4 1.9


1 114.4 29.3 .00


1 114.4 29.3


. 00


5 35.7 9.1 .00


27.2 65.0 31.5


7.0 16.6 8.1


.00 . 00 .00


8 4.4 1. 1 .35


3.5 6.9 0.9


0.9 1.8
0.2


.41 .14
.78


3-Way Interactions


1 By 2 By 3


35.2 35.2


4 8.8 2.3 .06


4 8.8 2.3


363.3 18 20.2 5.2


Explained Residual


949.0 243 1312.3 261


Total


Covariate T1 Attitude


Raw Regression Coefficient


.30


. 06 .00


3.9 5.0












among those receiving the high-relevance manipulation (LI

4.8 for both the control group and contradictory poll conditions; see Table 8-6). Among persons receiving the low-relevance version, those in the control group were less approving of the issue (M = 5.6) than those receiving the contradictory poll (M = 4.8, t(87) = -1.7, p<.05).4

An effect opposite the one hypothesized is found when considering the supporting poll. For persons in the lowrelevance condition, those receiving the poll were less approving (M = 4.8) than those in the control group (M

5.6, t(85) = 2.2, p<.02). Among those told the issue was


Table 8-6

Means of T2 Attitudes (Change Scores) on Capital Gains Tax by Poll and Relevance Manipulations



Poll Manipulation Received


Contradictory Poll Control Supporting Poll



Low Relevance 4.8 (+0.1)a 5.6 (+0.8)ab 4.8 (+0.1)b
N = 46 N = 42 N = 44

High Relevance 4.8 (+0.5) 4.8 (-0.1)c 5.4 (+0.9)c
N = 43 N = 45 N = 41


Note. Interaction F(2,254) = 3.2, p<.04. Entries sharing
a superscript are significantly different at the
.05 level.













of high personal relevance, those receiving the poll were more supportative (M =5.4) than those in the control group (M = 4.8, :t(84) =-1.8, p<.04). Jobs and recession issue

There was no significant two-interaction between relevance and poll found on the jobs issue (E(2.243)=

0.9, p<.41; Table 8-7). No differences were found between the control group and those receiving the contradictory poll regardless of relevance condition. Also, no difference between the control group and the supporting poll group was found among those in the high-relevance condition (M = 6.8 and M = 7.0, respectively). However, a


Table 8-7

Means of T2 Attitudes (Chanae Scores) on Jobs by Poll and Relevance Manipulations



Poll Manipulation Received


Contradictory Poll Control Supporting Poll



Low Relevance 5.3 (-1.2) 5.6 (-l.O)a 6.7 (0.0)a
N =45 N =46 N =43

High Relevance 6.7 (+0.1) 6.8 (+0.8) 7.0 (+1.0)
N =40 N =45 N =43


Note. No significant interaction. Entries sharing a
superscript are significantly different at the .05
level by t-tests.












t-test of the mean scores among those told the issue was of little relevance showed those in the control group less supportive (M = 5.6) than those receiving the supporting poll (M = 6.7, t(87) = -2.5, p<.01). Summary of hypothesis 1 results

The poll appeared to influence subjects on the more relevant jobs issue, but not on the less relevant capital gains tax issue.

Looking at within-issue relevance, the contradictory poll influenced only those in the low relevance condition, while the supporting poll influenced persons in both relevance conditions, but not in a way that would be hypothesized -- those in the low relevance condition demonstrating more approval when shown the supporting poll, while those told the issue was very relevant showed less approval.

Hypothesis 2 - Polls and Uncertainty orientation

It was hypothesized that uncertainty-oriented persons would be less persuaded by a poll than certainty-oriented persons.

Capital gains tax issue

A two-way interaction near the traditional level of significance was found for poll manipulation and uncertainty-orientation (f(4,242) = 1.9, p<.10; see Table 8-8).












Looking first at the contradictory poll, certaintyoriented persons in the control group were more supportive (M = 5.5) than those receiving the poll. (N = = 4.4, M(54) = -2.1, p<.03). The mean scores of uncertainty-oriented persons were in the expected direction, but were not statistically signficant, with those showing greater approval when in the contradictory poll condition (K =

5.3) than in control group (H=4.7, t(55) = -1.1, p<.14).


Table 8-8

Means of T2 Attitudes (Chanae Scores) toward Capital Gains Tax by Poll and Uncertainty Orientation



Poll Manipulation Received


Contradictory Poll Control Supporting Poll



Certainty 4.4 (+0.2)a 5.5 (+0.5)a 5.1 (+0.6)
Oriented N =29 N =29 N =31

Moderate 4.7 (-0.2) 5.3 (+0.2) 5.6 (+1.2)
Uncert-O. N =26 N =36 N =25

Uncertainty 5.3 (+0.7) 4.7 (+0.4) 4.5 (-0.2)
Oriented N =29 N =27 N =29



Note. Interaction F(4,242) = 1.9, R<.10. Entries sharing
a superscript are significantly different at the
.05 level by t-tests.













No significant differences were found between the control group and those receiving the supporting poll. Jobs and recession issue

No significant two-way interaction between poll and uncertainty orientation was found on the jobs issue (.E(4,243) = 1.8, R<.14; see Table 8-9). T-tests of the expected differences found no significant poll influence among certainty-oriented persons. Uncertainty-oriented persons were influenced by both poll versions as compared to the control group (M=5.7), with those receiving the


Table 8-9

Means of T2 Attitudes (change scores) toward Jobs and Recession by Poll and Uncertaintv Orientation



Poll Manipulation Received


Contradictory Poll Control Supporting Poll



Certainty 6.3 (-0.7) 5.8 (-0.6) 6.6 (+0.1)
Oriented N =27 N =40 N =18

Moderate 6.4 (+0.2) 7.3 (+1.0) 7.0 (+1.4)
Uncert-O. N =34 N =24 N =33

Uncertainty 4.5 (-.22)a 5.7 (.6)ab 6.8 (+0.1)b
Oriented N =24 N =27 N =35



Note. Interaction F(4,243) = 1.8, ns. Entries sharing a
superscript are significantly different at the .05
level by t-tests.












contradictory poll showing a decrease in support (M = 4.5, t(49) = -1.7, p<.05) and those receiving the supportive poll increasing in support (M = 6.8, t(66) = -1.9, p<.02). Summary of hypothesis 2 results

The capital gains tax issue offered some support for the second hypothesis. Among those receiving the contradictory poll, certainty-oriented persons were more likely to move in the direction of the poll while uncertainty-oriented persons were more likely to move in the opposite direction, as compared to the control group. No support for the hypothesis was found in the jobs issue. Hypothesis 3 - Polls, Relevance, and Uncertainty
Orientation

The third hypothesis predicted that certaintyoriented persons would be more sensitive to situations of high relevance and would be more influenced by polls in those situations.

capital gains tax issue

No significant three-way interaction was found

between uncertainty orientation, poll type, and personal relevance on the capital gains tax (F(4,242) = 0.3, p<.90; see Table 8-10). T-tests to explore the hypothesized relationship found a pattern in comparing certaintyoriented and uncertainty-oriented persons in the control group and contradictory poll condition, though neither of the findings reaches the traditional level of













significance. In these, certainty-oriented persons told the issue was of personal relevance showed more approval in the contradictory poll condition (1 = 4.1) than those in the control group (M = 4.9, _(55) = 1.1, p<.15). Uncertainty-oriented persons in the high-relevance condition, however, moved in the opposite Table 8-10

Means of T2 Attitudes (Change Scores) Toward Capital Gains Tax by Poll. Relevance. and Uncertainty Orientation



Contradictory Poll Control Supporting Poll


Cert.-Oriented 4.7 (+0.6)a 6.0 (+0.5)ab 5.1 (+0.5)b Low Relevance N = 14 N = 13 N = 15

Mod. Unc-O 4.4 (-0.4) 5.4 (+0.7) 5.0 (+0.2)
Low Relevance. N = 19 N = 20 N = 11

Uncert.-Oriented 5.5 (+0.2) 5.7 (+1.6)c 4.3 (-0.3)c Low Relevance N = 13 N = 9 N = 18


Cert.-Oriented 4.1 (-0.1) 4.9 (+0.5) 5.1 (+0.6) High Relevance N = 15 N = 13 N = 14

Mod. Unc-O 5.4 (+0.3) 5.3 (-0.4) 6.1 (+1.8)
High Relevance N = 10 N = 16 N = 16

Uncert.-Oriented 5.1 (+1.1) 4.1 (-0.3) 4.8 (0.0) High Relevance N = 18 N = 16 N = 11



Note. No significant interaction. Selected entries
sharing superscripts significantly different at the
.05 level by t-tests.













direction, with those receiving the contradictory poll demonstrating less approval (M = 5.1) than those in the control group (N = 4.1, _(32) = 1.5, p<.07). Jobs and recession issue

A significant three-way interaction was found between uncertainty orientation, poll type, and relevance on the


Table 8-11

Means of T2 Attitudes toward Jobs Issue by Poll. Relevance, and Uncertainty Orientation




Contradictory Poll Control Supporting Poll


Cert.-Oriented 4.9 (-1.9) 5.4 (-1.4)a 7.1 (+0.2)a
Low Relevance N = 14 N = 19 N = 14

Mod. Unc-O 6.0 (+0.1) 6.7 (+0.4) 6.7 (+0.6)
Low Relevance N = 21 N = 7 N = 14

Uncert.-Oriented 4.4 (-2.9) 5.4 (-1.3) 6.3 (-0.6)
Low Relevance N = 10 N = 20 N = 16


Cert.-Oriented 7.9 (+0.6)b 6.3 (+0.3)b 6.0 ( 0.0) High Relevance N = 13 N = 15 N = 11

Mod. Unc-O 6.8 (+0.3) 7.5 (+1.2) 7.4 (+2.2)
High Relevance N = 19 N = 17 N = 13

Uncert.-Oriented 4.6 (-1.4) 6.3 (+0.6) 7.3 (+0.7) High Relevance N = 8 N = 13 N = 19



Note. Interaction F(4,243) = 2.3), p<.07. Entries
sharing a superscript are significantly different
at the .05 level by t-tests.












jobs issue (f(4,243) = 2.3, p<.06; see Table 8-11). Instead of the hypothesized effect of high relevance on certainty-oriented persons, those receiving the contradictory poll were more approving of the jobs issue (M = 7.9) than those in the control group (M = 6.3, t(26) = 2.1, p<.03). No difference was found among certaintyoriented persons in the high relevance condition either in the control group of receiving the supporting poll. Poll influence was also seen in situations of high relevance for uncertainty-oriented persons (rather than certaintyoriented, as hypothesized). Among uncertainty-oriented persons told the issue was of high relevance, those in the control group showed more approval = 6.3) than those receiving the contradictory poll (M 4.6, t(19) = -1.4, p<.09) and less approval than those receiving the supporting poll version (M = 7.3, t(30) = -1.2, p<.13), though the last finding is not statistically significant. Summary of hypothesis 3 results

Little support for hypothesis 3 was found on the capital gains tax issue, but results from the jobs and recession issue were opposite the direction predicted. on the less relevant capital gains tax issue, certaintyoriented persons told the issue was of high relevance and who received a poll contradictory to the main thrust of the news article were more influenced in the direction of




Full Text

PAGE 1

EXPLAINING THE "BANDWAGON" AND "UNDERDOG" EFFECTS: A STUDY OF PERSONAL RELEVANCE AND UNCERTAINTY ORIENTATION AS FACTORS IN PUBLIC OPINION POLL INFLUENCE By BARRY A. HOLLANDER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1991

PAGE 2

ACKNOWLEDGEMENTS There are many people who deserve recognition for their assistance in my completing my graduate studies, which culminate in this dissertation. First, I would like to thank my committee: Drs. Mary Ann Ferguson, Michael Martinez, Leonard Tipton, Michael Weigold, and John Wright. It was a Leonard Tipton research seminar that first sparked my interest in public opinion and perceptions of opinion. No doubt this fascination will continue for some time. I would also like to thank Kim Walsh-Childers for acting as a substitue on my final dissertation. Thanks also to Michael Weigold and Julie Dodd for use of their students in the dissertation experiments. Mary Ann Ferguson deserves special note. She hired me as her research assistant at the Communication Research center when I was a lowly and hungry master's student barely able to find a computer's on-off switch. My three year apprenticeship with her at the Center is the reason I continued my graduate studies, and her research seminars provided me a theoretical direction. Her enthusiasm for research and teaching was contagious. If ever a graduate student bears the stamp of his major professor, I suspect ii

PAGE 3

I do. With luck and hard work, I may live up to this ideal. A number of fellow graduate students deserve acknowledgement, more than I have the time or space to mention. Geetu Melwani from the period of my work on a master's degree and fellow doctoral students Erika Engstrom, Art Emig, and Alan Fried come immediately to mind. There are many others. Finally, and most important, I thank my wife Edith for her patience, love, and support. She worked as I continued my graduate studies, making it possible for me to write this dissertation. I look forward to spending time with her again. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii ABSTRACT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vi CHAPTERS 1 INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 A Different Perspective...................... 2 Research Hypotheses. . . . . . . . . . . . . . . . . . . . . . . . . . 4 Summary of Dissertation Contents............. 5 2 PUBLIC OPINION AND POLLS..................... 6 Polls and Mass Communication................. 7 Bandwagon Research........................... 8 Conformity Research.......................... 16 The Poll Bandwagon--A Summary................ 19 3 PERSUASION RESEARCH. . . . . . . . . . . . . . . . . . . . . . . . . . 21 Cognitive Response Analysis.................. 21 Persuasion and Individual Differences........ 25 4 UNCERTAINTY ORIENTATION...................... 28 Theoretical Discussion....................... 28 Findings in Uncertainty Orientation.......... 33 Authoritarianism. . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 4 Other Personality Theories................... 35 on the study of Personality, Politics, and Communication.......................... 36 5 SYNTHESIS AND HYPOTHESES..................... 39 Bandwagon Research and Persuasion............ 39 Consensus as a Heuristic cue................. 43 Summary and Hypotheses. . . . . . . . . . . . . . . . . . . . . . . 4 4 Overview of the Studies...................... 46 iv

PAGE 5

6 STUDY 1 METHODOLOGY AND RESULTS.............. 47 Methdo 1 ogy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 7 Results ...................................... 52 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 o Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 7 STUDY 2 METHODOLOGY. . . . . . . . . . . . . . . . . . . . . . . . . . 6 6 The News Articles............................ 66 Measurement of Variables..................... 71 Uncertainty Orientation. . . . . . . . . . . . . . . . . . . . . . 7 2 8 STUDY 2 RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 7 Assumptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Manipulation Checks.......................... 78 Tests of Hypotheses.......................... 80 Other Significant Findings................... 94 Notes ........................................ 95 9 DISCUSSION..................... . . . . . . . . . . . . . . 96 summary of Study Findings.................... 96 Limitations of the Studies................... 100 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Implications for Bandwagon Research.......... 107 Future Research. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 O APPENDICES A NEWS ARTICLE EXAMPLES. . . . . . . . . . . . . . . . . . . . . . . . 113 B MEANS AND STANDARD DEVIATIONS OF ALL UNCERTAINTY ORIENTATION QUESTIONS.......... 126 C FACTOR LOADINGS OF ALL UNCERTAINTY ORIENTATION QUESTIONS. . . . . . . . . . . . . . . . . . . . . . 13 0 REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2 BIOGRAPHICAL SKETCH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 7 V

PAGE 6

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EXPLAINING THE "BANDWAGON" AND "UNDERDOG" EFFECTS: A STUDY OF PERSONAL RELEVANCE AND UNCERTAINTY ORIENTATION AS FACTORS IN PUBLIC OPINION POLL INFLUENCE By Barry A. Hollander December 1991 Chairman: Leonard Tipton Major Department: Mass Communication Concerns about the influence of public opinion polls have existed almost as long as polls themselves. "Bandwagon" research results have been inconsistent, however, even with an underdog effect being found. The conceptual framework of persuasion research from social psychology was used to review the bandwagon literature and to offer hypotheses. The individual difference variable of uncertainty orientation was presented as one explanation as to why some people are influenced in the direction of a poll while others are influenced in the opposite direction. Three hypotheses were presented. First, it was hypothesized that subjects would be more susceptible to poll influence in a low relevance situation as opposed to one of high relevance. Second, it was hypothesized that vi

PAGE 7

certainty-oriented persons would be more likely to move in a "bandwagon" direction, or with the poll influence, while uncertainty-oriented persons would be more likely to move in an "underdog" direction, or opposite the poll influence. Finally, it was hypothesized that when relevance was high, certainty-oriented persons would be susceptible to poll influence. A 2 (relevance) X 2 (uncertainty orientation) X 3 (type of poll) factorial design experiment was conducted using news articles with bogus poll results. Mixed support was found for all three hypotheses, primarily in situations where relevance was manipulated rather than between issues thought to differ in relevance. It was suggested that future research into public opinion poll influence take into consideration the relevance of the campaign or issue under consideration as well as the personality traits of those hearing such poll results. vii

PAGE 8

CHAPTER 1 INTRODUCTION In 1935, U.S. Rep. Walter Pierce of Oregon introduced legislation that would have prohibited use of the mail for gathering political straw ballots. If enacted, pollsters convicted under the law faced a maximum penalty of one year in prison or a $1,000 fine or both. Such polls are "powerful, subsidized propaganda," Pierce warned in a later Public Opinion Quarterly article, and should be recognized as "a potent, if not the most powerful, agency now used to influence public opinion" (Pierce, 1940, p. 241 and p. 243, respectively). While Pierce's bill failed, the potential influence of public opinion polls has been a focus of debate for more than 50 years. Pollsters assert that their efforts reflect rather than affect public opinion, arguing no "bandwagon" effect exists (Gallup & Rae, 1940; Robinson, 1937). Other early commentators such as Pierce (1940) and Lewis (1940), however, doubted that polls were so innocent and incapable of persuasion. Despite arguments, evidence from survey and experimental studies remains inconclusive. Reviews by Marsh (1983) and Merkle (1991) found that while some studies do favor a bandwagon interpretation, many do not. 1

PAGE 9

2 Laboratory studies of artificial elections coupled with bogus poll results produce an "underdog" effect, of people shifting to the minority rather than majority view. In general, there appears little support for the bandwagon thesis, particularly in controlled settings, which seem to favor an underdog effect. Possible reasons for such inconsistent conclusions range from the kinds of issues or candidates studied to how the "majority" and "minority" are identified. Few attempts have been made, however, to explain and systematically test why people may choose the bandwagon or underdog route or the process that takes place. Almost exclusively devoted to elections at the expense of issues, the bandwagon literature typically assumes an undifferentiated public, though some scholars have begun to investigate the influence of predispositions on likelihood to move with or against a poll (e.g., Lavrakas, Holley, & Miller, 1991). A Different Perspective Some public opinion scholars have hinted at the possibility that polls are influential in situations of less importance to the public. McBride (1991) suggests such influence is more likely in political races below the presidential level, while Hickman (1991) points to presidential primaries rather than elections as better opportunities to find the bandwagon effect.

PAGE 10

3 Another perspective is offered, this one from social psychology. Theorists such as Petty and Cacioppo (1986a, 1986b) outline two distinct modes of processing which bear directly on how and why persuasion takes place. The key to these persuasion models is whether people have the motivation and ability to process a message. Where motivation is low, people are thought to be more likely to use shortcuts in making decisions. An issue or political campaign of little personal relevance, then, is unlikely to provide the motivation necessary to carefully scrutinize all the questions involved. In such a case, a credible source or the presence of some other source characteristic can be of greater influence. In situations of high relevance, such source characteristics are thought to have little influence. With this conceptual framework in mind, the research in poll influence is reviewed and an attempt is made to organize the results along the lines seen in these persuasive models. As noted above, one criticism of bandwagon research is its focus on an undifferentiated public. One explanation for both the bandwagon and underdog effects may also be in people themselves. That is, some people may be more persuaded by polls, while others may be more open to the arguments presented by a minority. The individual difference variable uncertainty orientation is

PAGE 11

4 offered as one possible explanation (Sorrentino, Short, & Raynor, 1984). Briefly, Sorrentino and his colleagues posit that some people -certainty-oriented persons seek to maintain clarity about themselves and their environment. That is, they will avoid self-diagnostic situations if they feel that information is threatening, and, in fact, when facing a situation of personal importance they will rely on simple decision-making rules rather than consider all the information possible. Their polar opposites, uncertainty-oriented persons, are open to such information and in fact seek it out. Research Hypotheses Three research hypotheses can be generated from this discussion, all of which are discussed in more detail in later chapters. First, it is expected that issues of little personal relevance will be more influenced by poll results than issues of high personal relevance. When considering uncertainty orientation, it is hypothesized that certainty-oriented persons will be more likely to be persuaded by a poll than uncertainty-oriented persons. Finally, building on other work in uncertainty orientation, it is expected this relationship between uncertainty orientation and poll results will be more pronounced in situations of high relevance, which certainty-oriented persons would find threatening. That

PAGE 12

5 is, certainty-oriented persons would be even more persuaded by a poll in personally relevant situations rather than in low-relevant situations. summary of Dissertation contents Chapter 2 reviews the research on the influence of polls, most of which has been conducted by public opinion and mass communication scholars, while Chapter 3 discusses the persuasion literature from social psychology. Chapter 4 reviews the work on uncertainty orientation, and Chapter 5 provides a synthesis and the resulting hypotheses. Study 1 (Chapter 6) is designed mainly to select issues for use in study 2. The second study tests the research hypotheses discussed above. Chapter 7 presents the methodology of Study 2 and Chapter 8 the results. The final chapter provides an overall discussion of the findings.

PAGE 13

CHAPTER 2 PUBLIC OPINION AND POLLS Separating the poll from public opinion is difficult, and public opinion sometimes defined as that which a public opinion poll measures. Public opinion is often viewed as the aggregation and distribution of individual opinions, simply "any collection of individual opinions" (Childs, 1939, p. 331). Others regard public opinion as a communication process (e.g., Cooley, 1909: Price 1989). Price and Roberts (1988) distinguish between the pollster's view of public in public opinion as a noun and their conceptualization of public as an adjective describing the opinion process. An important feature of this conceptualization is F. D. Allport's (1937) view of public opinion formation occurring through an individual's observation of his social environment, a theoretical orientation extended by others, notably Noelle-Neumann's "spiral of silence" theory (Noelle-Neumann, 1974, 1977). Noelle-Neumann (1974, p. 51) suggests individuals possess a "quasi-statistical organ" that allows them to assess the opinion environment, largely through information provided by the mass media. However, it appears from research such as the false consensus effect (Ross, Greene, & House, 1977: Marks & Miller, 1987) and 6

PAGE 14

7 pluralistic ignorance (Fields & Schuman, 1976; O'Gorman & Garry, 1976) that if such an "organ" exists, it is indeed a faulty one. False consensus research has demonstrated that people tend to believe others think as they do, even when those individuals are obviously in a minority. Pluralistic ignorance has to do with misunderstanding of true opinion distribution. Polls and Mass communication Key to Noelle-Neumann's spiral of silence theory is how people learn about the distribution of opinion. A likely source is public opinion polls, a popular staple of the news media since early in the century. More than one third of newspapers surveyed reported the use of polls to gather information (Rippey, 1980), and about 15 percent of newspaper election stories focused on polls (Stovall & Solomon, 1984). While there has been some improvement in how the media conduct or report on polls (Salwen, 1985), others charge news organizations have fail to provide necessary explanatory information on polls and, when they do so, the reliance on such polls places a "horse-race metaphor" emphasis on elections (Broh, 1980). Content analyses of poll stories found newspapers often fail to provide the information deemed necessary by polling experts (Miller & Hurd, 1982; Sanders, Rollberg, & Buffalo, 1989), despite findings that inclusion of this

PAGE 15

8 information may improve polls' credibility (Paletz, et al., 1980; Salwen, 1987). People are aware of public opinion and polls. Gollin (1980) reported that 84 percent of a 1976 national sample had heard of polls. Kohut (1986) and Roper (1986) both report results suggesting the public regards polls and pollsters as generally credible. While polls are a popular means of people learning about the distribution of opinion on public issues, many question whether, rather than reflecting public opinion, polls can actually influence public opinion (see, for example, Bogart, 1972; Gollin, 1980; Roll & Cantril, 1980) . Bandwagon Research Early scholars and pundits termed the influence of actual or perceived majority opinion on election results as the "bandwagon" effect. A bandwagon effect occurs when the majority receives additional support from the publication of a poll; an underdog effect is just the opposite, in which people respond to an underdog's status and go against the majority or with the minority. This notion of a bandwagon effect has existed almost as long as public opinion polling, although much of the literature is simply speculative.

PAGE 16

9 Though a number of elegant theoretical models have been published arguing for the effect (Gartner, 1976; Henshel & Johnston, 1987; H. A. simon, 1954; Straffin, 1977; Zech, 1975), the literature is sprinkled generously with reports of null findings (Beniger, 1976; Dizney & Roskens, 1962; Fleitas, 1971; Lang & Lang, 1984). Adding to the confusion is the discovery of an "underdog" effect (Ceci & Kain, 1982; Laponce, 1966). As discussed in the previous chapter, early in this century congressional efforts to control straw ballots failed. Even the news media, a major user of polling today, argued for stricter regulation. A 1936 New York Times editorial, for example, cautioned that polls often "develop a bandwagon rush" and "in cases where public opinion is less definitely rooted, and the allegiance of large groups of voters is less definitely won, a 'bandwagon' tendency may play a decisive part in determining the result of an election" ("Straw Ballots," p. 22). Similar arguments were made by others (F. D. Allport, 1940; Perry, 1968; Stoler, 1986). Early pollsters brought forth more systematic though problematic evidence. Directly following Pierce's 1940 condemnation of polls in Public Opinion Quarterly. for example, Gallup and Rae (1940) presented sequential survey evidence finding no bandwagon effect.

PAGE 17

10 Research in public opinion poll influence typically takes one of two methodological directions. The first is the use of survey, panel, or trend survey data to uncover opinion movement; the second is the use of experiments or quasi-experimental methods in laboratory or field settings. First we review the survey evidence, then that from experiments. Survey Evidence one popular survey method is to ask respondents what influenced their decisions and whether, more specifically, polls had some influence (Hastings & Hastings, 1990; Roshwalb & Resnicoff, 1971)--a self-report method not without criticism as to whether individuals are aware or can accurately describe what influenced their opinions (Nisbett and Ross, 1977). Berelson, Lazarsfeld and McPhee (1954) found only about 10 percent mentioned polls as a source of information about changes in who they expected to win the presidential campaign. In asking directly about poll influence, Hastings and Hastings (1990) found between 3 and 11 percent of those polled willing to admit such influence. Roshwald and Resnicoff's (1971) survey of 1,000 respondents found no voluntary mentions of polls as being influential in a vote decision and, when asked, 66 percent of the respondents said polls were not at all important in their decision-making.

PAGE 18

11 Other survey research has attempted to track effects of polls through panel or trend studies. One of the earliest discussions of the effect on voting by Lazarsfeld, Berelson, & Gaudet (1948) found many voters attempting to sense the direction of public opinion and the outcome in order to vote "with the winner." Follow-up work (Berelson, Lazarsfeld, and McPhee, 1954) revealed that a bandwagon effect (perception guiding preference) and a projection explanation (preference guiding perception) carried about equal weight in presidential election voting. In another use of survey methodology, an extensive study by Beniger (1976) analyzed data from the date of gathering and publication of Gallup candidate preference polls and state primary elections from 1936 to 1972. Essentially, Beniger's sequential method was looking for whether a poll influenced the primary, or vice versa. In 83 of 183 preference polls followed by a subsequent poll, the ratings of the leading candidate increased from the first to the second poll, while in 81 cases it fell. Like many other sequential poll studies (Kavanagh, 1981; Roll & Cantril, 1980; McBride, 1991), Beniger found no consistent bandwagon effect and even hints of an equally inconsistent underdog effect, though others have successfully demonstrated more indirect poll effects such

PAGE 19

12 as those on campaign contributions, volunteer work, or endorsements by key political actors (Henshel & Johnston, 1987) . Non-experimental cross-sectional evidence cannot resolve the crucial question of causality. Trend studies provide a slightly clearer picture, but the results do not support the intuitive appeal and theoretical justification of a bandwagon effect of poll influence. Other methodologies have not improved matters. Experimental and Quasi-Experimental Evidence Given the weaknesses noted above, Campbell (1958) asserted "the question of a bandwagon effect resulting from publication of poll results can only be settled by experimentation" (p. 252). Even controlled settings have not been kind to the bandwagon thesis, however, with findings as mixed as those from survey research. To investigate the question, Campbell suggested a national manipulation of poll results, a difficult and costly task in the 1950s and one far too expensive--if even possible -today. A similar but less ambitious study reported by Gallup (1972) compared cities with and without newspapers carrying the Gallup Poll and found no differences in candidate preferences. No further details were provided.

PAGE 20

13 An early experiment by Cook and Welch (1940) used college students and as a stimulus the infamous 1936 Literary Digest poll showing Alf Landon leading Roosevelt in the presidential election. Students informed about the Digest poll were more likely to favor Landon than those in the control group not told about the poll. A second experiment conducted over a longer time period produced results in the direction of a bandwagon prediction, though the results were not statistically significant. These results led Cook and Welch to suggest any bandwagon effect may be a small one, reminiscent of Robinson's (1937) cautioning that the bandwagon thesis contained "more an element of fancy than fact" (p. 49). Katz (1972) argued that it was not sheer numbers from polls that resulted in influence, "but the character of the majority in relation to the individual's psychological group membership" (p. 24). That is, a poll of similar others was thought to have more influence than a poll of a more general public. Atkin (1969) found some support by using "student preferences" on one issue, though no effect was found on two other issues. Two British studies favored an underdog effect on likelihood to turn out to vote, indirectly affecting results. The first, cited in Teer and Spence (1973), varied information provided to three treatment groups of

PAGE 21

14 400 voters before an election. There was no effect on preference, though supporters of the party they were told was behind at the polls said they were more likely to turn out to vote. In the second study, Gaskell (1974) varied information given about who was ahead in the polls, but rather than a no-information control group, he told respondents the polls were close. Fewer supporters said they would vote if a clear-cut victory was predicted, but this finding was stronger for supporters of the party predicted to win. However, deBock's (1972) investigation of the one-sided 1972 presidential election with a field experiment suggested the trailing candidate may suffer loss of support intensity and turnout motivation after supporters have been exposed to election poll results. Navazio (1977) divided a sample of 500 persons into two groups, one a control and the other an experimental group that was told of bogus national poll results. The mail survey found no poll effect. In a similar vein, Roper, in a study reported by Cantril (1980), conducted a split-ballot trial with presidential approval ratings by introducing half of the questions with "As you know, all the polls have been showing support for Carter going down. We'd like to get your opinion about him." Few technical details were provided by the author, though it was

PAGE 22

15 reported that the experimental manipulation resulted in only a 1 percent movement from Carter. Ceci and Kain (1982) asked students their preferences for Carter and Reagan in the 1980 presidential election during morning classes. Before being asked, some sections of the class were told a survey of "college-educated persons" showed support for Carter (first section of the class), Reagan (second section), or no result (third section). Later that night, confederates posing as national pollsters telephoned students and requested a presidential preference. Prior to this, students were again provided bogus poll data. Despite the methodological problems of this study's use of intact groups, the results point to movement toward the underdog. A quasi-experimental panel study by Cloutier, Nadeau, and Guay (1989) found 75 percent of their 1,005 subjects had the same attitude over a one-month period. However, 14 percent changed in the direction of the poll manipulation, while 11 percent went with the underdog or minority view. Fewer studies combine experimental control with a representative sample. A two-wave national panel survey conducted during the 1988 Bush/Dukakis presidential campaign by Lavrakas, Holley and Miller (1991) included an experimental manipulation of informing respondents about

PAGE 23

16 actual public opinion poll results. Some interesting results were obtained. Persons who did not . graduate from high school and who were told that Bush was leading in the polls showed a significant increase in uncertainty about which candidate they would vote for compared to those with similar education who were not told Bush was leading. No effect was seen on those with higher education. The authors conclude better-educated persons were more resistant to the external source of influence. summarizing the experimental research, Marsh (1983) concluded no experimental evidence exists for a bandwagon hypothesis, though there is some support that a perception of growing strength may aid in the bandwagon process. conformity Research In a related field of study, research on conformity has long documented the influence of the majority on the minority in small-group settings (Allen, 1965; Kiesler & Kiesler, 1969). The majority's superior size, power, and status provides a firm base for establishing social reality and a position in which to reward those who agree (Festinger, 1950). Recent findings have led many to question whether the minority must be willing and helpless recipients or whether a minority cam instead bring about modification and change (Moscovici & Personnaz, 1980), speculation not unlike the underdog thesis. It has even

PAGE 24

17 been suggested that majority and minority influence processes differ (Moscovici, 1980; Nemeth, 1987), though many others maintain a single process is in effect (Doms, 1983; Latane & Wolf, 1981; Wolf, 1987). Latane's (1981) social impact theory is one example of single process argument, proposing that influence by either a majority or minority is a multiplicative function of the strength, immediacy, and number of its members. Typically these studies rely on a single confederate who disagrees with the majority on some perception task, the results showing a persistent minority can influence both the public and private responses of a majority. In one study, Moscovici & Personnaz (1980) presented subjects a series of blue slides consistently labeled as green by the confederate. On each trial, subjects were required to indicate the color of the slide and the color of the afterimage perceived on a screen following removal of the slide. The confederate was presented as either a member of the majority or minority. The minority viewpoint was found to influence subjects' judgments of the afterimage. The authors conclude that the persistent minority causes the majority to start a validation process by considering the deviant response and its reasoning. "In other words, as a result of trying to see or understand what the minority saw or understood, the majority begins to see and

PAGE 25

18 understand as the minority would" (Moscovici & Personnaz, 1980, p. 272). It appears the minority, at least in these experimental settings, can indeed influence the majority, confirming Asch's original suggestion (Asch, 1951) that some persons distort their perceptions of an unambiguous stimuli as a function of influence of others. In Moscovici's (Moscovici, 1980) theory, a consistently stated minority opinion can lead to careful scrutiny by the group, a detailed processing of the minority position similar to that of systematic or central processing. This influence is held to be latent, indirect, or covert, an internalized change of opinion not necessarily revealed publicly. Influence by the majority is held to bring about manifest, direct, or overt influence, what researchers call public compliance. One study in this paradigm stands out as closest to the bandwagon question (Kaplowitz, Fink, D'Alessio, & Armstrong, 1983). Subjects were presented six issue statements with different bogus poll results about those issues. Subjects were either told they would either have to discuss their attitudes publicly or were given a private response option. No conformity effect was found for those who reported high commitment to their attitudes about the issues, even those responding privately.

PAGE 26

19 However, the authors did find the bogus-poll technique could produce a conformity effect on those with little commitment to their attitudes. A previous study (Tyson & Kaplowitz, 1977) concluded that if people respond privately, results of public opinion polls will not directly influence responses to issues on which they have strong commitment to their views. The Poll Bandwaaon--A summary Throughout the accumulation of these assorted findings on poll influence, a number of factors have been suggested for opinion movements, including perception of majority trends (Marsh, 1984), the strength or unexpectedness of these perceptions (Aronson, Turner, & Carlsmith, 1963), and the commitment to an opinion (Kaplowitz, Fink, D'Alessio, & Armstrong, 1983). In some research a bandwagon effect is found, in others an underdog effect. some studies find no results, while others find both effects occurring. Why the difficulties? Merkle (1991) suggests a number of methods of uncovering the underlying process: more powerful experimental designs, care in looking for both bandwagon and underdog effects, assessing or controlling for previous perceptions, analysis of the relevance of reference groups, the kinds of topics used as a stimulus, investigation of personality characteristics,

PAGE 27

20 and consideration of how subjects interpret the poll results. Another way to approach the question is from a single theoretical perspective that attempts to consolidate findings in persuasion research in general and to apply that to poll influence studies in particular. The next chapter focuses on the work by cognitive response analysis theorists and two specific and complementary models of persuasion.

PAGE 28

CHAPTER 3 PERSUASION RESEARCH Soon after Gordon Allport declared attitudes to be "the most distinctive and indispensable concept in contemporary American social psychology" (Allport, 1935, p. 798), the construct blossomed as a research focus. Motivated by World War II, early scholars in social psychology investigated how persuasion and propaganda influenced attitudes (Hovland, Lumsdaine, & Sheffield, 1949; Hovland, Janis, & Kelley, 1953). The framework described by Lasswell (1948a) and elaborated by Hovland et al. (1953) most closely resembles the work of cognitive response theorists and is the theoretical direction taken in this dissertation. It has enjoyed remarkable growth in the past 10 years. The early approach focused on "who says what to whom with what effect," the investigation of the communicator, the message, the audience, and the effects of attitude change. Cognitive Response Analysis Cognitive response analysis delves further into the audience, arguing that attitude change is primarily a function of an individual's personal elaboration of external stimuli (Fiske & Taylor, 1984; Petty, Ostrom, & Brock, 1981). A stimulus causes a cognitive effect, which 21

PAGE 29

22 results in a response. Variables in this research include audience involvement, message repetition, message comprehension, and communicator credibility. Critical to the cognitive response perspective is audience involvement or personal relevance, the variable most mentioned as a determinant of processing styles. Research in the social judgment perspective shows high involvement to be associated with less attitude change, while the cognitive response to persuasion model shows just the opposite, a contrast noted by Petty and Cacioppo (1979a) and explained by differences in conceptions of involvement (O'Keefe, 1990). Without involvement, social cognitive theorists argue, an individual will not use thoughtful, central, systematic information-processing strategies and instead will rely on peripheral, heuristic, or shortcut strategies that depend heavily on such irrelevant cues as communicator characteristics (Chaiken, 1980; Petty & Cacioppo, 1979b). Simply put, less involved persons are more likely to operate "on automatic," while highly involved persons will operate in a more "controlled" mode (Fiske & Taylor, 1984, p. 354). This research into the two distinct styles of making social judgments is best exemplified by the elaboration likelihood model (ELM) of persuasion and the heuristic systematic model (HSM) of persuasion (see Petty &

PAGE 30

23 Cacioppo, 1986a or 1986b, for more complete descriptions of the ELM. For a similar view on HSM, se~ Chaiken, Liberman, & Eagly, 1989}. Elaboration in the ELM is generally described as "the extent to which a person thinks about the issue-relevant argument contained in a message" (Petty & Cacioppo, 1986b, p. 128). In their conceptualization, when elaboration likelihood is high, more cognitive resources should be used by the receiver. Use of issue-relevant elaboration typically results in the new arguments, or a personal translation of them, being integrated into the underlying belief structure or schema for the attitude object. Current social psychological research provides support for two general modes of processing discussed above, termed by these theorists as central and peripheral routes to persuasion. In this view, the likelihood of elaboration depends on a person's motivation and ability to evaluate the communication. Motivation is influenced by such variables as personal relevance or personal responsibility, while ability to process can be influenced either by personal characteristics such as prior knowledge or by such variables not related to the person as message comprehension or distraction in the communication situation. With both motivation and ability present,

PAGE 31

24 elaboration likelihood is high. With only one or neither present, elaboration likelihood is low. This model presents a continuum from no thought about the issue-relevant information to complete elaboration on the issues. On this theoretical continuum's elaboration side is the central route to persuasion, on the opposite end the peripheral route. The former has persuasion achieved through close scrutiny to message arguments and consideration of other issue-relevant material, the latter is conducted through use of some simple decision rule or heuristic principle to evaluate an advocated position. Persuasion through the central route is posited to be long lasting, while persuasion through the peripheral route is relatively temporary. Examples of heuristics are number of arguments presented, use of expert sources, communicator credibility, likability, and attractiveness. The two routes are not mutually exclusive; rather they represent the extremes of the elaboration likelihood continuum. For example, a mixture of central and peripheral processes are theorized to occur at moderate levels of elaboration (Petty & Cacioppo, 1986a). A series of experiments by Petty, Cacioppo, and their colleagues show that high relevance increases the likelihood of issue-relevant thinking and invoking the

PAGE 32

25 central route to persuasion (Petty & Cacioppo, 1979a, 1979b, 1984; Petty, Cacioppo, & Goldman, 1981). The HSM (see Chaiken, 1980, 1986, 1987; Chaiken & Eagly, 1983) similarly views heuristic processing as involving more limited cognitive effort and capacity. It goes further, however, by describing heuristics as learned knowledge structures that may be used by social perceivers. Such rules can range from "Experts' statements can be trusted" to "If most people think so, it must be right." In general, both models agree that two modes of processing exist, and that heuristics such as source expertise or consensus can lead to less systematic processing of arguments. Persuasion and Individual Differences The Petty, Cacioppo, and Goldman (1981) study of the effects of argument strength and communicator expertise on persuasive effectiveness provides a clear example of the cognitive response perspective. Subjects were presented a message with an issue that was either relatively involving personally (designed to prompt high elaboration) or not involving (presumably to cause less elaboration). Personal relevance or involvement was manipulated by making receivers believe the recommendations in the message for comprehensive examinations for senior undergraduates were being considered for adoption at their

PAGE 33

26 university the following year (high involvement) or 10 years later (low involvement). Also manipulated were argument quality (strong versus weak) and the expertise of the communicator (high versus low). For those in the high-relevance condition, strong arguments produced more attitude change than weak arguments, but source expertise did not affect persuasion. In the low-relevance condition, subjects were more persuaded by an expert than by a non-expert source, but the quality of the arguments in the message did not affect subjects• attitudes. It appears that when issue relevance is high, people are more persuaded by message content (the central route); but when relevance is low, they are more persuaded by attributes or cues (peripheral route). But is this always the case? Little work has been conducted on individual differences in the effects of personal relevance, though Cacioppo, Petty, and Morris (1983) demonstrated that high need for cognition, defined as the tendency for people to engage in and enjoy thinking, resulted in more central processing than for those low in the need for cognition. Need for cognition was not found to interact with the manipulations and only enhanced the likelihood for central processing. Others questioned whether personal relevance might actually reduce the likelihood for elaboration for some people

PAGE 34

27 (Sorrentino, Bobocel, Gitta, Olson, & Hewitt, 1988). Uncertainty orientation, identified by Sorrentino, Short, and Raynor (1984), is one such possible dimension.

PAGE 35

CHAPTER 4 UNCERTAINTY ORIENTATION Uncertainty orientation is an individual difference occupying a niche called the "warm look," an interaction and interdependence of motivation and cognition to explain behavior (Sorrentino & Higgins, 1986; Sorrentino & Short, 1986). The construct remains in an early state of development and researchers have concentrated less on a definition than on describing those who are certainty and uncertainty-oriented, conceding "we are still learning about them ourselves" (Sorrentino & Roney, 1990, p. 242). This chapter describes the theoretical and operational rationale behind the construct and sketches the research to date. Theoretical Discussion Definitions proposed for uncertainty orientation (and as we will see later, the measurement itself) have changed as well, though Sorrentino maintains the variable is a cognitive rather than motivational one (Sorrentino & Short, 1986). For example, need for cognition (Cacioppo & Petty, 1982) has been described as a variable measuring motivation to think, while uncertainty orientation is a measure of when to think (Sorrentino, et al., 1988). 28

PAGE 36

29 Uncertainty orientation has been described as the degree to which situations of certainty versus uncertainty are cognitively relevant (Sorrentino, Short, & Raynor, 1984); as involving the degree of uncertainty surrounding the outcome of one's activity (Sorrentino & Short, 1986); as how one orients him or herself toward the uncertainty concerning the self or the environment (Sorrentino & Hancock, 1987); as people's relative interest in either maximizing information gain or maintaining clarity (Roney & Sorrentino, 1987); and as an individual's approach to uncertainty or ambiguity (Sorrentino, et al., 1988). In general, uncertainty orientation is the concern or interest in finding out new things about the self and the environment and the tolerance of ambiguity about the self or environment. What should interest mass communication scholars is Sorrentino's contention that uncertainty orientation is primarily concerned with information value and the maintenance and preservation of current knowledge. Interest in the variable arose from Atkinson and Raynor's (1974) theory of achievement motivation and replies from cognitive informational theorists (Trope, 1975; Weiner, 1972). In short, it was argued that situations that attain clarity about the self result in the greatest arousal of achievement-related motives for uncertainty

PAGE 37

30 oriented people, and situations that maintain clarity lead to the greatest arousal of these motives for certainty oriented people. One series of experiments showed the theory of achievement motivation only held for uncertainty-oriented persons (Sorrentino, Short, & Raynor, 1984), that is, certainty-oriented people demonstrated no interest in using a diagnostic achievement situation to learn about themselves. Generally, it is posited that the uncertainty oriented person seeks to attain clarity about the self and the environment, while the certainty-orientated individual seeks to maintain present clarity about the self or the environment. To attain clarity, one is willing or likely to deal with information that may even threaten the self perception or challenge beliefs. Maintaining clarity is just the opposite, the tendency to hold on to beliefs and to avoid situations that are self-diagnostic. Attaining or maintaining clarity is done through use or failure to use or process information that is available. Historical Perspective The prototype for uncertainty orientation is traced to Rokeach's The Open and Closed Mind (1960), which presents a continuum of "gestalt types" to "psychoanalytic types." Rokeach characterized the former as having a "need for a cognitive framework to know and understand"

PAGE 38

31 and the latter as having the "need to ward off threatening aspects of reality" (Rokeach, 1960, p. 67) Thus the open-minded person possesses a belief system oriented toward the new, the closed-minded a belief system oriented toward the familiar, a dichotomy similar to the Freudian notion of basic trust and mistrust of the world. The more closed the system, according to this argument, the more difficult it is to distinguish between information received about the world and information about the source--a perspective not unlike that of persuasion theorists discussed in the previous chapter that emphasizes elaboration of issue matter versus influence by source characteristics. Rokeach's work included the famous "Joe Doodlebug" experiments to explore differences between those Rokeach called "closed" and "open" persons. The experiments showed that persons with "closed" systems had more difficulty with integrating new beliefs and often showed early closure. These results were not associated with intelligence or ideology. Sorrentino also notes the influence of Kagan (1972), who described uncertainty resolution as one of four primary determinants of behavior. The motive to resolve uncertainty, he writes, is the "wish to know" (p. 54). While not discussing individual differences, Kagan

PAGE 39

32 suggested that people in a state of uncertainty will either assimilate, remove, or escape from an event that is discrepant from an established schema. Certainty-oriented persons would attempt to remove or escape such discrepant information, while uncertainty-oriented persons would attempt to assimilate such information. Cognitive Perspective The term schema shows up in Sorrentine's work as defense for uncertainty orientation as a largely cognitive variable with motivational roots. Building on the work of construct assessability (Bargh 1984; Higgins & King, 1981), King and Sorrentino (1988) found that uncertainty oriented and certainty-oriented persons differ in the chronic accessibility of positive and negative constructs reflecting certainty (e.g., "cautious" and "stubborn") and uncertainty (e.g., "adventurous" and reckless"). In perhaps the most complete explanation of the construct, the authors write that their model assumes that uncertainty-oriented persons are those who have been rewarded for autonomous exploratory behavior. Such persons develop general schemas for situations that allow resolution of uncertainty about the self and the environment. When placed in these situations, they are affectively charged to resolve this uncertainty. Certainty-oriented persons, on the other hand, have not been rewarded for autonomous, exploratory behavior and in fact might have been punished for such behavior. Consequently, these persons gravitate toward and develop schemas for safe and familiar situations or schemas that do

PAGE 40

33 not require dealing with uncertain situations. Hence, they are affectively charged by situations involving certainty and the self and the environment (Sorrentino, et al., 1988, p. 358) An "affective charge" closely resembles the language of motivation scholars who describe motivation as the drive that energizes, orients, and guides behavior. Uncertainty-oriented individuals will be motivated in situations that involve uncertainty about the self or the environment, while certainty-oriented persons should be more motivated in situations that involve certainty about the self or the environment, setting into motion respective cognitive responses. Findings in Uncertainty orientation Research The initial test of this hypothesis by Sorrentino et al. (1984) found that for uncertainty-oriented persons, relevant sources of motivation were most aroused in performance situations that contained uncertainty about the self or the environment. The reverse was found for certainty-oriented persons. The authors argued their findings integrate existing cognitive and motivational interpretations of achievement behavior. This first study was aimed at explaining problems arising from motivational research. In work more closely related to social influence, Sorrentino & Hewitt (1984) tested whether there would be differences in how

PAGE 41

34 uncertainty-oriented and certainty-oriented people would approach information that was personally r~levant. As predicted, uncertainty-oriented persons chose activities aimed at resolving uncertainty about a new and potentially important ability, whereas certainty-oriented persons actually chose an alternative activity that would tell them nothing new about this ability. This finding held even when the information was likely to be positive. Again studying personal relevance, Sorrentino and Roney (1986) demonstrated that motivational arousal would occur in situations relevant to a person's uncertainty orientation. Achievement-related motives of uncertainty oriented persons were most aroused in a diagnostic task about a new and possibly important ability (mental flexibility), but those of certainty-oriented persons were most aroused in a non diagnostic task. Authoritarianism An important aspect of uncertainty orientation is use of authoritarianism, a construct elaborated by Adorno, Frenkel-Brunswik, Levinson, & Sanford (1950) and ably criticized by Shils (1954) as dealing only with authoritarianism of the "right." The preoccupation is understandable, however, given the political climate of

PAGE 42

35 the 1940s. Still, most of the early questions measuring the construct were directed at the political right. One of the earliest and most influential studies linking psychological and political concepts,~ Authoritarian Personality. identified a group of individuals who possessed anti-Semitic attitudes and then expanded this view to generally ethnocentric aversions for all people. Theoretical development of the F-scale (for fascism) was drawn largely from Freudian concepts of superego, ego, and id, and was traced to childhood development (Sanford, 1971). Authoritarians were found to display considerable cognitive rigidity and intolerance for ambiguity (Block & Block, 1951; Frenkel-Brunswick, 1949; Jones, 1954; Rokeach, 1948; Steiner & Johnson, 1963), to believe others thought the same way they do (Granberg, 1972; Scodel & Mussen, 1953; Simons, 1966), to favor conservative political attitudes (Adorno, et al., 1950), and to vote for conservative or authoritarian candidates (Higgins, 1965; Milton, 1952; Poley, 1974; Wrightsman, 1965). Other Personality Theories There is a long history of attempts to provide an exhaustive taxonomy of personality (Cattell, 1943; Digman, 1990; Fiske, 1949; Norman, 1963). Some consensus for a "Big Five" model appears to be emerging (Digman, 1990).

PAGE 43

36 Of these five factors, one dimension appears closely related to uncertainty orientation. This dimension goes by a number of labels, but perhaps can best be described as an openness to experience (Costa & Mccrae, 1985; Coan, 197 4) Another similar personality dimension is repression sensitization (Byrne, 1964), where persons are thought to deal with threatening stimuli in one of two ways--a person who approaches threats is a sensitizer, a person who avoids threats is a repressor. Repressors focus on supportative information and avoid information that is not supportative (Olson & Zanna, 1979). on the study of Personality. Politics. and Communication "To talk about politics without reference to human beings," Lippmann wrote, "is just the deepest error in our political thinking" (1913, p. 2). Early work by Lasswell emphasized the need to bring psychology into political analysis (Lasswell, 1930, 1948b, 1951). Yet personality variables, which indicate an organized, stable internal predisposition a person brings to a situation, have not enjoyed a great deal of success in political communication research. Even supporters of the theoretical perspective concede personality variables are unlikely to account for much of the variance when looking at such behaviors as joining a major political party or in voting, though

PAGE 44

37 evidence exists that such variables are "selectively felt at a number of points in the political process" (Knutson, 1973, p. 44). Much of this "vast and uneven work" falls under the externalization hypothesis, which treats political opinions as externalized versions of the mind's inner conflicts (Kinder & Sears, 1985, p. 676). Sorrentino argues for extending the work in individual differences from the achievement area to that of social influence and persuasion. Reviewing the work of persuasion theorists, Sorrentino and others questioned whether uncertainty orientation may interact with personal relevance in a persuasive context. As discussed earlier, research has typically shown that perceivers process information more thoughtfully when the subject has high rather than low personal relevance (Petty & Cacioppo, 1981: Fiske & Taylor, 1984; Nisbett & Ross, 1980). The authors replicated the work of persuasion theorists, primarily that of Petty & Cacioppo (1984) and Petty et al. (1981), to study the effects of personal relevance on attitude change as a function of one's uncertainty orientation. Sorrentino et al. (1988) predicted that, unlike uncertainty-oriented persons, high personal relevance would make certainty-oriented persons less careful or systematic in their processing of message arguments and

PAGE 45

38 more dependent on heuristics, or persuasion cues, than would low personal relevance. The investigators found that personal relevance does not appear to increase systematic processing for all persons--that certainty oriented persons are more motivated to consider a matter further when personal relevance decreases. These individuals do not prefer to reason matters out when facing a situation with uncertainty about themselves or their environment, the authors conclude. "When forced into these situations, they rely on heuristics rather than use their own judgment" (p. 368). This reversal of a principle "thought to be universal," they add, highlights the value of an individual-differences approach in social influence (p. 369). The question is not one of merely varying the strength of a persuasive effect, then, but varying the direction of the effect. "In other words, variables assumed to enhance systematic forms of information processing may work for some, but may actually dampen such tendencies in others" (Sorrentino & Hancock, 1987, p. 265)

PAGE 46

CHAPTER 5 SYNTHESIS AND HYPOTHESES Bandwagon Research and Persuasion Reviewing the bandwagon research from the perspective of persuasion theory offers some clues as to why the bandwagon effect occurs in some cases but not in others. A key variable in this paradigm is personal relevance. A look at many bandwagon and poll studies finds some influence in what we might suspect are low relevance issues, and little or no persuasive (or bandwagon) effects on what we might suspect are high relevance issues. For example, Atkin's (1969) use of a "student preference" poll found effects only on an income tax surcharge issue but not on the issues of marijuana or reopening the Warren Commission investigation (controversial matters at the time). While no measures of relevance are provided in the study, it seems likely that the tax surcharge issue was less important to his college student subject pool that the use of marijuana or the Warren Commission issue. Atkin used a three-week period between measures, thus it may be that the poll had more influence on the less relevant issue but no effect on the higher relevance issues. 39

PAGE 47

40 A look at Table 5-1 provides a look at some selected poll persuasion studies. Though drawn largely from Merkle (1991), the table organizes the list chronologically with bandwagon or bandwagon/underdog findings listed first, underdog-only effects listed second, and then null studies listed third. Most null findings occurred in the 1970s, a time when belief in persuasive effects of mass communication were in their decline. studies conducted in the 1980s have uncovered greater support for the bandwagon/underdog thesis, perhaps due to a reliance on experiments and quasi-experiments. Looking closely at the table of the thirteen studies finding a bandwagon or bandwagon/underdog effect, only five were elections or candidate evaluations. Both of the studies finding only an underdog effect were elections, while seven of eight studies finding no poll influence were of an election or candidate evaluation. It appears elections offer less an opportunity to demonstrate a poll's influence. This hardly seems surprising if, as is suggested by the ELM and other persuasion models, people are more highly involved in presidential or other elections or hold more closely to candidate evaluations than they would an abstract issue, therefore less likely to be persuaded by a heuristic such as a poll result.

PAGE 48

41 Table 5-1 Chronological Listing of Selected Bandwagon/Underdog and consensus Persuasion Studies Authors Subjects Topic Conclusion Cook & Welch College Pres. Some bandwagon ( 1940) students election support Allard (1941) College Issues Support for students bandwagon effect Lazarsfeld, General Pres. Support for et al., (1944) public election bandwagon and and Berelson, underdog effect et al., (1954) Atkin (1969) College Issues Some support for students bandwagon effect Roshwalb & General State Indirect support Resnicoff (1971) public election for bandwagon de Bock (1972) College Pres. Support for students election indirect bandwagon Meyers, et al. Church Issues Support for (1977) members bandwagon effect Kaplowitz, College Issues Bandwagon support et al. (1983) students on low-commitment Marsh (1984) General Abortion Trend produced public issue bandwagon effect Mackie (1987) College Issues Support for students bandwagon/underdog Cloutier, College Issue Support for et al. (1989) students bandwagon/underdog Lavrakas, General Pres. Support for et al. ( 1990) public election bandwagon/underdog

PAGE 49

42 Table 5-1--continued Authors Laponce (1966) Ceci & Kain (1982) Dizney & Roskens (1962) Sales Research Services Ltd. (1964) Cited in Teer & Spence (1973) Fleitas (1971) Gaskell (1974) Beniger (1976) Tyson & Kaplowitz (1977) Navazio (1977) Roper (1980) cited in Cantril (1980) Subjects Topic College Hypoth. students election College Pres. students election College students General public College students General public Gallup polls College students General public General public Pres. election British election Hypoth. election British election Primary elections Issues Attitudes on Nixon Attitudes on Carter Conclusion Support for underdog effect Some support for underdog effect No support for bandwagon/underdog No support for bandwagon/underdog No support for bandwagon/underdog No support for bandwagon/underdog No support for bandwagon/underdog No support for bandwagon effect No support for bandwagon/underdog No support for bandwagon/underdog

PAGE 50

43 Looking at issues, Kaplowitz, et al. (1983) found no conformity effect on what subjects described as high conunittment issues, but some conformity on low committment issues, a finding that would be anticipated in the persuasion perspective. Atkin (1969) also found support for the bandwagon thesis, but only on one of three issues (one less likely to be relevant to respondents). consensus as a Heuristic cue A number of different heuristic principles have been suggested in persuasion studies, but three have received the most research attention: the credibility, liking, and consensus heuristics (O'Keefe, 1990). The consensus heuristic can be broadly defined as "if other people believe it, then it's probably true" (O'Keefe, 1990, p. 107). Two recent studies using the cognitive response conceptual framework to investigate consensus effects arrive at different, though not contradictory, conclusions. Axsom, Yates, & Chaiken (1987) studied audience response as a consensus cue in order to clarify a history of mixed results, finding these reactions influenced opinions only under low involvement conditions: in high involvement conditions only argument quality influenced

PAGE 51

44 opinions, a result similar to that of Kaplowitz, et al. (1983). In a series of experiments more closely aligned to this dissertation, Mackie (1987) investigated the kind of processing induced by consensus information. Mackie found subjects simultaneously exposed to a majority poll result with which they disagreed and a minority with which they agreed showed more issue-relevant processing of the majority message. She concluded that the majority provides validity to the arguments presented, directing attention to them and resulting in considerable processing. That is, disagreeing with a majority results in more attention being paid to the issue-relevant facts themselves, leading to systematic processing. Majority status, then, can in some cases provide the motivation for recipients to systematically process communications in certain circumstances. summary and Hypotheses Polls influence people--sometimes. The conceptual framework of the ELM and HSM provides one hope of unraveling these tangled findings. As noted by Marsh (1983), research on the bandwagon effect or poll influence in general has tended to focus on elections at the expense of issues. This dissertation will explore issues and poll influence using the persuasion paradigm, which typically

PAGE 52

45 uses dependent variables other than a "yes" versus "no" or one candidate versus the other, instead reiying on an attitude measure. If an issue is personally relevant, then a consensus heuristic is predicted to have less influence. If an issue is not relevant, presentation of consensus should have more chance to sway attitudes. Following this, it is hypothesized that an issue poll will be more persuasive in lowrather than high-relevance issues. Second is the matter of uncertainty orientation. In a series of speculations about the possible interactive effects of uncertainty orientation, Sorrentino and Hancock (1987) argue that uncertainty-oriented persons, exposed to consensus information, would try to discover why a minority insisted on its side of an issue in the face of a majority. Certainty-oriented persons would be more likely to ignore the inconsistency and "go along with the crowd" (Sorrentino & Hancock, 1987, p. 256). Here we come to the bandwagon versus underdog effect. If this theorizing holds, uncertainty-oriented persons should be less likely to be influenced by a poll or move in the "underdog" direction, while certainty-oriented persons should be more persuaded by poll. Finally, based on the work of Sorrentino, et al. (1988), it is hypothesized that in situations of high

PAGE 53

46 relevance, certainty-oriented persons threatened by such relevance would turn to a heuristic cue such as a poll and be more influenced by a poll than would uncertainty oriented persons. overview of the studies This dissertation is organized around two studies. Study 1 largely focuses on the testing of issues for use in Study 2, though it also includes analysis of two experimental manipulations helpful in deciding how best to proceed with the second phase.

PAGE 54

CHAPI'ER 6 STUDY 1 METHODOLOGY AND RESULTS Methodology Choosing the Issues Two steps were taken to decide what issues to use. First, students from four mass communication writing classes, of about 20 students each, were asked in fall 1990 and spring 1991 to list the three most personally important social or political issues. In the first issue-listing phase, the Persian Gulf crisis received the most mentions (26), followed by abortion (15), the environment (14), AIDS (11), the economy (11), education (9), censorship, drugs, elderly and child abuse, and energy (each with 6). Next, from this list, issue statements were presented to two sections of a large introductory mass communications class for attitude and personal relevance measures. Two questions accompanied each issue statement, the first asking for the subjects' attitude toward the issue, the second for the personal relevance of that issue. Both were measured on l-to-5 scales as provided by the machine read answer sheets (bubble sheets). An issue attitude could range from a 1 for "disagree" to a 5 for "agree." 47

PAGE 55

48 Relevance measures could range from l-to-5 as well, a 1 representing "not relevant" and a 5 "very relevant." There were two reasons for this phase of the research. First, two issues rated as low and high relevance by students themselves would be presented to the subjects in the later study. Second, those issues had to be relatively unrelated to uncertainty orientation. An association between the issues and uncertainty orientation might have occurred due to a number of factors, the most likely being ideological differences between certainty oriented and uncertainty-oriented persons. Uncertainty Orientation Measures of uncertainty orientation resembles those used to assess achievement-related motives (Atkinson, 1958; Atkinson & Feather, 1966). Frederick and Sorrentino (1977) developed a projective measure of n-uncertainty from which to infer to relevance of uncertain situations. Early attempts to unconfound authoritarianism and acquiescence response sets (Christie, Havel, & Seidenberg, 1958; Couch & Keniston, 1960) have led many to question the usefulness of the construct. Sorrentino and his colleagues adopted the measure of authoritarianism of Byrne and Lamberth (1971) to infer the relevance of certain situations. Previous research has shown that people high in authoritarianism are more intolerant of

PAGE 56

49 ambiguity (Kirscht & Dillehay, 1967) and have less experience with uncertain situations (Kelman & Barclay, 1963). Interrater reliabilities for then-uncertainty projective measure have been found to be above .90 by Frederick and Sorrentino (1977) and Sorrentino (1977) had similar results for the authoritarianism measure. Generally, sentence leads are used for the projective test, with the open-ended responses coded for certainty and uncertainty imagery (see Fredrick & Sorrentino, 1977, for detailed scoring instructions). The acquiescence-free measure of authoritarianism results was used with then-uncertainty measures as opposing elements of uncertainty orientation. This resultant measure is used, according to the authors, because persons scoring low on n-uncertainty may not necessarily be high in certainty orientation, and vice versa. The method is not unusual in the motivational literature, for example, where two uncorrelated measures are used to measure achievement motivation (Atkinson & Feather, 1966). Briefly, z-scores are computed on the two measures for a resultant measure and a tertile split of the scores are used. The middle third is typically excluded from further analysis because of inconsistent results with this middle group.

PAGE 57

50 Projective measures are time consuming and in some research settings, such as surveys, not feasible. A 35item questionnaire, still being validated with the projective measures, has been produced by Sorrentino and his colleagues and is used in the research reported here. This instrument contains two sets of questions. The first set is designed to measure need for uncertainty, using questions rather than the previous projective measure. The second set of questions is the same as the original instrument and is designed to measure authoritarianism. Sorrentino and his colleagues believe the new instrument essentially taps the same construct and initial testing is positive, though it is also thought that the pen-and-pencil measure does not have the explanatory power of the projective method (Sorrentino, R. M., personal communication, 1989, 1991). 1 Independent and Dependent Variables Following the issues on the Study 1 instrument were measures of media reliance, attitudes about poll credibility, importance of opinions of others, demographics, political ideology, and uncertainty orientation. Finally, two of the issue statements were presented again for attitude measures (comprehensive exams and agricultural pesticide reforms). Bogus poll results

PAGE 58

51 were built into the issue statements and these versions randomly assigned to subjects. One manipulation involved the poll sample. Some subjects received a version that said the poll was of college students, others received a version that stated the poll was of Americans. A second manipulation had subjects either being told a single poll result or a version that added that a trend was seen in the direction of the poll result. A third factor was dropped due to question wording problems, reducing the number of subjects available for analysis. 2 To summarize, subjects received both issues (senior comprehensive exams and agricultural reforms in pesticide use) and were randomly assigned to one of four conditions: (1) Poll of Students; (2) Poll of Americans; (3) Trend-Poll of Students; and (4) Trend-Poll of Americans. Uncertainty orientation was added to this, making it a 2 (uncertaintyor certainty-oriented) X 2 (poll or trend and poll) X 2 (students or Americans as sample) analysis design. A control group was also included, these subjects receiving only the issue statements a second time with no poll results. An example of the dependent variable question is one from the poll-student condition, which stated: "A public opinion poll shows college students favor university senior comprehensive exams to graduate, with 76 percent of

PAGE 59

52 college students in favor. What do you think?" Subjects could disagree or agree on the same 1-to-5 scale as above. The second question in this condition stated: "The same poll shows college students in favor of agricultural reforms in pesticide use, with 78 percent of college students in favor. What do you think?" The slight difference in percentages of the bogus poll results was used to decrease suspicion. These issues differed significantly in relevance as reported by the subjects, with comprehensive exams being slightly higher in relevance (M = 3.9) than pesticide reform (M = 3.6, ~(190) = 2 . 6 , p< . 01 ) . Results One hundred and twenty-eight students in the two sections of an introduction to mass communication class took part in Study 1 on March 5, 1991. Nearly 63 percent were female. A plurality listed themselves as advertising majors (45, or 36.3 percent). Issue Statements Students were first provided the issue statements (Table 6-1 contains the issue statements and Table 6-2 the attitude and relevance mean scores and the association of these scores with uncertainty orientation). Subjects were presented 17 issue statements followed by the attitude and relevance questions after each issue. A bogus

PAGE 60

53 agricultural act received the lowest relevance score (M = 2.4), while worries about the recession and job availability produced the highest score (M = 4.6). Relevance scores differing by about one-tenth of a point are significantly different at the .05 level. The recession/jobs issue met the criteria of both high relevance and apparently no relationship to uncertainty orientation. Of the "low relevance" issues, the best choice appeared to be the capital gains tax (M = 2.9), also not significantly related with uncertainty Table 6-1 Issue Questions and Mean Relevance and Attitude Scores Relev. Attit. 1. Congress should pass the recentlyproposed Agricultural Act of 1991. 2.4 3.9 2. The United States can help its own economy and that of poor nations with increased use of the International Monetary Fund. 2.9 4.4 3. The capital gains tax should be changed. 2.9 ,., 4. Mandatory AIDS testing should be required for some people. 3.1 3.1 5. Weak federal regulations are to blame for the U.S. savings and loan crisis. 3.1 4.1

PAGE 61

54 Table 6-1--continued 6. Reforms are needed in the use of pesticides in agriculture. 7. Taxes are so high that they must be cut even if it means a reduction of public services. 8. Senior comprehensive exams for graduating seniors is a good idea. 9. Women should have the right to an abortion if they wish. 10. We should support the allied Relev. Attit. 3.7 3.0 3.7 4.4 3.9 3.3 4.0 3.2 efforts in the Persian Gulf War. 4.1 3.3 11. U.S. universities should toughen education requirements to be competitive in the world. 4.2 3.8 12. President Bush is doing a good job. 4.2 2.4 13. Americans need to sacrifice more in order to cut our energy use. 4.2 3.0 14. We should do more to protect the environment, even at the cost of some businesses. 4.3 4.2 15. Jobs should go to the best person regardless of race or sex. 4.6 2.5 16. The proposed University of Florida tuition increase is not necessary. 4.6 4.0 17. The recession ••ans good jobs will be hard to come by for the next few years. 4.6 4.6 Note Issues in boldface are those selected for use in study 2. Mean scores of approximately 0.1 difference are significantly different at the .05 level.

PAGE 62

55 Table 6-2 Attitude and Relevance scores by uncertainty orientation Attitude Relevance Issue c-o u-o t c-o u-o t 1. 4.1 4.1 -0.2 2.5 2.3 1.0 2. 3.0 2.7 1.6 2.7 3.1 -2.1a 3. 3.2 3.2 -0.2 2.8 3.0 -1.1 4. 4.2 3.6 2.5b 2.8 3.3 -1.9 5. 3.3 3.3 0.1 3.0 3.1 -0.5 6. 4.0 4.0 -0.4 3.4 3.9 -1. 7a 7. 2.6 2.0 2.4b 3.4 4.0 -2.5b 8. 2.4 2.8 -1.4 3.7 4.2 -2.1a 9. 4.2 4.4 -1.0 3.5 4.4 -3.2c 10. 4.6 4.3 1.2 4.1 4.1 -0.2 11. 3.3 3.0 1.0 3.9 4.4 -2.oa 12. 4.2 3.7 2.4b 4.0 4.3 -1.2 13. 4.1 4.1 0.2 4.0 4.4 -2.la 14. 4.2 4.5 -1.4 4.0 4.3 -1.4 15. 4.6 4.4 1.1 4.3 4.7 -1.0a 16. 3.4 3.1 1.3 4.4 4.6 -0.8 17. 4.1 4 .1 -0.2 4. 5 4.7 -0.9 Note c-o is certainty-oriented, u-o is uncertainty oriented. a p<.05; b p<.01; c p<.001

PAGE 63

56 orientation. These two issues were significantly different in personal relevance (~(191) = 20.1, R<.001) and were selected for further use in the later study. uncertainty orientation A subset of the questions provided by Sorrentino and his colleagues was used after factor analyses in bothStudy 1 and Study 2 were inconclusive. The subset of six need for uncertainty and six authoritarianism items were used (a more detailed discussion of the factor analyses is provided in Chapter 7). The need for uncertainty scale had a mean of 4.0 (~ = 0.7) and the authoritarianism scale a mean of 2.8 (SD= 0.7). As expected, the two had a significant negative association(~= -.22, p<.001). The need for uncertainty measure demonstrated fair internal consistency (standardized Cronbach's Alpha of .77). The authoritarianism scale was less internally consistent, with a Cronbach's Alpha of .59. To create a resultant uncertainty orientation measure, scores on the authoritarianism scale were subtracted from the need for uncertainty scale, resulting in a scale that ranged from -1.5 to 3.5 (M = 1.3, SD= 1.1). Individuals high in this resultant measure are considered uncertainty-oriented, while those low are considered more certainty-oriented. As noted in earlier

PAGE 64

57 discussion of the construct, the middle tertile is typically dropped from further analysis, leaving the extremes of uncertainty-oriented and certainty-oriented persons but reducing the number of subjects available. This procedure was not done in this research. Instead, the middle tertile is included to keep these subjects in the analysis. Uncertainty-oriented subjects were more liberal (M = 3.7) than certainty-oriented subjects (2.7, ~(82) = 3.9, p<.001) However, uncertainty-oriented persons did not differ from certainty-oriented persons on newspaper or television reliance, the credibility they placed on polls reported in newspapers, or in the importance they place in other people's opinions. Assumptions Analysis of variance was used to test for significant interactions. There are three main assumptions in the use of analysis of variance: that there be normally distributed treatment populations, homogeneity of error variance, and independence of error components (Keppell, 1982). The F-distribution is robust against violations of the first two conditions, though such violations are more serious in situations with unequal cell sizes. In general, Monte Carlo studies have shown large sample sizes will offset any moderate violation of normality, with

PAGE 65

58 significance usually misinterpreted by the F-test by only a small amount (for example, a reported significance of .os when actually .06). Various statistical tests are available to compare variances (e.g., Hartley, Cochran, and Bartlett), though as Keppell (1982) notes, many share in common a sensitivity to departures from normality as well as homogeneity of variance. That is, though experimental groups may demonstrate homogeneity of variance, a lack of normality will result in these homogeneity statistical tests reporting a lack of homogeneity among groups. Use of the SPSS-X statistical package confines the tests to the Bartlett-Box F, Cochran's c, and Hartley's F max. (SPSS-X, 1988). Examination of study 1 results revealed no threats to the assumptions of homogeneity of variance or normality. This allows us to confidently examine the unequal cell sizes found in this study. The third assumption, independence of error components, means that treatment groups are in no way related to any other observation in the experiment. Random assignment, used in the experiment reported here, achieves independence. Tests of statistical significance are at the traditional probability level of .OS, though some results below the .10 level are reported when they provide insight into relationships.

PAGE 66

59 Attitudes on Two Issues As noted earlier, a pair of issues were selected to help study the effects of two poll story manipulations. The first manipulation had subjects receiving either a single-shot poll result or a poll with additional information that the trend of opinion was also in that direction. The second manipulation was of the poll sample, with the poll either being of Americans in general or of students. A control group with no manipulations was asked for an opinion. No significant interactions were found on the comprehensive exams issue (see Table 6-3 for the F-table, Table 6-4 for the cell means). Only one near-significant two-way interaction was found on the agricultural pesticides issue, that between uncertainty orientation and the student versus American poll manipulation (f(2,81) = 2.5, p<.09). This result offers little in terms of interest since it is apparently the "middle tertile" group causing the relationship. No main effects were found between the types of poll manipulations. However, those in the control group showed less agreement with pesticide reforms (M = 3.6) than those receiving any of the poll conditions (M = 4.1, 1(116) = 2.3, 2<.02). However, when looking at the more relevant comprehensive exams issue, no significant difference was

PAGE 67

60 found between persons in the control group and those receiving some kind of poll condition. In general, comparisons with the control group showed poll influence only on the less relevant issue (see Table 6-7). summary Two issues -jobs during a recession and the capital gains tax -were selected for Study 2 based on the requirements that they differ in relevance and that the Table 6-3 F-Table of Analysis of Variance of Poll Manipulations on the Agricultural Pesticides Issue Sum Of Mean Sig. Source of Variation Squares OF Square F Of F Main Effects 3.5 4 0.9 1.0 .39 1 Poll vs. Trend 0.4 1 0.4 0.5 .50 2 Students vs. Americans 0.1 1 0.1 0.1 .75 3 Uncertainty Orientation 3.0 2 1.5 1.8 .18 2-Way Interactions 5.5 5 1.1 1.3 .27 1 By 2 0.4 1 0.4 0.4 .51 1 By 3 0.8 2 0.4 0.5 .62 2 By 3 4.2 2 2.1 2.5 .09 3-Way Interactions 0.8 2 0.4 0.4 .62 1 By 2 By 3 0.8 2 0.4 0.4 .62 Explained 9.9 11 0.9 1.1 .40 Residual 68.3 81 0.8 Total 78.1 92 0.8

PAGE 68

61 attitude and relevance measures not be associated with uncertainty orientation. study 1 included a pilot experiment investigating whether using students as a sample in the bogus poll would be more influential than Americans in general, and whether a report of a single poll or poll with reports of a trend would be more influential. Few differences were expected given the mild manipulations. Table 6-4 Attitude about Agricultural Pesticide Reforms by Poll Manipulations and Uncertainty Orientation. Poll Trend Cert.-Oriented 4.0 4.2 Students Polled N = 10 N = 9 Moderate U-O 3.8 3.2 Students Polled N = 4 N = 5 Uncert.-Oriented 4.4 4.5 Students Polled N = 10 N = 8 Cert.-oriented 4.8 4.1 Americans Polled N = 4 N = 7 Moderate U-O 4.2 3.8 Americans Polled N = 9 N = 10 Uncert.-oriented 4.0 4.0 Americans Polled N = 10 N = 7 Note. Interaction F(2,81) = 0.4, ns.

PAGE 69

62 Table-6-5 F-Table of Analysis of variance of Poll Manipulations on the comprehensive Exams Issue sum Of Mean Sig. Source of Variation Squares OF Square F Of F Main Effects 9.2 4 2.3 1.1 .38 1 Poll vs. Trend 4.3 1 4.3 2.0 .16 2 Students vs. Americans 1.4 1 1.4 0.7 .42 3 Uncertainty Orientation 4.2 2 2.1 1.0 .38 2-Way Interactions 5.2 5 1.0 0.5 .79 1 By 2 0.9 1 0.9 0.4 .52 1 By 3 1.1 2 0.5 0.3 .78 2 By 3 3.0 2 1.5 0.7 .49 3-Way Interactions 1.4 2 0.7 0.3 .71 1 By 2 By 3 1.4 2 0.7 0.3 .71 Explained 15.9 11 1.4 0.7 .76 Residual 173.1 81 2.1 Total 189.0 92 2.1

PAGE 70

63 Table 6-6 Attitude about Senior comprehensive Exams by Poll Manipulation and Uncertainty orientation. Poll Poll/Trend Cert.-oriented 2.1 2.2 Students Polled N = 10 N = 9 Middle Tertile 2.3 3.0 Students Polled N = 4 N = 5 Uncert.-oriented 2.6 2.6 Students Polled N = 10 N = 8 Cert.-oriented 2.0 3.3 Americans Polled N = 4 N = 7 Middle Tertile 2.0 2.6 Americans Polled N = 9 N = 10 Uncert.-oriented 2.9 3.1 Americans Polled N = 10 N = 7 Note. No significant interaction.

PAGE 71

64 Table 6-7 Mean Attitudes (SD) Toward Two Issues by Poll/Trend and Student/American Manipulations Type of Issue Comp. Exams Pesticides Poll 2.4 (1. 4) 4.2 (0.9) Trend 2.7 (1. 5) 4.0 (1. 0) Students 2.4 (1. 4) 4.1 (1. 0) Americans 2.7 (1. 4) 4.1 (0.9) Poll By students 2.3a (1.4) 4.1 (0.9) Poll By Americans 2.4 (1. 4) 4.2 (0.9) Trend By Students 2.5 (1.4) 4.1 (1. 0) Trend By Americans 3. oab ( 1. 5) 4.0c (0.9) Control 2.4b (1. 4) 3.6c (1. 1) Note. Column entries sharing a superscript significantly different at the .05 level. 1 NOTES Because a particular type of computer-readable answer sheet was used, the scales had to be a l-to-5 measure instead of the usual six-point scale provided for uncertainty orientation. Both scales were presented in terms of statements where respondents could either answer a 1 for "disagree" across to a 5 for "agree." Responses of 2 or 4 were for "somewhat disagree" or "somewhat agree," while a 3 was for "neither disagree or agree." The typical z-score method of previous uncertainty orientation research was not used since

PAGE 72

2 65 both scales making up the variable have the same range (l-to-5) and the reporting of the means is perhaps more meaningful than that of z-scores. A third factor manipulated the direction of the poll result. Half of the subjects were to be told the poll supported senior comprehensive exams and reforms in agricultural pesticide use, while half were to be told the poll shows most people against the issues. A problem with question wording on the "against" manipulation, making it unclear what was being asked, was discovered after a first session with students. These 35 cases were dropped from analysis and a second session with another set of students did not include versions with this manipulation, leaving a total number of subjects at 128.

PAGE 73

CHAPTER 7 STUDY 2 METHODOLOGY The News Articles News Article Preparation Two short news articles about the capital gains tax and jobs were prepared with a desktop publishing system to resemble genuine newspaper articles (see Appendix A for versions of each article). Both articles contained two line headlines and bogus "jumps" informing the reader to continue reading on another page. In addition, a breakout quote or piece of information was presented in larger format to make the bogus poll results more accessible. The articles were written in a similar fashion as to structure, the placement and length of quotations, the use of sources, and placement of manipulated information. The Manipulation of Relevance Issues, no matter how careful the presentation, may differ in ways other than those intended or manipulated. Despite the relevance differences inherent in the issues as reported by Study 1 subjects, the first stimulus has to do with personal relevance. It is apparent that the capital gains tax and the possibility of a tight job market are very different, not only in the relevance people may place in them, but also 66

PAGE 74

67 in the kind of issues they represent. The capital gains tax is a largely political matter. The jobs issue can be viewed politically, but it is likely subjects would consider it on other, more personal grounds. While this may seem to fit the "relevance" definition, it is possible subjects might fit the jobs issue into a different kind of category. Persuasion research in the cognitive response paradigm typically manipulates relevance on the same issue (see Petty & Cacioppo, 1986a, 1986b, for examples; O'Keefe, 1990, for a discussion). By manipulating relevance through telling subjects that the issue is or is not important to them at the present, researchers are able to avoid the pitfalls that come with differences between issues other than relevance itself. The question is one of theoretical and methodological control versus a more "real-world" test. This dissertation attempts both by providing subjects two issues apparently differing in personal relevance and then manipulating relevance within each issue. The manipulation occurred in the second paragraph of each version of the article. In the "low relevance" condition of the news article on the capital gains tax, for example, readers were told that "experts predict the final decision will not affect college students." In the jobs and recession article with a "high

PAGE 75

68 relevance" manipulation, readers were told that "analysts predict most college students will be greatly affected." The Manipulation of consensus The second manipulation had to do with perceptions of consensus through use of poll results. Subjects could receive one of three kinds of stories: a poll result in the direction of the article's chief arguments, a poll in the opposite direction, and a control group with no poll results. For example, in the capital gains tax article, subjects in the supporting poll condition--where poll results were in the direction of the majority of arguments presented--the fourth paragraph informed readers that a "recent nationwide public opinion poll showed 81 percent of Americans support the reduction of the capital gains tax, with 15 percent against and 4 percent undecided." In addition, the "blurb" or "pull out" added that "polls show most Americans are for a reduction in the capital gains tax." Subjects could also receive a version with the polls results contradicting the article's argument. The same sentence on poll results was provided, but instead the majority argued either against a capital gains tax or against the likelihood that jobs would be difficult to find for college students. A control group received a slightly shorter article (minus the paragraph on the poll

PAGE 76

69 results) and with a "blurb" pulled from the article itself. No hypotheses were presented concerning possible differences between how people process a supporting or contradictory poll or differences in the persuasive abilities of either type of poll. On the one hand, we might expect an contradictory poll to promote central processing, perhaps making the poll less influential (similar to the argument of Mackie, 1987). On the other hand, when the issue is of little relevance, we might expect the contradictory poll to persuade people as a consensus heuristic. This dissertation does not directly probe the different processes, but does use this research as a guide in explaining results. In general, it is expected that the contradictory poll will act in a similar fashion as the supporting poll. Summary of Manipulations Volunteers from an introductory advertising class participated in Study 2 over a two-week period for extra credit. Subjects could participate on March 19 or 20, 1991, for what will be called from this point Tl. During this time subjects were measured on attitudes about the two issues selected for further study, uncertainty orientation, demographics, and other variables.

PAGE 77

70 The design was a 3 (uncertainty versus middle versus certainty oriented) X 2 (low and high relevance) X 3 (poll result for or against or control) factorial for both issues. Subjects therefore could be assigned to any of six conditions for each article (three poll manipulations and two relevance manipulations). This meant a 6 X 6 or 36 different questionnaire versions. In addition, the order of article presentation was randomized to ensure no order effects would be present, resulting in 72 different questionnaire versions. Based on their categorization on uncertainty orientation, subjects were assigned to any of these 72 versions. Each subject's computer-readable bubble sheet from Tl was clipped to the assigned T2 version and subjects picked up these instruments at the T2 stage. To simplify presentation of the results, the condition where the poll agrees with most of the news articles will be called the "supporting poll" condition; the condition in which the poll results are the opposite of the arguments will be called "contradictory." Before returning one week later for T2, uncertainty orientation scores were determined and subjects randomly assigned to experimental conditions. The T2 instrument opened with additional media use questions then presented the first news article followed by a thought-listing task

PAGE 78

71 and attitude questions. A set of filler questions followed, and then the second article was presented along with its questions. The order of the two articles was randomized to ensure no order effects occurred. The chief independent variable measured at Tl was uncertainty orientation, handled in the same way as in Study 1 except for differences in scaling. Rather than a l-to-5 scale, a different computer-readable answer sheet version was used, this version ranging from Oto 9. This scale was adopted because it allows more movement on the attitude measures, a greater range for uncertainty orientation responses, and the possibility of answers beyond the l-to-5 range constraints such as age or days a week subjects read a newspaper. Measurement of Variables This study views attitudes as the evaluation of an object and as representing a residue of experience; different from moods, which are more temporary, attitudes are relatively enduring though susceptible to change (O'Keefe, 1990). Attitude was measured by the same question at Tl and T2 on both issues. The results of the Tl attitude measure were used as a covariate and the results at T2 were used as the dependent variable on each issue in an analysis of

PAGE 79

72 covariance. This allowed the statistical control of attitudes before the experimental manipulation. Subjects were asked, on a o-to-9 scale, to strongly disagree or strongly agree to the following statements: "The recession means jobs for college graduates will be hard to come by for the next several years" and "The federal government should cut the capital gains tax." The questions at Tl were listed with a number of other attitude measures to ease suspicion. A second approach to this dependent variable, and one perhaps more closely resembling traditional bandwagon research, is to look for differences between Tl and T2. To investigate this possibility, a change score was also computed by subtracting the Tl attitude from the one at T2. Therefore, a negative score represents a decrease in support for the issue statement and a positive score an increase in support. Uncertainty orientation As noted briefly in the previous chapter, factor analysis of the full 35-item index provided by Sorrentino and his colleagues was problematic. In both studies, a principal components and principal factor analysis using varimax and oblique rotations arrived at a 10-factor solution, some factors consisting of only a single item. This analysis, however, could be interpreted as a three

PAGE 80

73 or perhaps two-factor solution (Appendix B contains all questions with means and standard deviations, Appendix C the 10-factor solution). The first factor appears to tap the general construct of need for uncertainty, although one authoritarianism question also loaded on this factor. The second factor was made up of only three questions asking about respect for authority, parents, and friends. The third factor are all authoritarianism questions generally aimed as beliefs and values. Forcing a two-factor solution from all these questions resulted in some authoritarianism items loading on the need for uncertainty scale, though no uncertainty items loaded on the authoritarianism factor. After scrutiny of the factor analysis results and using the larger Study 2 subject pool, a subset of 12 items was selected and a second set of factor analyses conducted (factor analysis of the Study 1 pool resulted in roughly the same solution). The six need for uncertainty questions asked whether subjects enjoyed spending time discovering new things or finding out why things happen. The six selected authoritarianism measures focused on the need for discipline, obedience, and respect for authority. A principal factors analysis with a varimax rotation showed the items explaining 36.5 percent of the variance with a correlation between the factors of -.24. Both

PAGE 81

74 scales demonstrated moderate internal consistency, with Cronbach's Alpha of .81 for the need for uncertainty scale and at .73 for the authoritarianism scale. The mean score for need for uncertainty was 6.3 (fill= 1.4) and for authoritarianism was 4.5 (s..12 = 1.6). As in study 1, the authoritarianism score for each subject is subtracted from the need for uncertainty orientation score and an overall index constructed. A tertile split was conducted with the middle third included in analysis (see Table 1 for means and factor loadings of the 12 questions used in this study). The shortened instrument resulted in little movement of subjects from being categorized as either uncertainty oriented or certainty-oriented under the original format (X 2 (4) = 238.5, p<.0001). Only six persons found themselves classified differently, three in each direction. That is, three originally classified as certainty-oriented were later classified as uncertainty oriented, and vice versa.

PAGE 82

75 Table 7-1 Factor AnalYsis Results of Study 2 uncertainty Orientation Questions Question I enjoy spending time discovering new things. I like to find out why things happen. I like to fool around with new ideas, even if they turn out later to be a total waste of time. I enjoy thinking about ideas that challenge my views of the world. I usually try to learn about something I don't understand, even if I might end up in trouble because of it. I often put myself in situations in which I could learn something new. What the youth needs most is strict discipline, rugged determination, and the will to work and fight for family and country. Mean SD 6.5 1.8 6. 9 1. 8 6.1 2.0 6.9 1.9 5.6 2.2 6.5 1. 7 4.6 2.3 Fl F2 .71 -.16 67 -.05 .66 -.13 .61 -.24 .60 -.23 .59 -.17 -.17 .62

PAGE 83

Table 7-1--continued Question There is hardly anything lower than a person who does not feel a great love, gratitude, and respect for his or her parents. Young people sometimes get rebellious ideas, but as they grow up they ought to get over them and settle down. Obedience and respect for authority are the most important virtues children should learn. No sane, normal, decent person could ever think of hurting a close friend or relative. Books and movies ought not to deal so much with the unpleasant and seamy side of life; they ought to concentrate on themes that are entertaining or uplifting. Eigen Value Percent Variance Explained 76 Mean so 4.9 2.5 4.9 2.3 4.4 2.5 4.8 2.7 3.6 2.5 Fl -.09 -.17 -.12 -.15 -.12 2.8 23.3 F2 .60 .57 .55 .53 .45 1.6 13.2 Note Loadings above are from the rotated structure matrix of a principal factors analysis with oblique rotation. The correlation betwen factors was -.24.

PAGE 84

CHAPTER 8 STUDY 2 RESULTS The experiment in study 2 was conducted over a twoweek period. 1 Two-hundred and eighty-nine subjects participated in the first phase of the experiment (Tl). Two hundred and sixty-three returned for T2, a loss of approximately 9 percent. The average age of all subjects was 20.6 (SD= 2.7). Fifty-nine percent of the subjects were female and 41 percent male. Nearly half reported their academic standing as being juniors (137, or 47.4 percent). Assumptions Analysis of covariance were the primary statistical tests used in analysis. 2 As in study 1, examination for assumptions found no violations of normality or homogeneity of variance. Study 1 was used to select issues of differing relevance and with no relationship to uncertainty orientation. Association of simple issues statements with uncertainty orientation may differ in news articles as compared to simple issue statements. T-tests of pretest attitudes (before the manipulations) revealed no significant difference in attitudes toward either issue and uncertainty orientation. 77

PAGE 85

78 Manipulation Checks In a manipulation check of the poll condition, subjects were asked to agree or disagree with a statement that most Americans agree with the stance argued in the article. If the manipulation was successful, those receiving the condition where poll results favored the article should provide a higher score (measured on a o-to9 scale) than those in the control group, and those receiving a poll in the opposite direction (a contradictory poll) should demonstrate less agreement as compared to the control group. T-tests show significant differences in all cell means in the expected directions (see Table 8-1). Manipulations of relevance were mixed, however. A question after each article asked subjects to agree or disagree (on a o-to-9 scale) as to whether "the article was very relevant to me personally." As Table 8-2 shows, the manipulation appears to have worked on the capital gains tax article, with those receiving the low relevance paragraph (M = 3.1) reporting less personal relevance than those receiving the high relevance paragraph (M = 3.7, t(261) = -2.5, p<.01). However, there was no significant difference in relevance reported for the highly relevant jobs issue manipulation, perhaps due to a ceiling effect of the already high relevance associated with that

PAGE 86

79 Table 8-1 Means (SD) of Manipulation Check Asking How. Likely Most Americans Agree by Poll Type and Issue Poll Manipulation Received Issue Contradictory Poll Control Supporting Poll Jobs For 4. 8 ( 3. 2) 6. 4 ( 2. 0) 7.6 (1.6) Graduates Capital 3.7 (2.9) 4.6 (1.7) 7.1 (1.8) Gains Tax Note. All row entries are significantly different from each other at the .01 level by t-tests. Table 8-2 Means CSD) of Manipulation Check of Effects of Experimental Manipulation of Personal Relevance on Relevance Scores by Issue Relevance Manipulation Received Issue Low High Jobs For Graduates 7.1 ( 2. 0) 7.4 ( 2 .1) Capital Gains Tax 3.la ( 2. 1) 3.7a (2.3) Note. Entries sharing a superscript are significantly different at the .01 level by t-tests.

PAGE 87

80 issue. 3 Relevance scores associated with the jobs issue are significantly higher than those associated with the capital gains issue regardless of relevance manipulation. Tests of Hypotheses This study has three chief hypotheses. The first, based on cognitive response models of persuasion (e.g., the elaboration likelihood model of Petty & Cacioppo, 1986a, 1986b), predicted that poll results would be more influential on attitudes about less relevant issues as compared to more relevant issues. The second hypothesis had to do with explaining why the bandwagon or underdog effects are seen in different studies. Following the reasoning of Sorrentino (Sorrentino & Hancock, 1987), it was hypothesized that uncertainty-oriented persons are more likely to wonder why a minority thinks the way it does, to "go with the underdog," to react against a poll result as compared to certainty-oriented persons, while certainty-oriented persons would be more likely to "jump on the bandwagon," to be influenced by a poll. Third, following the work of Sorrentino, et al. (1988), it was hypothesized that in high-relevant situations, certainty oriented persons would be even more likely to be influenced by a poll as compared to uncertainty-oriented

PAGE 88

81 persons. The results section is organized around these three hypotheses. In the following results, T2 attitudes are used as the dependent variable and the Tl attitudes treated as covariates. For the capital gains tax issue, the Tl attitude had a raw regression coefficient of 0.18. For the jobs and recession issue, the raw regression coefficient was 0.30. Where change scores are reported in tables (usually in parentheses), these are the result of a subtraction of T2 attitude from T1 attitude. Statistical tests are for the attitude measures. Change scores are provided as a guide for change between Tl and T2. Finally, before moving to the results, recall that if the poll is influential, those in the contradictory poll condition should demonstrate lower attitude scores as compared to the control group. on the other hand, an influential supporting poll should result in attitude scores greater than those seen among those in the control group. Hypothesis 1 Poll Effects and Relevance The first hypothesis predicts that polls are more influential in situations of low relevance (Petty & Cacioppi, 1986a, 1986b). However, poll influence was found only on the more relevant jobs issue (see Table 83). Those receiving the supporting poll showed greater

PAGE 89

82 approval of the article's arguments (M = 6.8) than those in the control group (M = 6.2, t(l75) = -2.0, p<.05) or those receiving the contradictory poll (M = 6.0, t(l70) = -2.5, p<.01). No significant differences were found on the capital gains tax issue (F-tables for both issues are found in Table 8-4 and 7-5). capital gains tax issue Looking within issues, a two-way interaction between poll manipulation and personal relevance was found on the capital gains tax issue (!(2,242) = 3.2, p<.05). First, looking at the contradictory poll, no influence was seen Table 8-3 Means of T2 Attitudes (Change Scores) by Issue and Poll Manipulation Received Issue Capital Tax Jobs For Graduates Poll Manipulation Received Contradictory Poll Control 4. 8 ( +0. 3) N = 88 6.0 (-0.6)a N = 85 5.2 (+0.3) N = 87 6. 2 (-0. 2) b N = 90 Supporting Poll 5.1 (+0.5) N = 85 6.8 (+0.6)ab N = 86 Note. Row entries sharing superscript are significantly different at the .05 level by t-tests. Cell means for the jobs issue are significantly higher than those on the capital gains tax issue.

PAGE 90

83 Table 8-4 F-Table of Analysis of covariance of capital Gains Tax Issue by Poll and Relevance Manipulations and uncertainty Orientation Sum Of Mean Source of Variation Squares DF Square Covariates 31.4 1 31.4 Tl Attitude 31.4 1 31.4 Main Effects 9.4 5 1.9 1 Poll Type 4.6 2 2.3 2 Relevance o.o 1 0.0 3 Uncertainty Orientation 4.2 2 2.1 2-Way Interactions 70.3 6 8.8 1 By 2 22.5 2 11.3 1 By 3 27.0 4 6.8 2 By 3 17.0 2 0.9 3-Way Interactions 3.8 4 0.9 1 By 2 By 3 3.8 4 0.9 Explained 114.9 18 6.4 Residual 840.0 242 3.5 Total 954.9 260 3.7 Covariate Raw Regression Coefficient Tl Attitude .17 F 9.0 9.0 0.5 0.7 o.o 0.6 2.5 3.2 1.9 0.3 0.3 0.3 1.8 Sig. Of F .00 .00 .74 .51 .91 .55 .01 .04 .10 .90 .90 .90 .02

PAGE 91

84 Table 8-5 F-Table of Analysis of covariance of Jobs Isue by Poll and Relevance Manipulations and uncertainty orientation Sum Of Mean Source of Variation Squares OF Square Covariates 114.4 1 114.4 Tl Attitude 114.4 1 114.4 Main Effects 178.5 5 35.7 1 Poll Type 54.4 2 27.2 2 Relevance 65.0 1 65.0 3 Uncertainty Orientation 63.0 2 31.5 2-Way Interactions 35.2 8 4.4 1 By 2 6.9 2 3.5 1 By 3 27.4 4 6.9 2 By 3 1.9 2 0.9 3-Way Interactions 35.2 4 8.8 1 By 2 By 3 35.2 4 8.8 Explained 363.3 18 20.2 Residual 949.0 243 3.9 Total 1312.3 261 5.0 Covariate Raw Regression Coefficient Tl Attitude .30 F 29.3 29.3 9.1 7.0 16.6 8.1 1.1 0.9 1.8 0.2 2.3 2.3 5.2 Sig. Of F .00 .00 .00 .00 .00 .00 .35 .41 .14 .78 .06 .06 .00

PAGE 92

85 among those receiving the high-relevance manipulation (M = 4.8 for both the control group and contradictory poll conditions; see Table 8-6). Among persons receiving the low-relevance version, those in the control group were less approving of the issue (M = 5.6) than those receiving the contradictory poll (M = 4.8, ~(87) = -1.7, p<.05). 4 An effect opposite the one hypothesized is found when considering the supporting poll. For persons in the low relevance condition, those receiving the poll were less approving (M = 4.8) than those in the control group (M = 5.6, t(85) = 2.2, p<.02). Among those told the issue was Table 8-6 Means of T2 Attitudes (Change scores) on Capital Gains Tax by Poll and Relevance Manipulations Poll Manipulation Received Contradictory Poll Low Relevance High Relevance 4.8 (+0.l)a N = 46 4.8 (+0.5) N = 43 Control Supporting Poll 5 6 ( + O 8 ) ab N = 42 4.8 (-0.l)c N = 45 b 4.8 (+0.1) N = 44 5. 4 ( +0. 9) c N = 41 Note. Interaction ~(2,254) = 3.2, p<.04. Entries sharing a superscript are significantly different at the .05 level.

PAGE 93

86 of high personal relevance, those receiving the poll were more supportative (M = 5.4} than those in the control group (M = 4.8, 1(84} = -1.8, p<.04}. Jobs and recession issue There was no significant two-interaction between relevance and poll found on the jobs issue (~(2.243} = 0.9, p<.41; Table 8-7). No differences were found between the control group and those receiving the contradictory poll regardless of relevance condition. Also, no difference between the control group and the supporting poll group was found among those in the high-relevance condition (M = 6.8 and M = 7.0, respectively). However, a Table 8-7 Means of T2 Attitudes {Change Scores) on Jobs by Poll and Relevance Manipulations Poll Manipulation Received Contradictory Poll Low Relevance High Relevance 5.3 (-1.2) N = 45 6. 7 (+0.1) N = 40 Control Supporting Poll 5.6 (-1.0}a N = 46 6.8 (+0.8} N = 45 6.7 (O.O)a N = 43 7.0 (+1.0} N = 43 Note. No significant interaction. Entries sharing a superscript are significantly different at the .05 level by t-tests.

PAGE 94

87 t-test of the mean scores among those told the issue was of little relevance showed those in the control group less supportive (M = 5.6) than those receiving the supporting poll (M = 6.7, t(87) = -2.5, p<.01). summary of hypothesis 1 results The poll appeared to influence subjects on the more relevant jobs issue, but not on the less relevant capital gains tax issue. Looking at within-issue relevance, the contradictory poll influenced only those in the low relevance condition, while the supporting poll influenced persons in both relevance conditions, but not in a way that would be hypothesized -those in the low relevance condition demonstrating more approval when shown the supporting poll, while those told the issue was very relevant showed less approval. Hypothesis 2 Polls and Uncertainty orientation It was hypothesized that uncertainty-oriented persons would be less persuaded by a poll than certainty-oriented persons. Capital gains tax issue A two-way interaction near the traditional level of significance was found for poll manipulation and uncertainty-orientation (f(4,242) = 1.9, p<.10; see Table 8-8) .

PAGE 95

88 Looking first at the contradictory poll, certainty oriented persons in the control group were more supportive (M = 5.5) than those receiving the poll (M = = 4.4, t(54) = -2.1, p<.03). The mean scores of uncertainty-oriented persons were in the expected direction, but were not statistically signficant, with those showing greater approval when in the contradictory poll condition (M = 5.3) than in control group (M = 4.7, t(55) = -1.1, p<.14). Table 8-8 Means of T2 Attitudes
PAGE 96

89 No significant differences were found between the control group and those receiving the supporting poll. Jobs and recession issue No significant two-way interaction between poll and uncertainty orientation was found on the jobs issue (f(4,243) = 1.8, ~<.14: see Table 8-9). T-tests of the expected differences found no significant poll influence among certainty-oriented persons. Uncertainty-oriented persons were influenced by both poll versions as compared to the control group (M = 5.7), with those receiving the Table 8-9 Means of T2 Attitudes (Change Scores} toward Jobs and Recession by Poll and Uncertainty Orientation Poll Manipulation Received Contradictory Poll Control Supporting Poll Certainty 6.3 (-0.7) 5.8 (-0.6) 6.6 (+0.1) Oriented N = 27 N = 40 N = 18 Moderate 6.4 (+0.2) 7.3 (+1.0) 7.0 (+1. 4) Uncert-O. N = 34 N = 24 N = 33 Uncertainty 4.5 (-2.2)a 5.7 (-0.6)ab 6.8 (+O.l)b Oriented N = 24 N = 27 N = 35 Note. Interaction f(4,243) = 1.8, ns. Entries sharing a superscript are significantly different at the .05 level by t-tests.

PAGE 97

90 contradictory poll showing a decrease in support (M = 4.5, t(49) = -1.7, p<.05) and those receiving the supportive poll increasing in support (M = 6.8, t(66) = -1.9, p<.02). Summary of hypothesis 2 results The capital gains tax issue offered some support for the second hypothesis. Among those receiving the contradictory poll, certainty-oriented persons were more likely to move in the direction of the poll while uncertainty-oriented persons were more likely to move in the opposite direction, as compared to the control group. No support for the hypothesis was found in the jobs issue. Hypothesis 3 Polls. Relevance. and Uncertainty Orientation The third hypothesis predicted that certainty oriented persons would be more sensitive to situations of high relevance and would be more influenced by polls in those situations. Capital gains tax issue No significant three-way interaction was found between uncertainty orientation, poll type, and personal relevance on the capital gains tax (E(4,242) = 0.3, p<.90; see Table 8-10). T-tests to explore the hypothesized relationship found a pattern in comparing certainty oriented and uncertainty-oriented persons in the control group and contradictory poll condition, though neither of the findings reaches the traditional level of

PAGE 98

91 significance. In these, certainty-oriented persons told the issue was of personal relevance showed more approval in the contradictory poll condition (M = 4.1) than those in the control group (M = 4.9, t(55) = 1.1, p<.15). Uncertainty-oriented persons in the high-relevance condition, however, moved in the opposite Table 8-10 Means of T2 Attitudes (Change scores) Toward capital Gains Tax by Poll. Relevance. and uncertainty orientation Contradictory Poll Cert.-Oriented Low Relevance Mod. Unc-o Low Relevance. Uncert.-Oriented Low Relevance Cert.-Oriented High Relevance Mod. Unc-O High Relevance Uncert.-Oriented High Relevance 4. 7 ( +o. 6) a N = 14 4.4 (-0.4) N = 19 5.5 (+0.2) N = 13 4.1 (-0.1) N = 15 5.4 (+0.3) N = 10 5.1 (+1.1) N = 18 Control Supporting Poll 6.0 (+0.5)ab N = 13 5. 4 (+0. 7) N = 20 5.7 (+1.6)c N = 9 4.9 (+0.5) N = 13 5.3 (-0.4) N = 16 4.1 (-0.3) N = 16 b 5. 1 c +o. 5) N = 15 5.0 (+0.2) N = 11 4. 3 (-0. 3) C N = 18 5.1 (+0.6) N = 14 6.1 (+1.8) N = 16 4.8 (0.0) N = 11 Note. No significant interaction. Selected entries sharing superscripts significantly different at the .05 level by t-tests.

PAGE 99

92 direction, with those receiving the contradictory poll demonstrating less approval (M = 5.1) than those in the control group (M = 4.1, t(32) = 1.5, p<.07). Jobs and recession issue A significant three-way interaction was found between uncertainty orientation, poll type, and relevance on the Table 8-11 Means of T2 Attitudes toward Jobs Issue by Poll. Relevance. and Uncertainty Orientation Contradictory Poll Cert.-Oriented Low Relevance Mod. Unc-o Low Relevance Uncert.-Oriented Low Relevance Cert.-oriented High Relevance Mod. Unc-o High Relevance Uncert.-Oriented High Relevance 4.9 (-1.9) N = 14 6.0 (+0.1) N = 21 4.4 (-2.9) N = 10 7.9 (+0.6)b N = 13 6.8 (+0.3) N = 19 4.6 (-1.4) N = 8 Control Supporting Poll 5.4 (-1.4)a N = 19 6.7 (+0.4) N = 7 5.4 (-1.3) N = 20 6.3 (+O.J)b N = 15 7.5 (+1.2) N = 17 6.3 ( +0. 6) N = 13 7. 1 ( +o. 2) a N = 14 6.7 (+0.6) N = 14 6.3 (-0.6) N = 16 6.0 ( 0. 0) N = 11 7.4 (+2.2) N = 13 7.3 (+0.7) N = 19 Note. Interaction E(4,243) = 2.3), R<.07. Entries sharing a superscript are significantly different at the .05 level by t-tests.

PAGE 100

93 jobs issue (~(4,243) = 2.3, p<.06; see Table 8-11). Instead of the hypothesized effect of high relevance on certainty-oriented persons, those receiving the contradictory poll were more approving of the jobs issue (M = 7.9) than those in the control group (M = 6.3, t(26) = 2.1, p<.03). No difference was found among certainty oriented persons in the high relevance condition either in the control group of receiving the supporting poll. Poll influence was also seen in situations of high relevance for uncertainty-oriented persons (rather than certainty oriented, as hypothesized). Among uncertainty-oriented persons told the issue was of high relevance, those in the control group showed more approval (M = 6.3) than those receiving the contradictory poll (H = 4.6, t(l9) = -1.4, p<.09) and less approval than those receiving the supporting poll version (H = 7.3, t(30) = -1.2, p<.13), though the last finding is not statistically significant. Summary of hypothesis 3 results Little support for hypothesis 3 was found on the capital gains tax issue, but results from the jobs and recession issue were opposite the direction predicted. On the less relevant capital gains tax issue, certainty oriented persons told the issue was of high relevance and who received a poll contradictory to the main thrust of the news article were more influenced in the direction of

PAGE 101

94 the poll while uncertainty-oriented persons shifted in the opposite direction. On the jobs issue, certainty-oriented persons told the issue was of high personal relevance moved in the opposite direction of the poll result, while uncertainty-oriented persons showed movement with the poll. Other Significant Findings A near-significant two-way interaction was found for uncertainty orientation and relevance on the capital gains tax issue (~(2,242) = 2.5, p<.09}. The effect of relevance on those who scored in the middle tertile of the uncertainty orientation measure appears to be cause of the interaction, with those receiving the high-relevance manipulation showing more approval for the tax (M = 5.6} than those told the issue was of little relevance (M = 4. 9) Main effects were found only on the jobs and recession issue. A significant main effect was found the the manipulations of the poll (already discussed} and relevance (~(1,243) = 16.6, p<.001). Those told the issue was of high relevance showed less approval (M = 5.8} than those told the issue was of little relevance (M = 6.8). Finally, a main effect was found on the same issue for uncertainty orientation (~(2,243} = 8.1, p<.001}, with the middle tertile (M = 6.8) again showing differences from

PAGE 102

95 either those certainty-oriented (M = 6.2) or uncertainty oriented (M = 5.9). 1 2 3 4 Notes Approximately 75 subjects attended each of four possible sessions on one week and returned for one of four sessions the following week. All sessions were held in the afternoons on the same days and in the same classroom with only the primary investigator present. The "classic experimental approach" was used (SPSS-X, 1988). The analysis of covariance method first assessed and adjusted for the effects of the covariate, then for the factors, and finally for the two-way interactions and three-way interactions. Logarithmic transformations of the data using both a common (base 10) and natural (base e) system were conducted to see if the ceiling effect could be removed (Cohen & Cohen, 1975). The analysis found no significant difference on the relevance manipulation on the jobs issue). T-tests for simple main effects among the cells are conducted only among hypothesized relationships. For example, not-tests were conducted on differences among those categorized as "moderate" on uncertainty orientation.

PAGE 103

CHAPTER 9 DISCUSSION summary of Study Findings This dissertation investigated a possible theoretical explanation for inconclusive results found in public opinion poll research. Its second purpose was to explain the occurrence of occasional bandwagon and underdog findings with the individual difference variable uncertainty orientation. The first study was aimed at setting up study 2, where three main research hypotheses were presented. These are summarized below, along with the evidence from the results that bear on each one and a brief discussion. The study's limitations and problems follow, along with a discussion of uncertainty orientation and future research concerns. Hl--Relevance and Poll Influence First, it was hypothesized that one explanation for the number of mixed findings from poll influence studies may be found through use of a different theoretical perspective {i.e., cognitive response persuasion research). This conceptual framework proposes personal relevance as a key variable in providing the motivation for careful scrutiny of a message. Persuasion may still 96

PAGE 104

97 occur in situations of high relevance, but it is more likely to be due to thoughts about the issue itself and less likely to be the result of extraneous or source characteristics. In a low relevance situation, where little motivation for careful scrutiny exists, persuasion is more likely to occur through use of some heuristic or shortcut device, such as a credible source or that most other people think that way on an issue. Specifically, it was hypothesized that public opinion polls would be more influential in issues of low as compared to high relevance. Study 1, though not designed for hypothesis testing, provides some support. No influence was found on the more-relevant comprehensive exams issue, but the poll manipulations were found to have some influence on attitudes about the less-relevant pesticide reform issue. More important was the support from Study 2. Issues differing a priori in relevance were presented to subjects in an experimental setting. Also, the relevance of each issue was itself manipulated through a paragraph in each bogus news article. A poll was found to be influential on the more relevant jobs issue but not on the less relevant capital gains tax issue, only in the case of a poll with results congruent to the main thrust of the article. This between-issue attitude difference held when looking

PAGE 105

98 at the change scores "corrected" from the control group scores. Looking at the within-issue manipulation of relevance, matters become more complicated. On the capital gains tax, the incongruent poll influenced persons in the low relevance condition, but not those in the high relevance condition. This finding supports the general thesis that polls will be influential in situations of low relevance. However, the congruent poll led to increased support in the high relevance condition but decreased support in the low relevance condition, a result opposite of that hypothesized. To put this in bandwagon/underdog terms, it appears a congruent poll resulted in a bandwagon effect among those told an issue is personally important, but an underdog effect for those told an issue was not personally important. On the jobs issue, there was no difference among the personal relevance manipulations in the incongruent poll condition. However, among those receiving the congruent poll, there was movement in the direction of the poll only among those told the issue was not of great personal importance. Thus, at least for a congruent poll, relevance appears to explain a bandwagon effect. To summarize, there is some support for this hypothesis. When looking at issues differing in relevance

PAGE 106

99 we find poll influence in the high-relevant rather than low-relevant issue. When looking at withi~-issue differences in relevance, however, we find support for the notion that polls are more influential in situations of low relevance. H2--uncertainty orientation and Poll Influence Conformity research has found some instances of minority influence. As Moscovici and Personnaz (1980) note, consideration of minority viewpoint leads to influence by a minority or at least less influence by a majority. It was hypothesized that uncertainty-oriented persons would be more likely to consider a minority viewpoint and would therefore be less influenced by a poll and move in the "underdog" direction as compared to those who were certainty-oriented. Again, the evidence was mixed. On the capital gains tax, certainty-oriented persons demonstrated almost no change regardless of the poll. Attitudes of uncertainty-oriented persons reading the congruent poll results, however, moved in the opposite or underdog direction as hypothesized. No support was found in the jobs issue. H3--Uncertainty Orientation and Personal Relevance Finally, following the work of Sorrentino et al. (1988), it was hypothesized that certainty-oriented persons would, in situations of high relevance, be more

PAGE 107

100 susceptible to poll influence as compared to uncertainty oriented persons. Some support for this hypothesis was found among attitudes of those receiving the incongruent poll on the capital gains tax issue. Certainty-oriented persons told the issue was of personal importance showed an attitude shift in the direction of the poll; the attitudes of uncertainty-oriented persons shifted in the opposite or underdog direction. The jobs and recession issue provided no support for the hypothesis. overall, there was little support for this hypothesis. Limitations of the studies Reliability and Validity Reliability is the "extent to which an experiment, test, or any measuring procedure yields the same results on repeated trials" (Carmines & Zeller, 1979, p. 11). There are a number of means of assessing reliability, from the methods of retest, alternative-form, split halves, and internal consistency. The latter was used to explore uncertainty orientation, with Cronbach's alpha as the test statistic. The internal consistency of these two variables was of moderate strength. Carmines and Zeller (1979) define validity as the extent to which any measuring instrument measures what it is intended to measure. But, as Cronbach (1971) notes,

PAGE 108

101 "One validates, not a test, but an interpretation of data arising from a specified procedure" (p. 447). Therefore, an instrument may be valid for measuring one kind of phenomenon but not valid for another kind. Three types of validity discussed by the authors are criterion-related validity, content validity, and construct validity. Criterion-related, or predictive, validity is of importance when an instrument is said to estimate some behavior outside of the measuring instrument itself. If the projective test of uncertainty orientation had been used then this would present less of a problem in this study. The projective measure has a short but strong explanatory research record. No use of the new instrument has been published. content validity refers to the degree to which a measure covers the range of meanings included within the concept. One way to assess content validity is to submit the measure to the scrutiny of experts or some other panel. This was not done here. Uncertainty orientation can be broken into two aspects, need for uncertainty and authoritarianism. As discussed earlier, authoritarianism is a concept that has existed since the 1940s. While measures of authoritarianism have been strongly criticized by some, others have pointed to it as a valid measure of the authoritarian nature of subjects. Finally, construct

PAGE 109

102 validity is based on the way a measure relates to other variables within a system of theoretical relationships. Experiments and correlational studies with the index have produced what seems to be a reasonable number of expected relationships (see Chapter 4). Overall, some doubt exists about the validity of uncertainty orientation as originally presented. Even the subset of questions used offers problems, such as the moderate correlation with ideology. Recent unpublished work by Sorrentino, however, has shown uncertainty orientation differences as expected among students at an early age and even differences between gifted students and "average" people. Recall that the original 35-item questionnaire provided by Sorrentino and his colleagues resulted in an inconclusive 10-factor solution when two factors--need for uncertainty and authoritarianism--should have been found. It was decided that a subset of the questions that loaded most strongly together, and that on their face seemed essentially to tap the constructs of interest, be selected for further analysis. The 10-factor rather than 2-factor solution calls into question the validity of the original 35-item questionnaire.

PAGE 110

103 Internal and External Validity A number of concerns and criticisms have been raised about experiments as an appropriate means for the study of complex social beings (Cronbach, 1975; Gergen, 1973, 1976, but also see rebuttals by Greenwald, 1976, and Manis, 1976). The attraction of the experiment is its power to probe causal relationships, the chief reason for Campbell's (1958) assertion that experimental studies were necessary to unravel the mystery of the elusive bandwagon effect. But experiments come with their own problems. Typically in experiments we are more concerned with internal rather than external validity, though the question of generalizability is never definitively settled. Campbell and Stanley (1963) and Cook and Campbell (1979) point to several sources of internal invalidity. Briefly, this experiment's use of random selection, a standardized experimental setting, and use of a control group minimizes the majority of these threats. There is deserved concern about the failure of the relevance manipulation check for the jobs and recession issue. While in the direction expected, the maniuplation was not strong enough to achieve statistical significance. This may be due to some "ceiling effect" of the scale on

PAGE 111

104 the highly relevant issue, or because none of the subjects believed the issue to be of little personalrelevance. Even if an experiment "works," is it an accurate gauge of the "real world?" The greatest weakness of a laboratory experiment is its artificiality. This experiment consisted of groups of approximately one hundred students at a time (hardly generalizable to the population as a whole) sitting in a classroom reading two bogus news articles. In defense, these articles were created to appear genuine. Discussion with subjects uncovered no suspicion about the articles and their cover story. Still, there remains the possibility of an interaction of testing with the stimulus as a threat to external validity. That is, subjects read the news articles in a situation unlike typical newspaper reading, making it possible some interaction existed between the situation and how they processed the information. In many ways, this subject pool could be looked on as largely homogeneous, with little variation in age, socio economic status, and educational background. But the selection of college students as subjects, while taking away from the generalizability of the results, adds to the effective control of other sociological variables that may have interfered with the experimental manipulation.

PAGE 112

105 conclusions If bandwagon research could be simply summed up, it would be that public opinion polls influence peoplesometimes. A large number of studies find no influence, others find some. And such influence is not necessarily in the poll's direction (the bandwagon effect), but can also be in the opposite direction (the underdog effect). A closer look at bandwagon and underdog studies reveals a potential explanation through the relevance of the political campaign or issue studied. Many involve presidential elections or controversial issues where it is hardly surprising that little poll influence is found. And studies looking for an underdog effect have offered little explanation for the phenomenon. Guided by the work of persuasion theorists such as Petty and Cacioppo (1986a), polls on issues differing in relevance were found to influence the high rather than low relevance issue. When looking at within-issue differences of relevance, however, evidence points to more influence in situations of low relevance. Differences were also found between a poll supporting the majority of information presented in the news article and a poll not supportive of the information. Two questions arise--first, the differing effects of congruent and incongruent polls, and second, whether the

PAGE 113

106 between-issue and within-issue differences is one of methodological or theoretical concern. No hypotheses were generated as to poll support. It is possible that a supportive poll would promote attention to the minority viewpoint, and hence more mental effort, leading to less persuasion than might be seen by a supporting poll. There was less persuasion by the contradictory poll as compared to the supporting poll on the jobs issue, but no difference on the capital gains tax. In speculating about the underlying process, it seems likely that if a contradictory poll leads to more consideration of the issues, there should be more effort reported by subjects. No such difference was found. It also seems likely that certainty-oriented persons, facing news reports of experts arguing for continued difficulty in the job hunt while a poll reports most people disagree, had early closure of consideration of the jobs issue. This could explain the lack of support for the relevance and uncertainty orientation hypotheses found on the jobs issue. Indeed, the jobs issue manipulation check on poll influence, when broken down by uncertainty orientation, shows the contradictory poll demonstrating less influence on certainty-oriented persons as compared to uncertainty-oriented persons (f(2,176) = 2.6, p<.08),

PAGE 114

107 though it should be stressed that this finding too is not at the established .05 level of significance. Persuasion research often manipulates relevance within an issue, avoiding the problems that arise when using issues that may differ not only in relevance but in other ways as well. Such a manipulation takes something away from the generalizability of a result, since it is rare that anyone is available to tell a reader that this issue is personally important, while this one is not. This study does nothing to settle this methodological question. In fact, the very different findings when looking at between-issue relevance and within-issue relevance only adds to the problem. It may be that, when possible, research should use different issues as this research did and add a within-issue relevance manipulation. Implications for Bandwagon Research Research into the bandwagon and underdog effects should take into account the relevance a political campaign or public issue has with those individuals being studied. While there has been some hint of this from public opinion and mass communication scholars, few are clear in their conceptualization or measurement of relevance.

PAGE 115

108 In addition, other factors of the audience must be considered in order for reasonable theory-building. Lavrakas, et al. (1991) recently conducted one test of political predispositions and the influence of polls. In their investigation of the 1988 presidential election, the authors provided a manipulation in their survey informing repondents of poll results in favor of George Bush or Michael Dukakis. They found women, liberals, the lower income, and the unemployed to be likely to move in the underdog direction. Young adults, those not identifying with the Democratic Party, and those employed were more likely to move in the bandwagon direction. Obviously, the poll brings out latent predispositions of certain people. Those with a predisposition against Bush, told of a poll in his favor, shift in the underdog direction. Persons making their minds up late in the campaign were more likely to make up either the bandwagon or underdog movement. Differences in personality variables must also be considered. This study used uncertainty orientation with limited success to explain why the underdog and bandwagon effects occur. How useful is uncertainty orientation to the study of public opinion poll influence? The results here, while not overwhelming, are not so bleak as to lead us to toss out the construct.

PAGE 116

109 A key hypothesis of this study was that in situations of high relevance, where one would expect more careful consideration of the issue and less poll influence, certainty-oriented people would be more persuaded by a poll. The hypothesis did not receive strong support. It is interesting to note, however, that uncertainty-oriented certainty-oriented persons only in the high relevance condition (~(1,177) = 4.1, p<.05). This suggests that when relevance is high, certainty-oriented are less likely to undertake careful consideration of a message. No political or mass communication studies using uncertainty orientation have been published. This may be due to the original projective measure of need for uncertainty or simply a lack of interest in the construct. There seem to be a number of mass communication possibilities for uncertainty orientation. For example, it is possible that a narrative news style, thought to be more involving for the reader than the traditional inverted pyramid presentation, may actually result in early closure by certainty-oriented persons. Using such an approach to persuade readers to undertake some change in lifestyle, then, may not be as successful as once believed. This construct, or similar ones such as open

PAGE 117

110 mindedness, should be pursued further before they are set aside. Future Research More work is needed to tie together the various findings from research in persuasion, poll influence, and the bandwagon and underdog effects. One interesting possibility is combining the theoretical concepts of cognitive response analysis with work on elections by Lavrakas, et al. (1991), who showed the importance of political predispositions in bandwagon and underdog movement. The authors showed that a poll can "excite" or prime predispositions, causing movement in a predisposed direction among persons with no stated opinion. The poll used by the authors was only in one direction, and they do not address the matter of relevance. It may be that these persons who moved either with or against the poll saw less relevance in the presidential election and were therefore more open to poll influence. Or, like Mackie (1987), the poll caused more careful consideration of the issues, leading to persuasion. These possibilities must be investigated further, with more attempts to uncover the process. In particular, do persons with little interest in a political contest and who receive a poll showing the candidate they are predisposed toward is losing a race act

PAGE 118

111 in the same way as persons who care a great deal or are very interested in the same political contest? Lavrakas, et al. (1991) suggest political predispositions explain both the underdog and bandwagon effects. Personality differences are not considered. Returning to uncertainty orientation, it may be that certainty-oriented persons would act differently than uncertainty-oriented persons facing the losing proposition above. It would be interesting to tease out the relative strength of political, demographic, and personality differences when exposed to one-sided polls. A large scale discriminant analyisis strategy would be helpful in probing differences among those who show a shift in opinion from before and after exposure to a poll. Other constructs such as open-mindedness may have more potential for explaining political persuasion. Moving from candidates to issues, one must consider the possible impact of agenda-setting. Little work has been done merging persuasion and agenda-setting, which at first seem to be at odds with each other. This need not be the case. Issues new on the "agenda" of the public (and individuals) may behave differently in a persuasive context than issues that have been before the public for some time. Also, extensive or intensive media coverage of an issue, pushing it higher on the "agenda," may dampen

PAGE 119

112 the ability of polls to influence the public by making that issue more relevant. It may also be argued that such coverage would increase a poll's influence since the issue has become more important, thus drawing the attention of the reader or viewer, thus allowing the poll to spark opinion formation or change along the lines of existing political predispositions. These competing hypotheses deserve further attention. Finally, two other research possibilities present themselves. The first has to do with expectations, the second with how poll information is presented. Poll results flagarantly beyond the expectations of an undecided person who tends to favor the minority seems more likely to cause that person to move with the underdog than a poll result closer to what that person would expect. such a presentation to someone in the majority, but undecided, could possibly cause an underdog shift depending on such characteristics as personality. Moving to presentation of information, it seems likely that graphics such as pie charts, or detailed information about survey procedures, would influence the persuasiveness of a poll. For example, can a lopsided pie chart presentation lead to more of a bandwagon or underdog effect? All of these notions deserve attention.

PAGE 120

APPENDIX A NEWS ARTICLE EXAMPLES

PAGE 121

New push to reduce capital gains tax expected in 1991 A renewed effort to reduce the capital gains tax is expected later this year w11h supporters arguing such cuts would increase investments and spur a laggmg economy. Experts predict the final decision will affect college students greatly. Unuer the plan. most capital assets would be eligible for tax breaks. While supporters argue this wmdfall would prompt investment and economic growth, others say the proposal actually provides for little encouragement for such action and would benefit only the wealthy. A recent nationwide public opinion poll showed 81 percent of Americans against the reduction of the capital gains tax. with 15 percent for and 4 percent undecided. The tax reduction would create jobs and growth." said one economist. ''This is a way for the government to get the economy back on track." The U.S. tax code accorded capital gains preferential treatment until about Polls show most Americans are against a reduction in the capital gains tax. five years ago, when the 1986 Tax Refonn Act created two basic tax rates of 15 percent and 28 percent for income and capital gains alike. As much as a $12.5 billion tax windfall for the U.S. Treasury could be seen from the reduction over the next five years. some estimate. However, tax economists disagree over the affects of a capllal gains tax reduction. Supporters claim a lower tax would promote economic efficiencv bv encouragmg investors to cash in old gains on assets that are no longer productive and plow the proceeds back into more promising markets. See Tax on paf?e 4.4 114

PAGE 122

115 New push to reduce capital gains tax expected in 1991 A renewed effort to reduce the capital gains tax is expected later this year with support e rs argurng such cuts would increase in v estments and spur a lagging economy . Experts predict til e final decision will not aff e ct rnllel!e ~tudents . Und e r the plan . nws1 capital assels would be e ligible t o r tax breaks . While s upporters argue tllis windfall would prompt invcs1mem and economic growth , others say tile proposal actually provides for little e ncouragement for such action and would benefit only the wealthy. A recent nationwid e public opinion poll showed 8 1 percent of Americans against the reduction of the capital gains tax, with 15 percent for and percent undecided . 'The tax reduction w ould create jobs and growth, " s aid one economist. "This is a way for the government lo get the economy back on track.' ' The U.S. tax -:ode accorded capital gains preferential treatment until about Polls show most Americans are against a reduction in capital gains tax. tive year.; ago . when th e 1986 Tax Reform Act created two basi c tax rates of 1 5 percent and 28 percem for income and capital gains alike. As much as a $12.5 billion tax windfall for the U . S. Treasury could be seen from the reduction over the next five years. s ome estimate . Howe v er , tax economists disagree over the affe c ts oi a capital gains tax reduction. Supporters claim a lower tax would promote l'.conom1c efficiency by encouraging investors to cash in old gains on assets that are no longer productive and plow the proceed~ back into more promising markets. See Tax on page 4 r \

PAGE 123

116 New push to reduce capital gains tax expected in 1991 A renewed effort to reduce the capital gains tax is expected later this year with supporters arguing s uch cuts would increase investments and spur a lagging economv. Experts predict the final Je c ision will affect college s1ude1115 ~rea1Jv. Cnder the plan. most L"apllal asset s would be eligible for tax breaks. While ~ upporters argue this w1ndfall would prompt investmelll and economic growth . o thers say the proposal actuallv provides for linle encourag e ment tvr s uch action and would benefit only the wealthy. " The tax reduction would create jobs and growth ," said on e economi s t. "This is a way for the government t o get the eco nomv back on track .'' The U.S . t;ix code acrnrd e d capital gains preferential treatment until about fi v e years ago, when the 1%6 Tax Refonn A ct creat e d two basic tax rates of 15 Supporters claim a lower tax would promote economic efficiency iJerccnt and 2 8 percent t,H income and ,apital gams alike. As much as a $12 . .S billion tax windfall tor the U.S. Treasury couid be seen from the reduction over the next fiv e years, some estimate. However. tax economists disagree over the affects of a capital gains tax reduction. Supporters claim a lower tax would promote economic e fficiency by encouraging inv es tors to c ash in o ld gains on assets that a re no longer productive and plow the proceeds back into more promising markets . See Tax on page 4A

PAGE 124

117 New push to reduce capital gains tax expected in 1991 . ..\ renewed effort to reduce the capttal gains tax 1s expe c ted l a t e r this ycar with ~ upponers arguing su c h cuts would increase investments and spur a lagging erunomv. Experts predict the final dension will not affect college students. U nder the. plan . mo s t capital a~sets w ould be eligible fo r tax breaks. Whik ~ upporters argue thi s w indfall w nuid prompt investment and n : onomic groWlh , o thers sav the proposal a ctually pro\ ides fo r little encouragem e nt for such action and would benefit only the wealthy . "The tax reduction would create jnbs and growth," said one e conomist. "This is a way for the government to get the cconom~ back on track ." The L . s . tax code accorded capital gains preferential treatment until about liw \'ears ago , when the 1 9 86 Tax Reform ...\ct created two ba s i c tax rates ( 1f 15 Supporters claim a lower tax would promote economic efficiency. percent aud 28 .pe.rccnt for income and c apital gains alike . ...\s much as a $12 . 5 billion tax w indfall fo r the t. : .s. Treasury l. ould be s cen from the reduct i on over the next iive v e ars , s ome e s llmate . However, tax economists disagree ( 1\' er the affects of a capital gains tax reducti o n . Supporters claim a lower tax would promot e economic efficiency by encouraging investors to cash in old gains n n asset s r bat are no longer productiv e and plow 1he proceeds back into more promisin g markets. See Tax on page -IA

PAGE 125

118 New push to reduce capital gains tax expected in 1991 A renewed effort to reduce the capital gains tax is expected later this year with supporters argumg such cuts would increase investments and spur a lagging economy. Experts predict the final decision will affect collei;e students g reatly . L' nder the plan. most capi1al asse ts wo uld be digible for tax breaks. \\'hilt supporters argue this wrndfall would promp1 rnvestme111 and t'conomic growlh. others say the proposal ac tuallv provides for little encourageml'nt tor such action and would benefit only the wealthy. A recent nationwide public opinion poll show.:d 81 percent of Americans support the reduct1011 uf tbc cap11al gams tax, \\'ith 15 pnn:nt ; 1gainst and perc.:nt undecided. "The t;ix reduction would L'rtatt' johs and grO\vth." said nne economist. "This is a wav for the government to get the cconomv hack on track. ' The L .. S. tax code atcorded capttal gains preferential treatment until about Polls show most Americans are for a reduction in the capital gains tax. tiw war.; ilt! < >. when the 1986 Tax Reform .-\cl ueated two basic tax rates of 15 percent and 28 percent for income and capital ga11ts alike. As much as a $12.5 billion tax \vindfall for the L'.S. Treasury could be seen from the reduction over the next five years, some estimate. However, tax economists Jisagree ,1 \ er the affects of a capital gains tax redur11on. Suppnrtns daim a lown tax would promot t ' t,onom1c dlic1enc v b v encouraging investors to cash in old gains on assets that are no longer productive and plow the proceeds back inlo more promisiru; markets. Si:e Tax on page .. u

PAGE 126

119 New push to reduce capital gains tax expected in 1991 A renewed effort to reduce the capital gains tax is expected later this vear with supporters arguing such cuts would increase investments and spur a lagging economv. Experts predict the final dec1s1on will not affect college students. Under the plan. most capita I asset s would be d1gible for tax breah s. \Vhile supporters argue this windfall \\ouil.1 prompt rnvestment am.t ccono1111c ~rowth, others say the proposal actually provides for little encouragement for such action and would benefit only the wealthy. A recent nationwide public opinion poll showed 81 percent of Americans support the reduction of the capital gains tax, witl.J 15 percent against and 4 percent undecided . " The tax reduction would create jobs and growth," s aid one economist. .. This is a way for the government to ge t the economy back 011 track. " The U.S. tax code accorded cJpital gains !)referential treatment unlll about Polls show most Americans are for a reduction in the capital gains tax. tive ye ars ago. when the ~9~6 Tax Refonn Act created two baste tax rates of 15 percent and 28 percent for 111come and ~ap1tal gains alike . As much as a $12.5 billion tax windfall for the U.S. Treasury could be seen t'rom the reduction over the next five years, some estimate. However, tax economists disagree over the affects oi a capital gains tax reduction. Supporters claim a lov.:rr tax would promote economic dficiencv bv encouraging investors to cash in old gains nn assets that are no longer productive and plow tbc proceeds baci,; into more promising markets. See Tax on page 4A

PAGE 127

120 Recession may mean fewer jobs for college graduates The contmuing U.S. economic slump has some experts predicting Jobs will be hard to come by for some l'Oliege ,;raduatcs over the next tew vears. Analysts predict most college srudents -.viii not be affected. The economv is grow11w at Its siowest pace and L.S. industry b opnat,n.r Jl its il)West capacity 111 eight \ears. Jnd as many as 8.5 million could be unemploved and looking for work. A recent nauonwide public op111ion poll showed 82 percent of A.mencans believe jobs will be hard to come by for college students graduating in the next few years, with 14 percent disagreeing and percent undecided. "The economy is in a slump and that will intluence hiring decisions.• one analyst said. "These effects could be seen for years." \1ost economists agree that the record borrowing birn;e of the 1 YSOs has idt the l'.S. economvs private sector unprepared tor tough times. Polls show most Americans think jobs won't be plentiful during the recession. This means few businesses can expanu and open space for new employees. Some industries have found themselves financially unable to fill positions opened by retirement or employees taking ether positions elsewhere. If the economy grows more slowly than the Administration predicts, or should it actually shnnk, the budget deficit wJII balloon because of declming tax revenues. This in turn influences the hiring market. Many college students are considenng an extra year of education 111 order to wan out" the economic hard times. Administration officials. howe\'er. ma111tain the job market \VIII be ~trong rnough to provide jobs. See Jobs on par;e •U

PAGE 128

121 Recession may mean fewer jobs for college graduates The continuing U.S. economic slump has some experts predicting jobs will be bard to come by tor some college graduates over the nexr few years. Analvsts predict most college students will be grea tlv affected. The cconomv 1s growing at its slowest pace and L '. .S. industrv 1 s operat111g at its lowest capac11y in CJ ,! lll vears . and as many as 8.5 million could be unemployed and looking for work. A recent nallonw1de DUblic op111ion poll showed 82 percent ,ii . .\rnencans believe jobs will be bard to L'l>me by for college students graduating in the next few vears. with 14 percent disagreeing and ..t percent undecided. " The economv is in a slump and that will influence hiring detisions . " ,,ne analyst said . "These effects ,ould be seen for years .. , \1ost economists agree that the record borrowing binge of the 1980s bas left the l'.S. economys private sector unprepared for tough times. Polls show most Americans think jobs won't be plentiful during the recession. This means lc'W hus111esst'S can expand and llpc:n ~ p a ce t c lr new employees. Some industries t1ave found themselves linancially unable to fill positions opened hy retirement nr employees taking other positions elsewhere. If the economy grows more slowly than the Admin1strat1on predicts, or should it actually shrink. the budget deficit wiil balloon because of declining tax revenues . This in rum influences the hiring market. :\1any college students are considering an extra year o t education in order to .. wan out" the economic bard times. Administration officials. however, maintain tile Job market will be strong enough to pro\'lde Jobs . . ","ee Jobs on page 4A

PAGE 129

122 Recession may mean fewer jobs for college graduates The continuing U.S . economic slump bas some experts predicting Jobs will be hard to come by for some college graduates over tbe next few years. Analysts predict most college students will be greatly affected. The economy is growing at its slowest pace and U.S. industry is opcra1111g al its lowest capacity in eight years. and as many as 8 . 5 milliun could be unemployed and looking for work. A recent nationwide public opinion poll showed 82 percent of Americans believe jobs will not be hard to come by for college students graduatmg in the next few years, with 14 percent disagreeing and 4 percent undecided . 'The economv is in a slump and Iha! will influence hiring decisions," one analyst said. " These effects could be seen for vears . " Most economists agree that the record borrowing binge of the l l/80s has left the U.S. economv s private sector unprepared for tough times. Polls show most Americans believe jobs will still be plentiful despite recession. This means few businesses can expand and open space for new empioyees. Some industries have found themselves financially unable to fill positions opened by retirement or employees taking other positions elsewhere. If the economy grows more slowly than the Administration predicts . or should it actually shrink . the budget deficit will balloon because of declining tax revenues . This in tum 11ttluences the hiring market. Many college students arc c01L-;idering an extra year ol education in order to wait out" the economic hard timi:s . Administration officials. lrnwever. mamtain the JOb market will be strong enough to provide Jobs. See Jobs on page 4A

PAGE 130

123 Recession may mean fewer jobs for college graduates The continuing U.S. economic slump bas some experts predicting Jobs will be bard to come by for some college graduates over the next few years. Analvsts predict most college students will be greatly a llected. Tbe c:conomy is growing at its slowest pace and U.S. rndustry 1s operat111g at its l,1west Lapac1tv in eight years. ;ind JS many as ~-5 million could be unemployed and looking for work. "The economy is in a slump and that will inilucnce hiring dec1s1ons, ., ,rnc analyst said. "These effects could be seen for years." .\1ost economists agree that the record borrowmg binge of the 1980s has left the L'..S. economy's private sector unprepared tor tough times. This means tew businesses can expand and open space for new employees. Some industries have found them~dves Businesses may have fewer job openings as a result of the recession. !inannally unable to fill pos1110115 opened bv retirement or employees taking other poslliot15 elsewhere. [f the economy grows more slowly than the Administration predicts, or should it actually shnnk, the budget deficit will balloon because of declining tax revenues. This in tum iniluences the hiring market. Many college students arc considering an extra year of education in order to "wan out" the economic bard times. Administration officials, however. maintain tbe job market will be strong enough to provide jobs. See Jobs on page .JA

PAGE 131

124 Recession may mean fewer jobs for college graduates The continuing U.S. economic slump has some experts predicting jobs will he hard to come by for some college graduates over the next few years. Analysts predict most college student5 will not be affected. The economy 1s grow111g at its siowts1 pace and ll.S. indusrrv is operat111g at 1b lowest capacitv in eight ytars. and as many as 8.5 nullion could be unemployed and looking for work. A recent nauonw1de public opinion poll showed 82 percent ol Amencans believe jubs will not be ha rd to come by for college students graduating in the next few years, with 14 percenl disagreerng and 4 percent undecided. "'The economy is in a slump and Iha! wlll intlutnce hiring decisions." one analyst said. "These effects could he seen for vears." Most economists al.!ree that the record borrowmg binge of rbe 1980s has left the L:.S. economy's private sector unprepared for tough tunes. Polls show most Americans believe jobs will still be plentiful despite recession. This means tew businesses can expano and open space for new emplovees. Somt: 111dus1rics have found themseh-es financially unable to fill positions opened by ret1rcment or employees taking other positions elsewhere. If the economy grows more slowly than the Administration predicts, or should ii actually shnnk. the budget deficit \\ ill balloon because of declimng tax revenues. This in tum influences the hiring market. Many college students arc co1151denng an extra year of cducat10n in order to "'\va11 out" the economic hard times. Administration officials, however. maintam the job market will be strong enough to provide jobs. See Jobs on page -1,l

PAGE 132

125 Recession may mean fewer jobs for college graduates The continuing l'.S. economic slump bas some experts predicting Jobs will be bard to come by tor some college graduates over the next few years. Analysts predict most college students wi II not be a ifected. The economv 1s growing at its slowest pace and U . S. industrv is ()perating at its lowest capacitv 1n e 1 ~ht vcars. :1 11d :is many as 8.5 million could he unemployed and looking for work. "The economy b in a s lump and that will influence hiring decisions." lrne analyst said. "These effects could be seen for years." Most economists agree that the rl'cord borrowing binge of th, 1980s has ldt the U . S . economy''.> pri \ atc sector unprepared for tough times . This means kw b us1n..:~ses , :i n l".\panJ and open space for new cmplovees. Soml'. industries have t,,unJ thcmselvl's Businesses may have fewer job openings as a result of the recession. tinanciallv unable :o (ill positions opened by n ttrement or ..:moloyees taking other positions elsev,:here. If the economy grows more slowly than the Administration predicts. or should it actually shrink. the budget deficit will balloon because of declining tax revenues. This in tum in.tluences the hiring market. \1any college students are considering an extra year of education in order to "wait out" the economic hard times . Administration ,ilficials. however. maintain the Job mHKet will be strong enough to provide Jobs. See Jobs on page -IA

PAGE 133

APPENDIX B MEANS AND STANDARD DEVIATIONS OF ALL UNCERTAINTY ORIENTATION QUESTIONS Question Mean 1 I would rather do something at which I feel confident and relaxed than something which is challenging and difficult. 2 I would prefer a job which is important, difficult, and involves a 50 percent chance of failure to a job which is something important but not difficult. 3 I more often attempt difficult tasks that I am not sure I can do than easier tasks I know I can do. 4 There is no reason to punish any crime with the death penalty. 5 I like to think about philosophical issues, even if there are no real answers. 6 I usually try to learn about something I don't understand, even if I might end up in trouble because of it. 7 Obedience and respect for authority are the most important virtues children should learn. 8 A person who has bad manners, habits, and breeding can hardly expect to get along with decent people. 9 I like to fool around with new ideas, even if they turn out later to be a total waste of time. 126 4.5 4.4 4.7 7.2 5.9 5.6 4.4 4.7 6.1 SD 2.4 2.2 2.1 2.8 2.8 2.2 2.5 2.4 2.0

PAGE 134

127 Question 10 It is possible that creatures on other planets have founded a better society than ours. 11 I enjoy spending time discovering new things. 12 It's all right for people to raise questions about even the most sacred matters. 13 When they are little, kids sometimes think about doing harm to one or both of their parents. 14 If I were able to return to one of two incomplete tasks, I would rather return to the difficult one than the easy one. 15 Some of the greatest atrocities in history have been committed in the name of religion and morality. 16 I believe it is important for us to challenge our beliefs. 17 Anyone who would interpret the Bible literally just doesn't know much about geology, biology, or history. 18 If I find a concept too confusing, I generally cease to be curious about it. 19 I often put myself in situations in which I could learn something new. 20 There is hardly anything lower than a person who does not feel a great love, gratitude, and respect for his or her parents. Mean SD 3.9 2.9 6.4 1.8 1.7 1.9 3.8 2.6 4.9 2.4 2.1 2.0 6.6 2.1 4.5 2.7 5.7 2.4 6.5 1.7 4.8 2.5

PAGE 135

128 Question 21 An insult to our honor should always be punished. 22 No sane, normal, decent person could ever think of hurting a close friend or relative. 23 The findings of science may some day show that many of our most cherished beliefs are wrong. 24 People ought to pay more attention to new ideas, even if they seem to go against the American way of life. 25 If I come across something I find puzzling, I spend a lot of time trying to understand it. 26 Books and movies ought not to deal so much with the unpleasant and seamy side of life; they ought to concentrate on themes that are entertaining or uplifting. 27 If people would talk less and work more everybody would be better off. 28 Insults to our honor are not always important enough to bother about. 29 I enjoy thinking about ideas that challenge my views of the world. 30 The prisoners in our corrective institutions, regardless of the nature of their crime, should be humanely treated. 31 In this scientific age the need for a religious belief is more important than ever before. Mean SD 3.4 2.0 4.8 2.7 3.1 2.6 2.6 1.9 5.8 1.7 3.6 2.5 3.6 2.4 3.5 2.1 6.1 1.9 3.5 2.4 4.8 2.5

PAGE 136

129 Question 32 I like to find out why things happen. 33 I dislike having to solve unanswered questions. 34 What the youth needs most is strict discipline, rugged determination, and the will to work and fight for family and country. 35 Young people sometimes get rebellious ideas, but as they grow up they ought to get over them and settle down. Mean SD 7.0 1.8 5.4 2.1 4.6 2.3 4.9 2.3

PAGE 137

APPENDIX C FACTOR LOADINGS OF ALL UNCERTAINTY ORIENTATION QUESTIONS Fl F2 F3 F4 FS F6 F7 F8 F9 FlO 1 .74 2 .69 3 .65 4 .60 -.34 .34 -.49 .34 5 .60 .38 .39 6 .58 .54 7 .57 .40 8 .42 -.36 9 -.27 10 .61 -.32 11 .50 .34 12 .40 .39 13 .65 14 .63 .36 15 .61 16 -.36 .55 .32 17 .32 .48 .48 -.35 18 .47 19 .83 -.32 .33 20 .43 21 .31 .59 22 .57 23 .so 24 .33 .28 25 .43 26 .37 27 -.81 28 .32 -.40 -.40 130

PAGE 138

29 30 31 32 33 34 35 Eigen Fl F2 F3 F4 -.43 -.30 .41 -.41 .31 .44 131 F5 F6 F7 F8. F9 Fl0 .66 -.33 .40 -.60 .52 .41 .65 .45 .59 Value 6.0 2.5 1.8 0.9 0.9 0.7 0.7 0.6 0.5 0.4 % Var. 17.3 7.2 5.3 2.7 2.6 2.1 2.0 1.8 1.5 1.3 Note. The numbers above correspond to questions from the Appendix B. Factor loadings are from a principal factors analysis with oblique rotation. Loadings less than .30 not listed unless the factor that item loads on is less than that amount. Total variance explained= 43.6%

PAGE 139

REFERENCES Adorno, T., Frenkel-Brunswik, E., Levinson, D. & Sanford, N. (1950). The authoritarian personality. New York: Harper. Allard, W. {1941). A test of propaganda values in public opinion surveys. Social Forces, 2..Q, 206-213. Allen, V. L. (1965). Situational factors in conformity. In L. Berkowitz (Ed.), Advances in experimental social psychology, (Vol 2., pp. 133-175). New York: Academic Press. Allport, F. D. {1937). Toward a science of public opinion. Public Opinion Quarterly. i, 7-23. Allport, F. D. {1940). Polls and the science of public opinion. Public Opinion Quarterly.~, 249-257. Allport, G. w. (1935). Attitudes. Inc. Murchison (Ed.), Handbook of social psychology. Worcester, MA: Clark University Press. Aronson, E., Turner, J. A., & Carlsmith, M. (1963). Communicator credibility and communication discrepancy as determinants of attitude change. Journal of Abnormal and Social Psychology, 67, 31-36. Asch, S. E. (1951). Effects of group pressure upon the modification and distortion of judgments. In H. Geurtzkow (Ed.), Groups, leadership and men (Reprinted in 1963). Pittsburgh, PA: Carnegie Press. Atkin, c. K. (1969). The impact of political poll reports on candidate and issue preferences. Journalism Quarterly, 46, 515-521. Atkinson, J. W. (Ed.). (1958). Motives in fantasy, action, and society. Princeton, NJ: Van Nostrand. Atkinson, J. w., & Feather, N. T. (Eds.). (1966). A theory of achievement motivation. New York: Wiley. 132

PAGE 140

133 Atkinson, J. W., & Raynor, J. o. (1974). Motivation and achievement. Washington, DC: Winston. Axsom, D., Yates, s., & Chaiken, s. (1987). Audience response as a heuristic cue in persuasion. Journal of Personality and Social Psychology. -21., 30-40. Bargh, J. A. (1984). Automatic and conscious processing of social information. In R.S. Wyer & T. K. srull (Eds.), Handbook of social cognition. (pp. 1-43). Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers. Beniger, J. R. (1976). Winning the presidential nomination: National polls and state primary elections, 1936-1972. Public Opinion Quarterly. 40, 22-38. Berelson, B., Lazarsfeld, P.F., & McPhee, w. (1954). Voting. Chicago: University of Chicago Press. Block, J., & Block, J. (1951). An investigation of the relationship between intolerance of ambiguity and ethnocentrism. Journal of Personality. 1..2., 303-311. Bogart, L. (1972). Silent politics: Polls and the awareness of public opinion. New York: Wiley. Box, G. E. P. (1953). Non-normality and tests on variances. Biometrika, 40, 318-335. Broh, C. A. (1980). Horse race journalism: Reporting the polls in the 1976 presidential election. Public Opinion Quarterly, 44, 515-129. Byrne, D. (1964). Repression-sensitazation as a dimension of personality. In. B. A. Maher (Ed.), Progress in experimental personality research. New York: Academic Press. Byrne, D., & Lamberth, J. (1971). The effect of erotic stimuli on sex arousal, evaluative responses, and subsequent behavior. In Technical reports of the commission on Obscenity and Pornography (Vol. 8, pp. 4168). Washington, DC: U.S. Government Printing Office. Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, il, 116-131.

PAGE 141

134 Cacioppo, J. T., Petty, R. E., & Morris, K. J. (1983). Effects of need for cognition on message evaluation, recall, and persuasion. Journal of Personality and Social Psychology, 45, 805-818. Campbell, D. T. experimenting International (1958). on the possibility of with the "bandwagon" effect. Journal of Opinion and Attitude Research, 2, 251-260. Campbell, D. T., & Stanley, J. (1963). Experimental and quasi-experimental designs for research. Chicago: Rand McNally. Cantril A.H. (1980). Polling on the issues. Cabin John, MD: Seven Locks Press. Carmines, E.G., & Zeller, R. A. (1979). Reliability and validity assessment. Beverly Hills, CA: Sage. Cattell, R. B. (1943). The description of personality: Basic traits resolved into clusters. Journal of Abnormal and social Psychology. J.a, 476-506. Ceci, s. J., & Kain, bandwagon with the polling behavior. 242. E. L. (1982). Jumping on the underdog: The impact of polling on Public Opinion Quarterly. 46, 228Chaiken, S. (1980). Herustic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752-766. Chaiken, s. (1986). Physical appearance variables and social influence. Inc. P. Herman, E.T. Higgins, & M. P. Zanna (Eds.), Appearance and social behavior: The Ontario Symposium, (Vol. 3, pp. 143-177). Hillsdale, NJ: Lawrence Erlbaum Associates. Chaiken, S. (1987). The heuristic model of persuasion. In M. P. Zanna, J.M. Olson, & c. P. Herman (Eds.), Social influence: The Ontario Symposium (Vol. 5, pp. 339). Hillsdale, NJ: Lawrence Erlbaum Associates. Chaiken, s., & Eagly, A.H. (1983). Communication modality as a determinant of persuasion: The role of communicator salience. Journal of Personality and Social Psychology, 45, 241-256.

PAGE 142

135 Chaiken, S., Liberman, A., & Eagly, A.H. (1989). Heuristic and systematic information processing within and beyond the persuasion context. In (J~ s. Uleman & J. A. Bargh (Eds.), unintended thought: Limits of awareness. intention, and control (pp. 212-252). New York: Guilford. Childs, H. (1939). By public opinion I mean .... Public opinion ouarterly. ~, 53-69. Christie, R., Havel, J., & Seidenberg, B. (1958). Is the F scale irreversible? Journal of Abnormal social Psychology. 22, 143-159. Cloutier, E., Nadeau, R., & Guay, J. (1989). Bandwagoning and underdogging on North-American free trade: A quasi-experimental panel study of opinion movement. International Journal of Public opinion Research, i, 206-220. Coan, R. w. (1974). The optimal personality. New York: Columbia University Press. Cohen, J., & Cohen, P. (1975). Applied multiple regression/correlation analysis for the behaviorial sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Cook, s. W. and Welch, A. c. (1940). Methods of measuring the practical effect of polls on public opinion. Journal of Applied Psychology. 24, 441-454. Cook, T. D., & Campbell, D. T. (1979). Quasi Experimentation: Design and Analysis Issues for Field Settings. Chicago: Rand McNally. Cooley, c. H. (1909). Social organization: A study of the larger mind. New York: Charles Scribner's Sons. Costa, P. T., & Mccrae, R. R. (1985). The NEO personality inventory. Odessa, FL: Psychological Assessment Resources. Couch, A., & naysayers: variable. 151-174. Keniston, K. (1960). Yeasayers and Agreeing response set as a personality Journal of Abnormal Social Psychology, 60,

PAGE 143

136 DeBock, H. (1972). Influence of in-state election poll reports on candidate preference. Journalism Quarterly, 21, 457-462. Digman, J.M. (1990). Personality structure: Emergence of the five-factor model. Annual Review of Psychology, 41, 417-440. Dizney, H. F., & Roskens, R. w. (1962). An investigation of the bandwagon effects in a college straw election. Journal of Educational sociology, .ll, 108-114. Doms, M. (1983). The minority influence effect: An alternative approach. In w. Doise ands. Moscovici (Eds.), current issues in European social psychology (Vol. 1, pp 1-32). Cambridge, UK: Cambridge University Press. Festinger, L. (1950). Informal social communication. Psychological Review, 57, 271-282. Fields, J.M., & Schuman, H. (1976). Public beliefs and beliefs of the public. Public Opinion Quarterly. 40, 427-448. Fiske, D. w. {1949). Consistency of the factorial structures of personality ratings from different sources. Journal of Abnormal Social Psychology. 44, 329-344. Fiske, s. T., & Taylor, s. E. (1984). Social cognition. New York: Random House. Fleitas, D. W. (1971). Bandwagon and underdog effects in minimal information elections. American Political Science Review, 65, 434-438 Frederick, J.E., & Sorrentino, R. M. (1977). A scoring manual for the motive to master uncertainty (Research Bulletin No. 410). London, Ontario, Canada: University of Western Ontario. Frenkel-Brunswik, E. (1949). Intolerance of ambiguity as an emotional and perceptual personality variable. Journal of Personality, ll, 108-143. Gallup, G. (1972). The sophisticated poll watcher's guide. Princeton, NJ: Princeton Opinion Press.

PAGE 144

137 Gallup, G., & Rae, s. F. (1940). Is there a bandwagon vote? Public opinion Quarterly,~, 244-249. Gartner, M. (1976). Endogenous bandwagon and underdog effects. Public Choice, il, 83-139. Gaskell, G. !, 23-24. (1974). Polls and the voters. New Society. Gollin, A. E. (1980). Exploring the liaison between polling and the press. Public opinion Quarterly, 44, 445-461. Granberg, D. (1972). Authoritarianism and the assumption of similarity to self. Journal of Experimental Research in Personality,, 1-4. Hastings, E. H., & Hastings, P. K. (1990). Index to international public opinion. New York: Greenwood Press. Henshel, R. L., & Johnston, w. bandwagon effects: A theory. il, 493-511. (1987). The emergence of Sociological Quarterly, Hickman, H. (1991). Public polls and election participants. In P. J. Lavrakas & J. K. Holley (Eds.), Polling and presidential election coverage, (pp. 100153). Newbury Park: Sage Publications. Higgins, E.T., & King, G. (1981). Accessibility of social constructs: Information processing consequences of individual and contextual variability. In N. Cantor and J. F. Kihlstrom (Eds.), Personality. cognition. and social interaction, (pp. 69-122). Hillsdale, NJ: Lawrence Erlbaum Associates. Higgins, J. (1965). Authoritarianism and candidate preference. Psychological Reports, 16, 603-604. Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and persuasion. New Haven, CN: Yale University Press. Hovland, C. I., Lumsdaine, A. A., & Sheffield, F. D. (1949). Experiments on mass communication. Princeton, NJ: Princeton University Press.

PAGE 145

138 Jones, E. E. (1954). Authoritarianism as a determinant of first-impression formation. Journal of Personality, .u, 107-127. Kagan, J. (1972). Motives and development. Journal of Personality and social Psychology. il, 51-66. Kaplowitz, s. A., Fink, E. L., D'Alessio, o., & Armstrong, G. B. (1983). Anonymity, strength of attitude, and the influence of public opinion polls. Human Communication Research, .l.Q, 5-25. Katz, D. (1972). Attitude formation and public opinion. In D. Nimmo & c. Bonjean (Eds.), Political attitudes and public opinion. (pp. 47-62). New York: McKay. Kavanagh, D. (1981). Do opinion polls influence elections? Parliamentarian. il, 199-203. Kelman, K. c., & Barclay, J. (1963). measure of breadth and perspective. and social Psychology, il, 608-615. The F scale as a Journal of Abnormal Keppel, G. (1982). Design and analysis: A researcher's handbook (2nd ed.). Englewood Cliffs, NJ: Prentice Hall. Kiesler, c. A., & Kiesler, s. B. (1969). Conformity. Reading, MA: Addison-Wesley. Kinder, D.R., & Sears, D. o. (1985). Public opinion and political action. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., Vol. 2, pp. 659742). New York: Random House. King, G. A. & Sorrentino, R. M. (1988). Uncertainty orientation and the relation between individual accessible constructs and person memory. Social Cognition,~, 128-149. Kirscht, J., & Dillehay, R. (1967). Dimensions of authoritarianism. Lexington: University of Kentucky Press. Knutson, J. N. (1973). Handbook of political psychology. San Francisco: Jossey-Bass Publishers. Kohut, A. (1986). Rating the polls: The views of media elites and the general public. Public Opinion Quarterly. 50, 1-9.

PAGE 146

139 Lang, K., & Lang, G. E. (1984). The impact of polls on public opinion. The Annals of the American Academy of Political and social science. m, 129-14~. Laponce, J. A. (1966). An experimental method to measure the tendency to equibalance in a political system. American Political Science Review. 60, 982-993. Lasswell, H. o. (1930). Psychopathology and politics. Chicago: University of Chicago Press. Lasswell, H. D. (1948a). The structure and function of communication in society. In L. Byron (Ed.), communication of ideas, (pp. 37-52). New York: Harper. Lasswell, H. o. (1948b). Power and personality. New York: w. w. Norton. Lasswell, H. D. (1951). Political writings of Harold Lasswell. Glencoe, IL: Free Press. Latane, B. (1981). The psychology of social impact. American Psychologist, .J.2, 343-356. Latane, B., & Wolf, S. (1981). The social impact of majorities and minorities. Psychological Review, .fill, 438-453.Lewin, K. (1947). Group decision and social change. In T. M. Newcomb & E. L. Hartley (Eds.), Readings in social psychology (pp. 330-344). New York: Holt, Rinehart, & Winston. Lavrakas, P. J., Holley, J. K., & Miller, P. V. (1991). Public reactions to polling news during the 1988 presidential election campaign. In P. J. Lavrakas & J. K. Holley (Eds.), Polling and presidential election coverage. Newbury Park, CA: Sage Publications. Lazarsfeld, P. F., Berelson, B., & Gaudet, H. (1948). The people's choice. New York: Columbia University Press. Lewis, G. F. (1940). The congressmen look at the polls. Public Opinion Quarterly,~, 229-231. Lippmann, W. (1913). Preface to politics. New York: Mitchell Kennerly.

PAGE 147

140 Mackie, D. M. (1987). Systematic and nonsystematic processing of majority and minority persuasive communications. Journal of Personality and social Psychology . .2.l, 41-52. Marks, G., & Miller, N. (1987). Ten years of research on the false-consensus effect: An empirical and theoretical review. Psychological Bulletin, .rn, 72-90. Marsh, c. (1983). Do polls affect what people think? In c. F. Turner and E. Martin (Eds.), survey measurement of subjective phenomena. Report of the panel on survey measurement of subjective phenomena, National Research Council, National Academy of Sciences, 2 vols. New York: Sage. Marsh, c. (1984). Back on the bandwagon: The effect of opinion polls on public opinion. British Journal of Political science, ll, 51-74. McBride, F. w. (1991). Media use of pre-election polls. In P. J. Lavrakas & J. K. Holley (Eds.), Polling and presidential election coverage, (184-199). Newbury Park: Sage Publications. Merkle, D. M. (1991). The effects of opinion poll results on public opinion; A review and synthesis of bandwagon and underdog research. Paper presented to the International Communication Association Convention, Chicago. Meyer, P. (1975). Precision journalism: A reporter's introduction to social science methods. Boston: Houghton Mifflin Co. Meyers, D. G., Wojcicki, s. B., & Aardema, B. s. (1977). Attitude comparison: Is there ever a bandwagon effect? Journal of Applied Social Psychology, 2, 341-347 Miller, M. M., & Hurd, R. (1982). Conformity to AAPOR standards in newspaper reporting of public opinion polls. Public Opinion ouarterlv, .il, 244,246. Milton, o. (1952). Presidential choice and performance on a scale of authoritarianism. American Psychologist, 2, 597-598.

PAGE 148

141 Moscovici, s. (1980). Toward a theory of conversion behavior. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 13, pp. 209-239). New York: Academic Press. Moscovici, s. & Personnaz, B. (1980). studies in social influence. V: Minority influence and compulsive behavior in a perceptual task. Journal of Experimental and Social Psychology, ll, 210-282. Navazio, R. (1977). bandwagon research. 225. An experimental approach to PUblic opinion Quarterly. il, 211Nemeth, c. J. (1987). Influence processes, problem solving and creativity. In M. P. Zanna, J.M. Olson, & c. P. Herman (Eds.), Social influence: The Ontario Symposium (Vol. 5, pp. 237-246). Hillsdale, NJ: Lawrence Erlbaum Associates. Nisbett, R., & Ross, L. (1980). Human inference: Strategies and shortcomings of social judgment. Englewood Cliffs, NJ: Prentice-Hall. Noelle-Neumann, E. (1974). theory of public opinion. 43-51. The spiral of silence: A Journal of communication, 24, Noelle-Neumann, E. (1977). Turbulences in the climate of opinion: Methodological applications of the spiral of silence theory. Public Opinion Quarterly. 41, 143-158. Normal, W. T. (1963). Toward an adequate taxonomy of personality attributes: Replicated factor structure in peer nomination personality ratings. Journal of Abnormal social Psychology. 66, 574-583. O'Gorman, H.J., with Garry, s. L. (1976). Pluralistic ignorance--a replication and extension. Public Opinion Quarterly, 40, 449-458. O'Keefe, D. J. (1990). Persuasion: Theory and research. Newbury Park, CA: Sage. Olson, J.M., & Zanna, M. P. (1979). A new look at selective exposure. Journal of Experimental Social Psychology, 15, 1-15.

PAGE 149

142 Paletz, D. L., Short, J. Y., Baker, H., Campbell, B. C., Cooper, R. J., & Oeslander, R. M. (1980). Polls in the media: Content, credibility, and consequences. Public opinion Quarterly, il, 495-513. Perry, J. L. (1968). Polling: Self-fulfilling prophecy and bandwagon effects. Public Opinion Quarterly, ll, 478. Petty, R. E., & Cacioppo, J. T. (1979a). Effects of forewarning of persu ::: .3i ve intent and involvement on cognitive responses and persuasion. Personality and Social Psychology Bulletin, 2, 173-176. Petty, R. E., & Cacioppo, J. T. (1979b). Issue involvement can increase or decrease persuasion by enhancing message-relevant cognitive responses. Journal of Personality and social Psychology. 37, 1915-1926. Petty, R. E., & Cacioppo, J. T. (1981). Attitudes and persuasion: Classic and contemporary approaches. Dubuque, IA: Wm. c. Brown Company. Petty, R. E., & Cacioppo, J. T. (1984). The effects of involvement on responses to argument quantity and quality: Central and peripheral routes to persuasion. Journal of Personality and Social Psychology. 46, 69-81. Petty, R. E., & Cacioppo, J. T. (1986a). Communication and persuasion: Central and peripheral routes to attitude change. New York: Springer-Verlag. Petty, R. E., & Cacioppo, J. T. (1986b). The elaboration likelihood model of persuasion. In L. M. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 123-205). New York: Academic Press. Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement as a determinant of argument-based persuasion. Journal of Personality and Social Psychology, 41, 847-855. Petty, R. E., Ostrom, T. M., & Brock, T. C. (Eds.) (1981). cognitive responses in persuasion. Hillsdale, NJ: Lawrence Erlbaum Associates. Pierce, W. M. (1940). Climbing on the bandwagon. Public Opinion Quarterly, i, 241-243.

PAGE 150

143 Poley, W. (1974). Dimensionality in the measurement of authoritarian and political attitudes. Canadian Journal of Behavioral science,~, 83-94. Price, v., (1989). Social identification and public opinion: Effects of communicating group conflict. Public opinion Quarterly. ll, 197-224. Price, V., & Roberts, D. F. (1988). Public opinion processes. In Berger & s. H. Chaffee (Eds.), Handbook of communication science. Beverly Hills, CA: sage. Rhine, R., & Severance, L. (1970). Ego-involvement, discrepancy, source credibility, and attitude change. Journal of Personality and social Psychology. 1.., 11s19o. Rippey, J. N. (1980). Use of polls as a reporting tool. Journalism Quarterly, 57, 642-646, 121. Robinson, c. E. (1937). Recent developments in the straw-poll field, Part II. Public Opinion Quarterly, i, 42-52. Rokeach, M. (1948). Generalized mental rigidity as a factor in ethnocentrism. Journal of Abnormal Social Psychology. il, 259-278. Rokeach, M. (1960). The open and closed mind: Investigations into the nature of belief systems and personality systems. New York: Basic Books. Roll, C. W., & Cantril, A.H. (1980). Polls: Their use and misuse and politics (2nd ed.). Cabin John, MD: Seven Locks Press. Roper, B. W. (1986). Evaluating the polls with poll data. Public Opinion Quarterly .2.Q., 10-16. Roshwalb, I. & Resnicoff, L. (1971). The impact of endorsements and published polls on the 1970 New York senatorial election. Public Opinion Quarterly. 35, 410414. Ross, L., Greene, D., & House, P. (1977). The "false consensus effect": An egocentric bias in social perception and attribution processes. Journal of Experimental social Psychology. 18, 279-301.

PAGE 151

144 Salwen, M. B. (1985). The reporting of public opinion polls during presidential election years, 1968-1984. Journalism Quarterly. ll, 212-211. Salwen, M. B. (1987). Credibility of newspaper opinion polls: Source, source intent and precision. Journalism Quarterly, il, 813-819. Sanders, L. W., Rollberg, J. N., & Buffalo, M. D. (1989, August). Down to the wire; How six newspapers reported public opinion polls during the 12sa presidential campaign. Paper presented to the Association for Education in Journalism and Mass Communication, Washington, DC. Sanford, N. (1971). The approach of "The authoritarian personality." In F. I. Greenstein & M. Lerner (Eds.), A source book for the study of personality and politics, (pp. 139-170). Chicago: Markham Publishing Company. Scodel, A., & Mussen, P. (1953). Social perceptions of authoritarians and nonauthoritarians. Journal of Abnormal social Psychology, ll, 101-104. Shils, E. A. (1954). Authoritarianism: "Right" and "left." In R. Christie & M. Johoder (Eds.), studies in the scope and method of "The Authoritarian Personality. (pp. 24-49). Glencoe, IL: Free Press. Simon, H. A. (1954). Bandwagon and underdog effects and the possibility of election predictions. Public Opinion Quarterly, 18, 245-253. Simons, H. w. (1966). Authoritarianism and social perceptiveness. Journal of Social Psycology, 68, 291297. Sorrentino, R. M. (1977). [Test-retest reliabilities: Authoritarianism scores]. Unpublished raw data. Sorrentino, R. M., & Hancock, R. D. (1987). Information and affective value: A case for the study of individual differences and social influence. In M. P. Zana & c. P. Herman (Eds.), Social influence: The Ontario Symposium (Vol. 5, pp. 247-268). Hillsdale, NJ: Lawrence Erlbaum Associates.

PAGE 152

145 Sorrentino, R. M., Bobocel, R., Gitta, M. Z., Olson, J. M., & Hewitt, E. c. {1988). Uncertainty orientation and persuasion: Individual differences in the effects of personal relevance on social judgments. Journal of Personality and social Psychology.~' 357-371. Sorrentino, R. M., & Hewitt, E. C. (1984). The uncertainty-reducing properties of achievement tasks revisited. Journal of Personality and social Psychology. il, 884-899. Sorrentino, R. M., & Higgins, E.T. (1986), Handbook of motivation and cognition. New York: Guilford Press. Sorrentino, R. M., & Roney, c. J. R. (1986). Uncertainty orientation, achievement-related motives, and task diagnosticity as determinants of task performance. Social Cognition,~, 420-436. Sorrentino, R. M., & Roney, c. J. R. (1990). Uncertainty orientation: Individual differences in the self inference process. In J.M. Olson & M. P. Zanna (Eds.), Self-inference processes; The Ontario Symposium. Vol. 6 (pp. 239-257). Hillsdale, NJ: Lawrence Erlbaum Associates. Sorrentino, R. M., & Short, J. c. (1986). Uncertainty orientation, motivation, and cognition. In R. M. Sorrentino & E.T. Higgins (Eds.), Handbook of motivation and cognition (pp. 379-403). New York: Guilford Press. Sorrentino, R. M., Short, J. c., & Raynor, J. o. (1984). Uncertainty orientation: Implications for affective and cognitive views of achievement behavior. Journal of Personality and Social Psychology. 46, 189-206. SPSS Inc. (1988). SPSS-X User's Guide (3rd ed.). Chicago: Author. Steiner, I. D., & Johnson, H. (1963). Authoritarianism and "tolerance of trait inconsistency." Journal of Abnormal Social Psychology. 67, 388-391. Stoler, P. (1986). The war against the press. New York: Dodd, Mead and Company. Stovall, J. G., & Solomon, J. H. (1984). The poll as a news event in the 1980 presidential campaign. Public Opinion Quarterly, 48, 615.

PAGE 153

146 straw ballots. (1936, November 13). The New York Times, p. 22. Straffin, P. o. (1977). The bandwagon curve. American Journal of Political science, ll, 695-709. Teer, F., polls. & Spence, J. o. (1973). Political opinion London: Hutchinson University Library. Trope, Y. (1975). Seeking information about one's own ability as a determinant of choice among tasks. Journal of Personality and social Psychology. ll, 1004-1013. Tyson, H. L., & Kaplowitz, s. A. (1977). Anonymity and conformity. Public Opinion ouarterly, 41, 226-234. Weiner, B. (1972). Theories of motivation: From mechanism to cognition. Chicago: Markham. Wolf, S. (1987). Majority and minority influence: A social impact analysis. In M. P. Zanna, J.M. Olson, & c. P. Herman (Eds.), Social influence: The Ontario svmposium (Vol. 5, pp. 207-235). Hillsdale, NJ: Lawrence Erlbaum Associates. Wrightsman, L. s. (1965). Attitudinal and personality correlates of presidential voting preferences. Paper presented at the meeting of the American Psychological Association, Chicago. Zech, C. E. (1975). applied to voting. Leibenstein's bandwagon effect as Public Choice, 21, 111-122.

PAGE 154

BIOGRAPHICAL SKETCH Barry Hollander was born in Lawrenceburg, Tennessee, and attended the University of North Alabama. He barely graduated with a bachelor of science degree in journalism in 1981. A news reporter for six years in Mississippi, Louisiana, and Florida, Hollander won a number of honors, including first-place awards for "best investigative reporting" from the Mississippi Press Association in 1981 and for "best spot news coverage" from the Louisiana Press Association in 1983. Hollander and his wife, Edith, moved to Florida when he took a position with the Sarasota Herald-Tribune. In 1987, they moved to Gainesville, Florida, where both attended graduate school. Hollander completed his master's degree in mass communication in the summer of 1988 and continued his studies toward a doctorate in mass communication. He will graduate in December of 1991. In August 1991 the couple will move to Athens, Georgia, where Hollander has accepted a position as an assistant professor of journalism at the College of Journalism and Mass Communication. 147

PAGE 155

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degr~e of Doctor of Philosophy. p , :> .!. ~""'' J :-, '-< i) >1 c "-Leonard P. Tipton, Chair Professor of Journalism and Communications I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the d1J: of DL 1 Philos~ M ry Ferguson, c Associate Professor of Journalism and Communications I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. \ ' i ;-,)<'! Yw~ _ ;,.,,.___ ? ' John Wright Professor of Journalism and Communications I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy~ / . / I . ,,, # ~-/ / ~ ,,,,~_ ~ Michael Weigold , Assistant Professor of Journalism and Communications

PAGE 156

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. I ' ' ' Michael Martinez Associate Professor of Political Science This dissertation was submitted to the Graduate Faculty of the College of Journalism and Communications and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. ~-: December 1991 L, . College Journalism Communications Dean, Graduate School

PAGE 157

UNIVERSITY OF FLORIDA lllllllllll llllll lllll lllll llllllllllllllllllll 11111111111111111 3 1262 08553 4914


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EVW1YGHAG_1E7E8S INGEST_TIME 2017-07-20T21:18:41Z PACKAGE UF00102734_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES