Shaping Public Opinion: Personal Sources of Information on a Major Political Issue

ABSTRACT - The increasing interest on the application of consumer behavior concepts in the political arena motivated the study of a political issue with national implications through the opinion leadership concept. It was concluded that opinion leaders have the potential to cause significant shifts in public opinion.

Citation:

Glenn S. Omura and W. Wayne Talarzyk (1983) ,"Shaping Public Opinion: Personal Sources of Information on a Major Political Issue", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 484-489.

Advances in Consumer Research Volume 10, 1983      Pages 484-489

SHAPING PUBLIC OPINION: PERSONAL SOURCES OF INFORMATION ON A MAJOR POLITICAL ISSUE

Glenn S. Omura, Michigan State University

W. Wayne Talarzyk, Ohio State University

[The authors gratefully acknowledge research support provided by the Fred B. and Mabel Dean Hill fund and Consumer Mail Panels of Market Facts, Inc.]

ABSTRACT -

The increasing interest on the application of consumer behavior concepts in the political arena motivated the study of a political issue with national implications through the opinion leadership concept. It was concluded that opinion leaders have the potential to cause significant shifts in public opinion.

INTRODUCTION

Consumer research has increasingly focused attention on the political arena (see, for example, Advances in Consumer Research, Volume VIII, 1981). Two areas which have been the object of much attention in the study of individual responses to political events are voter behavior (see, for example, Chapman and Palda 1981a , b) and public policy (see, for example, Van Liere and Dunlap 1980; Bernhardt and Stiff 1981). The combination of these two areas have also been investigated, including such topics as voting on bottle bills (Crosby, Gill and Taylor 1981), government expenditures (Noam 1981), and energy conservation policies (Journal of Consumer Research, December, 1981, special issue on energy).

These studies take on greater significance as politicians become increasingly sophisticated in their use of marketing approaches to sway public opinions (see, for example, Mauser 1982). The application of consumer behavior theories in the political arena can increase understanding of the dynamics of public opinion and allow prediction of voter responses to various political issues. The purpose of this paper is to examine a political issue from the perspective of interpersonal influence, specifically through application of the opinion leadership concept. The emphasis is on understanding the dynamics of the issue and not on the theoretical concept. Opinion leadership is used to provide understanding of the issue and r.of vice versa. Hence, the research problem focuses on the issue, and the problem is addressed by applying the concept.

BACKGROUND FOR THE STUDY

The research report here is part of a larger national project studying the attitudes and opinions of the public toward key social issues in early 1974. The basic objective of the project was to gain a better understanding or the characteristics and behavior patterns of people on alternative sides of major contemporary social and political issues. The issue under study here is the then impending impeachment of Richard M. Nixon. The concern was to study the interpersonal flow or influence of those who favored versus those w4ho were against impeachment. [For a recent application of the opinion leader concept in an area or political research, see Robinson (1976). Numerous authors have studied the concept while focusing on politically-related national events (see, for example, Mendelsohn 1964; Greenberg, Brinton and Farr 1965).]

The concern was not to test or construct theories of interpersonal influence, rather, to investigate the role interpersonal potentially could play in shaping public opinion on the presidential impeachment issue. The importance of opinion influence on shaping public opinion is made evident in Nixon's memoirs (Nixon 1978). Nixon felt that public opinion was the primary force endangering his position. If public opinion leaned heavily to impeachment, he felt that impeachment would be imminent. Consequently, he developed a media strategy specifically to influence the polls (Lang and Lang 1980).

The primary research question was:

* What are the characteristics and behavior patterns of those who favor and those who are against impeachment, particularly with respect to the trait of opinion leadership? Other variables or interest are mass media exposure and perceptions or the mass media, and socioeconomic characteristics.

The secondary research question was:

* What combination of the variables can be most effective in discriminating between those who favor and those who are against impeachment?

While the key theoretical concept of the study was opinion leadership, due to the historic nature of the issue, it was felt that documentation of the mass media and socioeconomic variables may be of interest to future researchers. These variables will be discussed in relation to opinion leadership. Further, research related to but reported after the data for this study was collected will be discussed in the Results and Conclusion sections. It should be noted that while a body of literature evolved from the crisis, no study was round which addressed the same questions of the present study. However, causal inferences about the relationship between the trait of opinion leadership and opinion on the issue or impeachment are not straightforward in this or any other study. The conclusions drawn from the results of this study are based on simple correlational statistics.

RESEARCH METHODOLOGY

A national sample of 1,000 households from the Consumer Mail Panel of Market Facts, Inc., was used to provide a balanced national sample parallel to census data for geographic divisions and within each division by total household income, population density, degree of urbanization, and age. The questionnaires were mailed during the first week or March, 1974. Each household contacted was specifically asked to have either the panel member, a female or her spouse, if married, to complete the questionnaire in order to balance respondents by sex. Of the total 1,000 original households contacted, 640 usable questionnaires were returned. Panel members received a small girt in appreciation for their participation.

As a check on the representativeness or the respondents, a chi-square analysis was made comparing the original 1,000 households contacted and the final 640 respondents on five characteristics: age, education, household income, occupation or husband, and geographic residence. The analysis indicated that the older ate groups were over-represented. Sixty respondents were randomly selected from the older groups and removed to provide an adjusted sample. The final sample base of 580 respondents was found to be representative of the original population across the five characteristics.

The measure of interpersonal influence used in this study was an opinion leadership scale similar to the self-report method used by Rogers (1961), discussed by Rogers and Cartano (1962) and modified by King and Summers (1970). [The Rogers and Cartano version has been tested for construct validity (Jacoby 1974) and the King and Summers version for reliability (Yavas and Riecken 1989) and found acceptable on both measurement tests.] The seven self-reported opinion leadership scale questions as they appeared in the questionnaire are presented in Table 1 as part of the results section. The mass media variables were measured according to self-reported media exposure and self-reported perceived accuracy of the media. In order to measure relative exposure to alternative media, respondents were asked to report approximately how many hours per week on the average they devoted to each of the seven media categories. [The seven categories were described as: magazines (all types except news), magazines (news), television (all shows except news), television (news), radio (all shows except news), radio (news), and newspapers. It should be noted that the media variables were designed with the objective of serving the entire project and not only the needs of the study reported here. Hence, the questions determining exposure and perceived accuracy have to do with general viewing behavior and perception, and not necessarily only with the specific issue of Nixon's impeachment.] Perceived accuracy was measured by asking each respondent to indicate how accurate they believed radio, newspapers, television and magazines were in reporting the news. A seven point response scale, ranging from very inaccurate (1) to very accurate (7) was used. Standard socio-economic variables were available for each panel member.

Opinion about impeachment was obtained in the following manner: "Recently in the news there has been a lot of discussion concerning the impeachment of President Nixon. In your opinion, should President Nixon be impeached? Yes 2 No 2 ." Respondents who left the question blank or wrote-in undecided or no opinion were classified as non-usable in arriving at the final sample of 580 since the basic objective of the research was to understand the similarities and differences between those who favored and those who were against impeachment as opposed to those who were undecided.

RESEARCH RESULTS

Of the 580 usable respondents to the survey, 219 (37.8 percent) were in favor of impeachment while 361 (62.2 percent) were against such an action. This finding is within the range of other national polls taken about the same time (Lang and Lang 1980). The following four sections will compare these two groups of respondents across four different measures. A last section will present the results of an attempt to discriminate between the two groups using a linear combination of the four sets of measures.

Interpersonal Influence

Table 1 displays the questions used to form the opinion leadership score along with percentage responses to each question for those favoring and those against impeachment as well as for the sample as a whole. This table also compares the two groups in terms of their composite opinion leadership scores. The chi-square statistics for all of the cross classifications in Table 1 are significant at the .01 level.

As can be seen, there are major differences between those for and those against impeachment both in response to the individual seven questions making up the opinion leadership measure and the composite opinion score itself. In terms of responses to the seven questions, it is clear that the interpersonal communication characteristics of those favoring and those against impeachment are very different. When compared with respondents against impeachment those favoring impeachment were more likely to: (1) enjoy talking about the issue, (2) perceive themselves as providing information to friends about the topic, (3) talk with friends about impeachments (4) perceive themselves as more likely to be asked for information about the impeachment issue, (5) try to convince friends of their ideas about impeachment, (6) tell friends about some aspect of the impeachment issue, and (7) perceive themselves as regarded by friends as a-good source of information about the impeachment issue.

While 26.0 percent of the total sample were classified as "high" in self-reported levels of opinion leadership, 41.6 percent of those in favor of impeachment were so classified. [Levels of opinion leadership were divided into three categories. Those "high" in opinion leadership scored in the top 25% on the composite index. Those "medium" in opinion leadership scored in the middle 50% on the composite index. Those "low" in opinion leadership scored in the lower 95% on the composite index.] This compares with only 16.6 percent of those against impeachment being classified as "high" in level of opinion leadership. Thus, 41.6 percent of those favoring impeachment (219), or 16.2 percent of the total sample, may be viewed as the population segment which could have shaped public sentiment toward impeaching then President Nixon.

Media Exposure Patterns

No statistically significant relationships were found to exist between the amount of time per week devoted to each of the media and opinion about the impeachment issue. The similarities in media behavior patterns for both groups may be observed in Table 2. Those against impeachment seem more likely than those for impeachment to be in the heaviest exposure category of media behavior for all media except television news. This result on television viewership is similar to that found by Holm, Kraus, and Bochner (1974), Laing and Stevenson (1976), and Stevenson and Laing (1976). The first study focuses on Nixon's image during the Watergate hearings and measure media exposure similar to the present study. The latter two studies are impeachment issue specific and refer to-television viewing (only) of the impeachment hearings specifically.

Perceived Media Accuracy

For all classification of media, respondents who favored impeachment rated accuracy in reporting the news to be higher than did those who were against impeachment. Table 3 presents the details of this analysis phase. The chi-square statistics for both electronic media in this table are significant at the .01 level. It is interesting to note that both groups were somewhat consistent in their relative perceptions of media accuracy. Television was rated as "accurate" most often and magazines as "accurate" least often by both groups of respondents. These findings are comparable to those of Robinson's (1974) who states that television reports of the Watergate situation was perceived as fairer and more accurate than newspaper accounts.

TABLE 1

RESPONSE TO IMPEACHMENT ISSUE CROSS CLASSIFIED WITH QUESTIONS IN OPINION LEADERSHIP SCALE AND COMPOSITE OPINION LEADERSHIP SCORE

Socio-Economic Characteristics

While none of the chi-square statistics for the cross tabulations in Table 4 are significant at the .01 level, visual analysis suggests some directional relationships. Those in favor of impeachment, when compared to those against, are more likely to be younger, less likely to have higher incomes, and slightly more likely to have less education.

Multiple Discriminant Analysis

To examine the question or what combination of the variables used in the previous four sections could be most effective in discriminating between those who favor and those who are against impeachment, stepwise multiple discriminant analysis was employed. The objective was to develop a profile of the characteristics that best aids in distinguishing between the two groups of respondents. In addition, this analysis yielded a model for the prediction of group membership given knowledge of a person's responses to the differentiating characteristics.

Validation samples were used to test the efficacy of the derived discriminant statistical model. The individuals for each validation sample were randomly drawn from the two respective groups prior to the discriminant analysis. Approximately 95 percent of each original group went to the validation samples.

Based upon a significance level of .05, four variables combined to form the discriminating statistical model. Respondents who favored impeachment, compared to those against impeachment, were found to:

- have higher scores on the opinion leadership scale

- devote less time to reading newspapers

- devote less time to listening to radio (all programs except news)

- have high perceived accuracy of television

In order to determine how well the variables used in the discriminant model predict group membership, the proportional chance criterion Cpro was calculated. The criterion serves as an index of how well both groups were correctly identified given unequal group sizes. If the percentage of the total number correctly classified exceeds Cpro then the model derived is more effective than what might have occurred through chance (Morrison 1969).

For both the original and the validation samples, the derived discriminant model yielded high levels or correct classification as shown in Table 5. The 66.7 percent correct prediction of those for impeachment and the 69.4 percent prediction of those against impeachment exceeded the Cpro value for the original sample. The percentage of correct group classifications for the validation groups also exceeded the Cpro value for that sample. Thus. the variables of opinion leadership, perceived accuracy of television, exposure to newspapers and radio are partially validated in terms of their ability to discriminate between those favoring and those against impeachment.

TABLE 2

RESPONSES TO IMPEACHMENT ISSUE CROSS CLASSIFIED WITH PATTERNS OF MEDIA EXOOSURE

DISCUSSION

The results of this research demonstrate that there were major characteristic differences between those who favored and those who were against the impeachment of President Nixon. In the univariate statistical analysis, while both groups of respondents were relatively similar in their media behavior patterns and to some extent in their socio-economic characteristics, they were very dissimilar in their self-designated levels of opinion leadership and their perceived accuracy of certain news media. In the multivariate statistical analysis, based upon a combination of the classification information, it was found that certain variables such as level of opinion leadership, perceived accuracy of television, and exposure to newspaper and radio news were the most powerful in collectively discriminating between those for and those against impeachment. A basic linear model using information on these characteristics for each respondent yielded successful predictions of opinion on impeachment well above chance.

TABLE 3

RESPONSES TO IMPEACHMENT ISSUE CROSS CLASSIFIED WITH PERCEIVED ACCURACY OF MEDIA

TABLE 4

RESPONSES TO IMPEACHMENT ISSUE CROSS CLASSIFIED WITH SELECTED SOCIO-ECONOMIC CHARACTERISTICS

TABLE 5

VALIDATION OF DISCRIMINANT FUNCTIONS FOR IMPEACHMENT ISSUE RESPONSES

Several key points may be concluded from this study:

1. Had President Nixon not resigned from office, the results suggest that opinion in favor of impeachment could have increased. This conclusion is based on the result that respondents in favor of impeachment rated themselves higher on opinion leadership and were involved in talking about the topic and trying to convince others of their point of view. Meanwhile, those against impeachment perceived themselves to be lower in opinion leadership and were not so much involved in discussing the topic with others. Hence, if interpersonal influence is assumed to be a factor in determining opinion on impeachment, as the discriminant analysis predicted, interpersonal influence appeared to work in favor of impeachment. This conclusion is partially supported by related research. Holm, Kraus and Bochner (1974) found through a cross-lagged analysis of a Cleveland panel that those who held a positive image of the President tended to keep quiet whereas those more negatively predisposed tended to disseminate their opinions. In diagnosing the potential impact of any major social issue, opinion leadership can contribute substantially to the theoretical understanding of the formation of public opinion.

2. The finding of no difference in media exposure time between those favoring and those against impeachment suggests that the amount of media-based general information flow was identical for both groups. However, the differences found between the groups on the perceived accuracy of the media suggest that media influence may have been more strongly supportive of favoring impeachment than against. The perceived accuracy of the media by those favoring impeachment and conversely, the perceived inaccuracy of the media by those against impeachment, suggest that the media may have been perceived to report information tending more toward supporting impeachment. Unfortunately, it cannot be determined from the data collected in this study whether the media were biased and influential or simply relayed neutrally-affecting information which provided the basis for the public's opinions about impeachment. Research which is more specific provide ambiguous support for these alternatives. The findings from these studies may be interpreted from the perspectives of information processing, perception, or ecologicalism. Holm, Kraus, and Bochner (1974) show that non-Democrats who supported Nixon during the Watergate hearings avoided television coverage when there was substantial interpretive content, and shifted instead to newspapers where they could select the news items to read. Laing and Stevenson (1976) conclude that television viewers of the impeachment hearings had more information and this may be the cause of their predisposition toward favoring impeachment. These two studies suggest a future test or an interdependent or nonrecursive model of media use and influence concerning major social and political issues. .Along perceptual lines, Edelstein and Tefft (1974) conclude that the medium with the greatest degree of audience trust of Watergate information was television and that substantial variance existed among the media. In contrast to viewer perceptions, Larson (1974) content analyzed the television and newspaper media and concludes that there was no systematic bias reflecting Nixon's credibility, innocence, or guilt. Ecological cues projected by the media appear to be unbiased. Hence, while audience perception toward media reporting on the impeachment issue is inconclusive, it seems that the perception of the information relayed by the media may be a key to explaining voter behavior.

The impact of the media-on important social and political issues is still largely unknown and is in need of further study. When the issue is a national crisis, the role and responsibilities of the media appear to need further study and clarification. The significance of the association between perceived accuracy of certain media and opinions held about the impeachment issue in this study emphasizes the importance of media as a potential behavior modifier in regard to contemporary social and political issues.

3. While socioeconomic characteristics such as age, income and education tended to be related to opinions about the impeachment issue, they were not as strong determinants as were other respondent characteristics. With this in mind, political studies that attempt to profile alternative groups of people solely on the basis of socioeconomic variables may be limited in explanatory power.

4. Characteristics which by themselves may not show major differences between groups may be significant in combination with other variables in explaining and predicting group membership. In this study, levels of newspaper readership and radio news listenership were not significant in relationship to opinions about the impeachment issues when analyzed through univariate statistics. However, in combination with opinion leadership and perceived accuracy of television through a multivariate analysis, these two additional variables were significant in predicting group membership.

The research reported here indicates that it is important to study issues of high social and political impact in order to gain a better understanding or the dynamics of public opinion of those issues. While causal relationships cannot be demonstrated, the findings suggest that in situations where interpersonal influence variables can differentiate between those on separate sides of major social and political issues, the potential for those variables to shape overall public opinion appears great.

REFERENCES

Bernhardt, Kenneth L. and Ronald Stiff (1981), "Public Policy Update: Perspectives on the Federal Trade Commission," in Kent B. Monroe, ed., Advances in Consumer Research, Volume VIII, Ann Arbor: Association for Consumer Research, 452-454.

Chapman, Randall O. and Kristian S. Palda (1981a), "Estimating the Impact of Campaign Expenditures on Voting Behavior in a Simultaneous Equations Context," in Kenneth Bernhardt et al., eds., Educators' Conference Proceedings, Chicago: American Marketing Association. 256-260.

Chapman, Randall O. and Kristian S. Palda (1981b), "Voting Participation in a Public Consumption Perspective," in Kent B. Monroe, ed., Advances in Consumer Research, Volume VIII, Ann Arbor: Association for Consumer Research. 530-533.

Crosby, Lawrence A., James D. Gill; and James R. Taylor (1981), "Consumer/Voter Behavior in the Passage or the Michigan Container Law," Journal of Marketing, 45 (Spring), 19-32.

Edelstein, Alex S. and Diane P. Tefft (1974), "Media Credibility and Respondent Credulity with Respect to Watergate," Communication Research, 1 (October),426-439.

Greenberg, Bradley S., James E. Brinton, and Richard S. Farr (1965), "Diffusion of News about an Anticipated Major News Event," Journal of Broadcasting, 9 (Spring), 129-142.

Holm, John, Sidney Kraus, and Arthur P. Bochner (1974), "Communication and Opinion Formation, Issues Generated by the Watergate Hearings," Communication Research, 1 (October), 368-390.

Jacoby, Jacob (1974), "The Construct Validity of Opinion Leadership," Public Opinion Quarterly, 38 (Spring), 8189.

King, Charles W. and John 0. Summers (1970), "Overlap of O?inion Leadership Across Consumer Product Categories," Journal of Marketing Research, 7 (February), 43-50.

Laing, Robert B. and Robert L. Stevenson (1967), "Public Opinion Trends in the Last Days of the Nixon Administration," Journalism Quarterly, 53 (Summer), 294-302.

Lang, Gladys Engel and Kurt Lang (1980), "Polling on Watergate: The Battle for Public Opinion," Public Opinion Quarterly, 44 (Winter), 530-547.

Larson, Charles C. (1974), "A Content Analysis of Media Reporting or the Watergate Hearings," Communication Research, 1 (October), 440-448.

Mauser, Cars A. (1989), Winning Elections, New York: Praeger.

Mendelsohn, Harold (1964), "Broadcast vs. Personal Sources of Information in Emergent Public Crises: The Presidential Assassination," Journal of Broadcasting, 8 (Spring), 147-156.

Morrison, Donald G. (1969), "On the Interpretation of Discriminant Analysis," Journal of Marketing Research, h (May), 156-163.

Nixon, Richard M. (1978), RN: The Memoirs of Richard Nixon, New York: Grosset and Dunlap.

Noam, Eli M. (1981), "Measuring Subjective Valuation and Demand for Government Services," in Kent B. Monroe, ed., Advances in Consumer Research, Volume VIII, Ann Arbor: Association for Consumer Research, 534-538.

O'Keefe, Garrett J. Jr. and Harold Mendelsohn (1974), "Voter Selectivity, Partisanship, and the Challenge of Watergate," Communication Research, 1 (October), 345-367.

Robinson, John P. (1974), "Public Opinion During the Watergate, Crisis " Communication Research, 1 (October), 391-405.

Robinson, Michael J. (1975), "American Political Legitimacy in an Era of Electronic Journalism: Reflections on the Evening News," in Richard Adler, ed. Television as a Social Force: New Approaches to TV Criticism, New York: Praeger, 97-149.

Robinson, John P. (1976), "Interpersonal Influences in Election Campaigns: Two Step Flow Hypotheses," Public Opinion Quarterly, 40, 304-319.

Rogers, Everett (1961), "Characteristics of Agricultural Innovators and Other Adopter Categories," Wooster, Ohio: Ohio Experiment Station. Research Bulletin. 882.

Rogers, Everett and David G. Cartano (1967), "Methods or Measuring Opinion Leadership," Public Opinion Quarterly, 26 425-431.

Stevenson, Robert L. and Robert B. Laing (1976), "The Audience for the Impeachment Hearing," Journal of Broadcasting, 20 (Spring), 159-168.

Van Liere, Kent D. and Riley E. Dunlap (1980), "The Social Bases of Environmental Concern," A Review of Hypotheses, Explanations, and Empirical Evidence," Public Opinion Quarterly, 44 (Summer), 181-197.

Yavas, Ugur and Glen Riecken (1982), "Extensions of King and Summers' Opinion Leadership Scale: A Reliability Study," Journal of Marketing Research, 19 (February), 154-155.

----------------------------------------