************************************************************************
This paper was presented at 1993 annual meeting of the Midwest Association for Public Opinion Research
************************************************************************
INTRODUCTION
Prior to his 1983 article on "The Third Person Effect in Communication," W. Phillips Davison "became involved in the local phase of a national election, serving as a volunteer for his preferred candidate's organization." In reaction to receiving a leaflet of impressive quality from the opposition, Davison "procured a pile of political literature from his own party's local office and spent the rest of the day distributing it door to door." However, he found that "informal post-election analyses suggested that neither set of propaganda materials had exerted much influence on the voters." Davison asked himself why he had assumed that the rival candidate's leaflet would be so effective. The result of his inquiry led to his formulation of the "third person effect hypothesis." (Davison, 1983)
This young political scientist found himself involved in a similar experience, that of managing the election campaign for a local circuit judgeship. In preparation for the campaign, this campaign manager interviewed a number of party officials and former candidates, both successful and unsuccessful, from both parties, in order to devise a winning strategy. He was particularly interested in developing an effective media campaign. In spite of the discovery of Lazarsfeld, et. al., (1944) that very few voters are converted by political propaganda, the local politicos insisted that certain media were particularly influential. Could their "seat-of-the-pants" or "gut instinct" advice be accurate? Or, were the politicians and citizens overestimating the influence that the media have on the attitudes and behaviors of others? The answers were important to the campaign manager and to his candidate.
In his landmark article, Davison reported a series of minor and informal efforts to test whether "an individual who is exposed to a persuasive communication via the mass media will see this communication as having a greater effect on other people than on himself or herself." (1983, p. 4) Davison wrote:
Much of the testing of the third person effect has been experimental. For example, Cohen, et. al. (1988), in an experiment involving audience and jury assessments of the effects of defamatory communications, found the existence of a third person effect. They found that defamatory communications are perceived to affect others more than oneself and that as the definition of others is broadened the perceived effect on others increased. Tyler and Cook (1984) noted, in their investigation of the "impersonal impact hypothesis," that people distinguish between two possible levels of judgment, societal and personal. Based on a series of experiments, Tyler and Cook concluded:
The alternative to the third person effect most often discussed in the literature on pluralistic ignorance is the "looking glass" perception, the belief that others think the same as oneself (Taylor, 1982; O'Gorman, 1980, 1979, 1975; O'Gorman & Garry, 1976; Fields & Schuman, 1976; Breed & Ktsanes, 1961). Fields and Schuman have stated: "In the absence of strong counter-forces, a large proportion of people feel that the world they live in agrees with their own opinions on public issues" (1976, 445). The looking glass perception can have the majority accurately perceiving the position of the majority or the minority accurately perceiving the position of the minority. The latter has been termed the "false looking glass" or the "false consensus" (Scheff, 1975).
Lemert tested the looking glass phenomenon through several surveys prior to voting on two controversial ballot measures in Oregon (1986). The study compared the accuracy of election predictions by politicians with those by members of the voting public. He hypothesized and found that "politicians were better than voters at predicting the outcomes of two controversial ballot measures, and they displayed fewer and weaker signs of the Looking-Glass effect" (p. 219).
Little work has explored the differences in perceived influence among the various media. The purpose of this study was to test Davison's third person effect hypothesis with regard to rural voters' perceptions of the influence of various media. Unlike prior studies, this study combined examination of both interpersonal and mass media. During interviews in preparation for the campaign, the political leaders and candidates identified eleven "media" by which they believed political information is transmitted and through which voting decisions are influenced: meeting the candidates face-to-face (through door-to-door campaigning and the endless fried buffalo fish dinners), meeting campaign workers and political party officials (who often serve as candidate surrogates), family and relatives, friends and neighbors, co-workers, the local (and only) AM farm radio station, the "throw-away shopper" all-advertising newspaper, the regional newspaper (published out of county, but with a weekly local supplement), mailed literature, handouts and leaflets, and yard signs/store signs and bumper stickers. The five interpersonal sources were suggested without any knowledge of Katz and Lazarsfeld's two-step flow model of public opinion (1955). Although the region borders the statistical metropolitan area (SMA) of St. Louis and many residents are employed in the greater metropolitan area, no candidate could afford to utilize the St. Louis newspapers, radio stations or television stations, thus they were not included in this study.
When interviewed, the politicos offered two consistent responses. First, they agreed that the primary source of political information and most consumed medium in the county was the "shopper." As a result and following established tradition, every candidate advertised primarily and heavily in the weekly "throw-away" paper. Because of the advertising demand, the paper, delivered by the postal service to almost every household in the county, added a second section during the weeks prior to November 1990 election. Second, they agreed that the most influential element in any local race was face to face contact between the candidate and the voters. As a result and following established tradition, every candidate attempted to knock on every door of every farmhouse, townhouse or shanty, as well as attend every church picnic and buffalo-fish fry, and walk in every parade. The author's candidate and his wife distributed nearly a thousand pencils, each with a double-headed eraser that looked like a gavel and the imprinted slogan, "Keep the Gavel in the Hand of Experience."
This study hypothesized that, should the third person effect be operating, the voters will declare that others are more influenced by each of the various media than are they. Furthermore, if the local officials were correct, the candidates and the "shopper" should be rated as the most influential of the media.
METHOD
This study is based on results of exit polling during the 1990 elections in Jersey County, Illinois. Jersey is a small rural Illinois county (population just over 20,000) located at the confluence of the Mississippi and Illinois rivers about fifty miles north of St. Louis, Missouri. The county's population is divided between rural areas and the county seat of Jerseyville. There are 12 voting precincts in the city of 7,500 and 17 precincts in the rest of the county. Based on the voting lists, it was found that the numbers of registered voters and of those regularly voting are evenly distributed between county seat and the rural county.
Four precincts, two rural and two city, that have been most consistent with the county's voting patterns were selected for polling. Jersey County, much like the state of Illinois, is a split or swing county. Sheriff, county clerk and state representative are Republicans while the circuit clerk, state's attorney, coroner and state senator are Democrats. Out of the 22 contested races in the last three general elections (1988, 1986, 1984), one of the rural precinct voted 100% with the county-wide winners, whether Republican or Democrat. The three other precincts voted with the county 21 out of 22 times (95.4% consistency). The choice of the four precincts was reconfirmed with the results of the November 1990 general election. All four precincts voted 100% with the county in the seven major contested state and county-wide races.
The poll was a three page survey, which was self-administered, offered to every voter at each of the four precincts. A total of 542 persons out of 980 voters completed at least part of the survey. The questions designed to explore the third person effect were part of a larger set of questions exploring media use patterns in local political elections. For each of the eleven media identified in the preliminary interviews, each respondent was asked two questions: "How much did these sources influence how you voted?" and "How much do you think these sources influenced how others voted?" The respondents answered by circling "a lot," "some" or "none" on a three element Likert-scale. There were 478 valid cases for the questions on media influence.
The results for each medium were subjected to crosstab analysis resulting in six cell tables. The results were recoded into four cell format most appropriate to third-person effect analysis. Both possible combinations, "none/some" vs. "a lot" and "none" vs. "some/a lot," were calculated. With only one exception, the "none/none" cell was substantially larger (by an average factor of 3.5) than the "a lot/a lot" cell. Therefore, the latter combination of separating out the "none" answers and combining the "some" and "a lot" answers was deemed more usable for identifying the different effects.
RESULTS
The looking glass effect occurs when people believe that others behave or are influenced in the same manner or to the same degree. The "true" looking glass effect occurs when the majority accurately gauges the behavior of or influence on the majority. The alternative, when the minority accurately gauges the minority, is a "false" looking glass, because it mirrors the minority, not the majority.
Pluralistic ignorance occurs when people believe that others behave or are influenced in a different manner, regardless of their majority or minority status. Conservative bias is the form of pluralistic ignorance in which people believe that others behave or are influence to a lesser degree, hence the reference to conservative. The third person effect is a form of pluralistic ignorance, specifically related to influence, in which people believe that others are influenced to a greater degree than themselves.
Therefore, those who respond that they are not influenced and others are not influenced by a certain medium are demonstrating the looking glass effect. Likewise, those who respond that they are influence some or a lot and others are influenced some or a lot are also demonstrating the looking glass effect. The larger or majority group is the true looking glass while the smaller or minority group is the false looking glass. Those who believe others are influenced less than they are (none vs. some or a lot) exhibit a conservative bias. Conversely, those who claim they are influenced less than others (none versus some or a lot) are exhibiting the third person effect. (See Table I)
Six Jersey County media revealed true looking glass effects: family and relatives, political party officials, candidates, the local throw-away paper, mailed literature, and handouts/leaflets. Only one of the media, the candidates themselves, revealed a true looking glass effect for the majority of the respondents. The other five revealed a true looking glass effect for only a plurality of those answering. Five of the media revealed third person effects: friends and neighbors, co-workers, local radio station, regional newspaper, and yard signs/bumper stickers. Only the co-workers category revealed a third person effect for the majority of the respondents. The other four media revealed third person effects for a plurality of the respondents. Conservative bias was minimal for all media (ranging from 1.2% to 3.8%). (See Table II) As a result, when total looking glass effects (both true and false) are compared to pluralistic ignorance (both third person effect and conservative bias), only co-workers demonstrated pluralistic ignorance. All other media demonstrated the looking glass effect. (See Table III) In each case, the effect was significant (df 1, p<.001). (See Table IV).
DISCUSSION
Why is it that the third person effect appears to have so little impact? It could be, as suggested by Glynn and Ostman (1988), that the effect has little societal impact. However, there is a stronger argument consistent with the pluralistic ignorance hypothesis. Pluralistic ignorance implies people do not know what others think. This may be, in fact, just the opposite in Jersey County.
Jersey County is a close-knit, extended family community. In a prior election, the incumbent sheriff was challenged by his son-in-law's father, a retired state trooper. The county treasurer and the state's attorney for the neighboring county are brothers of the deputy chairman of the county board and former chairman of the county Democratic party. Their sister is married to another member of the county board. Everyone is on amiable terms and each relative has to run on his or her own merits. For example, the brother-in-law of the county clerk has repeated challenged the deputy chairman of the board for his seat. And, the member of a large and well-known farming family (with members on the school board and board of trustees of the community college) failed to unseat the county treasurer.
My candidate was unique in that he was born, raised and educated in California. Our primary election opponent campaigned on a heritage of 177 years of family in the county and membership on the best basketball team ever fielded by our high school. Our general election opponent's mother is editor of the regional newspaper's county supplement. Everybody knows everybody. As a result, appears to be little if any pluralistic ignorance in Jersey County. The local politicos may not have quantitative skills, but they do know their constituents.
Since third person effect is a component of pluralistic ignorance, low pluralistic ignorance means low third person effect. Cohen's jury experiment indicated that the more distant the audience, the stronger the third person effect. Perhaps the opposite is true; the closer the audience, the lower the third person effect. These results are also consistent with Lemert's study of the third person effect on two ballot issues, in which the politicians did a better job in predicting the outcome of the votes than did citizens. Politicians, contrary to our cynical tales, may actually have a good idea of what is occurring.
As hypothesized, the candidates were considered the most influential medium, followed by the shopper and political party officials.
My candidate won.
WHAT INFLUENCES ME
| None | Some or A Lot |
| 3. Looking Glass (false)
2. Third Person Effect |
4. Conservative Bias
1. Looking Glass |
2 + 4 = pluralistic ignorance
| SOURCE OF INFLUENCE | LOOKING GLASS | THIRD PERSON | LOOKING GLASS-F | CONSERV. BIAS |
| Political candidates | 69.4% | 17.2% | 9.6% | 3.8% |
| Local "throw-away" paper | 48.8 | 34.6 | 14.7 | 1.9 |
| Political party officials | 48.3 | 31.8 | 17.0 | 2.9 |
| Family and relatives | 48.2 | 38.8 | 11.8 | 1.2 |
| Literature by mail | 46.0 | 35.2 | 16.2 | 2.6 |
| Handouts and leaflets | 45.8 | 33.9 | 16.9 | 3.3 |
| Friends and neighbors | 40.6 | 45.1 | 13.1 | 1.2 |
| Regional newspaper | 39.4 | 40.9 | 18.3 | 1.4 |
| Yard signs & bumper stickers | 35.5 | 40.5 | 22.1 | 1.9 |
| Co-workers | 29.7 | 52.1 | 16.5 | 1.7 |
| Local radio station | 29.5 | 47.2 | 22.1 | 1.2 |
| SOURCE OF INFLUENCE | LOOKING GLASS EFFECT | PLURAL. IGNOR. EFFECTS |
| Political candidates | 79.0% | 21.0% |
| Political party officials | 65.3 | 34.7 |
| Local "throw-away" paper | 63.5 | 36.5 |
| Handouts and leaflets | 62.7 | 37.2 |
| Literature by mail | 62.2 | 37.8 |
| Family and relatives | 60.0 | 40.0 |
| Regional newspaper | 57.7 | 42.3 |
| Yard signs & bumper stickers | 57.6 | 42.4 |
| Friends and neighbors | 53.7 | 46.3 |
| Local radio station | 51.6 | 48.4 |
| Co-workers | 45.2 | 53.8 |
| SOURCE OF INFLUENCE | CHI SQUARED
(P<.001) |
DF | SIGNIFICANCE |
| Political candidates | 65.7 | 1 | yes |
| Political party officials | 55.9 | 1 | yes |
| Local "throw-away" paper | 51.9 | 1 | yes |
| Yard signs & bumper stickers | 49.3 | 1 | yes |
| Regional newspaper | 47.6 | 1 | yes |
| Literature by mail | 46.8 | 1 | yes |
| Handouts and leaflets | 45.6 | 1 | yes |
| Family and relatives | 41.1 | 1 | yes |
| Local radio station | 38.8 | 1 | yes |
| Friends and neighbors | 32.2 | 1 | yes |
| Co-workers | 20.7 | 1 | yes |
| SOURCE OF INFLUENCE | TRUE LOOKING GLASS | THIRD PERSON | ALL LOOKING GLASS* |
| Political candidates | 69.4% | 17.2% | 79.0% |
| Local "throw-away" paper | 48.8 | 34.6 | 63.5 |
| Political party officials | 48.3 | 31.8 | 65.3 |
| Family and relatives | 48.2 | 38.8 | 60.0 |
| Literature by mail | 46.0 | 35.2 | 62.2 |
| Handouts and leaflets | 45.8 | 33.9 | 57.6 |
| Friends and neighbors | 40.5 | 45.1 | 53.7 |
| Regional newspaper | 39.4 | 40.9 | 57.7 |
| Yard signs & bumper stickers | 35.5 | 40.5 | 57.6 |
| Local radio stations | 29.5 | 47.2 | 51.6 |
| Co-workers | 28.7 | 52.1 | 45.2 |
* includes both "true" and "false" looking glass effects.
* Breed, Warren and Thomas Ktsanes (1961). "Pluralistic Ignorance in the Process of Opinion Formation." Public Opinion Quarterly 25(3):382-392.
* Cohen, Jeremy, Diana Mutz, Vincent Price and Albert Gunther (1988). "Perceived Impact of Defamation: An Experiment on Third-Person Effects." Public Opinion Quarterly 52(2):161-173.
* Davison, W. Phillips (1983). "The Third-Person Effects in Communication." Public Opinion Quarterly 47:1-15.
* Fields, James M. and Howard Shuman (1976). "Public Beliefs About the Beliefs of the Public." Public Opinion Quarterly 40(4):427-448.
* Glynn, Carroll J and Ronald E. Ostman (1988). "Public Opinion About Public Opinion." Journalism Quarterly 65(2):299-306.
* Katz, Elihu and Paul F. Lazarsfeld (1955). Personal Influence. Glencoe, IL: Free Press.
* Lasorsa, Dominic L. (1989). "Real and Perceived Effects of 'Amerika'." Journalism Quarterly 66(2):373-378, 529.
* Lazarsfeld, Paul, B. Berelson, and H. Gaudet (1944). The People's Choice. New York, NY: Columbia University Press.
* Lemert, James B. (1986). "Picking the Winners: Politician vs. Voter Predictions of Two Controversial Ballot Measures." Public Opinion Quarterly 50:208-221.
* Mutz, Diana C. (1987). "Perceptions of Others in the Public Opinion Process: The Third Person Effect and the Spiral of Silence." Paper presented to the annual convention of the International Communication Association, Montreal.
* O'Gorman, Hubert J. (1980). "False Consciousness of Kind: Pluralistic Ignorance among Aging." Research on Aging 2(1):105-128.
* O'Gorman, Hubert J. (1979). "White and Black Perception of Racial Values." Public Opinion Quarterly 43(1):48-59.
* O'Gorman, Hubert J. (1975). "Pluralistic Ignorance and White Estimates of White Support for Racial Segregation." Public Opinion Quarterly 39(3):313-330.
* O'Gorman, Hubert J. and Stephen L. Garry (1976). "Pluralistic Ignorance: A Replication and Extension." Public Opinion Quarterly 40(4):449-458.
* Perloff, Richard M. (1989). "Ego-Involvement and the Third-Person Effect of Televised News Coverage." Communication Research 16(2):236-262.
* Scheff, Thomas J. (1975). "Toward a Sociological Model of Consensus." American Sociological Review 32(1):32-46.
* Taylor, D. Garth (1982). "Pluralistic Ignorance and the Spiral of Silence: A Formal Analysis." Public Opinion Quarterly 46:311-355.
* Tyler, Tom R. and Fay Lomax Cook (1984). "The Mass Media and Judgments of Risk: Distinguishing Impact on Personal and Societal Level Judgments," Journal of Personality and Social Psychology 8:195-200.