Perceptions of Public Opinion.

This study examines the influence of individuals' attitudes on their perceptions of the distribution of attitudes among others. Specifically, it focuses on the relationship between preferences for political c and idates during the 1980 presidential election campaign and perceptions of public opinion about the c and idates. Two data sets were employed: (1) a survey of a systematic r and om sample of undergraduates at the University of Michigan; and (2) a survey of a national sample of American adults (American National Election Study). The findings reveal a strong positive relationship between individuals' preferences for political c and idates and their perceptions of current public opinion. Also, an extremely strong relationship between preferences for political c and idates and predictions of the election outcome was obtained. This propensity for individuals to perceive or estimate their own preferences to be uncommonly prevalent among others is referred to here as the "projection effect." Two factors were hypothesized to influence the projection effect: attitude intensity and exposure to information about the actual distribution of opinion. Intensity was predicted to enhance the projection effect and exposure to information was expected to diminish the effect. The data reveal rather consistent although modest effects of attitude intensity, indicating that, by and large, the projection effect is stronger among those with intensely held preferences than among those who feel less intensely. Little evidence was found to confirm the expectation that exposure to information attenuates the effect. Some analyses suggest support for the hypothesis in that they produced the predicted pattern of results, but the trends do not reach statistical significance. Some of the previous research about the relationship between individuals' political attitudes and their perceptions of the distribution of others' attitudes has been concerned with the influence of public opinion polls on voters' behavior--the "b and wagon effect." The projection effect predicts the same overall positive relationship between these two variables but the correlation is assumed to be due to individuals' preferences. The present findings are more supportive of a projection interpretation than a b and wagon interpretation. Finally, evidence of a different type of effect of polls--a "viability effect" when there are three or more c and idates--was found.PhDSocial psychologyUniversity of Michiganhttp://deepblue.lib.umich.edu/bitstream/2027.42/159162/1/8304470.pd