Upward Social Projection and Attitude Certainty

Abstract

Predicting the attitudes of a powerful out-group can influence attitude strength and the perceived status of in-group members. American farmers indicated their own attitude positions on eight issues relevant to conflict with the Environmental Protection Agency (EPA), and then estimated the positions either of their in-group, the EPA, a control out-group, or participated in a no-estimation control condition. The estimation of attitude disagreement with EPA agents elicited negative thoughts about their power and control which mediated a decrease in self-certainty. Assumed attitude contrast also diminished the perceived status of farmers. Discussion focuses on relational and instrumental sub-group connections.

Portions of this research were presented at the 2006 annual meeting of the Midwestern Psychological Association in Chicago, Illinois. Support for this work was provided in part by an Ohio State University Seed Grant.

Notes

¹Although predictions of the attitudes of others are commonly referred to as social projection (Clement & Krueger, 2002; Riketta, 2006; Robbins & Krueger, 2005), the term attitude estimation avoids confusion with any Freudian connotations to the projection label. Hence, the term social projection will be retained in the title of this paper to maintain its association with previous research, but the term attitude estimation will be used for the sake of clarity in the narrative. The terms assumed agreement/similarity and assumed contrast/disagreement/dissimilarity describe attitude estimations in respective intragroup or intergroup domains.

²These eight issues dispute priorities and methods by which the environment can be preserved rather than the importance of environmental protection per se. Although participants perceived a clash between their goal to profit from farming and the EPA's goal to expand its bureaucracy and control, they also agreed with the need to preserve the environment.

³Analyses of the Fisher Z scores converted from the between-participant correlation coefficients produced results similar to those found by analyzing the within-participant Fisher Z scores. Between-participant correlations indicate the relationship between the farmers' own attitude positions and their estimations of a target's position separately for each of the eight issues (cf. Clements & Krueger, 2002; Holtz & Nihiser, 2008). A one-way ANOVA conducted on these scores showed the main effect for attitude estimation target, F(2, 21) = 16.26, p < .001, η² = .61. The estimation of self-other attitude similarity to in-group members (Fisher Z: M = .62, SD = .27; r = .52) exceeded the estimation of similarity to the control out-group (Fisher Z: M = .24, SD = .25; r = .22), t(21) = 2.88, p < .01, d = .88, and also exceeded the estimation of contrast from the EPA out-group (Fisher Z: M = −.19, SD = .32; r = −.17), t(21) = 5.41, p < .001, d = 1.88. Additionally, the estimation of self-other contrast from the EPA out-group differed from assumed similarity to the control out-group, t(21) = 2.98, p < .01, d = 1.00.

Note. Higher means indicate greater frequency of thoughts. Within each column means with different subscripts differ significantly from each other.

⁴Thoughts related to the issues and the testing situation showed no differences between target conditions and no relationships with the other variables in this study. Also, no differences were found between thought valences involving these types of thoughts. In addition, no hypotheses were ventured concerning the frequency of self-related thoughts; although it makes sense that participants reported more positive thoughts about themselves (M = .41) than negative (M = .21) or neutral thoughts (M = .13), χ ² (2, N = 89)= 24.91, p<.001 (w = .53).

⁵This general pattern of results was also replicated in the two control conditions. However, in the control out-group target condition, the number of positive in-group related thoughts exceeded only the number of neutral in-group related thoughts, χ²(1) = 6.40, p < .02; w = .52 [χ²(2, N = 24) = 6.65, p < .04 (w = .52)]. Also, in the no-estimation control condition, positive in-group-related thoughts only exceeded the frequency of neutral in-group thoughts, χ²(1) = 11.00, p < .01; w = .69 [χ²(2, N = 23) = 14.10, p < .01 (w = .78)]. As the means presented in Table 1 suggest, positive in-group bias may have been accentuated by the estimation of opinions attributed to either the farmer in-group or the relevant EPA out-group.

Note. Lower means indicate lower certainty. Means with different subscripts differ significantly from each other. Cell ns range from 20 to 24.

⁶Within the in-group target condition, assumed similarity to farmers predicted increments in opinion certainty, r(20) = .42, p < .05; and positive thoughts about farmers predicted greater certainty, r(20) = .48, p < .03. Nevertheless, these thoughts did not mediate the relationship between Fisher Z scores and certainty. Overall, Fisher Z scores predicted positive thoughts about farmers, β = .25, t(64) = 2.10, p < .04; but Fisher Z scores remained a significant predictor of opinion certainty, β = .28, t(64) = 2.22, p < .03, when considered simultaneously with the influence of positive in-group thoughts on opinion certainty, β = .06, t(64) = .51, ns.

Note. Lower means indicate lower status. Means with different subscripts differ significantly from each other.