Better than better-than-average (or not): Elevated and depressed self-evaluations following unfavorable social comparisons

Abstract

Two experiments were designed to investigate perceivers' self-evaluations when they received objectively positive above-average performance feedback but were told about another coactor who performed either moderately or much better than the participant. Results indicated that participants responded negatively to this comparison information even though they received better-than-average performance feedback. Participants were given the opportunity to evaluate themselves relative to another coactor who was described as performing at an average level. When the negative implications of the unfavorable social comparisons were relatively mild, both low and high self-esteem participants raised their self-evaluations vis-à-vis the inferior coactor who performed at an average level on the task. However, when the upward comparisons were especially unfavorable (i.e., when there was a large discrepancy between the performance level of the participant and the coactor—the comparison target), only high self-esteem participants raised their self-evaluations. Results provided evidence for active compensation and relatively passive spreading activation, supporting a schema-maintenance through compensation model (e.g., Seta & Seta, 1982, 1993; Seta, Seta, & McElroy, 2003).

Notes

1 Individuals also are motivated to avoid events that are inconsistent with important schemata.

2 Seta et al. (1999) found that low self-esteem persons demonstrated a stronger self-enhancement response relative to high self-esteem persons on a task that was low in self-relevance but only high self-esteem persons were able to self-enhance in a highly self-relevant task. The low self-relevance task used in the Seta et al. (1999) study was conceptually similar to the mildly dissonance producing situation employed in the Steele, Spencer, & Lynch (1993) study, in which low self-esteem persons were seen to be especially inclined to rationalize the negative implications of a decision.

3 In the superior and very superior conditions, we also asked participants questions about their performance in relation to the superior performer. Importantly, their expectation of the score they would receive was not correlated with their comparative self-evaluation suggesting that comparative evaluations were not the result of an association with the superior performer. We have not discussed these measures in detail because, in Experiment 2, we provided a more direct test of whether performers' elevated self-ratings were a response to an upward comparison threat or a result of an association with a superior performer.

4 Several participants appeared to misunderstand our manipulation check questions. Inspection of the original question for the first 32 participants indicated that a few participants did not respond. Therefore, we revised our question to make it more understandable for participants and substituted this wording for the remainder of the participants. Clarity was achieved by rewording the question and providing a midpoint on the scale. We based our manipulation check analysis on the revised question.

5 For exploratory purposes, we also performed an ANOVA considering the task related (e.g., performance ratings) versus indirectly related (dependability, maturity) attributes as repeated measures as an additional component of the design. Although there was a marginally significant main effect of the measure variable F (1, 137) = 3.06, p < .10, there were no significant interactions involving either self-esteem or performance condition and the measure variable. This dimension may not have mattered in the present study because the attributes of dependability and maturity do have some implications for task performance, given the SCAT involved making judgments about the likelihood of couples remaining in relationships and was described as measuring both social and cognitive skills. Thus, these measures involve the same construct. In addition, these questions were answered in close temporal proximity and were not presented as distinctly different from one another. The major findings involving the significant interaction between self-esteem and condition, of course, still held in this analysis and provided support for the major hypothesis underlying this research.

6 Our affect measure involved participants' ratings of how characteristic each of four adjectives (two negatively valenced and two positively valenced ones) was of their present feelings. The measure was effective in that large differences were obtained across conditions. However, we asked participants to consider their present feelings in light of their performance and that of the other person. A question arises as to whether similar findings would be obtained if we did not give them this directive. In an earlier pilot study participants were asked to rate adjectives in terms of how characteristic they were of their present feelings without being asked to consider their feelings in light of their performance vis-à-vis the other performer. This study included a neutral control condition and one in which participants were compared to a mildly superior performer. As in the present study participants reported higher levels of negative affect and high self-evaluations relative to an inferior in the mildly superior condition than in the control condition. Rating how characteristic adjectives are to participants' feelings is a common measure of affect. However, it can be less sensitive than physiological measures in that participants may attempt to conceal their true feelings and behave in a socially appropriate manner. Thus, even though this measure was sensitive in the present study, a more subtle measure may be needed in situations in which participants' are attempting to mask their affective reactions.

7 Although not the primary question motivating this research, the design of Experiment 2 allowed us to assess how the target of rating (self and average other) was affected by social comparison information. Specifically, individuals can compensate for unfavorable social comparisons by elevating/accessing positive aspects about themselves and/or by creating/accessing negative aspects of others. The general findings from the analysis of self – other ratings indicated that only high self-esteem persons were able to elevate their self-ratings following an unfavorable social comparison whereas all participants were able to derogate the inferior “average other.”

Additional information

Notes on contributors

Todd McElroy

We thank Judy Allison and Teresa Hill for their valuable help in preparing this manuscript. In addition, we thank Ashleigh Haire, Will Fleeson, and Stephen Fischer, for their helpful suggestions.