Abstract: In many social comparisons, people know exactly how they and others do. These comparisons induce a self-other gap. A variety of important decisions are made on the basis of judgments of the gap between ourselves and other people. Existing research indicates biased judgments of self-other gaps, with unknown absolute performance of others. However, the question we are interested in is whether judgments of a self-other gap will be accurate when both absolute performance of oneself and others are specified. This research investigated how the self-other gap was shaped by absolute and relative performances. We proposed the generalization effect, in which individuals generalized their absolute performance to rate their relative position to others though the actual self-other gap was specified.
We conducted seven studies (N = 2766) to test our proposed generalization effect on perceived self-other gap. Study 1 adopted a 2 (absolute performance: gain or loss) × 2 (relative performance: gain or loss) between-subjects design. The participants, who were informed their performance as well as their classmate’s performance in a test, rated the gap between themselves and the classmate. The result indicated that absolute gain caused a larger perceived self-other gap for relative gain (“I am far ahead of her”) than for relative loss (“I am not far behind her”). Conversely, absolute loss caused a larger perceived self-other gap for relative loss (“I am far behind her”) than for relative gain (“I am not far behind her”).
Studies 2 and 3 replicated the results in Study 1 with investment and social media scenarios. Besides, Study 2a excluded the influence of information order and Study 2b excluded the effect of emotion. Studies 3a and 3b ruled out the alternative explanations of numeric size.
Study 4 tested the association mechanism by cutting off the associations between multiple dimensions. We adopted a 2 (association: cutting-off or control) × 2 (absolute performance: gain or loss) × 2 (relative performance: gain or loss) between-subjects design. In the cutting-off condition, we designed a debiasing intervention where general associations among multiple dimensions were cut off. As a result, the effect found in Studies 1 to 3 persisted in the control condition but disappeared in the cutting-off condition where associations among multiple dimensions were cut off. The result indicated that generalization among dimensions accounted for the effect we found. The result also ruled out the explanations of egocentrism and focalism.
Study 5 manipulated the reference point in social comparison and found a null effect for reference point on the generalization effect, which ruled out the explanation of reference point.
We reveal that assessments of relative performance are biased even when people have sufficient information about their own and others’ absolute performances because people generalize their absolute performance to relative performance. The generalization effect reflects the overgeneralization bias in social comparison. People fail to realize that absolute performances are not necessarily related to relative performances. Moreover, the current research offers a feasible approach to reduce such a bias.

Key words: self-other difference, social comparison, overgeneralization, judgment and decision making