The Liking-Similarity Effect: Perceptions of Similarity as a Function of Liking

Abstract

In these two studies, we examined whether the inferences people make about likable and dislikable targets align with the predictions of balance theory. We hypothesized that people exhibit a liking-similarity effect by perceiving greater similarity with a likable person than a dislikable person. To test this hypothesis, we manipulated the likability of a target person and then assessed participants’ perceptions of similarity to that target person. In both studies, people rated likable others as more similar to themselves than dislikable others across a variety of domains (e.g., attitudes, personality characteristics, behaviors). In Study 2, individual differences in self-concept clarity, self-esteem, and preference for consistency moderated the liking-similarity effect.

Keywords:

• balance theory
• impression formation
• liking
• similarity

Notes

1. 1. We did not analyze the four profiles (political values, personality traits, attachment style, and behavioral intentions) separately for three reasons: (a) We hypothesized that people would generally rate likable others more similar to themselves than they would dislikable others, but did not have hypotheses that this would be stronger or weaker in any single domain. (b) Profile correlations are more robust to the extent that they are assessed across multiple items. Just as one can better trust a correlation observed in 40 participants than one observed in four participants, one can better trust a profile correlation observed across 40 items than one observed across four items. (c) Examining profile correlations across all items prevented the inflation of Type 1 error that would have resulted from examining four separate sets of items.

2. 2. As mentioned earlier, all generalized normative profiles are comprised of the average profile of one type (i.e., self, dislikable target, likable target), as such there is no variance in these correlations.

3. 3. When we split the sample in Study 2 in half and examine only the target that each participant evaluated first, our results replicate between subjects. Specifically, those who evaluated the likable individual projected both raw (M_z(34) = .74, SD_z(34) = .48) and distinctive (M_z(34) = .52, SD_z(34) = 1.79) profiles more than those who evaluated a dislikable individual projected their raw (M_z(34) = .07, SD_z(34) = .34) and distinctive (M_z(34) = –.16, SD_z(34) = .33) profiles, ts(202) > 3.78, ps < .001.

Additional information

Notes on contributors

Brian Collisson

Brian Collisson is an Assistant Professor in the Department of Psychology, Marian University.

Jennifer L. Howell

Jennifer L. Howell is a Doctoral Candidate in the Department of Psychology, University of Florida.