2010 Kenny SimilarityAgreementSelfOther
Kenny and West (2010) conduct a comprehensive meta‑analysis of 24 round‑robin studies (118 variables, 2,992 participants) to examine whether self‑perception differs fundamentally from other‑perception. They organize person‑perception into two orthogonal processes: agreement (self‑other agreement vs. consensus) and similarity (assumed similarity vs. assimilation). Using the Social Relations Model (SRM) they derive four observable indices (self‑other agreement, consensus, assumed similarity, assimilation) and two novel regression‑based parameters (k for assumed similarity, q for self‑other agreement). They test a suite of four moderators – visibility, evaluativeness, familiarity (acquaintance), and group size – and their interactions, first via hierarchical regressions and then with a structural‑equation‑model (SEM) that treats the moderators as predicting a latent “agreement” or “similarity” construct. The meta‑analytic results show that (1) consensus and assimilation are on average larger than self‑other agreement and assumed similarity, respectively; (2) self‑other agreement correlates moderately with consensus (r = .61) while assumed similarity correlates modestly with assimilation (r = .40), and self‑other agreement and assumed similarity are essentially unrelated (r = .03); (3) moderators affect the four processes similarly, with only two notable differences: visibility × group‑size interacts for self‑other agreement but not consensus, and group size has a stronger negative effect on assumed similarity than on assimilation. The SRM parameters k and q correlate strongly with their traditional correlational counterparts but are less contaminated by consensus and assimilation, and only group size predicts k (negative) while no moderator predicts q. The authors conclude that self‑ and other‑perception share many underlying determinants, supporting Bem’s (1972) claim of similarity, yet systematic differences (lower mean levels, specific moderator interactions) indicate that the self is not identical to an “other” perceiver. They discuss theoretical implications for accuracy vs. bias, the role of familiarity, and limitations such as reliance on college samples, lack of close‑relationship data, and exclusion of content‑specific moderators.
Topics
Framework of Agreement vs. Similarity in Person Perception
Defines the four cells of the perception matrix: self‑other agreement, consensus, assumed similarity, and assimilation, and explains how they map onto bias vs. accuracy debates.
Meta‑Analysis of 24 Round‑Robin Studies
Describes the inclusion criteria, sample characteristics (24 studies, 118 variables, 2,992 participants), and the extraction of the four SRM‑based indices.
Moderators of Agreement and Similarity
Presents visibility, evaluativeness, familiarity (acquaintance), and group size as the four primary predictors, along with their main effects and higher‑order interactions.
Structural Equation Modeling for Construct Validity
Details the SEM that tests whether the pattern of moderator effects is identical for self‑other agreement vs. consensus (agreement construct) and for assumed similarity vs. assimilation (similarity construct).
Round‑Robin Design and the Social Relations Model (SRM)
Explains how round‑robin data allow decomposition of variance into perceiver, target, and relationship effects, and how these feed into the four perception indices.
Empirical Findings: Mean Levels, Correlations, and Moderator Effects
Summarizes key results: consensus > self‑other agreement, assimilation > assumed similarity; moderate correlation between agreement indicators, low between similarity indicators; specific moderator differences (visibility×group‑size, group‑size on similarity).