Cross‐categorization effects in intergroup discrimination: Adding versus averaging

Abstract

In two between‐participants experiments (Ns = 216 and 260), Chinese in Singapore rated competence and attractiveness of strangers described (a) by cross‐categorizations of race (Malay: out‐group vs. Chinese: in‐group) and nationality (Malaysian: out‐group vs. Singaporean: in‐group), (b) by simple categorization of either race or nationality, and (c) by no‐group label. An additive effect of cross‐categorizations was obtained: Both the race and nationality categorizations produced main effects but no interaction effect. Analyses of variance of the simple and cross‐categorization data together also retained the additive effect as predicted by the adding rule. However, the higher ratings of the in‐group than of the out‐group in both the simple and cross‐categorization conditions were based on an out‐group derogation instead of an in‐group bias. Explanations of the out‐group derogation and suggestions for additional work on the cross‐categorization effects are discussed.

The crossed-categorization hypothesis: Evidence of reductions in the strength of categorization, but not intergroup bias

“…These revealed that the double ingroup versus double outgroup contrast (II versus OO) was consistently stronger than either the double ingroup versus partial outgroups contrast (II versus P) or the partial outgroups versus double outgroup contrast (P versus OO). Urban and Miller (1998;see also Hewstone et al, 1993;Singh et. al., 1997;Vanbeselaere, 1991) labeled this pattern of results an ''additivity'' model and concluded that most research supports that model.…”

“…The second question, which has been the focus of much contemporary work, involves only the crossed contexts and introduces a moderation question (see discussion below). This question asks whether bias is reduced when outgroups on one dimension are ingroup rather than outgroup members on the second dimension (e.g., Brewer, Ho, Lee, & Miller, 1987;Hagendoorn & Henke, 1991;Hewstone, Islam, & Judd, 1993;Singh, Yeoh, Lim, & Kim, 1997;Urban & Miller, 1998).…”

Crossed categorization versus simple categorization and intergroup evaluations: a meta‐analysis

“…These a priori F-tests were then transformed into standard indicators of signi®cance levels and effect sizes and subjected to meta-analytic integration (see Mullen, 1989;Mullen & Rosenthal, 1985;Rosenthal, 1991). These selection criteria rendered a total of 8 papers (Brewer, Ho, Lee & Miller, 1987;Eurich-Fulcer & Scho®eld, 1995;Hagendoorn & Henke, 1991;Hewstone et al, 1993;Migdal & Mullen,`Crossed categorization and evaluative complexity', unpublished manuscript, 1993;Singh, Yeoh, Lim & Lim, 1997;Vanbeselaere, 1987Vanbeselaere, , 1991. These eight papers yielded 16 sets of tests of the various de®nitions of ingroup bias in crossed categorization, representing the responses of 1250 participants.…”

Similarity and Liking Effects on Interpersonal Attraction: Test of the Two-Dimensional Trust-Respect Model

“…1 can also be explained by an alternative integration rule of adding (Singh, Yeoh, et al, 1997) instead of averaging with constant inferences about the missing liking information. The adding rule does not constrain the relative weights to sum to 1 (Anderson, 1981), and thus it predicts uniform effects of information presented alone and that presented with another piece of information.…”

“…1. Such methodological caution came from Singh, Yeoh, Lim, and Lim (1997) who advocated that the sample size should be larger in a single-category design than in the crossed-category design.…”