Social Identity Theory contains two seemingly incompatible predictions regarding the relationship between self-esteem and ingroup bias. The first focuses on low self-esteem as motivation for bias, predicting that low self-esteem individuals exhibit more ingroup bias. The second posits that high self-esteem results from exhibiting bias, thus, high self-esteem individuals exhibit greater bias. A meta analysis examined the relationship between self-esteem and ingroup bias. Additionally, the project examined methodological issues such as the lack of consistency in measurement of self-esteem, artificial dichotomization of self-esteem scores, classification of individuals as low self-esteem, and theoretical considerations such as the use of different ingroup bias strategies and the role of social category salience. Thirty-four studies yielding 102 effect sizes from 6660 subjects were included in the analysis.Results indicated a consistent pattern whereby high self-esteem individuals exhibited more ingroup bias than did individuals with low self-esteem. However, this result was moderated by ingroup bias strategy. When using ingroup bias strategies that required ratings of ingroup superiority, high self-esteem individuals showed more ingroup bias than individuals with low self-esteem. However, when using "indirect" strategies, such as rating groups that the individual did not contribute to, differences between low and high self-esteem individuals were not found. This result leads to the conclusion that both groups exhibit ingroup bias; however, individuals with low self-esteem are limited in the types of bias they exhibit. The pattern of results held for all self-esteem measures except for the Collective Self-Esteem Scale (CSES). No differences between low and high self-esteem individuals were found when scores on the CSES defined self-esteem. These results may however be an artifact of interactions with social identity salience. Methodological shortcomings were found in the definition of low self-esteem. Individuals were most commonly classified as "low self-esteem" based on median splits of self-esteem scores. This strategy resulted in classification of some individuals as low self-esteem despite relatively high self-esteem scores. Results of the current analysis are interpreted as indicating that individual level phenomena such as self-esteem may be predictive of group level behaviors. Implications for social identity theories are discussed.
Manipulated taking the agent versus the recipient perspective seems not to affect the relationship between agency-communion and self-esteem: A small-scale meta-analysis
