An Alternative Procedure for Subgroup Analyses in the Think Manager–Think Male Paradigm

Abstract. Studies in the think manager–think male paradigm ( Schein, 1973 ) routinely observe gender–leader similarity variation across participant subgroups (e.g., men and women). The traditional use of intraclass correlation coefficients (ICCs) to measure similarity hinders researchers’ ability to fully understand the nature of these relationships. This article advocates for a regression framework ( Edwards, 1995 ), which avoids problems associated with ICCs. A think manager–think male study based on rating content from Fischbach et al. (2015) was conducted to demonstrate the advantages of this procedure. The results were then compared against the traditional ICC analysis. Overall, it is argued that a regression framework is more useful in testing hypotheses about the relationship between participant characteristics and gender–leader similarity than ICCs.