Meta-Analytic Five-Factor Model Personality Intercorrelations: Eeny, Meeny, Miney, Moe, How, Which, Why, and Where to Go

Abstract

© 2020 American Psychological Association. Meta-analysis is frequently combined with multiple regression or path analysis to examine how the Big Five/Five-Factor Model (FFM) personality traits relate to work outcomes. A common approach in such studies is to construct a synthetic correlation matrix by combining new meta-analyses of FFM-criterion correlations with previously published meta-analytic FFM intercorrelations. Many meta-analytic FFM intercorrelation matrices exist in the literature, with 3 matrices being frequently used in industrialorganizational (I-O) psychology and related fields (i.e., Mount, Barrick, Scullen, & Rounds, 2005; Ones, 1993; van der Linden, te Nijenhuis, & Bakker, 2010). However, it is unknown how the choice of FFM matrix influences study conclusions, why we observe such differences in the matrices, and which matrix researchers and practitioners should use for their specific studies. We conducted 3 studies to answer these questions. In Study 1, we demonstrate that researchers' choice of FFM matrix can substantively alter conclusions from meta-analytic regressions or path analyses. In Study 2, we present a new meta-analysis of FFM intercorrelations using measures explicitly constructed around the FFM and based on employee samples. In Study 3, we systematically explore the sources of differences in FFM intercorrelations using second-order meta-analyses of 44 meta-analytic FFM matrices. We find that personality rating source (self vs. other) and inventory-specific substantive and methodological features are the primary moderators of meta-analytic FFM intercorrelations. Based on the findings from these studies, we provide a framework to guide future researchers in choosing a meta-analytic FFM matrix that is most appropriate for their specific studies, research questions, and contexts.