Meta‐analysis of Personality‐job Performance Relations: A Reply to Ones, Mount, Barrick, and Hunter (1994)

Tett, Jackson, and Rothstein's (1991) meta‐analysis identified higher average correlations for personality in predicting job proficiency than did Barrick and Mount's (1991). Ones et al. suggest that discrepancies between the two studies involving the Big Five personality dimensions are due to certain procedural differences. In this reply, we show that their arguments do not adequately explain the noted discrepancies. We also show that, because personality traits correlate significantly with job performance both positively and negatively beyond chance levels, use of absolute values, contrary to Ones et al., is important in meta‐analyses involving personality. Addressing all of Ones et al.3 statistical concerns, re‐analysis of Tett et al.k main data set results in slightly lower mean validities (e.g., .24 vs.29 for fully corrected values based on confirmatory estimates), and renders non‐significant the job analysis/no job analysis distinction found to be significant in the original study. Tett et al.'s main conclusions, however, remain unchanged. We suggest that Barrick and Mount's lower mean validities may be due to their averaging signed correlations, pooling exploratory and confirmatory findings, and to the use of different inclusion criteria for selecting source studies.