Personality-performance correlations at work: Individual and aggregate levels of analyses

In the occupational community, there is a widespread faith in the utility of personality assessment for selection, development, etc. This faith has been immune to arguments, supported by empirical evidence, regarding the poor correlation between personality and performance in the workplace (these correlations rarely exceed the 0.2–0.3 level). The difference between perception of utility and the actual empirical reality is large. We investigated one possible source of this perceived-actual discrepancy. In two separate samples, we compared the magnitude of validity coefficients from individual and aggregate (i.e. organizational) levels. Our results indicated that strong actual personality-performance correlations exist at the aggregate level of analysis, but not at the individual level of analysis. We suggest that this aggregate-individual correlation discrepancy may, in part at least, account for the perceived-actual discrepancy noted above. We conclude that the continued faith in personality testing in the workplace may be a consequence of test usersʹ sensitivity to actual aggregate level personality-performance correlations. However, we warn of the danger of drawing inferences from aggregate level correlations when making decisions about individuals, and point out the statistical artefacts that may account for some of the magnitude increase in aggregate level correlations. Several foci for further research are indicated.