A total of 37 studies were initially identified as satisfying the inclusion criteria. From those, five studies were not included in the final pool because of inadequate data for the calculation of effect size. Overall, a total of62 effect sizes calculated from 32 studies were finally processed. The meta-analysis revealed a positive moderate effect size, which was significantly different from zero (effect size ¼ .48, 95% confidence intervals (CI) ¼ .38, .58). The percentage of variance accounted for by sampling error was 57%, showing considerable heterogeneity among the effect sizes from the different studies, thus justifying the investigation of the hypothesized moderators (Hunter & Schmidt, 1994). The fail-safe statistic indicated that it would be unlikely that a sufficient number of unpublished studies (K0 ¼ 102) would exist to reduce the effect to a trivial size (.15). The results of all analyses are summarized in Table 1.