Limitations and Directions for Future Research
Although the number of studies included in this meta- analysis
is relatively large (K 99), our use of moderator analyses
led to relatively small numbers and low power for some analyses.
For instance, we found that the analysis of time periods
was not statistically significant; however, the pattern of the
effects did show support for the hypothesis proposed by RCT.
We found that men were considered slightly more effective
leaders in older studies, and women were favored in studies
conducted more recently. This is consistent with a recent metaanalysis
that found that leadership definitions were seen as less
masculine over time (Koenig et al., 2011). A post hoc power
analysis was conducted to determine if a low sample size (99
studies) could explain the nonsignificance of the effect. Consistent
with this idea, the observed power was only .35. Nonetheless,
Hunter and Schmidt (2004) have indicated that although
a small number of studies in a meta-analysis may
increase the likelihood of secondary sampling error, if the
overall sample size (N) across those studies is fairly large, this
should reduce the second order sampling error. Fortunately, in
our meta-analysis, even the analyses with a low number of
studies include fairly large overall sample sizes.