Many researchers overlook the importance of testing for potentially
spurious relationships among meta-analytic moderators
(Lipsey, 2003). Lipsey (2003) noted that confounding can occur if
moderator variables are related to each other and to effect size
estimates in meta-analytic research. We found two instances of
confounding that may influence how the primary findings of this
study are interpreted. The first involved the overlap between
studies examining senior leaders and those conducted in education
settings (K 29; 79%). Future research should examine the extent
to which the female gender role and the leader role align in
educational settings with leaders of different levels in order to
tease apart the influences of these two moderators. The second
important example of confounding in this study involved the large
amount of studies that examined lower level supervisors, between
1962 and 1995 (73%). Thus, it is difficult to know whether the
time period was the most important factor in men rating themselves
as more effective than women rated themselves, or whether
being in lower level positions was the more important moderator.
Future research should examine the feminine and masculine behaviors
needed for success at different hierarchical levels in order
to better understand how level moderates gender differences in
leadership effectiveness.