This study has several limitations. First, the set of variables included
in this meta-analysis is not exhaustive. There are certainly other relevant
variables that were omitted for lack of sufficient numbers of studies.
However, by including a total of 39 correlates, this study does broaden the
existing research focus beyond interview, performance appraisal, and promotion
ratings. Second, there were relatively few studies included in our
meta-analysis that primarily sampled older workers approaching retirement
age. For instance, even though more than half of employees sampled
852 PERSONNEL PSYCHOLOGY
here would be defined as old using the ADEA definition, only 7% of
employees were between 50 and 62 years old. Therefore, future research
should focus more on workers at the higher end of the age distribution
(e.g., ages 60–65) who are more vulnerable to negative age stereotypes
than employees at the low end of the older worker age distribution (e.g.,
ages 40–45). Third, some of our results are based on only a fewstudies, and
the results there are suggestive rather than conclusive. At the same time,
the consistency of findings observed across most relationships examined
suggests that our results are theoretically meaningful. Fourth, although
our meta-analysis examined the relationship of age with a wide range of
correlates that capture the six study stereotypes, we did not have direct
measures of these six stereotypes per se. Last here, we could not address
moderators (besides average age of the sample) in this meta-analysis.
Without large enough sets of studies, a meta-analysis cannot systematically
look at moderators in ways that yield highly stable results. Future
research needs to address moderating effects in much more depth.