Nearly all articles included more than one measure of age-based attitudes. In order of make use of all available information, we did not exclude any measures from included studies for which these data were available. However, to maximize both assumptions of independence and the utility of available data, we employed a shifting unit of analysis approach to effect size computation (Cooper, 2010). This technique computes overall effects by averaging within-study effects, but also incorporates all individual effect sizes—even if they came from the same study—in computing moderator analyses. (This was particularly useful to test the moderating impact of country, given that many articles used samples from multiple countries within the same broad—i.e., East or West—region.) In combining this level of analysis with the omnibus analysis described earlier, the shifting unit of analysis approach is an accepted compromise between maximizing available information from each study and maintaining assumptions of independence to the extent possible