Average effect size:-. were calculated acros~ all the ~tudies ami for subgroups of studies by study design and by category of dependent variable !Screening. Risk Reduction and Adherence to Medical Regimen). Each -.ubgroup was tested for homogeneity. For co~ts. -.everal of the studie!. !(lund positive correlations with their respective dependent variabks. As it was possible that the wording of their items was reversed (as several indicated). we treated the effect siLe for ea within the same study had opposite -.igns. these were not changed. Thu .... the mean effect siLcs for perceived cost" were possibly intlated. and the homogeneity test statistic reduced. making heterogeneity more liJ..cly. Table I gives the effect si7es used to represent each study.