Date created: | Last Updated:

Category: Project

Description: Although meta-analysis has been widely used in psychology and education, there is an increasing concern about its validity due to selective reporting of primary study results, which causes biased meta-analytic estimates. Although a broad array of methods has been proposed for addressing selection bias, there remains a lack of statistical development for adjustment methods in meta-analysis with multiple, dependent effect sizes. This dissertation examines novel adaptations of selective reporting adjustment methods in three studies. The first study adapts several adjustment methods based on robust variance estimation method with a multivariate working model and weighting scheme that correct for selective reporting bias while handling dependencies among effect sizes. The second study explores several potential selection mechanisms in which suppression occurs at the outcome level or at the study level. In addition, the second study evaluates the performance of promising novel adjustment methods proposed in the first study under different selection mechanisms. The third study provides a tutorial in the R statistical environment on the implementation of these novel adaptations. Taken as a whole, the research established in this dissertation contributes to the field by developing new methods to address selection bias in dependent effect sizes context, highlighting future research directions for methodological development in selective reporting correction, and providing tutorial to meta-analysts on applying the proposed adjustment methods.