A Study on Multilevel Meta-Analysis Methods

Main purpose of this study is to examine different models that are used to analyze meta-analytic data sets. Particularly, it was aimed to introduce the multilevel models that have been proposed as one of the best ways to handle dependency problem. Firstly, traditional meta-analysis techniques were described then new approaches provided in meta-analysis were explained with the focus on multilevel meta-analysis models. Three different methods (traditional meta-analysis, two-level and three-level meta-analysis models) were used to analyze a meta-analysis data set that has been used in a previous study. This data set included effect sizes collected from a common phenomenon in special education which examines the relationship between creativity and one of the psychological disorders, psychoticism. Analyses with three different methods showed that results of two-level meta-analysis yielded the same results as traditional meta-analysis did. In addition, it was shown that how a three-level meta-analysis model can be used to solve dependency problem as an extension of two-level meta-analysis model. Necessary information about the new approaches developed in meta-analysis literature was provided for researchers in Turkey.