The empirical review of meta-analysis published in Korea

Abstract

Meta-analysis is a statistical method that is increasingly utilized to combine and compare the results of previous primary studies. However, because of the lack of comprehensive guidelines for how to use meta-analysis, many meta-analysis studies have failed to consider important aspects, such as statistical programs, power analysis, publication bias, model selection, test of heterogeneity, and identification of heterogeneity. Therefore, the current study reviewed 84 meta-analysis studies conducted in Korea to examine proper application of the six categories named above. With regard to the issue of effect sizes, it was found that most of the meta-analysis studies obtained more than ten effect sizes, which seem to be an adequate number for representing an issue. However, many studies failed to consider the other issues: power analysis, publication bias, model selection, test of heterogeneity, and identification of heterogeneity.