Statistical Analysis for Meta-analysis

DOI Web Site
  • Shiroshita Akihiro
    Department of Respiratory Medicine, Ichinomiyanishi Hospital Scientific Research Works Peer Support Group
  • Kataoka Yuki
    Scientific Research Works Peer Support Group Department of Internal Medicine, Kyoto Min-iren Asukai Hospital

Bibliographic Information

Other Title
  • メタアナリシスの統計解析手法
  • メタアナリシス ノ トウケイ カイセキ シュホウ

Search this article

Description

<p>Meta-analysis is a statistical analysis that summarizes the findings. To compare the findings of various studies regardless of the sample size, we use effect size, which standardizes findings across studies, so that they can be compared independent of sample size. We weigh the effect sizes based on the accuracy (standard error) and combine them. There are two combining rules : the fixed effect model and the random-effects model. The fixed effect model assumes that one true population effect exists, while the random-effects model assumes variability in the population effects. Generally, the random-effects model is preferred because of the conservative results. Our article used a systematic review by Yang et al. to explain the common effect size indices (mean difference and standardized mean difference), combining rules (fixed effect model and random-effects model), and statistical heterogeneity.</p>

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top