Higher-Order Asymptotics for Random Effects Meta-Analysis: An Empirical Evaluation

• Activity Number: 656
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Joseph Beyene*+
• Companies: McMaster University
• Keywords: meta-analysis ; higher-order asymptotics ; random effects model

Abstract

A random-effects modeling approach is often used to dealing with unexplained heterogeneity in meta-analysis but may not perform well with small number of studies. We applied higher-order asymptotic methods to a series of meta-analyses using data extracted from the Cochrane Library. We focused on continuous outcomes and compared traditional methods with a second-order likelihood method based on Skovgaard's statistic. We investigated three effect measures (mean difference (MD), standardized mean difference (SMD), Ratio of Means (RoM)), and three methods of estimation for the heterogeneity parameter (DerSimonian-Laird (DL), Maximum Likelihood (ML), and Restricted Maximum Likelihood (REML)). A total of 66 meta-analyses were used in which the effect measure was MD. The largest average discrepancy in p-values was between the Skovgaard's method and a p-value based on Z test and ML estimator (mean difference = 0.05, SD of difference = 0.09). For SMD, 106 meta-analyses were available for analysis. Once again the largest discrepancy occurred between Skovgaard's and Z test with ML (mean difference = 0.03, SD of difference = 0.04). The results for ROM were similar to that of the SMD.