|
Abstract:
|
In random effects meta-analysis, estimation of the heterogeneity variance followed by inference on the overall treatment effect parameter is an outstanding problem that has been addressed over the years without consensus. This study focuses on methods for estimating the heterogeneity variance in the random effects model. A detailed simulation study is performed to compare eighteen heterogeneity variance estimators with respect to bias and mean square error for normal and binary responses. Measures of effect size considered include mean differences, probability differences, and log odds ratios. AN example from the extant literature is used to illustrate the application of the procedures.
|
