Abstract:
|
Meta-analyses often include studies that report multiple effect sizes based on a common pool of subjects, inducing dependence among the effect size estimates. Robust variance estimation (RVE) provides a method for pooling dependent effects, even when information on the exact dependence structure is not available. When the number of studies is small or moderate, however, test statistics and confidence intervals based on RVE can have inflated Type I error. This paper describes and investigates several small-sample adjustments to F-statistics based on RVE, including both multi- and single- parameter (i.e., t-tests) tests. Simulation results demonstrate that one such test, which approximates the test statistic using Hotelling's T^2 distribution, is level-alpha and has Type I error closer to nominal than the others. An empirical application demonstrates how results based on this test compare to the large-sample F- and t-tests.
|