Paper Summary
Share...

Direct link:

Comparing Multivariate Meta-Analysis Methods for Estimating Between-Study Variance

Fri, April 14, 11:40am to 1:10pm CDT (11:40am to 1:10pm CDT), Radisson Blu Aqua Hotel, Chicago, Floor: 2nd Floor, Caribbean

Abstract

We used a simulation to compare the performance of three methods for estimating the between-study variance of a multivariate meta-analysis: the robust variance estimation (RVE) method; the Riley method; and a three-level model. First, we found that the RVE method yields similar results when estimated using the metafor and robumeta R packages. Second, we found that the RVE method is less biased than the Riley method and the three-level model in most conditions. Moreover, as between-study variance increased, under-estimation by the RVE method became more severe.

Authors