Simulations for the Q statistic with constant and inverse variance weights for binary effect measures
Abstract
Cochran's statistic is routinely used for testing heterogeneity in meta-analysis. Its expected value (under an incorrect null distribution) is part of several popular estimators of the between-study variance, . Those applications generally do not account for the studies' use of estimated variances in the inverse-variance weights that define (more explicitly, ). Importantly, those weights make approximating the distribution of rather complicated. As an alternative, we are investigating a statistic, , whose constant weights use only the studies' arm-level sample sizes. For log-odds-ratio, log-relative-risk, and risk difference as the measure of effect, these simulations study approximations to the distributions of and , as the basis for tests of heterogeneity. We present the results in 132 Figures, 153 pages in total.
Cite
@article{arxiv.2206.08907,
title = {Simulations for the Q statistic with constant and inverse variance weights for binary effect measures},
author = {Elena Kulinskaya and David C. Hoaglin},
journal= {arXiv preprint arXiv:2206.08907},
year = {2023}
}
Comments
Version 2 includes new Appendix C on empirical power of Q for log-odds-ratio