Answers are grounded in this paper and cite their sources.
This article presents the results of a simulation study of the estimation of a common odds ratio in multiple-contingency tables when the data are correlated within clusters. The correlation is modeled by the beta-binomial distribution. The three objectives of the study were to determine (a) the variances, biases, and relative efficiencies of the beta-binomial maximum likelihood estimator (MLE), the binomial MLE, and the Mantel-Haenszel estimator of the common odds ratio; (b) the variances and biases of two estimators of the intracluster correlation coefficient, ρ; and (c) the effectiveness of adjusting variance estimates of the binomial and Mantel-Haenszel estimators using the two estimators of ρ. The simulation shows that increasing the size of the intracluster correlation coefficient increases the variance and the positive bias of all three estimators of the odds ratio. The Mantel-Haenszel estimator displays less variance and bias than the MLE except for large values of Ψ and ρ. An unexpected but important observation, however, is that applying a cluster-inflation factor to the formula for the large-sample variance of the Mantel-Haenszel estimator often results in an overestimate of the variance. The only circumstances in which the unadjusted variance is an underestimate of the observed variance occur when the clustering is high and the odds ratio is unity or when the contingency tables represent a wide band of success probabilities. This observation, coupled with the fact that the Mantel-Haenszel estimator is expressed as a closed formula, leads to the recommendation of the large-sample variance formula as a conservative estimator in most cases. A comparison of the moment and analysis of variance (ANOVA) estimators of intracluster correlation coefficient ρ shows that the ANOVA estimator is consistently positively biased, but that both its bias and variance are less than those of the moment estimator, especially when ρ is small. The moment estimator is extremely positively biased when the correlation is low, and negatively biased when it is high.
No takes yet. Share an insight, caveat, or question.
Donald et al. (1990) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: