Maximum Likelihood Estimation of the Risk Ratio in Case-Cohort Studies

Abstract

In this paper, maximum likelihood estimation of the risk ration in case cohort studies is considered. In the crude analysis it gives the closed-form estimator of the risk ratio and the large-sample variance estimator of its logarithm. Although the previously proposed crude rsik ratio estimator generally retains high efficiency, its asymptotic efficiency relative to the new estimator falls below .9 in some parameter combinations. In the stratified analysis, maximum likelihood estimation of the common risk ratio and the standardization methods based on stratum-specific maximum likelihood estimators are given.