Sequential Estimation in Binary Response Models with Fixed and Random Covariates

摘要

This paper discusses sequential maximum-likelihood estimation in binary response models arising in medical applications. Using earlier results by the same authors, it is shown that a properly chosen sequential maximum-likelihood estimate can be utilized for setting up a fixed-size multivariate confidence region for the unknown model parameters. Constructed sequential procedures are shown to closely approximate the optimal fixed-sample procedure in terms of coverage probability and average number of observations. The obtained results are applied to estimation problems in binary response models arising in prospective and retrospective studies.