An approximately optimal non-parametric procedure for analyzing exchangeable binary data with random cluster sizes

For the exchangeable binary data with random cluster sizes, we develop an approximately optimal non-parametric procedure for obtaining estimates of the moments of all orders. Moreover, based on this procedure, we can also obtain efficient estimates of underlying parameters of moments of all orders. An application is made to data sets from a developmental toxicity study. Simulation results show that our procedure is valid and performs better than Bowman and George’s procedure and the pairwise likelihood procedure. Copyright Springer-Verlag Berlin Heidelberg 2013