U-estimability under sequential binomial sampling (Q803697)

Bernoulli random variables are sampled up to some well-defined stopping time. The stopping rule is supposed to depend only on the number of successes and the number of failures. Let S be the corresponding sample space, and let (a,b) be a point lying on some path (trajectory described by the number of successes and failures in S). The results of this paper deal with the existence of unbiased estimators for monomials of the form \(p^ aq^ b\), where p denotes the success probability and \(q=1-p\). These results are complementary to those by \textit{E. L. Lehmann} [Theory of point estimation (1983; Zbl 0522.62020)]. The main theorem is a necessary and sufficient criterion for monomials of the above form to have unbiased estimators.