On the Simple Inverse Sampling with Replacement

In this paper we derive some unbiased estimators of the population mean under simple inverse sampling with replacement, using the class of Hansen-Hurwitz and Horvitz-Thompson type estimators and the post-stratification approach. We also compare the efficiency of resulting estimators together with Murthy's estimator. We show that in despite of general belief, the strategy consisting of inverse sampling with Murthy's estimator is highly less efficient when the target population is rare, whereas it can be more efficient when subpopulation means are closed. In fact, for inverse sampling to be highly efficient design one should know the population structure and then use an appropriate estimator.