(Received June 15, 1994; revised March 3, 1995)
Abstract. A simple random sample is drawn over a finite population which is composed of several subpopulations. Each subpopulation consists several domains. The minimax estimator under squared error loss function for the domain totals over a subpopulation is derived, in which the number of sample units falling into the subpopulation is random.
Key words and phrases: Random sample size, subpopulation, domain totals.