Rather than an estimate of the population mean or total sometimes an estimate of the mean or total of a subpopulation or domain is desired. For such problems the number of units in the sample that fall into the domain is a random variable. This causes some complications for the usual frequentist approach. Here we give a simple and coherent noninformative Bayesian approach to domain estimation.
- Domain estimation
- Polya posterior and noninformative Bayes
- Sample survey