Stratified sampling. We want to estimate the mean J1. = E(X) of a population that has been divided (stratified) into s mutually exclusive parts (strata) (e.g., geographic locations or age groups). Within the jth stratum we have a sample of i.i.d. random variables and a stratum sample mean We assume that the s samples from different strata are independent. Suppose that the jth stratum has 100pj% of the population and that the jth stratum population mean and variances are and Let and consider the two estimators
where we assume that
(a) Compute the biases, variances. and MSEs of How should be chosen to make unbiased?
{b) Neyman allocation. Assume that are known (estimates will be used in a later chapter). Show that the strata sample sizes that minimize are given by
(c) Show that with minus MSE with nk given by (1.7.3) is