Asymptotic normality and the bootstrap in stratified sampling (Q795425)

The asymptotic distributions of the estimates of a linear combination of stratum means in stratified sampling are considered. Two situations are discussed: (a) The populations are assumed arbitrary and the sampling is with replacement. (b) The populations are assumed finite and the sampling is without replacement. Both the number of strata and their size is arbitrary. Lindeberg conditions are shown to guarantee asymptotic normality and consistency of variance estimators. The same conditions also guarantee the validity of the bootstrap approximation for the distribution of the t-statistics. Via a bound on the Mallows distance, situations will be identified in which the bootstrap approximation works even though the normal approximation fails.