Abstract
We study the possibility of using the Horvitz-Thompson (HT) estimator from a sample collected by the Rao-Hartley-Cochran (RHC) scheme. We derive expressions for several alternative unbiased variance estimators of the estimators given by RHC and HT for estimating a population total. For a real life problem, on finding well correlated auxiliary variate values, we consider selecting the first stage units by the RHC scheme and the second stage units by equal probabilities without replacement. Using real data, we compare the relative accuracies in estimation by dint of simulation, using the classical and the new variance estimators. Our criteria for comparison are the actual coverage percentage (ACP) and the average coefficient of variation (ACV). We demonstrate that as against the RHC estimator, the HT estimator is a viable contender in estimation from an RHC sample.
AMS (2000) Subject Classification : 62D05.
Get full access to this article
View all access options for this article.