Authors: Nuanpan Lawson
Abstract: Estimating variance for a population total or population mean under nonresponse requires a known or estimated response probability. In this paper an alternative method for estimating variance in a population total, in the presence of nonresponse, under probability proportional to size sampling is proposed. The Taylor linearization procedure has been used to transform the proposed estimator under reverse framework. We can show in theory that the proposed estimator is approximately unbiased and does not require either known or estimated response probability when a sampling fraction is negligible.
Keywords: probability proportional to size sampling; Taylor linearization procedure; reverse framework; response probability