Determination of an efficient sampling design for rare and clustered population using design based estimators

Abstract

A sampling design that provides estimates of population mean and abundance with small variance is important to researchers. Estimates that are accurate even with minimal sampling efforts allow researchers to easily and confidently investigate rare populations. In the determination of efficient sampling design for rare and clustered population, mean square errors have been applied in many previous research works. However, this method only captures the variability of the estimator and fails to capture their reliability. This study obtained the interval estimates based on the design based estimators, the HT and HH estimators. The study examined the behavior of the Horvitz Thompson (HT) and Hansen Hurwitz (HH) estimators under the ordinary adaptive cluster sampling design (ACS) and adaptive cluster sampling with data driven stopping rule (ACS’) design using artificial population that is designed to have all the characteristics of a rare and clustered population and another population that does not have those characteristics. The efficiency of HT and HH estimators were used to determine the most efficient design in estimation of population mean in rare and clustered population. The coverage probability confidence intervals of population mean based on HT estimators and the HH estimators were examined. Results of the simulated data show that the adaptive cluster sampling with stopping rule is the more efficient sampling design for estimation of rare and clustered population than ordinary adaptive cluster sampling.