On regression technique and adaptive cluster sampling for estimation of population mean yield in case of clustered population

Abstract

Adaptive cluster sampling (ACS) provides an effective framework for including neighboring or related units while sampling rare and highly clustered populations. Introduced by Thompson, ACS is a reliable sampling technique under the assumption that the data are free from outliers. However, conventional ACS-based estimators can yield distorted results in the presence of outliers. Motivated by this limitation, the present study develops adaptive estimators within the ACS framework using ordinary least squares (OLS), Huber M, and Mallows GM estimation methods. The proposed regression-type estimators incorporate these functions to improve performance. For illustration, real-life numerical data of the Blue-winged teal has been utilized. The mean square error (MSE) properties of both the adapted and proposed estimators are derived and compared to assess their efficiency.