Diagonal Systematic Sampling with Two Random Starts

Abstract

Linear Systematic Sampling, one of the oldest sampling schemes using label distances in the sampling process is known to perform well for populations exhibiting specific patterns with respect to the variable under study. As a competitor to Linear Systematic Sampling, Subramani(2000) proposed Diagonal Systematic Sampling and proved it is more efficient in terms of variance when compared to Linear Systematic Sampling in the presence of linear trend. In this paper, an extension of diagonal systematic sampling has been considered using two random starts instead of one random start as done in the case of original diagonal systematic sampling and the performance of the usual expansion estimator has been compared for the proposed sampling scheme with two random starts. Further, Yates corrected estimator is also developed for the proposed method and a detailed numerical study has been carried out using super population models suitable for populations with linear trend with reference to Gautschi’s (1957) Linear Systematic Sampling (LSS) with two random starts. Outcomes of the comparative study indicate that the proposed method performs better than Linear Systematic Sampling with two random starts for populations having linear trend.