NEW PARTIALLY SYSTEMATIC SAMPLING
Ching‐Ho Leu, Kam‐Wah Tsui · 1996
When the sample mean, y, is used to estimate the mean of a finite popula- tion, the usual systematic sampling procedure cannot provide an unbiased estimator of V (y), the variance ofy. In this paper, we propose a new partially systematic sampling procedure which provides an unbiased estimator of V (y). Moreover, the population size N is not required to be a multiple of the sample size used. We also compare the efficiency of this new sampling procedure with simple random sam- pling, the usual or circular systematic sampling procedures, and systematic sam- pling with multiple starts for populations exhibiting certain characteristics, such as random, linear, or autocorrelated trends, or periodic variation.