Mathematics · Statistics and Probability
Selection of Work Sampling Observation Times: Part II-Restricted Random Sampling
Joseph J. Moder, Henry D. Kahn
See on the mapAbstract
Abstract This paper considers the problem of sampling from a finite population defined on the interval (1, 2, …, T), using a restricted form of random sampling in which the spacing between consecutive observations must be at least some minimum value, M. For this type of restricted random sample, of size N, the value M can be arbitrarily specified in the range from zero, corresponding to simple random sampling with replacement, to a maximum of T/N, corresponding to systematic sampling. An efficient procedure is developed for generating a sample for specified values of N, M and T. It is then shown, by computer simulation, that values of M in the range from 60 to 90 percent of T/N retain most of the advantages of systematic sampling, while avoiding most of the disadvantages for some special processes., Also, the error variance associated with stratified sampling is comparable to restricted random sampling with a minimum spacing of about 70 percent, and both methods are almost as efficient as systematic sampling.