Three computer methods, namely the Random Draw, the Sequential and the Reservoir Methods, to obtain a simple random sample without replacement are reviewed. Several implementations of each method are given and criteria for the selection of an implementation are discussed. The most promising ones are compared for efficiency in their execution times. It is concluded that for small sampling fractions, Ernvall and Nevalainen's (1982) implementation of the Random Draw Method is most efficient when N, the population size, is known. When N is unknown, an Algorithm Z of Vitter (1985), an implementation of the Reservoir Method, is recommended. For larger sampling the algorithm of Bissell (1986) and Vitter (1984) is quite efficient.
Tijdschrift beschrijving