SAMPLE SIZE FOR CONTINUOUS SUPERPOPULATIONS

Abstract

The objective of this research was to determine the sample size without using the estimation of the finite population?s unknown parameters by preliminary sampling and normality assumption. The method was based on KolmogorovSmirnov?s distance between the empirical distribution functions of both population and sample. Unequal order statistics assumption is necessary. For N£100 the distribution of the distance was obtained; for N>100 the asymptotic approximation of the distribution of the distance between the empirical distribution of two samples was used. For a given distance d, it was determined the sample size that yields a sample with distance smaller than d with probability 1-a. For N£100, d=0.22 can be used, and for N>100 a table with values of d for a sampling proportion of 0.05, 0.10, 0.15 and 0.20 with probability 0.95 is given.