Taipei : Graduate Institute of Management Science, Tamkang University
An attempt will be made here to extend Dalenius'(1950) theory of univariate stratification for one population to multiple population. If k mutiple populations have the same or different domains, then they can be divided simultaneously into L strata according to the variable y under study. The gross population is composed of the k populations. The ith stratified simple random sample of size $n_i$ ia drawn from the ith population,i= 1, 2,...,k. The gross population mean is estimated by the k sample means. The variance of the gross sample mean will be taken as a measure. We will call a system of stratification, if it minimizes the variance of the sample mean. This paper will discuss the method for finding optimum stratification points for multiple populations under Neyman allocation. A numerical illustration is also given.
International journal of information and management Sciences 1(2), pp.1-8