Abstract
This article deals with the determination of compromise integer strata
sample sizes using goal programming in multivariate stratified sam-
pling. Firstly, the problem of determining optimum integer strata sam-
ple sizes is formulated for the univariate case, and then based on these
individual optimal solutions, individual goal variances are calculated. A
new compromise criteria is defined for the goal programming approach
based on predetermined or calculated goal variances. It is shown that
the proposed approach provides relatively more effcient and feasible
compromise integer strata sample sizes for multivariate surveys.