Optimal Stratification of Sampling Frames in a Multivariate and Multidomain Sample Design

Abstract

So far, in stratified sampling the problem of determining the optimal size and allocation of units in strata has been solved considering the stratification of population as given; and, conversely, the definition of an optimal stratification has been investigated without considering the optimisation problem of sampling size and allocation. An interesting proposal has been advanced in the recent past, concerning the joint solution of both problems: it is based on a tree search in the space of possible strata configurations, solving for each visited node the corresponding multivariate allocation problem by applying the Bethel-Chromy algorithm. Here we propose a non deterministic evolutionary approach, making use of the genetic algorithm paradigm. Having implemented both methods, we applied them to a real sampling survey (the Italian Farm Structure Survey), and evaluated the obtained results, by comparing them to the ones related to solution adopted in 2003 survey.