COMPARISON OF RELATIVE EFFICIENCY FOR OPTIMAL STRATIFICATION IN THE AGRICULTURAL PRODUCERS TYPOLOGY

Abstract

The typology of producers is one of the most used techniques in the design of differentiated programs for the agricultural sector, its success depends on efficiency for optimal stratification. The objective of this study was to compare Dalenius and Hodges’ techniques and K-means in terms of relative efficiency for optimal stratification in the typology of agricultural producers in Mexico. The hypothesis was that the technique of Dalenius and Hodges is more efficient than that of K-means. Information from two surveys was analyzed and contrasted; one of them applied to malanga (Arum esculentum L.) producers and the other one to cattle farmers with dual-purpose cattle. The information was captured and analyzed in the Statistica v. 7 software. The variables analyzed were area, production, wages, and income from the sale of malanga or cattle. Three strata or groups were formed, and by main components were stratified with the first main component and the technique of Dalenius and Hodges, which was contrasted with the of K-means method, located in the module of cluster analysis, at the same Statistica interface. For the comparison of the relative efficiency of the variables, the area sown with malanga and size of the ranch (ha), the ratio of the variances of both techniques was used according to the relative efficiency formula. Results showed that both techniques coincided 87.5% in malanga producers and 91.6% in producers with cattle. However, the typology obtained by the Dalenius and Hodges technique is more efficient than that of the K-means method with cluster analysis because it minimizes the variance among strata. This situation is similar to what happens when stratified sampling is used, from which the Dalenius and Hodges technique is derived, aiming to study heterogeneous populations.