Application of Sampling to Agricultural Statistics with Emphasis on Stratified Samples

Abstract

DATA on agricultural items, in general, are extremely heterogeneous, and any efficient sampling program to obtain information about agriculture almost invariably requires stratification. This whole discussion could be devoted to the ways in which the theories of stratification are being utilized to increase the accuracy of agricultural data collected on a sample basis. We assume, however, that it would be of more interest to this group to have presented to it a specific sampling problem in agriculture in which the sampling theory of stratification has been applied. This will be done to (i) point out some of the problems encountered in applying the principles; (2) show how it was possible to circumvent some of these difficulties; (3) indicate what was gained by stratifying in this particular instance; (4) show how the gains were calculated. The Pre-Harvest Wheat Surveys have been selected as a basis for the discussion. There were two specific objectives of the surveys. The first was the development of a sampling technique that would, at harvest and prior to market-