Size of sampling unit in yield surveys

Abstract

In a previous communication1 I directed attention to the risk of obtaining biased estimates of crop yields with sampling units of small size. Further results confirming these findings are now available. Of particular interest are the results of the investigations on paddy (rice) in the Gaya district in Bihar and the Kistna district in Madras. The Gaya district has a geographical area of 4,766 sq. miles, of which approximately 35 per cent is under paddy. It is divided into four sub-divisions. The plan of sampling consisted in selecting at random 108 villages, 36 from the Gaya division and 24 from each of the other sub-divisions. In each selected village two paddy-growing fields were selected at random, and in each selected field the following five sampling units were marked: (a) one rectangle of size 33 ft.×16½ ft., area 544.5 sq. ft.; (b) two equilateral triangles of side 15 links, area 42.4 sq. ft.; and (c) two isosceles right-angled triangles with the equal sides equal to 5 ft. area 12.5 sq. ft.