Inverse Binomial Sampling

Abstract

IN a recent communication discussing the application of the sampling method described by Haldane¹, Tweedie² states that the method was described as a technique for using hæmocytometers. This is not so. Hæmocytometers are designed for estimating the population density per unit volume, and the method under discussion is adapted for estimating the incidence of an attribute in a population consisting of members possessed of that attribute and members not possessed of it. This is an entirely different problem. The method, which has been in use in this Laboratory for a considerable time, and has proved of great value, was designed for estimating the incidence of certain abnormal forms of erythrocytes in blood films, this incidence often being as low as 0·2 per cent. By using this sampling technique, increased accuracy is obtained at these low incidence levels, and a saving of time and labour occurs at high incidence levels.