A Generalised Measure of Disparity for Dichotomous Populations

Many populations are intrinsically dichotomous and many others are meaningfully dichotomized. For gauging the level of disparity between two sections with respect to a certain trait, many disparity indices have been devised. They are all based on aggregate sectional rates. They all err ignoring the relative size of the two sections, thus tacitly assuming that the two sections an of equal size. This paper devises a new positive disparity measure through a simple geometrical device and generalises it for the situation of unequal size sections. The measures are applied to female-male and rural-urban dichotomies in the case of literacy for the States of India. It is found that a disparity measure that ignores the issue of relative size, grossly underestimates/overestimates the magnitude of disparity.