Approximating Discrete Distributions, with Applications
D. V. Gokhale · Journal of the American Statistical Association · 1973
This article presents a generalized procedure of finding the discrete distribution which minimizes, subject to a set of linear constraints on the probabilities, the “discrimination information” with respect to a given probability distribution. A convergent iterative algorithm is modified to serve this purpose. Many applications are discussed including analyses of contingency tables and some discrete analogues of the one-sample and several-samples problem. A test for the validity of the imposed constraints is provided by a test-statistic distributed asymptotically like a chi square.