A note on exact calculation of the non central hypergeometric distribution |
| |
Authors: | Bruce Barrett |
| |
Institution: | Department of Information Systems, Statistics, and Management Science, University of Alabama, Tuscaloosa, AL, USA |
| |
Abstract: | Direct calculation of the non central hypergeometric (NH) distribution and its moments can present computational issues in both efficiency and accuracy. In response, several methods, both approximate and exact, for calculating the NH mean and variance have appeared in the literature. We add to this body of work, a straight-forward, exact method that is easily programed, efficient, and computationally stable. Specifically, by considering the logs of the values of the NH probability mass function (pmf) and then shifting the exponents so that, prior to normalization, the mode acquires a value of 1, concerns for overflow are eliminated. |
| |
Keywords: | 2 × 2 tables conditional inference non central hypergeometric |
|
|