The Detection of Partial Association, I: The 2 × 2 Case
M. W. Birch · Journal of the Royal Statistical Society Series B (Statistical Methodology) · 1964
Summary A criterion is derived for testing null hypotheses of conditional independence of two dichotomous random variables. This test has optimum properties when the alternative hypothesis is that the association of the two random variables in the conditional distribution is, in a certain sense, constant.