Partial Association Measures and an Application to Qualitative Regression
定义了两种偏关联度量，通过随机变量函数的无相关性刻画条件独立性，并推广了相关比分解，应用于定性变量回归。
We define partial ф 2 and additive partial ф 2 measures of association between two random variables and we characterize conditional independence by the lack of correlation between functions of the random variables. We study decompositions of ф 2 which generalize the correlation ratio decomposition. Application to regression on qualitative variables is an illustration.