I am reading "An Introduction to Categorical Data Analysis, Second Edition" by Alan Agresti. Data I am trying to fit is presented in Table 7.3 in the book. The log-linear regression by PROC GENMOD is in Table 7.6. I fitted the same model by PROC CATMOD and the chi-sqaure statistic is different for the main effect term but the same for the interaction term. The estimates and SEs are all different. It seems to me that PROC GENMOD is just fitting a log linear model with intercept, but what model PROC CATMOD is fitting here? And why some of the results are the same while the others are not?