In this lesson, we introduced the log-linear model and showed how it can be used to explain the associations among a set of two or more categorical variables. Although many of the same results as those from a logistic model can be obtained, the main advantage of the log-linear model is that by not treating anyone variable as the response, it can estimate relationships among any of the variables, whereas the logistic model focuses only on relationships involving the response. The advantage of the logistic model, however, is that it can accommodate quantitative variables.

In the next lesson, we continue the discussion on log-linear models to include ordinal data and "dependent" data, which is effectively the categorical version of matched pairs.