In this lesson, we explored the structure and potential associations between 3 categorical variables. We extended the concepts of testing for independence and associations from two-way to three-way tables. We introduced various types of associations, including marginal and conditional, and showed how the presence of a confounding variable can change or even reverse the relationship between the others.

Up to this point, we have been focusing on significance testing and summary measures of association. In what follows, we will switch gears and learn how to explain these associations by way of models and focus on estimating the parameters of the models. The advantage of the modeling approach is the additional flexibility to accommodate more variables and even quantitative predictors.