11: Advanced Topics I

Overview Section

In the previous lesson we have covered the basic applications of loglinear models: independence and various types of association. In this lesson we continue with further applications involving ordinal variables and dependent data. Ordinal variables allow for a linear relationship to exist, and we can accommodate with the log-linear model by including a slope parameter interpreted in much the same way as that in a regression model.

The dependent data consideration is the categorical version of matched pairs or repeated measures, where an association between variables is taken for granted, and the questions of interest focus more on the type of agreement between the variables.

Upon completion of this lesson, you should be able to:

  Objective 11.1

Use the log-linear model to measure association between ordinal variables.

  Objective 11.2

Recognize the difference between independent and dependent samples involving a categorical response and measure the strength of agreement between two categorical variables.

  Objective 11.3

Distinguish between association and agreement when dealing with dependent categorical and use log-linear models to describe the agreement.