ABSTRACT
Although still dealing with analysis of repeatedly observed categorical observations, this chapter requires a change in perspective. Thus far, we have focused on variable relationships when introducing readers to log-linear models and χ2 decomposition. An approach oriented somewhat more toward differential research was presented with Prediction Analysis (PA) which links patterns of predictor variables with patterns of criterion variables. Yet, PA still expresses results in terms of variable relationships. The present section, introducing Configural Frequency Analysis (CFA), focuses on cases, that is, subjects or observations. Similar to Prediction Analysis, variable relationships are of interest only in that a log-linear base model must be defined, relative to which subjects can be described.
