Categorical Data Analysis Using The SAS® System, 2nd Edition

9.3. Nominal Response: Generalized Logits Model

9.3.1. Methodology

When you have nominal response variables, you can also use logistic regression to model your data. Instead of fitting a model to cumulative logits, you fit a model to generalized logits. Table 9.6 redisplays the data analyzed in Section 6.3.3. Recall that schoolchildren in experimental learning settings were surveyed to determine which program they preferred. Investigators were interested in whether their response was associated with their school and their school day, which could be a standard school day or include afterschool care.

Table 9.6. School Program Data

		Learning Style Preference
School	Program	Self	Team	Class
1	Regular	10	17	26
1	After	5	12	50
2	Regular	21	17	26
2	After	16	12	36
3	Regular