Does ANYONE understand the symmetry model for categorical data analysis by Agresti? I have a Likert scale that places students into 4 categories. I am doing a pretest-posttest design, so I will give them the scale at T1, then again later at T2. I will construct a 4x4 table. How do I find out if significant change occurred between categories from T1 to T2? I can't use chi square, must use symmetry model and SAS software (Unless someone knows how to do matched pair analysis in SPSS...). Any suggestions?

I happen to be starting this book and had it laying on my desk next to my pc when I read your post :P

Unfortunately I can't offer too much information since you're quite far ahead of me in the text, but the text does mention for 2x2 tables marginal homogeneity implies symmetry. See pages 358 - 365 for testing info.

Hope this helps a little