The SAS code:

proc genmod data=diets descending;

class ID DIET;

model TASTE = DIET / dist=mult link=clogit;

repeated subject=ID /type=IND CORRW COVB;

RUN;

I have the following output and I was wondering how best to interpret the following results.

*PROC GENMOD is modeling the probabilities of levels of TASTE having LOWER Ordered Values in the response profile table.*This is how I have understood it based on this general expression:

Given the output stating “PROC GENMOD is modeling the probabilities of levels of TASTE having LOWER Ordered Values in the response profile table,” –

The intercept1 (5 on the Likert scale) beta estimate is 0.07. In terms of odds that becomes exp0.07exp0.07.

In reference to D_DIET, if the subject is on A_DIET, the odds of having a rating of 5 then becomes exp0.07+0.15exp0.07+0.15.

Extending the same logic, for intercept2 (4 on Likert scale) beta estimate is 1.07. In terms of odds that becomes exp0.07exp0.07.

In reference of D_DIET, if the subject is on A_DIET, the odds of having a rating of 4 then becomes exp1.07+0.15exp1.07+0.15.

However, to estimate the OR (odds ratio) between ratings of 5 vs. ratings of 4, for the subjects who are on A_DIET in ref. to D_DIET, the OR is: exp0.07+0.15−exp1.07+0.15exp0.07+0.15−exp1.07+0.15. Simplifying this further, exp0.07−1.07exp0.07−1.07.

If one is interested to estimate the OR between A_DIET and B_DIET for having received a rating of 5, then the estimated should be exp0.07+0.15−0.07–(−0.08)=exp0.15+0.08exp0.07+0.15−0.07–(−0.08)=exp0.15+0.08

I would appreciate your comments or suggestions if these interpretations are correct.

Thank you