Dear All,
I am working on a multivariable polytomous regression
model with 3 levels of outcome.
Level 1 is the reference and there are 2 sets of effects
for each explanatory factor.
My question is, can I apply some sort of post-hoc test
to assess whether there are differences between levels
2 and 3 of the outcome, for each factor?
I guess I could simply repeat the model with level 3 as
reference to get my effect sizes but I would probably need
to make the p-values more conservative.
Any help gratefully recieved
Jon
|