Hi,

I have a trouble fitting a model for a categorical data. My dependent variable recorded by a t point likert scale, from very disagree to agree, with independent variables such as age, education (factor variable), income, general health interest, and gender (factor dummy).
I have tried to run OLS regression yet the residuals are not normally distributed and heterokedasticity.
I have tried ordered logit model yet the parallel trend assumption is violated.
I have came to multinominal logit model yet the hausman test for IIA in Stata does not work with factor variable.

Do you have any suggestion to my problem?

Thank you so much.

Lanna