Regarding nominal tests in ordinal package clm() and brant test for Mass package polr()

[SidMiddela]

Dear Prof. Miller,

I would like to thank you for helping students like me, through your GitHub repositories. I have few questions related to the nominal tests in Ordered Logit models.

1. Are the nominal tests a replica of the brant tests?
2. Do we also need to do scale test to test the proportional odds assumption?
3. How to interpret the age and sex variables in the model "clm(natfare_f ~ collegeed + rincom16 + pid7 + polviews, nominal = ~ age + sex, data=subset(gss_spending, race == 1))"?
4. The variables that have a p-value <0.05 should be used as nominal. I did not follow one or more of the covariates may have non-constant effects at all levels. The non-constant effects means the coefficient at different levels will differ?
5. In the brant test, the probability of the Omnibus should be greater than 0.05, if I understood it clearly.
   So if we build same model using MASS package polr function and Ordinal package clm() function, the results of brant test and combined nominal & scale tests will/ should match?

Thanks for your consideration.

Regards,
Siddartha
PhD student