Test for Ologit and Gologit

[Valerie Aulia]

Hello everyone

In my recent study, I have an ordinal dependent variable (Health Complaints) ranging from 1 to 4, and binary independen var (Car Ownership) with 0 and 1. This ordinal dependent var make me choose ologit as my main model. But, my model cant fill the parallel trend assumption, thus i choose gologit model.

I'm still confused with the test that i need to run:
1. Do I need to run Goodness of Fits (with Hosmer-Lemeshow test)? If so, when did i have to run this, after running the ologit, or after i run the gologit estimation?
2. After running ologit, i do oparallel test (for brant test to check the parallel trends, with the result that this assumption is rejected thus i choose gologit for my model). I also do multicollinearity using VIF. Are these correct?
3. Do I also need to run VIF after gologit?
4. Are there any other test that i need to run for this ologit and gologit (except brant test and VIF)?

Im really glad if you can help me to understand this :')

Thank you!

[Maarten Buis]

The way you described your model you have only one explanatory/independent/x/right-hand-side-variable. If that is true than you cannot have multicollinearity, so there is no need to test for that.

[Valerie Aulia]

My control variables are: Education level, age, sanitation, main source for drinking water, smoking, travel behaviour, working hour, water and food deprivation, health insurance, social insurance. With these control variables, i still need to perform multicollinearity and GOF test, right?