Hello community,
I wanted to ask about the assumption of proportional odds and the Brant test. i have two questions. firstly, when assessing the brant tests violations do you only consider the overall result for all the explanatory variables?
to be specific the overall results for the brant test is more than 0.05 but i have noticed that some of independent variables are less than 0.05.
chi2 p>chi2 Df
All 41.04 0.339 38
1.sex 4.88 0.087 2
1.education
2.education
3.education		 1.18
0.38
0.59		 0.555
0.828
0.746		 2
2
2
1.marital
2.marital
3.marital		 6.76
3.45
7.60		 0.034
0.178
0.022		 2
2
2
1.employment
2.employment		 3.09
0.50		 0.213
0.780		 2
2
1.tobacco 3.83 0.147 2
1.reside 6.67 0.036 2
1.income
2.income		 2.00
0.12		 0.368
0.942		 2
2
1.agegroup
2.agegroup
3.agegroup		 0.50
0.37
2.55		 0.778
0.831
0.280		 2
2
2
1.moderate 2.81 0.245 2
1.vservings 3.60 0.165 2
1.alcohol 1.41 0.494 2
I'm wondering whether i should use the gologit2 model instead because of the two variables that are violating the assumptions or should stick to the overall results and proceed with the proportional odds model.

my second question is based the fitstat(when doing model comparison):when i do a comparison between my nested model and full model,i noticed that there is no difference between them, could this be an issue with my data?