for a homework paper I have analyze some data. Its about how social networks are related to health. My depended var is an ordinal variable with 5 categories. I used an ordered logistic regression (ologit in STATA12) to render some models. The only problem: according to omodel, the proportional odds assumption is violated
I used this tutorial here and sticked to the solution with the gologit2 pack. This works, but the problem is: i have to idea how to interpret these tables there, because now I get back a coefficient for every category of the dep. var. Also: according to the output for a certain var:
"Constraints for parallel lines imposed for var33 (P Value = 0.1698)"

But when I checked the tables and the coefficient is the SAME in every category for var33! Then why is the assumption violated for this var?

I'd like just to use the normal ologit model. Is there any help to justify this? My paper cant be longer than 15 pages, so i'd like to keep it short.

Thank you!