I am running an ordinal logistic analysis in SPSS (command GENLIN or PLUM, have used both).
At the end, I would like to perform a hierarchical ordinal logistic model and hope that there is a way to deal with violating the proportional odds assumption...
Because running a multinomial logistic regression instead of an ordinal logistic regression doesn't make sense in my opinion because the nature of my dependent variable is ordinal and I only have a small sample size.
I appreciate any help or inputs. Thanks
Hi Corinne,
Can you please show a sample of your data, specifying the dependent and independent variables?
You can conduct tests to confirm that the assumption has been violated. If so, you can try a generalized ordered logistic model.
Dear Watheq and Faye, thanks tons for your answers and trying to help me. I guess, I will just remove variables with a high VIF and use Partial Least Squares Regression or Principal Components Analysis to reduce the data. But I don't know the disadvantages of the two when having ordinal data. But I am sure I will figure that out with the help of google.
Regarding your question Watheq: the dependent variable is comparative advantage through sustainable strategy and the independent variables are a bunch which definitely have to be cut such as pollution prevention techniques (with the idea that this cuts the costs), employee activities, pressure of large companies (the thesis focuses on SME), activities of compeditors, etc.
Showing the data doesn't make much sense since everything is in German, sorry.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS