I am conducting the ordinal regression model using the "clmm" function of the package "ordinal". As the tutorial said, we should use the "nominal_test" and "scale_test" in the "clm" function to check the assumption.
Like the result of the example in this tutorial, unfortunately, the p-values of some predictors are significant in "nominal_test" and "scale_test" in my case. According to Fabian Bross (2019, www.fabianbross.de/mixedmodels.pdf.), "if the proportional odds assumption fails, the results of the model will not be reliable."
Then I also tried to check the assumption by the "brant" function in the package "brant." Again, some predictors are significant, which means my model is not reliable.
As I have to fit the ordinal regression model with random factors, it is difficult for me to give the "clmm" model up. I am unsure about the importance of checking assumptions and what I should do next, as the model failed to check the assumption.
