I am rnning a multinomial logistic regression in spss. My outcome variable is a categorical variable with 4 categories, I have several categorical predictors and age as a continuous predictor. I need to check for the linearity of the logit assumption.
Following what is described in "Discovering statistics using SPSS" by Field, I ran the model as it is adding the interaction between age and Ln(Age) to check if the linearity of the logit assumption is met. The interaction term is significant <0.001 in all categories. Firstly, does that mean that the assumption is not met? If so, what are my options? Ideally, I'd like to work on SPSS but I can also use R if necessary. Is there a way to loosen the assumption or are there any other usefull model alternatives? Please keep in mind that I need to keep the model as simple to interpret as possible.