I'm struggling with quite a simple sample/cell size issue. I am fitting an ordinal logistic regression, and I have 2 predictors - both categorical - and an outcome with 3 categories. I am only testing main effects
When I add both predictors to the model, should I focus on the cell counts separately for each predictor or be concerned about the cell counts as a function of both predictors. That is, should I be focusing on the cell counts in P1*P2*Outcome categories or P1*outcome and P2*outcome separately. I hope I've made some sense.