I have a dataset with matched pairs from a retrospective cohort (not all cases have a match). I wanted to know the best way to analyse the data. Most of my outcomes are categorical. Should I code each dataset as case/control and then go ahead with chi square? Or should I be using conditional logistic regression/mc nemar test?
Also, I have cases/controls with multiple occurences of the outcome of interest (the time to occurence is not relevant to the study), while their matched pair may have zero or a different number of occurences of the outcome of interest. How do I analyse this? Here the chi square wont work since 1 case/control will be counted more than once, right?