I have come across a problem at deciding which hypothesis test I should use.
I have personal previous research (great part of which is lost); of it, I have all the n and the proportions for each variables but not individual data. I also have new research on the topic (say, adherence to certain medical practices) with all complete data.
I am trying to find out if adherence to those practices is statistically different or equal across time. Data is dependent, but not because the observed cases are observed at different times but because the ones executing the practices are the same and since more or less all physicians adhere to some level to guidelines. So...not formally dependent cases but still. So I am not sure if I should use McNeamar's.
Also, I should check, but there is the chance for small superposition (say 2 or 3 cases in both datasets), so I have also looked into Cochran's Q.
I would be very grateful if you could guide me.