I have been reading a lot on this site and I think I am going to analyze my data using a mixed model ANOVA- random effect= subject and 2 fixed effects- comparison and pricing. All of my subjects were in all 4 conditions (a cross between the 2 factors) and answered 3 items in each condition. The catch is that my DV is dichotomous. I read that it is not a good idea to assign a mean (i.e. proportion in this case) of the 3 items for each subject and then do the ANOVA using that. Does this mean that I will have number of subjects * 3* 4 observations?
I spoke with my old advisor who specializes in multi-level modeling and he told me to restructure the data so that each column is a subject and each subject has 12 observations. He then told me to go ahead and run an analysis with MIXED. He then told me to do a GENLINMIXED, which can handle dichotomous outcomes. This means there will be (# of subjects)*3*4.