I need to create a table that compares pre-test/post-test change for dichotomous variables across demographic categories. I used PROC MI to impute missing data with fcs discrim for the dichotomous variables.   I used PROC SurveyFREQ to conduct the McNemar Chi-square. I have requested ODS output for both DiscordDiff and CrossTabs.    For the McNemar test, I can get the difference estimate and test statistics using the following: proc mianalyze data=[DiscordDiff dataset]; modeleffects Estimate; by female; stderr StdErr ; run;   I saw that the DiscordDiff dataset doesnt include the time 1 and time 2 mean scores. So I included CrossTabs but that doesnt have a stderr variable   Is there a way that I can can get the imputed mean scores (percentages, frequencies) for the pre-test  and post-test?        ... View more