I have binary data that I want to run a factor analysis on in Stata. I can run this fine using tetrachoric correlation, but I have nearly 50% missing data, so I really need to impute the data. I have been using MI estimate for my regression analysis, but the command cannot be used with factor analysis. Has anyone got any ideas on how I overcome this please? If I use listwise deletion the results are interesting and I have managed to condense 29 variables in to two, but I want to check this is still valid with the missing data imputed.
Many thanks in advance.