I am conducting an EFA of approx. 40 binary variables, some of which indicate very rare medical conditions. Some responses are missing. My hope is to use the default estimator (WLSM as I understand it) as well as ULS. I am using VARIMAX and PROMAX rotations. When I attempt to run the analysis with either estimator, I get an error message indicating that one of my variables has less than one category. This variable represents the rarest condition but it does occur (approx 70 out of 55,000 observations). I thought that perhaps the error message could be due to missing data, but my understanding is that, by default, V5 of MPlus should not be performing a listwise deletion unless I request it. So my question is, what could be causing this error message?