The sample reduces when running logistic regressions
This post has NOT been accepted by the mailing list yet.
I am running a logistic regression on different samples using multiple imputations:
mi estimate, or post: logit Y X1 X2, vce(cluster PSU)
I have run the regression for the first samples and Stata gave me back results where all the observations are included, except for one sample for which I have got 1332 observations but 1292 observations were taken by Stata for the regression. In addition, the categorical variable education (i.education which takes values 0, 1, 2, and 3) presents a problem: 0.education is the baseline, for 1.education and 2.education Stata gives me back the relative coefficient, whereas 3.education is omitted although there are 168 observations.