I estimated a structural equation model, using data constructed for Mplus 6.11, and got the "number of observations" of 14,322. I needed couple of new variables in the data, so I added the variables, using SPSS for Windows (as I always have done), and converted the SPSS data file into Mplus data, using Stat Transfer Eleven (which I usually do). When I ran the initial file (i.e., DATANAME.inp), its output told me the number of observations is 14,322. But, when I estimated the structural equation model with the new variables added, the total number of observations in its output is only 7,161, which happens to be exactly the half of what was expected. If I run the same model without the new variables, using the "old" data, I get 14,322. I checked the new variables' frequencies, which look fine. I don't understand why I'm having this problem. Would this have anything to do with Mplus 7, which I'm now using? Or, perhaps Stat Transfer Even not being able to "fully" prepare data for Mplus 7? This doesn't make sense.