I have another question about fitting a two -part mixed-effects model (TPMM) in Mplus.
I tried to fit the same data set for two versions of Mplus code for the TPMM. One approach is creating two sets of variables (the binary part & the continuous part seperately) which is what people usually did using an earlier version of Mplus. Another approach is writting out the "DATA TWOPART" statements and using the raw data, which is shown in Mplus manual version 4 & 5. For the rests of the code, they are identical. What surprised me is that, I get very different results from these two approaches.
I am wondering what causes the difference? If there is anything different, which approach would you suggest me to do?