Ting Dai posted on Friday, May 10, 2013 - 10:17 am
Hi Drs. Muthen,
I tried running the ex 11.4 model, and noticed that although in the data set there are missing data indicators, z1-z5, for the y1-y5 (y0 is fully observed), you still have Mplus create d1-d5 to indicate the missingness for y1-y5. I ran the model without creating d1-d5 but use z1-z5 to indicate missing, and the model estimates are different from the original example.
1) How was d1-d5 created, mathematically speaking? 2) Why do they, when put in the model, lead to different estimates from the model with observed missing indicators (z1-z5)?