Kaisa Kirves posted on Wednesday, November 27, 2013 - 11:31 am
I run the same input file with the same dataset with two different computers but with the same version of Mplus (7.11) and the results were partly different because of a different seed value. How is this possible? The model was factor mixture model with two time points. The best loglikelihood value was the same. Also the identified groups were the same but in different order. The largest differences were in auxiliary results: the groups means were different and thus also the chi values were different.