I am running initial growth mixture models to predict a binary event outcome.
1. What is the interpretation of a signifcant threshold value in the main model results for each separate latent class, does that simply mean there was event occurance observed within the estimated latent class?
2. In the "results in probability scale" is the significance of category 1 and 2 interpreted as signficant prediction of correct classification of nonevent and event?
Should both categories be signifcant to be meaninful interpretation?
3. Is the Odds ratio for Class 1 compared to Class 2, e.g., 2.705*, simply mean Class 1 is 2.7 times more likely to have the event?
And for .17*, 2 compared with 4, simply mean class 2 is 83 percent less likely than class 4 to have event?
You will find answers to these questions in our Mplus course handouts on our web site. An example of binary distal outcome with a GMM is given in the handout of Topic 6, slides 146-157, especially slide 151.
See also the discussion of logodds, odds, and odds ratios in Topic 2.