|
|
Mixed Latent Markov Modeling |
|
Message/Author |
|
|
Greetings: We are attempting to expand upon the mover-stayer model (see ex. 8.14 in the MPLUS user's guide). Here are the specifics: We have a variable measured at four points in time and the variable has three categories. We hypothesize three latent classes for each time point and two latent transition chains. The problem we are running into deals with assumptions. What is the logic behind fixing the logit intercepts (@10 and -10?) and how can it be extended to more than two time points? We're also not sure to specify constraints in the mover chain of the measurement portion of the model. A specific problem is that we are getting the following error "ordered thresholds for class x and y for indicator z are not increasing" Many thanks, Tom |
|
|
I would suggest looking at the following papers which are available on the website: Kaplan, D. (2008). An overview of Markov chain methods for the study of stage-sequential developmental processes. Developmental Psychology, 44, 457-467. Nylund, K. (2007). Latent transition analysis: Modeling extensions and an application to peer victimization. Doctoral dissertation, University of California, Los Angeles. |
|
Back to top |
|
|