|
![](http://www.statmodel.com/discussion/icons/csc/1px.gif) |
Limitation to Latent Transition Analysis |
![Previous](http://www.statmodel.com/discussion/icons/csc/navgen_arrow_left.gif) ![](http://www.statmodel.com/discussion/icons/csc/1px.gif)
|
Message/Author |
|
Siny Tsang posted on Friday, November 14, 2014 - 3:05 pm
|
|
|
Hello, I am trying to fit a LTA model to data that consists of 22 dichotomous items, each measured 8 times. I was wondering if there is a limitation with respect to the size of the contingency table for conducting LTA? I got the 2-class model to run, but the 3-class model seems to be stuck at reading the input file; the iterations of estimation have yet started. I was wondering if you have any suggestions? Thanks! |
|
|
8 time points is heavy for LTA even if you had fewer items per time point than your 22. You can try the following approach: 1. Analyze each time point separately using LCA and save estimates using the SVALUES option. These runs should use many STARTS. 2. Put together the SVALUES for a joint LTA run of all time points, making sure that the classes are ordered in the same way, specifying measurement invariance, and requesting STARTS = 0. This may still give slow computations in which case you want to simplify by first trying fewer time points, say 1st and 8th; 1st, 4th, and 8th; etc. |
|
Back to top |
|
|