Message/Author |
|
Rohan Palmer posted on Thursday, February 07, 2013 - 11:56 am
|
|
|
I am working on a Latent Class Analysis using twin data. In order to account for the nonindependence of twins, I am using the command "CLUSTER = ". As my outcome may vary by age, I would also like to include age as a covariate for class membership in the model. However, I am having difficulty running the model with clustering along with a covariate. Is this possible in MPlus? If so, are there unique specifications I should consider? |
|
|
Are your data in a wide format where you have variables both twins or in a long format where you have one variable and a variable that identifies each twin. If you have the wide format, you do not need to control for clustering because the multivariate analysis does that. You can have both clustering and covariates. |
|
|
Sorry for the delay. I was watching my inbox and not the site. Our data are in long format and that we have a variable that identifies each twin and each family |
|
|
You can turn your long data into wide using DATA LONGTOWIDE and then use the Fig 4 approach of Prescott http://www.statmodel.com/download/BG-Prescott2004.pdf where you modify this to LCA in line with http://www.statmodel.com/download/Topic%205.pdf |
|
Back to top |