Lily Wang posted on Thursday, December 09, 2010 - 6:34 pm
Hi, I am looking for a software that can help me analyze an SEM model with two related binary dependent variables. Both are single-indicator variables, e.g. enrolled versus didn't enroll. Both are presumed to be influenced by a number of common independent variables, but there is no temporal relationship between the two. However, I do think one could also predict the other. I am not sure if Mplus could help me analyze this. Please advice. Thanks a lot!
You can use either probit or logistic regression to estimate a model where both categorical dependent variables are regressed on a set of independent variables and include the residual correlation between the two categorical dependent variables.
Lily Wang posted on Friday, December 10, 2010 - 9:56 am
Thank you for your reply and sorry about double posting. Won't do it again.
When you mentioned probit or logistic regression, did you mean probit or logistic model in a traditional sense, or is there a way of doing that in an SEM context? My independent variables are largely latent.