Dear all,
I am interested in running a latent class regression model. I need to
estimate the class-specific regression models for the outcome variable
y (with covariates x) and the coefficients for the class prediction
equation (with covariates z). Before I can embark on such a project in
Stata, I need to do some more reading though. The standard
econometrics textbooks are silent on the matter of latent classes, and
it is unclear to me how the likelihood function is constructed. If
someone could recommend a good book/article detailing this method,
that would be great.
My specific questions are:
- Do I need to use the EM algorithm, or can I use standard maximum likelihood?
- Can a latent class model accommodate cross-sectional time series data?
- Can the sets of covariates x and z be overlapping?
- Can x contain lagged values of y? Do I need to use instruments for those?
- Can I have clustered data within the latent classes? (standard
errors adjusted for clustering, as in standard Stata regression
commands -cluster(...)- )
I'd be very grateful for any hints.
Thanks a lot,
Eva
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/