st: re: Mplus 7 and Stata 12

["Airey, David C" <david.airey@vanderbilt.edu>]

.

Alan wrote:

> Dave,
> Can you be more specific? The stata2mplus command saves a data set for Mplus and writes a simple Mplus program to verify that it worked. Stata's own sem command can do a lot with quantitative endogenous variables in Stata 12.1.
> 

Phil wrote:

> We, the UCLA Statistical Consulting Group, use Mplus when we have CFA
> or SEM models with binary, ordinal or multinomial outcomes.  We also
> use it to get the Satorra-Bentler scaled chi-square.  And also for
> latent class models.  We use Stata for all of our Mplus data
> management and send it to Mplus using -stata2mplus-.  Mplus 7 runs on
> my Macbook Pro just fine.



I am interested in what is not common between the two software choices for
analysis of longitudinal and hierarchical data, mostly.

I was just skimming the growth curve chapter in Rabe-Hesketh and Skrondal's text and 
noted how you can model continuous outcome longitudinal data using either -sem- or
-xtmixed-, and was curious if this is also true for categorical outcomes.
So Mplus allows categorical outcomes, but so does Stata (xtmelogit, xtmepoisson), just
not in the context of SEM or -sem-.

I was considering purchasing an Mplus license which is used in a longitudinal 
data analysis course I'm going to take soon. However, even if you go back to school
and are a full time student, Mplus must be purchased at expensive non-student prices if you gained
an advanced degree in a non-quantitative field. They have this strange stipulation
in their license. So I'm just balancing that purchase with a potential new
version of Stata coming out next summer with SEM additions, etc.

Mplus 7 looks very good.

Cheers,

-Dave




*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/