st: Goodness of Fit Statistics using -gsem- in Stata 13.1 SE

["Christopher P. Salas-Wright" <salaswright@utexas.edu>]

Hello all,

I am attempting to run a path analysis using the -gsem- function in Stata
13.1 SE.  All of the variables used in my model are binary (0 = no, 1 =yes).
As an example, my syntax looks as such:

gsem (badactor-> badactson badactdaughter), logit

I am running into two problems:

*** Problem #1 ***

First, I am unable to get goodness of fit statistics using the -estat gof,
stats(all)-  post-estimation command that normally works with the regular
-sem- command.  From what I can tell, it seems that I cannot get goodness
of fit statistics using the -gsem- command.  Is this correct?  If so, are
there other options?  For example, could I run the same model simply using
-sem- rather than -gsem-?  I am hesitant to do so, even though when I do
run my model this way it does seem to have acceptable fit per the RMSEA,
CFI, etc.

*** Problem #2 ***

I am having difficulty allowing error terms to covary using the -gsem-
command as well.  Take, for example, what happens when I use the following
syntax:

gsem (badactor-> badactson badactdaughter), logit cov
(e.badactson*e.badactdaughter)

When I do this I get the following error message: "invalid covariance
specification; e.badactdaughter does not identify an error for an
uncensored gaussian response with the identity link, r(198);"
Note: I don't get this error message when I run this model using the
"sem" command
or when I use the -gsem- command but don't specify "logit".

Does anyone have any insight regarding these two issues?

My sincere thanks!

Chris
*
*   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/