st: IV tests using SVY

["Kujawa Rhoads, Jennifer T." <jkujaw2@uic.edu>]

Dear Statalist,

I am running a two stage least squares (2SLS) model in Stata using
?ivregress 2sls- and I am using the   -SVY- prefix for the analysis given
that my data are complex survey data.   I?m trying to figure out how I can
test the validity of my instruments given that I am using SVY.

In order to generate test statistics to test the need for and validity of
the instruments, Stata offers the options ?estat firststage-, -estat
endogenous-, and ?estat overid- for ivregress.  However, these options are
only available if you do NOT use the ?SVY- prefix (you can only specify
the pweight).  From what I can tell, there aren?t any equivalent options
to generate instrumental variable test statistics when using SVY.

This brings me to my question:  Can the test statistics generated using
these estat options for ivregress (and specifying the pweight only) be
used to evaluate my 2SLS model that?s estimated using SVY?   Without using
the psu and strata variables, which way will these test statistics be
biased?   Alternatively, is there a way to generate instrumental variable
test statistics for a 2SLS model that is run using the ?SVY- prefix?

I am using Stata/SE 10.1 for Unix (Born 18 Aug 2009).

Thank you,
Jennifer Rhoads


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