Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: boostrapping psmatch2, llr kerneltype(normal)


From   "Jef LJP Leroy" <jll27@cornell.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: boostrapping psmatch2, llr kerneltype(normal)
Date   Wed, 29 Jun 2005 17:32:22 -0500

I have succesfully used Stata's bootstrapping command to get the SE's for
psmatch2. 
Today, I tried 
 
replace order=uniform()
sort order
psmatch2 cinter_auto, out(resdiffhaz) pscore(prop_auto) comm llr
kerneltype(normal)
local att=r(att)
drop if _weight==.
bs "psmatch2 cinter_auto, out(resdiffhaz) pscore(prop_auto) comm llr
kerneltype(normal) " "r(att)", /// 	reps(500) cluster(claveofi)

The "drop if _weight==." is there to make sure that bootstrap only uses the
observations that were actually used to calculate att. While this works fine
for other matching methods (neirest neighbor, kernel), the program above
yields the following error:

"insufficient observations to compute bootstrap standard errors
no results will be saved"

I get an empty table bs table, and the strange thing is (and this is where I
am really confused) is that the "Number of obs " the table reports is not
the total number (treatment + control), but only the control observations.
Does anybody have any idea of what is going wrong?
Many Thanks.

Jef

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index