[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: oaxaca with subpop in svy

From   Austin Nichols <>
Subject   Re: st: oaxaca with subpop in svy
Date   Wed, 15 Jul 2009 15:05:28 -0400

Maggie Hicken<> :
Make variables that are defined only for your desired subsample, e.g.

webuse nhanes2, clear
g male=sex==1 if sex<.
g mblack=black if male==1
oaxaca bpsystol weight, by(mblack) svy
* not:
oaxaca bpsystol weight, by(black) svy
* compare :
svy, subpop(male): reg bpsystol black
svy: reg bpsystol mblack

On Wed, Jul 15, 2009 at 2:51 PM, Maggie Hicken<> wrote:
> Thanks -- Yes, I understand that I can use 'svy' with 'oaxaca' -- but I need
> to specify subsamples of the entire NHANES.  For example, I want to analyze
> only women who are black or white.  For 'regress', I would specify the
> 'subpop' option with the 'svy' prefix.  But for the 'oaxaca' command, I get
> an error message when trying to specify a subpopulation with the 'svy'
> option (saying that it doesn't support the subpop option).
> -----Original Message-----
> From:
> [] On Behalf Of Richard Williams
> Sent: Wednesday, July 15, 2009 3:43 PM
> To:;
> Subject: Re: st: oaxaca with subpop in svy
> At 01:01 PM 7/15/2009, Maggie Hicken wrote:
>>I would like to use the 'oaxaca' command using 'svy' in NHANES, but need to
>>use the 'subpop' option of 'svy' to have the correct SEs.  Is there a way
> to
>>use the 'subpop' option?  Or does anyone have an alternative that would
>>yield the SEs that I need in Oaxaca with a complex survey design?
>>Thanks for your help!
> It says how to do it in the oaxaca help file.  Instead of an svy
> prefix, you use an svy option.

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index