That makes sense, but there is another hitch....
The reason for needing the correct sample size is that I need to compute the population standard deviation. I have been using the command suggested in http://www.stata.com/support/faqs/stat/supweight.html
The post estimation command after I use the svymean is:
di sqrt(e(N) * el(e(V_srs),1,1))
Additionally, I will be using some by commands as well. For example:
svymean var1, subpop(samp) by(sex)
and then I want to compute the population standard deviation for var1 for both categories of sex. Without the sample size, I cannot get the correct standard deviation using the suggested post estimation command.
----- Original Message -----
From: jpitblado@stata.com (Jeff Pitblado, StataCorp LP)
Date: Friday, May 19, 2006 10:14 am
Subject: Re: st: subpop and the mysterious sample size
> ROBERT BOZICK <rbozick1@jhem.jhu.edu> is concerned abou the
> 'Number of obs'
> output in the header of -svymean- (replaced by -svy: mean- in
> Stata 9):
>
> > I am working on a project where I need to use survey commans to
> estimate the
> > standard error correctly --- my sample uses a stratified cluster
> design. I
> > created a variable called samp to indicate the analytic sample.
> When samp =
> > 1, then the respondent will be included in the analysis; when
> samp = 0 then
> > the respondent will not be included in the analysis.
> >
> > The frequency of samp is shown below:
> >
> > tab samp
> >
> > samp | Freq. Percent Cum.
> > ------------+-----------------------------------
> > 0 | 6,917 42.25 42.25
> > 1 | 9,456 57.75 100.00
> > ------------+-----------------------------------
> > Total | 16,373 100.00
> >
> > As you can see, there should be 9,456 in my analysis.
> > When I use the svy commands to estimate means for my analytic
> sample using
> > the subpop command, the output reports that there are 15,548
> used in the
> > analysis. Intuitively, that cannot be correct. Does anyone
> know what is
> > going on here? How can I fix this so that it reports 9,456
> instead of
> > 15,548? Thanks!
> >
> > svymean var1, subpop(samp)
> >
> > Note: 11 strata omitted because they contain no subpopulation
> members>
> > Survey mean estimation
> > pweight: f1pnlwt Number of obs
> = 15548
> > Strata: strat_id Number of
> strata = 350
> > PSU: psu Number of PSUs
> = 729
> > Subpop.: samp==1 Population
> size = 3312561.5
> > -----------------------------------------------------------------
> -------------
> > Mean | Estimate Std. Err. [95% Conf. Interval]
> Deff> ---------+---------------------------------------------------
> -----------------
> > var1 | 46.72176 .2878163 46.15584 47.28768
> 4.473885> ---------------------------------------------------------
> ---------------------
>
> In Stata 8, unlike -svyregress-, -svymean- does not display the
> subpopulationsample size. However Robert could use -svyregress-
> to compute a subpopulation
> mean
>
> . svyregress var1, subpop(samp)
>
> and have the subpopulation size and its sample size get reported
> in the
> header. Note that the subpopulation mean estimate and its
> standard error are
> the same whether you use -svymean- or -svyregress-.
>
> In Stata 9, all the -svy- estimation commands have a unified
> header, and this
> header includes the subpopulation size and its sample size too.
> The Stata 9
> syntax, using Robert's example is
>
> . svy, subpop(samp) : mean var1
>
> --Jeff
> jpitblado@stata.com
> *
> * 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/
>
*
* 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/