Re: st: generating std. devs. using svyset

[Steven Samuels <ssamuels@albany.edu>]

Kristen's simple suggestions and my complex one all give the same  
answer.  This makes sense--they are mathematically identical. Here  
are some numbers from nhanes2, which has n=10,351 and is stratified  
and clustered.

Variable      Mean           SD summ aweight  SD complex      SD e(V_rs)

height        168.45988           9.699          9.699          9.699
weight        71.900635          15.433         15.432         15.433
bpsystol      126.94579          21.401         21.400         21.401
bpdiast       81.017265          12.794         12.793         12.794

If one wants to compute a CI for the population SD, the answer will  
depend on the sample design and a more "complex" approach.

Steve

On Apr 26, 2007, at 4:23 PM, Kristin MacDonald wrote:


> Leslie Hinkson <lhinkson@Princeton.EDU> asked about calculating  
> standard
> deviations when working with survey data.  One way she can obtain the
> population standard deviation is to estimate the mean of the  
> variable using
> -svy:mean-.  This command returns a matrix e(V_srs) that contains  
> an estimate
> of the variance assuming data were collected from a simple random  
> sample
> without replacement.  The formula for V_srs is
>
>        V_srs = (s^2) / n
>
> Therefore, we can use this value along with the sample size in e(N)  
> to obtain
> an estimate of the population standard deviation.  Here is an example.
>
>        webuse nhanes2, clear
>        svyset
>        svy: mean height
>        di "standard deviation = " sqrt(e(N) * el(e(V_srs),1,1)
>
> It turns out that this standard deviation can also be calculated by
> -summarize- with aweights.  For further details on this, see
>
>        http://www.stata.com/support/faqs/stat/supweight.html
>
> -- Kristin
> kmacdonald@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/