You are assuming there is an underlying population SD, right? Which
means there is an underlying population mean (proportion) m and the SD
is sqrt(m*(1-m)), right?
or after the -vwls- command,
di sqrt(_b[_cons]*(1-_b[_cons]))
see also
http://www.stata.com/support/faqs/stat/supweight.html
(noting that you have aweights).
On Sun, Sep 27, 2009 at 8:02 AM, Frank Peter <[email protected]> wrote:
> Dear Statalisters,
>
> Please, I am trying to convert standard error to standard deviation from meta-analysis of proportion. Taking data from a recent post
> http://www.stata.com/statalist/archive/2009-07/msg00097.html
>
> input study cases total
> 1 20 1000
> 2 40 5000
> 3 30 1500
> 4 25 3300
> end
>
>
> Is it valid if I get the SD like this?
>
> r(seES)*sqrt(1000+5000+1500+3300)
>
> OR
>
> r(seES)*sqrt((1000+5000+1500+3300)/4)
>
> I need the sd to calculate alpha and beta parameter for beta function - ibeta(a,b,x)
>
*
* 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/
