Re: st: Re: standardized coefficient in survey weighted regression using spost9

[Steven Samuels <sjhsamuels@earthlink.net>]

--


In Stata's survey programs, estimates of regression coefficients and  
of population SD's depend only on the survey weights, not on other  
aspects of the design.  Therefore, contrary to Michael's point 2,   
Sandi can  use -reg- with -pweights- and the -beta- option to get  
standardized betas for any complex survey.
-Steve


On Nov 20, 2008, at 6:57 PM, Michael I. Lichter wrote:

> 1. Sandi implied that she was using the survey commands (by  
> mentioning svyset). They do not support the beta option.
> 2. However, if Sandi only has pweights and not a complex design  
> (with clustering or stratification), she can use "regular" - 
> regress- with a pweight and specify the beta option as Martin  
> suggested.
> 3. I assume there is a reason having to do with the survey method  
> of variance estimation that explains why the svy commands do not  
> produce betas and why programs like listcoef and betacoef don't  
> work with them, but I don't know what it is.
> -ml
>
> Martin Weiss wrote:
> > BTW, why not use the -beta- option to -regress- which does allow  
> > sampling weights?
> > *************
> > sysuse auto, clear
> > g r=5*runiform()
> > regress length price mpg rep78 [pweight = r], beta
> > *************
> >
> > HTH
> > Martin
> > _______________________
> > ----- Original Message ----- From: "Pruitt, Sandi"  
> > <SPRUITT@DOM.wustl.edu>
> > To: <statalist@hsphsun2.harvard.edu>
> > Sent: Friday, November 21, 2008 12:27 AM
> > Subject: st: standardized coefficient in survey weighted  
> > regression using spost9
> >
> > > I am running a weighted linear regression (using a sampling  
> > > weight or "pweight") using svyset commands. Following a previous  
> > > listserv discussion, I downloaded spost 9 and used the command  
> > > listcoef so I could obtain standardized betas for this  
> > > regression. (I have a reviewer specifically asking for this.)  
> > > However while it works, I get the following warning:
> > > "pweights not compatible with summarize; weights will be treated  
> > > as aweights"
> > > Can someone tell me how this might be affecting the standardized  
> > > beta that I want to report?
> > > Thanks.
> > >
> > > ________________________________________
> > > Sandi L. Pruitt
> > > Postdoctoral Research Associate
> > > Washington University School of Medicine
> > > Division of Health Behavior Research
> > > Campus Box 8504
> > > 4444 Forest Park Avenue, Suite 6700
> > > St. Louis, MO 63108
> > > Phone: 314-286-1921
> > > Fax: 314-286-1919
> > > spruitt@dom.wustl.edu
> > >
> > >
> > > *
> > > *   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/
> >
> > *
> > *   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/
> --
> Michael I. Lichter, Ph.D.
> Research Assistant Professor & NRSA Fellow
> UB Department of Family Medicine / Primary Care Research Institute
> UB Clinical Center, 462 Grider Street, Buffalo, NY 14215
> Office: CC 125 / Phone: 716-898-4751 / E-Mail: mlichter@buffalo.edu
>
> *
> *   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/
*
*   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/