Re: st: predictive margins not estimable in svy

["Michael I. Lichter" <MLichter@Buffalo.EDU>]

Ah, I see. (Susan said it was Stata 10.)

Three things. First, Susan uses these two commands:

	svy, subpop(subpopnodm): logistic pad i.x age56 sex01 race2
	margins i.x , subpop(if subpopnodm==1)

Are these the same subpopulation? They might have to be. 

Second, the online help for -margin- (http://www.stata.com/help.cgi?margins) says:

       *vce(unconditional)* specifies that the covariates that are not fixed
           be treated in a way that accounts for their having been sampled.
           The VCE is estimated using the linearization method.  This method
           allows for heteroskedasticity or other violations of
           distributional assumptions and allows for correlation among the
           observations in the same manner as *vce(robust)* and *vce(cluster*
           /.../*)*, which may have been specified with the estimation command.
           This method also accounts for complex survey designs if the data
           are *svyset*. See /Obtaining margins with survey data and/
           /representative samples/ in *[R] margins*.

This makes 
	
	margins i.x , subpop(if subpopnodm==1) *vce(unconditional)*

look likely, and if that doesn't work, try the advice at the end of that paragraph -- "See /Obtaining margins with survey data and/ /representative samples/ in *[R] margins*."

-ml


Martin Weiss wrote:
> <>
>
> Upgrade to 11 to get it...
>
>
> http://www.stata.com/stata11/margins.html
>
>
>
> HTH
> Martin
>
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Michael I. Lichter
> Sent: Samstag, 8. August 2009 02:25
> To: statalist@hsphsun2.harvard.edu
> Subject: Re: st: predictive margins not estimable in svy
>
> Susan,
>
> Where did you get this -margins- command? It's not part of Stata 10, and 
> -findit- doesn't show anything.
>
> -ml
>
> susan@hailpern.com wrote:
>   
> > Hello,
> >
> > I am having problems using the Stata 10 "margins" command with survey
> > data.  
> >
> > I have a created a subpop of ~2500 observations (total 5000+) and would
> > like to obtain the age- sex- race adjusted prevalence of my dichotomous
> > outcome by cystatin-c quartile. Data is svyset; svy logistic model
> > including age, sex and race and cystatin-c was run, followed by the
> > marginals command.  
> >
> > The output says (not estimable). (below)
> >
> > Can anyone help?  I do get estimations if I take it out of survey mode,
> > Thanks,
> > Sue
> >
> > . svy, subpop(subpopnodm): logistic pad i.x age56 sex01 race2
> > (running logistic on estimation sample)
> >
> > Survey: Logistic regression
> >
> > Number of strata   =        28                 Number of obs      =     
> >  5684
> > Number of PSUs     =        57                 Population size    = 
> > 230532837
> >                                                Subpop. no. of obs =     
> >  2472
> >                                                Subpop. size       =
> > 90812359.8
> >                                                Design df          =     
> >    29
> >                                                F(   6,     24)    =     
> > 28.73
> >                                                Prob > F           =    
> > 0.0000
> >
> > ------------------------------------------------------------------------------
> >              |             Linearized
> >          pad | Odds Ratio   Std. Err.      t    P>|t|     [95% Conf.
> > Interval]
> > -------------+----------------------------------------------------------------
> >            x |
> >           2  |   1.521746   .5364044     1.19   0.243     .7400256    
> > 3.12923
> >           3  |   4.057542   2.135947     2.66   0.013     1.382564   
> > 11.90805
> >           4  |   9.510348   3.931692     5.45   0.000     4.083106   
> > 22.15145
> >              |
> >        age56 |   4.655526   2.164545     3.31   0.003     1.798843   
> > 12.04881
> >        sex01 |    1.14324    .236113     0.65   0.522      .749361   
> > 1.744149
> >        race2 |   2.314983   .5035135     3.86   0.001     1.483732   
> > 3.611936
> > ------------------------------------------------------------------------------
> >
> > . margins i.x , subpop(if subpopnodm==1)
> >
> > Predictive margins                                Number of obs   =     
> >  3009
> >                                                   Subpop. no. obs =     
> >  2472
> > Model VCE    : Linearized
> >
> > Expression   : Pr(pad), predict()
> >
> > ------------------------------------------------------------------------------
> >              |            Delta-method
> >              |     Margin   Std. Err.      z    P>|z|     [95% Conf.
> > Interval]
> > -------------+----------------------------------------------------------------
> >            x |
> >           1  |  (not estimable)
> >           2  |  (not estimable)
> >           3  |  (not estimable)
> >           4  |  (not estimable)
> > ------------------------------------------------------------------------------
> >
> >
> >
> > *
> > *   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. <mlichter@buffalo.edu>
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 126 / Phone: 716-898-4751 / FAX: 716-898-3536

*
*   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/