Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: nldecompose and svy subpopulation
From
Austin Nichols <[email protected]>
To
[email protected]
Subject
Re: st: nldecompose and svy subpopulation
Date
Tue, 5 Jun 2012 16:37:07 -0400
meenakshi beri <[email protected]>:
Or switch to -oaxaca- on SSC, described in
http://www.stata-journal.com/sjpdf.html?articlenum=st0151
webuse nhanes2, clear
oaxaca highbp age black orace, logit svy by(female)
oaxaca highbp age, logit svy(,subpop(black)) by(female)
On Tue, Jun 5, 2012 at 4:07 PM, meenakshi beri
<[email protected]> wrote:
>
> Thanks Steve! I will write to the authors.
>
> Best Regards,
> Meenkashi
>
>
> ________________________________
> Meenakshi Beri
> Graduate Teaching Assistant
> Department of Economics
> Wayne State University
> [email protected]
>
>
> ----------------------------------------
>> Subject: Re: st: nldecompose and svy subpopulation
>> From: [email protected]
>> Date: Tue, 5 Jun 2012 16:03:35 -0400
>> To: [email protected]
>>
>>
>>
>> -nldecompose- seems to ignore all the -svyset- information except the probability weights. Bootstrap standard errors,
>> the only ones available, completely ignore strata, clustering, (and subpop()), all of which affect standard error calculations in ordinary survey work.
>>
>> You can write to the author. Perhaps he, or someone else familiar with decomposition techniques, can explain why the survey design should not have a say in estimating standard error.
>>
>> Steve
>> [email protected]
>>
>> On Jun 5, 2012, at 2:35 PM, meenakshi beri wrote:
>>
>> Thanks a lot Steve!
>>
>> I am sorry about the inconvenience. I used -findit nldecompose- and installed st0152_1 package.
>>
>> Yes, you are right. My subpopulation specification was wrong. However, even after changing the subpop() specification as suggested, it still gives me an error whereas the code runs perfectly if I use svy regression without subpopulation specification. My logic is whether svy subpopulation works with nldecompose or not(Probably it runs with svy commands only, and not with the svy subpop)?
>>
>> ----------------------------------------------------
>> nldecompose, by(female) bs bsoptions(reps(100)) regoutput: svy, subpop(sample_oaxaca) : logit x y z, or
>>
>> r(101)
>> factor variables and time-series operators not allowed
>> -------------------------------------------------
>> Regards,
>> Meenakshi
>>
>> ________________________________
>> Meenakshi Beri
>> Graduate Teaching Assistant
>> Department of Economics
>> Wayne State University
>> [email protected]
>>
>>
>> ----------------------------------------
>> > Subject: Re: st: nldecompose and svy subpopulation
>> > From: [email protected]
>> > Date: Tue, 5 Jun 2012 14:07:14 -0400
>> > To: [email protected]
>> >
>> >
>> > As a frequent poster to Statalist, you should know to tell us where to get a non-officical command (SSC?
>> > somewhere else?). Your error is in the subpop() specification, which takes either the name of a 0/1 variable
>> > which indicates the subpopulation or an "if" expression. See Example 1 of the "subpopulation
>> > estimation" section of the Survey Manual.
>> >
>> > Thus: "subpop(if sample_oaxaca==1)" or, if sample_oaxaca is 0/1, "subpop(sample_oaxaca)" should work.
>> >
>> > Steve
>> > [email protected]
>> >
>> >
>> > On Jun 5, 2012, at 1:26 PM, meenakshi beri wrote:
>> >
>> > Hello Statalist,
>> >
>> > Is svy subpopulation logit analyses compatible with nldecompose command (non-linear decomposition package by Sinning, Hahn and Bauer)?
>> >
>> > I have tried the following but it does not work:
>> > nldecompose, by(female) bs bsoptions(reps(100)) regoutput: svy, subpop(sample_oaxaca == 1) : logit x y z, or
>> >
>> > Am I doing something wrong or nldecompose does not work with svy subpopulation? Any help would be highly appreciated.
*
* 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/