Re: st: How to test for equality of variance in data with sampling weights

[Steven Samuels <ssamuels@albany.edu>]

Johannes, I would be very surprised if the sample was a Simple Random  
Sample.  The text you provided was not in the link, which is a  
spreadsheet "doc_bases2.xls".  However there appears to be a  
technical document Buato302.doc at the link "Recomendaciones t�cnicas  
para el uso de la BUA"  on the page. http://www.indec.mecon.gov.ar/ 
dbindec/default_ampliada.asp. I cannot read Spanish, so I'm unable to  
determine if this will be helpful.

At this point, I suggest that you consult the staff or others who  
have used the EPH for guidance on what variables, if any, designate  
the multistage sampling units in the single urban survey.  At the  
very least,  determine if more than one person was surveyed in each  
household. Determine if there is a household identifier as well as a  
person identifier.  Most of these surveys provide a codebook with  
definitions of variables.

The identification of the specific sampling design would not be  
necessary if all you wanted to do is ESTIMATE standard deviations for  
different groups.  However you will need to compute standard errors  
for the estimated SD's, and for the standard errors, specification of  
the correct design is crucial.

Good luck!

Steven


On May 24, 2007, at 10:22 AM, jmetzler@worldbank.org wrote:


> Dear Steven,
>
> thanks for your reply. The short version of the answer is, (after  
> going back to
> documentation, even though it is not specifically mentioned): the  
> survey seems
> in fact to be stratified by geographical region in the country.  
> There is no
> mentioning of clustering. However, in order to obtain a longer time  
> series, I am
> also working with the same survey reduced to the capital city only  
> that should
> be unstratified.
>
> So, given my original question (about how to test for the equality  
> of variance
> of my labor wage variable  for the whole sample and for specific  
> subgroups over
> time, and for subgroups at the same point in time) help would be very
> appreciated on both fronts, but ignoring strata for the moment is  
> probably
> easier.
>
> Thanks in advance,
> Johannes
>
>
>
> PS: Here is the long answer: I am working with two survey series  
> from Argentina,
> the EPH. The full Argentinean survey is collected in urbean areas  
> throughout the
> country. The survey documentation is in Spanish only, and very  
> unspecific about
> sampling. The relevant section reads:
>
> La E.P.H.  es una encuesta por muestreo, desarrollada en  
> aglomerados urbanos;
> esto significa que para conocer las diversas caracter�sticas del  
> total de los
> hogares de un aglomerado urbano se encuesta una peque�a fracci�n  
> representativa
> de los mismos
>
> Even though it is not mentioned specifically, the notion of  
> representativeness
> by urban area implies stratification. The full document can be  
> found on
> http://www.indec.mecon.gov.ar/dbindec/login.asp, using key
> DF939RFEA45YLHN01DC1VK1G8W6T4J, going to "Base Usuaria Ampliada EPH  
> - Puntual"
> and following the link "El detalle de las bases disponibles se  
> puede consultar
> aqu�".
>
> As mentioned before, if I only use the series for one of the urban  
> areas, there
> is no stratification left.
> There is no mentioning of clustering, but it does not specifically  
> state random
> sampling either.
>
>
>
>
>
>
>
>              Steven Samuels
>              <ssamuels@alban
>               
> y.edu>                                                          To
>              Sent by:                statalist@hsphsun2.harvard.edu
>              owner- 
> statalist                                                 cc
>              @hsphsun2.harva
>               
> rd.edu                                                     Subject
>                                      Re: st: How to test for  
> equality of
>                                      variance in data with sampling  
> weights
>              05/22/2007
>              09:38 AM
>
>
>              Please respond
>                    to
>              statalist@hsphs
>              un2.harvard.edu
>
>
>
>
>
>
> On May 26, 2007, at 9:16 AM, Johannes Metzler wrote:
>
>
> I am working with household survey data, where the sample is not
> clustered, not
> stratified, but has sampling weights. Ultimately, I would like to
> test the
> equality of the variance of my labor wage variable...
> ___________________________________________
>
> Johannes:
>
> Can you please give some more detail about the sampling design?
> Please include information about the target population and sampling
> procedure, whether the same households appeared more than one time in
> the survey, and how the weighting was done. A URL reference to the
> original methodology document, hopefully in English, would be
> sufficient.  If necessary, send the document to me privately.
>
> I have never encountered a houshold survey that was not done in at
> least two stages, and I have never seen a national or regional survey
> which was not stratified in some way.  Without being sure of your set
> up, I am reluctant to try to answer your question.
>
> Steven
>
> ----- Forwarded by Johannes Metzler/Person/World Bank on 05/17/2007  
> 09:45 AM
> -----
>
>              Johannes
>              Metzler
>                                                                        
>        To
>              05/16/2007              statalist@hsphsun2.harvard.edu
>              08:24  
> PM                                                        cc
>              35070   LCSPP
>                                                                        
>   Subject
>                                      How to test for equality of  
> variance in
>                                      data with sampling weights
>
>
>
>
>
>
>
>
>
> Dear statalist users,
>
> I am faced with a problem for which I have not been able to find a  
> solution yet,
> neither in statalist archives nor through googling, (even though it  
> does not
> strike me as something extremely fancy that I am trying)
>
> I am working with household survey data, where the sample is not  
> clustered, not
> stratified, but has sampling weights. Ultimately, I would like to  
> test the
> equality of the variance of my labor wage variable
> 1) for the whole sample and for specific subgroups (such as age  
> groups) over
> time
> 2) for subgroups at the same point in time.
>
> I am aware of the command "robvar" but
> a) it does not feature weights
> b) I am wondering if it is statistically appropriate to pool my  
> data for two
> years and specify 'year' as the groupvar in the by() option.
>
> What would you advise me? Is there another command implemented?
> Thank you very much in advance,
>
> Johannes
>
>
>
>
> *
> *   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/