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: dhs data in stata number of obs and population size
From
Steven Samuels <[email protected]>
To
[email protected]
Subject
Re: st: dhs data in stata number of obs and population size
Date
Wed, 18 May 2011 10:00:16 -0400
Michele:
I was incorrect in my advice: if you have an external estimate of the total relevant population in the country, then you can create a weight that will allow you to estimate totals with -svy: total-. Call the external estimate "pop_tot" and the sample size samp_tot
*********************************************
gen double wt_final = w3*pop_tot/samp_tot
********************************************
If you are using one of the specialized modules, then "samp_tot" should be the count of people who got the module.
Steve
[email protected]
Michele
Your question is not clear to me. The statement in the 2007 Bangladesh Final Report is:
"The final weights were normalized in order to give the total number of un-weighted cases equal to the total number of weighted cases at national level, for both household weights and individual weights."
This appears to refer to the normalization (I called it "standardization", incorrectly) that was applied to the weights before their distribution.
If you are asking: can you apply a correction to the weights so that you can estimate population *totals*, the answer is "No". If you want to convert percentages totals, you must multiply them by an external estimate of the total for the population. This, to my mind, is one of the most serious problems with using normalized rates-that they do not provide independent estimates of totals.
Steve
[email protected]
On May 18, 2011, at 4:43 AM, michele savini zangrandi wrote:
Thank you Steve,
Does that mean that if I want to get population estimates (with large standard errors of course) I have to normalize the weights to the population size as reported on the country final report?
Michele
On 17/05/2011 23:12, Steven Samuels wrote:
> According to the DHS Recode Manual (http://www.measuredhs.com/pubs/pdf/DHSG4/Recode1DHS.pdf), the weights are standardized to sum to sample size. You should be very familiar with this manual.
>
>
> -Steve
>
> Steven J. Samuels
> Consulting Statistician
> 18 Cantine's Island
> Saugerties, NY 12477 USA
> Voice: 845-246-0774
> Fax: 206-202-4783
> [email protected]
>
>
> On May 17, 2011, at 12:42 PM, Michele Savini Zangrandi wrote:
>
> This is a question to anyone who has some familiarity with the DHS survey.
>
> I am trying to use dhs survey data in stata, but am unsure on how to
> parametrize it correctly,
>
> So far, using the household members recode file, I have done the following
>
> pweight: w3
> VCE: linearized
> Single unit: missing
> Strata 1: hv022
> SU 1: hv021
> FPC 1:<zero>
>
> where w3=hv005/1000000
>
> Under this parametrization, when I issue an "svy: tab" command for any
> variable, Stata reports a population size very similar to the number
> of observations. This seems a little strange to me. I am doing
> something wrong?
>
> I would really appreciate your help.
>
> Regards
>
> Michele
> *
> * 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/
*
* 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/