Re: st: weight

[Steven Samuels <ssamuels@albany.edu>]

On Jul 3, 2007, at 1:27 PM, Sebastian Kruk wrote:
> I have a Stata database and a Excell spreadsheet.
> In the first one,  I have a survey microdata of households (v.g.
> income, quality of life, poverty, education, employment, housing and
> equipment).
> In the second one, I have a excell archive with population
> projections. I will like to use it to compute a weight for a data in
> Stata.
> How I form groups by ages and then I use a weight by group of age?
What you want to do is called "poststratification."

1.  Create age group categories for your data to match categories in  
your spreadsheet.

Suppose the spreadsheet categories and corresponding population  
projection counts are:

20-29    140,000
30-39    120,000
40-49    300,000
50-59    200,000
60-69    100,000
70-100    50,000

In your Stata data set, create a grouped age variable with the same  
categories:

egen agegp=cut(age), at(20,30,40,50,60,70)
// Creates a grouped variable & names the categories for the LH  
endpoints

label define agegp 20 "20-29" 30 "30-39" 40 "40-49" 50 "50-59" 60  
"60-69" 70 "70+"
label values agegp agegp

2. Create a variable in Stata to hold the projection counts. You can  
do this by hand or you can add an "agegp" column to the spreadsheet  
and use -insheet- to bring them into a stata data set, then -merge-  
with your data by agegp

By hand:
gen     age_ct=140000 if agegp==20
replace age_ct=120000 if agegp==30
replace age_ct=300000 if agegp==40
.
.
replace age_ct=50000  if agegp==70


3. Finally, -svyset- your data.  As Friedrich stated, there may  
already be a "weight" variable.  There may also be a stratum  
variable, and primary sampling unit variable in your microsurvey.  
These should be documented in the survey codebook.
I assume you have all three, named "final_wt","stratum", "psu".

The -svyset- command is:

"svyset psu [pweight=finalt_wt], strata(stratum) poststrata(agegp)  
postweight(age_ct)"

4. If you you have separate population counts on age and gender, then  
you might create a combination age-gender variable and post-stratify  
on this 12 category variable.  Be aware that if you have few people  
in some categories you may need to combine them.

For example:
gen     age_gender = 21  if agegp==20 & gender==1
replace age_gender = 22  if agegp==20 & gender==2
etc.

-Steven


*
*   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/