Creating a survey weight with survwgt

[Tom Prosser]

Dear Forum,
I want to create weights for survey data (using the characteristics of gender, age and ethnicity) and was just wondering if I was using the survwgt package (and its raking function) correctly? Any comments on the code below would be most gratefully received. Best and thanks,
Thomas


gen Weight = 1

// generate totals for the weights using demographic data

gen GenderTotal=.
replace GenderTotal = 0.49 if Gender == 1
replace GenderTotal = 0.51 if Gender == 2

gen AgeTotal=.
replace AgeTotal = 0.12 if Age_group == 1
replace AgeTotal = 0.17 if Age_group == 2
replace AgeTotal = 0.17 if Age_group == 3
replace AgeTotal = 0.16 if Age_group == 4
replace AgeTotal = 0.38 if Age_group == 5

gen EthnicityTotal=.
replace EthnicityTotal = 0.63 if Ethnicity == 1
replace EthnicityTotal = 0.12 if Ethnicity == 2
replace EthnicityTotal = 0.06 if Ethnicity == 3
replace EthnicityTotal = 0.11 if Ethnicity == 4
replace EthnicityTotal = 0.08 if Ethnicity == 5

// rake the weights using the Stata survwgt package

survwgt rake Weight , by(Gender Age_group Ethnicity) totvars(GenderTotal AgeTotal EthnicityTotal) generate(rakedweight)

[Tom Prosser]

Would be so so grateful for help with this, if anybody is feeling generous

[Chen Samulsion]

I think you can compare your result with other command, thus verifing if it is correct. The -ipfweight- command below is from SSC.

ipfweight Gender Age_group Ethnicity, values(0.49 0.51 0.12 0.17 0.17 0.16 0.38 0.63 0.12 0.06 0.11 0.08) gen(ipfweight) maxiter(20)
corr rakedweight ipfweight

. corr rakedweight ipfweight
(obs=200)

             | rakedw~t ipfwei~t
-------------+------------------
 rakedweight |   1.0000
   ipfweight |   1.0000   1.0000

[Tom Prosser]

Thanks so much! I appreciate it immensely :-)