Why can't individual WI data and values be released [message #25438] |
Wed, 19 October 2022 04:36 |
Marco Stamazza
Messages: 3 Registered: October 2022
|
Member |
|
|
I see that many of us are trying to reproduce the wealth indices from past surveys. It is painful and time consuming.
I don't understand why the values (and possibly the recoded data) can't be made available. It would save us a lot of time and likely mistakes.
Is there a way for asking for these files?
|
|
|
|
|
Re: Why can't individual WI data and values be released [message #25496 is a reply to message #25475] |
Fri, 28 October 2022 12:57 |
Janet-DHS
Messages: 899 Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member Tom Pullum:
The wealth index is based on household-level variables and is the same for every person in the same household. Your use of the word "individual" is a little misleading.
The continuous index, hv271, is used to construct the quintiles, hv270 for all households, hv270a for urban households, and hv270b for rural households. The difference between the three versions of hv270 is just in the specification of the 4 cutpoints that divide the continuous index into 5 pieces. The bottom quintile, for example, applies to the households that have the lowest values of hv271, up the scale to include the bottom 1/5 of weighted de jure household members. For hv270a the application is to urban households only, and for hv270b to rural households only. There is no use of regression. Stata includes a "pctile" command to find the cutpoints.
The variables that are used to construct hv271, with principal components analysis, vary somewhat from one survey to another but are specified on the website for each survey. Not sure why you would want to reconstruct the wealth index. Yes, it's a lot of work.
|
|
|
|
Re: Why can't individual WI data and values be released [message #25604 is a reply to message #25497] |
Wed, 16 November 2022 09:15 |
Janet-DHS
Messages: 899 Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member Tom Pullum:
I just checked, and it appears that this survey was typical of DHS surveys. A variable hv117 in the PR file is "eligibility for the women's interview". It is 1 if the woman is de facto (i.e. available for the interview) and age 15-49. I constructed a check for the coding of hv117
gen hv117_test=0
replace hv117_test=1 if hv103==1 & hv104==2 & hv105>=15 & hv105<=49
tab hv117_test hv117,m
The "test" variable matches hv117 perfectly. There don't seem to be any other criteria. Some of the eligible women will not actually appear in the IR file, usually because they refused or they were not (after all) available to meet with an interviewer. But there doesn't seem to be any systematic reason for eligible women to be missing from the IR file. Hope this answers your question. If not, let me know.
|
|
|