| Clarification on applying sampling weights in Stata using India NFHS-5 [message #30908] |
Thu, 17 April 2025 14:18  |
Withadel
Messages: 2
Registered: April 2025
|
Member |
|
|
I am currently working with the India NFHS-5 Individual Recode (IR) dataset for a project analyzing maternal health service utilization. While going through the guide on sampling weights, I still have some doubts and would appreciate any clarification.
Context of my analysis:
Dataset: IAIR7DFL.DTA
Population of interest: women aged 1549 who had a live birth in the past 5 years
Main outcomes: number of ANC visits (categorical), institutional delivery (binary)
Software: Stata 17
I understand from the DHS documentation that v005 should be used as the weight variable, and that it needs to be divided by 1,000,000 before applying. However, I am still confused about how to properly specify this in the svyset command in Stata, and how clustering and stratification should be handled for India, which has a complex sampling structure.
I came across some posts and papers suggesting using v023 as the stratification variable instead of v022, especially for India. Could someone confirm which one is appropriate here?
Given that NFHS-5 has state-specific sampling frames and uses oversampling in some states/urban areas, is there any adjustment needed beyond this basic svyset structure? Or does DHS already account for that in v005?
|
|
|
|
|
|
|
|
|
|
| Re: Clarification on applying sampling weights in Stata using India NFHS-5 [message #31072 is a reply to message #30908] |
Wed, 11 June 2025 22:21  |
miaowschool
Messages: 1
Registered: June 2025
|
Member |
|
|
v022 vs v023 is a common source of confusion. For the India DHS (NFHS) dataset, `v023` is recommended as the stratification variable instead of `v022`. `v022` is the standard DHS stratification variable, but in the case of India, due to state and urban/rural sampling, `v023` more accurately captures the actual strata used in the sample design. So your `svyset` command should be:
svyset [pw=weight], psu(v021) strata(v023)
Survival Race
|
|
|
|
| Re: Clarification on applying sampling weights in Stata using India NFHS-5 [message #31108 is a reply to message #31072] |
Mon, 30 June 2025 05:42 |
nytwordlehints
Messages: 1
Registered: June 2025
|
Member |
|
|
For India NFHS-5, v021 is the PSU (cluster), and v022 is the standard stratification variable (usually urban/rural within state). Although some suggest v023, DHS recommends v022 for stratification.
The weight v005 already accounts for oversampling and complex design, so no further adjustment is needed. Just ensure to subset your sample (women 15-49 with a live birth in 5 years) before analysis.
|
|
|
|
Members
Search
Help
Register
Login
Home
