What is the difference between v005 and sweight DHS India? [message #25217] |
Sun, 18 September 2022 06:07 |
teeb
Messages: 11 Registered: July 2022
|
Member |
|
|
In DHS India, the BR files have two weights: v005 (sample weight) and sweight(state samp weight). Same for HR files. Could anyone explain in what ways these weights are different? Is the hv005 weight in HR file same as v005 weight in BR file?
|
|
|
Re: What is the difference between v005 and sweight DHS India? [message #25226 is a reply to message #25217] |
Mon, 19 September 2022 08:52 |
Bridgette-DHS
Messages: 3203 Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS staff member, Tom Pullum:
The weights in the IR and MR files (v005 and mv005) are similar to those in the PR and HR files (h005) but are adjusted for nonresponse by women and me, respectively, and renormalized to have means of 1 (and then multiplied by 1000000 to avoid a decimal point).
In the India PR files, the only difference between the national weight (hv005) and the state weight (shweight) is that the national weight has a mean of 1 in the entire PR file and the state weight has a mean of 1 within each state (ignoring the factor of 1000000). If you do "correlate hv005 shweight if hv024==1", for example, you get a correlation of exactly 1.
The only time when you would use shweight is when doing an analysis of a specific state. And even when doing a within-state analysis, with Stata you will get the same results with hv005 and shweight. Stata automatically re-normalizes weights (at least with pweight and iweight) to have a mean of 1.
I never use (and have never needed) shweight. I think the India surveys are the only DHS surveys that include separate weights for the level 1 administrative units.
|
|
|