The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Domestic Violence » India DHS 2019-21: Mismatch in number of observations (India DHS 2019-21: Mismatch in number of observations)
India DHS 2019-21: Mismatch in number of observations Sun, 29 May 2022 02:32
 Anjana Messages: 3Registered: April 2020 Member
Dear team,

I am working with the India DHS (NFHS) 2019-21 on the domestic violence module. I encounter a problem when trying to match with Table 15.9 of the national report, i.e. Percentage of ever-married women age 18-49 who have experienced various forms of violence committed by their husband ever or in the 12 months preceding the survey, India 2019-21.

The table in the report mentions number of observations to be 62,381. However, I get the number of observations to be 63,815. The unweighted command I use is the following:

ta phy_viol if v044==1 & v502~=0 & v012>=18 & v012<=49

Please let me know if there is any error in the above. Thank you.
Re: India DHS 2019-21: Mismatch in number of observations [message #24564 is a reply to message #24547] Wed, 01 June 2022 09:06
 Janet-DHS Messages: 150Registered: April 2022 Senior Member
Following is response from DHS Research & Data Analysis Director, Tom Pullum:

The numbers given in the final report are weighted. You are close enough that the difference could just be due to weighting. Please add "[iweight=v005/1000000]" to your command and see if you get a match.
Re: India DHS 2019-21: Mismatch in number of observations [message #24587 is a reply to message #24564] Sat, 04 June 2022 02:57
 Anjana Messages: 3Registered: April 2020 Member

Following the suggestion, I added the weight as follows:

ta phy_viol [iweight=v005/1000000] if v044==1 & v502~=0 & v012>=18 & v012<=49

But the total number of observations after weighting comes to 64,061.945, which is also different from the number mentioned in the report (62,381).

Please let me know any suggestions in this regard. Thanks.
Re: India DHS 2019-21: Mismatch in number of observations [message #24619 is a reply to message #24587] Wed, 08 June 2022 11:49
 Janet-DHS Messages: 150Registered: April 2022 Senior Member
Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

For any table that includes variables from the domestic violence module, you should use d005 as the weight variable. You may be able to get a match if you use d005, rather than v005.
Re: India DHS 2019-21: Mismatch in number of observations [message #24636 is a reply to message #24619] Sun, 12 June 2022 13:42
 Anjana Messages: 3Registered: April 2020 Member
Yes, I was able to get the correct figures now. Thank you for your help.
 Previous Topic: Can I use R to open files? Next Topic: Discrepancies in results
Goto Forum:

Current Time: Tue Aug 9 09:31:28 Coordinated Universal Time 2022