The DHS Program User Forum: Domestic Violence » MULTILEVEL WEIGHT FOR DOMESTIC VIOLENCE MODULE

Home » Topics » Domestic Violence » MULTILEVEL WEIGHT FOR DOMESTIC VIOLENCE MODULE

Show: Today's Messages :: Show Polls :: Message Navigator

MULTILEVEL WEIGHT FOR DOMESTIC VIOLENCE MODULE [message #30582]

Wed, 01 January 2025 22:25

wavogo
Messages: 14
Registered: December 2015
Location: Normal, Illinois

Member

Dear DHS,
I am working on empowerment and Intimate Partner Violence (IPV) and have appended data from nine countries in West Africa (Benin, Burkina-Faso, Ivory Coast, Gambia, Ghana, Mali, Nigeria, Liberia, and Sierra Leone) that conducted the domestic violence module. As recommended, I used DHS coding to account for complex survey design as follows by setting up the svyset command in each country before appending.
**Set up weights
**Add PSU variable
gen psu = v021
**Add strata variable
egen strata=group(v025 v024)
**Generate weight and all-woman weight to use for ever-married samples
gen wt=d005/1000000
svyset [pweight=wt], psu(v021) strata(v023) singleunit(centered)

However, I am interested in multi-level logistic regression models of IPV experiences.
Level-1 (individual-level) and Level-2 (cluster-level) weights by de-normalizing the weights and creating adjusted weights based on a specific normalization factor.

I have used the recommended code in https://www.dhsprogram.com/pubs/pdf/MR27/MR27.pdf that was posted in the DHS forum by Bridgette-DHS on Friday, October 29, 2021 at 15:42 but get errors e.g. "gen d_HH = DHSwt * (M/m_c)
m_c not found"
Can you please help me figure out what I need to correctly run this code to create the level weights?

Thank you so much and standing by to clarify my challenges if you have questions.

Report message to a moderator

Re: MULTILEVEL WEIGHT FOR DOMESTIC VIOLENCE MODULE [message #30591 is a reply to message #30582]

Thu, 02 January 2025 14:13

wavogo
Messages: 14
Registered: December 2015
Location: Normal, Illinois

Member

Dear DHS,

I wanted to add that the reason the svy: logit models are insufficient for me is that I have aggregated individual measures e.g. education, employment, and poverty within each Primary Sampling Unit (PSU)- which I term "community characteristics".

However, reviewers are concerned that this assumes that PSU is homogenous in terms of for example education levels, which might that be the case as there might be intra-community differences leading to potentially inaccuracies in the representation of community-level characteristics.

I need help with setting up weights that account for (1) the survey design effects of DHS (2) the fact that the domestic violence module was not administered to all women (3) multi-level weights that adjust for intra-community (PSU) differences.

Any insights and guidance and STATA coding to help accomplish this is very gratefully appreciated.

Thank you!

Report message to a moderator

Re: MULTILEVEL WEIGHT FOR DOMESTIC VIOLENCE MODULE [message #30606 is a reply to message #30591]

Tue, 07 January 2025 08:26

Janet-DHS
Messages: 917
Registered: April 2022

Senior Member

Following is a response from DHS staff member, Tom Pullum:
If you look at the Stata code, just a few lines before "gen d_HH = DHSwt * (M/m_c)" you will find this line: "* m_c total number of completed households (added from the HR dataset)".

To get m_c for a specific survey, open the HR file and enter "codebook hv000". The output will include one number, and that's the number you want.

The instructions to get the level weights for the dv weights are also given in MR27, on the bottom of page 8. I'm afraid that it will be a fair amount of work, with virtually no impact on your results.

Report message to a moderator

Previous Topic:

Partner's Alcohol Consumption in Women's Individual file

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Tue Jan 14 11:37:47 Coordinated Universal Time 2025