The DHS Program User Forum: Weighting data » create a weighted population variable

Home » Data » Weighting data » create a weighted population variable

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

create a weighted population variable [message #24504]

Wed, 25 May 2022 06:49

marghetambu
Messages: 1
Registered: May 2022

Member

Hi,

I merged all the individual files(men and women) of all the African countries. I created a variable "perwt" which containes the variable v005/1000000 for women and mv005 (reweighted using the procedure described here https://userforum.dhsprogram.com/index.php?t=msg&goto=13 158&S=Google).
Moreover, I generated a variable "population" equal to 1 for all the individuals.
I aggregate all the observations at province/year level and I create the variable "population_sum" as the weighted sum (population*weights). However, since the weights are decimals, the aggregate population variable obtained contains decimal numbers.
How can I obtain a weighted variable "population" which calculates all the individuals interviewed in a specific province and in a specific year?
Thank you so much.

Have a nice day!

Report message to a moderator

Re: create a weighted population variable [message #24526 is a reply to message #24504]

Thu, 26 May 2022 16:39

Janet-DHS
Messages: 938
Registered: April 2022

Senior Member

Following is response from DHS Research & Data Analysis Director, Tom Pullum:

I'm not clear what you mean by "a weighted variable "population" which calculates all the individuals interviewed".... The number of people interviewed is the unweighted number in the sample. Are you looking for an estimate of the population? For that, I would recommend that you get the estimated national population from the UN Population Division website (https://population.un.org/wpp/), and multiply that by the weighted proportion in the DHS sample in the province. The UN estimates are for July 1 of a specified year. The DHS estimates can be assigned to the median date of interview. You may want to interpolate.

Report message to a moderator

Previous Topic:	M_h data for Myanmar, M_h & M data for Pakistan
Next Topic:	Weighting after pooling multiple countries

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Wed Oct 22 10:09:39 Coordinated Universal Time 2025