The DHS Program User Forum: Sampling » Working with Wealth Index Quintiles

Home » Data » Sampling » Working with Wealth Index Quintiles

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

Working with Wealth Index Quintiles [message #2991]

Mon, 29 September 2014 10:20

Ryan
Messages: 6
Registered: May 2013
Location: Washington

Member

I am working with the wealth index quintiles and trying to exploit the fact that, by definition, each quintile captures an equal share of the population. It seems to me that I should be able to take the (weighted) count of individuals in the poorest quintile in each region and divide that by the (weighted) sum of all individuals in the poorest quintile to get the fraction of the poorest quintile living in each region in a county. This would be very useful for me, because I could then make statements about where people of different wealth levels live inside a country. The figures would be comparable between, for example, richest and poorest because each group represents 20% of the population.

Before I do this, I was wondering if I was making any conceptual mistakes in my treatment of the weights. I was also trying to think through if these calculations should be done on the individuals or household dataset. I think I need to work with the individuals, but presumably then I also need to merge the male and female datasets?

Any advice on this general idea would be appreciated.

Thank you

Report message to a moderator

Re: Working with Wealth Index Quintiles [message #3030 is a reply to message #2991]

Mon, 06 October 2014 10:29

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Following is a response fron Senior DHS Specialist, Shea Rutstein:

The quintiles are based on one-fifth of the dejure household population, estimated by using the household weight times the number of usual household members. It is different from the number of individuals interviewed since it includes children under age fifteen and persons older than 49 as well as interviewed women and men. Tabulate households by quintile by region, weighted by hhmemwt, which is hv005*HV012. That should give you what you want.

Report message to a moderator