The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Wealth Index » sub-national wealth quintiles in STATcompiler
sub-national wealth quintiles in STATcompiler [message #23690] Sat, 06 November 2021 05:16 Go to next message
rame is currently offline  rame
Messages: 1
Registered: November 2021
Dear Sir/Madam
I am doing DHS data analysis. According to my research, I need to use sub-national (provincial-level) poverty rate data to verify my model's prediction which is established with DHS survey data like AOHR71FL for Angola.
I found that STATcompiler provide the wealth quintiles by provincial-level, I think these data maybe can be used as verification data. I would appreciate if you could offer me more information about provincial-level wealth quintiles in STATcompiler.

a)Is there any document explain the method to get the wealth quintiles at provincial-level ?
b)How do wealth quintiles calculated with DHS dataset?
c)In SAR20 report, DHS also provide district-level data. Has DHS verified the accuracy when sum all district-level data to provincial-level data? Does one provincial poverty rate is exactly equal to ∑(districts in this province)?

Would appreciate guidance in this regard.

[Updated on: Sat, 06 November 2021 21:51]

Report message to a moderator

Re: sub-national wealth quintiles in STATcompiler [message #23701 is a reply to message #23690] Tue, 09 November 2021 08:58 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3117
Registered: February 2013
Senior Member

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

The wealth quintiles (hv270 in the HR and PR files) are calculated from the continuous index (hv271) by identifying four specific values of hv271, the boundaries between adjacent quintiles, that will produce the same number of weighted de jure household members in each quintile. You could apply the same procedure to subpopulations, but that may not really meet your needs.

I suggest the following. You would EXPECT 20% of the weighted de jure household members in each of the subnational units to be in the bottom quintile. You could easily calculate the OBSERVED number. The ratio of observed to expected would be a good aggregate indicator of poverty. Alternatives would be to use the bottom TWO wealth quintiles. You could do urban and rural areas separately.

The wealth quintiles describe the relative position of the households at the time of the survey in terms of household assets and services. They don't really specify a poverty line or poverty rate.

Previous Topic: Comparative Wealth Index - Madagascar
Next Topic: Problem with recreating Asset quintiles
Goto Forum:

Current Time: Wed Jul 24 16:35:00 Coordinated Universal Time 2024