The DHS Program User Forum: Nutrition and Anthropometry » PMT INDEX SCORE TO WEALTH INDEX SCORE/WEALTH INDEX

Home » Topics » Nutrition and Anthropometry » PMT INDEX SCORE TO WEALTH INDEX SCORE/WEALTH INDEX

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

PMT INDEX SCORE TO WEALTH INDEX SCORE/WEALTH INDEX [message #22658]

Sat, 17 April 2021 19:36

Mister Denis
Messages: 3
Registered: April 2021

Member

Hi Members
I have a problem of converting Proxy means test scores (ranging from around 7.0 scores implying poorest or very poor to around 16 implying rich) to wealth index similar to that of DHS to reflect the same thing Poorest to richest. However, the reverse conversion can also suit the purpose. Any one with an idea please.
With thanks
Denis

Report message to a moderator

Re: PMT INDEX SCORE TO WEALTH INDEX SCORE/WEALTH INDEX [message #22670 is a reply to message #22658]

Mon, 19 April 2021 14:33

Bridgette-DHS
Messages: 2265
Registered: February 2013

Senior Member

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

The DHS staff cannot help with this posting. Other forum users are encouraged to respond.

Report message to a moderator

Re: PMT INDEX SCORE TO WEALTH INDEX SCORE/WEALTH INDEX [message #22671 is a reply to message #22658]

Mon, 19 April 2021 14:44

Reduced-For(u)m
Messages: 292
Registered: March 2013

Senior Member

One very simple way would be to use the DHS wealth quintiles and then put the proxy means-test scores into quintiles as well and just use those (which quntile of the outcome is the household in)...they would be truly comparable if and only if the two measures generate the same ranking of households (from poorest to richest). This is surely not true (they will rank differently for reasons of design and reasons of noise), but you could then go an check some aspects of the data to see how similar they are.... say by looking at the distribution of other household characteristics/assets across quintiles (if, say, you have e.g. maternal education and flooring material in both surveys you could check the proportion of "highly educated" or "has concrete floor" in each quintile in each survey).

But I don't think there is every going to be any perfect way to compare the two, unless maybe they somehow ask for exactly the same information in both surveys and then you could generate your own index using both surveys at the same time.

Hope that sorta helps.

Report message to a moderator