The DHS Program User Forum: Unmet Need » Unable to replicate Statcompiler results for Kenya

Home » Topics » Unmet Need » Unable to replicate Statcompiler results for Kenya

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

Unable to replicate Statcompiler results for Kenya [message #27697]

Thu, 21 September 2023 12:43

lsenderowicz
Messages: 2
Registered: September 2023

Member

I am calculating unmet need for all women for seven countries: Burkina Faso, Chad, CIV, DRC, Kenya, Nigeria, and Uganda for an analysis I'm running. As a data/code check, I'm comparing the results that I'm getting by running the unmet need code (for Stata, copied and pasted directly from Github), to results from Statcompiler, and the results for six of the seven countries are right on target.

For Kenya (2014), however, the results are way off. When we run the code (again, copied directly from the DHS' Github page) on the Kenya data, we get an "all women" unmet need of 5.93%, while Statcompiler gives an "all women" unmet need of 13% for that country and year.

Since we're applying DHS code directly to unedited DHS data, we cannot figure out where this discrepancy might be coming from. And the fact that we're not running into this same issue for any of the other six countries makes it all the more puzzling.

Any insight folks might be able to provide about the nature of this discrepancy would be very much appreciated!

Report message to a moderator

Re: Unable to replicate Statcompiler results for Kenya [message #27841 is a reply to message #27697]

Mon, 09 October 2023 18:08

Trevor-DHS
Messages: 803
Registered: January 2013

Senior Member

The Kenya 2014 DHS used a split sample with a long and a short questionnaire. Unmet need can only be calculated for this survey for those cases that used the long questionnaire. The easiest way to test for the long questionnaire is to use v044 != .

If you try selecting for v044 != . then I believe you should be able to match. Check the denominator to check you are matching the denominator given in STATcompiler or the report.

Report message to a moderator

Re: Unable to replicate Statcompiler results for Kenya [message #27844 is a reply to message #27841]

Tue, 10 October 2023 12:19

lsenderowicz
Messages: 2
Registered: September 2023

Member

This is really helpful -- thank you so much Trevor!

Report message to a moderator

Previous Topic:	Mauritania
Next Topic:	HOW TO CALCULATE CONTRACEPTIVE DISCONTINUATION VARIABLE

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Nov 21 16:41:55 Coordinated Universal Time 2024