The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Problem Generating A_h; M_h & M (Approximating Level 1 & 2 Weights - Issues)
Problem Generating A_h; M_h & M [message #27352] Tue, 25 July 2023 01:35 Go to next message
Yawo is currently offline  Yawo
Messages: 45
Registered: February 2019
Member
Hello:

I continue to make good progress on approximating level 1 & 2 weights for my datasets. However, I have encountered two main challenges

1. Unable to Generate M: The following datasets did not report the number of households in sampling frame:

-- Burkina Faso 2010
-- Cameroon 2018
-- Congo Democratic Republic 2013
-- Chad 2014
-- Cote d'Ivoire 2011
-- Gabon 2019
-- Kenya 2008
-- Liberia 2013
-- Mali 2012
-- Niger 2012
-- Eswatini/Swaziland 2006
-- Togo 2013

2. Missing A_h & M_h: The following countries did not report number of EA and/or the average size of households needed to calculate M_h and/or A_h. :

-- Congo Democratic Republic 2013
-- Kenya 2008
-- Mali 2012
-- Eswatini/Swaziland 2008

I combed through the reports and unable to find the appropriate values, and therefore unable to generate M, A_h and M_h as required for the approximation.

Are there some other sources for these values, or a different way of generating them. Thanks in advance for your assistance and advice on how to proceed.

cheers, cY

Re: Problem Generating A_h; M_h & M [message #27355 is a reply to message #27352] Wed, 26 July 2023 08:27 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3188
Registered: February 2013
Senior Member

Following is a response from Senior DHS staff member, Tom Pullum:

Good to hear that you are making progress, but DHS no longer has staff who can respond to questions about calculating multi-level weights. For some specific surveys, the information is simply not available. Keep in mind that the procedure described in Methodological Report #27 only gives estimates of the true level weights, because it never happens that all the information needed to calculate the true weights is available. When some types of data required by the procedure are not available, it should be possible to simplify the procedure further to adapt to less and less complete data.

It is inefficient for users to have to calculate the level weights independently. I hope that later in 2023 the DHS analysis team will be able to coordinate the compilation of these estimates and make them centrally available.
Re: Problem Generating A_h; M_h & M [message #27356 is a reply to message #27355] Wed, 26 July 2023 11:04 Go to previous messageGo to next message
Yawo is currently offline  Yawo
Messages: 45
Registered: February 2019
Member
Thanks so much, Tom and Bridgette: I appreciate your quick response. Given that we are approximating level weights here, which would be a better option? Should I just assume a weight2==1 for those countries missing complete information? Or substitute the "number of selected households" for "the number of households in the sampling frame"?

Thanks - CY
Re: Problem Generating A_h; M_h & M [message #27360 is a reply to message #27356] Thu, 27 July 2023 08:16 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3188
Registered: February 2013
Senior Member

Following is a response from Senior DHS staff member, Tom Pullum:

I do not feel comfortable recommending either of those options at this time. I suggest that you run some analyses with one option. Then repeat with the other option and see whether the results are different. Select the option which gives more conservative results, that is, coefficients that are closer to zero and less significant. Please stay tuned because I hope we can come up with some better defaults soon.
Previous Topic: Sex ratio at 2nd birth order
Next Topic: weighting urban rural divide , multiple countries years
Goto Forum:
  


Current Time: Thu Nov 7 11:16:57 Coordinated Universal Time 2024