The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Multilevel model for SPA data
Multilevel model for SPA data [message #26134] Fri, 10 February 2023 10:12 Go to next message
Lucia is currently offline  Lucia
Messages: 3
Registered: April 2020
Member
Hello, I am conducting an analysis of the labor and delivery observations using data from SPAs conducted in Kenya 2010, Malawi 2013-14, and Nepal 2021.
I want to carry out a multilevel mixed-effect model for each country, however, I have read different opinions about it in the forum.

I saw that some people suggest setting the clwgt = 1 and just using the fcwgt for the multilevel model, this is because the clwgt already takes into account the fcwgt.
Others suggest using both weights with this code: svyset inv_id, weight(fcwgt) strata(strata) singleunit(centered) ||_n, weight(clwgt).

I also saw that there is a document that explains how to approximate level weights, however, it is for DHS data and not SPA data.


I was wondering if you could help me in understanding which method should I use, and if the document on DHS data is applicable to SPA data.


many thanks,
Kind regards,
Lucia
Re: Multilevel model for SPA data [message #26171 is a reply to message #26134] Tue, 14 February 2023 09:50 Go to previous message
SaraDHS is currently offline  SaraDHS
Messages: 44
Registered: December 2020
Member
Hi Lucia,

So first, the document on approximating the level weights for DHS data is not applicable to the SPA data since the two surveys have very different sampling approaches.

Second, in general, and especially in this case since you are using the labor and delivery observations, I would recommend using facility weight as the first level weight, and assume equal weight for all clients (clwt=1) in the same facility. Although the latter assumption is not always true, in the case of the three surveys you are using, in over 90% of facilities ALL labor and delivery clients at the facility were included in the observation, so it is pretty close in these cases.

I hope this is helpful,

Sara


Sara Riese, PhD Senior Demographic and Health Researcher, DHS Program
Previous Topic: Sex ratio at birth using IR file , India, NFHS-5
Next Topic: Melogit and Weights
Goto Forum:
  


Current Time: Fri Mar 29 07:32:52 Coordinated Universal Time 2024