The DHS Program User Forum: Weighting data » Multilevel model with region random effects

Home » Data » Weighting data » Multilevel model with region random effects (Multilevel model with region random effects)

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

Re: Multilevel model with region random effects [message #26217 is a reply to message #26208]

Fri, 17 February 2023 08:54

Bridgette-DHS
Messages: 3202
Registered: February 2013

Senior Member

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

In the setup for a multilevel model, level 1 would be the household and level 2 would be the psu (cluster). In 2022 we prepared a methodological report describing how to do this (https://www.dhsprogram.com/pubs/pdf/MR27/MR27.pdf). There have been several postings on the user forum which will help with this subject.

However, if your higher-level variable is (or variables are) at the level of the region, I don't think you need a multilevel model. If you had cluster-level variables, yes, but region is not part of the sample design. An individual-level model should be sufficient.

Report message to a moderator

[Message index]

		Multilevel model with region random effects By: sophia614 on Thu, 16 February 2023 09:08
		Re: Multilevel model with region random effects By: Bridgette-DHS on Fri, 17 February 2023 08:54
		Re: Multilevel model with region random effects By: Aline Semaan on Wed, 29 March 2023 19:59
		Re: Multilevel model with region random effects By: Bridgette-DHS on Fri, 31 March 2023 07:57

Previous Topic:	Weighting in IV approach
Next Topic:	Why I am getting different total observations when using iweight for tabulating a variable

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Mon Dec 2 09:04:25 Coordinated Universal Time 2024