The DHS Program User Forum: Weighting data

Home » Data » Weighting data » Weighting in IV approach (Egypt - panel data - 2005-2008-2014)

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

Re: Weighting in IV approach [message #26513 is a reply to message #26507]

Tue, 28 March 2023 09:19

Bridgette-DHS
Messages: 3230
Registered: February 2013

Senior Member

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

We recommend always using weights, and also the adjustments for clustering and stratification, which have been discussed in many forum posts. The estimates change slightly when you use weights, and the standard errors change slightly when you adjust for clustering and stratification. The weights tend to increase the standard errors and occasionally a coefficient will be significant when the weights are not used and non-significant when the weights are used. When this happens it is especially important to make the adjustments for clustering and stratification (as you are doing with svyset).

I hesitate to describe successive surveys in the same country as panels. The surveys are independent cross-sections. The individuals, households, and clusters are different in every survey.

You are on your own when it comes to including instrumental variables in your analysis. I hope you are confident in making the required assumptions. I am not familiar with "ivregress" but if it does not allow svy then I'd be very cautious in the interpretation of borderline significance. "Cautious" means interpreting a borderline value as "not significant", even if it's disappointing.

Report message to a moderator

[Message index]

		Weighting in IV approach By: Alanood on Mon, 27 March 2023 16:56
		Re: Weighting in IV approach By: Bridgette-DHS on Tue, 28 March 2023 09:19
		Re: Weighting in IV approach By: Alanood on Tue, 28 March 2023 17:35
		Re: Weighting in IV approach By: Alanood on Tue, 28 March 2023 19:40
		Re: Weighting in IV approach By: Bridgette-DHS on Wed, 29 March 2023 08:47
		Re: Weighting in IV approach By: Alanood on Thu, 30 March 2023 06:37

Previous Topic:	Weighting Combined Individual Data for Logistic Regression Analysis
Next Topic:	Multilevel model with region random effects

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu Feb 27 21:23:54 Coordinated Universal Time 2025