The DHS Program User Forum: Sampling » AREG Model and Svyset

Home » Data » Sampling » AREG Model and Svyset

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

AREG Model and Svyset [message #12121]

Wed, 29 March 2017 16:14

Nayab_M
Messages: 3
Registered: March 2017
Location: Pakistan

Member

I am using the Pakistan DHS (2012-13) and trying to run an AREG model. Apparently the svyset command does not work with this. Is there an alternative command that would work. I can account for weights and the clustering but the stratum option only works with svyset as pointed out by Tom Pullum in this thread http://userforum.dhsprogram.com/index.php?t=msg&th=4908& amp;goto=9278&S=Google

If I choose to ignore the stratum stratification but accounted for the weights and clustering, would it impact my results? The AREG model is essential since I am including regional effects for the 471 psus in this dataset.

Any help would be appreciated.

Thank you.

Report message to a moderator

Re: AREG Model and Svyset [message #12126 is a reply to message #12121]

Thu, 30 March 2017 13:50

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Following is a response from Senior DHS Stata Specialist, Tom Pullum:

I'm not familiar with "areg" but it appears to be the same as usual regression with "i.". For example, the syntax would be "svy: regress v201 i.v013". If you have lots of categories, as with v001, then you maay need to reset matsize, as with "set matsize 10000" or some other larger number instead of 10000. You could also consider a mixed model with random effects (a random intercept) for v001.

Report message to a moderator

Re: AREG Model and Svyset [message #12128 is a reply to message #12126]

Thu, 30 March 2017 21:01

Nayab_M
Messages: 3
Registered: March 2017
Location: Pakistan

Member

Thank you so much for your help.
I wanted to ask which is the correct strata variable in Pakistan DHS 2012-13, v023 or v022. I used this command:

svyset [pweight=weight], psu(v021) strata(v023)

v023 is simply provinces classified as rural or urban while, v022 are 78 strata (numbered), so which one is recommended in the svyset command?

If v023 is the correct variable and I use it and I include province level dummies in the regression, would the inclusion of the province dummies be rendered useless?

Your assistance will be very valuable to me.
Thank you.

Report message to a moderator

Re: AREG Model and Svyset [message #12170 is a reply to message #12128]

Tue, 04 April 2017 17:07

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Following is a response from Senior DHS Sampling Specialist, Ruilin Ren:

The labels for the variables V022 and V023 are confusing. The right variable for stratification for the Pakistan 2012 survey was given by V022. However there are less than 78 strata because the numbering is not continuous.

Report message to a moderator

Previous Topic:	Set up DRC data in Stata
Next Topic:	District-level aggregates

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Wed Nov 27 07:19:12 Coordinated Universal Time 2024