The DHS Program User Forum: India » NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis

Home » Countries » India » NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis

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

NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis [message #2134]

Wed, 07 May 2014 10:45

drsanmis
Messages: 2
Registered: May 2014

Member

Hi,
I am looking at a categorical outcome in the NFHS 3 data. Do I need to use PROC LOGISTIC or PROC SURVEY LOGISTIC in SAS. ( Published papers have used either of them).
If proc survey logistic is used the syntax requires using STRATA and CLUSTER statements. I am not sure which variables in the dataset signify strata and cluster.
It will be very helpful if someone can guide me on this.

Report message to a moderator

Re: NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis [message #2167 is a reply to message #2134]

Thu, 15 May 2014 08:58

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Our SAS Specialist is reviewing your posting, and we will get back to you soonest.

Thanks.

Report message to a moderator

Re: NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis [message #2169 is a reply to message #2134]

Thu, 15 May 2014 09:21

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Following is a response from our Senior SAS Specialist, Ruilin Ren.

For the use of the logistic regression, it really depends on your purpose of analysis. If you are interested in exploring patterns of the survey data themselves, but do not intend to extrapolate the entire survey population, you can use proc logistic. But if your intention is to explore patterns among the entire survey population, you need to use proc survey logistic. In the later case, you need to declare CLUSTER (variable HV001 at household level, V001 at individual level), STRATA (HV022 at household level, V022 at individual level). You also need to declare the sampling weight WEIGHT (HV005 divided by 1000000 at household level, and V005 divided by 1000000 at individual level).

Report message to a moderator

Re: NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis [message #2660 is a reply to message #2169]

Sat, 02 August 2014 22:22

ABLR
Messages: 10
Registered: August 2014

Member

Just to doublecheck, the strata variable would be v022 and not v023 or a combination of v024 and v025? Isn't it unusual in DHS data for v022 to be the strata variable. [I'm using Stata, but I assume this wouldn't differ across software.] Thanks for checking!

Report message to a moderator

Re: NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis [message #2739 is a reply to message #2660]

Tue, 19 August 2014 11:45

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Another response from Ruilin Ren:

As for stratification variables for NFHS-3, the stratification should be HV024 (V024) crossing HV025 (V025). HV022 (V022)is not the stratification variable.

Report message to a moderator

Re: NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis [message #6751 is a reply to message #2739]

Thu, 09 July 2015 06:20

Kathryn Kershaw
Messages: 3
Registered: July 2015
Location: London, UK

Member

Hi,

My query is clarification of this one. In Stata I used the following commands to survey set my data:

gen wgt = v005/1000000
svyset v021 [pweight = wgt], strata(v023)

But I found this thread and thought maybe I should create a new strata variable:

egen strata = group(v024 v025)

This produces a different variable from v023 with separate groups for the rural and urban aspects of each state, which v023 doesn't have.
Could you clarify which one I should use please?

Many thanks!

Kathryn

Report message to a moderator

Re: NFHS 3 data - what variables define STRATA and CLUSTER for SAS analysis [message #6755 is a reply to message #6751]

Thu, 09 July 2015 17:15

Reduced-For(u)m
Messages: 292
Registered: March 2013

Senior Member

Check out the country report for the year/country you are using. It should tell you how the sampling was done (see the intro or the appendix). From that, you should be able to determine which the correct stratification is - it is not the same for every country, but my experience is that the stratification by regionXurban [group(v024 v025)] is the most commonly used.

Report message to a moderator

Previous Topic:	2014-2015 DHS
Next Topic:	Village codes for educational facility in 1992 and 1998

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Sat Nov 23 11:22:19 Coordinated Universal Time 2024