The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » General » stratification, sample weights and clustering in pooled ata
stratification, sample weights and clustering in pooled ata [message #15664] Tue, 28 August 2018 20:51 Go to next message
pramesh is currently offline  pramesh
Messages: 39
Registered: November 2015
Location: australia
Member
Hi,

I am trying to do the analysis by pooling NDHS for the year 2006, 2011, and 2016.

Since I was looking at the data, i found similar sampling stratification used in 2001, 2006, and 2011. However, in 2016 this has been changed.


Since my analyses is pooled NDHS 2001-2016, i want to know if my coding for taking (stratification, sample weights and clustering) into account in the pooled dataset is correct. Below is the coding i have used:

gen weight = v005 / 1000000
egen stratid = group (region v102), label
egen wave_psu = group (v001), label
svyset wave_psu [pweight=weight], strata(stratid)


Thak you for your help.

P
Re: stratification, sample weights and clustering in pooled ata [message #15673 is a reply to message #15664] Wed, 29 August 2018 09:26 Go to previous messageGo to next message
Liz-DHS
Messages: 1381
Registered: February 2013
Senior Member
Dear User, you are referring to Nepal?
Re: stratification, sample weights and clustering in pooled ata [message #15679 is a reply to message #15673] Wed, 29 August 2018 15:13 Go to previous messageGo to next message
pramesh is currently offline  pramesh
Messages: 39
Registered: November 2015
Location: australia
Member
Dear Liz,

Sorry I did not mention it before. Yes, this is for Nepal.

Thank you.
Re: stratification, sample weights and clustering in pooled ata [message #15711 is a reply to message #15664] Tue, 04 September 2018 12:30 Go to previous message
Liz-DHS
Messages: 1381
Registered: February 2013
Senior Member
A response from Senior Sampling expert, Dr. Mahmoud Elkasabi:
Quote:

I assume the user is interested in pooling the data to explore data trends where every survey is represented as one point. In this case, you can use v023 as stratification variable; the variable is coded to reflect the correct stratification in each survey.

Best,
Mahmoud

Previous Topic: % of children who were dropouts
Next Topic: Mozambique DHS 2015 sampling design/frame different from other DHS
Goto Forum:
  


Current Time: Fri Nov 16 20:17:35 Eastern Standard Time 2018