The DHS Program User Forum: Sampling » Calculating Sampling Errors at the State Level

Home » Data » Sampling » Calculating Sampling Errors at the State Level

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

Calculating Sampling Errors at the State Level [message #6937]

Wed, 05 August 2015 11:56

amaustin
Messages: 20
Registered: February 2015

Member

Are state-level sampling errors available for the 2013 Nigerian DHS? While the published reports document National, Urban/Rural, and Regional estimates for the sampling errors they do not go down to the state level.
Thank you!

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #6976 is a reply to message #6937]

Fri, 07 August 2015 10:16

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

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

It is true that the appendix C (estimation of sampling errors) in the final report of Nigeria DHS 2013 did not provide sampling errors at the state level, because it's a lot of tables.

We do not have the state level sampling errors in house either. However, it is true that the survey was designed to allow break down to state levels, so data users can calculate the sampling errors at the state level if they need to.

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #6977 is a reply to message #6976]

Fri, 07 August 2015 10:28

amaustin
Messages: 20
Registered: February 2015

Member

Thank you!
So using the SAS proc surveyfreq procedure should work for calculating sampling errors at the state level?

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #7035 is a reply to message #6977]

Fri, 14 August 2015 09:24

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

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

Yes, you can use the PROC SURVEYFREQ to calculate the sampling errors for simple proportions at the state level. You need to declare the survey parameters such as CLUSTER and STRATUM. I will suggest you use PROC SURVEYMEANS which calculates the sampling errors for means, proportions and ratios.

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #7036 is a reply to message #7035]

Fri, 14 August 2015 09:28

amaustin
Messages: 20
Registered: February 2015

Member

Thank you!

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #14785 is a reply to message #7036]

Sun, 06 May 2018 00:25

Hassen
Messages: 121
Registered: April 2018
Location: Ethiopia,Africa

Senior Member

Thank you all!!

Hassen Ali(Chief Public Health Professional Specialist)

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #16335 is a reply to message #7035]

Fri, 14 December 2018 16:32

PRB
Messages: 2
Registered: December 2018

Member

Hello, I am surprised to read that PROC SURVEYFREQ would calculate the sampling errors appropriately. Some (or many?) of the country reports have an appendix "ESTIMATES OF SAMPLING ERRORS" which outlines the complex calculations used to calculate the sampling errors, including that DHS staff wrote a special program to run a bootstrap to calculate the SEs. So are you saying that the SAS survey PROCS can replicate that same program?
I apologize for doubting, I just want to be sure I understand correctly.

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #16381 is a reply to message #16335]

Wed, 02 January 2019 12:12

Bridgette-DHS
Messages: 3199
Registered: February 2013

Senior Member

Please see the following response from our Senior Sampling Statistician, Ruilin Ren:

The SAS PROC SURVEYMEANS can calculate sampling errors for simple indicators such as means, proportions, and ratios. But it cannot calculate complex rates such as for the TFR, CMR and MMR. ICF developed SAS codes for the sampling errors presented in appendix B, which used the PROC SURVEYMEANS for simple indicators. For TFR, CMR and MMR, a special code was developed using Jackknife. Though the PROC SURVEYFREQ can also calculate sampling errors for proportions, but it is not used in the ICF's code for sampling errors. You can specify the variance estimation method in PROC SURVEYMEANS including Bootstrap and Jackknife, but again it is for simple indicators. ICF codes did not use replicate method for simple indicators.

Report message to a moderator

Re: Calculating Sampling Errors at the State Level [message #16382 is a reply to message #16381]

Wed, 02 January 2019 14:07

PRB
Messages: 2
Registered: December 2018

Member

Thanks!

Report message to a moderator

Previous Topic:	Oaxaca decomposition and stratification.
Next Topic:	Isolating the 15-24 female pop at district-level (Malawi)

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Mon Nov 25 11:22:54 Coordinated Universal Time 2024