Home » Data » Sampling » Strata specification across multiple surveys
Strata specification across multiple surveys [message #9233] |
Thu, 25 February 2016 18:05 ![Go to next message Go to previous message](/theme/default/images/down.png) |
lmcdougal
Messages: 17 Registered: March 2013
|
Member |
|
|
I am doing a pooled analysis across 31 recent DHS surveys, all from 2010 - present. I'm trying to set up the survey design for each survey; psu (v021) and denormalized weights are clear to me, but I'm having trouble determining with certainty what the correct strata variable to use for each survey is. I know that the most reliable way is to look in the final reports to see individual survey designs, but across this number (and variable languages), that's a challenge.
Based on what I've read in this forum, it seems that generally crossing district and urban/rural (ex. <group(v024 v025)>) is an appropriate methodology, but there are some instances where that is not correct, for example Bangladesh 2011, where there were 3 levels of stratification. Can v022 be used in general, but replaced with the cross of v024/v025 when it is missing? If not, is there a centralized list of strata across surveys that could be shared? Any other suggestions?
Many thanks for your help.
PS - The surveys I'm using are: Bangladesh 2011, Benin 2011-12, Comoros 2012, Congo Brazzaville 2011-12, Congo Democratic Republic 2013-14, Cote d'Ivoire 2011-12, Dominican Republic 2013, Egypt 2014, Gabon 2012, Ghana 2014, Guinea 2012, Haiti 2012, Honduras 2011-12, Indonesia 2012, Jordan 2012, Kenya 2014, Kyrgyz Republic 2012, Liberia 2013, Mali 2012-13, Namibia 2013, Nepal 2011, Niger 2012, Nigeria 2013, Pakistan 2012-13, Peru 2012, Sierra Leone 2013, Tajikistan 2012, Togo 2013-14, Uganda 2011, Yemen 2013 and Zambia 2013-14.
|
|
|
Goto Forum:
Current Time: Fri Aug 16 21:44:44 Coordinated Universal Time 2024
|