| Reclassifying admin areas [message #29053] |
Wed, 17 April 2024 17:03  |
Wolf
Messages: 2
Registered: June 2022
|
Member |
|
|
I am pooling 6 DHS surveys in Bangladesh together. In order to do this, I must harmonize some variables, including adminastrative areas (Divisions), as the number of Divisions have changed over time (in 1999, there were 5, now there are 8).
I have plotted the GPS coordinates of the PSUs onto a map, with the current (8) Division boundaries, and classified each PSU based on which Division the clusters falls within.
The question I have is, do you think this is a problem for survey weights in the pooled dataset?
These changes will not impact the 2017/18 survey as it was conducted after the boundaries were redrawn, but it will impact earlier surveys, particularly the 1999/2000 survey which at the time it was conducted, only had 5 admin areas.
Thanks in advance!
|
|
|
|
| Re: Reclassifying admin areas [message #29059 is a reply to message #29053] |
Thu, 18 April 2024 08:57  |
Bridgette-DHS
Messages: 3208
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
Congratulations on harmonizing the administrative areas. That's a lot of work.
You should not have to change the weights at all. They reflect the sampling fractions for the strata as they were defined at the time of each survey. That is, they are correct for each cluster as it was classified at the time. The change in boundaries would not alter those sampling fractions.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
