Information on Regions Collected Over Time [message #20484] |
Tue, 10 November 2020 13:06 ![Go to next message Go to next message](/theme/default/images/down.png) |
CassandraC
Messages: 1 Registered: November 2020
|
Member |
|
|
Hello,
Is there English-language documentation for each country where there may be differences in the regions surveyed during different rounds of the DHS? I've consulted the final reports for each country of interest to me, but because some are in Portuguese, it is sometimes difficult to check if a given survey sampled only from certain regions vs. the country as a whole.
For example, the 2012-13 DHS for Mali was not able to survey several regions due to conflict, thus if I wanted to accurately compare something like % of children ever vaccinated for Mali across all rounds of the DHS, it would perhaps be advisable to only compare using the regions that were surveyed across all rounds.
Thank you!
|
|
|
|
Re: Information on Regions Collected Over Time [message #20514 is a reply to message #20484] |
Fri, 13 November 2020 07:58 ![Go to previous message Go to previous message](/theme/default/images/up.png) |
Bridgette-DHS
Messages: 3152 Registered: February 2013
|
Senior Member |
|
|
Following is another response from Senior DHS Directors, Tom Pullum & Livia Montana:
The great majority of DHS surveys have been nationally representative. Mali is the main exception. Even there, the areas that were omitted in different surveys had only a small percentage of the population. The effect on national estimates is small. A more common challenge is when a country re-aligns its level 1 administrative units (what we call regions). This has happened in several countries, including the Philippines, Uganda, and Nepal. When this happens, the survey continues to be nationally representative, but regions cannot be tracked over time (unless you work with the GPS coordinates of the clusters, which are given in the spatial covariates files). Sometimes the numerical codes for regions will change from one survey to the next will change, even if the boundaries don't change. This also will not affect national estimates, but will make it trickier to track sub-national changes from one survey to the next. Virtually all of the main survey reports include a map and lists of the regions. It should be clear from the maps whether there have been omissions or changes, even in the few survey reports that are not in English.
[Updated on: Fri, 13 November 2020 08:08] Report message to a moderator
|
|
|