The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Weighting data » Disaggregating to lower Administrative Division
Disaggregating to lower Administrative Division [message #11595] Fri, 13 January 2017 16:48 Go to next message
adesh91 is currently offline  adesh91
Messages: 1
Registered: January 2017
Member
Hello,


I am trying to calculate proportions for access to improved sanitation stratified by urban-rural as well as at the first administrative division level (admin1). Ideally I would like to have separate estimates for urban and rural within each admin1 unit. Do strata usually correspond to admin1 units? Additionally can I use the weights as is to calculate this proportion?

Additionally if I want to obtain estimates at a lower administrative division level (such as districts/admin2) would I have to adjust the sample weights as multiple strata may overlap with a single admin2 unit and are DHS surveys designed such that they are representative at lower administrative division levels as well beyond urban/rural and regional?
Re: Disaggregating to lower Administrative Division [message #11619 is a reply to message #11595] Tue, 17 January 2017 16:04 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3208
Registered: February 2013
Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:

Yes, in general admin 1 is region (v024) and the strata are v024 x v025 (urban/rural). There are only a few exceptions, especially in recent years, to this.

Yes, although admin 2 areas are nested in admin 1 areas, because of the role of v025 it is possible for admin 2 areas to straddle two strata.

You never have to adjust the weights or any part of the svyset command. As you may know, it is slightly better to use "svy, subpop(...):" than "svy:" together with "if...."

DHS data are representative for any subpopulations or areas in the sense that they are unbiased (when weighted). However, as you go to smaller areas, the number of clusters will fall off rapidly, and the standard errors of estimates will increase. In that sense, the quality of the estimates will deteriorate, but they are still unbiased.
Previous Topic: Weighting data after merging survey rounds with different levels of representation
Next Topic: Do I really need to Weight the DHS data
Goto Forum:
  


Current Time: Fri Dec 13 00:14:18 Coordinated Universal Time 2024