| Validation challenges: v169b with DHS Statcompiler [message #29304] |
Fri, 24 May 2024 10:24  |
researcher_dhs
Messages: 3
Registered: April 2024
|
Member |
|
|
I have been calculating summary statistics for the use of mobile phones for financial transactions by women across DHS surveys using the microdata as part of a larger project.
My calculations from microdata align with the reported values in the DHS final reports. However, as I am calculating this for a large number of surveys, I conducted a validation exercise with the Statcompiler values. I find large variations.
For instance, the reported value for total percent of women using their mobile phones for financial transactions in GHA2022DHS is 70.37% in the microdata, 70.4% reported in table 15.6.1 in the final DHS report, but 82.5% according to Statcompiler. This is true across BFA2021DHS (30.25% in the microdata, 31% in the report (I translated the relevant paragraphs), and 39.5% in Statcompiler). It is also true for other countries.
I just wanted to flag this as a potential issue, or try to understand in case I'm making an error in computations.
|
|
|
|
| Re: Validation challenges: v169b with DHS Statcompiler [message #29424 is a reply to message #29304] |
Mon, 17 June 2024 07:00  |
Bridgette-DHS
Messages: 3094
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Staff Member, Trevor Croft:
Basically, there are two indicators with the same numerator, but different denominators. See table 15.6.1 in the report. The indicator in STATcompiler is the last one with the restricted denominator. Ultimately, we will have both in the STATcompiler, but currently only the second one is there.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
