Age at marriage [message #30500] |
Tue, 10 December 2024 17:30  |
Varsha
Messages: 54 Registered: November 2020
|
Senior Member |
|
|
Hello,
I was going over the codes provided here for 4+ ANC visits:
https://github.com/DHSProgram/DHS-Indicators-Stata/blob/mast er/Chap09_RH/RH_ANC.do
Why are we including "don't know/missing" response in the 0 category. I mean, why are we not treating it as a missing value and excluding it from the calculations. Only when I include it, I'm exactly able to match the estimate for 4+ ANC visits given in Table 8.5 on page 217 in NFHS-4 national report.
The same goes for the variable: ANC before 4 months.
Please help.
|
|
|
Re: Age at marriage [message #30504 is a reply to message #30500] |
Wed, 11 December 2024 07:10   |
Bridgette-DHS
Messages: 3230 Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
When estimating a percentage of responses that are "yes", it is general practice at DHS to classify responses such as "don't know" and "maybe" with "no". You will find many examples. For example, the standard questionnaire includes a question "are you pregnant now?" The possible responses are "yes", "no", and "unsure". In the tables, "unsure" is classified with "no". As another example, in the calculation of HIV or malaria prevalence, ambiguous test results are grouped with negative.
Years ago, to be frank, I got into a debate at DHS regarding this practice. In particular, I argued that ambiguous HIV results should be removed from the denominator rather than being grouped with the negative results--the sort of argument you would make for ANC visits. However, I lost the argument.
Let's just say that having a standard policy means that you don't have to go through variable-by-variable interpretation of what is meant by "unsure", "maybe", "don't know", "not stated", etc. You just always group them with "no," producing a conservative estimate of the percentage "yes".
|
|
|
|
Re: Age at marriage [message #31143 is a reply to message #30500] |
Wed, 09 July 2025 23:20  |
trusteepioneer
Messages: 1 Registered: July 2025
|
Member |
|
|
In the DHS Stata code you referred to (RH_ANC.do), responses such as "Don't know" or missing are often included in the 0 category when calculating indicators like 4+ ANC visits or early ANC (before 4 months).This is intentional -- and here's why:
The DHS standard practice is to include all women in the denominator except those who did not have a live birth in the reference period (typically last 5 years). That means:
Women who responded "Don't know" or had missing values for ANC visits are counted as not having received adequate ANC.
This deflates the estimate slightly, which is reflected in official tables (like Table 8.5 in NFHS-4).
|
|
|