| KR file, Missing anthropometric indicators in the 2022 Ghana DHS [message #30294] |
Tue, 29 October 2024 23:06  |
Josie
Messages: 4
Registered: October 2024
|
Member |
|
|
Hi,
I am exploring the 2022 Ghana DHS for child malnutrition outcomes. I am using the KR file because I am interested in children under 5 of interviewed women. I find that in this file, the total sample is 9353 but there are missing values for a little over 4900 children on anthropometric indicators.
The 2022 Ghana DHS report indicated that 5045 children were eligible for the measurements of whom 98% had valid measures. In the KR file, information is available for about 4417 children, which is understandable since not all children had interviewed mothers.
What are the reasons for the substantially high number of missing children in the 2022 KR file (except those that were reported dead == 291)? How do I subsample to get the right sample for my work? Do I just exclude all the missing information? It would be worthwhile getting a justifiable reason to acknowledge in my research paper for my sample. I think I will also need to perform a comparison analysis on the socio-demographic characteristics of children and their mothers with missing data and those without to confirm it is a representative sample. Is this right?
I am using R for my analysis.
This was not the case for the 2003, 2008 and 2014 GDHS, which I felt was very odd (especially as I am also doing a trend analysis).
|
|
|
|
| Re: KR file, Missing anthropometric indicators in the 2022 Ghana DHS [message #30311 is a reply to message #30294] |
Thu, 31 October 2024 17:22  |
Janet-DHS
Messages: 891
Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member, Tom Pullum:
When a variable is coded with a dot "." it means the variable was NA or Not Applicable for that case. In this survey there was subsampling for anthropometry. Measurement of children's weights and heights only applied to half of the households, alternating from one household to the next within each cluster. A variable v042, "household selected for hemoglobin" applies to weight (hw2) and height (hw3), not just blood tests for hemoglobin concentration. If v042=0, then hw2, hw3, and all variables constructed from hw2 and hw3, such as hw70, hw71, and hw72, are coded with a dot for NA.
You can proceed to analyze the data without any concern for the subsampling. There is no need to compare the children with v042=1 (selected) vs the children with v042 (not selected) because the selection was random.
|
|
|
|
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
