| Filtering children's data for nutrition indicators [message #30337] |
Fri, 08 November 2024 05:10  |
Mark_H_22
Messages: 3
Registered: September 2024
|
Member |
|
|
Hello,
I have a question about filtering children's datasets to calculate dietary diversity. In the Guide to DHS statistics (DHS8), the following description for calculating the denominator is given (page 11.61):
To select the cases for the denominator, first filter or keep only living children living with the mother born in the preceding 24 months (keep/select if b19 < 24 & b9 = 0), then keep only the youngest child (keep/select if the first entry in the dataset (_n = 1) or the first for this respondent (caseid ≠ caseid[_n-1]) as the youngest child is the first listed for the respondent in the data file), then finally restrict to children age 6-23 (note that you cannot select for children age 6-23 in the first step as this would include some cases that were not asked the questions.
I understand this as the 3 following steps:
1) Filter to keep only living children, living with their mother, under 24 months old.
2) Keep only the youngest child per mother.
3) Remove children younger than 6 months.
I am unsure why filtering the data to only contain children aged 6-23 months must be the last step, and not the first. In step 2, only the youngest child is kept, and all older siblings are removed. Therefore, if the youngest child is younger than 6 months, this child is also removed in step 3 and I end up with no children for this particular respondent (mother).
Surely the data should be filtered to keep children in the range 6-23 months first, and then keep only the youngest child?
Any clarification on this would be really helpful. Thank you.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
