The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Child Health » How is sampling done for child development and child diet?
How is sampling done for child development and child diet? [message #23689] Sat, 06 November 2021 04:07 Go to next message
jejimel is currently offline  jejimel
Messages: 2
Registered: November 2021
Member
Hello DHS forum,

I'm working with several DHS datasets to examine the associations between child development and diet in children 36-59 months of age. Both child development and diet are not collected for all children in this age range. I'm trying to find information on how children are sampled for these two assessments. Since both are optional, there seems to be virtually no information on the DHS website or in final reports.

Thank you!

Lily
Re: How is sampling done for child development and child diet? [message #23703 is a reply to message #23689] Tue, 09 November 2021 19:54 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3017
Registered: February 2013
Senior Member

Following is a response from DHS Research & Data Analysis Director, Tom Pullum:

The selection process is described in the questionnaires in an appendix to the final report. When there is subsampling of children, the selected child is either the youngest child in an age range who is living with the mother (b9=0) or a randomly selected child based on criteria such as being in a specified age range. For the latter, the random selection (out of all children in the household who meet the criteria) are selected with a Kish grid to ensure randomness.

You will have a problem if the child selected for the child development module and the child for whom you have the diet information are different children, but you can shift from a child-level to a household-level analysis (with some attenuation of relationships). Please identify one of the surveys you are using.


Re: How is sampling done for child development and child diet? [message #23707 is a reply to message #23703] Thu, 11 November 2021 02:10 Go to previous messageGo to next message
jejimel is currently offline  jejimel
Messages: 2
Registered: November 2021
Member
Thank you for your quick response. This is very helpful! I'm working with 15 surveys, two of which are Uganda 2016 and Burundi 2016-2017 (as examples). Using the questionnaires in the appendices, I was able to confirm that the ECDI was administered to the youngest child 36-59 months of age. However, I am still struggling to understand the sampling process for child diet for this age group. The questionnaire instructions and skip patterns indicate that data should be collected for the youngest child born between 2014-2016, corresponding to the youngest child <24 months of age. However, there is diet data available for older children 36-59 months of age and it is not clear how they were sampled.
Re: How is sampling done for child development and child diet? [message #23754 is a reply to message #23707] Wed, 24 November 2021 13:54 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 3017
Registered: February 2013
Senior Member
Following is a response from Senior DHS Specialist, Trevor Croft:

The variables v409-v414w in the women's data (the IR file) refer to the youngest child under 2 living with the mother. When the KR file is constructed, with one record for each child born in the last 5 years, those variables (along with other v variables in the IR file) are copied onto the records for all children born in the last 5 years, even when the variables do not apply to that child. Thus you must either use the IR file, or use the KR file after selecting just the youngest child under 2 living with the mother.

To correctly analyze the variables with the KR file, follow the selection documented in the Code Share Library:
Go to https://github.com/DHSProgram/DHS-Indicators-Stata; Select Chapt11_NT

*Selection of the youngest child under 2 years living with the mother is done with lines 96-100 in INTmain.do:

* Selecting for youngest child under 24 months and living with mother
keep if age < 24 & b9 == 0
* if caseid is the same as the prior case, then not the last born
keep if _n == 1 | caseid != caseid[_n-1]
For the Uganda 2016 survey, age can be replaced with b19.
Previous Topic: Computing ARI
Next Topic: Post Natal Care check
Goto Forum:
  


Current Time: Fri Mar 29 11:51:10 Coordinated Universal Time 2024