| KR duplicates in household line number (b16) [message #31752] |
Mon, 02 February 2026 12:18 |
jackglaser
Messages: 2
Registered: February 2026
|
Member |
|
|
Hey folks. I'm working with a number of DHS files and have noticed duplicates in variable b16 (household line number, used to merge with the PR recode) within KR recodes. Specifically, these will be children in the KR file with the same v001 (cluster), v002 (household ID), and b16. This is after dropping any observations with b16 missing or 0.
Note that these children will all have unique values for the combination v001 v002 v003 bidx.
Usually (as in the Senegal 2020 and Burkina Faso 2021 surveys), these duplicates will have different mothers (so, v003 "mother's household member line number" will differ between them); although for some surveys (notably Rwanda 2000) they will have the same mother.
In each survey there are only a small number of these cases (~5-10 usually, never more than 80). But I'm curious as to how this might have occurred and whether my plan of dropping them for any PR-based analysis (since they don't uniquely merge onto the PR) but keeping them for any KR-based analyses (since they do have unique bidx) seems like the right call.
Thanks in advance!
|
|
|
|
Members
Search
Help
Register
Login
Home
