/*IPUMS MICS Webinar on Linking across units of analysis 
Presenter: Anna Bolgrien, Project Manager of IPUMS MICS 

April 15, 2026 
*/

/*File preperation for Stata example 

Create 3 data extracts with the sample Sierra Leone 2017. 

Create extract for:
	Household Member (HL) unit of analysis with the variables sample, cluster, hhno, lineno, linenomom, linenodad, age, sex, relate
	Children age 0-4 (CH) unit of analysis with the variables sample, cluster, hhno, linech, linemc, agech, sexch
	Woman (WM) unit of analysis with the variables sample, cluster, hhno, linewm, agewm
Download these extract, unzip, run the provided harmonization do file, rename. 
	sl2017a_hl.dta
	sl2017a_ch.dta
	sl2017a_wm.dta
*/

*Starting with the HL dataset 
use sl2017a_hl.dta, clear

*Look at some example families

*This family has 2 parents and 3 children. 
	list lineno relate sex age linenomom linenodad if cluster==1 & hhno==9

*Man with 2 wives with multiple children and 6 foster/adopted children 
	list lineno relate sex age  linenomom if cluster==20 & hhno==9
	*within this HH we can see that some children's fathers have died, so non are orphans. But the children's mothers/fathers don't live in HH (Fostered)
	
*This family has multi-generational relationships 
	list lineno relate sex age linenomom linenodad if cluster==4 & hhno==15
	*This is the family we will look at for the rest of the examples. 
	
*Close out of the household member file. Open the Child age 0-4 file. 
use sl2017a_ch.dta, clear
*Find the 2 children in the household listed above (Cluster 4, HHno 15)
	list linech  sexch agech linemc if cluster==4 & hhno==15
	*Create a linking key back to the HL file. 
	gen lineno=linech
	save sl2017a_ch_1.dta,replace

*Open the WM data file. 	
use sl2017a_wm.dta, clear

*Identify the 2 women from the household listed above (cluster 4, hhno 15)
	list linewm agewm if cluster==4 & hhno==15
	*Create a linking key to the biological mother among children in the household roster (HL)
	gen linenomom=linewm
	save sl2017a_wm_1.dta, replace


*Returning to the HL file. 
use sl2017a_hl.dta, clear
	*View the example household (cluster 4, hhno 15)
	list lineno relate sex age if cluster==4 & hhno==15
	
*merge CH information onto children in the household roster. 	
	merge 1:1 sample cluster hhno lineno using sl2017a_ch.dta
	rename _merge _mergech
	*View the example household (cluster 4, hhno 15) to identify that the 2 children now have linked information from the CH record. 
	list lineno age sex agech sexch if cluster==4 & hhno==15
	
*From the HL roster, merge information about the WM if they are listed as the biological mothers for children 0-17 in the household  
merge m:1 sample cluster hhno linenomom using sl2017a_wm.dta
	sort cluster hhno lineno
	*View the example household (cluster 4, hhno 15) to see the information about the women in the household has been linked to the appropriate children in the household. 
	list lineno age sex agech sexch  linenomom linemc linewm agewm  if cluster==4 & hhno==15

*To identify caregivers that are not biological mothers, 
*Create a flag variable that identifies when the caregiver is the same as the biological mother 
	gen biomom=1 if linemc==linenomom & linemc!=.
	*Create the inverse flag relationship when the caregiver is not the biolgoical mother 
	replace biomom=0 if linemc!=linenomom & linemc!=.

	*View the flag variable 
	tab biomom

*Look at an example household where the caregiver is not the biological mother for some of the children
	list lineno relate sex age linenomom linemc biomom if cluster==20 & hhno==9