clear
//Enter the filepath where you have downloaded data from IPUMS PMA, I recommend this is also your R working directory
cd "[enter filepath to your downloaded data from IPUMS PMA]"
//Create a data file with the variables RESULTFQ, FERTPREF, FULLPANELWEIGHT, STRATA, EAID, PHASE, GEOCD, GEONG
//Update the ### with the name of your data file from IPUMS PMA
use pma_00###.dta

//Keep only observations where the female respondent was interviewed in all three phases
keep if resultfq_1 == 1 & resultfq_2 == 1 & resultfq_3 == 1

//Create new geography variable to separate DRC and Nigeria sub-national samples
gen geog = country
replace geog = 3 if geocd == 2 & country == 2
replace geog = 8 if geong == 2 & country == 9
label define geography 1 "Burkina Faso" 2 "Kinshasa, DRC" 3 "Kongo Central, DRC" 7 "Kenya" 8 "Lagos, Nigeria" 9 "Kano, Nigeria"
label values geog geography
//Revert No Response cases to missing
foreach var of varlist fertpref_* {
	recode `var' (98/99=.)
}

//Revert No Response cases to missing
foreach var of varlist fertpref_* {
	recode `var' (98/99=.)
}
//Generally following the 2-phase example of Zimmerman et al. 2023
label define general_category 1 "Wants a child" 2 "No more children" 3 "Don't know"


forvalues i = 1(1)3 {
	gen gen_cat_`i' = .
	replace gen_cat_`i' = 1 if fertpref_`i' == 1 
	replace gen_cat_`i' = 2 if fertpref_`i' == 2
	replace gen_cat_`i' = 3 if fertpref_`i' == 97 | fertpref_`i' == 3
	label values gen_cat_`i' general_category
}


//alluvial figure
keep country geog urban subnational eaid_* strata_* gen_cat_* fullpanelweight
drop if gen_cat_1 == .
drop if gen_cat_2 == .
drop if gen_cat_3 == .
gen seqnum = _n

export delimited "alluvial_wide.csv", replace