| Re: Survival status of preceding birth [message #25364 is a reply to message #25342] |
Wed, 12 October 2022 12:17   |
Janet-DHS
Messages: 893
Registered: April 2022
|
Senior Member |
|
|
Following is a response from DHS staff member Tom Pullum:
Your question is outside the range of the DHS users forum but I already had a program to merge successive siblings and I just needed to adapt it for your purposes. I have modified your outcome variable, but just by omitting cases of type 3. Because most deaths occur at a very early age, there are many more cases of type 1 than of type 2.
* Program to construct pairs of children, a reference child plus next earlier sibling
* This example does not include the mother's variables
* b3 is cmc of birth
* b7 is (imputed) months of age at death.
* b11 is the interval between the two siblings, in months
* Specify a path to a workspace
cd e:\DHS\DHS_data\scratch
use v001 v002 v003 v008 b* using "...BDBR7RFL.DTA", clear
* To get b7r, add 6 months if b7 is >=24 and a multiple of 12.
* Do not change if the change would be later than the date of interview.
gen b7r=b7
replace b7r=b7+6 if b7>=24 & b7==12*int(b7/12) & b7+6<v008
egen sibsize=count(bidx),by(v001 v002 v003)
* construct two files, one for the younger child in a pair and one for the older child
sort v001 v002 v003 bidx
save temp_younger.dta, replace
rename b* b*_sib
* attach the child with bidx=2 to the child with bidx=1
* attach the child with bidx=3 to the child with bidx=2
* etc.
gen bidx=bidx_sib-1
sort v001 v002 v003 bidx
merge v001 v002 v003 bidx using temp_younger.dta
tab _merge
* the child with bidx=1 does not get attached to a younger sibling
* the child with bidx=sibsize does not get attached to an older sibling
* remove these two types of cases
drop if bidx==0 | bidx_sib==.
tab _merge
drop _merge
keep v001 v002 v003 bidx* b0* b3* b4* b5* b7* b11
gen sib_died=.
replace sib_died=0 if b5_sib==1
replace sib_died=1 if b5_sib==0 & b3_sib+b7r_sib<b3-9
replace sib_died=2 if b5_sib==0 & b3_sib+b7r_sib>=b3-9
label variable sib_died "Outcome for younger sib"
label define sib_died 0 "Alive" 1 "Died before" 2 "Died after"
label values sib_died sib_died
* "before" and "after" are relative to the month of conception of the reference child
order v001 v002 v003 bidx* b0* b3* b4* b5* b7_sib b7r_sib sib_died
list v001 v002 v003 bidx* b0* b3* b4* b7_sib b7r_sib b11 sib_died if _n<=50, table clean compress
tab sib_died
* save with some filename and then merge with the BR or KR file to get weights and other variables
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
