| NFHS-4 Median Birth interval mismatch [message #18002] |
Tue, 13 August 2019 13:03  |
dfnbcc
Messages: 3
Registered: June 2019
|
Member |
|
|
Hello,
I am trying to calculate the median birth interval using STATA from the NFHS-4. However, I am getting the wrong median compared to the report. The report says the median in India is 32 months, however, I am getting 28 months.
I am subtracting 9 months from the variable as I want birth-to-pregnancy interval, not birth-to-birth. But even if I don't subtract 9, I get different values.
use IABR74FL.dta
gen weight=v005/1000000
rename sdistri shdistri
tostring shdistri, replace
gen strata = shdistri
svyset v021 [pweight=weight], strata (strata)
keep shdistri v008 b3 bord b0 b11 v106 hv024 v025 v190 v447a v133 s301 hv001 hv002 hvidx v005 v021 weight strata
*Births in the month of interview are included and births 60 months before the interview are excluded.
gen monthssncbirth=v008-b3
replace monthssncbirth=. if monthssncbirth<=60
drop if monthssncbirth==.
*First-order births (and their twins) are excluded from both numerators and denominators.
svy: tab bord
gen birth_order=bord if (b0 > 1)
svy: tab birth_order
*subtract by 9 months to find birth-to-pregnancy interval (WHO)
gen birth_interval=b11-9
replace birth_interval=. if birth_interval<0
Members Search Help Register Login Home Home » Topics » Fertility » NFHS-4 Median Birth interval (Median birth interval does not match NFHS-4 report)
Show: Today's Messages :: Show Polls :: Message Navigator
Switch to threaded view of this topic Create a new topic Submit Reply
- NFHS-4 Median Birth interval [message #17992] Sun, 11 August 2019 11:28
dfnbcc is currently offline dfnbcc
Messages: 2
Registered: June 2019
Member
Hello,
I am trying to calculate the median birth interval using STATA from the NFHS-4. However, I am getting the wrong median compared to the report. The report says the median in India is 32 months, however, I am getting 28 months.
I am subtracting 9 months from the variable as I want birth-to-pregnancy interval, not birth-to-birth. But even if I don't subtract 9, I get different values.
Code: [Select all] [Show/ hide]
use IABR74FL.dta
gen weight=v005/1000000
rename sdistri shdistri
tostring shdistri, replace
gen strata = shdistri
svyset v021 [pweight=weight], strata (strata)
keep shdistri v008 b3 bord b0 b11 v106 hv024 v025 v190 v447a v133 s301 hv001 hv002 hvidx v005 v021 weight strata
*Births in the month of interview are included and births 60 months before the interview are excluded.
gen monthssncbirth=v008-b3
replace monthssncbirth=. if monthssncbirth<=60
drop if monthssncbirth==.
*First-order births (and their twins) are excluded from both numerators and denominators.
svy: tab bord
gen birth_order=bord if (b0 > 1)
svy: tab birth_order
*subtract by 9 months to find birth-to-pregnancy interval (WHO)
gen birth_interval=b11-9
replace birth_interval=. if birth_interval<0
Would appreciate some help please.
Thanks in advance.
|
|
|
|
|
|
| Re: NFHS-4 Median Birth interval mismatch [message #18152 is a reply to message #18041] |
Mon, 30 September 2019 06:19  |
Bridgette-DHS
Messages: 3199
Registered: February 2013
|
Senior Member |
|
|
Following is a response from DHS Research and Data Analysis Director, Tom Pullum:
Instead of the BR file, use the KR file to get births in the past five years. Look at the distribution of b11. There are some very short intervals and I would edit them out. Then enter the following line in Stata:
summarize b11 [fweight=v005] if b11>9, detail
The 50th percentile is 32 months. That matches the report.
To get a birth-to-pregnancy interval you could subtract 9 months, but that would be the interval to pregnancies that resulted in a live birth, and presumably you would want to include all pregnancies, not just those that resulted in a live birth. For that you would need to use the calendar, and probably include current pregnancies. That's more complicated, partly because of left censoring.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
