The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Bangladesh » Bangladesh infant mortality replication
Bangladesh infant mortality replication Tue, 09 June 2015 13:30
 busbyj2 Messages: 3Registered: June 2015 Location: Austin Texas Member
We're trying to replicate the infant mortality data from 2011 DHS survey for Bangladesh for the Divisions but we're getting different numbers than what's in the report.

Here is a piece of our STATA code. Would welcome any troubleshooting.

gen prev5yr = v008 - b3 < 60
gen prev3yr = v008 - b3 < 36

gen infdeath = b6 <= 211
replace infdeath = . if b6 == 997 | b6 == 998 | b6 == 999

gen chldeath = b6 <= 305
replace chldeath = . if b6 == 997 | b6 == 998 | b6 == 999

gen attbirth = m3a == 1
replace attbirth = 1 if m3b == 1
replace attbirth = 1 if m3c == 1
replace attbirth = 1 if m3d == 1
replace attbirth = 1 if m3e == 1
replace attbirth = 1 if m3f == 1

gen delhfac = m15 != 11
replace delhfac = 1 if m3g == 1

************** 1 - INFANT MORTALITY (PREV 5 YEARS) **********************

gen indic=1

svy: mean infdeath, subpop(prev5yr)
* show size of subpopulations
estat size
* store Part_Results in various variables
matrix m=e(b)
matrix s=e(V)
matrix su=e(_N)
estat effects, deft
matrix d=e(deft)

gen natl=m[1,1]
gen senatl=s[1,1]
gen sizenatl=su[1,1]
gen deftnatl=d[1,1]

svy: mean infdeath, over(region) subpop(prev5yr)
estat size
matrix mq=e(b)
matrix sq=e(V)
matrix suq=e(_N)
estat effects, deft
matrix dq=e(deft)

forvalues i = 1/`numreg'{
gen reg`i'=mq[1,`i']
}
forvalues i = 1/`numreg'{
gen sereg`i'=sq[`i',`i']
}
forvalues i = 1/`numreg'{
gen sizereg`i'=suq[1,`i']
}
forvalues i = 1/`numreg'{
gen deftreg`i'=dq[1,`i']
}
savesome indic-deftreg`numreg' if _n==1 using birth_result1, replace
matrix drop m s su mq sq suq d dq
drop natl senatl sizenatl deftnatl indic

forvalues i = 1/`numreg'{
drop reg`i'
}
forvalues i = 1/`numreg'{
drop sereg`i'
}
forvalues i = 1/`numreg'{
drop sizereg`i'
}
forvalues i = 1/`numreg'{
drop deftreg`i'
}
Re: Bangladesh infant mortality replication [message #5580 is a reply to message #5570] Wed, 10 June 2015 11:07
 Bridgette-DHS Messages: 2019Registered: February 2013 Senior Member
Following is a response from Senior DHS Stata Specialist, Tom Pullum:

It happens that in January I prepared a version of my Stata under-five mortality program for this survey. The program and the log file are attached. I do not use the Stata do-file editor. You will probably want to put it into that. Please check the paths. The approach you are using will get close but will not give an exact match to the DHS procedures. Let me know if you have questions.

Re: Bangladesh infant mortality replication [message #5582 is a reply to message #5580] Wed, 10 June 2015 11:16
 busbyj2 Messages: 3Registered: June 2015 Location: Austin Texas Member
Thank you! Will be in touch should we have additional queries.
Re: Bangladesh infant mortality replication [message #5620 is a reply to message #5582] Wed, 17 June 2015 11:45
 Bridgette-DHS Messages: 2019Registered: February 2013 Senior Member
You are welcome.
Re: Bangladesh infant mortality replication [message #15101 is a reply to message #5620] Sun, 03 June 2018 12:05
 abpromiti Messages: 20Registered: December 2017 Location: Dhaka Member
hello,
How can i calculate infant mortality using spss?should i use B6 or B7?i m using IR file from BDHS 2014.
Re: Bangladesh infant mortality replication [message #15125 is a reply to message #15101] Wed, 06 June 2018 11:22
 Bridgette-DHS Messages: 2019Registered: February 2013 Senior Member

Following is a response from Senior DHS Stata Specialist, Tom Pullum:

All the under-five mortality rates are calculated from the BR file, NOT the IR file. The calculation uses b7, which is a recode of b6. The calculation is complicated. Some details are given in the Guide to DHS Statistics ( https://www.dhsprogram.com/pubs/pdf/DHSG1/Guide_to_DHS_Stati stics_29Oct2012_DHSG1.pdf). To match DHS you need a special program which is not generally distributed.

 Previous Topic: Deal with missing data in BMI Next Topic: Number of households
Goto Forum:

Current Time: Tue Nov 24 01:02:54 Coordinated Universal Time 2020