The DHS Program User Forum      
Discussions regarding The DHS Program data and results
Home » Topics » Fertility » Wanted Total Fertility Rate in DHS
Wanted Total Fertility Rate in DHS [message #2448] Thu, 19 June 2014 09:14 Go to next message
AbdoulM
Messages: 2
Registered: June 2014
Member
Hello!

I am working on fertility using the DHS data from several african countries
The Total Fertility Rate (TFR) values I computed are exactely the same as in the DHS reports
But when computing the Wanted TFR, the values are not matching the ones published in the reports
The method consist in excluding the unwanted births before using the Schoumaker's stata software module (tfr2)

Find below how I made it
Please help me understand where the differences come from
Thanks,

I use the definition of "wantedness" as given in the "Guide to DHS Statistics" (Page 87, Updated Sept. 2006)
Let's remind that:
*b3_ is the date of birth in CMC;
*b5_ is the survival status (1 or 0);
*b7_ is the age at death, mounth imputed;
*v613 is the ideal number of children.

A birth is considered wanted if the number of living children at the time of conception of the child i (NLC_i) is less than the ideal number of children (INC), that is: i is wanted if NLC_i < INC
The time of conception for a child i is: dconcep_i = b3_i - 9

A preceding child (let's say j) is considered living at the time of conception of the child i if:
**C1) j is born before the date of conception of i, That is: b3_j < dconcep_i
**C2) j is living at the time of the survey, That is: b5_j==1
**C3) OR j died at or before the date of conception of i, That is: b3_j + b7_j <= dconcep_i

According to these conditions, I create a dummy variable (G) as:
For every child j preceding i,
*G_ij =1 if {C1 AND (C2 OR C3)}
*If no, G_ij=0

The number of living children at the time of conception of i is computed as NLC_i = Sum(G_ij) over j.
For every i, we can now compare NLC_i with the INC
*If NLC_i < INC, then i is wanted
*If no, i is not wanted

I then recode every b3_i (i=1 to 6) as: recode b3_i (0/21600=0) if NLC_i > INC
This is supposed to exclude all the unwanted births when computing the TFR (with the Schoumaker's tfr2)

I am aware that the missing values in Stata can distort the resluts, so I write the commands accordingly.
Find attached the do-file, it may help understanding the process

Regards,

AbdoulM
University of Geneva
  • Attachment: Wanted_TFR.do
    (Size: 25.97KB, Downloaded 297 times)
Re: Wanted Total Fertility Rate in DHS [message #2466 is a reply to message #2448] Tue, 24 June 2014 14:32 Go to previous messageGo to next message
Liz-DHS
Messages: 1290
Registered: February 2013
Senior Member
Dear User,
We are currently researching your post and will get back to you as soon as we can.
Thank you!
Re: Wanted Total Fertility Rate in DHS [message #2467 is a reply to message #2448] Tue, 24 June 2014 15:05 Go to previous messageGo to next message
Liz-DHS
Messages: 1290
Registered: February 2013
Senior Member
Dear User,
Here is a response from one of our technical experts,Dr. Shea Rutstein:
Here is the STATA instruction given:

A preceding child (let's say j) is considered living at the time of conception of the child i if:
**C1) j is born before the date of conception of i, That is: b3_j < dconcep_i
**C2) j is living at the time of the survey, That is: b5_j==1
**C3) OR j died at or before the date of conception of i, That is: b3_j + b7_j <= dconcep_i
But a child who died at or before conception of the next child is dead at the time of conception.
The correct code should be
**C3) OR j died after the date of conception of i, That is: b3_j + b7_j > dconcep_i

Shea
Re: Wanted Total Fertility Rate in DHS [message #2474 is a reply to message #2467] Tue, 24 June 2014 18:23 Go to previous messageGo to next message
AbdoulM
Messages: 2
Registered: June 2014
Member
Liz, Shea,

Thank you very much for this effective answer!

Regards,
AbdoulM
Re: Wanted Total Fertility Rate in DHS [message #12989 is a reply to message #2448] Thu, 31 August 2017 10:05 Go to previous messageGo to next message
Mercysh is currently offline  Mercysh
Messages: 22
Registered: April 2014
Member
Do you mind to share your do file for TFR, I see your results are comparable to those published in DHS reports.

Mercy
Re: Wanted Total Fertility Rate in DHS [message #12995 is a reply to message #12989] Mon, 04 September 2017 09:20 Go to previous messageGo to next message
Maurel
Messages: 1
Registered: September 2017
Location: Cameroon
Member
Hello!
I am working on the impact of socioeconomic status on fertility trends desired and achieved among Cameroonian women using the DHS surveys of 2004 and 2011. I used the Moumouni's do file to make the corrections as suggested above but I still do not get the same results as DHS. could someone help me find the problem?

thank you!
Re: Wanted Total Fertility Rate in DHS [message #13032 is a reply to message #12995] Sun, 10 September 2017 13:18 Go to previous messageGo to next message
Liz-DHS
Messages: 1290
Registered: February 2013
Senior Member
Dear User,
I think to get any real assistance, you would need to post your code. Other users may be able to look at it and help you troubleshoot. In addition you may want to read the paper by Bruno Schoumaker which may help. It is titled "A Stata module for computing fertility rates and TFRs from birth histories: tfr2"
https://www.demographic-research.org/volumes/vol28/38/
Thank you!

[Updated on: Sun, 10 September 2017 13:18]

Report message to a moderator

Re: Wanted Total Fertility Rate in DHS [message #14943 is a reply to message #2474] Fri, 18 May 2018 01:42 Go to previous messageGo to next message
ashwini is currently offline  ashwini
Messages: 1
Registered: May 2018
Member
Hello,

I wanted to compute the Wanted Fertility rate and followed the instruction and the method posted by the one of the user but results are not matching with the report.

Can I get help on this?

These are the steps followed-
A preceding child (let's say j) is considered living at the time of conception of the child i if:
**C1) j is born before the date of conception of i, That is: b3_j < dconcep_i
**C2) j is living at the time of the survey, That is: b5_j==1
**C3) OR j died at or before the date of conception of i, That is: b3_j + b7_j <= dconcep_i
But a child who died at or before conception of the next child is dead at the time of conception.
The correct code should be
**C3) OR j died after the date of conception of i, That is: b3_j + b7_j > dconcep_i

Thank you.
Re: Wanted Total Fertility Rate in DHS [message #15025 is a reply to message #14943] Fri, 25 May 2018 14:58 Go to previous message
Liz-DHS
Messages: 1290
Registered: February 2013
Senior Member
A response from Dr. Shea Rutstein:
Quote:

Our code for the wanted TFR table is

for ch in REC21_EDT do { Births }
birth = B3;
if birth in lowcm:higcm then
agegroup = int( (birth-V011)/60 ) - 2; { Age group at birth }
if agegroup in 1:7 then
chi = 0;
do i = soccurs( REC21 ) while i > ch by (-1)
if B5(i) = 1 then
chi = chi + 1
else
deathm = B7(i);
if !special( B6(i) ) & int( B6(i)/100 ) = 3 then { if age at death in years }
deathm = B7(i) + 6; { assume death occur in the middle }
endif;
if B3(i) + deathm >= B3 - 9 then
chi = chi + 1;
endif
;
endif;
enddo;
if V613 in 0:90 & chi < V613 then
colt606 = 1;
xtab( t606b, rweight )
endif;
colt606 = 2;
xtab( t606b, rweight );
endif;
endif;
enddo;


Note the highlighted section:

It says add the age at death of a dead child to the date of birth of that child to get the date of death. If that sum (the date of death) is equal to or greater than the date of birth of the next child minus nine months (date of conception) then the child was living at the time of conception. The number of the living children is then compared with the desired number of children and, if less, then the next child was wanted. If the same or higher than the desired number, then the next child was not wanted at that time. The division into wanted-unwanted is done for all children born in the time period for the fertility rates.

So the user is correct except for children who died in the same month as the conception. These children are considered as living at the time of conception since the dates of conception and dates of death are a little fuzzy due to only being precise as to month.

I looked up the Guide to DHS Statistics which needs to be corrected on page 90:

It says:

1. Wantedness--A birth is considered wanted if the number of living children at the time of conception of the birth is less than the ideal number of children as reported by the respondent. For the calculation of the time of conception, nine months are subtracted from the date of birth. A preceding child is considered living at the time of conception if it was born before the date of conception and a) is living at the time of the survey or b) died at or before the date of conception (calculations based on century-month codes). The date of death of preceding children who died is calculated by adding the age at death in months to the child's birth date in century-month code. If a child's age at death is reported in years, then the child is assumed to have died at ages 6 months higher than the number of years (30 months for children who died at 2 years of age, 42 months for children who died at 3 years of age, etc.).
The highlighted word should be "after" instead of "before". The Guide needs to be corrected. The CSPRO program is OK.


Previous Topic: Factors affecting the duration of breastfeeding
Next Topic: Sex of preceding child
Goto Forum:
  


Current Time: Sat Aug 18 06:01:13 Eastern Daylight Time 2018