The DHS Program User Forum      
Discussions regarding The DHS Program data and results
Home » Topics » General » India NFHS 2015, data on age at marriage
India NFHS 2015, data on age at marriage [message #13927] Thu, 25 January 2018 14:45 Go to next message
cmisunas is currently offline  cmisunas
Messages: 3
Registered: June 2016
Member
I am writing with a question regarding the data on age at marriage from the recently released India NFHS 2015-16.

I understand that for India, to account for the practice of gauna, surveys have always used two country-specific questions: one question to record age/date at first marriage and the standard DHS question to record age/date at first cohabitation. I assume this explains why the values in StatCompiler and the NFHS report differ slightly.

I have been able to reproduce the estimates currently on StatCompiler using the standard DHS variable on age at first cohabitation (v511): 37% of women currently aged 20-49 reported their age/date at first cohabitation as occurring by age 18.

Unfortunately, I have been unable to reproduce estimates from the report on the percent of women married by age 18 using the additional variables on age at first marriage (s308-309): 41.1% among women aged 20-49.

When using variables 308 and 309, excluding inconsistencies (women who have provided the same year of marriage as birth) as well as women who reported don't know, the estimate is different than what is included in the final report (41.5%). Can you please advise or share syntax on how to proceed in reproducing the estimate from the report for this indicator?

**********************
rename v012 age
rename v511 ageatmar

gen weight=v005/1000000
svyset v021 [pw = weight ]

gen Agemarried=int(( s308c - v011 ) / 12)

gen wmarby18=0
replace wmarby18=1 if ageatmar<18
replace wmarby18=1 if s309<18
replace wmarby18=1 if Agemarried<18 & (s308c~=9997) & (s308c~=9998) & (s308y~=v010)

[Updated on: Thu, 25 January 2018 15:16]

Report message to a moderator

Re: India NFHS 2015, data on age at marriage [message #13981 is a reply to message #13927] Wed, 31 January 2018 11:39 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 1359
Registered: February 2013
Senior Member
Following is a response from Senior Data Processing Specialist, Mianmian Yu:

It is correct to apply S308C and S309 in your calculation. Below are some core codes used for tabulating Table 6.2 in the final report. They are in CSPro, but you may try and revise "gen Agemarried=int(( s308c - v011 ) / 12)" according to these codes.

BTW, I didn't see 41.5% but 40.1% on P.164.

{ ************************************************************************* }
  { Table 6.2 - part 2 of 2 (women) and Table 6.3.1 }

  { Generic logic to determine age at first *marriage* for NFHS-4 }
  if S308C in 500:1500 then
    v511x = int((S308C - V011) / 12);
  elseif S309 in 0:96 then
    v511x = S309;
  else
    v511x = notappl;
  endif;
  recode v511x => T602Col;
          <15  => 1;    { Thus marriage dates before birth are included }
         15-17 => 2;
         18-19 => 3;
           20  => 4;
         21-24 => 5;
               => default;
  endrecode;
  NevrMar = (V501 = 0);

  xtab(T602, rweight);

  { Marriage }
  if v511x = notappl | v511x > 30 then
    MedAcum = 30;
  elseif v511x < 8 then
    MedAcum = 8;
  else    
    MedAcum = v511x;
  endif;
  xtab(T602m1, rweight);
  if V012 >= 20 then
    xtab(T6031m, rweight);
  endif;

  { Cohabitation (= DHS marriage) }
  if V511 = notappl | V511 > 30 then
    MedAcum = 30;
  elseif V511 < 8 then
    MedAcum = 8;
  else    
    MedAcum = V511;
  endif;
  xtab(T602m2, rweight);
Re: India NFHS 2015, data on age at marriage [message #14013 is a reply to message #13927] Fri, 02 February 2018 12:16 Go to previous messageGo to next message
cmisunas is currently offline  cmisunas
Messages: 3
Registered: June 2016
Member
Dear Mianmian,

Many thanks for this response. I have revised the variable according to your codes. Can you please confirm if my understanding is correct, especially for the third estimate?

Estimate #1: The estimate currently on StatCompiler (37%) uses the standard DHS variable on age at first cohabitation (v511) and represents the percent of women currently aged 20-49 who reported first cohabiting/living with their partner by age 18.

Estimate #2: The estimate in the NFHS report (40.1%) uses the standard variables on age at first marriage (s308c & S309) and represents the percent of women currently aged 20-49 who reported first entering into marriage by age 18.

Estimate #3: The percent of women currently aged 20-49 who reported first marrying OR cohabiting/living with their partner by age 18 would be a combination of both Estimates 1 & 2 using (v511,s308c & S309). When I run this, it produces a value of 41.5%


*Estimate 1: Cohabitation (37.0%)*
gen Estimate1=0
rename v511 Agecohabited
replace Estimate1=1 if Agecohabited<18
svy: tab Estimate1 if age>=20 & age<=49

*Estimate 2: Marriage (40.1%)*
gen Agemarried=int(( s308c - v011 ) / 12) if (s308c>=500 & s308c<=1500)
replace Agemarried=s309 if s309<=96
gen Estimate2=0
replace Estimate2=1 if Agemarried<18
svy: tab Estimate2 if age>=20 & age<=49

*Estimate 2: Cohabitation and/or Marriage (41.5%)*
gen Estimate3=0
replace Estimate3=1 if Estimate1==1 | Estimate2==1
svy: tab Estimate3 if age>=20 & age<=49


Re: India NFHS 2015, data on age at marriage [message #14048 is a reply to message #14013] Tue, 06 February 2018 15:49 Go to previous messageGo to next message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 1359
Registered: February 2013
Senior Member

Following is a response from Senior Data Processing Specialist, Mianmian Yu:

The logic and syntax of the first two estimates look fine. Though Estimate 3 may not be labeled as "Cohabitation and/or Marriage", instead it should better be described as "Cohabitation OR Marriage", otherwise its syntax and result would not make sense to me.
Re: India NFHS 2015, data on age at marriage [message #14242 is a reply to message #14048] Sat, 10 March 2018 07:15 Go to previous messageGo to next message
pintupaul383@gmail.com is currently offline  pintupaul383@gmail.com
Messages: 3
Registered: September 2017
Location: New Delhi, India
Member
Hi, This is regarding calculation of age at first marriage using NFHS-4.
Syntax for calculation:
replace s308c=. if s308c==9997
replace s308c=. if s308c==9998
gen agemarried=int(( s308c - v011 ) / 12)
replace agemarried=s309 if s309<=96

*This syntax I have constructed from the understanding of previous posts. If we calculate age at first marriage using the above syntax the results will be matched with national report.
*But I am little wonder about the variable "s309". I could not understand that fact that why do we replace this variable with the newly created variable "agemarried". And what this variable (s309) stands for? Can anybody clear it?
Thak you.


M.Phil. Researcher
Centre for the Study of Regional Development
School of Social Science
Jawaharlal Nehru University
New Delhi-110067
India
Re: India NFHS 2015, data on age at marriage [message #14293 is a reply to message #14242] Mon, 19 March 2018 12:28 Go to previous message
Bridgette-DHS is currently offline  Bridgette-DHS
Messages: 1359
Registered: February 2013
Senior Member
Following is another response from Senior Data Processing Specialist, Mianmian Yu:

S309 is a reported age in years at the time of marriage, while S308C is century month code of reported date of marriage. Using S308C and V011 to calculate age at marriage is more accurate.

Previous Topic: Weighting in pooled dataset, multiple countries and years
Next Topic: Label lists for Uganda DHS 2016 data variables
Goto Forum:
  


Current Time: Sun May 20 09:54:18 Eastern Daylight Time 2018