The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Countries » Other countries » age at first cohabitation/v511
age at first cohabitation/v511 [message #23619] Thu, 21 October 2021 11:48 Go to next message
skanse is currently offline  skanse
Messages: 4
Registered: October 2021
Member
Hello,

After reading this post: https://userforum.dhsprogram.com/index.php?t=msg&goto=19 905&&srch=age+at+first+union#msg_19905


1) I wanted to ask if v511 has been updated and includes ever married women for all countries that include ever married women for the women's surveys?


2) I see that some questionaires include a question "How old were you when you first started living with him?" but I do not see a variable for that. I see that v511 is 'age at first cohabitation' but the recode manual states that the variable v511 is calculated from the century month code of the date of start of first marriage or union and the century month code of the date of birth of the respondent. Could you help me understand why v511 is calculated when there is a question asking the age directly and why that's not inlcuded?

3)In the dataset labels, v511 is labelled as 'age at first cohabitation' --does this include informal and formal marraiges/unions?

4) Lastly, when I created variables for child marriage <18 years and >18 years using v511 and checked it with the age variable hv105, I see that the age for v511 could be much higher than the actual age from hv105. v511 seems to align more with the ages represented using v012. Is there any guidance on this? I checked the DHS github, but I did not see anything related to child marriage using R. So I am not sure what the best way to calcualte the prevalance would be.

Thanks!
Re: age at first cohabitation/v511 [message #23621 is a reply to message #23619] Thu, 21 October 2021 16:48 Go to previous messageGo to next message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
You don't say which country and survey you are talking about, but I am guessing that you are talking about the India NFHS based on the link you included.
1) The dataset has not yet been updated.
2) v511 includes both formal marriages and informal unions where the woman and her partner live together as if married. The response to "How old were you when you first started living with him?" is included in v511 as this information is used to impute the century month code from which v511 is created.
3) As note in 2) v511 includes both formal marriages and informal unions.
4) How are you doing the comparison of v511 to hv105? Are you correctly linking the record for the respondent from the PR file to the IR file? For the most part, you should not see cases where v511 is much higher than hv105 if the cases are properly linked. Code for child marriage in R is not available yet, but you could refer to the code in Stata to see how it is calculated. Or refer to the Guide to DHS Statistics.


Re: age at first cohabitation/v511 [message #23622 is a reply to message #23621] Thu, 21 October 2021 17:26 Go to previous messageGo to next message
skanse is currently offline  skanse
Messages: 4
Registered: October 2021
Member

I am refering to the following countries/year:
-Mali, 2018
-Uganda 2016
-Pakistan, 2017-2018
- Timor-Leste, 2016
-Haiti 2016-2017
-South Africa, 2016
-Albania, 2017-18
-Senegal, 2019
-Angola, 2015
-Colombia, 2015
-Bolivia, 2008
-Cambodia, 2014


I have merged the PR and IR files, here is a short portion of the R code for the merge: right_join(mergepr,mergeir,by=c("hv001"="v001","hv002"= "v002","hvidx"="v003"),
copy=FALSE)

For example with Pakistan, I have restricted it to women ages 20-24 using hv105. If I then check the crosstabs of v511 and hv105, there are at least 9 cases that the v511 is higher than the age from hv105 of 24.

However, if I use restrict it using v012, and then check the cross tabs of v511 and v012, then the ages are all below age 24 for v511.

This is similar to some of the other countries that I have checked, varying in amount of cases.

In either way, there seems to be a quite a few differences beetween hv105 and v012 if I check those two variables by crosstabs.

Is there something wrong with the merge? or should I be using v012 instead of hv105?


Re: age at first cohabitation/v511 [message #23623 is a reply to message #23622] Thu, 21 October 2021 17:59 Go to previous messageGo to next message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
1) The issue with updating the dataset only applies to the India NFHS dataset.
2) V012 is the age of the respondent as reported by the respondent. hv105 is the age of the household member as reported by the household respondent. The household respondent is often a different person than the individual respondent, and may report a different age in hv105 than the respondent herself gave in hv105. If you crosstab hv105 and v012 you would see many differences in the reported age in some countries (in some countries they are quite consistent, but in countries where ages are not well known they can differ substantially). We treat the household and individual interviews as separate interviews and do not attempt to correct the age reported in these separate interviews. We would recommend using v012 rather than hv105 when disaggregating this data by age.


Re: age at first cohabitation/v511 [message #23624 is a reply to message #23623] Thu, 21 October 2021 18:07 Go to previous messageGo to next message
skanse is currently offline  skanse
Messages: 4
Registered: October 2021
Member
Great, thank you so much for the explantion!

My last questions just to be sure I understand--

1) all other countries would included ever married women within v511?

2) for v012, you mentioned that v012 is the age as reported by the respondent. The recode manual states "Current age in completed years is calculated from the century month code of the date of birth of the respondent (V011) and the century month code of the date of interview (V008). In a few cases the age in the data file will be different from that reported by the respondent when the respondent's birthday was in the month of interview, but she had not yet had her birthday. If the respondent correctly reported her age at her last birthday (and not her age at her next birthday) then the calculated age was rounded up from the reported age, to avoid inconsistencies between the age and the century month code for the birth. "

To my understanding, this is calculated, and the respondent is not actually asked her age. Would that be a correct understanding? So it is calucalted from the birth date given by the responde, and then the date of interview? The variable hv105, is the actual age reported by the head of the household, and it is not calculated.
Re: age at first cohabitation/v511 [message #23625 is a reply to message #23624] Thu, 21 October 2021 18:21 Go to previous message
Trevor-DHS is currently offline  Trevor-DHS
Messages: 787
Registered: January 2013
Senior Member
1) Yes, all other surveys include this data for ever married women.
2) Please see the questionnaires for each survey. Both the date of birth and age are reported by the respondent during the interview. Sometimes some of that information is not fully reported, e.g. age may be reported without date of birth (or vice versa), or month of birth may not be reported but year of both is reported. All of this information is used to calculate and if necessary impute a century month code. The age in v012 is calculated from the imputed century month code of the date of birth and the date of interview. V012 will be the same as the age originally reported in almost all cases, but can differ in relatively rare cases where there are inconsistencies in the date and age reported. In almost all cases v012 is identical to the age reported in the women's questionnaire. HV105 is exactly as reported by the household respondent (who is not necessarily the head of the household), but this age is often reported by someone who does not really know the respondent's age, and can be off by a number of years. The household respondent can be any adult who lives in the household. See hv003 for the line number of the household respondent.
Previous Topic: Indonesia breastfeeding variables (m4 m5) 2012
Next Topic: Senegal men's recode urban sample
Goto Forum:
  


Current Time: Thu Mar 28 04:19:27 Coordinated Universal Time 2024