| Linking adult children (>15) to their parents [message #12784] |
Thu, 13 July 2017 06:18  |
Juliette
Messages: 3
Registered: February 2015
|
Member |
|
|
Dear DHS-experts,
I'm currently trying to link parents and children living in the same household in order to compare some answer in the women and men questionnaires, hence I want both parents and children who answered the "over 15" questionnaire in the same database.
Do you have any advice in creating such a database?
Here's what I've tried for the case Ivory Coast, DHS 1994.
Quote:
use "$path/CIPR35FL.DTA", clear
* no line number of parents if children are older than 15
* keep heads of hh, their spouses & children
keep if hv101==1 | hv101==3 | hv101==2
*
gen eligibility=1 if hv117==1 |hv118==1
gen sondaughter=1 if hv101==3 & eligibility==1
gen head=1 if hv101==1 & eligibility==1
gen spouse=1 if hv101==2 & eligibility==1
bysort hhid: egen nbsd=count(sondaughter)
bysort hhid: egen nbhead=count(head)
bysort hhid: egen nbspouse=count(spouse)
*
gen keephh=1 if nbhead==1 & nbsd>=1 & nbspouse>=1
recode keephh (.=0)
*
keep if keephh==1
* MERGING WITH DATA FROM WOMEN AND MEN FILES
* ID: cluster_hh_line nb_age_survey
egen wuid=concat(hv001 hv002 hvidx hv007), punct(_)
egen muid=concat(hv001 hv002 hvidx hv007), punct(_)
drop if eligi!=1
*
merge 1:1 wuid using "$path/CI_women_94.DTA"
drop if _merge==2
merge 1:1 muid using "$path/CI_men.DTA", gen(merge)
drop if merge==2
Many thanks for your help!
[Updated on: Thu, 13 July 2017 07:18] Report message to a moderator |
|
|
|
| Re: Linking adult children (>15) to their parents [message #12795 is a reply to message #12784] |
Fri, 14 July 2017 08:40  |
Bridgette-DHS
Messages: 3214
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
As you may know, with more recent surveys you can go to the BR file and use the variable b16 to find the line number of a child (of any age) who is living in the same household as the mother. The early surveys, such as the one you are using, do not include b16. What you are trying to do is more general, however, namely, to link a child with either parent.
I have not gone over your logic in detail, but I assume you have thought it through carefully. Several years ago I tried to do the same thing, and reached the conclusion that it is possible to find MOST such relationships but NOT ALL. The problem is that hv101 is not sufficient to catch--with certainty--a biological relationship between a pair of people when neither of them is the household head. I suggest that you first remove from the PR file everyone who is ineligible for the individual interview. I see that 8713 women were eligible (hv117=1) and only 2909 men were eligible (hv118=1). You could merge with the IR and MR files. On the basis of age, number of children ever born and relation to head you could remove a lot of respondents. Do a combination of thinking about the possible combinations and seeing which combinations actually occur in the data. Good luck. If you find a good way to do this, you could post it for others to use.
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
