Project

General

Profile

Support #1234

Variable for checking if parent has same ex-partner for all children

Added by Charlotte Edney over 1 year ago. Updated over 1 year ago.

Status:
In Progress
Priority:
Normal
Category:
-
Start date:
08/28/2019
% Done:

50%


Description

Hi,

I'm looking at waves 3, 5 and 7 of the individual file and the variable exsame. I noticed something that looks strange in changes from wave to wave. For some individuals they respond "no" to the question: "is the father/mother of all your children the same person?" in earlier waves 21/23 and then in wave 23/25 change their response to "yes" the father/mother of all your children is the same person. This doesn't seem to make logical sense (I'd understand/be able to check the other way round as respondents may have new children in the survey period). It seems to occur for enough observations not to just be a mistake by the respondent. Do you have any insights on this?

Thanks in advance,
Charlotte

#1

Updated by Alita Nandi over 1 year ago

  • Subject changed from Variable for checking if parent has same ex-partner for all childre to Variable for checking if parent has same ex-partner for all children
  • Status changed from New to In Progress
  • Assignee set to Charlotte Edney
  • % Done changed from 0 to 10
  • Private changed from Yes to No

Hello Charlotte,

I have run a quick check and compared c_exsame e_exsame g_exsame and found:
1. 62 cases where c_exsame=2 & e_exsame=1
2. 75 cases where c_exsame=1 & e_exsame=2
3. 90 cases where e_exsame=2 & g_exsame=1
4. 84 cases where e_exsame=1 & g_exsame=2

Is this what you find?

Best wishes,
Alita

#2

Updated by Charlotte Edney over 1 year ago

Hi Alita,

Thanks for looking into it.
I was using a merged dataset and looked at it observationally but also checked the individual file in wide format to be sure it wasn't coming from one of my merges.

I get similar but slightly different numbers:
1. 46 cases where c_exsame=2 & e_exsame=1
2. 60 cases where c_exsame=1 & e_exsame=2
3. 79 cases where e_exsame=2 & g_exsame=1
4. 74 cases where e_exsame=1 & g_exsame=2

also
5. 86 cases where c_exsame=2 & g_exsame=1
6. 93 cases where c_exsame=1 & g_exsame=2

Not quite sure why we've got a disrecpency as I only used the individual file in wide format to cross tabulate.

Best,
Charlotte

#3

Updated by Alita Nandi over 1 year ago

  • % Done changed from 10 to 50

For 5. & 6., my numbers are 91 and 97. You are consistently getting slightly fewer discrepancies than I am. Not sure why. Have you added any condition that drops some cases? Below is the code I used. Ignoring the difference in our numbers, the % of discrepancy between Wave 3-5 is around 7.6%, higher for the other waves (12.3% between 5 & 7, 15.4% between Waves 3 & 7). Have you checked whether the PIDP of the children reported are the same in the cases where you find the discrepancies - I mean could some of these be genuine cases of change in circumstances?

Hope this helps.

If you have further questions or follow-ups please let us know.
Best wishes,
Alita


foreach w in c e g {
qui use pidp exsame using "$m/`w'_indresp.dta", clear
save `w', replace
}
use c, clear
merge 1:1 pidp using e, nogen
merge 1:1 pidp using g, nogen

  • recoding missing values to system missing
    mvdecode _all, mv(-9/-1)

ta c_exsame e_exsame if c_exsame~=e_exsame, nol
ta e_exsame g_exsame if g_exsame~=e_exsame, nol
ta c_exsame g_exsame if c_exsame~=g_exsame, nol

generat diffce=0 if c_exsame<. & e_exsame<.
replace diffce=1 if c_exsame<. & e_exsame<. & c_exsame~=e_exsame
ta diffce

generat diffeg=0 if g_exsame<. & e_exsame<.
replace diffeg=1 if g_exsame<. & e_exsame<. & g_exsame~=e_exsame
ta diffeg

generat diffcg=0 if g_exsame<. & c_exsame<.
replace diffcg=1 if g_exsame<. & c_exsame<. & g_exsame~=c_exsame
ta diffcg

Also available in: Atom PDF