Wave 9 chmain file
I've just downloaded wave 9 of Understanding Society to add another wave of child maintenance data (chmain file) to my analysis.
I was quite surprised that there are only 456 entries in the i_chmain file, compared with over 3,000 in waves 3, 5 and 7.
Do you know why there are so few entries in this wave?
I noticed in the questionnaire documentation something written about a script error for some of the variables in the file and am wondering if this has anything to do with it?
#1 Updated by Alita Nandi 8 months ago
- Private changed from Yes to No
- % Done changed from 0 to 10
- Assignee set to Stephanie Auty
- Status changed from New to In Progress
Many thanks for your enquiry. The Understanding Society team is looking into it and we will get back to you as soon as we can.
#2 Updated by Alita Nandi 8 months ago
- % Done changed from 10 to 80
- Assignee changed from Stephanie Auty to Charlotte Edney
In Waves 5, 7 and 9, adult respondents (cases in w_indresp) for whom w_csacalc=1 & at least one w_csakidno[N]=1 OR w_cmcokidno[N]=1 OR w_cmvolkidno[N]=1 should have been asked the questions which are reported in w_chmain. But this filtering did not work properly in Waves 5 & 7. In those waves anyone who reported w_csacalc=1 or w_csacalc=2 were asked the questions - so there are more cases in e_chmain and g_chmain than should be. This error was corrected in Wave 9, and hence fewer cases in Wave 9 as comapred to Waves 5 & 7.
Note in Wave 3, the filtering was supposed to be w_csacalc=1 or w_csacalc=2 and so that worked ok.
The filtering rules can be found in the questionnaires.
#3 Updated by Charlotte Edney 8 months ago
I'm sorry but I'm not following you very well. I can't find the filtering information anywhere in the questionnaire - could you point me in the right direction?
I can only see the question routing (under universe) for each variable, is this what you mean? For example if I take a random variable from wave 9 e.g. cseeterm the questionnaire documentation says the routing is: "If(CMRoute = 1) // If biological/adoptive parent of child aged under 16 or child aged 16-19 in full-time education, whose other biological/adoptive parent is not currently resident in the HH And If(CSACalc = 1 | 2) // If child's non-resident biological/adoptive parent is alive" -
Doesn't this imply the filtering is also applied to w_csacalc=1 or w_csacalc=2 in wave 9?
#4 Updated by Alita Nandi 8 months ago
Just above the Universe this note is written:
Loop foreach child enumerated via CSAKIDNO, CMCOKIDNO and CMVOLKIDNO. NOTE: script error - the script for this loop
routes on all cases that respond to CSACALC only and the additional filter that loops through CSAKIDNO, CMCOKIDNO and
CMVOLKIDNO has not been applied. more information in User Guide. NOTE: CSEETERM, CSEEHOL, RECON replaced with
CSEETERM2, CSEEHOL2, RECON2. See User Guide for details."
Within those who qualify as per (CMRoute = 1) & (CSACalc = 1 | 2), these questions are asked only for those children mentioned in CSAKIDNO, CMCOKIDNO or CMVOLKIDNO. Although the note says this additional filter was not applied, it was detected during fieldwork and then applied correctly.
#5 Updated by Charlotte Edney 8 months ago
I did see this note and orginally thought it makes it seems like this is an error in just wave 9? And it was applied correctly in the previous 3 waves?
I'm also wondering where the observations come from if they didn't loop over the childrens names? Just one response for each respondent (not based on the children mentioned)? Because even then there seems to be very few observations considering the amount of people who respond to the csacalc question (1,724).
Thank you for your help with this query.
#8 Updated by Stephanie Auty 7 months ago
By using duplicates report pidp (in Stata) on the i_chmain data file, you can see that there is more than one record for some pidps, and duplicates report pidp i_childpno shows one record per adult and child combination, so the loop has been applied over the children's names.
I will reply again when I have more information.
#11 Updated by Stephanie Auty 6 months ago
- Target version set to M9
- Assignee changed from Alita Nandi to Charlotte Edney
- Status changed from In Progress to Feedback
- Category set to Questionnaire content
Apologies for the long delay in response.
The observations do loop over the children's names from CSAKIDNO, CMCOKIDNO and CMVOLKIDNO. However, if you look at the Universe for those questions, the children's names are only asked in CMCOKIDNO if there is also a CSA/CMS calculation, and in CMVOLKIDNO if there is also a CSA/CMS calculation or a Court Order for maintenance. Those with only a voluntary arrangement are not asked which children it covers, and so are not asked the questions for which data appears in i_chmain.
#12 Updated by Charlotte Edney 6 months ago
Thanks for your response. I think I understand it better now. I just want to clarify something from the thread. So the reason that there are so few observations in i_chmain compared with other waves is due to the routing on the csacalc variable - i.e. as mentioned previously and from what i can see in the data, the csacalc is only be routed on csacalc==1 in wave 9 and (csacalc==1 | csacalc==2) in the other waves. But in the survey documentation for wave 9 the universe says the questions are routed on (csacalc==1 | csacalc==2). This is why I'm so confused. Does this mean the documention isn't correct? And I'm wondering about which routing has been/will be applied in wave 11 as the documentation also says routing (csacalc==1 | csacalc==2).