Support #856
closed
how to deal with contradictory records
100%
Description
Dear Understanding Society team,
I am writing to seek your suggestions about contradictory data.
The first example is in the same wave. I ran cross-tabulation between a_qfhigh (highest qualification ever achieved) and a_fenow (age when leaving school / never went to college or university). I assume that the respondents who attained university degree would not be found in the category "never went to college or university" in the variable a_fenow. However, this is not the case.
. tab a_qfhigh a_fenow if a_dvage>=25
| still in further education
highest qualification | missing inapplica don't kno write in never wen at colleg | Total
----------------------+------------------------------------------------------------------+----------
missing | 1 17 0 0 0 0 | 18
refused | 0 3 0 2 1 0 | 6
don't know | 0 18 7 19 38 2 | 84
university higher deg | 0 16 3 3,805 53 236 | 4,113
1st degree level inc | 0 14 3 5,458 203 296 | 5,974
diploma in higher edu | 0 4 2 2,089 440 142 | 2,677
teaching qualificatio | 0 0 2 574 107 23 | 706
nursing or other medi | 0 1 2 556 335 37 | 931
a level | 0 11 3 1,443 1,156 152 | 2,765
welsh baccalaureate | 0 0 0 0 2 0 | 2
international baccala | 0 1 0 26 10 5 | 42
as level | 0 1 1 104 75 13 | 194
higher grade/advanced | 0 1 0 175 173 17 | 366
certificate of sixth | 0 0 0 60 49 6 | 115
gcse/o level | 0 18 10 3,198 5,037 262 | 8,525
cse | 0 1 1 529 1,360 41 | 1,932
standard/ordinary (o) | 0 1 0 183 430 21 | 635
other school (inc. sc | 0 7 0 355 967 35 | 1,364
none of the above | 1 392 16 1,676 10,915 287 | 13,287
----------------------+------------------------------------------------------------------+----------
Total | 2 506 50 20,252 21,351 1,575 | 43,736
The category that is shown as "never wen" is actually "never went to college/university".
As can be seen from the above table. there are 203+53 respondents who had university degrees, yet reported that they never went to university.
The second example is related to cross-wave youth data. I assume that if a respondent who reported yes to "ever smoked cigarettes" in wave 1 would not say no to the same question in wave 2. However, this is what I got:
tab a_ypevrsmo b_ypevrsmo
ever smoke |
cigarettes | ever smoke cigarettes at all
at all | missing yes no | Total
-------------+---------------------------------+----------
missing | 0 1 15 | 16
yes | 2 74 39 | 115
no | 27 209 2,416 | 2,652
-------------+---------------------------------+----------
Total | 29 284 2,470 | 2,783
39 respondents who said that he/she has ever smoked in wave 1 said never smoked in wave 2.
This is getting more complicated when I link wave 1-6 together.
I understand that these are measurement errors/recall errors. I could have simply recoded these contradictory records as missing. But the numbers of these recodes may seem trivial but as they add up over waves, they are not small numbers. I am not sure what I am supposed to deal with them.
Many thanks for your time,
Regards,
Min
Updated by Stephanie Auty about 7 years ago
- Status changed from New to In Progress
- % Done changed from 0 to 10
- Private changed from Yes to No
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.
Best wishes,
Stephanie Auty - Understanding Society User Support Officer
Updated by Stephanie Auty about 7 years ago
- Status changed from In Progress to Feedback
- Assignee changed from Alita Nandi to Min Zhang
- % Done changed from 10 to 70
Dear Min,
In the first case, it is possible that some of these respondents gained their degree through distance learning and so did not go to university. However, at least some will be a data inconsistency based on interviewer or respondent error. These two questions are not asked together in the questionnaire and there is no check implemented between them in the CAPI software.
In your second example, these differences will be due to respondent error as the youth questionnaire is self-completion. It could be that they had only smoked once and then forgotten about it by Wave 2, for example. You might decide that it’s more likely that someone would smoke and then forget than make up that they had smoked, or think that they had when they hadn’t, but you will need to decide which assumptions you are willing to make based on your research question.
Best wishes,
Stephanie Auty - Understanding Society User Support Officer
Updated by Stephanie Auty about 7 years ago
- Status changed from Feedback to Resolved
- % Done changed from 70 to 100
Updated by Stephanie Auty almost 7 years ago
- Status changed from Resolved to Closed