Explaining valid skips for AUDIT-C items
I'm using alcohol data (AUDIT-C) from Waves 7, 9 and 11 for a longitudinal study. Please could you provide any information on why participants would have a response of -8 (for valid skip) on AUDIT-C items (i.e, auditc3, auditc4, and auditc5) for reasons other than that they are non-drinkers (as indicated by "no" on 'auditc1')? There also appears to be some participants with this response on item auditc1 as well.
Any explanations would be appreciated, thank you.
Updated by Understanding Society User Support Team 2 months ago
- Category set to Data documentation
- Status changed from New to Feedback
- % Done changed from 0 to 70
- Private changed from Yes to No
You can find this information in these questions' universe (see the pdfs of the questionnaires here: https://www.understandingsociety.ac.uk/documentation/mainstage/questionnaires).
For example, in wave 7, the universe of auditc1 is "If ((GRIDVARIABLES.ModeType = 1 & CASISTART.scac = 1 Thru 3) | (GRIDVARIABLES.ModeType = 2 | 3)) //
Mode is face-to-face and has agreed to self-completion OR mode is telephone or web". This means that this question was not asked of respondents who gave a face-to-face interview but didn't complete the self-completion part. The variables mentioned in the universe are not available in the released data as these are "technical" variables created during the interview and later dropped. However, you can check this using other variables available in the End User Licence data, indmode ("Interview mode", =1 face-to-face) and scflag_dv (=1 when the self-completion part completed, =0 when not completed). You can then see that the overwhelming majority (save for a few inconsistencies) of -8 responses on auditc1 are for people for whom indmode==1 and scflag_dv==0. So, the above explains -8s on auditc1. The remaining questions in the module, auditc2-c5, were additionally filtered by auditc1 (for most of them the condition being auditc1=1), as you rightly observed. You can further check if the filtering rules were the same for other waves.
Please note that there is another category of missing values for these variables, namely -9 "missing". This category includes all sorts of accidental errors in the data collection and processing. Unfortunately, the number of such missingness is high in wave 12.
I hope this helps.
UKHLS User Support