Abstract
Aim
To compare data quality of two formats of the weekly recall measure (WR1 vs WR2) and the weekly quantity–frequency alcohol measure (QF1 vs QF2).
Design
Participants were randomly allocated to one of the four formats of alcohol measure.On aggregate level, formats were compared for mean number of alcohol units/drinking days and for item nonresponse. Respondents’ problems with completing the questionnaire were assessed by cognitive interviewing.
Findings
No differences in alcohol consumption were found between WR1 and WR2, and
item nonresponse was higher on WR2 than on WR1. QF2 yielded a higher mean number of drinking days/week than QF1 but no differences in number of units/week; QF2 had a higher item nonresponse rate than QF1. Most problems occurred in averaging and in reporting consumption according to the given instructions.
Conclusions
On an aggregate level, there were no differences in alcohol consumption
between WR1 and WR2, but WR2 had a higher likelihood of item nonresponse. According to the ‘more is better’ principle, QF2 is preferred, but also has a higher item nonresponse than QF1. Interviewing uncovered problems and misreporting that could not be revealed by comparing aggregate scores.
To compare data quality of two formats of the weekly recall measure (WR1 vs WR2) and the weekly quantity–frequency alcohol measure (QF1 vs QF2).
Design
Participants were randomly allocated to one of the four formats of alcohol measure.On aggregate level, formats were compared for mean number of alcohol units/drinking days and for item nonresponse. Respondents’ problems with completing the questionnaire were assessed by cognitive interviewing.
Findings
No differences in alcohol consumption were found between WR1 and WR2, and
item nonresponse was higher on WR2 than on WR1. QF2 yielded a higher mean number of drinking days/week than QF1 but no differences in number of units/week; QF2 had a higher item nonresponse rate than QF1. Most problems occurred in averaging and in reporting consumption according to the given instructions.
Conclusions
On an aggregate level, there were no differences in alcohol consumption
between WR1 and WR2, but WR2 had a higher likelihood of item nonresponse. According to the ‘more is better’ principle, QF2 is preferred, but also has a higher item nonresponse than QF1. Interviewing uncovered problems and misreporting that could not be revealed by comparing aggregate scores.
Original language | English |
---|---|
Pages (from-to) | 164-169 |
Journal | Journal of Substance Use |
Volume | 8 |
Issue number | 3 |
DOIs | |
Publication status | Published - 2003 |