close Icon

Effects of Question Type and Order When Measuring Peak Consumption of Risky Drinking Events.

Yeung JT, Livingston M, Callinan S, Wright C, Kuntsche E, Room R, Dietze P

VIEW FULL ARTICLE
  • Journal Alcohol and alcoholism (Oxford, Oxfordshire)

  • Published 18 Aug 2021

  • Volume 55

  • ISSUE 6

  • Pagination 631-640

  • DOI 10.1093/alcalc/agaa076

Abstract

There is new interest in measuring alcohol consumption during risky drinking events, but there is little guidance on how to best ask such questions. In this study, we contrast two different types of questions on peak consumption over a single heavy drinking occasion. We used a general question that ask respondents to recall the total amount consumed (total consumption question), and location-specific questions that ask respondents to recall consumption in each drinking location (location-specific peak consumption, LSPC).

Heavy drinkers (≥11 Australian Standard Drinks (ASD) per occasion for males, ≥8 for females) from the second wave of a prospective cohort study were recruited via landline random digit dial from Melbourne in 2012. Respondents were randomly assigned to surveys of different question order, and either first received total consumption (n = 127) or LSPC questions (n = 147). T-tests compared peak consumption between categories stratified by sex and consumption tercile.

Mean peak consumption was 12.5 ASD. Irrespective of question order, consumption amounts for total consumption and LSPC questions were not significantly different for both sexes. However, drinkers in the highest tercile asked LSPC questions first provided significantly higher consumption estimates in response to the total consumption question than in response to the LSPC questions.

At a population level, LSPC and total consumption questions produce similar estimates of peak consumption for risky drinking events. Except for heavy drinkers, general consumption questions may be sufficient when asking about these drinking events in consumption surveys, without the greater response burden of longer LSPC questions.