Can—and should—automaticity be self-reported using a single item? A secondary analysis of 16 datasets

Abstract

Understanding the influence of habit on health behaviour, or the formation or disruption of health habits over time, requires reliable and valid measures of automaticity. The most used measure, the Self-Report Behavioural Automaticity Index (SRBAI; derived from the Self-Report Habit Index [SRHI]), comprises four items, which may be impractical in some research contexts. Responding to demand from fellow researchers, this study sought to identify whether and which single items from the SRBAI adequately detect hypothesised effects of automaticity, via secondary analysis of 16 datasets, incorporating 16,838 participants and seven different behaviours. We assessed construct validity through correlations between each item and the full SRBAI (and where possible, the SRHI) and predictive validity by examining correlations with behaviour. All four single-item measures independently met construct and predictive validity criteria. We recognise compelling conceptual and methodological arguments regarding why people should not attempt to assess automaticity via a single, self-report item. However, where circumstances require brief measures, three SRBAI items each offer a credible and practical one-item measure that can substitute for the SRBAI or SRHI. We recommend one item in particular—‘Behaviour X is something I do automatically’—because it tended to most closely replicate the effects of the four-item SRBAI.