ABSTRACT
The methods employed to measure behaviour in research testing the theories of reasoned action/planned behaviour (TRA/TPB) within the context of health behaviours have the potential to significantly bias findings. One bias yet to be examined in that literature is that due to common method variance (CMV). CMV introduces a variance in scores attributable to the method used to measure a construct, rather than the construct it represents. The primary aim of this study was to evaluate the impact of method bias on the associations of health behaviours with TRA/TPB variables. Data were sourced from four meta-analyses (177 studies). The method used to measure behaviour for each effect size was coded for susceptibility to bias. The moderating impact of method type was assessed using meta-regression. Method type significantly moderated the associations of intentions, attitudes and social norms with behaviour, but not that between perceived behavioural control and behaviour. The magnitude of the moderating effect of method type appeared consistent between cross-sectional and prospective studies, but varied across behaviours. The current findings strongly suggest that method bias significantly inflates associations in TRA/TPB research, and poses a potentially serious validity threat to the cumulative findings reported in that field.
Acknowledgements
We thank Megan Andrews, Madalyn Oliver, Alexander Svenson and Celeste Coltman for help in screening the literature, data extraction and coding and Murad Safadi for designing and maintaining the database used to handle extracted data. We also thank two anonymous reviewers and the journal editor for their helpful suggestions on earlier drafts.
Disclosure statement
No potential conflict of interest was reported by the authors.
ORCID
Máirtín S. McDermott http://orcid.org/0000-0002-2640-3161