Abstract
Recent research has shown that comparative risk perceptions account for unique variance in concern and behavioural intentions, above and beyond absolute self risk perceptions. However, comparative risk perceptions have been measured both directly (i.e. a single question requiring a self-to-peer comparison) and indirectly (i.e. separate, absolute questions about the self and one's peers), yet no study has examined which measure is more predictive of concern and intentions. Two studies examined this issue in the context of several health risks (e.g. cancer, heart attacks). Study 1 showed that direct comparisons were generally more predictive of concern and intentions than indirect comparisons. Study 2 replicated this finding and revealed asymmetries in how people responded to absolute self versus peer questions. Implications for risk screening in health contexts are discussed.
Acknowledgements
I would like to thank Marie Helweg-Larsen, Bill Klein, Paul Windschitl, and anonymous reviewer for their helpful coments on a previous version of this manuscript.
Notes
Notes
1. Although considerable research has suggested that risk perceptions influence concern/worry, the opposite may also be true: concern/worry may impact risk perceptions (Lipkus, Klein, Skinner & Rimer, Citation2005; Loewenstein, Weber, Hsee & Welch, Citation2001). As will be described later, the current research utilised a cross-sectional design and it is therefore impossible to distinguish between these different directions of causality in the current studies. However, it is notable that some research has suggested that worry mediates the relationship between risk perception and preventative behaviours (Chapman & Coups, Citation2006).
2. The language used here implies that there is, in fact, a full and deliberate comparison being made by participants when answering direct comparative questions. Some research has suggested that direct comparative estimates are highly conflated with absolute self estimates and that instead of making a true ‘comparison’, participants may simply use their absolute self-assessments as a proxy (e.g. Gold, Citation2007; Klar & Giladi, Citation1999). Although it is clear that there is substantial overlap between these measures, there is evidence elsewhere to suggest that responses to ‘comparative’ estimates do involve some consideration of both the self and the referent (Windschitl, Rose, Stalkfleet, & Smith, in press) and that absolute self and direct comparative estimates are not perfectly correlated (Lipkus et al., Citation2000). Further separating these constructs is evidence that absolute self and direct comparative estimates independently predict health outcomes (Zajac, Klein, & McCaul, Citation2006).
3. It is important to note that absolute self estimates were also quite predictive of concern and intentions (the average of the rs across all events was 0.51 for concern and 0.36 for intentions). However, it is also notable that there was evidence that direct comparative estimates and absolute self estimates were both uniquely predictive of concern and intentions (see also, Klein, Citation2002; Lipkus et al., Citation2000; Zajac et al., Citation2006). These results will be returned to in the ‘General discussion’.
4. After participants answered the ‘statistical estimate’ questions, they were also asked about the extent to which they went with their ‘gut’ when answering each risk question. This was included because previous research has shown that gut-level estimates are more related to behaviours than numeric-level estimates (see, e.g. Windschitl & Wells, Citation1996), and it was speculated that direct measures may be answered in a more ‘gut-level’ fashion. However, there were no significant differences found across the three question types, thus these analyses were not included in the text to save space.
5. As in Study 1, it is important to note that absolute self estimates in Study 2 were also quite predictive of concern and intentions (the average of the rs across all events was 0.49 for concern and 0.21 for intentions).