Perceived severity also forms part of threat perception in both Protection Motivation Theory (PMT) and the Extended Parallel Process Model (EPPM). In PMT, severity and vulnerability promote health motivation along with efficacy beliefs, but this is offset by the intrinsic and extrinsic rewards associated with 'unhealthy' behaviour and the costs associated with performing the recommended behaviour. In EPPM, however, the focus is solely on the balance between threat and efficacy beliefs. If efficacy beliefs exceed threat levels then health protective advice is followed (‘danger-control’), whereas if threat beliefs exceed efficacy levels then efforts are focused on managing fear (‘fear-control’).
A number of commentators have observed that if the likelihood of experiencing a particular health problem, its perceived severity, or its perceived controllability is zero, then an individual's motivation to act should also be zero (e.g. Feather, 1982; Weinstein, 2000). One conclusion that some researchers have drawn is that vulnerability, severity and efficacy should combine multiplicatively, so that if any one of these three variables holds a value of zero, motivation will be nil. In line with this reasoning, Rogers' (1975) original formulation of PMT held that severity, vulnerability and response efficacy combined multiplicatively. But the lack of empirical support for the predicted relationships led to a revised model in which severity and vulnerability were summed (Rogers, 1983). Rogers did, however, retain the view that there would be second-order interaction effects between threat appraisal and efficacy appraisal.
In the EPPM, Witte also proposes an additive model, suggesting that vulnerability and severity should be summed, but she argues that threat should be subtracted from efficacy (no interaction between the two is proposed). However Witte also states that threat perceptions need to reach a certain threshold level before people become motivated to consider health protective action, though she does not specify this level in numerical terms (Witte, 1998).
More complex theories. Rogers and Witte opted for additive models following the lack of evidence to support interaction effects between vulnerability and severity, but other researchers have argued that there is a lack of good empirical data to properly test the proposed multiplicative relationships between these two variables. Weinstein (2000) observed the expected multiplicative relationship between vulnerability and severity, but he pointed out the difficulty of demonstrating such a relationship in between group analyses with a sample of less than 400. However, Weinstein (2000) also found evidence for a model that was more complex than a multiplicative one, with the latter model only applying when likelihood judgements were less than 50:50. Maddux and Rogers (1983) also found evidence for a complex model (which they describe as ‘sub-additive’) to describe the relationship between vulnerability, response efficacy and self-efficacy, so the issue of how severity, vulnerability and efficacy combine remains an empirical question.
Evidence for The role of severity in predicting behavioural intentions and behaviour. A number of systematic reviews of the predictive utility of severity have been conducted. These have assessed the value of particular theories (e.g. the HBM: Harrison et al., 1992; and PMT: Milne et al., 2000) or the role of perceived vulnerability and severity perceptions in motivating particular behaviours (e.g. uptake of vaccination: Brewer et al., 2007; condom use: Albarracin et al., 2005). In the review of research using Protection Motivation Theory to explain the performance of early detection and prevention behaviours, small but significant associations were observed between perceived severity and both intention and concurrent behaviour, but no significant relationship was found between severity and subsequent behaviour (Milne et al., 2000). In Harrison and colleagues' review of Health Belief Model variables in predicting health behaviours, evidence was found for a small but significant relationship between perceived severity and behaviour in prospective studies (Harrison et al., 1992). Although both of these reviews included studies where the relationship between severity and behaviour may have been obscured because of methodological weaknesses, a meta-analysis looking at the relationship between vulnerability, severity and behaviour, which omitted poorer quality studies, found a small to moderate association between severity and uptake of vaccinations in prospective studies (Brewer et al., 2007). Each of these three reviews examined direct associations between severity and intention/behaviour; they did not assess how severity may interact with other beliefs such as perceived vulnerability. In addition, they all focused on non-experimental studies, thus providing only indirect evidence that severity plays a causal role in behaviour. One review of experimental studies designed to increase condom use found that perceived severity was related to condom use under certain conditions (Albarracin et al., 2005). However the mediational analysis, looking at how the interventions actually worked, combined severity together with likelihood under the broader umbrella of 'threat' and did not report the findings specific to severity alone.