Drive Theory

The fear-as-acquired drive model (often referred to as the drive model) argues that stimuli that evoke fear result in a drive to avoid the unpleasant emotional state (Hovland, Janis, & Kelly, 1953). Under this model, any behavior that effectively alleviates fear will then become automatically associated with relief from the aversive state, and that behavior will become a habitual response to threatening stimuli. Hovland, Janis, and Kelly noted that habituated behaviors can be adaptive in that they target the threat itself and aim to reduce or eliminate it. These behaviors can also, however, be maladaptive if they do not result in behavioral change but instead involve avoiding the topic or becoming defensive toward the message source.

In an early test of this model, Janis and Feshbach (1953) examined the connection between fear arousal and the persuasiveness of dental-hygiene messages. In their experiment, participants who saw the most graphic and frightening images of tooth decay and poor oral hygiene rated the material as the most worrisome and interesting compared to those in other conditions who saw less disturbing images or just heard a lecture about cavities. However, the participants in this high-fear condition were actually less likely than those who saw milder images about tooth decay to improve their brushing and flossing habits in a post-test one week after exposure to the initial fear appeal.

In light of findings like these, Janis (1967) later argued that the relationship between the elicitation of fear and adoption of the target behavior would be an inverted U-shaped curve. That is, too little fear or too much fear would push individuals toward maladaptive behaviors that quickly reduce the aversive state, such as problem minimization or source derogation, without offering true protection from the threat. However, a moderate amount of fear could be just enough to motivate more adaptive actions to neutralize the threat. Empirical research, however, did not readily support the predicted curvilinear relationship between fear arousal and influence.

Parallel Process Model (PPM)

When a growing body of empirical evidence found a positive linear relationship between fear and persuasion (that is, the more frightened participants were, the more persuasive the message, generally speaking), the drive model no longer seemed a plausible explanation for why fear appeals affect audiences (Beck & Frankel, 1981; Giesen & Hendrick, 1974; Mewborn & Rogers, 1979). As the drive model lost favor, researchers began exploring alternative explanations for why stronger fear appeal messages could lead to greater adaptive responses, while still accounting for the evidence that sometimes, fear appeals lead to maladaptive responses. Most notably, Leventhal (1970, 1971) argued that fear appeals evoked two separate but simultaneous processes: an emotional fear control reaction and a primarily cognitive danger-control reaction. In his parallel process model (PPM), Leventhal argues that one of two responses to a fear appeal is likely. If a person focuses on managing the threat they perceive in the environment, Leventhal argues that danger-control processes are engaged to manage the threat. As such, adaptive outcomes are likely. However, if a person focuses on controlling the emotion of fear, fear control processes are engaged, which ultimately result in maladaptive outcomes, such as avoidance, denial, and reactance. Though conceptually appealing, the PPM did not specify when individuals were more likely to rely on fear-control or danger-control processes. Thus, the model proved unhelpful for both predicting responses to fear appeals as well as to fear appeal message design.

Protection Motivation Theory (PMT)

Given that theories of motivation were losing favor to those that focused on cognition in the field of psychology in the 1970s, it is unsurprising that the next major advance in fear appeal theorizing focused on the cognitive elements associated with fear appeal effectiveness. The protection motivation theory (Maddux & Rogers, 1983; Rogers, 1975) elaborated on the danger-control branch of the PPM, explicating four different cognitive reactions to fear appeals. These included two perceptions related to the potential threat: severity of the threat and susceptibility to it. The other two perceptions related to the target behavior: the effectiveness of the target behavior for avoiding the threat (response efficacy) and the ability of the individual to enact the target behavior (self-efficacy). Maddux and Rogers (1983) argued that individuals who were high in both threat and efficacy perceptions would be motivated to protect themselves from the threat at hand and therefore the most likely to engage in adaptive, danger-control behaviors. Though the four cognitions identified in the PMT proved important to persuasive outcomes in response to fear appeals, the specific relationships among them as asserted by the theory were not fully supported by empirical evidence. Despite accumulating evidence that fear appeals could be persuasive, the field remained without a good explanation for the mechanisms of these message effects.

Extended Parallel Process Model (EPPM)

In light of the limitations of previous theorizing, and coinciding with an increased interest in affect and emotion in the psychology literature (Lazarus, 1991), Witte (1992) proposed the extended parallel process model (EPPM). The EPPM represented a merger of the PPM and the PMT, but it also recognized the importance of the emotion of fear itself in understanding message effects. Broadly, the EPPM posits that message-relevant fear arises in response to threat appraisals. More specifically, the EPPM argues that if a message contains information on threat severity and susceptibility, fear will be aroused. Although fear may predispose audiences to engage in defensive motivations, it also prompts consideration of efficacy potential. If the threat appraisals are more dominant than the efficacy appraisals, individuals will seek to diminish their aroused fear through maladaptive behaviors, such as denigrating the message source or rationalizing their unhealthy behaviors. However, if efficacy appraisals are strong enough to dominate threat appraisals, the EPPM predicts that individuals will feel capable of addressing the threat and will then enact adaptive, danger-control behaviors.

Although the predicted interaction between threat and efficacy appraisals lies at the heart of the EPPM, the full model includes 12 propositions detailing numerous intricate relationships between the different types of cognitions, fear, and message effectiveness (Witte, 1992). Visually, the EPPM is represented as a path model whereby the four message components in the external stimuli (i.e., self-efficacy, response efficacy, susceptibility, and severity) predict their related cognitive perceptions. Next, these cognitive perceptions form a feedback loop with fear, as well as interact with each other to either predict protection motivation (which then predicts message acceptance) or defensive motivation (which then predicts message rejection). From a message-design standpoint, the EPPM suggests that fear-arousing messages should include both threat and efficacy information if they are to persuade audiences to take adaptive actions. That is because threat perceptions are necessary to elicit fear but efficacy perceptions are necessary to promote protection motivation.

As with previous models, though EPPM offers an appealing explanation for when fear appeals may be more or less persuasive, research has not supported all of the EPPM’s key predictions. That is, though significant relationships between fear and persuasion have been identified, meta-analyses have found mixed support for the threat by efficacy interactions predicted by the EPPM (de Hoog, Stroebe, & de Wit, 2007; Witte & Allen, 2000). Further, many of the 12 propositions originally articulated within the EPPM have not been consistently tested. Indeed, in a systematic review of the EPPM literature, Popova (2012) found that two of Witte’s (1992) propositions, (a) when perceived efficacy is at a moderate level, perceived threat will have an inverted-U-shaped effect on message acceptance, and (b) when perceived efficacy is at a high level, there is a reciprocal relationship between perceived threat and fear, have yet to be empirically tested. In order to fully support or reject tenets the original EPPM, additional research is needed.

Cognitive Functional Model (CFM)

Although no new models of fear appeals have specifically been advanced since the EPPM, theorists have considered how a range of emotions, including fear, might generate persuasive effect. For example, Nabi (1999) proposed the cognitive-functional model (CFM) to advance the literature on negative emotional appeals generally, including specific predictions regarding the role of fear in mediated persuasion. Combining aspects of cognitive-response theories of persuasion with the appraisal-based perspectives on emotion, the CFM focuses on the role of three concepts: motivated attention, motivated processing, and message expectations. Motivated attention refers to the degree of approach or avoidance response to the message based on the receiver’s initial emotional response; motivated processing refers to how motivated the message receiver is to process the message carefully; and message expectations pertain to the audience’s degree of certainty that the message will offer reassurance or not.

The CFM predicts that because fear’s action tendency motivates individuals to avoid threatening situations, message-related fear should predispose audiences to want to avoid giving further attention to the message. Yet fear’s associated motivation for protection from the threat may still lead to message processing, depending on the audience’s expectations regarding the remainder of the message content. Specifically, the CFM predicts that an individual experiencing message-relevant fear who is also uncertain as to whether the rest of the message will contain reassuring information (i.e., it may help alleviate fear, but full exposure is necessary to know for sure) would be most likely to carefully process the message. This deeper processing in the face of uncertainty occurs, according to the CFM, because the audience member is trying to find some piece of information that might satisfy the fear-induced goal of protection. In an initial test of the model, the predictions that fear promotes less careful message processing relative to anger and uncertainty of message reassurance results in more careful processing that certainty of reassurance were supported. However, the prediction that fear, but not anger, would be associated with less careful information processing when expectation of reassurance was high was not supported. Still, the CFM introduces the notion that the emotion of fear influences how carefully persuasive messages are processed, which in turn has implications for the direction and endurance of persuasive effect.

Emotions as Frames Model (EFM)

A second more general approach to considering the influence of discrete emotions, including fear, on persuasive outcome is the emotions-as-frames perspective, which argues that an audience member’s emotional state provides a lens through which individuals pay attention to and process message information (Nabi, 2003). Rooted in appraisal theories of emotions (e.g., Lazarus, 1991), the EFM argues that when a message contains information that is relevant to an emotion’s core relational theme, or the essential perception that underlies an emotional experience, that particular emotion is aroused. For example, the core relational theme of fear is imminent threat. So, if an audience member appraises a message as representing an imminent threat to her own well-being, she will likely experience fear. The EFM then predicts that once an emotion is experienced, emotion-consistent information will be made accessible from memory. Additionally, audiences will be motivated to attend to message information that is consistent with the goals of the aroused emotion (e.g., protection, in the case of fear). Finally, given the expectation that emotional experience directs information processing and information accessibility, the EFM predicts that decision-making and action will be heavily influenced by these emotion-driven processes. In short, message consumers’ emotional states frame how they respond to the rest of the message, which in turn results in emotion-consistent decision-making and action.

Multiple studies have supported the predictive power of the emotions-as-frames perspective, demonstrating not only that different emotion frames lead to different ways of viewing problems and preferred solutions but also that emotion is an important mediator of that process (e.g., Kühne & Schemer, 2015). Although it is applied more in the context of news than persuasion, the EFM has the potential to assist in understanding how fear-based messages might generate persuasive influence by emphasizing how emotion-relevant information is more likely to be attended to later in the message and how post-message behaviors are likely to be consistent with emotional motivations.

Emotional Flow

Throughout the history of the study of fear appeals, any message that evoked negative emotion would often be categorized as a fear appeal. More recently, scholars have adopted a more nuanced approach, noting that messages that include both threat components and efficacy components, and thus are well-categorized as fear appeals, can evoke other emotions in addition to fear and that these other emotions have unique influence on persuasive outcomes. The recently advanced emotional flow perspective (Nabi, 2015; Nabi & Green, 2015) builds on this recognition by arguing that messages, including fear appeals, evoke multiple emotions in sequence as the contents of the messages unfold. It further asserts that the ordering or shifts in emotional states in response to changing message content may be critical to understanding persuasive outcomes. For example, fear appeals begin with threat information followed by efficacy information. The emotional flow perspective suggests that the threat information results in fear arousal, and the efficacy information is associated with feelings of hope. As such, the emotional flow perspective suggests that the emotional shift from fear to hope may help to explain the conditions under which fear appeals are more likely to be effective. Although as yet untested, the emotional flow perspective is unique in highlighting that emotional states like fear can be fleeting and rapidly evolve into other states, and this sequencing could be useful in persuasive message design.

Gender

Evidence from a meta-analysis of fear appeals found that appeals targeting female-dominated audiences are more effective than those targeting male audiences, likely because women tend to be more prevention focused in that they are more sensitive to avoiding negative outcomes than to seeking positive outcomes (Tannenbaum et al., 2015). A review of the literature on more general gender differences in fear may also help explain potential gender differences in responses to fear appeals. This body of work suggests that, in general, women report being afraid of more items or events and having more severe fear reactions to frightening items or events than men. However, these gender differences are likely moderated by gender-role socialization processes such that expressions of fear are typically looked down on for men but not women, resulting in a cycle whereby fearful reactions by women are both tolerated and expected by other members of society (McLean & Anderson, 2009). This raises the question of how gender-role socialization may influence how men, in particular, respond to fear appeals in terms of experienced fear versus reported fear, and the implications for both message design and application.

Age

The use of fear to motivate behavior change is well-documented across the lifespan. While some work suggests age is not a moderator of fear-appeal effects (Witte & Allen, 2000), the role of developmental stages in the processing of and behavioral reactions to fear appeals has yet to be thoroughly or rigorously tested. Indeed, with developmental stages in mind, there is reason to believe that different message structures would be more appropriate for different age groups.

Beginning with the very young, testing the influence of fear-based messages on children is a difficult area of empirical inquiry given the ethical implications of potential psychological trauma. However, cautionary tales, which tell the stories of people who ignore the dangers of a forbidden act and suffer as a result, have frequently been used to encourage compliant behavior in children (P. J. Miller & Moore, 1989). As such, it is evident that fear is used to persuade even very young audiences. Although not specifically documented, the pervasiveness and endurance of cautionary tales in children’s literature suggest that such narratives are believed to be influential; and in light of the growing evidence of the persuasive influence of narratives on adults (Green, 2006; Green & Brock, 2000), there is reason to believe such stories do, indeed, have the desired influence on children. Thus, it is reasonable to imagine that fear-evoking narratives—whether in the media or told by parents—are effective vehicles for persuading children, perhaps even more so than didactic fear-based messages. Further, at a more general level, fear-based messages that alert children to the importance of making self-protective decisions while they are still developing their schemas for health-related behaviors, such as regular teeth brushing or healthy eating, may result in the creation of sustainable habits that eventually result in improved health across the lifespan.

The use of fear appeals to discourage adolescents from engaging in risky health behaviors, such as smoking and drug use, is well-documented and unsurprising, given that adolescents are often assumed to perceive themselves as less vulnerable to risk. However, some evidence suggests a more nuanced picture. That is, adolescents may not perceive themselves to be less vulnerable to risk as much as they actually perceive a range of health-threatening behaviors as less risky than they actually are, compared to their adult counterparts. Moreover, they may perceive themselves to be less able to avoid the harms associated with risky health behaviors (Cohn, Macfarlane, Yanez, & Imai, 1995). Thus, highlighting both actual risk of health-threatening behavior, especially occasional ones, to enhance perceived severity, as well as boosting self-efficacy to protect oneself from harm, may be particularly important message goals to match the psychological profile of adolescents. Further, there is evidence that campaigns targeting adolescents’ risky health behaviors are at high risk for failure due to the likelihood of reactance, or resistance to having behavioral choices controlled by others. Given that fear-based messages, which may be perceived as a more manipulative message style, carry a higher likelihood of defensive processing and reactance, fear appeals may be a risky message strategy to use for this age group. As such, careful attention to both the contexts in which fear appeals are most appropriate for adolescent audiences and how such messages are structured is warranted.

At the other end of the age spectrum, there is reason to believe that the elderly may also have unique reactions to fear-evoking messages compared to younger people. Socioemotional selectivity theory argues that the elderly and the terminally ill are more likely to pay attention to positive information than to negative information (Carstensen, Fung, & Charles, 2003). This is because present-oriented goals are more important than future-oriented, protection-related goals when one perceives her timeline to be limited. This theoretical perspective also argues that older individuals have better memory for positive information than for negative information, suggesting that the threat component of fear appeals may not be as memorable as the efficacy component for this particular audience. Thus, it may be that fear appeals that emphasize efficacy over threat may be more successful for this demographic, though future research is needed to address this question directly.

Based on the literature and existing theory spanning from children to the elderly, we suggest that though fear-based messages may be useful to address health issues across the lifespan, there are unique challenges to consider for each age group. From children and adolescents to young adults, middle-aged individuals, and finally the elderly, using diverse samples with wide age ranges or investing in long-term cohort and longitudinal studies may help reveal potential differences in fear-arousal and fear appeal effects across developmental stage. Such research would, in turn, help determine what message structures would be best suited for which age groups and within what health contexts.

Personality Traits

Although little attention has been given to how specific personality traits impact reactions to fear appeals, there are good reasons to believe that they matter and should be looked at more carefully in the future. Given personality traits influence perceptions of events, and given emotions are based on such perceptions, personality traits could influence whether or not a fear appeal is likely to evoke fear, to what degree, and toward what end. In essence, different types of audiences with different predispositions are likely to respond differently to fear appeals.

One trait variable that has been studied for its contribution to fear arousal is trait anxiety, or the dispositional tendency toward experiencing anxiety and worry across situations. However, there is mixed evidence as to trait anxiety’s impact on fear appeal effects. Witte and Allen (2000), in their meta-analysis, did not find a significant relationship between trait anxiety and post-message attitudes, intentions, or behaviors. However, trait anxiety has been linked to other outcomes of fear appeals. For example, in a study of fear appeals about condom use and AIDS prevention, Witte and Morrison (2000) found that whereas trait anxiety was not related to threat perceptions, efficacy perceptions, or behavior, participants low in trait anxiety reported greater defensive avoidance than did those high in trait anxiety. This relationship could indicate that individuals who are already in an anxious state of mind when presented with a fear appeal may be more accepting of the message than those who come to a message with less anxiety.

Another individual difference that may influence reactions to fear appeals is monitoring versus blunting responses to information (S. M. Miller, 1987). People cope with potentially threatening information in different ways. Some individuals, known as monitors, tend to experience more anxiety and prefer more, rather than less, information about health threats; whereas blunters instead avoid information related to threats. As such, monitors and blunters may respond differently to fear appeal messages. Although little research speaks to this issue, Williams-Piehota, Pizarro, Schneider, Mowad, and Salovey (2005) found that participants classified as blunters responded with less negative affect to mammography messages that were simple and direct versus lengthy and detailed in their presentation of threat and efficacy information. Further, blunters were significantly more likely to actually obtain a mammogram in response to the more direct messages. Monitors, however, responded to both message types similarly. Thus, it may be that blunters may respond better to shorter, more direct fear appeals whereas monitors may be more willing to engage with longer or more detailed fear-inducing messages. However, far more research is needed to confirm this finding and to explore other ways in which monitoring and blunting, and indeed other individual differences, suggest implications for fear-based messaging.