information about probability in a simple visual reference.
Figure 3.2 National Hurricane Center Cone of Uncertainty for 2007 Tropical Storm Dean.
While these models seek to translate risk estimates and projections by subject matter experts into warnings and alerts that can inform the public and persuade them to take appropriate protective action, theory and associated research have focused on the ways in which the public processes this information. It is important to recognize that principles of risk and risk perception as described in Chapter 8 play an important role in how warning messages are received and interpreted. Public perceptions of risk are influenced by a number of factors, including previous experience of the risk, age, institutional factors, media reporting, the technical nature of the risk, and cultural factors (Covello, 2009; Mileti & Sorenson, 1990; Tansey & Rayner, 2009). In addition, risks may be amplified by social factors. According to Kasperson et al. (1988), “Social amplification of risk denotes the phenomenon by which information processes, institutional structures, social-group behavior, and individual responses shape the social experience of risk” (p. 181). Thus, some risks may be elevated in their significance while others are downplayed.
In addition to relative levels of uncertainty and social amplification of risk, two other variables are fundamental to warnings: time and width of diffusion. Risks are time-bound phenomena and the ability to issue a warning quickly is associated with the ability of the public to take the desired protective action in time to avoid the risk. Some threats, such as tsunamis and tornadoes, may emerge very quickly, and warning systems must be structured for very rapid response. In general, the timelier a warning message, the greater the capacity of the public to take protective action.
Related to questions of timing are variables determining how broadly the message is diffused. Factors such as intensity, availability of channels, the channel(s) employed, and time of day all influence the width of diffusion. Warning messages will typically follow the typical S-shaped curve of information diffusion, where the distribution will generally start slowly, build, and then taper off (Rogers & Sorensen, 1991). Messages of warning are also subject to repetition through word of mouth and increasingly through social media such as retweets on Twitter. While systems employing multiple channels have the broadest and most rapid diffusion, some proportion of the public, including the homeless, will not receive a warning message in a timely manner. Theory then generally frames warnings as a specialized communication process and links this process to larger decisional systems and processes. As a form of communication, basic concepts of reception, understandability, consistency, and credibility are important, as is the diminished capacity, or mental noise, that may accompany a risk situation (Covello, 2009). In addition, because warnings are generally inconsistent with the status quo, they often are met with skepticism. Drabek (1999) notes that most often the first response to a disaster warning is denial. Most theories see warning as more than a simple stimulus response process. Rather, the process is typically characterized as involving individuals, messages, behaviors, attributes, perceptions, and social structures.
Hear-Confirm-Understand-Decide-Respond Model
Sociologists exploring the phenomenon of a community’s response to a warning have offered a number of important insights about how warning messages are received and processed. Much of the research on warnings examines the social-psychological response by individuals during the period of hearing a warning until acting or choosing not to act as a consequence (Sorensen, 2000). Theory has sought to explain the warning process and improve practice by structuring messages more strategically and by integrating warning systems. These approaches seek to understand warnings as more than a simple stimulus response phenomenon but as a complex social process that involves interpreting, personalizing, assessing, and confirming the risks and warnings (Mileti, 1995). These processes – both for natural hazard events, such as earthquakes and floods, and technology-based risks, such as nuclear plant accidents – have been described by Mileti and Fitzpatrick (1992), Mileti and Peek (2000), and Sorensen (2000). Warnings, like all human communication, begin with message creation by a sender and message reception by a receiver, who then interprets and responds. Mileti and Sorensen (1990) describe a process of “Hear-Confirm-Understand-Decide-Respond” as fundamental to risk communication in the public response component of public warnings. This framework is consistent with basic models of communication, including reception, interpretation, and response, but has been adapted specifically to the processing of public warning messages.
Mileti and Peek (2000) argue that a public warning system consists of three interrelated subsystems: a detection subsystem, a management subsystem, and a public response subsystem. The detection subsystem consists of the processes of initially identifying a hazard and the potential for severe harm. In many cases, detection occurs through some formalized monitoring system managed by a government agency or organization. In other cases, risks are identified through more informal means. Risk detection is a complex process involving the integration and interpretation of information, often from diverse sources. A number of factors affect the warning system, including the level of noise, failures in foresight, inability to interpret risk cues, breakdowns in vigilance, and various forms of distraction (Seeger et al., 2003). The management subsystem refers to the decision-making processes involved in weighing the risks and determining protective warnings and actions. These processes are most often managed by a response agency or organization and rely heavily on subject matter experts. As described earlier, the implications of issuing warnings are often weighed in a cost-benefit analysis before decisions are made to issue a warning. Public warnings often have significant costs including economic costs associated with social disruption. Risk communication in the detection and management of subsystems typically takes place among officials, often with little direct inclusion of the public. Risk communication in the public response subsystem includes warning the public and takes account of public perceptions, processing of messages, and actions. This final public response system is critical in that public actions, such as evacuations, shelter in place, or boil water, are often the central strategy for mitigating and limiting harm.
Some of the theories that could be employed to understand the public response subsystem include the extended parallel process model (EPPM), fear appeals, the health belief model, and the theory of reasoned action. The health belief model, for example, explains health behaviors as a function of individual perceptions, attitudes, and beliefs (Rosenstock, 1966). Attitudes, beliefs, and perceptions about risk can similarly influence risk mitigation behaviors such as evacuations or shelter in place. The EPPM begins with the assumption that threat is a primary motivator of action. Fear is an emotion while threat is a cognitive response. The EPPM seeks to incorporate the drive for defensive action through behavioral change as well as the ability to take the action (Roberto et al., 2009; Witte, 1992). These and similar approaches seek to explain how information is processed and how messages may influence behavior and thus complement the Hear-Confirm-Understand-Decide-Respond model.
Applications of the Hear-Confirm-Understand-Decide-Respond Model
Mileti’s warning process model has been very influential in the examination of basic questions regarding warning communication. Many of these investigations have used case studies and survey methodologies to examine warning systems for natural disasters such as earthquakes, tornadoes, and potential radiological events. For example, Aguirre et al. (1991) examined the warning system failures associated with the 1987 Saragosa, Texas, tornado that killed 30 people and injured 121. It was found that hearing a warning is facilitated if it occurs in one’s native language, if a strong social network is present, and if the message comes from officials. Mileti and Darlington (1995) investigated the public’s response to earthquake warnings in the San Francisco area, a region prone to earthquakes. They found the public is more likely to hear and respond to a warning message when it is delivered through multiple channels. Clarity of the message also facilitates understanding. The public is likely to respond to a warning message if it is delivered
