4.3: The Protective Action Decision Model

A review of theories on social influence, persuasion, behavioral decisionmaking, attitude-behavior relationships, protective action, and innovation processes reveals a wide variety of perspectives providing useful accounts of the ways in which risk communication can influence disaster response and hazard adjustment (Lindell & Perry, 2004). Although these theories overlap to some extent with the findings of research on hazards and disasters, all of them provide valuable insights that can extend our understanding of ways in which people respond to the threat of environmental hazards. The relevant elements of these complementary approaches have been integrated with the findings of disaster research to produce a model of the factors that influence individual’s adoption of protective actions against natural and technological hazards and disasters. This integrated model is the Protective Action Decision Model (PADM).

Figure 4-2. Communication Network Model.

According to Lindell and Perry (2004), the PADM is most directly based upon a long history of research on disasters that has been summarized by many authors (Barton, 1969; Drabek, 1986; Fritz 1961; Janis & Mann, 1977; Lindell & Perry, 1992; Mileti, et al., 1975; Mileti & Peek, 2001; Mileti & Sorensen, 1987; Perry,et al., 1981; Tierney, et al., 2001). This research has found sensory cues from the physical environment (especially sights and sounds, see Gruntfest, Downing & White, 1978) or socially transmitted information (e.g., disaster warnings) can each elicit a perception of threat that diverts the recipient’s attention from normal activities. Depending upon the perceived characteristics of the threat, those at risk will either resume normal activities, seek additional information, pursue problem focused actions to protect persons and property, or engage in emotion focused actions to reduce their immediate psychological distress. Which way an individual chooses to respond to the threat depends upon evaluations of both the threat and the available protective actions.

The findings of previous disaster research can be combined with propositions drawn from the theories listed earlier in this chapter to express the PADM in terms of a flow chart that provides a graphic representation of the model (see Figure 4-3). The process of decisionmaking begins with environmental cues or risk communication messages that initiate a series of predecisional processes. In turn, these predecisional processes stimulate either a protective action decisionmaking process or an information seeking process. To proceed through the successive stages of either process, the individual must arrive at an affirmative answer to the questions posed. The dominant tendency is for environmental cues and risk communication messages to prompt protective action decisionmaking, but information seeking occurs when there is uncertainty about the answer to the critical question at a given stage in the protective action decisionmaking process. Once the question is resolved, processing proceeds to the next stage in the protective action decisionmaking process.

Figure 4-3. Information Flow in the PADM.

The model attempts to characterize the way people “typically” make decisions about adopting actions to protect against environmental hazards. The stages within the protective action decisionmaking process are sequential, as are those within the information seeking process. However, few people follow every step in the model in the exact sequence listed in Figure 4-3. For example, an extremely credible (or powerful) source might obtain immediate and unquestioning compliance with a directive to evacuate an area at risk—even if there were no explanation why evacuation was necessary or what alternative protective actions were feasible (Gladwin, Gladwin & Peacock, 2002). Such an order would, of course, be quite improbable in contemporary American society, but compliance with such an order would bypass all of the intermediate stages in the PADM. Other situations can be imagined in which some, but not all, decision stages would be bypassed. The important lesson is that—unless risk communicators have an extreme amount of credibility or power to compel compliance—the more stages in the PADM they neglect, the more ambiguity there is likely to be for message recipients. In turn, greater ambiguity is likely to lower compliance and cause warning recipients to spend more time in seeking and processing information rather than preparing for and implementing protective action. Indeed, ambiguity can initiate a repetitive cycle of decision processing and information seeking that postpones the initiation of protective action until it is too late to be completed before hazard onset.

Predecisional Processes

Both environmental cues and risk communication from other persons prompt three predecisional processes that are needed to bring information to conscious awareness. These are exposure to, attention to, and interpretation of environmental cues or—alternatively—reception of, attention to, and comprehension of socially transmitted information (Fiske & Taylor, 1991). Environmental cues and risk communications are somewhat independent of each other, so one household might only observe environmental cues, whereas another might receive only warnings. Still other households might have access to both environmental cues and warnings. Regardless of whether information comes from environmental cues or social warnings, all three pre-decisional processes are necessary. That is, information from the physical environment will not lead to the initiation of appropriate protective actions unless people are exposed to, heed, and accurately interpret the environmental cues. Similarly, information from the social environment will not lead to the initiation of appropriate protective actions unless people receive, heed, and comprehend the socially transmitted information.

These predecisional processes are critical because some of those at risk who are exposed to environmental cues will heed this information, but others will not. Whether or not people heed the available information is determined by their expectations, competing attention demands, and the intrusiveness of the information. Specifically, expectations of threat are established when people have advance information that leads them to believe the potential exists for a significant environmental impact. For example, many people in tornado-prone areas know the months of the year in which there is a peak level of activity. Consequently, they check weather forecasts frequently and attend to environmental cues such as cloud formations. Competing demands are important because attention is limited, so absorption in one task will tend to prevent the processing of information associated with other tasks. Continuing with the example of tornadoes, people who are engaged in tasks that require intense concentration are less likely to notice gathering storm clouds and might not notice a warning even if they have a radio turned on. Of course, the perceptual intrusiveness of hazard information affects attention because it disrupts cognitive processing of the primary task at hand. Those who did not notice the gathering storm clouds, or even an approaching funnel cloud, are certain to notice the roar of the wind or will notice a warning if it is preceded by a loud signal from a radio or a nearby siren. Finally, interpretation of environmental cues is critical because this requires an understanding of the hazard. For example, some coastal residents have lost their lives because they did not understand that a sudden recession of water is the trough phase of a tsunami. The naïve reaction to receding water has frequently been to confuse it with a sudden low tide and to take advantage of an unexpected opportunity to collect stranded fish. Of course, those who have been properly trained recognize this as a sign of danger and immediately evacuate to high ground.

The predecisional processes for warnings are similar to those of environmental cues. First, people must receive information from another person through a warning channel and attend to this information. Accordingly, the characteristics of the warning channel itself can have a significant impact on people’s reception and attention to warning message content. Once a warning has been received and heeded, some people will comprehend the available information, whereas others will not (what Turner, Nigg & Heller-Paz, 1986, call “hearing and understanding”). The comprehension of warning messages will depend upon whether the message is conveyed in words they understand. Quite obviously, warnings disseminated in English are unlikely to be understood by those who understand only Spanish. In addition, however, comprehension also affected by more subtle factors. A warning source cannot achieve comprehension of a warning message if it uses technical terms that have no meaning for those at risk. For example, phrases such as “hypocenter”, “Saffir-Simpson Category”, “oxidizer”, and “millirem” are specialized terms that will not be understood by all who hear them. Specialized terms cause confusion and distract people from processing the information in the rest of the message. If such terms must be used in warning messages, they should be explained—ideally before any emergencies arise.

Decision Stages

Once the three predecisional processes have been successfully completed, cognitive processing turns to the decision stages in the core of the model presented in Figure 4-3—risk identification, risk assessment, protective action search, protective action assessment, and protective action implementation. In addition, information seeking activities include information needs assessment, communication action assessment, and communication action implementation. Each of the decision stages in the PADM is discussed in detail below.

Risk identification.

According to the PADM, people’s decisions about how to respond to a hazard or disaster begin with risk identification, which can be interpreted as the initial step in what Lazarus and Folkman (1984) call primary appraisal. As noted earlier, this process can be initiated by the detection of environmental cues, but the most important sources of risk identification usually are warning messages from authorities, the news media, and peers such as friends, relatives, neighbors, and coworkers. Conversely, the first step emergency managers must take when promoting the adoption of hazard adjustments is to disseminate their message widely to attract the attention of those at risk and inform them of the potential for environmental extremes that could threaten their health, safety, and property.

In both disaster response and hazard adjustment, those at risk must answer the basic question of risk identification, “Is there a real threat that I need to pay attention to?” (Anderson, 1969a; Janis & Mann, 1977; Mileti, 1975; Perry, 1979a). The importance of the resulting threat belief is supported by research showing individuals routinely try to maintain their definition of the environment as “normal” in the face of evidence that it is not (Drabek, 1986). Researchers have found a positive relationship between level of threat belief and disaster response across a wide range of disaster agents, including floods (Mileti, 1975; Perry, Lindell & Greene, 1981), volcanic eruptions (Perry & Greene, 1982; Perry & Hirose, 1991), hazardous materials emergencies (Lindell & Perry, 1992), hurricanes (Baker, 1991), earthquakes (Blanchard-Boehm, 1998), and nuclear power plant emergencies (Houts, Cleary & Hu, 1988; Perry, 1985).

Risk assessment.

The next step, risk assessment, refers to the process of determining the likely personal consequences that the disaster or hazard could cause (Otway, 1973; Perry, 1979a). Decades of research have shown the perception of personal risk—the individual’s expectation of personal exposure to death, injury, illness, or property damage—is a critical variable in explaining disaster response (Mileti & Sorensen, 1987). This process of assessing personal relevance, which Mileti and Sorensen (1987) refer to as the “personalization of risk”, has been recognized as an important factor by persuasion theorists as well as disaster researchers (Eagly & Chaiken, 1993). In the risk assessment stage, a positive response to the question, “Do I need to take protective action?” elicits protection motivation whether the risk involves a disaster response or long-term hazard adjustment (Fritz & Marks, 1954; Perry, 1983). Some of the factors associated with people’s personalization of risk include “the probability of the impending event occurring [and] the severity, to the individual, of such a development” (Withey, 1962, p. 104; see also Neuwirth, Dunwoody & Griffin, 2000 and Lindell & Perry, 2000 for reviews of relevant research).

The immediacy of a threat is also important because warning recipients must understand that the message describes a threat whose likely consequences will occur in the very near future. Thus, immediacy is related to forewarning, which is the amount of time between the arrival of the warning (or personal detection of environmental cues) and disaster onset. For emergency managers, the amount of forewarning received from hazard detection agencies such as the National Weather Service and US Geological Survey affects their choice of message content, the channels feasible for delivery, and the number of times the warning can be repeated. For those at risk, the amount of forewarning received from emergency managers affects their sense of urgency to act. Other factors being equal, the likelihood of immediate disaster response increases as the amount of time until impact decreases. However, people tend to devote this additional time to other activities such as information seeking and expedient property protection when they believe there is more time before impact than the minimum necessary to implement protective action. Information seeking can ultimately increase compliance with recommended protective actions but does, inherently, delay it. Similarly, the amount of time risk area residents devote to expedient property protection also delays their initiation of personal protective action. In both cases, the delay in protective action might be dangerous because the time of disaster impact cannot be predicted with perfect accuracy. For many of the events studied by disaster researchers, warnings were issued when impact was imminent, thereby reducing the extent of these other activities. Ultimately, increasing the amount of forewarning changes the risk communication from a disaster warning to a hazard awareness message (Perry, et al., 1981; Nelson & Perry, 1991).

Previous research has addressed people’s beliefs about other temporal dimensions of hazard impact, as well. The duration of impact, which refers to the length of time hazard impacts will persist, has been addressed principally in connection with studies of technological risk perception (Slovic, Fischoff & Lichtenstein, 1980; Lindell & Earle, 1983; Lindell & Barnes, 1986). For radiological and toxic chemical hazards, it appears many people are concerned that long-term contamination could prevent them from returning to their homes for a long period of time after a disaster (Lindell, 1994c).

In general, research has shown simple measures of risk perception are positively correlated with disaster response (Drabek, 1999), but it also is important to qualify this finding with one further consideration. Specifically, the hazards most frequently studied by disaster researchers are ones whose principal physical impacts are property damage and traumatic injuries. In such cases, the exposure paths from the hazard agent through the environment to those at risk are relatively simple and well understood by the general public. Physical proximity to the hazard increases risk, so safety increases with distance from the point of impact. Indeed, Kunreuther, et al. (1978) reported proximity, along with certainty and severity, was an important threat characteristic influencing the purchase of hazard insurance. The correlations between risk perception and behavioral responses to that object might not be so high in cases where the exposure paths are more complex than those involved in simple proximity. For example, food contamination and infectious diseases both involve complex exposure paths that might be difficult for most people to understand. For example, some have found it difficult to understand why AIDS can be transmitted by infected needles but not by mosquito bites because the two types of exposure seem to be similar (both involve injection through the skin surface).

Similar issues must be considered in examining hazards that have different types of impacts. For example, Perry and Montiel (1997) reported the magnitude of perceived risk was higher for threats affecting life safety and property than for those affecting property alone. Another issue concerns the definition and measurement of perceived risk. Some studies have used very global measures of risk, whereas others have used more specific measures. Early studies of evacuation compliance that defined risk in terms of three components—certainty, severity, and immediacy—of the threat have reported high positive correlations between risk perception and disaster response (Perry, et al., 1981). However, some researchers have applied these characteristics to the occurrence of a disaster, whereas others have applied them to personal hazard exposure, and still others have applied them to the consequences of that exposure. In some cases, there are essentially no differences among disaster impact, personal hazard exposure, and personal consequences. For example, people living close to a volcano might think the occurrence of a major eruption is highly likely to occur within the next year, their chance of severe exposure is high because they live so close to the volcano, and their chance of experiencing severe adverse health consequences within that time interval is high because the effects of blast and ash will be felt immediately. In other cases, the differences among disaster impact, personal hazard exposure, and personal consequences could be profound. For example, people living in the vicinity of a toxic chemical facility might think the occurrence of a major release to ground water is highly likely to occur within the next year, but also believe their chance of severe exposure within that time interval is low because they live upstream from the release point. Even if they thought the chances of personal exposure were high, they might believe their chance of experiencing severe adverse health consequences within that time interval is low because it would take many years to develop cancerous tumors.

The differences among disaster impact, personal exposure, and personal consequences are important because a number of investigators have found many people have an unrealistic sense of optimism about their ability to avoid danger—in extreme cases, this results in a sense of total invulnerability. For example, data from Lindell and Prater (2000) indicate people’s perceptions that there is a significant probability of an earthquake in their community do not necessarily imply that they believe there is a high probability of being personally affected by that earthquake. Moreover, some studies have indicated perceptions of severity also can be quite complex. Research on earthquake hazard has revealed perceptions of severity to be multidimensional because people are concerned about death, injury, property damage, and disruption to work and daily activities (see Lindell & Perry, 2000, for a review). Other research on risk perceptions regarding radiological and toxic chemical hazards indicates people are also concerned about delayed health effects such as cancers and genetic effects (Lindell, 1994; Lindell & Barnes, 1986; Perry & Montiel, 1997).

Protective action search.

If a threat is judged to be real and some unacceptable level of personal risk exists, people turn to protective action search—which involves retrieving one or more feasible protective actions from memory or obtaining information about them from others. The relevant question in protective action search is “What can be done to achieve protection?” and its outcome is a decision set that identifies possible protective actions. The results of some studies (e.g., Jackson, 1977) suggest risk area residents’ first attempt to answer this question often involves a search for what can be done by someone else to protect them against the hazard. When there is insufficient time to find someone else to provide protection—as is usually the case during disasters—or when such a search is unsuccessful, households must rely on their own resources to achieve protection. In many instances, an individual’s own knowledge of the hazard will suggest what type of protection to seek (e.g., sheltering in the basement following a tornado warning). People are especially likely to recall actions they have taken on previous occasions if they have had personal experience with that hazard. Alternatively, they might consider actions they have taken in the course of their experience with similar hazards—recognizing, for example, the impact of a volcanic mudflow is similar to that of a flood and, thus, protective responses to flood are likely to be effective for a mudflow as well.

Information about protective actions also can be received from a variety of external sources. Specifically, those in the risk area are likely to become aware of alternative protective actions by observing the behavior of others. This occurs, for example, when neighbors are seen packing cars in preparation for hurricane evacuation or employing contractors to reinforce their homes against earthquake shaking. People also are likely to consider actions with which they have had vicarious experience by reading or hearing about others’ actions in response to a hazard. Such vicarious experience is frequently transmitted by the news media and relayed by peers—friends, relatives, neighbors, and coworkers. Finally, people also are made aware of appropriate protective actions by means of disaster warnings and hazard awareness programs that carry protective action recommendations from authorities. Specifically, a well designed warning message will assist recipients in constructing a decision set by providing guidance in the form of one or more protective action recommendations (Mileti & Sorensen, 1988). Nonetheless, authorities should not assume warning recipients will implement the official protective action recommendation even if only one protective action is mentioned in the warning message. People will always be aware that continuing normal activities is an option and they might think of other alternatives by recalling such actions from memory or observing the actions of others.

Protective action assessment.

After people have established that at least one protective action is available, they pass from the protective action search stage to protective action assessment. This involves examining alternative actions, evaluating them in comparison to the consequences of continuing normal activities, and determining which of them is the most suitable response to the situation. At this point, the primary question is “What is the best method of protection?” and its outcome is an adaptive plan.

As noted earlier, choice is an inherent aspect of emergencies because those at risk generally have at least two options—taking protective action or continuing normal activities. Comparing alternatives with respect to their attributes leads, in turn, to a balancing or trade-off of these attributes with respect to their relative importance to the decision maker. Under some conditions, those at risk can only take one action and, therefore, must make a choice among the alternatives. Evacuation maximizes the protection of personal safety, but abandons property to the action of the hazard agent or, as some evacuees have erroneously feared, to looters (Perry, et al., 1981; Lindell & Perry, 1990). On the other hand, emergency measures to protect property (e.g., sandbagging during floods) require the property owner to remain in a hazardous location. This problem also exists in the context of long-term hazard adjustment but is significantly reduced because households have time before disaster onset to carefully consider trade-offs among alternative protective actions and to implement multiple actions. Even when there is only a moderate amount of forewarning, households might be able to engage in a combination of actions. For example, if a flood has been forecast to arrive within a few hours, people could perform emergency floodproofing and elevate contents to higher floors to provide as much property protection as possible, yet evacuate family members before the floodwater reaches a dangerous level.

When households assess the salient characteristics of alternative protective actions, they are likely to consider a set of characteristics that have been identified by previous research on disaster response and hazard adjustment. In reviews of disaster studies conducted since the 1940s, Fritz (1961), Sorensen and White (1980), Sims and Bauman, (1983), Drabek (1986), and Tierney, et al. (2001) have noted that a protective action is unlikely to be considered unless it is considered to be effective in reducing the negative consequences associated with disaster impact. Thus, efficacy, which is measured by the degree of reduction in vulnerability to the hazard, refers to success in protecting both persons and property (Cross, 1980; Kunreuther, et al., 1978). In some cases, such as sandbagging during floods, property protection is the specific objective of the protective action. In other cases, however, people consider the implications for property protection of actions whose principal goal is to protect persons. For example, many researchers have found that those who fail to comply with an evacuation recommendation do so because of concerns about protecting their property from looting.

Research also suggests people evaluate protective actions in terms of their safety—that is, the risks that might be created by taking that protective action. For example, some research has reported that those who have not complied with recommendations to evacuate did so because they were concerned about the traffic accident risks involved. As a general rule, the traffic accident risks of evacuation appear to be no greater than those of normal driving (Lindell & Perry, 1992). However, it is important to recognize warning recipients’ behavior is determined by their beliefs about safety, not the historical evidence about safety. Thus, it is important for local authorities who want to increase compliance with evacuation recommendations to ensure people are aware that evacuation accident rates are low.

Alternative protective actions also can be assessed in terms of their perceived time requirements for implementation, which are a function of the number and duration of the steps required to complete a given action. Evacuation is typically time consuming, requiring unification of the family, preparation for departure, selection of a safe destination and route of travel, and transit out of the risk area (Lindell & Perry, 1987; Lindell, Prater, Perry & Wu, 2002). By contrast, time requirements for in-place protection are small—requiring only that occupants shut off sources of outside air, such as doors, windows, chimney dampers, and forced air circulation systems for heating and cooling (Lindell & Perry, 1992). A major problem in large scale evacuations such as those for hurricanes is people’s underestimation of the amount of time needed to reach their destinations. Kang, Lindell and Prater (in press) found coastal residents have reasonably accurate expectations about the time requirements of familiar tasks under their control (e.g., packing bags and shuttering windows), but they substantially underestimate the amount of travel time needed to clear the risk area. The problem seems to be that they plan to take familiar routes to familiar destinations and assume it will take the usual amount of time to get there. Unfortunately, they fail to account for the fact that an evacuation might have ten times as much traffic on that route as they normally encounter, thus turning a two hour trip into a 20 hour trip.

Perceived implementation barriers

The perceived implementation barriers affecting protective action decisions arise from resource constraints precluding the selection of a preferred protective action, as well as obstacles that are expected to arise between the decision to take a protective action and the achievement of protection. In the former category, resource constraints include a lack of knowledge and skill, tools and equipment, or social cooperation required to achieve protection (Lindell & Prater, 2002). In the case of evacuation, this may include a lack of knowledge of a safe place to go and a safe route to travel. Related barriers include the lack of access to a personal vehicle (e.g., those who are routinely transit dependent or families in which one spouse has the only car during the workday) or lack of personal mobility due to physical handicaps. These were clearly factors affecting the alarming death toll in Hurricane Katrina. In some instances, the separation of family members will be considered to be an evacuation barrier. Until family members have been reunited or separated family members can establish communication contact and agree upon a place to meet, evacuation is unlikely to occur (Killian, 1952; Drabek & Boggs, 1968; Haas, Cochrane & Eddy, 1977). Of course, separation of family members is unlikely to be a significant problem during incidents, such as hurricanes, that have ample forewarning.

Finally, a variety of researchers (Cross, 1980; Fritz, 1961; Kunreuther, et al., 1978; Sorensen & White, 1980) have reported the perceived cost of actions to protect personal safety is a consideration in protective actions decisions. Such costs include out-of-pocket expenses (gasoline, food, and lodging), opportunity costs (e.g., lost pay from workdays missed during evacuation), effort, personal sacrifice, and aesthetic cost (e.g., the unattractive appearance of houses that are elevated out of the flood plain). The high cost of protective action can lead people to delay its implementation until they are certain it is necessary. For example, many households delay hurricane evacuation because they want to avoid incurring evacuation expenses if possible. These averaged $262 per household during the Hurricane Lili evacuation (Lindell, Prater, Lu, Arlikatti, Zhang & Kang, 2004).

A significant impediment to the assessment of protective actions arises when none of the available alternatives dominates the others (i.e., is superior to the others on all of the evaluation attributes). For example, Lindell and Perry (1992) reported evacuation was rated higher than sheltering in-place and expedient respiratory protection in efficacy for protecting persons (a positive consequence). However, evacuation also was judged to be higher in its resource requirements for time, effort, skill, cost, and barriers to implementation (all negative consequences). This suggests people must sometimes make a difficult choice between the higher effectiveness of evacuation and its higher resource demands against the lower effectiveness of the alternative protective actions (sheltering in-place and expedient respiratory protection) and their lower resource demands.

The importance of perceived attributes in the protective action assessment stage should alert risk communicators to the potential for differences between the judgments of experts and the public, especially in connection with protective actions that are not well known to those at risk. Sheltering in-place can substantially reduce toxic gas exposure to safe levels (Wilson, 1987, 1989), but its effectiveness does not seem to be recognized outside a relatively narrow circle of experts. Moreover, attempts to evacuate immediately prior to tornado impact, which are contrary to scientific recommendations (Glass, et al., 1980), are probably due to the recognition that sheltering in-place during the tornado does not guarantee survival. This observation also holds true for many victims of fires in high-rise buildings who have attempted unsuccessfully to evacuate when sheltering in their rooms would likely have saved their lives.

The end result of protective action assessment is an adaptive plan, but people’s adaptive plans vary widely in their specificity, with some being only vague goals (e.g., “We’ll stay with my sister’s family”) and others begin extremely detailed. At minimum, a specific evacuation plan includes a destination, a route of travel, and a means of transportation. More detailed plans include a procedure for reuniting families if members are separated, advance contact to confirm the destination is available, consideration of alternative routes if the primary route is unsafe or too crowded, and alternative methods of transportation is the primary one is not available.

Research has documented a tendency for those who lack a ready adaptive plan to experience more negative disaster outcomes (Quarantelli, 1960; Perry, 1979b; Drabek, 1986). A classic example in the literature on floods lies in the Hamilton, Taylor, and Rice (1955, p. 120) interview with the recipient of an evacuation warning that contained no information on safe evacuation routes or safe destinations: “We couldn't decide where to go... So we grabbed our children and were just starting to move outside...if it had just been ourselves, we might have taken out. But we didn't want to risk it with the children.”

Protective action implementation.

The fifth step, protective action implementation, occurs when all the previous questions about risk reduction have been answered satisfactorily. Specifically, those at risk have determined action should be taken, at least one available option is likely to be effective in achieving protection, and that option is logistically feasible. In general, the implementation of protective actions consumes resources people would prefer to allocate to other activities, so those at risk frequently delay implementation until they have determined that the immediacy of the threat justifies the disruption of normal activities. Thus, people often ask the question, “Does protective action need to be taken now?” The answer to this question, whose outcome is the threat response, is crucial because people sometimes postpone the implementation of protective action even when there is imminent danger. As noted earlier, recipients of hurricane warnings have often been found to endanger their safety because too many of them wait until the last minute to begin their evacuations. Unfortunately, they fail to recognize that adverse weather conditions and a high volume of traffic can significantly reduce the average speed of evacuating vehicles, thus running the risk that their evacuation will not be completed before the arrival of storm conditions (Baker, 1979, 1990, 1993; Dow & Cutter, 1998, 2002; Prater, Wenger & Grady, 2000). The problem of procrastination is even more severe in connection with long-term hazard adjustment than it is in disasters with ample forewarning because hazard awareness programs cannot specify even an approximate deadline by which action must be taken. For example, an earthquake prediction might indicate a 75% chance of a damaging earthquake within the next 20 years.

Information needs assessment.

At all stages of the protective action decision process, people who are responding to the threat of disaster must act on the basis the available information, even if it is insufficient for a confident appraisal of the threat or the available protective actions. However, when people think time is available, they cope with the lack of information by implementing three additional stages involving information search. The process of information search begins with an information needs assessment arising from an individual’s judgment that the available information is insufficient to justify proceeding further in the protective action decision process. The research literature indicates ambiguity at any point in the protective action decision process will tend to initiate information seeking, especially when the probability of disaster impact reaches a critical threshold. Thus, if any of the questions cannot be answered with an unequivocal yes or no, people will ask “What information do I need to answer my question?” so they can generate an identified information need. As is the case with a lack of information about a threat, information seeking can also resolve a lack of information about appropriate protective actions. In particular, additional information about alternative protective actions could make it clearer which action would be most appropriate for that situation. Such information seeking is frequently needed because, as noted earlier, those at risk are rarely aware of all of the alternatives available to them.

Communication action assessment.

Identification of a need for information does not necessarily suggest where the needed information can be obtained. Thus, the next question in the information seeking process is “Where and how can I obtain this information?” Addressing this question leads to information source selection and information channel selection, which constitute an information search plan. The sources from which information is sought are likely to differ depending upon stage of the protective action decision process that has generated the need for information. For example, uncertainty about risk identification and risk assessment can stimulate questions directed to officials and, more likely, the news media (see Lindell & Perry, 1992). The high level of reliance on the news media appears to be due to people’s desire to confirm the information they initially received in a warning message from one source by contacting a different source (Drabek, 1969). By contrast, uncertainties about protective action search, protective action assessment, and—especially—about protective action implementation are likely to prompt questions directed to peers.

The sources sought are likely to be affected by the available channels, which in many disasters precludes the use of the telephone because circuits are so overloaded that it is impossible to obtain a dial tone for hours or even days. Further, attempts to reach authorities sometimes prove futile because emergency response agencies are busy handling other calls. Thus, people are often forced to rely on the mass media and peers even when they would prefer to contact authorities. This distinction between risk area residents’ preferred channels of information receipt and their actual channels of information receipt also can be seen in connection with long-term hazard adjustment. For example, Lindell and Perry (1992) reported residents of communities downstream from the Mt. St. Helens volcano revealed some significant disparities between their preferred and actual channels of information receipt in the years after the 1980 eruptions. However, there also were significant differences between the two communities of Toutle and Lexington in both their preferred and actual channels of information receipt. Unfortunately, the available research does not reveal any general principles of source and channel preference that can be assumed to apply across a broad range of communities.

Communication action implementation.

The final step in the information search process is communication action implementation, which provides decision information by answering the question, “Do I need the information now?” If the answer to this question is positive, that is, they are threatened by an imminent disaster, people will actively seek the needed information from the most appropriate source through the most appropriate channel. Drabek’s (1969; Drabek & Stephenson, 1971) research indicates people will go to great lengths, contacting many people over a period of minutes to hours, if the prospect of an imminent disaster needs to be confirmed. However, information seeking will be less frequent and less active if the location is specific but the time of impact is ambiguous. Perry, Lindell, and Greene (1982) reported many residents of the area around Mt. St. Helens monitored the radio four or more times a day after the initial ash and steam eruptions led authorities to believe increased activity might indicate an increased probability of a larger eruption. By contrast, the absence of locational specificity and time pressure inherent in a hazard awareness program provides little need for those at risk to obtain immediate answers, so they are likely to forego active information seeking in favor of passive monitoring of the situation. Unfortunately, the absence of a deadline for action means this passive monitoring is likely to continue until an imminent threat arises (as in the case of hurricanes and floods) or until a disaster strikes (as in the case of earthquakes).

Communication action implementation can have one of three outcomes. If the query elicits a message that meets the information needs that initiated the search, then information seeking has been successful and the decision maker can return to the point in the protective action decision process that generated the information search. However, if the source is unavailable, the query produces no additional information, or the query produces no useful information at all, then information seeking is unsuccessful. The response to this situation is likely to depend upon an individual’s expectations for success in obtaining the desired information from another source or through another channel. Optimism regarding either of these is likely to motivate further information seeking. Pessimism regarding the success of obtaining the needed information is likely to force the decision maker to attempt a protective action decision on the basis of the information available.

In summary, The PADM provides a framework that identifies the critical stages of information processing relevant to household adoption of protective actions and—for each stage—the activities performed, the typical question asked, and the outcome (see Table 4-1). If an individual cannot determine a satisfactory answer to the question posed at one of the decision stages, then progress toward implementation of a protective action is likely to be delayed and possibly even terminated. If the process terminates due to a negative answer about risk identification, then the decision maker is likely to return to normal activities. If the process terminates due to a negative answer about risk assessment, then the decision maker is likely to monitor the situation. If the process terminates due to a negative answer about the availability or acceptability of protective actions, then the decision maker is likely to enter a state of either denial or panic (Janis & Mann, 1977). Which of these emotion-focused coping strategies is used depends upon a person’s susceptibility to distraction, with the most distractible being inclined to denial and the least distractible being inclined to intense fear. Nonetheless, extensive research reveals a very low incidence of panic in disaster (Drabek, 1986).

Table 4-1. Warning Stages and Actions.

Stage	Activity	Question	Outcome
1	Risk identification	Is there a real threat that I need to pay attention to?	Threat belief
2	Risk assessment	Do I need to take protective action?	Protection motivation
3	Protective action search	What can be done to achieve protection?	Decision set (alternative actions)
4	Protective action assessment and selection	What is the best method of protection?	Adaptive plan
5	Protective action implementation	Does protective action need to be taken now?	Threat response
6	Information needs assessment	What information do I need to answer my question?	Identified information need
7	Communication action assessment and selection	Where and how can I obtain this information?	Information search plan
8	Communication action implementation	Do I need the information now?	Decision information

Source: Lindell & Perry (2004).