Professional Documents
Culture Documents
Lindell Perry PADM
Lindell Perry PADM
net/publication/271767724
CITATIONS READS
204 5,093
2 authors:
All content following this page was uploaded by Michael K Lindell on 17 June 2015.
The Protective Action Decision Model (PADM) is a multistage model that is based on find-
ings from research on people’s responses to environmental hazards and disasters. The PADM
integrates the processing of information derived from social and environmental cues with
messages that social sources transmit through communication channels to those at risk. The
PADM identifies three critical predecision processes (reception, attention, and comprehen-
sion of warnings or exposure, attention, and interpretation of environmental/social cues)—
that precede all further processing. The revised model identifies three core perceptions—
threat perceptions, protective action perceptions, and stakeholder perceptions—that form the
basis for decisions about how to respond to an imminent or long-term threat. The outcome
of the protective action decision-making process, together with situational facilitators and
impediments, produces a behavioral response. In addition to describing the revised model
and the research on which it is based, this article describes three applications (development
of risk communication programs, evacuation modeling, and adoption of long-term hazard
adjustments) and identifies some of the research needed to address unresolved issues.
KEY WORDS: Protective action decisions; protective action perception; risk perception
616 0272-4332/12/0100-0616$22.00/1
C 2011 Society for Risk Analysis
The Protective Action Decision Model 617
provide a brief overview of the model, which will be and relevant stakeholders. These perceptions pro-
followed by more detailed discussion of its compo- vide the basis for protective action decision making,
nents in Sections 2–5. Section 6 will provide a theo- the outcome of which combines with situational facil-
retical assessment of the PADM and Section 7 will itators and impediments to produce a behavioral re-
describe its applications in three areas (development sponse. In general, the response can be characterized
of risk communication programs, evacuation model- as information search, protective response (problem-
ing, and adoption of long-term hazard adjustments). focused coping), or emotion focused coping. In many
The findings of studies on individual response cases, there is a feedback loop as additional environ-
to environmental hazards and disasters can be dia- mental or social cues are observed or warnings are
grammed in a flow chart that provides a graphic rep- received. The dominant tendency is for such informa-
resentation of the model (see Fig. 1). The process of tion to prompt protective action decision making, but
protective action decision making begins with envi- information seeking occurs when there is uncertainty
ronmental cues, social cues, and warnings. Environ- at a given stage in the protective action decision-
mental cues are sights, smells, or sounds that sig- making process. Once the uncertainty is resolved,
nal the onset of a threat whereas social cues arise processing proceeds to the next stage in the process.
from observations of others’ behavior. Warnings are The stages in the PADM characterize the way
messages that are transmitted from a source via a people “typically” make decisions about adopting
channel to a receiver, resulting in effects that de- actions to protect against environmental hazards.
pend on receivers’ characteristics. The relevant ef- These stages are sequential, as are those within the
fects are changes in receivers’ beliefs and behaviors, information-seeking process. However, few people
whereas receivers’ characteristics include their phys- are likely to follow every step in the model in the
ical (e.g., strength), psychomotor (e.g., vision and exact sequence listed in Fig. 1. For example, an ex-
hearing), and cognitive (e.g., primary and secondary tremely credible (or powerful) source might obtain
languages as well as their mental models/schemas) immediate and unquestioning compliance with a di-
abilities as well as their economic (money and ve- rective to evacuate an area at risk—even if there
hicles) and social (friends, relatives, neighbors, and were no explanation why evacuation was necessary
co-workers) resources. or what alternative protective actions were feasi-
Environmental cues, social cues, and socially ble.(14) The important lesson is that—unless warn-
transmitted warnings initiate a series of predecisional ing sources have an extreme amount of credibility or
processes that, in turn, elicit core perceptions of the they have substantial power to compel compliance—
environmental threat, alternative protective actions, the more stages in the PADM they neglect, the more
618 Lindell and Perry
ambiguity there is likely to be for message recipients. One crucial aspect of the social context is the
In turn, greater ambiguity is likely to cause warning network of organizations and individuals that com-
recipients to spend more time in seeking and pro- prise the warning network.(6,22) An original source
cessing information rather than preparing for and im- can transmit a message by means of a broadcast pro-
plementing protective action.(9,15) Indeed, ambiguity cess directly to ultimate receivers (e.g., households)
can initiate a repetitive cycle of information process- and also by means of a diffusion process through in-
ing and information seeking that persists until it is termediate sources who, in turn, relay messages to
too late to complete a protective action before haz- ultimate receivers.(23) These ultimate receivers might
ard onset. also transmit messages to each other, thus resulting
in some people receiving multiple warnings, others
receiving only a single warning, and some people re-
ceiving no warnings. The nature of the warning net-
2. ENVIRONMENTAL AND
work has a significant impact on protective action de-
SOCIAL CONTEXT
cision making because multiple sources often deliver
The physical environmental component com- conflicting messages that require searching for addi-
prises the geophysical, meteorological, hydrological, tional information to resolve the confusion.
or technological processes that generate a hazard and The warning network can communicate informa-
transport it to the locations where people are ex- tion through a variety of different types of channels.
posed (see Ref. 16, chapter 5, for further discussion). These include print (newspapers, magazines, and
These processes generate hazards that vary in the brochures), electronic (commercial radio and televi-
speed of onset, and magnitude, scope, and duration sion, telephone, route alert (broadcast from a moving
of their impacts. One important hazard character- vehicle), tone alert radio, siren, and Internet), and
istic that is often quite relevant for people’s emer- face-to-face (dyadic conversation or group presenta-
gency response is the availability of environmental tion). These channels differ in characteristics such as
cues such as sights and sounds that indicate hazard dissemination rate and precision, penetration of nor-
onset. In tornadoes, for example, the sight of a funnel mal activities, message specificity/distortion, sender
cloud and a roar “like a freight train” (a common de- and receiver requirements for specialized equipment,
scription by tornado victims) provide unmistakable and feedback/receipt verification (see Ref. 5, chapter
indications of imminent threat. In other cases, such 4, for a detailed discussion). Each channel has advan-
as the recession of coastal waters before a tsunami, tages and disadvantages, with channels that provide
the cues are much more ambiguous and likely to be the fastest dissemination often providing the least in-
misinterpreted. Still other hazards, such as ionizing formation (e.g., mechanical sirens). Moreover, peo-
radiation and some toxic chemicals, provide no envi- ple differ in their channel access and preferences.
ronmental cues whatsoever. For example, tornado warnings broadcast over an
The transmission of social information is based English-language radio station missed the population
upon the classic six-component communication of Saragosa, Texas that routinely listened to Spanish-
model of source-channel-message-receiver-effect- language stations.(24) In addition to the official warn-
feedback.(17−21) Sources in the social context consist ing systems, however, peers typically relay informa-
of other people who may transmit information about tion through informal warning systems. Even when
hazards and protective actions, as well as providing peers do not explicitly transmit warning messages,
assistance to reduce the hazard or providing material their behavior—especially obvious preparations for
resources that assist protective response. For exam- evacuation—can serve as social cues for protective
ple, authorities, news media, and peers (i.e., friends, action.(25)
relatives, neighbors, and co-workers) can provide in-
formation about environmental threats and alterna-
3. PSYCHOLOGICAL PROCESSES
tive protective actions. Moreover, emergency man-
agers can dispatch city and school buses to evacuate Psychological processes are defined by three sets
those who lack their own means of transportation. of activities—(i) predecisional processes; (ii) core
They can also provide public shelters to those who perceptions of the environmental threat, alterna-
lack the money needed to stay in commercial facili- tive protective actions, and social stakeholders; and
ties or lack nearby friends or relatives with whom to (iii) protective action decision making. The three
stay. predecisional processes of exposure (whether people
The Protective Action Decision Model 619
receive information), attention (whether they heed ple, in many places along the Oregon coast, moun-
it), and comprehension (whether they understand tains prevent people from receiving signals from
it) are largely automatic processes that take place National Oceanographic and Atmospheric Adminis-
outside of conscious processing.(26) The three core tration Weather Radio transmitters.(29) Even when a
perceptual objects—environmental threats, alterna- warning has been received and heeded, some people
tive protective actions, and societal stakeholders— will fail to comprehend the available information—
can elicit either automatic or reflective judgments, what others called “hearing and understanding.”(30)
depending on the degree to which an individual The comprehension of warning messages depends
has schemas that provide readily accessible and co- upon whether the message is conveyed in words
herent beliefs about those objects. When some- they understand. As the Saragossa tornado indicated,
one has a schema—a generic knowledge struc- warnings disseminated in English are unlikely to
ture defined by instances, attributes that differen- be understood by those who understand only Span-
tiate these instances, and interrelationships among ish.(24) In addition, however, comprehension also af-
the attributes—beliefs about objects encompassed by fected by more subtle factors. A warning message
that schema are rapidly accessed to produce an over- cannot be comprehended if it uses esoteric terms
all judgment that is congruent with the available in- that have no meaning for those at risk. For example,
formation about the situation. Finally, contrary to phrases such as “hypocenter” (earthquake), “Saffir-
widespread belief, panic rarely occurs.(27) Instead, Simpson Category” (hurricane), “oxidizer” (chem-
protective action decision making is often a reflective ical), and “millirem” (radiological) are specialized
process that assesses the available information about terms that will not be understood by all who hear
the threat, alternative protective actions, and social them.
stakeholders to choose a behavioral response. The
research literature suggests that inappropriate disas-
3.2. Perceptions of Threats, Protective Actions,
ter responses are more frequently due to inadequate
and Stakeholders
information than to defective cognitive processing.
Unlike comprehension, which provides the lit-
eral meaning of the words in a warning message, per-
3.1. Predecisional Processes
ceptions of the threat, alternative protective actions,
Regardless of whether information comes from and stakeholders involve broader associations that
environmental cues or social warnings, three pre- are integral parts of the schemas or mental models(31)
decisional processes are necessary to produce a within which the threat, protective actions, and stake-
protective response. Information from the physical holders are embedded. It should be quite obvious
environment will not lead to the initiation of appro- that people will differ from each other in the com-
priate protective actions unless people are exposed prehensiveness of their schemas about these objects.
to, heed, and accurately interpret the environmen- That is, some people will have highly differentiated
tal cues. Similarly, information from the social envi- schemas whereas others have poorly differentiated
ronment will not lead to the initiation of appropri- schemas about an object.
ate protective actions unless people receive, heed,
and comprehend the socially transmitted informa-
3.2.1. Perceptions of Environmental Threats
tion. Whether or not people heed the available infor-
mation is determined by their expectations, compet- The essential attributes of people’s perceptions
ing attention demands, and the intrusiveness of the of environmental threats are generally considered
information.(26) These processes are known to be af- to be probability and consequences, but some well-
fected by the warning recipient’s age,(28) but the ef- known approaches to perceived risk include factors
fects of other demographic characteristics have not such as dread and unknown risks.(32) These addi-
been reported. tional dimensions of perceived risk appear to be
In the case of warnings, people must first receive useful in explaining people’s responses to a broad
information from another person through a warn- range of technologies and societal activities, but it
ing channel and attend to this information. Accord- is not completely obvious how they are related to
ingly, the characteristics of the warning channel itself the perception of, and response to, imminent threats
can have a significant impact on people’s reception from environmental hazards.(33) Research on haz-
and attention to warning message content. For exam- ards and disasters has emphasized a definition of
620 Lindell and Perry
perceived risk in terms of people’s expectations of flooding, and storm wind will cause their death or
the personal impacts from an extreme environmen- injury, kill or injure their loved ones, destroy their
tal event.(7,8) These expected personal impacts in- property, or disrupt their jobs or basic services such
clude death, injury, property damage, and disruption as electric power and water.(25,38)
to daily activities such as work, school, and shop-
ping.(34) Moreover, most research on hazards and dis-
3.2.2. Perceptions of Hazard Adjustments
asters has found that risk perception predicts warning
responses such as evacuation(35) and long-term haz- A substantial amount of research has focused on
ard adjustments.(36) These protective responses have people’s perceptions of natural hazards but there is
been studied for hazards such as earthquakes,(37) also a need for studying their perceptions of natu-
hurricanes and other coastal storms,(38) floods,(39) ral hazard adjustments.(52) Such studies are needed
and volcanic eruptions.(40) However, there have been because the theory of reasoned action (TRA) posits
some studies in which risk perception was unrelated that one’s attitude toward an object (e.g., seismic haz-
to hazard adjustment.(34,40−42) ard) is less predictive of behavior than one’s attitude
In addition to the certainty and severity of the ex- toward an act (seismic hazard adjustments) relevant
pected personal impacts, environmental threats can to that object.(53) Thus, to understand the adoption of
differ in their degree of intrusiveness, which is the hazard adjustments, it is just as important to under-
frequency of “thoughts generated by the distinc- stand the perceived attributes of the hazard adjust-
tive hazard-relevant associations that people have ments as the perceived attributes of the hazard itself.
with everyday events, informal hazard-relevant dis- The identity of these perceived attributes can
cussions with peers, and hazard-relevant information be surmised from studies on the adoption of haz-
received passively from the media” (Ref. 13, p. 125). ard adjustments, which have found support for at-
Hazard intrusiveness is correlated with the adoption tributes such as effectiveness,(54) cost,(55) required
of earthquake hazard adjustments(34,41) and expecta- knowledge,(56) and utility for other purposes.(57)
tions of participating in hurricane mitigation incen- Lindell and Perry(5,13,52) summarized this line of re-
tive programs.(43) search by proposing that hazard adjustments can be
Expected personal impacts and hazard intru- defined by hazard-related and resource-related at-
siveness are related to the recency, frequency, and tributes, both of which differentiate among hazard
intensity of people’s personal experience with haz- adjustments.(58) Hazard-related attributes, such as ef-
ard events.(34,43−45) Such experience can involve ca- ficacy in protecting people and property and use-
sualties or damage experienced by the respondent fulness for other purposes, have been found to be
him/herself, by members of the immediate or ex- significantly correlated with adoption intention
tended family, or by friends, neighbors, or co- and actual adjustment.(41,59,60) Resource-related at-
workers.(34) In turn, hazard experience is often corre- tributes (cost, knowledge and skill requirements,
lated with proximity to earthquake,(46) hurricane,(47) time requirements, effort requirements, and required
and flood(48) sources. In addition to the indirect effect cooperation with others) generally have the pre-
of hazard proximity on risk perception (via hazard dicted negative correlations with both adoption in-
experience), there can also be a direct relation be- tention and actual adjustment, but these have been
tween hazard proximity and perceived personal risk small and nonsignificant in studies conducted to date.
that is determined by a perceived risk gradient relat- The hazard adjustments in these studies have gener-
ing increasing proximity to increased risk.(49) How- ally had small resource requirements, so it is unclear
ever, the resulting risk judgments can be quite inac- if the lack of support for the significance of these at-
curate because there are cases in which people have tributes is due to this factor.(58)
limited ability to identify their location in risk ar-
eas.(50,51)
3.2.3. Perceptions of Social Stakeholders
Information from environmental cues and social
warnings, together with prior beliefs about the haz- Previous research has characterized stakehold-
ard agent, produces a situational perception of per- ers as authorities (federal, state, and local gov-
sonal risk that is characterized by beliefs about the ernment), evaluators (scientists, medical profession-
ways in which environmental conditions will produce als, universities), watchdogs (news media, citizens’,
specific personal impacts. In hurricanes, for example, and environmental groups), industry/employers, and
risk perceptions have been characterized by people’s households.(2,61,62) The interrelationships among
beliefs about the degree to which storm surge, inland stakeholders can be defined by their power over
The Protective Action Decision Model 621
each other’s decisions to adopt hazard adjustments. to the series of decision stages—risk identification,
French and Raven posited that power relationships risk assessment, protective action search, protective
can be defined in terms of six bases—reward, co- action assessment, and protective action implemen-
ercive, expert, information, referent, and legitimate tation. In addition, information-seeking activities in-
power.(63,64) Reward and coercive power are the clude information needs assessment, communication
principal bases of regulatory approaches, but these action assessment, and communication action imple-
require continuing surveillance to ensure rewards mentation. Each of the decision stages in the PADM
are received only for compliance and that punish- is discussed in detail below.
ment will inevitably follow noncompliance.(65) Such
surveillance is rarely feasible, so authorities need to
3.3.1. Risk Identification
rely on bases of power other than reward and co-
ercion. French and Raven’s conceptions of expert According to the PADM, people’s threat percep-
(i.e., understanding of cause and effect relationships tions lead to risk identification, which is equivalent to
in the environment) and information (i.e., knowl- what Lazarus and Folkman call primary appraisal.(72)
edge about states of the environment) power sug- In both emergency response and long-term hazard
gest assessing perceptions of stakeholders’ hazard ex- adjustment, those at risk must answer the basic ques-
pertise. French and Raven’s conception of referent tion of risk identification: “Is there a real threat that I
power is defined by a person’s sense of shared iden- need to pay attention to?”(4,73,74) The importance of
tity with another,(66) which is related to that per- the resulting threat belief is supported by research
son’s trustworthiness. Although trust has been de- showing individuals routinely try to maintain their
fined many different ways,(67) fairness, unbiasedness, definition of the environment as “normal” in the
willingness to tell the whole story, and accuracy are face of evidence that it is not.(2) Researchers have
central.(68) found a positive relationship between level of threat
French and Raven defined legitimate power by belief and disaster response across a wide range of
the rights and responsibilities associated with each disaster agents, including floods,(9,73) volcanic erup-
role in a social network, which raises questions about tions,(15,75) hazardous materials emergencies,(5) hur-
what households consider to be the responsibility ricanes,(25) earthquakes,(76) and nuclear power plant
of different stakeholders for protecting them from emergencies.(77,78)
seismic hazard.(63) This is reinforced by research
on stakeholders’ perceived protection responsibility,
3.3.2. Risk Assessment
which dates from research that attributed low rates
of seismic adjustment adoption to respondents’ be- The next step, risk assessment, refers to the pro-
liefs that the federal government was the stakeholder cess of determining expected personal impacts that
most responsible for coping with earthquakes.(69) a disaster could cause.(8,74) The process of assessing
Much later, respondents had come to believe earth- personal relevance has also been recognized as an
quake preparedness was an individual’s responsibil- important factor by persuasion theorists.(66) In the
ity.(70) The conclusion that a perception of personal risk assessment stage, a positive response to the ques-
protection responsibility leads to a higher level of tion “Do I need to take protective action?” elicits
seismic adjustment adoption is supported by similar protection motivation whether the risk involves a dis-
findings on tornado adjustment adoption.(71) Recent aster response or long-term hazard adjustment.(79,80)
research has found that stakeholders differ signifi- Some of the factors associated with people’s person-
cantly in these perceived characteristics (expertise, alization of risk include “the probability of the im-
trustworthiness, and protection responsibility) and, pending event occurring [and] the severity, to the in-
moreover, these characteristics have significant pos- dividual, of such a development” (Ref. 81, p. 104).
itive correlations with hazard adjustment intentions As this quote indicates, the immediacy of a threat
and actual adjustment adoption.(41,67) is important because warning recipients must under-
stand that the message describes an event whose
consequences are likely to occur in the very near
3.3. Protective Action Decision Making
future. Thus, immediacy is related to forewarning,
Once the three predecisional processes have which is the amount of time between the arrival of
been completed and the three types of core percep- a warning (or personal detection of environmental
tions have been activated, cognitive processing turns cues) and disaster onset. People tend to engage in
622 Lindell and Perry
activities such as information seeking and expedi- quently fail to mention other protective actions such
ent property protection when they believe there is as chlorinating the water or using bottled water.(85)
more time before impact than the minimum neces-
sary to implement protective action.(9,82) Successful 3.3.4. Protective Action Assessment
warning confirmation can ultimately increase compli-
After people have established that at least one
ance with recommended protective actions but does,
protective action is available, they pass from pro-
inherently, delay them. Similarly, the amount of time
tective action search to the protective action as-
that risk area residents devote to expedient property
sessment stage. This involves examining alternative
protection also delays their initiation of personal pro-
actions, evaluating them in comparison to the con-
tective action. In both cases, the delay in protective
sequences of continuing normal activities, and deter-
action might be dangerous because the time of dis-
mining which of them is the most suitable response
aster impact cannot be predicted with perfect accu-
to the situation. At this point, the primary question
racy.(83)
is “What is the best method of protection?” and its
outcome is an adaptive plan.
3.3.3. Protective Action Search Choice is an inherent aspect of emergencies
because those at risk generally have at least two
If a threat is judged to be real and some un-
options—taking protective action or continuing nor-
acceptable level of personal risk exists, people are
mal activities. Comparing alternatives with respect to
motivated to engage in protective action search—
their attributes leads, in turn, to a balancing or trade-
which involves retrieving one or more feasible pro-
off of these attributes with respect to their relative
tective actions from memory or obtaining informa-
importance to the decisionmaker. Under some con-
tion about them from others. The relevant question
ditions, those at risk can only take one action and,
in protective action search is “What can be done to
therefore, must make a choice among the alterna-
achieve protection?” and its outcome is a decision
tives (e.g., either evacuate or shelter in-place; Ref. 5,
set that identifies possible protective actions. In many
p. 155). In other cases, people can take multiple ac-
instances, an individual’s own knowledge of the haz-
tions and must choose which ones to implement first.
ard will suggest what type of protection to seek (e.g.,
For example, people in earthquake prone areas are
evacuation from floods or sheltering in the basement
advised to have a working transistor radio with spare
from a tornado).
batteries, at least 4 gallons of water in plastic con-
In addition, those in the risk area might become
tainers, a complete first-aid kit, and a 4-day supply of
aware of feasible protective actions by observing so-
dehydrated or canned food for themselves and their
cial cues such as the behavior of others. This occurs,
families. They are also advised to strap water heaters,
for example, when neighbors are seen packing cars in
tall furniture, and heavy objects to the building walls
preparation for hurricane evacuation.(84) People also
and to bolt the house to its foundation.(57)
are likely to consider actions with which they have
The end result of protective action assessment
had vicarious experience by reading or hearing about
is an adaptive plan, but people’s adaptive plans vary
others’ protective actions. Such vicarious experience
widely in their specificity, with some being only vague
is frequently transmitted by the news media and re-
goals and others begin extremely detailed. At mini-
layed by peers. Finally, people also are made aware
mum, a specific evacuation plan includes a destina-
of appropriate protective actions by means of dis-
tion, a route of travel, and a means of transporta-
aster warnings and hazard awareness programs that
tion.(9) More detailed plans include a procedure for
carry protective action recommendations from au-
reuniting families if members are separated, advance
thorities. Specifically, a well-designed warning mes-
contact to confirm the destination is available, con-
sage will assist recipients by providing guidance in
sideration of alternative routes if the primary route
the form of one or more protective action recom-
is unsafe or too crowded, and alternative methods of
mendations.(8) However, such guidance is often in-
transportation if the primary one is not available.(86)
adequate. For example, water contamination advi-
sories often suggest boiling drinking water, but fail
3.3.5. Protective Action Implementation
to explain whether boiled water is also required for
making ice, brushing teeth, rinsing vegetables, brew- The fifth step, protective action implementation,
ing coffee, cooking pasta, taking showers, or wash- occurs when all the previous questions about risk
ing dishes, counters, hands, or clothes. They also fre- reduction have been answered satisfactorily. In
The Protective Action Decision Model 623
steam eruptions.(97) By contrast, the absence of lo- plement a protective action that they have selected.
cational specificity and time pressure inherent in a Finally, people might relay warnings to others or dis-
hazard awareness program provides little need for cuss the implications of the information they have
those at risk to obtain immediate answers, so they received.(2) As is the case with the adoption of long-
are likely to forego active information seeking in fa- term hazard adjustments, a perceived lack of urgency
vor of passive monitoring of the situation. Unfortu- in obtaining information about a hazard or protective
nately, the absence of a deadline for action means actions can lead to procrastination.
this passive monitoring is likely to continue until an
imminent threat arises (as in the case of hurricanes
6. THEORETICAL ASSESSMENT OF
and floods) or until a disaster strikes (as in the case
THE PADM
of earthquakes).
As the previous sections indicate, there has been
a substantial amount of research that has been con-
4. SITUATIONAL IMPEDIMENTS
ducted within the framework of the PADM or on
AND FACILITATORS
variables that are part of the PADM (e.g., other re-
The actual implementation of behavioral re- search on the relationship of risk perception to evac-
sponse depends not only on people’s intentions to uation). In addition, there has been some limited
take those actions but also on conditions in their discussion of the relationship between the PADM
physical and social environment that can impede ac- and the findings from research in domains other
tions that they intended to take or that can facili- than hazards and disasters—including persuasion, so-
tate actions that they did not intend to take.(98) In cial conformity, behavioral decision theory, attitude-
most cases, the lack of correspondence between in- behavior theory, and information seeking. However,
tentions and behavior seems to be caused by imped- there has only been limited discussion of the PADM
iments rather than unexpected facilitators; there are in relation to other theories of protective action.
many instances in which people have wanted to evac-
uate but lacked a safe place to go and a safe route to
6.1. PADM Comparison to Other Theories of
travel.(9) Other impediments include a lack of access
Protective Action
to a personal vehicle (e.g., those who are routinely
transit dependent or families in which one spouse has The PADM identifies a series of information-
the only car during the workday) or a lack of personal processing stages relevant to household adoption
mobility due to physical disabilities.(99,100) In cases in- of protective actions and—for each stage—the typ-
volving rapid onset disasters, the separation of fam- ical activity performed, question asked, and out-
ily members can also be an evacuation impediment. come. The characterization of the PADM as a stage
Until family members have been reunited or sepa- model of protective action warrants a comment
rated family members can establish communication in light of Weinstein, Rothman, and Sutton’s cri-
contact and agree upon a place to meet, evacuation tique of stage models of health behavior.(103) Like
is unlikely to occur.(101,102) Janis and Mann’s(4) conflict model, the PADM dif-
fers from stage models such as the Transtheoreti-
cal Model (TTM)(104) and the Precaution Adoption
5. FEEDBACK
Process Model (PAPM)(105) in three important ways.
The final stage in the PADM is a feedback loop First, risk area residents implement emergency re-
directed by the communication action assessment sponse actions such as evacuation and sheltering in-
that returns to the initial inputs—environmental and place temporarily rather than permanently. Thus, the
social cues, and information sources, channel access PADM lacks an equivalent to the TTM’s mainte-
and preferences, and warning messages. This feed- nance stage and the idea of a “termination stage”
back loop is extremely common in sudden onset dis- has a qualitatively different meaning for the situa-
asters because people seek to confirm or contradict tions addressed by the PADM. In the TTM, termi-
any warnings they have received, typically by con- nation is “the stage at which individuals have zero
tacting a different source using a different channel.(9) temptation”(104) to revert to their previous behavior.
Alternatively, some people might be looking to ob- Thus, the TTM is clearly directed toward (health)
tain additional information about the threat, about behaviors that must be maintained indefinitely
which protective action to select, or about how to im- through continuing effort. In the PADM, by contrast,
The Protective Action Decision Model 625
“termination” of an inherently temporary emergency Finally, there are some similarities as well as
response would be the stage at which it is safe for distinct differences between PADM and protection
individuals to revert to their previous (“normal”) motivation theory (PMT).(108−110) PADM’s concep-
behavior. tion of hazard-related attributes (protection of per-
Second, emergency response actions have signifi- sons, protection of property, and utility for other
cantly different time scales for their decision-making purposes) is similar to, but broader than, PMT’s re-
processes than do health behavior changes. During sponse efficacy. However, PADM’s conception of
emergencies, people might pass through all of the resource-related attributes (cost, time and effort
stages of the PADM in a matter of minutes, whereas requirements, knowledge and skill, and required co-
the TTM and PAPM address stages of change that operation) is distinctly different from PMT’s self-
are more likely to unfold over months or years. Fi- efficacy because the latter seems most closely re-
nally, the PADM is typically applied to situations lated to the knowledge and skill component of the
in which emergency managers are transmitting in- resource-related attributes. Moreover, the resource-
formation concurrently to large numbers of people related attributes are characteristics of a protective
who are responding to a single “focusing event”(106) action whereas self-efficacy is a characteristic of the
rather than situations in which health professionals person. An advantage of the PADM formulation is
conduct personal interventions that are tailored to that assessing the perceived characteristics of differ-
individuals in different stages of a behavioral change ent protective actions makes it possible to determine
process. Nonetheless, the PADM can be applied to if people have erroneous perceptions of the alterna-
long-term hazard adjustment as well as emergency tive actions and correct those misperceptions if they
response actions. In such cases as deciding to insure exist. In addition, identifying the characteristics of
houses, elevate them above flood level, or bolt them different protective actions makes it possible to iden-
to their foundations, there are more notable overlaps tify which actions are most likely to be adopted and
between the situations addressed by the PADM, on to initially focus a risk communication program on
the one hand, and the TTM and PAPM, on the other increasing their adoption. Once the most resource-
hand. However, the extent of the overlap needs to be effective hazard adjustments have been adopted (the
examined in future research. ones that provide the greatest efficacy in protect-
In addition to examining the PADM’s compat- ing persons and property for a given level of re-
ibility with stage models of health behavior, fu- sources), the risk communication program can be
ture research also needs to systematically examine redirected toward the more resource-intensive haz-
its theoretical similarities to and differences from ard adjustments. Nonetheless, a focus on (personal)
other psychological models. For example, hazard self-efficacy might be more useful than a focus on
adjustment attributes in the PADM are equiva- (task) resource requirements in some situations. For
lent to TRA’s attitude toward the act. However, example, an emphasis on self-efficacy might be useful
although disaster research has found that confor- when one must implement a single protective action
mity with the behavior of others can be a sig- (e.g., weight loss) whereas a focus on task demands
nificant influence on people’s evacuation decisions, (resource requirements) might be more useful when
there has been no PADM research that has ex- someone can choose among multiple protective ac-
plicitly addressed the subjective norm as it has tions.
been conceptualized in the TRA. Moreover, the
PADM’s premise that people can vary in the ex-
6.2. Identifying the Core Perceptions
tent to which they base their protective action deci-
that Determine Protective Actions
sions upon thoughtful consideration of information
about environmental cues, social context, warning One major implication of the literature cited in
source characteristics, and warning message charac- the previous section is that, despite extensive theo-
teristics or, alternatively, upon unquestioning com- rizing and data collection, it still is not entirely clear
pliance with an authority’s recommendation is sim- what motivates people to take protective action.
ilar to ideas about central versus peripheral routes One of the major contributions of PADM research
to persuasion in the Elaboration Likelihood Model is the finding that, though risk perception is usu-
(ELM).(107) The findings of PADM research that are ally an important determinant of protective action,
compatible with the ELM need to be replicated and there are other perceptions that are also important—
extended.(41,67) and sometimes even more important than risk
626 Lindell and Perry
perception. There is a need to consider other as- vidual mental models can be aggregated into a con-
pects of threat perception, such as hazard intrusive- sensus model. In a somewhat similar manner, the
ness, and also to consider perceptions of protective hazard beliefs approach(30,112) presents some very
actions and stakeholders. Nonetheless, it is not en- broad questions about a hazard to a pretest group,
tirely clear whether the PADM’s core perceptions of identifies common themes in the responses, and de-
threat, protective actions, and stakeholders are them- velops an inventory of fixed-response items that as-
selves sufficient to account for an adequate percent- sess both accurate and erroneous beliefs. Both of
age of the variance in people’s hazard adjustment these inductive approaches are likely to prove most
adoption. useful in characterizing people’s beliefs about new
This issue is particularly important when hazard technologies such as nanotechnology.(113)
experts, such as seismologists and earthquake engi- One of the major impediments to assessing how
neers, try unsuccessfully to communicate event prob- people think about hazards is that they vary in the
abilities and consequences to risk area residents in differentiation of their beliefs about a domain and,
an effort to increase the adoption of hazard adjust- among those with differentiated beliefs, there is vari-
ments. Lindell(111) addressed this problem by char- ation in the degree to which those beliefs are in-
acterizing risk area residents’ perceptions of three tegrated.(114) At one extreme are those who have
hazards (a volcanic eruption of Mt. St. Helens, a few or no beliefs about the domain whereas, at the
toxic chemical release from a railroad tank car, and other extreme, are those who have highly differenti-
a radiological materials release from a nearby nu- ated belief systems characterized by a large number
clear power plant) in terms of four categories of haz- of elements that vary in the strength of their link-
ard characteristics that were consistent with previous ages.(115,116) Consistent with these analyses in other
theorizing about technological hazards(32) and dis- domains, there is evidence that risk area residents
aster impacts.(1) Respondents rated the three haz- vary in the degree to which they have differentiated
ards in terms of hazard agent characteristics (like- schemas or mental models of environmental haz-
lihood of a major release, ease of risk reduction, ards. This variation in people’s hazard models has
and likelihood of release prevention), impact charac- led some researchers to measure respondents’ reac-
teristics (speed of onset, existence of environmental tions to hazards simply in terms of a single question-
cues, scope of impact, and duration of impact), ex- naire item of concern about the problem.(117) If risk
pected personal impacts (immediate death, delayed area residents have only very diffuse conceptions of
cancer, genetic effects, and total property loss), and seismic threat, then a global construct such as con-
affective/behavioral reactions (dread, frequency of cern might be a more accurate characterization of
thought about the hazard, and frequency of discus- their beliefs than the specific dimensions assumed
sion about the hazard). The results suggested that by PADM. Further research is needed to determine
perceived impact characteristics mediated the rela- what proportions of the population have specific be-
tionship between characteristics of the hazard agent liefs, global beliefs, and no beliefs at all about the en-
characteristics and expected personal impacts. Simi- vironmental hazards to which they are exposed and
lar results emerged from a study of household evac- the actions they can take to protect themselves.
uation decision making in response to Hurricane
Ike.(84) Perceived storm characteristics (local land-
7. PADM APPLICATIONS
fall, major intensity, and rapid onset) partially me-
diated the effects of coastal proximity and hurri- The PADM has mostly been applied in three
cane experience on expected personal impacts (surge areas—risk communication programs, evacuation
damage, inland flood damage, storm wind damage, modeling, and long-term hazard adjustment. Each of
and casualties), which, in turn, had a direct effect on these is discussed below.
evacuation decisions.
Another line of research that is likely to prove
7.1. PADM Application to Developing Community
fruitful in understanding people’s conceptions of haz-
Risk Communication Programs
ards is the mental models approach.(31) Such studies
interview individuals to elicit the elements of their Research on which the PADM is based has
beliefs about a specific hazard domain and the in- led to the development of an extensive set of
terrelationships among the elements of that domain. recommendations for risk communication, particu-
After interviewing a number of participants, the indi- larly in communities with ethnic minorities. These
The Protective Action Decision Model 627
recommendations for community risk communica- Table I. Tasks for the Continuing Hazard Phase
tion programs are based upon the distinct differences
Strategic analysis
between risk communication activities undertaken Conduct a community hazard/vulnerability analysis
during the continuing hazard phase (the time be- Analyze the community context
tween incidents) and those taken during an escalating Identify the community’s prevailing perceptions of the hazards
crisis (when there is adequate forewarning of disas- and hazard adjustments
ter impact) or emergency response (when forewarn- Set appropriate goals for the risk communication program
ing is absent). Risk communication during the con- Operational analysis
Identify and assess feasible hazard adjustments for the
tinuing hazard phase is directed toward encouraging
community and its households/businesses
long-term hazard adjustments such as hazard mitiga- Identify ways to provide incentives, sanctions, and
tion, emergency preparedness, and hazard insurance technological innovations
purchase. Risk communication during an escalating Identify the available risk communication sources in the
crisis or emergency response is directed toward en- community
Identify the available risk communication channels in the
couraging appropriate disaster responses. However,
community
risk communication during both phases requires the Identify specific audience segments
development of an effective risk communication pro- Resource mobilization
gram. Such programs need to be carefully developed Obtain the support of senior appointed and elected officials
during the continuing hazard phase because the hu- Enlist the participation of other government agencies
man and financial resources available for environ- Enlist the participation of nongovernmental (nonprofit) and
mental hazard management are usually limited until private sector organizations
Work with the mass media
a crisis occurs. Although resources are more read- Work with neighborhood associations and service
ily available during a crisis or emergency response, organizations
time is often severely limited so improvised ef- Program development for all phases
forts at risk communications can produce spectacular Staff, train, and exercise a crisis communications team
failures.(118) Establish procedures for maintaining an effective
There are five basic functions that should be ad- communication flow in an escalating crisis and in emergency
response
dressed in the continuing hazard phase. These are
Develop a comprehensive risk communication program
strategic analysis, operational analysis, resource mo- Plan to make use of informal communication networks
bilization, program development, and program im- Establish procedures for obtaining feedback from the news
plementation (see Table I). The purpose of strategic media and the public
analysis is to identify community constraints and set Program implementation for the continuing hazard phase
appropriate goals for the overall risk communication Build source credibility by increasing perceptions of expertise
program. The purpose of operational analysis is to and trustworthiness
Use a variety of channels to disseminate hazard information
use the elements of the classic communication model Describe community or facility hazard adjustments being
(source, channel, message, receiver, effect, and feed- planned or implemented
back) to identify the community resources that are Describe feasible household hazard adjustments
available for the risk communication program. The Evaluate program effectiveness
purpose of resource mobilization is to enlist the sup-
Source: Lindell & Perry (2004).
port of stakeholders in the community who are likely
to share an interest in using risk communication to
reduce hazard vulnerability. Program development authorities to determine if the risk communication
for all phases involves the use of available commu- program has been effective.
nity resources to develop a workable system that can The format of Table I might seem to imply that
implement risk communication in both the continu- the five risk communication functions form a simple
ing hazard phase and also in an escalating crisis or linear sequence, but some tasks will be performed
emergency response. As soon as the risk communi- concurrently and the entire process will frequently be
cation program has been developed, it is possible to iterative. For example, some resource mobilization
immediately begin program implementation for the tasks might take place concurrently with the opera-
continuing hazard phase. This function involves con- tional analysis, or tasks conducted during the opera-
ducting the activities that will encourage risk area tional analysis phase might be suspended temporar-
residents to adopt long-term hazard adjustments. It ily in order to return to the strategic analysis and
also involves conducting the activities that will allow refine it.
628 Lindell and Perry
Once authorities have determined that they are ists, obtain further information about the threat and
in an escalating crisis or emergency response, they alternative protective actions, and relay warnings to
need to implement the predetermined risk communi- peers.(2) The broad outlines of this process are under-
cation actions that were developed during the contin- stood,(13,14,122) but there appears to be no research on
uing hazard phase. These include activating the crisis the duration of psychological preparation for protec-
communication team promptly, determining the ap- tive action.
propriate time to release sensitive information, and Logistical preparation is the time it takes for
selecting the communication channels appropriate to a household to assemble the people and resources
the situation. An escalating crisis or emergency re- needed to implement protective action. Thus, evac-
sponse also requires authorities to maintain source uees need to perform tasks such as assembling fam-
credibility with the news media and the public, pro- ily members, packing bags for the trip, and securing
vide timely and accurate information to the news the house before leaving.(82) The time required for
media and the public, and evaluate performance mental preparation and logistical preparation both
through postincident critiques. For further details on contribute to total preparation time, but these com-
the application of the PADM to the development ponents do not seem to be additive because the two
of community risk communication programs, see components cannot be presumed to be mutually ex-
Ref. 13, chapter 5. clusive. That is, some household members might be
searching for information at the same time that oth-
ers are packing bags and still others installing hur-
7.2. PADM Application to Evacuation Models
ricane shutters. Available evidence has documented
Most of the research on the PADM and similar that households’ preparation times vary as a function
models has focused on the prediction of whether peo- of incident characteristics; for example, preparation
ple will engage in a particular behavior. Although times for evacuation from the Mt. St. Helens eruption
this is an important application, there are cases in were noticeably different from those for Hurricane
which it is important to predict when people will en- Lili.(82) Because of these and other differences, pre-
gage in a particular behavior. For example, the tim- dictions of evacuation departure times have been less
ing of a household’s evacuation departure is a critical successful than predictions of households’ evacuation
input to transportation analysts’ evacuation models decisions. Some researchers have tried to improve
because evacuation routes can become overloaded prediction by adopting estimation models such as
if too many vehicles enter the evacuation route sys- neural networks, sequential logit, and survival anal-
tem at the same time. Thus, local officials seeking to ysis but these have had modest success.(87,123−125)
manage the evacuation need to understand the rate Other research has extended the scope of the
at which people receive an initial warning and the PADM to examine evacuation logistics—people’s
time it takes them to prepare to evacuate (see Refs. actions between the time they decide to evacuate
119 and 120 for further discussion of evacuation time and the time they arrive at their evacuation destina-
components). Each warning mechanism (siren, tone- tions(86) as well as the return entry process.(126) The
alert radio, radio, television, and telephone) has a latter process is, in some respects, just the reverse of
characteristic rate of exposure over time.(5,23,121) In the evacuation problem. However, it is more com-
addition, warning mechanisms vary in terms of their plicated because of the logistical problems author-
ability to attract attention and provide comprehensi- ities experience in communicating reentry informa-
ble messages that will change risk area residents’ core tion to evacuees that might be scattered over dozens
perceptions of threat, protective actions, and stake- of towns in multiple states.
holders in the desired directions.
The time households spend in preparing to
7.3. PADM Application to Hazard
evacuate appears to involve two components, men-
Adjustment Adoption
tal preparation and logistical preparation. Mental
preparation is the time it takes to collect and pro- A recent review of research on household adop-
cess the information needed to decide an evacua- tion of hazard adjustments concluded that risk per-
tion is necessary. The existence of a period of men- ception is consistently related to the adoption of
tal preparation has been documented in research hazard adjustments, but perceptions of stakehold-
indicating that people engage in milling, during ers and hazard adjustments are also relevant and
which time they seek confirmation that a danger ex- deserve greater attention.(36) There is considerable
The Protective Action Decision Model 629
evidence that hazard experience increases hazard ad- and Chaiken’s(129) distinction between systematic
justment adoption, but hazard proximity and hazard and heuristic processes. Both theories assert that in-
intrusiveness also appear to play significant roles and cidental factors such as stakeholder perceptions af-
should be the focus of additional research. Finally, fect attitudes and behavior without affecting salient
demographic variables continue to be unreliable pre- beliefs or subjective norms (to use terms from TRA).
dictors of hazard adjustment adoption but should re- Further research is needed to determine if the fail-
ceive continuing attention to assess their effects on ures to find complete mediation are due to additional
risk perception, stakeholder perceptions, and hazard variables specified by the PADM, other variables not
adjustment perceptions, as well as hazard experience, specified by the PADM, or are simply manifestations
hazard proximity, and hazard intrusiveness. of the effects of peripheral/heuristic processing.
One of the major limitations of hazards and dis-
asters studies to date is that the PADM predicts
that some of the variables should form causal chains. 8. CONCLUSION
For example, hazard proximity is predicted to cause
This article has described an updated version
hazard experience, hazard experience is predicted
of the PADM and summarized the research that
to cause risk perception, and risk perception is pre-
supports its principal components. This review in-
dicted to cause hazard adjustment adoption. Another
cludes articles that were cited in Lindell and Perry,(13)
example is that perceptions of information sources’
but also includes other research conducted prior to
expertise, trustworthiness, and protection responsi-
and subsequent to that book’s publication. There
bility should affect risk area residents’ risk percep-
are varying degrees of support for different com-
tions, which, in turn, should affect their adoption
ponents of the PADM. Some components have ex-
of hazard adjustments. That is, each successive vari-
tensive support, whereas others have conflicting or
able is hypothesized to completely mediate the re-
counter-intuitive findings, and still others are largely
lationship between the variable that precedes it and
untested. Future research should seek to resolve the
the variable that follows it so that, for example,
conflicting findings, explain the counterintuitive find-
the estimated effect of stakeholder perceptions on
ings, and examine the untested propositions.
hazard adjustment adoption becomes nonsignificant
In addition to its predictive validity, any theoret-
when controlling for risk perception. Most hazards
ical model should be useful. Accordingly, this arti-
and disasters studies to date have used single equa-
cle has described the ways in which the PADM has
tion models to predict a single dependent variable
been applied to three different areas—risk commu-
such as household hazard adjustment or evacuation
nication, evacuation modeling, and long-term hazard
so their results cannot shed any light on the me-
adjustment. Although these applications are in their
diation hypotheses. Moreover, the few studies that
early stages, the evidence to date suggests that the
did report tests of the PADM’s mediation hypothe-
PADM is continuing to evolve into a useful frame-
ses have found only partial mediation where com-
work for managing societal response to environmen-
plete mediation was expected.(34,44) Nonetheless, the
tal hazards. Applications of the model would not
results did suggest that risk area residents’ aware-
need to produce large changes in behavior to be valu-
ness of the available hazard adjustments and accurate
able. Updating an argument made in an earlier arti-
perceptions of their attributes do mediate the rela-
cle,(41) major disasters are so costly—the economic
tionship between hazard experience and hazard ad-
impact of Hurricane Katrina was estimated to range
justment.(59,127) In addition, Lindell and Prater found
from $81 billion(130) to $140 billion(131) —that even a
evidence that hazard intrusiveness had an effect on
1% decrease in losses would save hundreds of mil-
hazard adjustment adoption that was independent of
lions if not billions of dollars.
perceived personal risk.(34)
In some cases, the failure of a complete me-
diation hypothesis has a theoretically logical expla-
ACKNOWLEDGMENTS
nation. For example, the inability of risk percep-
tion or hazard intrusiveness to completely mediate This work was supported by the National Sci-
the relationship between perceived stakeholder char- ence Foundation under Grants SES 0527699, SES
acteristics and hazard adjustment adoption is con- 0838654, and CMM 0927739. None of the conclusions
sistent with Petty and Cacioppo’s(128) distinction be- expressed here necessarily reflects views other than
tween central and peripheral routes to persuasion those of the authors.
630 Lindell and Perry
REFERENCES 21. Petty RE, Wegener DT. Attitude change: Multiple roles for
persuasion variables. Pp. 323–390 in Gilbert DT, Fiske ST,
1. Barton A. Communities in Disaster. New York: Doubleday, Lindzey G (eds). Handbook of Social Psychology, 4th edi-
1969. tion. New York: McGraw-Hill, 1998.
2. Drabek TE. Human System Responses to Disaster: An 22. Lindell MK, Prater CS, Peacock WG. Organizational com-
Inventory of Sociological Findings. New York: Springer- munication and decision making in hurricane emergencies.
Verlag, 1986. Natural Hazards Review, 2007; 8:50–60.
3. Fritz CE. Disaster. In Merton RK, Nisbet RA (eds). Con- 23. Rogers GO, Sorensen JH. Diffusion of emergency warnings.
temporary Social Problems. New York: Harcourt, Brace & Environmental Professional, 1988; 10:185–198.
World, 1961. 24. Aguirre BE. The lack of warnings before the Saragosa tor-
4. Janis I, Mann L. Decision Making: A Psychological Analysis nado. International Journal of Mass Emergencies and Disas-
of Conflict, Choice and Commitment. New York: Free Press, ters, 1988; 6:65–74. Available at: www.ijmed.org, Accessed 1
1977. May 2010.
5. Lindell MK, Perry RW. Behavioral Foundations of Com- 25. Baker EJ. Hurricane evacuation behavior. International
munity Emergency Planning. Washington, DC: Hemisphere Journal of Mass Emergencies and Disasters, 1991; 9:287–310.
Press, 1992. Available at: www.ijmed.org, Accessed 1 May 2010.
6. Mileti DS, Drabek T, Haas JE. Human Systems in Extreme 26. Fiske ST, Taylor SE. Social Cognition: From Brains to Cul-
Environments. Boulder, CO: University of Colorado Insti- ture. Boston: McGraw-Hill, 2008.
tute of Behavioral Science, 1975. 27. Committee on Disaster Research in the Social Sciences.
7. Mileti DS, Peek L. The social psychology of public response Facing Hazards and Disasters: Understanding Human Di-
to warnings of a nuclear power plant accident. Journal of mensions. Washington, DC: National Academy of Sci-
Hazardous Materials, 2000; 75:181–194. ences/National Research Council, 2006.
8. Mileti, D S, Sorensen JH. Why people take precautions 28. Mayhorn CB. Cognitive aging and the processing of hazard
against natural disasters. Pp. 296–320 in Weinstein N (ed). information and disaster warnings. Natural Hazards Review,
Taking Care: Why People Take Precautions. New York: 2005; 6:165–170.
Cambridge University Press, 1987. 29. Lindell MK, Prater CS. Tsunami preparedness on the Ore-
9. Perry RW, Lindell MK, Greene MR. Evacuation Planning in gon and Washington coast: Recommendations for research.
Emergency Management. Lexington, MA: Heath Lexington Natural Hazards Review, 2010; 11:69–81.
Books, 1981. 30. Turner R, Nigg J, Heller-Paz D. Waiting for Disaster. Los
10. Tierney K, Lindell, MK, Perry RW. Facing the Un- Angeles: University of California Press, 1986.
expected: Disaster Preparedness and Response in the 31. Morgan MG, Fischhoff B, Bostrom A, Atman CJ. Risk Com-
United States. Washington, DC: Joseph Henry Press, munication: A Mental Models Approach. Cambridge: Cam-
2001. bridge University Press, 2002.
11. Burton I, Kates, R, White GF. The Environment as Hazard, 32. Slovic P, Fischhoff B, Lichtenstein S. How safe is safe
2nd edition. New York: Guildford Press, 1993. enough: A psychometric study of attitudes toward techno-
12. Lindell MK, Alesch D, Bolton PA, Greene MR, Larson logical risks and benefits. Pp. 80–103 in Slovic P (ed). The
LA, Lopes R, May PJ, Mulilis J-P, Nathe S, Nigg JM, Palm Perception of Risk. London: Earthscan, 1980.
R, Pate P, Perry RW, Pine J, Tubbesing SK, Whitney DJ. 33. Terpstra T, Lindell MK, Gutteling JM. Does communicat-
Adoption and implementation of hazard adjustments. Inter- ing (flood) risk affect (flood) risk perceptions? Results of
national Journal of Mass Emergencies and Disasters Special a quasi-experimental study. Risk Analysis, 2009; 29:1141–
Issue, 1997; 15:327–453. Available at: www.ijmed.org, Ac- 1155.
cessed 1 May 2010. 34. Lindell MK, Prater CS. Household adoption of seismic haz-
13. Lindell MK, Perry RW. Communicating Environmental ard adjustments: A comparison of residents in two states.
Risk in Multiethnic Communities. Thousand Oaks, CA: International Journal of Mass Emergencies and Disasters,
Sage, 2004. 2000; 18:317–338. Available at: www.ijmed.org, Accessed 1
14. Gladwin CH, Gladwin H, Peacock WG. Modeling hurri- May 2010.
cane evacuation decisions with ethnographic methods. Inter- 35. Sorensen JH. Hazard warning systems: Review of 20 years
national Journal of Mass Emergencies and Disasters, 2001; of progress. Natural Hazards Review, 2000; 1:119–125.
19:117–143. Available at: www.ijmed.org, Accessed 1 May 36. Lindell MK. North American cities at risk: Household re-
2010. sponses to environmental hazards. In. Rossetto T, Joffe H,
15. Perry RW, Greene, M. Citizen Response to Volcanic Erup- Adams J (eds). Cities at Risk: Living with Perils in the 21st
tions. New York: Irvington, 1983. Century. Dordrecht: Springer, in press.
16. Lindell MK, Prater CS, Perry RW. Fundamentals of 37. Mileti DS, Fitzpatrick C. The causal sequence of risk com-
Emergency Management. Emmitsburg, MD: Federal munication in the Parkfield earthquake prediction experi-
Emergency Management Agency Emergency Management ment. Risk Analysis, 1992; 12:393–400.
Institute, 2006. Available at: www.training.fema.gov/ 38. Dash N, Gladwin H. Evacuation decision making and behav-
EMIWeb/edu/fem.asp, Accessed 1 May 2010. ioral responses: Individual and household. Natural Hazards
17. Lasswell H. The structure and function of communication Review, 2007; 8:69–77.
in society. Pp. 43–71 in Bryson L (ed). Communication of 39. Terpstra T, Gutteling JM. Households’ perceived respon-
Ideas. New York: Harper, 1948. sibilities in flood risk management in the Netherlands. In-
18. McGuire WJ. The nature of attitudes and attitude change. ternational Journal of Water Resources Development, 2008;
Pp. 329–334 in Lindsey G, Aronson E (eds). Handbook 24:555–565.
of Social Psychology, 2nd edition. Reading, MA: Addison- 40. Perry RW, Lindell MK. Volcanic risk perception and adjust-
Wesley, 1969. ment in a multi-hazard environment. Journal of Volcanology
19. McGuire WJ. The nature of attitudes and attitude change. and Geothermal Research, 2008; 172:170–178.
Pp. 233–256 in Lindsey G, Aronson E (eds). Handbook of 41. Lindell MK, Whitney DJ. Correlates of seismic hazard ad-
Social Psychology, 3rd edition. New York: Random House, justment adoption. Risk Analysis, 2000; 20:13–25.
1985. 42. Mileti DS, Darlington JD. The role of searching in shaping
20. O’Keefe D. Persuasion: Theory and Research, 2nd edition. reactions to earthquake risk information. Social Problems,
Thousand Oaks, CA: Sage, 2002. 1997; 44:89–103.
The Protective Action Decision Model 631
43. Ge Y, Peacock WG, Lindell MK. Florida households’ ex- 66. Eagly AH, Chaiken S. The Psychology of Attitudes. Ft.
pected responses to hurricane hazard mitigation incentives. Worth, TX: Harcourt, Brace College Publishers, 1993.
Risk Analysis, in press. 67. Arlikatti S, Lindell MK, Prater CS. Perceived stakeholder
44. Lindell MK, Hwang SN. Households’ perceived personal role relationships and adoption of seismic hazard adjust-
risk and responses in a multi-hazard environment. Risk ments. International Journal of Mass Emergencies and Dis-
Analysis, 2008; 28:539–556. asters, 2007; 25:218–256. Available at: www.ijmed.org, Ac-
45. Weinstein ND. Effects of personal experience on self- cessed 1 May 2010.
protective behavior. Psychological Bulletin, 1989; 105:31– 68. Meyer P. Defining and measuring credibility of newspapers:
50. Developing an index. Journalism Quarterly, 1988; 65:567–
46. Palm R, Hodgson M, Blanchard, RD, Lyons D. Earthquake 574, 588.
Insurance in California. Boulder, CO: Westview Press, 1990. 69. Jackson EL. Response to earthquake hazard: The West
47. Peacock WG, Brody SD, Highfield W. Hurricane risk per- Coast of North America. Environment and Behavior, 1981;
ceptions among Florida’s single family homeowners. Land- 13:387–416.
scape and Urban Planning, 2005; 73:120–135. 70. Garcia EM. Earthquake preparedness in California: A sur-
48. Preston V, Taylor SM, Hedge DC. Adjustment to natural vey of Irvine residents. Urban Resources, 1989; 5:15–19.
and technological hazards: A study of an urban residential 71. Mulilis JP, Duval TS. The PrE model of coping with threat
community. Environment and Behavior, 1983; 15:143–164. and tornado preparedness behavior: The moderating effects
49. Lindell MK, Earle TC. How close is close enough: Public of felt responsibility. Journal of Applied Social Psychology,
perceptions of the risks of industrial facilities. Risk Analysis, 1997; 27:1750–1766.
1983; 3:245–253. 72. Lazarus RS, Folkman S. Stress, Appraisal, and Coping. New
50. Arlikatti S, Lindell MK, Prater CS, Zhang Y. Risk area ac- York: Springer, 1984.
curacy and hurricane evacuation expectations of coastal res- 73. Mileti DS. Natural hazards warning systems in the United
idents. Environment and Behavior, 2006; 38:226–247. States. Boulder, CO: University of Colorado Institute of Be-
51. Zhang Y, Prater CS, Lindell MK. Risk area accuracy and havioral Science, 1975.
evacuation from Hurricane Bret. Natural Hazards Review, 74. Perry RW. Evacuation decision making in natural disaster.
2004; 5:115–120. Mass Emergencies, 1979; 4:25–38.
52. Lindell MK, Perry RW. Household adjustment to earth- 75. Perry RW, Hirose H. Volcano Management in the United
quake hazard: A review of research. Environment and Be- States and Japan. Greenwich, CT: JAI Press, 1991.
havior, 2000; 32:590–630. 76. Blanchard-Boehm RD. Understanding public response to
53. Fishbein MF, Ajzen I. Predicting and Changing Behavior: increased risk from natural hazards. International Journal of
The Reasoned Action Approach. New York: Psychology Mass Emergencies and Disasters, 1998; 16:247–278. Avail-
Press, 2010. able at: www.ijmed.org, Accessed 1 May 2010.
54. Mulilis J-P, Duval TS. Negative threat appeals and earth- 77. Houts PS, Cleary PD, Hu TW. The Three Mile Island Cri-
quake preparedness: A person-relative-to-event PrE model sis: Psychological, Social and Economic Impacts on the Sur-
of coping with threat. Journal of Applied Social Psychology, rounding Population. University Park, PA: Pennsylvania
1995; 25:1319–1339. State University Press, 1988.
55. Kunreuther H, Ginsberg R, Miller L, Sagi P, Slovic P, 78. Perry RW. Comprehensive Emergency Management: Evac-
Borkan B, Katz N. Disaster Insurance Protection: Public uating Threatened Populations. Greenwich, CT: JAI,
Policy Lessons. New York: John Wiley, 1978. 1985.
56. Davis MS. Living along the fault line: An update on earth- 79. Fritz CE, Marks, E. The NORC studies of human be-
quake awareness and preparedness in Southern California. havior in disaster. Journal of Social Issues, 1954; 10:26–
Urban Resources, 1989; 5:8–14. 41.
57. Russell L, Goltz JD, Bourque LB. Preparedness and hazard 80. Perry RW. Environmental hazards and psychopathology.
mitigation actions before and after two earthquakes. Envi- Environmental Management, 1983; 7:543–552.
ronment and Behavior, 1995; 27:744–770. 81. Withey S. Reaction to uncertain threat. Pp. 93–123 in Baker
58. Lindell MK, Arlikatti S, Prater CS. Why people do what they G, Chapman D (eds). Man and Society in Disaster. New
do to protect against earthquake risk: Perceptions of hazard York: Basic Books, 1962.
adjustment attributes. Risk Analysis, 2009; 29:1072–1088. 82. Lindell MK, Lu JC, Prater CS. Household decision making
59. Lindell MK, Prater CS. Risk area residents’ perceptions and and evacuation in response to Hurricane Lili. Natural Haz-
adoption of seismic hazard adjustments. Journal of Applied ards Review, 2005; 6:171–179.
Social Psychology, 2002; 32:2377–2392. 83. Lindell MK, Prater CS. A hurricane evacuation manage-
60. Terpstra T, Lindell MK. Citizens’ Perceptions of Flood Haz- ment decision support system (EMDSS). Natural Hazards,
ard Adjustments: An Application of the Protective Action 2007; 40:627–634.
Decision Model. College Station, TX: Texas A&M Univer- 84. Huang SK, Lindell MK, Prater CS, Wu HC, Siebeneck LK.
sity Hazard Reduction & Recovery Center, 2009. Household Evacuation Decision Making in Response to
61. Pijawka KD, Mushkatel AH. Public opposition to the siting Hurricane Ike. College Station, TX: Texas A&M University
of the high-level nuclear waste repository: The importance Hazard Reduction & Recovery Center, 2010.
of trust. Policy Studies Review, 1991; 10:180–194. 85. Lindell MK, Mumpower J, Huang S-K, Wu H-C. Percep-
62. Lang JT, Hallman WK. Who does the public trust? The case tions and Expected Responses to a Water Contamination
of genetically modified food in the United States. Risk Anal- Emergency. College Station, TX: Texas A&M University
ysis, 2005, 25:1241–1252. Hazard Reduction & Recovery Center, 2010.
63. French JRP, Raven BH. The bases of social power. Pp. 150– 86. Lindell MK, Kang JE, Prater CS. The logistics of household
167 in Cartwright D (ed). Studies in Social Power. Ann Ar- evacuation in Hurricane Lili. Natural Hazards, in press.
bor, MI: Institute for Social Research, 1959. 87. Fu H, Wilmot CG, Zhang H, Baker EJ. Modeling the hur-
64. Raven B. Social influence and power. Pp. 371–382 in Steiner ricane evacuation response curve. Transportation Research
I, Fishbein M (eds). Current Studies in Social Psychology. Record, 2007; 2022:94–102.
New York: Holt, Rinehart & Winston, 1965. 88. Baker EJ. Hurricane evacuation in the United States. Pp.
65. Raven BH. The bases of power: Origins and recent develop- 98–108 in Pielke R Jr, Pielke R Sr (eds). Storms, Vol .1. Lon-
ments. Journal of Social Issues, 1993; 49:227–251. don: Routledge, 2000.
632 Lindell and Perry
89. Dow K, Cutter SL. Crying wolf: Repeat responses to hurri- 112. Whitney DJ, Lindell MK, Nguyen DH. Earthquake beliefs
cane evacuation orders. Coastal Management, 1998; 26:237– and adoption of seismic hazard adjustments. Risk Analysis,
252. 2004; 24:87–102.
90. Dow K, Cutter SL. Emerging hurricane evacuation issues: 113. Bostrom A, Lofstedt RE. Nanotechnology risk communica-
Hurricane Floyd and South Carolina. Natural Hazards Re- tion past and prologue. Risk Analysis, 2010; 30:1645–1662.
view, 2002; 3:12–18. 114. Scott WA, Osgood DW, Peterson C. Cognitive Structure:
91. Southern California Earthquake Center. Uniform California Theory and Measurement of Individual Differences. New
Earthquake Rupture Forecast (UCERF), 2011. Available at: York: Wiley, 1979.
www.scec.org/ucerf/, Accessed 27 January 2011. 115. Converse PE. The nature of belief systems in mass publics.
92. Hanson S, Vitek JD, Hanson PO. Natural disaster: Long- Pp. 206–261 in Apter DE (ed). Ideology and Discontent.
range impact on human response to future disaster threats. New York: Free Press, 1964.
Environment and Behavior, 1979; 11:268–284. 116. Schuman H, Kalton G. Survey methods. Pp. 635–698 in
93. Drabek TE. Social processes in disaster: Family evacuation. Lindzey G, Aronson E (eds). The Handbook of Social Psy-
Social Problems, 1969; 16:336–347. chology, 3rd edition, Vol. 1. Reading, MA: Addison-Wesley,
94. Lindell MK, Perry RW. Risk area residents’ changing per- 1985.
ceptions of volcano hazard at Mt. St. Helens. Pp. 159–166 in 117. Dooley D, Catalano R, Mishra S, Serxner S. Earthquake pre-
Siccardi F, Nigg J, Nemec J (eds). Prediction and Perception paredness: Predictors in a community survey. Journal of Ap-
of Natural Hazards. Amsterdam: Kluwer Academic Publish- plied Social Psychology, 1992; 22:451–470.
ers, 1993. 118. Seeger MW, Sellnow TL, Ulmer RR. Communication and
95. Drabek TE, Stephenson J. When disaster strikes. Journal of Organizational Crisis. Westport, CT: Praeger, 2003.
Applied Social Psychology, 1971; 1:187–203. 119. Lindell MK. EMBLEM2: An empirically based large-scale
96. Morss RE, Hayden MH. Storm surge and “certain death”: evacuation time estimate model. Transportation Research,
Interviews with Texas coastal residents following Hurricane 2008; A 42:140–154.
Ike. Weather, Climate, and Society, 2010; 2:174–189. 120. Lindell MK, Prater CS. Critical behavioral assumptions in
97. Perry RW, Lindell MK, Greene MR. Threat perception and evacuation analysis for private vehicles: Examples from hur-
public response to volcano hazard. Journal of Social Psychol- ricane research and planning. Journal of Urban Planning and
ogy, 1982; 116:199–204. Development, 2007; 133:18–29.
98. Triandis HC. Values, attitudes, and interpersonal behavior. 121. Lindell MK, Perry RW. Warning mechanisms in emer-
Pp. 195–259 in Howe H, Page M (eds). Nebraska Symposium gency response systems. International Journal of Mass
on Motivation, Vol. 27. Lincoln, NE: University of Nebraska Emergencies and Disasters, 1987; 5:137–153. Available at:
Press, 1980. www.ijmed.org, Accessed 1 May 2010.
99. Heath SE, Kass PH, Beck AM, Glickman LT. Human and 122. Gladwin H, Peacock WG. Warning and evacuation: A night
pet-related risk factors for household evacuation failure dur- for hard houses. Pp. 52–74 in Peacock WG, Morrow BH,
ing a natural disaster. American Journal of Epidemiology, Gladwin H (eds). Hurricane Andrew: Gender, Ethnicity and
2001; 153;659–667. the Sociology of Disasters. London: Routledge, 1997.
100. Van Willigen M, Edwards T, Edwards B, Hessee S. Riding 123. Dixit VV, Pande A, Radwan E, Abdel-Aty M. Understand-
out the storm: Experiences of the physically disabled during ing the impact of a recent hurricane on mobilization time
Hurricanes Bonnie, Dennis, and Floyd. Natural Hazards Re- during a subsequent hurricane. Transportation Research
view, 2002; 3:98–106. Record, 2008; 2041:49–57.
101. Killian LM. The significance of multi-group membership in 124. Fu H, Wilmot CG. Survival analysis–based dynamic travel
disaster. American Journal of Sociology, 1952; 57:309–314. demand models for hurricane evacuation. Transportation
102. Drabek TE, Boggs K. Families in disaster: Reactions and rel- Research Record, 2006; 1964:211–218.
atives. Journal of Marriage and the Family, 1968; 30:443–451. 125. Wilmot CG, Mei B. Comparison of alternative trip genera-
103. Weinstein ND, Rothman AJ, Sutton SR. Stage theories tion models for hurricane evacuation. Natural Hazards Re-
of health behavior: Conceptual and methodological issues. view, 2004; 5:170–178.
Health Psychology, 1998; 17:290–299. 126. Siebeneck LK, Cova TJ. An assessment of the return-entry
104. Prochaska JO, Johnson S, Lee P. The transtheoretical model process for Hurricane Rita 2005. International Journal of
of behavior change. Pp. 59–84 in Gochman DS (ed). Hand- Mass Emergencies and Disasters, 2008; 26:91–111. Available
book of Health Behavior Research I: Personal and Social at: www.ijmed.org, Accessed 1 May 2010.
Determinants. New York: Plenum Press, 1997. 127. Norris FH, Smith T, Kaniasty K. Revisiting the experience-
105. Weinstein ND. The precaution adoption process. Health behavior hypothesis: The effects of Hurricane Hugo on haz-
Psychology, 1988; 7:355–386. ard preparedness and other self-protective acts. Basic and
106. Birkland TA. After Disaster: Agenda Setting, Public Policy Applied Social Psychology, 1999; 21:37–47.
and Focusing Events. Washington, DC: Georgetown Uni- 128. Petty RE, Cacioppo JT. Communication and Persua-
versity Press, 1997. sion: Central and Peripheral Routes to Attitude Change.
107. Petty RE, Wegener DT. The elaboration likelihood model: New York: Springer-Verlag, 1986.
Current status and controversies. Pp. 37–72 in Chaiken, 129. Chaiken S. The heuristic model of persuasion. Pp. 3–39 in
Trope Y (eds). Dual-Process Theories in Social Psychology. Zanna MP, Olson JM, Herman CP (eds). Social Influence:
New York: Guilford Press, 1999. The Ontario Symposium, Vol. 5. Hillsdale NJ: Erlbaum,
108. Floyd DL, Prentice-Dunn S, Rogers RW. A meta-analysis 1987.
of research on protection motivation theory. Journal of Ap- 130. Knabb RD, Rhome JR, Brown DP. Tropical Cyclone
plied Social Psychology, 2000; 30:407–429. Report: Hurricane Katrina 23–30 August 2005. Mimi, FL:
109. Neuwirth K, Dunwoody S, Griffin RJ. Protection motivation National Hurricane Center, 2006. Available at: http://www.
and risk communication. Risk Analysis, 2000; 20:721–734. nhc.noaa.govpdf/TCR-AL122005 Katrina.pdf, Accessed 25
110. Rogers RW. A protection motivation theory of fear appeals January 2011.
and attitude change. Journal of Psychology, 1975; 91:93–114. 131. Burton L, Hicks MJ. Hurricane Katrina: Preliminary Esti-
111. Lindell MK. Perceived characteristics of environmental haz- mates of Commercial and Public Sector Damages. Hunting-
ards. International Journal of Mass Emergencies and Dis- ton, WV: Marshall University for Economic and Business
asters, 1994; 12:303–326. Available at: www.ijmed.org, Ac- Research, 2005. Available at: http://www.marshall.edu/cber/,
cessed 1 May 2010. Accessed 1 January 2011.