Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
1!
The Protective Action Decision Model: !Theoretical, Methodological, and !Practical Issues!
Michael K. Lindell Texas A&M University Hazard Reduction & Recovery Center University of Washington Department of Urban Design and Planning
Hierarchy of PADM Variables
2!
!Cognitive processes!
Social/environmental processes!
Social psychological processes!
3!
Basic Elementsof the Cognitive Component!
Exposure! Attention! Comprehension!
Mental model!
4!
Basic Elements of the Social Psychological Component!
Source! Channel! Message! Receiver! Effect!
Feedback!
5!
Basic Elements of the Social Component!
Ultimate receiver 5!
Ultimate receiver 4!
Ultimate receiver 2!
Ultimate receiver 3!
Ultimate receiver 1!
Original source!
Intermediate 1!
Intermediate 2!
No information!
Information isolate!
High information
flow!
Basic Elements of the Environmental Component
6!
Physical environment!• Geophysical systems!• Meteorological systems!• Hydrological systems!• Technological systems!
Hazard adjustments• Hazard mitigation!• Disaster preparedness!• Emergency response!• Disaster recovery!
Societal stakeholders!• Households/communities!• Businesses/industries!• Local/state/federal government!
Efficacy! Risk!
Resource requirements!
7!
Protective Action Decision Model
Behavioral response!• Information search!• Protective response!• Emotion-focused coping!
Situational facilitators!
Situational demands/ constraints!
Receiver characteristics!
Social! cues!
Message!content!
Channel access and preference!
Source characteristics!
Environmental! cues!
Pre-decision processes!• Exposure!• Attention!• Comprehension!
Threat perception!Protective
action perception!
Stakeholder perception!
8!
Illustrative Data
l Longview (1985) data: Hazard perceptions – Within the 10 mile EPZ of a nuclear power, – Adjacent to a major interstate highway, rail line, and shipping
channel for hazardous materials, and – Forty miles west of the Mt. St. Helens.
l Six city (1997) Los Angeles and Seattle area data: Earthquake mitigation and preparedness.
l Houston (2003) data: Hurricane mitigation and preparedness.
l Hurricane Lili (2002), Katrina/Rita (2005), and Ike (2008) data: Evacuation decision making
l American Samoa (2009) earthquake and tsunami l Christchurch and Tohoku (2011) earthquakes
9!
Public Perceptions of Environmental Hazards
l Disaster research identifies the following dimensions of people’s mental models of hazards as determinants of their warning responses – Hazard agent characteristics – Impact characteristics – Personal consequences – Affective and behavioral reactions – Alternative protective actions
10!
Longview: Hazard Agent Characteristics
Minimal difference among hazards!
Volcano different from chemical and
nuclear hazards!
11!
Longview: Hazard Impact Characteristics
1
2
3
4
5
Speed of onset Environmental cues Scope of impact Duration of impact
MSH eruption
Chlorine tank car
Nuclear power plant
Volcano different from chemical and
nuclear hazards!
Nuclear different from volcano and chemical hazards!
12!
1
2
3
4
5
Immediate death Delayed cancer Genetic effects Total property loss
MSH eruption
Chlorine tank car
Nuclear power plant
Chemical different from volcano and nuclear hazards!
Longview: Expected Personal Consequences
Large differences among all three hazards!
13!
1
2
3
4
5
Dread Think about frequently Talk about frequently
MSH eruption
Chlorine tank car
Nuclear power plant
Minimal differences among hazards!
Longview: Affective and Behavioral Reactions
14!
Longview: Perceptions of Stakeholder Expertise on Multiple Hazards
1.00
2.00
3.00
4.00
5.00
Self Peers Newsmedia Local govt. State/fed govt.
MSH eruption
Chlorine tank car
Nuclear power plant
Government and news media are higher in expertise for the unfamiliar hazards, but the differences among stakeholders are small for the familiar hazard.!
Mea
n ra
ting!
Stakeholder!
Six City: Perceptions of SeismicStakeholders on Multiple Attributes!
15!
1.00
2.00
3.00
4.00
5.00
Mean
Rat
ing
Stakeholder
Hazard KnowledgeTrustworthiness Responsibility
Differences among stakeholders in expertise are small for this (familiar) hazard. The notable differences are in protection responsibility.!
Stakeholder!
Mea
n ra
ting!
16!
Public Perceptions of Protective Actions
l Some studies have collected free response data showing that some people are – unaware of any protective action for a given hazard – believe all recommended actions are ineffective, or – believe all recommended actions require excessive
resources.
l These findings underscore the importance of assessing people’s perceptions of different protective actions.
17!
Michigan State University: Perceptions of Toxic Chemical Protective Actions!
1
2
3
4
5
Efficacy Time Effort Obstacles Skill Money
Evacuation
Sheltering in-place
Expedient respiratory protection
Evacuation is much higher in efficacy but also much higher in resource demands!
Profiles for Seismic Hazard Adjustments With the Highest Adoption Intentions !
18!
The most popular adjustments have multiple uses and low resource requirements!
Profiles for Seismic Hazard Adjustments With the Lowest Adoption Intentions !
19!
The least popular adjustments have low efficacy and high resource requirements!
20!
The Social and Environmental Context of Hazard Adjustment Adoption
l Two studies modeled the adoption of hazard adjustments—one for earthquakes (Six city study) and the other for hurricanes (Houston study).
The Social/Environmental Context of Seismic Hazard Adjustment Adoption!
21!
Hazard intrusiveness
Hazard experience
Risk perception Hazard adjustment
Gender Age
Ethnicity
Marital status Income
Home ownership
The Social/Environmental Context of Hurricane Hazard Adjustment Adoption!
22!
Hazard proximity
Past tenure
Hazard experience
Tenure expectations
Perceived personal risk
Hazard adjustment
Gender Hazard information Ethnicity Income
Solid lines indicate complete support and dotted lines indicate partial support for hypothesized paths.!
23!
Sources of Information in Rapid Onset Disasters
l The American Samoa earthquake and tsunami provides an example of the prevalence of environmental cues, social cues, and social warnings in a rapid onset disaster.
0 5 10 15 20 25 30 35 40 45 50
Personal knowledge
Broadcast warning
Bell ringing
People evacuating
Personal warning
Phone warning
Ocean behavior
Tsunami wave
Animal behavior
24!
First Source of Information About the Tsunami
Warning Message Content (N = 72)
25!
0% 20% 40% 60% 80% 100%
Other Continued driving Stood in doorway
Stayed outside Dropped but did not cover
Continued what I was doing Pulled the car over
Tried to protect property Tried to protect people
Immediately left the building Dropped, covered and held
Stopped activity but stayed put
0% 20% 40% 60% 80% 100%
Other Continued driving Stood in doorway
Stayed outside Dropped but did not cover
Continued what I was doing Pulled the car over
Tried to protect property Tried to protect people
Immediately left the building Dropped, covered and held
Stopped activity but stayed put
0% 20% 40% 60% 80% 100%
Other Continued driving Stood in doorway
Stayed outside Dropped but did not cover
Continued what I was doing Pulled the car over
Tried to protect property Tried to protect people
Immediately left the building Dropped, covered and held
Stopped activity but stayed put
0! 10! 20! 30! 40! 50!
Safe areas!
Tsunami threat!
Protective action recommedation!
Affected areas!
Sources of assistance!
Sources of additional info!
Percent of respondents receiving a warning!
Further Information After the Shaking Stopped
26!
0% 20% 40% 60% 80% 100%
Other Continued driving Stood in doorway
Stayed outside Dropped but did not cover
Continued what I was doing Pulled the car over
Tried to protect property Tried to protect people
Immediately left the building Dropped, covered and held
Stopped activity but stayed put
0% 20% 40% 60% 80% 100%
Other Continued driving Stood in doorway
Stayed outside Dropped but did not cover
Continued what I was doing Pulled the car over
Tried to protect property Tried to protect people
Immediately left the building Dropped, covered and held
Stopped activity but stayed put
0% 20% 40% 60% 80% 100%
Other Continued driving Stood in doorway
Stayed outside Dropped but did not cover
Continued what I was doing Pulled the car over
Tried to protect property Tried to protect people
Immediately left the building Dropped, covered and held
Stopped activity but stayed put
0 10 20 30 40 50 60
Saw tsunami
Saw evacuations
Face-to-face info
Phone info
Radio info
TV info
TM info
Email info
Percent of respondents!
27!
Emotional and Behavioral Response to Rapid Onset Disasters
l The Christchurch and Tohoku earthquakes provide examples of the role of emotional response in a rapid onset disaster.
Emotional Response to Seismic Shaking
28!
1 2 3 4 5
Energetic
Relaxed
Optimistic
Passive
Annoyed
Depressed
Fearful
Nervous
Alert
Christchurch
Hitachi
Not at all Very great extent Extent
Shock!
Vigilance!
Fear!
Initial Behavioral Response to Seismic Shaking
29!
0% 10% 20% 30% 40%
Froze
Evacuated immediately
Took cover
Tried to protect people
Tried to protect property
Pulled the car over
Continued activity
Other
Christchurch
Hitachi
Percent
Emotion Profiles For Behavioral Responses to Seismic Shaking
30!
1
2
3
4
5
Continue Freeze ProPers Cover ProProp ImmEvac
Vigilance
Shock
Fear
31!
The Social/Environmental Context of Disaster Response
l Recent studies have begun to test multi-stage causal models of household hurricane evacuation.
Household Evacuation From Hurricane Ike: Hypothesized Model
32!
H4!
H3!
H1!
H2!
H9!H8!
H5!
Perceived evacuation impediments!
H6! H7!Perceived storm characteristics!
Expected personal impacts!
Evacuation decision and
timing!
“”Unnecessary” evacuation experience!
Environmental and social cues!
Official !warning!
Coastal !proximity!
Hurricane experience!
Female !Gender!
Household Evacuation From Hurricane Ike: Final Model
33!
Perceived evacuation impediments!
Perceived storm characteristics!
Expected personal impacts! Evacuation
decision!
“Unnecessary” evacuation experience!
Environmental and social cues!
Official !warning!
Coastal !proximity!
Hurricane experience!
Education!
34!
Some Criticisms of the PADM
l Some researchers and practitioners think the PADM is too rational an account of individual decision making in emergencies!– In fact, people are generally rational (although not always
adaptive) in emergencies; reports of irrational behavior such as panic and incapacitating shock are greatly exaggerated.!
– People can move through information processing stages quickly, especially when disaster impact is imminent.!
– Moreover, they might skip some stages and go directly to protective action (e.g., compliance with an authority).!
– In addition, they might return to earlier stages of the decision process—especially when the situation is uncertain and there seems to be enough time before impact.!
35!
Future Directions for PADM Research
l Examine the relationship between risk perception and hazard intrusiveness. – Risk perception involves the subjective probability of
personal consequences occurring within a given time interval.
– Hazard intrusiveness involves the frequency of thought and discussion about a hazard.
– They are related, but it is unclear which is is more important in changing behavior—emphasizing the likelihood of personal consequences or reminding people frequently about those likelihoods.
36!
Future Directions for PADM Research
l Examine the relationship between problem-focused and emotion-focused coping over time and between behavioral domains. – Problem-focused coping reduces hazard vulnerability but
only indirectly reduces fear—especially if hazard adjustments are not perceived to be 100% effective.
– Emotion-focused coping directly reduces fear through denial, distraction, or self-medication but generally does little to reduce hazard vulnerability.
– Fatalism, hopelessness, and denial can be interpreted as arising from lack of knowledge of protective actions that are effective and feasible.
37!
Future Directions for PADM Research
l Examine the impacts of task demands on the adoption of hazard adjustments – Frequency of repetition
• Habits: Keeping the gas tank full during hurricane season • Periodic behaviors: Renewing hazard insurance annually • Unique acts: Bolting a house to its foundation
l Develop better measures of the effects of “experience”. – This variable yields conflicting results because it is quite
complex (e.g., personal vs. vicarious) and has been measured in so many different ways.
38!
Future Directions for PADM Research
l Examine the group decision processes that take place within households. – We usually collect data from one respondent but many times
household members collectively negotiate investments in hazard adjustments, especially if they are resource intensive.
39!
Future Directions for PADM Research
l Develop better assessments of the “microscopic” cognitive processes involved in attention, comprehension, judgment, and decision making. – These can take place very rapidly (milliseconds to minutes)
and might not be susceptible to accurate recall in surveys conducted weeks or months after a major emergency.
– Laboratory experiments provide rapid measurement but lack the population diversity and situational characteristics of true emergencies.
– Web experiments using DynaSearch and Stormview provide an opportunity to examine the effects of different warning messages on protective action decision processes in diverse population samples.
40!
Future Directions for PADM Research
l Conduct multi-level analyses (within a single study) or statistical meta-analyses that examine the effects of contextual variables at disasters varying in – event characteristics (e.g., scope of impact, time of day) – geography (i.e., hills, valleys, rivers). – culture (including typical social, physical, and household
contexts), – household hazard education and emergency preparedness, – community warning technology (e.g., electronic sirens), and – disaster-relevant resources (e.g., access to cars and hazard
resilient structures such as steel reinforced parking structures for tsunami evacuation),
Lessons from the PADM for Risk Communication
l Perceptions of the risk of an event have less impact on behavior than expectations of personal consequences. – Perceived earthquake probability has a lower correlation with
long-term hazard adjustment than does the perceived probability of personal consequences from an earthquake.
– Perceived hurricane characteristics have lower correlations with evacuation than do expected personal consequences.
l Lesson: Don’t talk to people about the probability that an event will happen; talk to them about the probability of that event’s consequences for them.
41!
Lessons from the PADM for Risk Communication
l Risk perception arouses protection motivation but does not provide a means of resolving that motivation.
l Lesson: Don’t talk to people only about the probability of personal consequence; also talk to them about – Protective actions being taken by authorities, – Alternative protective actions they should consider, – The characteristics of those protective actions, and – Sources of assistance if this is needed to implement
protective action.
42!
Lessons from the PADM for Risk Communication
l Long-term hazard adjustments vary in their hazard-related and resource-related attributes.
l Lesson: Help people to identify the the most cost-effective hazard adjustments – Encourage them to develop a plan for adopting hazard
adjustments sequentially as money, time, and effort allow.
43!
I want to thank my colleagues (Ron Perry, Carla Prater, Marjorie Greene, Walt Peacock) and former and current students (Dave Whitney, Yang Zhang, Sudha Arlikatti, Jung Eun Kang, Seong Nam Hwang, Teun Terpstra, Yue Ge, Hao Che Wu, Shih-Kai Huang, Laura Siebeneck, Hung-Lung Wei) for their contributions to the development and testing of the model. I am also grateful to Bill Anderson, Dennis Wenger, and Bob O’Connor for support from the National Science Foundation under Grants CMS 9633595, CMS 0219155, SBE 0527699, SBE 0838654, CMM 1138612, and IIS 1212790. None of the conclusions expressed here necessarily reflects views other than those of the author.
44!
Acknowledgements