1. Human Response to Emergency Warnings
John Sorensen
Barbara Vogt Sorensen
AGU Fall Meetings
December, 2009

2. Definition of a Warning System
Communications system that maximizes the probability that people at risk take appropriate protective actions (PA) that minimize their exposure to damages
Warnings allowed residents to escape to a tornado shelter
Source: National Weather Service

3. Science Based Behavioral Studies Have Been Conducted on Warning Response In USA For Over 50 Years
Hurricanes:
Elena, Kate, Eloise, Camille, David, Frederick, Carla, Floyd, Andrew, Bertha, Fran, Georges, Brett, Bonnie, Ivan
Hazardous Materials:
Mississauga, Mt. Vernon, Denver, Confluence, Pittsburg, Nanticoke, West Helena, Graniteville
Nuclear Power:
Three Mile Island
Floods:
Denver, Rapid City, Big Thompson, Abilene, Clarksburg, Rochester, Sumner, Valley, Snoqualmie, Denver, Austin
Volcano:
Mt. St. Helen
Wildfire:
San Diego
Tsunami:
Hilo

4. Factors Impacting Warning Response
The message
Cues
Individual characteristics
Roles
Experience
Belief
Knowledge
Perceived risk
Social interaction

5. Factor 1: The Message Five Dimensions Channel Frequency Content Style
Source

6. Message Channel and Frequency
Number of Communication Channels
More is better
Type of Communication Channels
Personal channels work best
More diverse the better
Communication Frequency
The more the message is repeated & heard the better
Repetition fosters confirmation
Confirmation fosters belief
Belief fosters taking action

7. Message Content A description of the hazard and impacts
The current and predicted location of the impacts (and location of no impacts)
What specific actions to take (and not to)
When to take those actions
Sources of information

8. Message Style Clear: Worded in a manner that people understand
Specific: Contains precise information
Accurate: Even minor errors may cause problems
Certain: Is authoritative and confident
Consistent: Explains basis for change in information

9. Factor 2: Cues (Non-verbal Information)
Social Cues Help
- What neighbors, friends, & relatives are doing
- What organizations are doing
Physical Cues Help
Can confirm the risk (rain in flood warnings)
Non-verbal warning (fire)

10. Factor 3: Individual Characteristics (May Constrain Response)
Socio-economic Status
Little money, low education
Age
Being young or old
Gender
Being male
Ethnicity
Being non-Anglo
Acculturation
Not speaking English,
Born in another country

11. Factor 4: Roles (Incentives to Response)
Roles or Responsibility for Others:
Having children
Larger family size
Greater community involvement
More kin relationships
Family united

12. Factor 5: Experience
People interpret warning information based on their experience
Think the disaster they face will be like those experienced
Inclined to do what was appropriate for them as experienced in past events

13. Factor 6: Belief There is no single credible spokesperson
People have different ideas about who’s credible
Who’s credible is not that important
Many think spokesperson credibility will lead to message belief
How to achieve warning belief
One consistent message over diverse channels
From a panel of spokespersons
e.g., officials, Red Cross, scientists, familiar newscaster, & others
Repeated multiple times

14. Factor 7: Knowledge Multi-faceted concept
Past: what people have learned about the hazard
Present: what people think based on the information/cues they get during the event
Natural inclination: it is not going to effect me
Not static and can change rapidly
Manage it in warning messages
Provide warning information that overcomes differences in people’s past, present, & natural inclinations

15. Factor 8: Perceived Risk
It’s about perceived risk during the event
Different from pre-event threat perception
Denial is the first common perception
Usually a constraint to taking action
People don’t perceive they’re at risk
People perceive they’re safe
And search for information to confirm that they are
Moreover
People dichotomize risk into doing something vs. doing nothing
They don’t act in proportion to scientific probability estimates

16. Factor 9: Social Interaction
People talk about it with others
People confirm the warning, the risk, and what they can do about it
Few do something because they’re told to
People need to think it’s their own idea
People think it is there own best interest
People believe it will protect them

17. Key Findings: Warning Dissemination
In most emergencies 20 to 50 percent of first warnings are received from non-official sources (friends & relatives)
Diffusion characterized by logistic curse
Steepness influenced by urgency
Adequate data on how traditional technologies perform
Little data on how emerging technologies perform

18. San Diego Reverse 911
October, 2007, wildfires in San Diego City and County
County extensively used reverse 911 for evacuation notifications
Opportunity to asses the effectiveness of phone messages
Did not evaluate call completion rate
Conducted survey of 1200 households in evacuation zones

20. Methods of Warning Reverse 911 Route alert Media IP system
Weather radio
Hotlines (211)
Road signs

21. Reverse 911 Use Targeted to specific areas/subdivisions
County: 233,950 evacuation calls
Activated 36 times
City: 14,738 evacuation calls
Short 15 second message
Some mandatory, some advisories
Also supported reentry and health warnings after evacuations

22. Source of First Warning
Reverse 911 – 42.1%
Television – 7.7%
Route alert – 4.1%
Informal face-to-face – 3.8 %
Informal telephone call – 2.4%
Tone alert radio – 0.3%
IP device – 0.2 %
Radio – 0.2%

23. Factors Determining Evacuation
Receipt of reverse 911 call
Perception that residence was in an area threatened by wildfires
Adoption of mitigation measures
Receipt of informal warning
Lived in a rural area
Saw smoke or fire (environmental cues)

24. Conclusions Robust understanding of warning response
New technologies offer rapid forms of warning
Reverse 911 was effective in getting people to evacuate
Personal message
From authoritative source
People confirmed message