To Cross the Bridge or NOT

and the Perception of Risk

Michael J Kozuch, PhD2015

We are constantly making decisions

based on our capacity

to handle different levels of risk.

Our understanding of risk

is based on past experience

and our notion

of potential outcomes.

But, helping individuals chose

their level of acceptable risk for

any given hazard is critical.

Quantifyingthese levels of risk

“perception” can be

challenging.

Insurance companies quantify risk using

economic terms like

risk of ruin, damage or loss estimates.

However, not all risk is economically

or easily quantifiable.

How do you place a value on

your own life or

the loss of a loved one?

Would you cross or not cross a bridge

if there was an imminent threat to the bridge?

Would you even begin a journey across a bridge

if you could

not see where it went?

To explore the perception of risk in the “language ofprobabilities”, we posed a series of questions to students who were learning about hazards.

The student population was culturallyand ethnically diverse…

The Question we posed:

Given the probability X of

an earthquake of magnitude Y,

would you drive across a bridge?

(If not, the alternative is to drive 1 hour

out of your way).

For example, if there was a

25% chance of a M 4.0 earthquake, would you cross the bridge?

Magnitudes. No assumptions were made about student familiarity with Magnitudes and the associated damage. Theyrelied on what they heardin the media.

Probabilities. The students only knew that 30% probability meant that one third of the time there is a quake when the “conditions” were similar.

Or, what if there was a

25% chance of a M 7.0 earthquake, would you cross the bridge?

Or, what if there was a

75% chance of a M 3.0 earthquake, would you cross the bridge?

These questions were meant to provide a baseline to see what they understood about these hazards from the media, academic and government sources to help us see if improvements could be made in the messaging.

P ( X | Y )X = probability of occurrence

Y = earthquake magnitude

Several probability levels for

earthquakes of different magnitudes

were asked.

ResultsThe Results are interesting...

Major Outcomes:

The greater the magnitude the less likely they are to cross the bridge.

The greater the probability the less likely they are to cross the bridge.

Most students understood probabilities,but some misunderstood the terms altogether.

3

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

M3 M4 M5 M6 M7

25%

50%

75%

Results show the probability of earthquake

occurrence needed to invoke a response.

Earthquake Magnitude

Perc

ent

of

Ind

ivid

ual

s C

ross

ing

Bri

dge Probability of

Earthquake Occurrence

99 students

Consistent with literature:

Others* have shown similar risk distributionsfor other hazards and situations.

*See Jonkman, vanGelder and Vrijling, 2003, An Overview of of quantitative riskmeasures for loss of life and economic damage, J Haz Mat A99 (1-30).

Severity of hazard or accident

Probabilityof exceedance

Unacceptablerisk

Acceptablerisk

We can dissect the data further

to see gender differences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

M3 M4 M5 M6 M7

25%-female

25%-male

50%-female

50%-male

75%-female

75%-male

Earthquake Magnitude

Perc

ent

of

Ind

ivid

ual

s C

ross

ing

the

Bri

dge

Perceived Risk by Gender

25%

Probability of Earthquake Occurrence

75%

50%

68 females31 males

Our population showed gender-specific risk sensitivity.

Males took progressively more risk than females as the probability of an event increased.

One could argue for a biological link to testosterone levels in regulating the hypothalamus-pituitary-adrenal levels in flight-fight response.*

*Mehta, Jones, and Josephs, 2008, J Pers Soc Physhol 94(6): 1078-93.

Landslide

Risk

Photo by Eekster (Wikipedia)

We looked at another hazard:

We tackled a new situation:

New hazard: Landslides

Added a new factor: Responsibility

We posed 2 questions to see howmuch a response is affected byresponsibility.

The Question.

At what probability would you NOT cross the bridge?

0 10 20 30 40 50 60 70 80 90 100%

Landslidewill not happen

Landslidewill definitely

happen

Probably will

happen

Probably will nothappen

If they answer 40%:

At what probability would you NOT cross the bridge?

0 10 20 30 40 50 60 70 80 90 100%

Landslidewill not happen

Landslidewill definitely

happen

Probably will

happen

Probably will nothappen

acceptable risk unacceptable risk

The Question.

At what probability would you NOT cross the bridge ifyou had a child in the car?0 10 20 30 40 50 60 70 80 90 100%

Landslidewill not happen

Landslidewill definitely

happen

Probably will

happen

Probably will nothappen

Although the hazard remained the same, the perception of risk increased asthe subjects took on responsibility(such as travelling with a child).

For high probabilities of landslide occurrence:Significantly fewer students would cross the bridge when given more responsibility.

Results

Results show the probability of landslide

occurrence needed to invoke a response.

Simple questionnaires can be given to user

groups to understand their perception of risk

for any situation.

This has important implications for effective

communication and messaging

in hazard preparedness.

The Take-Away

Custom Messages. Messages can be tailored

to the level of hazard and perceived risk.

Probabilities. Messaging with probabilities is

perhaps best explained in simpler terms

(e.g. high, medium or low).

Designed by Stakeholders. The most effective

messages and graphics can be designed by

stakeholders to ensure clarity.

The Take-Away

A tsunami warning where the message is simple and visual.

But it doesn’t cover tsunami that are createdremotely.

A reminder that hazardscan change over time.

What does this say about our messaging?

Is this enough?

X

A Note about this study.

The data and plots are meant to show a trend for the given population of students and are not meant to be representative of a larger population.

Factors such as age, religion, educational level, and other indicators may contribute to subjective biases in how individuals respond in the presence of a hazard.

This preliminary study shows that there are quantitative ways to measure qualitative information.

As in any study, care must be given in presenting a statistical problem to individuals who are not familiar with probabilities and this topic will be explored further.

- Michael Kozuch( 2015)