Ris_Kim
Mr.Garcia
RiskPerception.pptxRisk Perception
An insightful comment
Reports that say that something hasn't happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns – the ones we don't know we don't know.
Donald Rumsfeld, February 12, 2002
Should you worry about any of these? Which ones?
Lightning
Tornadoes
Hurricanes
Heat
Blizzards
Some puzzles
Why do people fear genetically modified tomatoes but smoke cigarettes?
Why would someone pay more than twice as much for a product when the risk of minor injury associated with it is reduced from 15/10,000 to 10/10,000
(.0015 to .001) ?
Thinking about Risk
What risks do we face in everyday life?
How do people perceive risk?
What level of risk is acceptable?
How should we talk about risks?
Tradeoffs of costs and benefits
Value of human life
Bayes’ Theorem
Risk and Uncertainty
Risk: the distribution of possible outcomes is known
Uncertainty: the distribution of possible outcomes is unknown
Most people think of them interchangeably
How do experts talk about risk?
What is the hazard?
What happens when people are exposed to the hazard?
How many people are exposed to the hazard?
Tradeoffs between costs and benefits
Can you calculate whether a risk is worth taking?
How do insurance companies do this?
What about you?
Ebola in the US
Huge reaction in US media
A nurse returning from West Africa was forcibly quarantined after testing negative for Ebola.
Travel bans were urged by some politicians.
One college decided not to admit any African students.
2 people in the US died from the disease. Neither contracted it in the US.
Influenza in the US
Kills about 18,000 people annually
Annual vaccination recommended by CDC
Only about 40% of adults get the vaccine
Often available free
Why don’t people worry about the flu? Why do they worry about Ebola?
3 main factors in risk perception (Paul Slovic)
Dread
The unknown
Number of people affected
Plus availability heuristic
Dread risk
Dread
Perceived lack of control
Catastrophic potential
Fatal consequences
Unequal distribution of risks
The unknown
Unknown
Unobservable
New
Delayed harm
Number of people affected
Seems obvious
But we know that people are not good at estimating frequency
Other factors in risk perception
Trust
Source of hazard
Control
Identity of victim – age, celebrity status, relationship
Why don’t people worry about these?
Smoking
Texting while driving
Unsafe sex
How should we talk about risk?
Lifetime exposure to a certain pollutant is expected to kill 1.4 of 1,000 people.
Lifetime risk is 0.0014
Lifetime risk is 0.14%
Lifetime risk is 1 in 710
Out of 1,000 people, we could expect 1.4 more people to die.
Would concrete examples help?
Smoking kills about 480,000 people in the US annually.
Leading preventable cause of death
That’s more than 1,300 a day.
3 jumbo jet crashes per day, every day
The equivalent of the Titanic every day
16 times as many as are killed in car accidents
Innumeracy
It’s hard to understand the difference between 1/10,000 and 1/10,000,000
Lifetime risk of being struck by lightning?
1/12,000
1/9,000,000 chance of being struck twice
Chance of winning PowerBall?
1/292,000,000
For most of humanity’s existence, all we needed was to count
Most of what we needed to count was tangible.
Abstraction doesn’t come naturally
Germ theory of disease came from availability of microscopes
Doctors didn’t wash hands between surgeries until 1847
Technology enables us to detect risk
But few of us really understand its magnitude
Certainty has special status
Prospect theory
Difference between 0% and 5% perceived as much larger than difference between 25% and 30%
Same with 95% and 100%
Risk, Costs, and Benefits
Why are some risks ignored while others feared?
How much value do people place on reduced risk?
Willingness to Pay vs Willingness to Accept
WTP – WTA disparities
Norovirus is a highly contagious virus that causes nausea, abdominal pain, muscle aches and digestive problems. Most people recover within a week or so.
How much would you pay to reduce your risk of being infected with norovirus by 10%?
How much would you accept to increase your risk of being infected with norovirus by 10%?
WTP usually less than WTA for most situations
In some cases, WTA is infinite
No price is high enough to accept increased risk.
Why?
Value of a Human Life
http://www.lifehappens.org/insurance-calculators/calculate-human-life-value/
When do we place a value on human life?
How do we do it?
Tough questions: Medical care
The last year of life accounts for a large proportion of medical cost.
Humans are mortal.
Physicians are trained to keep people alive at any cost.
Most people say they want to die at home, with family, without tubes.
Most people die in a hospital, with tubes.
Suggestion to encourage end-of-life care discussions denounced as “death panels”
26
Oregon Health Plan
Establishes priorities for Medicaid spending
Clinical effectiveness and cost-effectiveness
Recognizes limited resources
Evidence based
Highest priority for preventive treatment and chronic diseases
Maternity and newborn care
Diabetes
How can you compute risks?
A doctor detects a lump in a patient’s breast and estimates a 1% chance that it may be cancerous, based on many years of experience. To be on the safe side, a mammogram is performed. Mammograms are 80% accurate in detecting cancerous tumors and 90% accurate in detecting benign tumors. The mammogram is positive.
What is the chance that the patient has cancer?
Confusion of the Inverse
Probability of cancer given a positive test ???
Probability of a positive test given cancer
Bayes’ Theorem
A way to incorporate base rates and evidence
P (Hypothesis True|Data) =
P(Data| H True) * P(H True)
[P(Data|H True) * P(H True)] +[ P( Data|H False) * P(H False)]
Hypothesis True = Patient has cancer
Hypothesis False = Patient doesn’t have cancer
Data| H True = positive test when patient has cancer
Data| H False = positive test when patient doesn’t have cancer
The numbers for this example
Hypothesis True = Patient has cancer = .01
Hypothesis False = Patient doesn’t have cancer = .99
Data| H True = positive test when patient has cancer = .8
Data| H False = positive test when patient doesn’t have cancer = .1
Bayes’ Theorem, applied
P(disease| positive test) =
P(disease) * P(positive| disease)
P(disease) * P(positive| disease) + P(no disease) * P(positive| no disease)
= (.01 * .8)/(.01*.8) + (.99 *.10) = .008/.107 = .075 = 7.5%
A positive test result doesn’t always spell doom!
How can you deal with risk?
Understand the base rates
Remember that what you want to be true isn’t necessarily the case
Don’t get sidetracked by availability
