reflections on the need for seeing beyond risk assessments and
TRANSCRIPT
Reflections on the need for seeing beyond risk assessments and
decision analysis tools indecision analysis tools in decision-making situations
involving high risks
Terje Aven
University of Stavanger, NNorway Conference on Nuclear
Risk and Public Decision MakingDecision-Making 14-16th November 2011Paris
What is Risk?
C,UC,U
C: the consequences of the activityU: uncertainty (what will C be?) y ( )
What is Risk?
A C UA,C,U
A: initiating event C: the consequences of the activityyU: uncertainty (what will C be?)
Where is the P b bilit ?Probability ?
Risk
General How to describe orconcept describe or measure risk
U … P …
Risk description p
Specific events, consequences,
ProbabilitiesProbabilitiesfrequencies, …
(C Q )(C’, Q, K)
C’ : Specific consequences Q: Measure of uncertainty (often P)Q: Measure of uncertainty (often P)K: Knowlegde that Q is based on
Risk Risk description
Threats,
p
,hazards,
ConsequencesC
Specific events, consequences,
ProbabilitiesC Probabilitiesfrequencies, …
Uncertainty
(C U) (C Q )(C,U) (C’, Q, K)
(A C U)C’ : Specific consequences Q: Measure of uncertainty (often P)
(A,C,U)
Q: Measure of uncertainty (often P)K: Knowlegde that Q is based on
Formula Optimal/rightFormula Optimal/right decision
Risk description What is acceptableRisk description What is acceptable risk
• ------ P =1 x 10-4
Risk-based decision-making
Expected utility theory
A BB
Eu(X) = 0.3 Eu(X) = 0.5
Choose alternative B
Analysis ManagementManagement
Risk analysisRisk acceptance criteria
Cost-benefit analysisDecision analysis
Managementreview and judgment
Decision
Limitations
Uncertainties
I f iRisk Assessment
P Decision-making
Other
Informing
Other concerns
Misconception (risk-based decision-making)
Formula Optimal/rightFormula Optimal/right decision
Misconception (decision analysis)??
Formula Optimal/rightFormula Optimal/right decision
Cost-benefit analysis
n
taNPV
tt
iNPV
1 t i0 1Expected net present value = E[NPV]
Uncertainties and risk not addressedUncertainties and risk not addressed beyond expected values
x
Law of large numbers
• The average converges to the gexpected value
• (X1 + X2 + … + Xn)/n EX1(X1 X2 … Xn)/n EX1
Analysis Management
Ri k l i
Management
Risk analysisRisk acceptance criteria
Cost-benefit analysisDecision analysis
Managementreview and judgment
Decision
Expected utility theory
• E[u(X)]
Decision problem
Consequences Preferences
Uncertainties Values
Decision problem
Consequences Preferences
Uncertainties Values
U t i ti ( )Uncertainties Adequate tool (P)
Probability
R l ti fRelative frequency Interpretation
P
Jugdmental/knowledge-based
Pfg
probabilities P
Knowledge-based probability
• P(A|K) =0.1
• The assessor compares his/her uncertainty (degree og belief) about the occurrence of the event Aog belief) about the occurrence of the event A with drawing a specific ball from an urn that contains 10 balls (Lindley, 2000).
K: background knowledgeK: background knowledge
The probability of an event is the price at• The probability of an event is the price at which the person assigning the probability is neutral between buying and selling a ticket that is worth one unit of payment if the eventthat is worth one unit of payment if the event occurs, and worthless if not
Knowledge-based probability
• P(A|K) =0.1
• The assessor compares his/her uncertainty (degree og belief) about the occurrence of the event Aog belief) about the occurrence of the event A with drawing a specific ball from an urn that contains 10 balls (Lindley, 2000).
K: background knowledgeK: background knowledge
• The need for seeingThe need for seeing beyond Pbeyond P
John offers you a game: throwingJohn offers you a game: throwing a die
• ”1 2 3 4 5”: 61,2,3,4,5 : 6
• ”6”: -24
What is your risk?
Risk
• Expected value – 24 x 1/6 + 6 x 5/6 = 1– 24 x 1/6 + 6 x 5/6 = 1
(C,P):6 5/6• 6 5/6
• -24 1/624 1/6
Is based on an important assumption – the die is fair
• While probabilities can always be assigned, the origin and amount of informationthe origin and amount of information supporting the assignments are not reflected by the numbers produced
Approaches reflecting the need for seeing beyond Pseeing beyond P
Interval probabilities p
0.1 ≤ P(A) ≤ 0.5
Aven, T. and Zio, E. (2011) Some considerations on the treatment of uncertainties in risk assessmentthe treatment of uncertainties in risk assessment for practical decision-making. Reliability Engineering and System Safety, 96, 64-74.
An adjusted approach
P,E P,E UF, , F
P ProbabilitiesE Expected values
UF Uncertainty factor assessment
Uncertainty factors
• How important are they? iti it- sensitivity
- uncertainties
Uncertainty factorimportance
Degree ofsensitivity
Significant 9 3 2,3sensitivity
Moderate 8 6 1,5
Minor 7
Mi M d t Si ifi tMinor Moderate Significant
Degree of uncertainty
Risk descriptions
P E S UP,E S UF
KK
Decision problem
Consequences Preferences
Uncertainties Values
Uncertainties Adequate tool (P)
Preferences, values Tool
Integration
Formula Optimal/rightFormula Optimal/right decision
Analysis Management
Ri k l i
Management
Risk analysisRisk acceptance criteria
Cost-benefit analysisDecision analysis
Managementreview and judgment
Decision
• Analysis informs, nothing more
• Always needs to see beyond the analyses
• Improvements of the analyses• Improvements of the analyses
The balance
Development - Protection
U
PreferencesPreferencesvalues
Decision analysis: yintegration
Analysis Management
Ri k l i
Management
Risk analysisRisk acceptance criteria
Cost-benefit analysisDecision analysis
Managementreview and judgment
Decision