Delphi Method for Expert Risk Assessment
Structured technique for gathering and refining expert judgment on risks and uncertainties.
Delphi Method for Risk Assessment
What is the Delphi Method?
The Delphi method is a structured communication technique for gathering and refining expert opinions through iterative rounds of anonymous feedback.
When to Use Delphi
- Limited historical data
- Complex, novel risks
- Need for consensus
- Expert judgment required
- Emerging risk assessment
Core Characteristics
- Anonymity: Experts respond independently
- Iteration: Multiple rounds of feedback
- Controlled Feedback: Aggregated responses shared
- Statistical Analysis: Results analyzed quantitatively
Traditional Delphi Process
Round 1: Initial Survey
- Open-ended questions
- Experts provide estimates independently
- No interaction between participants
Round 2: Feedback and Revision
- Share anonymized summary statistics
- Participants review group responses
- Revise estimates if desired
- Explain reasoning for outlier positions
Round 3+ : Convergence
- Continue iterations until:
- Consensus reached
- Stable disagreement understood
- Predetermined rounds completed
- Typically 3-4 rounds sufficient
Risk Assessment Applications
Likelihood Estimation
- Expert estimates of event probability
- Calibrate using reference classes
- Aggregate for consensus view
Impact Assessment
- Qualitative and quantitative estimates
- Multiple impact dimensions
- Range estimation (min, most likely, max)
Emerging Risk Identification
- Identify potential future risks
- Assess importance and timing
- Prioritize monitoring
Scenario Probability
- Assign probabilities to scenarios
- Validate scenario plausibility
- Inform stress testing
Facilitation Guidelines
Expert Selection
- Diverse expertise
- Relevant experience
- Independent thinkers
- 7-15 participants ideal
Question Design
- Clear, unambiguous wording
- Avoid leading questions
- Include calibration questions
- Use consistent scales
Feedback Presentation
- Summary statistics (median, IQR)
- Anonymized rationale for outliers
- Visual displays (box plots, histograms)
- Highlight areas of consensus/divergence
Convergence Criteria
- Interquartile range stability
- Kendall's coefficient of concordance
- Predetermined number of rounds
- Practical time constraints
Variations
- Real-time Delphi: Online, continuous
- Policy Delphi: Focus on options, not consensus
- Mini-Delphi: Single meeting, rapid iteration
- Decision Delphi: Paired with decision analysis
Limitations
- Time-intensive
- Expert availability
- Groupthink possibility (despite anonymity)
- Not suitable for well-quantified risks