Spiegelhalter: Communicating Risk
Key insights from David Spiegelhalter's work at the Centre for Risk Communication at Cambridge — and what investors can learn about presenting uncertainty honestly.
The communication problem
Spiegelhalter’s background is in Bayesian thinking, risk evidence, and the communication of uncertainty. His central insight for communicators of risk is deceptively simple: the aim should not be to increase trust, but rather to demonstrate trustworthiness.
O’Neill’s criteria for trustworthy communication are:
- •Intelligent openness about information
- •Accessible: people can find it
- •Intelligible: people can understand it
- •Useable: it can be checked and explained
- •Assessable: information is transparent about uncertainty
The key point: if you don’t know something, give some form of confidence interval. Communication of uncertainty is best done by being honest and transparent about what you don’t know.
Framing and formats
Giving people cold facts is not a good way of achieving informed decisions. Spiegelhalter focuses on how to nudge the audience in a certain direction without being manipulative — informing rather than persuading.
Key principles:
- •Test different formats — choose the one which fulfils your communication objectives
- •Framing is important: present both positive and negative framing
- •Use frequencies (this is rain: 70 out of 100 times) rather than abstract probabilities — frequencies framed by numerous journals and professional bodies appear to be effective communication tools
- •In areas like weather forecasting, public does appear to understand that forecasts are based on multiple possible futures
- •Humans spend their lives on one axis (severity of event) seeking to make low risk events even less likely — the other axis (probability of event) is often ignored
Levels of uncertainty
Spiegelhalter introduces a useful scale for expressing how knowable an event may be — from direct expression of uncertainty (where you can put numbers on it) through to epistemic uncertainty (where you genuinely don’t know). This connects directly to the three types of uncertainty framework discussed in the Art of Uncertainty piece.
His practical rating scale:
- •4+ = can know the probabilities (like a roulette wheel)
- •3+ = have good evidence as to the shape of the probability distribution
- •2+ = have an idea but now evidence could have a substantial impact on risk appraisal
- •1 = a total risk — unknown unknowns. It just makes sense to take a precautionary approach and focus on resilience
Application to investment
Several practical takeaways for investment communication:
- •Volatility as risk for some is a lovely dose of confirmation bias. We have a different definition of risk to others — could we avoid the word 'risk'?
- •Focus on plausibility/strength of evidence — have a matrix like the National Risk Register with magnitude on one axis and probability on the other
- •Potentially try and calibrate or compare our risk scores or investment outcomes against the general theoretical distribution for value stocks
- •Use scenario analysis to evaluate possible different futures and how you might respond to them differently
- •Bank of England GDP forecast graphs are a good example of expressing uncertainty — they give no central forecast, reducing anchoring on a specific number
- •Trustworthiness vs credibility: need to be a trustworthy source before a crisis occurs, not after
Key Takeaway