Presenting risk: does your audience understand the denominator?
In Berne, a presentation to managers in the federal administration, public health, telecommunications or precision manufacturing may need to inform a decision based on an imperfect signal. The managerial question is specific: how should that signal be presented so that the audience distinguishes its reliability from the probability that an alert actually indicates the problem being sought?
What the literature establishes
Gerd Gigerenzer and Ulrich Hoffrage (1995, Psychological Review) showed, in Bayesian reasoning tasks, that natural frequencies could facilitate inferences that participants handled less successfully when given probabilities. These frequencies describe nested counts drawn from a common population, rather than conditional percentages presented separately. For an oral presentation, this suggests making the starting population, the relevant cases and the resulting alerts explicit, rather than assuming the audience will spontaneously reconstruct their relationships.
A representation that can be learned
Peter Sedlmeier and Gerd Gigerenzer (2001, Journal of Experimental Psychology: General) investigated training in Bayesian reasoning that included constructing natural-frequency representations. Their findings support teaching a way of representing the problem rather than relying solely on a formula. The relevant mechanism for a speaker is therefore not cosmetic simplification: it is making the subsets needed for the calculation visible.
The hasty conclusion
One might conclude that all percentages should be replaced with counts. That would confuse natural frequencies with any presentation of whole numbers: their value lies in preserving the relationships between groups. A presentation that juxtaposes counts calculated from different populations may leave the very difficulty it claims to solve untouched. (our executive and employee training programmes)
What these findings do not demonstrate
These studies concern reasoning and learning tasks, not presentations to committees in Berne. They demonstrate neither that a format makes a speaker more credible nor that a better answer to a problem produces a better organisational decision. The choice of an alert threshold, the cost of errors and data quality remain separate questions that a clear presentation cannot settle.
A practical check in Berne
Within SHR’s “Public Speaking & impact” programme, an exercise to test in Berne would involve preparing two versions of the same presentation: one using conditional probabilities, the other natural frequencies, with comparable content and duration. The case could concern an administrative alert, screening, a telecommunications incident or a precision inspection, using documented data and a reference answer calculated beforehand. Randomly assigning participants to the versions and then asking each person what proportion of alerts actually indicates the target problem would provide a verifiable measure: the proportion of correct answers under a scoring rule set before the exercise. This local result would help select a presentation format, not establish the general superiority of a method. To go further: explore the Public Speaking & Impact training in Bern in the canton of Bern, or browse our executive and employee training programmes in Switzerland.
In pictures: Public Speaking & Impact in Bern
- public speaking & impact
- Bern
- research
- training bern