Engineering, production, governance, cost, communication · insight-1296
Use quantile dotplots, not only interval bars, when non-specialists decide under uncertainty
- Evidence type
- empirical finding
- Difficulty
- medium
- Roles
- Data analyst, Data scientist
- Industries
- aviation, logistics, energy
Summary: situation, action, reason, boundary
Situation
A gate-staffing planner receives delay forecasts as 80 percent interval bands and acts on the midpoint anyway (hypothetical).
Action
Show 20 to 50 equally likely outcomes of the predictive distribution as dots, mark the decision threshold, state the chance of crossing it, and test the display with two or three real users on a real decision before rollout.
Reason
In one incentivised bus-arrival experiment, quantile dotplots gave better and more consistent decisions than a no-uncertainty control, and text depended on which probability range it stated.
Boundary
Forecasts and risk scores read by planners, managers and customers, not by statisticians.
Symptom or risk
Bars are read as certainty inside and nothing outside, so tail risk and the chance of crossing the decision threshold are ignored.
Exceptions
Expert audiences may prefer fan charts, and a single-domain experiment may not transfer; test with your readers.
Trade-offs
Dots take more space than a bar; use small multiples with shared axes.
Pitfalls
Plotting dots from a poorly calibrated distribution, which presents false precision convincingly.
Lessons and concepts
Claims and their sources
Fernandes and colleagues report that in an incentivised bus-catching experiment quantile dotplots produced the best decisions, CDF plots nearly as good, and textual uncertainty depended on the probability range communicated.