Skip to content

Base-Rate Visibility Panel

Display tool — instantiates Salience-Significance Decoupling

Places the base rate and its denominator beside a vivid instance, so a striking case cannot be read as representative.

A vivid case arrives with no denominator attached, and the mind supplies the wrong one. Base-Rate Visibility Panel is the artifact that puts the missing denominator back on the screen: next to the striking instance it shows how often the thing actually happens, out of how many, against what background — the prevalence, the comparison group, the "and this many times it didn't." Its defining commitment is quantitative and always-on: it does not argue about the vivid case or hunt for counter-stories, it simply refuses to display an instance without the rate that makes it interpretable, so the reader literally cannot see the anecdote without seeing the base rate in the same frame.

Example

A clinic is fielding alarmed calls about a frightening side effect after a handful of vivid reports circulate locally. Each report, on its own, reads as a warning. The Base-Rate Visibility Panel reframes the numbers instead of the stories. Beside the reported cases it displays the prevalence — the event occurs in roughly 3 per 100,000 courses of the drug — and the two comparisons that make that number mean something: the same event's background rate in people not taking the drug (about the same), and the denominator this cluster came from (tens of thousands of prescriptions filled without incident).

Set beside the base rate, the cluster stops reading as a trend and starts reading as what a large denominator predicts. The panel doesn't tell anyone what to decide; it restores the comparison the vivid cases had stripped away, so clinicians and patients weigh the risk against a real baseline rather than against the vividness of a few frightening anecdotes.

How it works

  • Attach the denominator. Every salient count is shown as a rate — this many out of that many — never as a bare numerator, which is where availability does its damage.
  • Restore the comparison group. Put the relevant baseline next to the figure: the background rate, the without-treatment rate, the prior period, whatever answers "compared to what?"
  • Keep the un-events visible. Show the silent majority — the cases where nothing happened — because their absence from memory is precisely what inflates the vivid few.
  • Stay standing, not on-demand. The panel is a permanent fixture of the display, so the base rate is present before the vivid case has time to set the anchor.

Tuning parameters

  • Reference class — which denominator you choose as "out of." The single most consequential dial: a base rate is only as honest as the population it is drawn over, and a too-broad or too-narrow class quietly changes the story.
  • Comparison baselines — how many and which — background rate, prior period, peer group. More baselines add context but can clutter the panel into unreadability.
  • Time window — the span the rate is computed over. Short windows track change but are noisy; long windows are stable but can mask a real recent shift.
  • Uncertainty display — whether the rate carries an interval or stands as a point. Bands prevent a small-sample rate from being over-read, at the cost of a busier panel.

When it helps, and when it misleads

Its strength is that it neutralizes base-rate neglect structurally rather than by exhortation — you cannot forget the denominator when it is printed beside the number.[1] It is at its best against exactly the errors vividness produces: a scary-but-rare event, a lottery-sized success story, a "crime wave" that is a reporting wave.

It misleads when the denominator is wrong. A cleverly or carelessly chosen reference class can make a real problem vanish or a trivial one loom, and a base rate over-aggregated across genuinely different subgroups can hide a hazard concentrated in one of them. The classic misuse is denominator-shopping — sliding to whichever reference class flatters the point you already wanted to make. The discipline is to justify the reference class out loud, show it, and let it be contested; the panel's honesty lives entirely in that choice.

How it implements the components

This panel realizes the restore-the-baseline side of the archetype — the quantitative correction, not the qualitative hunt:

  • base_rate_panel — it is this component: the standing display of prevalence and denominator next to the vivid instance.
  • comparison_set_restoration — it rebuilds the baseline and comparison group the salient case was seen without, so the instance is judged against a real reference rather than against nothing.

It restores the quantitative baseline but does not hunt qualitative counter-cases — that is Counterexample Surface Scan — nor diagnose why the vivid item became prominent, which Attention-Capture Inference Test handles.

Notes

The panel corrects rate neglect, not frame error. It answers "how often, out of how many," but it trusts the reference class it is handed; if the population itself is drawn from a biased frame, the panel will faithfully display an honest rate over a dishonest denominator. Reconstructing that frame is Sample Frame Reconstruction's job, and a base-rate panel is only as trustworthy as the frame feeding it.

References

[1] The tendency to under-use general prevalence information in favour of specific, vivid case detail — Kahneman and Tversky's base-rate neglect. Making the base rate co-visible with the instance is the standard structural corrective. ↩