Targeted tasks that provoke rare behaviour

Description

Design an artificial task, with time pressure or added noise, that makes a rare error or behaviour occur often enough to compare designs statistically.

Some important errors are too rare to observe in a lab session. An expert slip, moving off an on-screen button before release so the action silently fails, happened about once a month in real use and never in careful short tests. After analysing the mechanism, researchers built a task that invites it: type a number on an on-screen keypad, confirm, then move quickly to a distant next button under time pressure. This produced up to a dozen missed errors per session even with novices, enough to show that a sonic click made every slip noticed. In the extreme demonic variant, random noise is added to pointer movement to force overshoots so recovery behaviour can be studied. The method trades ecological validity for statistical power, and generalisation rests on the analysed mechanism.

Attributes

Loss type
Latent
Effort
Weeks

Steps

  1. Analyse the rare behaviour's mechanism step by step from incidents, logs or theory.
  2. Design a task whose structure invites that behaviour, such as rapid sequences or distant targets.
  3. Add load (time pressure, secondary tasks) or perturbation (random pointer noise) to raise its frequency.
  4. Pilot until the behaviour occurs often enough per session for statistical comparison.
  5. Run the comparison of designs on the now-frequent behaviour.
  6. Argue generalisation from the mechanism and other evidence, and state the limits on real-use claims.

Sources

  • Statistics for HCI (Dix, 2020)

Also known as

  • Causing trouble with buttons
  • Slip-provoking task design
  • Demonic interventions
  • Targeted tasks that provoke rare behaviour (including demonic interventions)

Categories

  • Accessibility
  • Human Factors & Error