GOMS and KLM with Expert User Timing Studies

Description

Predicts expert task times by decomposing tasks into GOMS methods and keystroke-level operators, calibrated with timing studies of real expert users and devices.

GOMS (Goals, Operators, Methods, Selection rules) models skilled, error-free performance; its simplest variant, the Keystroke-Level Model, predicts execution time by summing standard operator times: keystroke (about 0.2 s for a skilled typist), pointing (about 1.1 s), homing hands between devices (0.4 s) and mental preparation (1.35 s), using placement rules for M operators. Predictions typically fall within about 20% of observed expert times. Its rigour, and its decline, lie in calibration: standard operator times are population averages from 1980s hardware. Running short timing studies with actual expert users on current devices to re-estimate operator times, especially for touch, and validating model predictions against observed times, makes KLM a cheap way to compare designs for high-frequency tasks without large experiments.

Attributes

Loss type
Vestigial
Effort
Days

Steps

  1. Empirically derive KLM parameters for your specific user group and device: time 20+ participants performing each operator type in isolation.
  2. Compute trimmed means.
  3. For touch interfaces: separate tapping (0.3–0.5s) from swipe (0.2–0.4s).
  4. For elderly users: K≈0.35s, P≈1.5s.
  5. For expert keyboard users: K≈0.12s.
  6. Apply updated parameters to KLM predictions for the target interface.

Sources

  • Human-Computer Interaction (Dix, Finlay, Abowd & Beale)
  • The Psychology of Human-Computer Interaction (Card, Moran & Newell)
  • Human-Computer Interaction — Alan Dix

Also known as

  • Keystroke-Level Model calibration
  • GOMS analysis
  • KLM timing

Categories

  • User Research
  • Task Analysis & Cognitive Modelling