Q2.09.3Inspection findings versus user evidencedesignresearch

Heuristic evaluation yields an issue list, not user data

Aliases: inspection issue list · expert prediction · user evidence

What it is

Heuristic evaluation produces inspection findings: principle-based predictions about where users may be hindered, not observations of failure, prevalence, or experience. The list supports early repair and hypothesis formation but is not evidence that “users proved” a problem.

Why it happens

Evaluators map interface states to principles and infer consequences without target users performing tasks. This exposes prelaunch risks but creates false positives users bypass and misses problems dependent on real knowledge, motivation, and setting.

Studying it

Record screen evidence, trigger, violated principle, predicted consequence, and confidence. Compare against usability sessions, support incidents, or telemetry for hits, unique findings, and false positives. Nonoccurrence in a small user sample is not proof of absence; validation must match state and population.

Where it stops holding

Absence of user data does not mean no evidence: safety, accessibility, and consistency can rely on standards and engineering facts. User tests are not the sole truth either. The requirement is to label evidence type and avoid presenting predicted prevalence or severity as observed.

Applying it

  • Label each item as an expert prediction with state, consequence, and uncertainty.
  • Fix clear low-cost issues; validate disputed high-risk ones with users.
  • Never infer affected-user proportions from issue counts.
  • Reinspect the principle after repair and test practical impact in target tasks.

Related

  • Same group: Q2.09.1 Independent then aggregated · Q2.09.2 Limited single-reviewer coverage
  • Adjacent: Q2.06 Usability testing · Q2.10.3 Expert review cannot replace user testing
  • Search terms: inspection finding · expert prediction · user evidence

Cards in the same group

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/handbook/Q2.09.3