Human Factors Analysis and Classification System (HFACS)

Description

Code the human contributions to incidents at four nested levels, from unsafe acts up to organisational influences, then aggregate codes across incidents to find systemic patterns.

HFACS operationalises Reason's Swiss cheese model as a coding scheme. Each incident is coded at four levels: unsafe acts (errors and violations); preconditions for unsafe acts (substandard conditions and practices of operators); unsafe supervision (inadequate supervision, planned inappropriate operations, failure to correct known problems, supervisory violations); and organisational influences (resource management, organisational climate, operational process). A later variant adds external factors such as economic, policy and regulatory pressures. Because every incident is coded with the same categories, frequencies can be compared across many incidents, shifting attention from individual blame to the latent conditions that recur. It has been adapted to rail, mining, maritime and process industries, and categories normally need tailoring to the domain, with coder agreement checked.

Attributes

Loss type
Siloed
Stage
Post-launch
Effort
Weeks

Steps

  1. Gather the incident record: reports, interviews, logs and timelines.
  2. Identify each unsafe act and classify it as an error (skill, decision or perceptual) or a violation.
  3. For each unsafe act, identify the preconditions (operator state, environment, team practices) that enabled it.
  4. Trace supervisory factors that allowed those preconditions to persist.
  5. Trace organisational influences (resources, climate, processes) and, in the extended form, external factors.
  6. Have two coders code independently on a sample and resolve disagreements to stabilise the categories.
  7. Aggregate codes across incidents and target interventions at the most frequent supervisory and organisational causes.

Sources

  • Handbook of Human Factors and Ergonomics (Salvendy & Karwowski, 2021)

Also known as

  • HFACS

Categories

  • Accessibility
  • Human Factors & Error