A map that broadens investigation
AcciMap is associated with Rasmussen and Svedung’s work on risk management in sociotechnical systems. It represents factors across levels and relationships that help explain an accident, from operations to relevant organisational or regulatory decisions.
It complements thinking about root cause analysis by avoiding concentration on one immediate action. It does not require a failure to be assigned to every level: relationships should only be included where the case evidence can support them.
Levels and interactions
The map distinguishes decision-making areas such as company management, supervision, operational work and process conditions. Legislation, regulators or other entities may be included where their actions or limitations help explain the event under investigation.
Positioning a factor on the map does not itself demonstrate blame or importance. What matters is explaining how it affected conditions at other levels, for example through objectives, resources, information, constraints or delays in responding to change.
Evidence-based construction
Investigation brings together chronology, documents, interviews and technical information. Each relationship should express an understandable mechanism rather than act as a decorative arrow linking any fact to the outcome. Confirmed relationships should be distinguished from hypotheses still requiring examination.
For example, a decision about deadlines may affect planning and supervisory availability. Inclusion requires evidence of that connection in the particular case; assuming that every accident results from production pressure would not be sufficient.
Using the map for improvement
AcciMap may reveal that changing only the last barrier would leave conditions that contributed to the event intact. Recommendations should explain which relationship they seek to change and who has the ability to do so, including coordination across organisations where appropriate.
The map does not calculate probabilities or replace specialist technical analysis. Nor should it expand until it becomes unreadable. Its value is a traceable and sufficiently broad explanation supporting discussion of measures and verification of their effects, without treating a graphical representation as automatic proof of causation in the event under investigation.
