ICAO addressed the systematic investigation of human factors in aviation accidents and incidents in the 1993 circular Human Factors Digest No 72. The main ideas highlighted in the circular concern contemporary approaches to the investigation of human factors in accidents and incidents including the application of the SHELL model and Reason's (19903) Accident Causation Model4 to the collection, analysis and presentation of human factors information. The application of these models suggest that human error is viewed as a symptom or effect of air transportation system issues as opposed to narrowly regarding operator error as the final cause of accidents (Dekker, 20021). Although the publication itself is relatively old, the ideas it describes are generic enough to remain relevant to present-day investigations and serve as an introduction to the investigation of human factors in aviation accidents and incidents. The information that follows is a synopsis of the contents of that ICAO circular.
15. By identifying the involvement of human factors in accidents and incidents, new and better measures can be implemented to reduce the frequency and minimize the consequences of repetitive human errors.
4. The events that led up to an accident or incident are the primary focus of the digest.
2. Thus, the digest is relevant to investigators and investigation authorities, civil aviation regulatory authorities, corporate management and other aviation personnel with an interest in the subject area.
Chapter 1. Need for, and purpose of, HF investigation
1.1. Human factors have, for many years, been recognized as a major contributing element in aviation incidents and accidents. Despite this knowledge, progress in adopting a uniform approach to the investigation of human factors has been slow. In some instances, the difficulty in dealing with the less tangible human factors issues has resulted in causes being attributed to “pilot factors” in the absence of any tangible technical causes.
1.2. Accident investigation reports are usually effective in describing what happened and when but in many cases the reports do not explain how and why breakdowns in human performance occurred. This narrow focus tends to ignore the root causes of the human error, or performance deficiencies, making it difficult to determine, analyse or understand the underlying human factors issues with any degree of accuracy or consistency.
1.5. Human operators, often the final barrier that stops the sequence of events from causing an accident, are only one element of the complex aviation system. Therefore, when events line up to cause an aircraft accident or incident, the investigating authority must investigate all elements of the system to fully understand how and why the occurrence happened and identify underlying deficiencies that might potentially cause another occurrence.
1.8. In other words, the investigating authority must adopt a systems approach to investigating human factors in accidents and incidents that examines the interactions between various aviation system components and how these components integrated to result in an accident/incident.
1.9. One such approach is provided by Reason's (19903) Accident Causation Model (or "Swiss Cheese Model").
1.22. This systems approach to human factors investigation encourages the investigator to go beyond the unsafe actions of front-line operators and identify latent failures present in the system that are capable of contributing to future occurrences.
1.28. Identifying these latent failures within the aviation system, whether they are at an organisational or regulatory level, allows the investigator to achieve the purpose of human factors investigation - to identify why the front-line operator committed unsafe act(s) and how operator actions led to failure in defences to determine the most effective corrective or preventive measures.
Chapter 2. Conduct of the investigation
2.1. An effective and efficient human factors investigation is methodical and complete; it should be well integrated and coordinated with other investigation elements and requires appropriate management of all available resources.
2.6. Planning and prioritising the various elements of the investigation are extremely important.
2.7. For instance, the investigator must assign high priority to the preservation and collection of evidence, especially information that is liable to being forgotten or disturbed, disappearing or becoming unavailable. This will ensure that human factors information will be available for analysis.
2.9. Within an investigation team, the human factors investigator or group must cooperate and interact with other team members who will also collect data on relevant human factors aspects during the course of their work.
2.10. In the case of a major aircraft accident, the human factors group is generally responsible for coordinating human factor elements, this includes ensuring appropriate and sufficient data are collected as well as producing meaningful results.
2.15. Investigators must collect information on the decisions, actions and behaviours of all people involved in the occurrence and the conditions under which these dimensions were carried out to obtain a full understanding of how the window of accident opportunity was created.
2.33. This information can be acquired from both primary and secondary sources. Primary sources reveal factual field information. Secondary sources provide additional empirical information about human factors that may be used to facilitate analysis of primary information.
2.16. The SHELL model facilitates data collection by providing a systematic approach to identifying human factors issues.
2.18. Information collected during the human factors investigation should allow a thorough analysis of each SHELL component and identify where mismatches, or interactions, between these components and the central human operator contributed to the occurrence.
2.20. Data collected on the human operator (liveware component) at the centre of the SHELL model can be broken down into physical, physiological, psychological and psychosocial considerations.
2.26. In determining how much information is enough, good judgement is required by the investigator.
2.27. Part of this judgement requires some form of ongoing analytical reasoning process to integrate and develop various aspects of the investigation.
2.29. However, the depth and detail of information collected during the human factors investigation is limited by available investigative resources and should exclude aspects pursued for a purpose unrelated to accident prevention.
2.30. For instance, data concerning who to blame is irrelevant to the purpose of investigating human factors and should not be collected as it fails to help explain how and why the causal event sequence was initiated and not stopped before the accident/incident occurrence.
2.46. Measurable human factors information will produce a deductive argument where conclusions are self evident, depicting straight-line cause and effect relationships. However, other less tangible human factors information provide conclusions that cannot be tested conclusively and are therefore less precise and vulnerable to dispute. To develop a more complete understanding of human factors involvement, investigators are often required to work within this domain of intangible information. Often, conclusions are based on inductive reasoning and analysis that draw inferences on the most probable or most likely explanations of behavioural events.
2.47. Further problems that investigators must consider when analysing human factors information include  assessment of relevance of abnormal/non-standard behaviour,  sensitivity and privacy considerations and  speculation avoidance.
2.50. The viability of inductive conclusions is based on the consistency and acceptance of the reasoning process used by the investigator and the weight of evidence supporting the conclusions.
2.52. A step-by-step reasoning process has been developed to ensure that the investigator considers all reasonable possibilities to deal with less tangible human factors evidence:
2.53. Step One - Test for Existence: Aims to establish or test the probability of existence of one or more human factors conditions.
2.54. Step Two - Test for Influence: Aims to establish or test the probability that a human factors condition or conditions did contribute to the sequence of events leading to an accident/incident.
2.55. Step Three - Test for Validity: Forces the investigator to draw conclusions of probability in a systematic way on the basis of indisputable empirical knowledge and verifiable evidence.
Chapter 3. Reporting and preventive action
3.2. Varied readership and sensitivity to different reader motivations are important factors to consider when writing the investigation report.
3.4. However, the most important readers are the decision-makers responsible for implementation of the report’s safety recommendations. These decision-makers must be convinced by the report if preventive actions are to be taken.
3.3. As a result, the report must state accident/incident causes, identify hazards uncovered during the investigation, indicate effective or ineffective handling of the hazards and offer recommendations to eliminate or control hazards so as to prevent future accidents.
3.5. Decision-makers are not going to act on a flawed or poorly substantiated report therefore for the sake of accident prevention, it is important that the report is complete and accurate.
3.7. The final report must be structured in an orderly and logical manner, guided by the standard format, consisting of various sections contained in the appendix to ICAO Annex 13.
3.8. Section 1 of the report concerns factual information that describes what happened as well as information pertinent to understanding the conditions and circumstances of the occurrence.
3.9. Section 1 and its subsections contain human factors information and issues, including history of flight, personnel information, aircraft information, additional information etc.
3.11. Throughout this section, deviations, discrepancies and hazards are compared to a recognised aviation standard/empirical evidence to provide the foundation for analysis of their influence on the accident sequence of events.
3.12. Section 2 of the report concerns the analysis, identifying the existence and notion of human factors involvement. It describes or validates the reasons why the circumstances resulted in the accident/incident thus creating a link between the factual information provided in Section 1 and the conclusions provided in Section 3.
3.17. Discussion of the causation chain, causal hazards and non-contributory hazards that warrant safety action is a key part of the analysis section.
3.14. The analysis section may be logically sequenced in any number of ways according to the particular circumstances of the accident/incident.
3.13. For instance, each sub-section may read like a mini accident report, developing the relationship between causal links.
3.15. Similarly, Reason's accident causation model may be used as a systematic framework to present the analysis of factual information.
3.18. Section 3 of the report concerns the conclusions. Conclusions should identify all hazards that need to be addressed, cause-related or not.
3.20. Conclusions about causes should consist of concise statements regarding the reasons why the accident happened.
3.21. Causal statements should be  listed with all causes considered,  formulated with consideration to corrective and preventive measures,  linked to safety recommendations and  written without hints of blame or liability.
3.25. Section 4 of the report concerns safety recommendations for preventive action and should address all identified hazards that need to be rectified, both matters directly associated with the causal factors of the occurrence and non-causal matters revealed by the investigation process.
3.33. Various alternative safety recommendations should be assessed for technical feasibility, acceptability, practicality, ease of implementation and appropriateness for the recipient of the recommendations.
3.27. Furthermore, in accordance with Reason's model, safety recommendations for preventive measures should focus on underlying system hazards created or ignored by management. These preconditions for unsafe acts are often the result of management decisions, action or inaction.
3.34. Safety recommendations should be general rather than authoritative. Thus, recommendations should not focus on how to make changes but rather what needs to change to  eliminate safety hazards (first and foremost),  modify the system to reduce hazard risk or  teach people how to cope with hazards that cannot be controlled.
3.28. Conversely, effective identification, elimination or mitigation of safety hazards is dependent on the response strategies that aviation companies, manufacturers or regulators adopt. Response strategies include  problem denial,  repairing the problem to prevent recurrence or  system reform/optimisation.
3.36. After the draft final report is produced, it is circulated to the States involved in an accident investigation for consultation and review. This consultative process could also occur within a State between parties involved in a domestic occurrence.
3.39. Finally, human factors data collected during an accident/incident investigation should be recorded in a database such as ADREP to facilitate future safety analysis.
3.42. In order to learn from the lessons of others there is a need to provide improved means of reporting and recording human factors data in a user-friendly manner.
Want to know more?
- AviationKnowledge - ICAO: fundamental human factors concepts
- This AviationKnowledge page provides a general introduction to human factors. A basic knowledge of human factors is required to conduct an effective human factors investigation.
- ICAO Human Factors Digest No7
- You can purchase the digest from ICAO as "ICAO - International Civil Aviation Organisation (1993). Human factors digest no 7: investigation of human factors in accidents and incidents, Circular 240-AN/144. ICAO (Montreal, Canada), 1993.
Contributors to this page
Melissa J PERRY (2009). School of Aviation, Massey University, New Zealand MJPerry