Literature Review on Aviation Safety

I spent two months without flying to work on a research class. My research topic was the role of pilot error in aviation safety, so I thought I’d share it with you… as an apology for being gone so long.

Literature Review

Aviation has always had its risks, but none have been so constant or difficult to remedy as that of pilot error. Since the days when the fabled Icarus flew too close to the sun and melted his wings, the primary cause of aviation accidents has not been failed mechanics or inclement weather, but pilot error. O’Hare, Wiggins, Batt and Marrison found in their 1994 study that the quantity of accidents due to pilot error ranges from 60% to 80%, making it the primary factor in aviation mishaps. Even more disconcerting is the notion that while the reliability of aircraft mechanisms, specifically engines, have been steadily improving, the rate of accidents due to pilot error has not. Contrary to the popular belief that pilot error is the ‘final frontier’ in aviation, a study by Hobbs in 2004 showed that the rate of accidents as a result of pilot error has remained stable, indicating that human factors have always been the primary issue regarding flight safety. “Although the nature of aviation changed dramatically over the first 100 years of powered flight, the primary place of people as determinants of safety remained remarkably constant” (p. 340). This review will identify the methods used to classify and analyze pilot errors, identify the types of pilot errors most commonly associated with aviation mishaps, and assess factors that may contribute to the prevalence of those error types.

Pilot error is recognized as the leading cause of accidents by regulatory agencies and industry experts. The Federal Aviation Administration states that “Despite all the changes in technology to improve flight safety, one factor remains the same — the human factor. It is estimated that approximately 75 percent of all aviation accidents are human factors related” (2003, p. 16–1). To address the issue, governments have implemented a variety of training and education programs over the last 30 years. O’Hare suggests that historical stability of pilot error as the primary cause of aviation mishaps is evidence that these programs are inefficacious (2000, p. 2003). He suggests that “The lack of definitions and criteria for coding human errors … may be one of the reasons why the proportion of accidents considered due to human error has remained stubbornly high for several decades” (2000, p. 2003). Taneja elaborates that “Human factors in aircraft accidents … are less tangible and quantifiable than the mechanical causes. Variability in the thoroughness and precision of human factors investigation then result in accident databases that may not be amenable to easy interpretation” (2002, p. 2).

The aviation industry has recognized the need to systematically review dangerous incidents and accidents in order to classify and understand pilot error. “In recent years, progressive civilian and military–mishap investigating organizations throughout the world have been insisting that human-performance analyses be undertaken as a routine part of their activities” (Diehl, 1991, p. 105). Structured analysis of accidents using various cognitive frameworks has highlighted several trends in aviation mishaps.

One such cognitive framework is Reason’s Model of Unsafe Acts (Reason, 1990). Wiegmann & Shappell, in their 1997 study, elaborate that this model classifies unsafe acts as either intentional or unintentional. Unintentional acts are comprised of “slips” due to attention failure and “lapses” due to memory failure. Intentional acts refer to “mistakes” and “violations”. Mistakes occur when rules and procedures are incorrectly applied, or when procedures do not exist for the given situation, while violations are actions that are performed to deliberately break rules and regulations (p. 71). When Weigmann & Shappell applied Reason’s model to a database of aviation accidents from the U.S. Navy and Marine Corps, the most common unsafe acts were mistakes — the incorrect implementation of procedures — comprising between 54% and 60% of pilot errors.

An alternative cognitive framework that has been widely used is Rasmussen’s Taxonomic Algorithm (Rasmussen, 1982). According to O’Hare et al. (1994), Rasmussen developed a model of information processing comprised of a series of steps related to decision-making. When an accident is analyzed, the user attempts to determine which step has been incorrectly performed or skipped. Cognitive errors are categorized as Information, Diagnostic, Goal, Strategy, Procedure and Action (pp. 1862–1863). These error categories are used to classify accidents from a database involving New Zealand civil aircraft from 1982-1991. O’Hare et al. found that the final two steps in the taxonomy, procedure and action, constituted 41% of pilot errors, with errors involving choice of procedure being the most frequent at 26% (p. 1864). In a similar study, Weigmann & Shappell (1997) applied Rasmussen’s Taxonomic Algorithm to the same database it had used for Reason’s model. They found that procedure errors and action errors comprised between 35% and 55% of the pilot errors examined (p. 76).

Both Reason’s Model and Rasmussen’s Taxonomy indicate that errors in implementation of procedures are the most common form of pilot cognitive error. While Reason’s Model is derived from the pilot motive, Rasmussen’s Taxonomy follows a chain of events that offers more insight into the nature of these errors. Wiegmann & Shappell (1997) show that implementation errors occur after many other cognitive steps have been successfully completed. The pilot has already identified a problem, chosen a reasonable goal, and decided on a strategy for achieving the goal. However, the most common error occurs when the pilot attempts, and fails, to execute the procedure that is consistent with the strategy selected. Thus, most pilot errors do not occur in the identification of the problem or the identification of the solution, but in the execution of that solution (p. 72).

O’Hare further explores the cognitive demands that affect pilot error using his own framework, the “Wheel of Misfortune” (2000). This framework combines Reason’s Model, Rasmussen’s Taxonomy, and Roth & Wood’s “cognitive triad” (1988). In O’Hare’s “Wheel of Misfortune” model, performance of tasks is constrained by three factors: the inherent demands of the task environment, the resources available to the pilot and the interface with the task environment. This model is rooted in the assumption that a task is achievable so long as “the resources available to complete the task are equal to, or exceed the demand of the task” (pp. 2008–2009). If the cognitive demands on a pilot are greater than resources he has available, he will not be able to complete the task. These frameworks enhance researchers’ understanding of a pilot’s mental limitations and highlight “the detrimental effect that cognitive load can have on user performance” (Busse & Johnson, 1998, p.36).

Industry studies regarding effects of cognitive workload on pilot performance have been limited to commercial and military aviation. Most often, studies assessing pilot performance relative to cognitive workload are evaluations of new cockpit systems. Singer & Dekker find that airline cockpit warning systems that significantly add to a pilot’s cognitive workload cause erroneous diagnoses of system failures (2000, p. 72). “Human performance gains become visible as soon as the warning system itself manages various failure in some way before presenting them” (2000, p. 74).

Outside of commercial aviation, Lichacz performed a landmark study on the cumulative effects of sleep deprivation, time pressure and workload on performance in a dynamic air traffic control (ATC) task. “The ATC task is representative of many of the practical characteristics faced by individuals in real-world operating environments” (2002, p. 48). In this experiment, Lichacz finds that subjects make more mistakes when confronted with high mental workload scenarios than in low workload scenarios, and they commit dramatically more errors in scenarios with high mental workload and high time pressure (2002, p. 54).

These studies suggest that increasing a pilot’s cognitive workload increases that pilot’s rate of cognitive errors. Future research is needed to establish causality between pilot workload and pilot errors. If causality exists, cognitive frameworks should be applied to categorize the type of errors generated by increases in workload. A strong correlation between workload and errors implementing procedures would imply that the aviation industry and regulatory agencies could reduce the rate of fatal accidents and dangerous incidents by focusing on cockpit resource management and other workload management techniques.

Works Cited

Busse, D., & Johnson, C. (1998). Using a cognitive theoretical framework to support accident analysis. Symposium conducted at the Human Error Safety and Systems Development Conference, Seattle, USA.

Diehl, A. (1991). Human performance and systems safety considerations in aviation mishaps. International Journal of Aviation Psychology, 1(2), 97-106.

Federal Aviation Administration. (2003) Pilot’s Handbook of Aeronautical Knowledge: FAA-H-8083-25. Washington, DC: U.S. Government Printing Office (GPO).

Hobbs, A. (2004). Human factors: The last frontier of aviation safety? International Journal of Aviation Psychology, 14(4), 335–341.

Lichacz, F. (2005). Examining the effects of combined stressors on dynamic task performance. International Journal of Aviation Psychology, 15(1), 45–66.

O’Hare, D. (1999). Safety is more than accident prevention: Risk factors for crashes and injuries in general aviation. In D. O’Hare (Ed.), Human performance in general aviation (pp. 265–289). Aldershot: Ashgate.

O’Hare, D. (2000). The ‘Wheel of Misfortune’: A taxonomic approach to human factors in accident investigation and analysis in aviation and other complex systems. Ergonimics, 43(12), 2001–2019.

O’Hare, D., Batt, R., Wiggins, M. and Morrison, D. (1994). Cognitive failure analysis for aircraft accident investigation. Ergonomics, 37, 1855–1869.

Rasmussen, J. (1982). Human errors: A taxonomy for describing human malfunction in industrial installations, Journal of Occupational Accidents, 4, 311–333.

Reason, J, (1990). Human error. New York: Cambridge University Press.

Roth, E.M. & Woods, D.D. (1988). Aiding human performance, I: Cognitive analysis. Le Travail Humain, 51, 39–64.

Singer, G., & Dekker, S. (2000). Pilot performance during multiple failures: An empirical study of different warning systems. Transportation Human Factors, 2(1), 63–76.

Taneja, N. (2002). Human factors in aircraft accidents: A Holistic approach to intervention strategies. Proceedings of the 46th Annual Meeting of the Human Factors and Ergonomics Society. Santa Monica, Human Factors and Ergonomics Society.

Wiegmann, D., & Shappell S. (1997). Human factors analysis of post accident data: Applying theoretical taxonomies of human error. International Journal of Aviation Psychology, 7(1), 67–81.

Photos from a Fine Week of Flying

I'm still scrambling to write entries from the last couple of weeks. But I have photos to share from that time period, so enjoy. I have some teasers below.

Airplane Show at Clow (1C5)
Long Solo Cross-Country