Is There Such a Phenomena as ‘Pilot Error’ in Aviation Accidents

The term ‘Pilot error’ has been attributed to 78%[1] of Army aviation accidents. Despite the technological advances in Rotary Wing (RW) aircraft i. e. , helicopters accidents attributed to technology failure are decreasing, whilst pilot error is increasing. Currently, RW accidents are investigated and recorded using a taxonomy shown to suffer difficulties when coding human error and quantifying the sequence of events prior to an air accident. As Human Factors (HF) attributed accidents are increasing, lessons aren’t being identified nor the root cause is known. Therefore, I propose to introduce Human Factors Analysis and Classification system (HFACS) an untried taxonomy to the UK military developed as an analytical framework to investigate the role of HF in United States of America (USA) aviation accidents. HFACS, supports organizational structure, pre-cursors of psychological error and actual error; but little research exists to explain the intra-relations between the levels and components, or the application in the military RW domain. Therefore, I intend to conduct post-hoc analysis using HFACS of 30+ air accidents between 1993 to present. Implications of this research are to develop a greater understanding of how Occupational Psychology (OP) can help pilots understand HF, raise flight awareness and reduce HF attributed fatalities. Introduction â€Å"On 2 June 1994 an RAF Chinook Mk2 helicopter, ZD 576, crashed on the Mull of Kintyre on a flight from RAF Aldergrove to Fort George, near Inverness. All on board were killed: the two pilots, the two crewmembers and the 25 passengers. This was to have been a routine, non-operational flight, to take senior personnel of the security services to a conference. The sortie was planned in advance; it was entirely appropriate for these pilots, Flt Lts Jonathan Tapper and Richard Cook, and for the aircraft, ZD576, to have been assigned this mission. An RAF Board of Inquiry (BOI) was convened following the accident and carried out a detailed investigation. BOIs are established to investigate the cause of serious accidents, primarily, to make safety recommendations but, at the time of this crash, to also determine if human failings were involved. Their conclusion, after an exhaustive investigation was there was not one single piece of known fact that does not fit the conclusion that this tragic accident was a controlled flight into terrain. † The BOI found no evidence of mechanical failure and multiple witnesses stated that the aircraft appeared to be flying at 100ft at 150 knots there was no engine note change, the aircraft didn’t appear to be in distress and at the crash scene the throttle controls were still in the cruise position (not at emergency power if collision with the ground was imminent). 2] So the causation moved to Human Factors (HF). But some questions remain unanswered, on that fateful day why did these seasoned and experienced pilots fly their aircraft and passengers into a hillside at 150 knots. If this accident was attributed to HF it now appears to some that the aircrew themselves are more deadly than the aircraft they fly (Mason, 1993: cited in Murray, 1997). The crucial issue therefore is to understand why pilots Flt Lts Jonathan Tapper and Richard Cooks’ actions made sense to them at the time the fatal accident happened. Relevance of Research So why is this topic relevant to OP research? The British Army branch of aviation is an organization called the Army Air Corps (AAC) and in keeping with the trends of the other two services the Fleet Air Arm of the Royal Navy and the Royal Air Force, it has seen a steep decline in accidents in recent years. However, accidents attributed to Human Factors (HF) have steadily risen and are responsible for 90% of all aviation accidents. [3]. This research will depart from the traditional perspective of the label â€Å"pilot error† as the underlying causation of Aviation accidents, whereby current theory and research purport a ‘systemic’ approach to human factors investigation of Aviation accidents. This approach is derived from Reasons Model of Accident Causation, which examines the causal factors of organizational accidents across a spectrum of sectors from; nuclear power industry (e. g. , Chernobyl), off-shore oil and gas production (e. g. Piper Alpha) to transportation (e. g. Charring Cross) (Reason 1990). This approach recognizes that humans, as components of socio-technical systems, are involved in designing, manufacturing, maintaining, managing and operating aviation systems including the methods of selecting and assessing potential employees to the aviation industry from Pilots, Cabin crew, Engineers and Baggage handlers. Therefore, our ability to identify, understand and manage these potential issues enables us to develop systems that are more error-tolerant, thus reducing risk and the potential for accidents. I intend to be able to provide a more consistent, reliable and detailed analysis of HF causal factors that attribute to aviation accidents within the AAC. On average, the AAC experiences around 6 major accidents per year, although a record year was recorded with only two accidents in 1993. However, in 1992 aviation accidents cost over ?10M[4] in taxpayer’s money. Usually the causation of accidents are classified (human error, technical failure or operational hazard). Whilst there was a reduced figure of ?1M for 1993, the satisfaction of this financial success was marred by the fact that one of the two accidents resulted in a fatality. However, it is the concept of human error or pilot error that dominates the outcome of most BOIs particularly when there are fatalities. Current taxonomies used to classify accident causal groups do not extend beyond this distinction although more recently organizational factors have been included to reflect a more systemic view of accident causation. However, the HF domain is extensive and current taxonomies employed by the AAC do not encapsulate this. By using HFACS (currently adopted by the US Navy, Army, Airforce, and Coast Guard), a human error orientated accident investigation and analysis process; I will conduct post-hoc analysis of 30+ category four and five accidents from 1993 to present day. Literature review Before we start to look at any reduction in Air Accidents we need to grasp an understanding of category of accident. How many times when we hear about air accidents, â€Å"it was pilot error†, merely noting HF was responsible doesn’t prevent repetition nor identify any critical lessons, plus the description is far too generic. The term pilot error doesn’t assist us in understanding the processes underlying what leads to a crash, nor does it give us a means to apply remediation or even identify lessons to prevent re-occurrence. The other issue is that it is very seldom one single factor caused the helicopter to crash. Professor RG Green (1996) uses a categorization method: Modes of failure, Aircrew Factors and System failures. Within each of these exist sub-categories. E. g. , in Modes of Failure category lists a number of common errors made by the individual or individuals from; selective attention, automatic behaviour, forming inappropriate mental models, affects of fatigue and perceptual challenges leading to spatial disorientation, particularly common to RW flight. Aircrew factors, refers to background factors relevant to individuals: decision-making, personality, problem solving, Crew composition, Cockpit Authority Gradient (CAG) and Life stress. Finally, the systems factors applicable to the organization that we serve under, termed enabling conditions such as: Ergonomics, Job pressures and Organizational Culture. Bodies of Research Now, human error doesn’t just happen, usually a sequence of events will unfold prior to the accident. Human error is often a product of deeper problems; they are systematically connected to features of the individual’s tools, tasks and the surrounding media (Dekker, 2001). Therefore, in order to provide remediation through the development of strategies it is vital that we understand the various perspectives experienced through flight and how these could effect a pilot; these range from: cognitive, ergonomic, behavioural, psychosocial, aeromedical, and the Organizational Perspectives (Weigmann and Shappell 2003). Within the environment of human performance error is a unique state of a pilot’s operational environment that could be affected by anyone of, or all of the perspectives. Rasmussen (1982) utilized a cognitive methodology to understanding aircraft accidents. O’Hare et al. (1994) described the system as consisting of six stages: ‘detection of stimulus; diagnosis of the system; setting the goal; selection of strategy; adoption of procedure; and the action stage'. The model was found to be helpful in identifying the human errors involved in aviation accidents and incidents (O’Hare et al. 1994). One draw back being that these models using cognition are operator centric and do not consider other factors such as; the working environment, task properties, or the upervisory and work organization (Wiegmann and Sappell, 2001c). Edwards (1972) developed the ‘HELS system' model, which was subsequently called the ‘SHEL' model. Citing that Humans do not perform tasks on their own but within the context of a system; initially SHEL was a system focusing on the ergonomics and considered the man-machine interface. A tool that can be appli ed to investigate air accidents through the evaluation of human-machine systems failure. The ‘SHEL' model categorizes failure into: software, hardware, liveware and environment conditions. However the SHEL model fails to address the functions of management and the cultural aspects of society. Empirical findings Bird’s Domino Theory (1974) views accidents as a linear sequence of related factors or series of events that lead to an actual mishap. The theory covers the five-step sequence First domain Safety/Loss of control, the second domain, basic causes, identifies the origin of causes, such as human, environment or task related. The immediate causes include substandard practices and circumstances. The fourth domain involves contact with hazards. The last domain could be related to personal injury and damage to assets (Bird, 1974; and Heinreich, et al. , 1980). It is much like falling dominos each step causes the next to occur. Removing the factors from any of the first three dominos could prevent an accident. This view has been expanded upon by Reason (1990). Reason’s ‘Swiss cheese' model fig 1, includes four levels of human failure: organizational factors, unsafe supervision, preconditions for unsafe acts and unsafe acts. The HFACS was developed from this model in order to address some of limitations. The starting point for the chain of event is the organization ‘Fallible decisions' take place at higher levels, resulting in latent defects waiting for enabling factors (Reason, 1990). Management and safe supervision underpins any air operation through flight operations, planning, maintenance and training. However, it is the corporate executives, the decision makers who make available the resources, finances and set budgets. These are then cascaded down through the tiers of management and to the operator. Now this sounds like an efficient and effective organization and according to Reason failures in the organization come about by the breakdown in interactions and holes begin to form in the cheese. Within an organization unsafe acts may be manifested by lack of supervision attributed to organizational cultures operating within a: high-pressure environment, insufficient training or poor communication. The latent conditions at the unsafe supervision level promote hazard formation and increase the operational risks. Working towards the accident, the third level of the model is preconditions for unsafe acts. Performance of the aircrew can be affected by fatigue, complacency, inadequate design and their psychological and physical state (USNSC, 2001; Shappell and Wiegmann, 2001a; Wiegmann and Shappell, 2003). Finally, the unsafe acts of the operator are the direct causal factor of the accident. These actions committed by the aircrew could be either intentional or unintentional (Reason, 1990). The ‘Swiss cheese' model sees the aviation environment as a multifaceted system that does not work well when an incorrect decision been taken at higher levels (Wiegmann and Shappell, 2003). The model depicts a thin veneer of cheese the veneer symbolizing the defence against Aviation accidents and the dotted holes portray a latent condition or active failure. It is a chain of events that usually lead to an accident however as errors are made the holes begin to appear in the cheese, a datum line penetrates the cheese and if all the holes pass through the line, then a catastrophic failure occurs and a crash ensues. These causal attributions of poor management and supervision (organizational perspective) may only be unearthed if equipment is found in poor maintenance (ergonomic). If the organizational culture is one of a pressured environment then this could place unnecessary demands on the aircrew producing fatigue (Aeromedical). Or management could ignore pilots’ concerns if the CAG was at imbalance (psychosocial perspective). All of these factors could hinder and prevent aircrew from processing and performing efficiently in the cockpit, which could result in pilot error followed later by an Air Accident. However, with Reasons model it doesn’t identify what the holes in the cheese depict. For any intervention strategy to function and prevent reoccurrence the organization must be able to identify the causal factors involved. The important issue in a HF investigation is to understand why pilots’ actions made sense to them at the time the accident happened (Dekker, 2002). HFACS was specifically developed to define latent and active failures implicated in Reasons Swiss Cheese model so it could be used an accident investigation and analysis tool (Shappel and Weigmann, 1997; 1998; 1999; 2000; 2001). The framework was developed and refined by analyzing hundreds of accident reports containing thousands of human causal factors. Although designed originally for use within the context of the military aviation HFACS has shown to be effective within the civil aviation arena as well (Wiegmann and Shappel, 2001b). Specifically HFACS describes four levels of failure; each one corresponds to one of the cheese slices of Reasons model. These are a) Unsafe acts b) Pre-conditions for Unsafe acts c) Unsafe supervision and d) Organizational influences (Weigmann and Shappel, 2001c) Methodology By using a combination of qualitative (i. e. the process of recoding causal factors based on individual and group discussions) and quantitative (causal factor analysis of recoded narratives against HFACS taxonomy) research methodologies to identify further causal groups to be used in classifying accidents and to assess the validity of the HFACS framework as a tool to classify and analyze accidents. Data to be used in this study will be derived from the narrative findings of AAC BOIs conducted between 1990 and 2006[5]. This should equate to approximately 30-35 narratives to be used in the analysis. Authority to access the Board of Inquiry library has been granted by the Army's Flight Safety and Standards Inspectorate, which is the AAC organization responsible for conducting Aviation accident investigations and analysis. Data will only be used that comprises of category 4 accidents (single fatalities and severe damage to aircraft) and category 5 (multiple fatalities and loss of aircraft). In addition to the narrative description in the report, the following information will also be collected: the type of mission in which the accident happened (e. . low-level flying, exercise, HELEARM[6]); the flight phase (e. g. take-off, in the hover, flight in the operational area, approach, and landing); the rank of the pilot(s) (to measure CAG and see if this is a contributory factor) involved and the type and category of aircraft. This study will concentrate on all Army helicopters; including all variants of the Lynx, Gazelle and Squirrel trainer. Coding frames will be developed and tested for use in the final recoding exercise. An Occupational Psychologist from the Human Factors epartment of the MOD will supervise the training and the coders will be a number of RW pilots with a minimum of 1000hours flying time at the time of the research. Each pilot will be provided with a workshop in the use of HFACS framework. This is to ensure parity and that all coders understand the HFACS categories. After the period of training the raters will be randomly assigned air accidents so that two independent raters can independently code each accident. It is intended to code the inter-rater reliability on a category-by-category basis. The degree of agreement (the inter-rater reliability) initially between the two coders will be achieved by Cohens Kappa (Cohen, 1960;Landis and Koch, 1977). SPSS v. 15. 0 will be used to quantify the frequency of causal factors of the 30+ narratives. It is also hoped to compare the inter-rater reliability between all the coders using Fleiss Kappa. Fleiss’s Kappa assessment method is used to measure the similarity agreement of observers and treats them symmetrically (Fleiss, 1981). The level of agreement between the raters is statistically measured against what could be achieved through chance. The Kappa level range would be classed as achieving moderate inter reliability if it were between 0. 41-0. 60. Cohen’s Kappa is based on the statistical measurement analysis of the level of agreement between raters in excess of (Landis and Koch, 1977). Discussion The research intends to apply an untried methodology not as yet sanctioned by the UKs Ministry of Defence in order to analyze a number of Air Accidents within the AAC between 1993 and present day. Thirty plus serious Category 4 and 5 accidents will be re-classified using the taxonomy of HFACS. It is intended where pilot error was the cause, to identify the HF associated and attribute to each accident. It is also hoped that the HFACS taxonomy can accommodate the HF identified during re-coding and therefore provide tangible evidence that HFACS could be used by the AAC as a reliable tool. It is hoped a number of comparison analysis can be achieved and are accidents more prevalent when flying in visual meteorological conditions (VMC) or poor visibility instrument meteorological conditions (IMC) therefore two sets of visual conditions; VMC and daylight or impoverished visual conditions IMC or twilight/nighttime. Wiegmann, D. A. and Shappell, S. A. (2003). What would also be interesting was the causation and aircrew behaviours of fatal and non-fatal accidents and are these more prevalent on operations or during training. The author was in Afghanistan 2006 and over 6-month period there wasn’t a single crash let alone fatality. But the AAC records 6 crashes a year so again this is worthy of investigation. The ranks of the pilot is also worthy of interest with regards to achieving a good CAG there may be causal evidence to indicate that an imbalance between ranks could have lead to an aircrash. The Organizational hierarchy will; also be researched is it one specific organization that keeps having crashes is there an issue with the pressures placed on the pilots by the organization. The inter-rater reliability will also be calculated by using Fleiss Kappa which will work for more than two raters, it is intended that an acceptable level of inter rater reliability will be recorded. In addition, the intra-rater reliability as a holistic measurement is hoped to be high in order to support the credibility of the results. An Organization could benefit from gaining a standardized, consistent coding methodology and that data can be used for identifying trends and intervention strategies can then target these trends in accident causation. It is hoped that granularity can be achieved beyond the label â€Å"pilot error† and identify the underlying causation of the accident. If successful and if HFACS is adopted UK military wide, perhaps the real cause of why ZD576 flew into the Mull of Kyntre could be unearthed. If other Military organizations can reap success then HFACS could be a reliable tool to identify causation and could be used in accident investigation. Ethics I will comply fully with the BPS[7] ethical principles when conducting research with human participants. All identifiable information relating to individuals discussed in the narrative findings will be removed in accordance with the data protection act, for the purposes of analysis and reporting. All participates will be fully appraised of my research, recognize that all the coders are volunteers and give informed consent before the research and to understand how the information will be used. The coders will be reviewing material depicting instances of fatalities therefore it is important that the coders do not come to any psychological harm, over and above the risk of harm in ordinary life (participants will be invited to contact me if participation causes concern at any time or to ask questions). Maintaining a good rapport particularly with the coders is also a desirable. Being an Aeronautical Engineer should also bridge any cultural gaps and maintain a good working relationship.