Aviation may be classified as a high-hazard industry: the safety of flight crew, and, in the case of commercial aviation, passengers depends on the precise execution of multiple high-risk maneuvers coordinated by a team of skilled pilots and technicians. Over the past several decades, the commercial aviation industry has garnered wide recognition for its safety record, taking its place among military and nuclear operations as an example of a high-reliability organization (HRO). HROs are organizations that exist in hazardous environments where errors carry high consequences, yet the error rate is extremely low [9].

Prior to the 1980, training of civilian and military pilots focused almost exclusively on the technical aspects of flight [10]. Prompted by a series of high-profile, high-fatality airline incidents, the aviation and aerospace industries began to take a critical look at why planes crashed. An example is the Tenerife disaster in 1977, where two fully loaded jumbo jets (KLM Flight 4805 and Pan Am Flight 1736) collided on the runway of a small Canary Island airport during a failed takeoff maneuver, killing 583 passengers and crew members—the most fatal disaster in aviation history. A large, international panel of experts was involved in the crash investigation, which implicated errors in communication, rushed and incomplete procedure, and failure to challenge the decision-making of senior KLM Captain van Zanten when it was obvious to crew members that his actions were hasty and ill-informed [11].

In 1976, the National Aviation and Space Agency (NASA) began collecting data on all aviation mishaps via the Aviation Safety and Reporting System (ASRS). Designed in close cooperation with the aviation industry, the ASRS was a voluntary reporting system created to assist in the analysis of aviation catastrophes and near misses [12]. Early reports from the ASRS implicated human error as the root cause of 60–80 % of all aviation incidents [13]. The US General Accounting Office reported that a lack of team coordination, failure to assign tasks, and lack of effective leadership and supervision were a contributing cause in nearly half of fatal accidents between 1983 and 1985 [14]. In 1984, Billings and Reynard published a review of 7 years of data and over 35,000 reports in the ASRS in an attempt to better understand and classify the nature of consequential cockpit errors. The majority of events could be attributed to either human or system factors: at the system level, over 70 % of reported incidents involved faulty information transfer [12]. Although information quality was typically not the culprit, failure to effectively communicate key data to the personnel responsible for tactical decision-making frequently led to erroneous decisions with dire consequences. Reported human factor error included failure of flight crews to effectively manage resources, plan for contingencies, question authority, and request clarification [12]. Whether or not human error resulted in a consequential accident depended largely on the environment in which the error occurred—as an example, the authors point out that an altitude error is more likely to result in an accident when the plane is close to the ground. Furthermore, errors were found to be more consequential if they occurred during periods of high workload or nonroutine operations [12]. In contrast, technical problems including aircraft malfunction and equipment failures were comparatively rare and, in general, had a less significant impact on aircraft accidents. Billings concludes:

The disease or disorder called “human error” causes half of the preventable deaths in both civil and military flying personnel. It is the largest single cause of premature mortality in this population. This disorder needs to be attacked as aggressively and effectively as we have attacked the physiological and medical disorders responsible for the remainder of preventable deaths [12].

The development of pilot training programs targeting human factors can be traced back to a 1980 NASA workshop entitled Resource Management on the Flightdeck [15]. Research presented at this conference explored the human and team behaviors most frequently implicated in air crashes. Key targets of cockpit team performance included interpersonal communication, decision-making, and leadership [16]. The term “cockpit resource management” (CRM) was created to refer to nontechnical (i.e., human resource) cockpit team training that focused on these specific aspects of human and team behavior [17]. CRM training was touted as an important tool to reduce the incidence of consequential aviation mishaps occurring as a result of human error.

Following the NASA workshop, and in response to recommendations made in the wake of the Tenerife investigation, aviation and aerospace moved quickly to develop CRM-based team training programs tailored to suit specific industry needs. Since that time, CRM training has undergone a significant evolution as it became integrated into the fabric of aviation safety culture and cockpit training. First-generation CRM programs mainly used seminar and tabletop exercises to engender elements of effective managerial style and professional behavior in the cockpit. More recent iterations have cut a broader and more inclusive swath, using realistic flight simulators to address human factor training in the context of simulated missions [16]. Third and fourth generations saw simulation-based line-oriented flight training (LOFT) become a required component of pilot training and expand its scope to include flight technicians, attendants, and mechanics in an attempt to engender common attitudes towards safety across disciplines [16]. Although each generation has had demonstrated successes in influencing flight crew’s attitudes towards human factors, LOFT training has had the most robust impact on achieving buy-in from participants by producing demonstrable changes in behaviors observed during formal Line Operational Evaluations (LOE) of crews in full mission simulation [17]. The Federal Aviation Administration (FAA) currently requires airlines to provide CRM and LOFT training for all flight crews, while still allowing for carriers to maintain a degree of flexibility in training under the Advanced Qualifications Program (AQP) [1, 18].

A common thread spanning generations of CRM training is the notion that its primary goal should be to reduce the incidence of consequential human error through focused instruction on effective communication, leadership, resource utilization, problem-solving, and situational awareness. Helmreich, Merritt, and Wilhelm have argued that even when these specific behaviors are being taught, the link between human factors training and error management needs to be explicitly drawn, such that CRM is regarded as a series of countermeasures with three lines of defense: the avoidance, capturing, and mitigation of error and its consequences [16]. In this framework, error is regarded as “ubiquitous and inevitable,” and instruction is refocused on the natural limitations of human ability, the nature of cognitive errors, and the effects of stress, fatigue, and work overload on team performance during both routine and crisis situations.

Proponents of CRM have pointed to a gradual but distinct decrease in airline incidents and fatalities occurring as a result of human error over the past 25 years as evidence of the effectiveness of formal team training to improve safety in aviation. However, aviation incidents are extremely rare events, and drawing specific conclusions about the impact of training in the accident rate per million flights is problematic [16]. Proposed surrogates have included direct observation crew attitudes and behaviors using formal LOE protocols. In the absence of a criterion standard for evaluating performance, it is not possible to say with certainty that CRM training, rather than advances in technology or “smarter planes,” can account for the bulk of safety improvements. Nevertheless, based on the strong face validity of CRM principles, the majority of aviation crews believe that CRM training has had a significant impact on flight safety [19].

In 2001, Salas, Burke, Bowers, and Wilson published a review of 58 reports of CRM training in search of evidence for its effectiveness in preventing consequential error in commercial and military aviation [13]. Drawing extensively from the framework for evaluating training programs developed by Kirkpatrick [20], Salas and colleagues argue that while there is reasonable evidence that aviators enjoy CRM training and learn about CRM as a result of LOFT and other programs, evidence of a demonstrable impact on outcomes measured in terms of organizational outcomes is nearly nonexistent. Although simulation-based LOFT and LOE do generally demonstrate that CRM training has an impact on behavior, the link between improved cockpit dynamics in a simulated environment and safety outcomes in the real world remains by in large theoretical. The most powerful reports are those that assess the impact of CRM training on multiple levels and include higher levels of evidence (i.e., influences behaviors and outcomes) in their analysis—not surprisingly, there is a paucity of such multilevel reports in the literature, yet these studies generally offer the most compelling evidence that CRM training can have a broad impact on attitudes, knowledge, behavior, and outcomes [13].

Overall, the weight of evidence suggests that CRM training programs do offer something of value in promoting safety and human factors in aviation, and there is a general belief that human factors training does contribute to “safety in the skies.” However, as Salas and colleagues point out, larger and more rigorous multilevel studies are needed in order to accurately assess the impact of team training on aviation safety.