Flight to the Future: Human Factors in Air Traffic Control (1997)

Teamwork among controllers and between controllers and pilots is critically important for safe and efficient air traffic control. The FAA, however, has generally considered the controller function to be an individual one and has therefore not focused on the teamwork aspects of controller tasks, selection, training, or performance appraisal. This chapter discusses teamwork and associated communications, supplementing the relatively meager literature on air traffic control with studies of cockpit teamwork, because pilots are part of the air traffic control team and because information pertaining to flight deck teamwork leads to promising hypotheses applicable to the study of air traffic control.

The definition of team we use in considering air traffic control functions is a broad one that includes individuals who are interacting face to face, by voice, or by written or graphic media to manage air traffic. The size of teams in air traffic control is variable. In addition to the primary actors (one or more pilots and a communicating controller), teams often use additional controllers sharing functions in a sector and may also include supervisors and instructors conducting on-the-job training. Controllers also interact with other controller teams to coordinate the management of flights. It is a characteristic of teams in this environment that they must interact with technology (i.e., radio, radar, computers) to do their jobs. Air traffic control teams are also faced with the responsibility of handling multiple tasks under time pressure at the group as well as the individual level (Waller, 1995).

Teams' functions in air traffic control are both transitory (the interaction between a controller and an aircraft) and relatively enduring (controllers sharing

FIGURE 7.1 A model of team performance in air traffic control. Source: Adapted from Helmreich and Foushee (1993).

functions in the same sector on the same shift). Teams, however composed and however enduring, do not function in isolation but within an organizational and environmental system that profoundly influences their behaviors. It may be useful to consider controller teams in the context of a conceptual model of controller performance. Figure 7.1 outlines input, process, and output factors adapted from a model of flight crew performance developed by Helmreich and Foushee (1993). Input factors that precede team interactions range from individual to organizational to environmental. Individuals bring to the team task their physical condition (for example, fatigue level), technical competence (influenced by the nature and quality of training), and experience. Teams also vary in their composition (the mix of individuals) and compatibility, and all of these factors influence group processes and performance outcomes.

At a broader level, the behavior of teams is influenced by the regulatory environment, including doctrine and procedures, and the organizational culture, including norms and climate (see Chapter 8). The physical environment includes the reliability and usability of equipment as well as external factors such as weather and traffic levels. Because of the varying nature of the airspace to be managed and the organizational structure of the FAA, particular facilities tend to have distinctive subcultures. The existence of multiple subcultures makes it difficult to generalize about team behavior in air traffic control.

Group processes include both technical activities, such as navigation, aircraft

separation, flow management, traffic situation monitoring, and response to user requests (Danaher, 1980). They also include interpersonal activities, such as the formation and psychological maintenance of a team concept and communications activities to maintain situation awareness and decision making at the group level. Group processes lead to multiple outcomes, including safety and the avoidance of operational error, the efficiency of traffic management, and work attitudes and morale. The model is recursive, in that process variables can influence input attitudes and outcomes and will affect both future input and process factors.

Research into air traffic team issues needs to take into account the multiple input and group process factors specified in the model to maximize the generality and validity of findings. Unfortunately, because of the variability of these factors across facilities, few, if any, investigations can control or measure all of them. These limitations need to be assessed when considering the implications of studies of controller team performance.

In this chapter, we describe team performance and interpersonal communications in the aviation system and illustrate them through their role in selected accidents and incidents. Relevant research into communications and teamwork in both air traffic control and on the flight deck is reviewed. Efforts to improve team performance through formal training programs in both air traffic and on the flight deck are described and evaluated. Finally, the implications of several types of automation for air traffic control teamwork and communications are discussed.

Under the definition of team we use here, any errors that involve interpersonal communication are classified as team-related. For example, failures in the transmission and receipt of clearances and separation errors associated with increased or decreased workload can be considered as team rather than individual cognitive or workload issues because they relate to the interface between humans. This is not just an academic distinction, because whether an error is classified as an individual rather than a team or system failure has implications for an organization's response strategy regarding sanctions, retraining, and work design.

Because the FAA has generally considered the controller function to be an individual one, in terms of skills and accountability, strategies for error reduction (aside from punishment of individuals found responsible) tend to focus on technological innovations such as automation (e.g., Helmreich and Schaefer, 1994). This is in common with practices in many technical endeavors, such as aviation and medicine. The logic implies that, if humans commit errors, their removal from the system should eliminate these errors. One result of this philosophy of error reduction has been minimal efforts to address team issues either in training or in work design. Recently, however, team aspects of error and superior performance

have been recognized, and new team training efforts (discussed below) have been initiated in air traffic control.

The nature of team communication breakdowns can be illustrated by describing several accidents in which the controller-pilot interface was identified as a contributory element, either positive or negative. Research into input and group process factors associated with incidents also contributes to an understanding of the team role.

The 1972 crash of a wide-bodied jet transport in the Florida Everglades provided an early look at flawed communication between an air traffic controller and a flight crew as well as the impact of distractions on flight crew performance (National Transportation Safety Board, 1973). In this accident, the cockpit crew became distracted from primary flying and monitoring duties while investigating a landing gear warning light. During this period, the autopilot was inadvertently disengaged and the aircraft began a gradual descent from its intended altitude. The controller on duty did not warn the crew in any way of its impending flight into terrain—but merely asked ''how are things going out there?" In the cockpit there was a failure to maintain active monitoring of flight controls and in air traffic control there was a failure to share situation awareness that could have prevented the accident.

In contrast with the preceding accident, at Sioux City in 1989, the handling of a DC-10 that lost all hydraulic systems and flight controls due to the catastrophic failure of an engine was exemplary (National Transportation Safety Board, 1990). During the in-flight emergency, the flight crew worked effectively with controllers to select an alternate airport and to mobilize emergency units prior to the attempted landing. In this emergency, there was appropriate exchange of information both within the flight deck and between the pilots and the air traffic controllers, and this was combined with sensitivity to workload issues and emotional support needs (Predmore, 1991). The flight crew attributed their successful management of the emergency to formal training in interpersonal human factors known as crew resource management.

In 1990 a B-707, en route from Medellin, Colombia, crashed near New

York's John F. Kennedy Airport after total fuel exhaustion. The accident followed repeated holds during the flight from South America (National Transportation Safety Board, 1991). Although flight crew coordination and decision making were egregious and were clearly major causal factors, the interactions with air traffic control suggest that cultural patterns in communication may have influenced the crew's behavior with disastrous consequences (Helmreich, 1994). Individuals from highly hierarchical cultures that exhibit high power distance (i.e., high relative difference in power at successive ranks) normatively avoid questioning the actions of superiors (Hofstede, 1980). Demonstrating this style of interaction, the Colombian flight crew maintained a subordinate-to-superior relationship with the controllers and unnecessarily accepted multiple holding patterns and, despite being in an extreme emergency just prior to the crash, accepted instructions that delayed their return to the airport after a missed approach. In addition, the crew was indirect in communicating the urgency of its fuel state to the controllers. This behavior is consistent with Colombian society's high power distance (Hofstede, 1980) and was mirrored within the aircraft in the interactions between subordinate crew members and the captain (Helmreich, 1994). Had the crew declared an emergency or refused a clearance, it could have received immediate assistance with either landing at JFK or diverting to an alternate field.

Controllers have no regulatory requirement to recognize cultural differences in communications, particularly when the crew fails to adhere to standard procedures. However, the accident illustrates the complexity of communications across cultural boundaries and raises concerns about future communications problems in an increasingly global aviation system.

One approach to the reduction of separation errors and midair collisions in terminal areas has been the introduction of an airborne computerized traffic alert and collision avoidance system (TCAS, discussed further in Chapter 12). Despite its important alerting function, TCAS also changes the nature of interactions between controllers and pilots, since the controller is not privy to the cockpit information provided by the system. A particular incident involved a Boeing 737-200 transport plane and a light twin-engine turboprop in a terminal area (S.G. Jones, University of Texas at Austin, personal communication, 1995). The heavily loaded 737 took off on a standard instrument departure (SID) and was cleared by air traffic control to climb to 15,000 feet. At an altitude of 5,500 feet, TCAS provided an aural alert and the 737 crew saw the smaller aircraft ahead at the same altitude and a range of about two miles. Just after the TCAS alert, the controller reported the traffic and instructed the 737 to "maintain visual separation." Immediately thereafter, TCAS issued the commands "Descend, descend now." In accordance with company policy, the 737 commenced a descent and the

small aircraft passed approximately 200 feet above the transport. The incident was initiated by an error by the controller, who failed to note the potential conflict between the transport and the light aircraft. However, the critical team issue rests in the fact that the controller did not have access to the TCAS actions and alerts and hence did not share the mental model of the pilots in the aircraft. In this case, the crew was obligated by policy to obey the command that led to a sudden departure from the assigned flight path—a maneuver that was unexpected by the controller. In terms of automation, the TCAS example illustrates how the actions of an automated system can constitute critical items of information required by all team members who must share situation awareness, suggesting that increased verbal interaction may be needed in some cases to share this information.

Many air traffic control positions are staffed by two controllers who work together, with one handling radar monitoring and communications tasks (R-side position) and the other dealing with flight data (D-side position). Thus a ground-based team manages the aircraft under its control, but a single individual usually communicates with the team's air traffic. This work design not only divides the task but also provides redundancy in the form of additional eyes and ears to maintain situation awareness. However, under low traffic conditions, supervisors frequently elect to increase the efficiency of resource utilization and to combat boredom by combining these duties, thus turning a team activity into an individual one. Although this practice does reduce staffing requirements, it also de-emphasizes the team function and can lead to inconsistency in defining duties at a particular position. It is noteworthy that a substantial proportion of operational errors occurs during periods of low workload, when such combined activities are in effect. Investigating changes in task and team resource allocation in a sample of 142 Canadian operational irregularities, Stager and Hameluck (1990) found that more than one-fourth occurred at combined R-side and D-side positions. The authors suggested that changing a team task into an individual one may alter the perception and organization of the task. One of the consequences of this kind of change can be a reduction in situation awareness. There is a need for study of the relationships among workload, teamwork, situational awareness, and operational errors using data obtained at American air traffic control facilities.

Most research on air traffic control communications has been conducted with an individual focus, centered less on the interpersonal component of communication than on its content and form (e.g., Cardosi, 1993; Kinney et al., 1977; Morrow et al., 1993; Nadler et al., 1993). Other research has focused on individual

demographic factors such as experience, position, and rated effectiveness (Human Technology, 1991). These lines of research, although valuable, do not deal directly with the team issues associated with controller performance.

Seamster and colleagues (1993) conducted an experiment involving the simulation of operational problems that allowed analysis of performance effectiveness in a controlled context. These investigators grouped subjects into pairs to encourage collaboration on problem solution sets. In some scenarios, the sheer volume of aircraft transmissions prevented controllers from verbalizing their strategies. Experienced controllers simplified these situations and reduced monitoring loads by managing their workload early. Teams that most efficiently handled high workloads did so through the use of situational inquiries, frequent observations, and statements of intent, along with direct responses to queries. Task prioritization, workload management, and contingency planning were most effective if conducted during low workload periods, before periods of high traffic began. Seamster et al.'s (1993) research is important in its demonstration that safe and efficient traffic management requires the practice of effective teamwork skills as well as cognitive ones.

As noted, despite the complex team nature of air traffic control, its organizational culture and procedures have not historically stressed the team aspects of the controller job. Accordingly, team issues are seldom addressed in training or evaluation. The fact that on-the-job training constitutes the major means of socializing and qualifying new controllers as well as maintaining job competence exacerbates this problem. On-the-job instructors are fully qualified in the technical aspects of the controller function, but they do not receive systematic instruction in how to evaluate, instruct in, or reinforce communications and team skills.

Research on the determinants of operational errors in the Southwest Region of the FAA is being conducted by Jones (1993, 1995), as part of an initiative to reduce and contain human errors known as ASSET (Air Safety System Enhancement Team). One of the primary elements of the research is a survey completed without jeopardy by individuals involved in incidents. The survey elicits information on factors surrounding the event.

Three team-related scales were derived from behavioral questions on the survey. These include elements of task management (i.e., planning and workload distribution), information exchange (i.e., description of situation factors and inventions), and interpersonal relations (i.e., interpersonal sensitivity and receptivity). These team scales reflect the concepts included in formal training programs known as crew resource management.

Preliminary analyses contrasted scale scores for operational mishaps with normative and exemplary data from days without incidents. The data suggest that team issues are critical factors in operational errors. There were significant

differences between incidents and normative and exemplary conditions on all three team behavior scales, with more positive scores associated with the absence of mishaps.

Data addressing these issues collected across a broad array of air traffic control facilities should prove useful for the determination of critical issues in team training. The data also suggest the potential value of formal programs to train controllers and their supervisors in effective teamwork and communications strategies. It also emphasizes the extent to which effective workload management is as much a team function (knowing when and how to shift task responsibility from one member to another) as it is an individual function, as described in Chapter 6.

Attitudinal data regarding flight deck management and crew coordination have been collected using a survey that measures the level of endorsement of CRM concepts, the cockpit management attitudes questionnaire (Helmreich, 1984; Gregorich et al., 1990). Data from this questionnaire have been used to isolate training needs and to measure the impact of CRM training (e.g., Helmreich and Foushee, 1993; Helmreich and Wilhelm, 1991). Preliminary data were collected in several air traffic control facilities using an adaptation of the questionnaire and its extension, the flight management attitudes questionnaire (Helmreich and Foushee, 1993). More than 500 controllers from 3 en route facilities and 1 TRACON completed the survey as modified for each environment. Although it would be inappropriate to generalize to the system from this limited sample, the data do demonstrate that the concepts captured in aviation can be reliably measured in the air traffic control environment (Sherman, 1992). Scales that parallel those isolated among flight crews were identified in factor analyses.

The data suggest that, among those queried, there is general acceptance of the importance of team coordination and open communications. However, attitudes regarding leadership responsibilities and the need for interactive leadership differed from those of U.S. pilots. Two items on the revised survey follow a description of four leadership styles: democratic, consultative, directive, and autocratic. These questions ask respondents to identify preferred and experienced leadership style. The majority of controllers would prefer a consultative leadership style, in which leaders seek the opinions of subordinates. However, nearly 60 percent in one facility at which these questions were asked reported experiencing autocratic leadership offering little consultation and explanation for actions. Only 11 percent reported working under consultative leadership.

Scores were also quite low on a scale formed of items showing recognition of the negative effects of stressors on performance, such as fatigue, personal problems, and inexperienced coworkers. Many controllers feel that their decision-making ability is as effective in emergencies as under normal conditions.

Similar denial of vulnerability to stress effects has been found in pilots and medical personnel (Helmreich and Foushee, 1993; Helmreich and Schaefer, 1994). The perception of being "bulletproof" when faced with stressors can result in a failure to use teamwork as a countermeasure to stress.

Another scale, derived in the controller sample, contains items reflecting a willingness to question and disagree with the actions and decisions of others. Not surprisingly, junior (developmental) controllers reported more reluctance to speak up. The results suggest that integrated human factors training with a special focus on leadership and stress issues could have the same beneficial effects measured among flight crews, at least for the groups surveyed.

Because of the paucity of research and empirical data regarding team performance and team training specific to air traffic control, experiences in a related domain, commercial aviation, are discussed in the next section before reviewing the steps that have been taken to introduce team training in air traffic control.

By the late 1970s, research had demonstrated that human error was associated with the majority of accidents and incidents in commercial aviation (Cooper et al., 1980; Murphy, 1980). The same data also indicated that these human failures tended to be in leadership and team communication and coordination rather than technical aspects of flight control. This has implications when deciding where new systems should focus their human support. These findings have remained robust overtime in their implication of interpersonal team factors as critical determinants of aviation safety (Helmreich and Foushee, 1993; Helmreich et al., 1993), and hence their implications for the necessity of preserving or enhancing team communication functions in system design changes.

Commercial and military aviation responded to these data, in cooperation with the National Aeronautics and Space Administration, by undertaking the development and evaluation of training programs in interpersonal human factors that were known initially as crew resource management (CRM) training. The meaning of the CRM acronym has subsequently changed to the broader designation of cockpit resource management to reflect the fact that team issues extend beyond the cockpit to include interfaces with air traffic control, cabin, dispatch, and ground operations (Helmreich and Foushee, 1993). Considerable empirical data have accumulated over the last decade indicating that CRM training can and does change attitudes and behavior among flight crews and that these changes increase the margin of safety in flight operations (Diehl, 1993; Helmreich and Foushee, 1993; Helmreich and Wilhelm, 1991). However, the data also indicate that some programs have greater impact than others and that a variety of causal factors determine behavioral outcomes and overall crew effectiveness (Helmreich and Foushee, 1993; Taggart, 1993, 1994). A partial listing of factors that influence the acceptance and practice of the concepts provided in this training is

relevant to consideration of team issues in air traffic control. In programs with high positive impact:

1. The organizational culture is supportive of human factors concepts and training.

2. Senior management demonstrates its strong support for the program.

3. The program is supported by unions as well as management.

4. Critical role models (instructors and evaluators) and managers practice and reinforce effective team communication and coordination. It is especially important that the concepts taught are evaluated and encouraged under operational conditions and are not expressed as abstract concepts. The failure of many traditional management development training programs (including total quality management efforts) to influence day-to-day behavior comes, at least in part, from a lack of connection with the mundane realities of individuals' jobs. Human factors training is operationally based in the domain, rather than imported and conceptual. Training needs to reflect the organizational culture and to be rooted in operational behavior and situated learning rather than addressed as vague psychological constructs, such as leadership and open communication. Many organizations survey flight crews before developing training to determine critical issues for training, and again following training to measure impact using standard measures of attitudes relating to flight deck management (e.g., Helmreich, 1984; Helmreich and Foushee, 1993; Helmreich and Wilhelm, 1991).

5. Instructors and evaluators receive special training in the evaluation and reinforcement of team concepts. Clearly, if the critical role models in an organization cannot evaluate and reinforce the concepts, the probability is low that they will become embedded in the organizational culture.

6. Human factors training is experiential rather than didactic. Trainees need to practice and experience the concepts being communicated rather than to receive lectures regarding effective behavioral practices.

7. Nonjeopardy simulation is provided to allow team members to practice concepts and receive feedback. Military and commercial aviation has embraced the concept of line oriented flight training (LOFT), in which full mission simulations are conducted under highly realistic conditions to allow crew members to practice concepts without threat to their licenses (Butler, 1993; Federal Aviation Administration, 1978).

8. Human factors data are collected in incidents and operational errors to provide empirical data for training development and evaluation. Instances of both deficient and highly effective team performance provide the most relevant material for both initial and recurrent training. By utilizing reality-based incidents with which participants can identify, the probability of acceptance is greatly enhanced.

The FAA has strongly endorsed CRM training for pilots. A total of airlines,

including most of the major carriers, will soon be regulated by a new special federal aviation regulation that allows training innovations and requires both recurring CRM training and LOFT (Federal Aviation Administration, 1990). The FAA has also issued a revised advisory circular that emphasizes many of the points listed above (Federal Aviation Administration, 1992). More recently, the FAA has issued a notice of proposed rule making that will require CRM training for all pilots covered by the Code of Federal Regulations, Volume 14, Parts 121 and 135 operations (applicable to most scheduled air carriers) and will extend the training to flight attendants and dispatchers. The National Transportation Safety Board (1994), in its investigation of the crash of an FAA aircraft, noted that the FAA does not provide CRM training for its own pilots and recommended its implementation.

As CRM training in aviation became widely implemented and enthusiastically accepted, working controllers at several facilities concluded that the issues involved were highly relevant to their duties and operational problems. With the cooperation of airlines having CRM programs in effect, grass-roots training programs were implemented beginning in 1988 at several facilities, including Seattle and Chicago. The locally developed programs were known as controller awareness and resource training (CART). These initiatives were supported by the controllers' union, the National Air Traffic Controllers Association, were well received by participants, and received some support from FAA management. However, these programs were clearly derivative adaptations of airline training, with most examples and exercises focused on cockpit rather than air traffic control issues. The program continued informally until 1993. During its existence, its implementation at a facility was entirely at the discretion of facility management and the controllers' union. Informally, it was widely recognized that many of the facilities with serious human factors problems were most resistant to this type of initiative.

Senior FAA officials in air traffic also became aware of the increasing growth and impact of airline CRM as well as the development of local programs for air traffic controllers. To signal organizational commitment to a system-wide human factors training program, a conference was held in October 1991 in Austin, Texas, to foster the exchange of information among research and operational personnel involved with major airline CRM programs, senior air traffic control management, members of the controllers' union, and facility managers. Following this meeting, in 1992, a steering committee was formed to develop a training program more focused on air traffic control, and a contractor was engaged to develop a national program for controllers in all facilities. This program is known as air traffic teamwork enhancement (ATTE). During 1992 and 1993, 150 workshop leaders (facilitators) were trained.

Following deliberations of a cross-sectional committee composed of labor and management, a revised curriculum for the program was approved in late 1994. Additional facilitators, who are working controllers, were given ATTE facilitator training during the first half of 1995 and nationwide implementation is under way. It is important to note, however, that the program is not mandatory, and the training costs are not a budget item. Funding for ATTE training must be provided by individual facilities. At present there are no data on the percentage of facilities that have initiated ATTE or the number of controllers who have completed the workshop.

The design of the curriculum for ATTE makes it a conventional basic awareness CRM program as defined in an advisory circular (Federal Aviation Administration, 1992). It is implemented as a three-day workshop to be attended by 6–20 controller participants. The curriculum is designed to be presented by two facilitators. The training also uses the same approach adopted by many airlines in having the material presented by facilitators who are working controllers rather than members of management or training professionals. The training strategy is participative rather than didactic. Its goal is to use exercises and experiences to demonstrate the importance of the concepts being presented. In these respects, ATTE complies with the FAA's recommendations for initial CRM training.

The manual for facilitators introduces the concept by pointing out that research into air traffic control incidents has concluded that approximately 70 percent involve human error and that more errors occur when sectors are staffed individually rather than by teams. Thus the framework is set for emphasizing the importance of teams and teamwork. Six major topics are included in the curriculum:

1. Understanding air traffic teamwork. This module has the expressed goals of identifying resources available to controllers, demonstrating how they can make better use of resources, and discussing the characteristics of effective teams.

2. Communicating with others. This module focuses on teaching skills for communicating effectively and providing feedback. It also discusses barriers to communication and provides practice in applying communications skills.

3. Being a resource. In this module, characteristics of controllers who are valuable resources to their teams are introduced and the importance of speaking up (assertiveness) is discussed.

4. Managing stress. This segment discusses current stress levels among controllers and the relationships among stress, health, and performance. Methods of reducing stress are discussed.

5. Managing conflict. The conflict module discusses how attitudes and values influence ways of dealing with conflict and the outcomes of destructive

and constructive conflict. Styles of conflict management and techniques for effective resolution of interpersonal disputes are also included. In the future, if aircraft become more autonomous in the development and execution of flight plans (for example, free flight navigation using global positioning satellites), the controller will need skills in negotiation, as will pilots, in order to establish optimal, safe routing. Negotiating skills are clearly related to conflict management, and training in this area can be made an extension of the present.

6. Summary. Insights and learning from the workshop are summarized and the session is evaluated.

One of the strengths of the program is its attempt to make the experience relevant to the domains of air traffic control. However, the syllabus does not include material on systems issues that may impede team coordination and performance, issues surrounding the interface between controllers and supervisors, or performance evaluation techniques. Although admirable in intent, the ATTE program is lacking in several of the factors discussed earlier that have been shown to influence the success of CRM programs. Specifically, the following concerns can be raised about ATTE:

1. Program development took place in the absence of empirical data regarding controller attitudes and incidents occurring in air traffic control facilities. Facilitators are charged with determining critical issues in their facilities and adapting the curriculum to reflect them. This is an extreme demand to place on individuals who are neither professional researchers nor professional educators.

2. The program does not demonstrate organizational commitment to the concepts by being budgeted and mandated at the national level and integrated into ongoing training and evaluation activities. Indeed, part of the job of the facilitator is to sell the program within his or her facility.

3. The use of peer facilitators has long been practiced in aviation under the working assumption that peers will be the most credible communicators of the nontechnical concepts associated with CRM and will be less threatening to participants. These ideas were certainly relevant to the climate of suspicion regarding psychological training that surrounded the introduction of CRM in the early 1980s. However, the situation has changed dramatically, and it is not uncommon to have the provision of CRM training made part of union contract demands. The unintended negative consequence of using peer facilitators is to dissociate CRM from both formal training and evaluation. In other words, it becomes seen as a training event but not as part of the culture, the "way we do business day to day."

4. Formal evaluation of the impact of ATTE has not been initiated using both behavioral and attitudinal outcome measures, nor is such validation planned at this time.

5. No additional training has been developed for managers, on-the-job training instructors, or evaluators to enable them to provide effective debriefing and reinforcement of human factors behaviors. With the use of peer facilitators, the formal training and evaluation structure is bypassed and leadership may relate negatively to the program as a result of being excluded.

6. The training is designed as a single-event program without provision for annual recurrent training. Effective programs provide updated annual training that reflects areas of concern isolated from incidents or research.

7. The program does not include nonjeopardy simulation to allow realistic practice of behaviors or to receive feedback on performance. Also, it does not include simulations that reproduce the team environment of facilities.

The strategies that have made CRM programs effective in air carrier operations would seem to apply directly to air traffic control programs. If this assumption is correct, it is unlikely that these programs will achieve their potential unless they are carefully tailored to air traffic control, and unless improved air traffic control programs reflect the concerns mentioned above.

Because team-centered simulation is at the heart of air carrier CRM programs, it is important to discuss the use of simulators in air traffic control training. Simulation, at least at radar positions, tends to concentrate on the ability of controllers to manage extremely high-density traffic as individuals. An important adjunct to basic human factors training would be simulation that includes coordination among positions and also includes supervisory personnel as active participants. Effective full mission simulation (LOFT), as practiced by air carriers, also involves structured human factors briefings and debriefings to reinforce the concepts practiced.

Extrapolating from air carrier experience, one of the most significant enhancements the FAA could make in its use of simulation would be to initiate team-oriented training utilizing scenarios that involve interactions with supervisors, other sectors, and on-the-job training. However, the agency is faced with a dilemma in implementing standardized simulation training on a system-wide basis because of the idiosyncratic nature of operations at the various facilities. Although the concepts involved are general, the specific cultures at different facilities may dictate differing emphases in training.

As the FAA introduces automated systems into the air traffic system, it is essential that the effects of such innovation on teamwork and interpersonal communications be addressed during the design phase. In this section, some of the

consequences of an already introduced system are discussed, along with the possible behavioral effects of two other systems.

Historically, it appears that the earliest forms of air traffic control automation, the ARTS and HOST computer systems (discussed further in Chapter 12) that provide more detailed information on target identification and position may have had the unintended consequence of reducing teamwork and acceptance of the importance of team coordination. Before the displays were automated, members of the team had to rely on verbal information transfer to ensure that both controllers and pilots maintained situation awareness. With the additional information provided by automation, understanding of the benefits of sharing mental models may have become lost, fostering the more individualistic role definition found today.

The near-midair collision discussed earlier illustrates unintended consequences of automation. TCAS provides flight crews with a visual display of traffic as well as aural warnings and commands. The warnings and commands it gives are not available to the controller handling the flight. Because of this discrepancy in information, controllers and pilots may not have the same mental model of the situation and the controller may not know in advance that an aircraft will deviate from an assigned altitude or heading. Thus it is possible, as in the example, for conflicting instructions to be issued. Although TCAS is proving to be a valuable tool for collision avoidance, it adds uncertainty to the controller role and may reduce the level of teamwork achieved.

Datalink is designed to provide electronic exchange of information between aircraft and controllers. Providing visual rather than aural information has the potential for reducing misunderstanding of clearances. However, it also may reduce the amount of information available to flight crews, often relayed via nonverbal cues in voice communications, and may have a deleterious effect on the development of teamwork between controllers and flight crews. Furthermore, one of the ancillary benefits of verbal communication between controllers and multiple aircraft on the same frequency is a great deal of information, albeit sometimes ambiguous, about conditions and traffic (Pritchett and Hansman, 1993). The party-line aspect of air traffic control communications provides information on traffic flow, weather, etc. For example, in the Avianca accident described earlier, there was a great deal of information regarding diversions and holding that, had it been processed, could have helped the Colombian crew decide to divert or declare an emergency prior to running out of fuel. The absence of verbal interaction between controller and aircraft may also make it harder to establish an effective team relationship under conditions in which datalink is not working or during emergency conditions when joint decision making is required.

Attitude surveys of flight crew members' reactions to automation have shown wide variability in their liking for automation, in the recognition that team communications requirements are changed by automation (the presence of an ''electronic

team member"), in perceptions of freedom to adjust the level of automation employed, and in concerns that the use of automation may degrade operational skills (Sherman and Helmreich, in press). These suggest a need to communicate organizational philosophies of automation to personnel and further the need for formal training in the team as well as technical aspects of automation use. These cockpit issues are likely to be mirrored for controllers as automation is increased in the air traffic control environment.

Teamwork, reflected in verbal communication among controllers and their supervisors and between controllers and flight crews, is likely to be a critical component of air traffic control for the foreseeable future. As in other technological endeavors, a high percentage of operational errors involves breakdowns in communications, coordination, and group decision making. Crew resource management training has proved to be effective in improving team coordination in flight crews and is being mandated on a worldwide basis. Similar training for air traffic controllers and their supervisors and trainers has the potential to provide similar enhancement of teamwork. This potential will only be realized if the necessary commitment by and support from FAA management becomes evident.

The automation of components of the air traffic system may influence team interactions and can, in some circumstances, have a negative effect on teamwork and the ability of controllers to maintain situation awareness. The panel has identified a number of approaches to improving team coordination and communication in the air traffic control system:

1. Making team issues a part of the organizational culture of the air traffic system by defining the nature of team coordination as part of the organization's task description. It is important to include evaluation of team as well as individual skills as part of performance assessment.

2. Focusing on team as well as individual factors in the investigation of operational errors in the air traffic control system.

3. Make team training a centrally funded program required at all air traffic control facilities.

4. Using empirical data, including analysis of team issues in operational errors, survey data on controller attitudes regarding team issues, behavioral measures of team performance in simulation, and participant evaluation of training programs to refine training programs, to ensure that critical issues are addressed in the curriculum, and to measure training impact.

5. Including interface issues (controller to supervisor) as well as controller to air crew as part of team training.

6. Providing additional training in human factors issues should be provided

for supervisors and on-the-job training instructors to allow them to evaluate team performance and reinforce effective behavior.

7. Providing recurrent training in team human factors and using team-oriented simulation as part of training.

8. Evaluating the impact of automation components on interpersonal communication and team performance before adopting systems.

9. Including team-related automation issues in team training.