Teams Embedded in Organizations Some Implications Daniel R. Ilgen Michigan State University

From the early 1980s to the present, organizations have increased their reliance on teams. Over the same time period, the amount of traditional small group-team research has decreased. The void has been filled by research that is focused on issues facing task-performing teams embedded in organizations. It is argued here that the organizational perspective has produced research that complements and extends past research on groups and teams. A case is made for differences in orientation between traditional team research and work aimed specifically at teams in organizations. Implications of these differences are explored. [Teams] are alive and well, but living elsewhere. —Levine and Moreland, 1990, p. 620

J

ohn Levine and Richard Moreland (1990) reached the above conclusion after reviewing small group research for the decade of the 1980s. Their upbeat conclusion regarding the health of team research was in stark contrast to that of Ivan Steiner (1986), who felt his high hopes expressed in the early 1970s (Steiner, 1974) were unfulfilled. He saw a decrease in the volume of work and felt much that did exist was built on a foundation of weak theory and on data restricted by methods too limited to meet the challenges presented to them (Steiner, 1986). In contrast to the divergent views on viability, they concurred on location. Steiner (1986) saw small groups as too much an integral part of human existence to be neglected forever. He warned that if those traditionally concerned with small groups and teams abandoned them, others would turn their attention to them. Others did. McGrath (1997) identified these others as those working in organizational behavior, speech communications, political decision making, group therapy, family studies, and education. Sundstrom, DeMeuse, and Futrell (1990) highlighted the growth of work on self-managed work teams, and Stout, Salas, and Fowlkes (1997) recently identified research on team training as an extremely active topic in the 1990s. All of self-managed work teams and most of team training falls within the organizational behavior domain mentioned by McGrath (1997). In the present article, I also focus on the same general domain—that of the behavior of teams in work organizations—and address what this domain of team

February 1999 • American Psychologist Copyright 1999 by the American Psychological Association, Inc. 00O3-066X/99/$2.0O Vol. 54. No. 2, 129-139

research has added to our understanding of team functioning.1 The 1980s saw heightened interest in teams embedded in public and private production or service organizations. A number of factors, both negative and positive, contributed to the interest. The crash of an Air Florida passenger airline taking off from National Airport in Washington, DC was attributed to crew members' lack of experience with winter flying conditions. Command and control team errors were blamed for two U.S. Armed Forces disasters in which aircraft were mistakenly fired upon and lives were lost. In one case, an Iranian commercial airline was shot down over the Persian Gulf. In another, a U.S. military helicopter was shot down by its own air force. Fortunately calamity did not drive most of the increased interest in teams. Major changes were and are occurring in the way work is structured. One of the most critical changes is the reengineering of work around teams

Editor's note.

William C. Howell served as action editor for this article.

Author's note. Some parts of this article were presented as part of a keynote address at the First Biannual meeting of the Australian Conference on Applied Psychology in Sydney (July 1995), and other parts were presented at a workshop at the Kurt Lewin Institute, University of Amsterdam, Amsterdam, the Netherlands (April 1997). Work on the article was supported in part by grants from the Office of Naval Research (N00014-93-0983) and the Air Force Office of Scientific Research (F49620-95-1-0314). The support was very much appreciated, but it is recognized that the support does not imply agreement or disagreement by the supporting agencies of the ideas expressed herein. I also gratefully acknowledge the helpful comments of Michael Frese, Beryl Hesketh, John R. Hollenbeck, and Steve Kozlowski. Correspondence concerning this article should be addressed to Daniel R. Ilgen, Department of Psychology and Department of Management, Michigan State University, East Lansing, MI 48824-1117. Electronic mail may be sent to [email protected]. ' In the literature, some authors have made a strong case for a distinction between small groups and teams (e.g., Morgan, Glickman, Woodward, Blaiwes, & Salas, 1986). McGrath and Gruenfeld (1993) make even finer distinctions in the domain typically described as teams. They partition the domain into teams, task forces, and crews. Both taxonomies can be useful. However, in much of the literature, the distinctions between small groups and teams are blurred; what is labeled a team by some is a small group, a task force, or a crew to others. Because it is not the purpose here to make fine definitional distinctions at the margins, I do not differentiate among them; I use the terms teams and groups interchangeably to represent collectives that are composed of two or more individuals who (a) interact, (b) are interdependent, and (c) share some common goal(s) or objective(s).

129

Daniel R. llgen

rather than individuals' jobs (llgen, 1994). Some of this change was fueled by attributions for Japanese success in the marketplace to their use of teams {llgen, Major, HoU lenbeck, & Sego, 1993). Highly publicized organizational redesigns, such as General Motor's structure for Saturn sales and manufacturing also helped to fuel the fires of team implementation. The level of interest is reflected in a number of cover stories in leading news and business magazines in the early 1990s touting the value of teams. Wilh rising expectations, some backsliding was inevitable. The September 5, 1994 title of me cover story of Fortune, a widely circulated weekly business magazine, was "The Trouble With Teams." Although the volume of literature on teams embedded in organizations supports Levine and Moreland's (1990) conclusion that team research is living elsewhere, it says little about the life being led. I will focus on the latter and argue that interest in this subset of teams has (a) shifted the focus in the general model of teams, (b) directed attention toward different team constructs (or altered the emphasis placed on constructs), and (c) increased tolerance for and interest in an expanded set of methods for studying teams. These changes have implications for the nature of theories about (earns that have resulted and the transfer of knowledge about teams to organizational settings.

Models Emphasis McGrath (1984) and Hackman (1987) described traditional small group research in classic systems theory terms with inputs, processes, and outputs arrayed across time. To incorporate dynamic change, feedback loops were added emanating primarily from the ouiput side of the model and feeding back to inputs or process as teams moved from 130

state t0 to ii and beyond. Numerous constructs within each of the three domains were addressed, depending on the theory. However, regardless of the nature of the authors' focus, the input domain usually included task characteristics, some elements of context, and people who composed the teams. Process constructs typically included interactions among members, communication, coordination, and interpersonal influence mechanisms, such as leadership. Outpuis clustered in task-focused (i.e., performance) and socioemotional domains. Although research and theory fit an input-processoutput model, all regions of the mode! did nof get equal billing nor did theory and research progress linearly from left to right. Rather, traditional approaches tended to focus on the development of psychological process theories. Team tasks, contexts, and composition (on the input side) often were of interest only as boundary conditions restricting the actors, behaviors, and contexts over which process theories generalized. Similarly, outputs often played the role of supporting actors with process theory center stage. The dominance of process over inputs and outputs is perhaps best illustrated by the large number of studies on team consensus decision making using the National Aeronautics and Space Administration (NASA) "moon problem task." In this work, teams made consensus decisions, and investigators observed the process of decision making, usually with little or no interest in the task or the task performance of the team. The task and performance on the task simply provided a setting and an activity on which team process could be observed. The process-dominated focus influenced the way theory was constructed and research progressed. Both began wilh process and worked to the outside, selecting input and output constructs after considering their role vis a vis the process theory. Anchoring team concerns in an organizational context shifted the focus in the input-process-output mode! from the center to the right or left (i.e., to outputs or inputs). In organizations, teams' task performance is critical; it is their raison d'etre. On the input side, embedding teams in the larger organizational system demands that they adapt to, model, or be influenced by the organizational environment in which they exist (Ancona, 1987). Also, the centrality of staffing concerns in work organizations forces attention on team composition. On the output side there are also restrictions. Among those who manage organizations, outcomes believed to impact the effectiveness of the team in the organizational context are considered more legitimate than others. Although there exist a large number of outcomes that fit this restriction, the selection of variables to be considered still must pass the "test" of relevance as the theory is constructed. The emphasis on inputs and outcomes not only influences the variables selected for study but also influences how theory develops. Temporally, theory development and research on organizationally relevant team issues is often initiated because of concerns about either input (e.g., context) or output (e.g., performance) issues that are identified as critical. Only then are theories constructed around them. That is. the behavior of individuals in teams and the beFebruary 1999 • American Psychologist

havior of teams themselves are first anchored in the contexts and outcomes. Then process theories are constructed to accommodate the inputs and outputs, constraining the system by the nature of the organizational setting. As a result, the domain of the organizational literature on teams tends to have a more richly articulated input and output space with the theories of team processes developed to support understanding input and output variables critical in organizations. Team process remains critical, but the nature of the process is bounded and grounded by inputs and outputs critical in organizational settings.

Normative Concerns A second difference in orientation is in the willingness to develop normative models. Most all of the earlier research on small groups is descriptive (Hackman, 1987). Research in this tradition seeks to generate knowledge about what actually happens in teams and, through induction, build general laws. It is hoped that these laws will be useful for guiding some practice, but applications are secondary interests. Normative models, in contrast to descriptive ones, usually start with a purpose to develop ways to improve teams so that behavior in them will meet some objective. Hackman's (1987) work on the design of work teams is a clear example of a normatively driven model. He is explicit in the goal to develop a model based on the scientific data available to him that will increase the probability that teams with characteristics outlined in the model will perform better. Once the model is constructed, the task of evaluating its normative value follows. The objective of most research on teams embedded in organizations is not simply to understand team behavior but understand it in a way that will shed light on ways that will increase the effectiveness of teams in organizations. Because all interventions in organizations can be roughly classified as falling into three domains—composition (selection and placement of members on the teams), engineering situations (e.g., job design, role specification, organizational structure, or setting), and training (changing people)—the theories that result from the study of teams in organizations are expected to have implications for one or more of three means of impacting team effectiveness. Theory and research are judged against criteria of validity and criteria of normative contribution. The result is often a mix of normative and descriptive theory, such as is illustrated in Hackman's (1990) edited book on teams in organizations or much of the work done on command and control teams in the military (e.g., Cannon-Bowers & Salas, 1998; Kleinman, Serfaty, & Luh, 1984).

Critical Constructs Performance Teams embedded in organizations exist to perform tasks. How they perform the tasks for which they are responsible is the primary concern of those who design and manage organizations as well as those who study these teams. Although performance was an important topic in traditional team research, it was often not the central concern. As a February 1999 • American Psychologist

result, the team performance construct itself was less systematically addressed. McGrath (1997) identified three schools of thought in small-group research, two of which were very much concerned with performance (those located at the University of Illinois and the University of Michigan). In both, the theory behind performance rather than performance itself was central. Consider, for example, Steiner's (1972) theory described in his classic book, Group Process and Productivity. Without a doubt, performance, termed "productivity," was extremely important. But the complexities of identifying the performance construct across teams and the measurement of performance as a construct received little attention. The focus was on team processes, specifically process loss. The study of teams embedded in organizations is often conducted by those with a strong individual difference view of work, in which the definition and measurement of performance is the focus of attention whenever the behavior of individuals at work is addressed. Developing measures of performance criteria that are valid and reliable and neither deficient nor contaminated is often the starting point of research in work settings. This emphasis appears in much of the research on work teams. It is unlikely that team-performance measures would have advanced as rapidly or as far if teams had not become the focus of attention of those accustomed to addressing criterion development problems. A number of approaches to team performance have surfaced. Four are mentioned here as illustrations of different ways to address team performance. Morgan and his colleagues (Baker & Salas, 1992; Cannon-Bowers, Tannenbaum, Salas, & Volpe, 1995; Mclntyre & Salas, 1995; Morgan, Glickman, Woodard, Blaiwes, & Salas, 1986) followed the classic criterion development model to construct their team evolution and maturation (TEAM) measure of team performance. Subject matter experts composed of team members and those who work with teams were asked to provide critical incidents for good and poor team behaviors. The incidents were then used to construct behavior-expectation scales in the standard fashion (see, e.g., Campbell, Dunnette, Arvey, & Hellervik, 1973; Smith & Kendall, 1963). Studying a sample of teams in a U.S. Navy training exercise, Morgan et al. (1986) identified two parallel performance tracks that develop over time. The task-work track focused on operations-related behaviors relatively idiosyncratic to the tasks to which teams were assigned. The other track, teamwork, included activities that strengthened the relationships, communication, and coordination within teams. Task-work and teamwork behaviors differentiated teams conducting training exercises on effectiveness. The fact that induction from observing teams to a set of behavior dimensions clustered into two sets, one with a major task component and the other more person-focused, is hardly new. Clearly there is convergence with "initiation of structure and consideration" (Stogdill, 1974), "concern for production and concern for people" (Blake & Mouton, 1969), and similar constructs of others who have looked at 131

leadership or leadership functions in teams (Katz & Kahn, 1978; Seashore & Bowers, 1970). Yet, Morgan and his colleagues' (1986) teamwork differed somewhat from the earlier studies. The four subdimensions of teamwork were performance monitoring, feedback, closed-loop communications, and backing-up behaviors. Members of effective teams monitored the performance of other team members. They kept track of how others were doing and stepped in to help out. Monitoring was not "spying." It was an accepted norm for keeping track of other team members. Implicit in monitoring was trusting that others will come to the support of a teammate and believing that he or she should do the same. The next step beyond monitoring was giving and accepting feedback. Members of effective teams did not see feedback as the prerogative of only leaders. Each member saw this activity as part of his or her own role. Implicit in this was the unspoken assumption that each team member needed to become reasonably familiar with every other team member's role. Without such knowledge, meaningful feedback is unlikely. Closed-loop communication was communications with a twist. Senders sent messages. Receivers received those messages and let the senders know that they had received them. Senders also checked to be sure that the communication was received as intended; they did not simply stop with the confirmation that their message was received. Finally, members of effective teams backed up their teammates. Team members were not focused only on their own performance. They had a sense of how other team members were doing, and they were willing to help others when their own tasks demanded less than full attention. Building on the model of Morgan et al. (1986), later work refined the measures of team performance (Baker & Salas, 1992, 1997; Cannon-Bowers, Salas, & Converse, 1993). Although this work modified slightly the specific dimensions and their measures, the general approach of capturing performance dimensions by induction from clustering descriptions of team behaviors scaled in terms of effectiveness units was used. The performance-measurement approach relates to two points raised earlier. With respect to the input-process-output model, the work was initiated as a study of team performance (an output), but key constructs, such as teamwork, that get mapped into performance could very reasonably be considered process. Teamwork was identified, in part, based on the use of a performance-measurement process with a long history of use in industrial and organizational psychology (see Smith & Kendall, 1963) that focuses attention on identifying critical behaviors for effective and ineffective behaviors and scaling the behaviors along an effectiveness dimension. Critical behaviors are the same behaviors that are studied as team process. I do not wish to argue that such behaviors should be considered outcomes rather than process but only to point out that behaviors are identified not from an initial focus on process but by an inductive process that begins with a concern for team effectiveness. In which box the behaviors fall in whose model is not really impor132

tant. The observation is that the nature of the variables identified as important is influenced, in part, by the normative orientation of the investigators. Recall that the initial purpose of the investigators was to identify dimensions of team performance for the purpose of training. As a result, performance behaviors are critical because knowledge about behavior is critical for responding with ways to train team members to be more effective.2 Pritchard's work provides a very different approach to team effectiveness (Pritchard, 1995; Pritchard, Jones, Roth, Stuebing, & Ekeberg, 1988). He capitalized on individual models of performance and mixed theory and practice in a way that makes the two hard to disentangle. His system is known as ProMES for "productivity measurement." Although the system can be applied to individuals, it is usually used at the team level. Key to ProMES is the assumption that team members form stable impressions of their team's performance, which, with help, they can scale along a quasi-ratio scale from low to high. Through a very labor-intensive process, teams identify the outcomes (called key products) they produce and estimate how each product contributes to the unit's effectiveness. Helping teams understand what they produce and how it relates to effectiveness is accomplished by feeding back performance data in the form of contingency graphs with performance in effectiveness units on the ordinal and units of products produced on the abscissa. The slopes of the lines provide a visual picture that shows how units produced relate to performance. From the figures it is usually clear to team members how the team should distribute its time and effort to perform its job effectively. If effectiveness increases sharply as more units of one kind of output are produced by the team and there is little or no change in effectiveness as more of another is produced, team members realize that they should shift attention to the former to improve effectiveness. Central to the ProMES system is giving feedback, applying goal-setting principles, and establishing reward systems for the team. ProMES is a motivationally focused procedure, with both descriptive and normative components.

2 Kozlowski, Gully, Nason, and Smith (in press) take a more radical stance on the process versus outcome distinction with respect to team performance. They argue that, under dynamic conditions where teams perform over time, performance is a process—a process of behaving, thinking, feeling, interacting, and reacting individuals. Yet, to measure performance, time frames in the life of the team are selected, usually arbitrarily (e.g., a year), and activities during the period are summarized as indicators of performance. These same behaviors, if observed in action, would be considered process but when frozen in time and "observed" retrospectively as a collection of activities, are treated as outcomes (e.g., the number, length, and tone of voice for calls made by a telemarketer during the time period). From Kozlowski et al.'s (in press) perspective, it is useful and necessary for performance to be treated as an outcome in order to measure it, give feedback about it, and evaluate behaviors, but those same activities can also be seen as a process if viewed dynamically in a network of relationships among constructs. Thus, the distinction between process and outcome resides in the purpose or point of view for addressing the phenomenon rather than in some inherent feature of the phenomenon itself.

February 1999 • American Psychologist

TEAM and ProMES provide an interesting contrast in approaches to work team performance, effectiveness, or productivity. TEAM represents the traditional criteriondevelopment approach of identifying performance dimensions, constructing psychometrically valid scales, and then using them to measure team performance. ProMES makes no attempt to discover general behavior dimensions of performance. The overarching feature of this system is to create a measure that is anchored idiosyncratically in each team, but then the idiosyncratic dimensions are transformed into a common effectiveness metric. Using this metric, the ProMES system can be extended to aggregate team performance up the organizational hierarchy so that an estimate of the percent of total productivity can be obtained at each level by combining all lower levels. Fleishman and his colleagues developed a more extensive taxonomy of team-performance dimensions (Cooper, Shiflett, Korotkin, & Fleishman, 1984; Nieva, Fleishman, & Rieck, 1978; Shiflett, Eisner, Price, & Schemmer, 1982). After a number of years of research on the structure of tasks and human abilities (see Fleishman & Quaintance, 1984, for the culmination of this work), Fleishman focused his attention on teams in a programmatic research effort that spanned a decade. Dimensions of team performance were identified and refined, and rating scales with behavioral anchors were developed for the dimensions. The importance of the work is that it simultaneously addressed contextual demands, task characteristics, and human abilities. It did this in a manner that allowed for the psychometrically sound measurement of team-performance dimensions. Table 1 describes a version of the taxonomy presented by Fleishman and Zaccaro (1992). Hackman (1990) presented a collection of case studies on a wide variety of teams, ranging from a corporate restructuring team to a children's theater company. Attempting to capture commonalities in team performance, Hackman (1990) concluded that three dimensions of performance were critical. The first was the typical task-work component—productive output. The two more unique ones were the enhancement of the team members' ability to work together and the growth and well-being of team members. These latter two emphasize features of teams embedded in organizations that are often ignored when studying teams in isolation, particularly ad hoc ones. First, teams in organizations exist over relatively long periods of time. As a result, one dimension of their effectiveness at time tn is behavior that facilitates the teams' ability to work well together at time tn+l. Second, team members must feel that they are gaining something personally from working in the team. Work in organizations is typically viewed as an exchange relationship in which people provide their time and effort in exchange for personal returns to them (Naylor, Pritchard, & Ilgen, 1980). Hackman (1990) simply recognized this by stressing that part of a team's performance is maintaining perceptions of a reasonable exchange in the eyes of its members. In summary, the study of teams embedded in organizations has focused attention on team performance. As a result, the construct of team performance has been more February 1999 • American Psychologist

Table 1 Fleishman's Taxonomy of Team Performance" Dimension one: Orientation Exchange of information on member resources and constraints Exchange of information on team tasks and goals Exchange of information on situational characteristics and constraints Task priority assignments Dimension two: Resource distribution Matching member resources to task requirements Load balancing across members Dimension three: Timing/Activity pacing General activity pacing Individual member activity pacing Dimension four: Response coordination Response sequencing Time and position coordination of responses Dimension five: Motivation Development of team performance norms Generating acceptance of performance norms Establish team-level performance to reward linkages Reinforcement of task orientation Balance team orientation with individual competition Resolution of performance-relevant conflicts Dimension six: System monitoring General activity monitoring Individual activity monitoring Adjustment to errors (both team and individual) Dimension seven: Procedure maintenance Monitor team level procedural activity Monitor individual level procedural activity Adjust to nonstandard activities a

From "Toward a taxonomy of team performance functions." In R. W. Sweezey & E. Salas (Eds.), Teams: Their Training and Performance (p. 51), by E. A. Fleishman and S. J. Zaccaro, 1992, Norwood, Ml: Ablex. Copyright 1992 by Ablex. Reprinted with permission.

fully articulated, and several new measures of team performance have emerged. Some of these are based on sound job analytic principles. All are bound by the recognition that the performance of teams in organizations is central to their existence and to recognizing that these teams exist over time, making one element of their performance their ability to work together in the future. In addition, as the team-performance domain was mapped, behaviors often considered team process were encapsulated into performance, in large part because of concerns for creating approaches to performance with normative or prescriptive value.

Team Training Normative biases combined with high demand have expanded the interest in training for teams embedded in organizations. In the last 15 years, the demand for team training has skyrocketed, in large part because of the increased use of teams in organizations. It has also been 133

influenced by the high visibility and interest in cockpit crew performance and the attribution of many aircraft accidents or near misses to human error. Much of the training has the traditional training focus on needs assessment, content, methods of presentation, learning objectives, transfer to the job, and training evaluation. These fundamentals are not unimportant; they deserve careful attention. But they do not represent the most significant contributions of this work to understanding the behavior of teams embedded in organizations. The primary value has been the focus on teamwork behaviors and the development of team training theory. Helmreich (1984), Foushee (1984), and their colleagues (Gregorich, Helmreich, & Wilhelm, 1990) studied the behavior of cockpit crews, particularly reports of near misses, and identified personality and attitudinal variables believed to be critical for team performance. From this research measures of the attitudinal constructs were developed and refined (Gregorich et al., 1990), along with a training program, cockpit resource management (CRM). This training has been adopted by a number of commercial airlines and has been adapted to other settings where high coordination is required among team members performing critical behaviors often under stressful conditions, for example, operating rooms in hospitals (Gaba, Howard, & Small, 1995). In spite of its popularity, the validation data on CRM is dominated by trainee judgments about its value with little actual team performance data (Salas, Fowlkes, Stout, Milanovich, & Prince, 1998). Nevertheless, when some of the same variables central to CRM were included in two studies of naval aircraft cockpit crews trained on the Aircrew Coordination Training Program directed at team communication, member assertiveness, mission awareness, and situational awareness, support was found for the impact of such training on both trainee reactions and performance on a simulator (Salas et al., 1998). New team training models are emerging that incorporate principles of cognitive psychology and the fact that learning is a continuous process over time. A great deal of credit for stimulating interest and support of these newer models goes to Drs. Salas and Cannon-Bowers of the Naval Air Warfare Center in Orlando, Florida. Three edited volumes reflect the recent trends. These are Decision Making Under Stress: Implications for Training and Simulation, edited by Janis A. Cannon-Bowers and Eduardo Salas (1998); Teams: Their Training and Performance, edited by Robert Swezey and Eduardo Salas (1992); and Improving Training Effectiveness in Work Organizations, edited by J. Kevin Ford et al. (1997). Behavior is governed by both controlled and automatic processes. Training is an intervention demanding controlled processing but one that, if effective, often must aid in creating automatized routines (Rogers, Maurer, Salas, & Fisk, 1997). Although such views of cognitive processing are well accepted for individuals, raising them to the team level is not always straightforward. Clearly teams do develop routines or habits (Gersick & Hackman, 1990), but transitions over time are anything but smooth (Gersick, 1988). 134

Team mental models have been introduced as critical to team performance and training. Just as mental models guide individual performance (Johnson-Laird, 1980), it is argued that teams whose members share a common view of the task and the interaction structure among team members should do better (Cannon-Bowers et al., 1993; Klimoski & Mohammed, 1994; Orasanu, 1990; Orasanu & Salas, 1993). From a training standpoint, the concern is with ways of training teams so that their members will share similar perceptions of critical task and interpersonal dimensions and assume that what is shared is a positive or helpful model. Although there are good reasons for the value of shared mental models in teams, to date there has been little empirical research that can unambiguously isolate the contribution of shared mental models to team performance. Expert and novice differences observed with individuals have also been adopted into team training (Salas, Dickinson, Converse, & Tannenbaum, 1992). Trainees typically are novices. Understanding qualitative shifts in the way information is processed once expertise status has been obtained is important if what trainees learn is to transfer to team performance in the organization. Expert and novice differences are not well understood at this time. But, for our purposes here, the specifics are less important than the view taken toward team training. This view begins with the assumption that expert team performance is developmentally dynamic and evolves over time as the team learns from its experiences (Kozlowski et al., in press). It is this perspective that is altering how team training is conceived and spills over to the general problem of understanding how teams adapt and change in their dynamic environments. Levels Organizations are social systems open to their environments. They serve as environments to teams and other collectives of individuals nested within them (Katz & Kahn, 1978). In a similar fashion, teams themselves are systems with team members nested within them. These teams and their members are also open to their respective environments. Systems frameworks are ubiquitous in organizational research (Kozlowski & Salas, 1997). As a result, most any consideration of teams in organizations immediately invokes a systems perspective for thinking about such teams. However, as Kozlowski and Salas point out, systems theory provides a readily available heuristic, but its complexities often strand the empirical researcher with a metaphor that provides few cues for resolving many of the complexities that acceptance of the metaphor demands. One exception to this state of affairs is work on levels of analysis. From a levels perspective, theoretical principles exist for disentangling constructs within systems in ways that allow them to be open to empirical research while not violating their systemic nature (Kozlowski & Salas, 1997). Decomposition is possible because those constructs higher in a system tend to operate more slowly than those further down in the system (Simon, 1973). Rousseau (1985) assumed a three-level system composed of individuals nested within units (teams in our February 1999 • American Psychologist

case), and units within organizations. She then described three basic types of models to address relationships among variables that exist at more than one level in the three-level system. Composition models were those that specified functional relationships between functionally similar variables existing at different levels. For example, task performance is a construct that exists both at the individual and team level. Although performance may share many elements in common at both levels, performance is also composed of some elements not identical at each level. Compositional models deal with the nature of similarities and differences between levels for similar constructs. Cross-level models link variables that exist at one level with those at another. Consider, for example, the relationship between cohesion and team member satisfaction. Cohesion, by its very nature, is a team-level variable. It cannot exist within a single individual. Satisfaction is a state of the individual and, as such, is an individual-level variable. Cross-level models link the two. Finally, multilevel models assume functional identity for constructs at different levels and for causal processes affecting them (Kozlowski & Salas, 1997). At least two major contributions critical to building theory and conducting research on teams in organizations have resulted from the growing literature on levels of analysis. One contribution is to theory. Levels of analysis issues are first and foremost issues of theory (Kozlowski & Salas, 1997). A levels perspective provides a template of issues to be considered as theory is built. This framework is valuable in construct definition and constructing models of the relationships among variables. The second contribution of a levels perspective is that of helping the researcher avoid misspecification and aggregation biases. Misspecification occurs when relationships are attributed to levels that do not match those in which the behaviors of interest occur (Rousseau, 1985). An example of such an error is that of using the mean of individual team members' descriptions of their leader's behavior toward them as an indication of the leader's behavior toward the team as a whole. Aggregation biases occur to the extent that the magnitude or form of a relationship between or among variables is an artifact of the method by which the data are combined. Organizations are obviously multilevel, and teams embedded in them are confronted with levels issues. In stand-alone teams critical input, process, and output constructs also need to be specified as to level, and problems of levels need to be addressed. However, levels issues in team research were often ignored until recently. For the most part, the attention paid to levels issues was stimulated by those working with teams embedded in organizations.

Time Teams embedded in organizations exist over time. They have a past and a future, both of which influence behavior in the present (Hackman, 1992; McGrath, 1990, 1991). To those whose interest is teams in organizations, issues associated with the passage of time are givens (Hulin, Henry, & Moon, 1989). Jobs and roles are learned and unlearned over February 1999 • American Psychologist

time (Ilgen & Hollenbeck, 1991). Dyadic interaction patterns between leaders and members develop, stabilize, and change over time (Graen & Scandura, 1987; Kozlowski, Gully, McHugh, Salas, & Cannon-Bowers, 1996). Some people stay actively involved in their work, and others withdraw, either temporarily (e.g., absenteeism, day dreaming) or permanently (e.g., voluntary turnover; Hulin, 1991). If a person does leave the team, she or he is often replaced, and the team-performance pattern continues with the replacement member, but the interpersonal dynamics in the team are not necessarily the same. Both the replacement member and the team itself adapt to the new member. In sum, organizations represent environments in which teams with stable and unstable memberships react to conditions that often require complex and rapid coordinated responses over time (Kozlowski et al., 1996). Recognition of temporal issues is not unique to those interested in teams embedded in organizations. McGrath (1990, 1991) has urged that all small group research be sensitive to issues of time, and he provides a theory of groups with time in mind. However, an organizational perspective brings time into sharp focus. Time cannot be ignored. Theories and methods for studying teams in organizations have confronted temporal issues. Graen's leadermember exchange (LMX) model (Graen, 1976; Graen & Cashman, 1975; Graen & Scandura, 1987) treats leadership as the development of dyadic relationships over time. Team members are socialized into the group, and leaders modify and build relationships with their subordinates. Gersick (1988, 1989) observed task force teams over time and discovered midpoint transitions in which the teams reevaluated their progress and often made major qualitative shifts in their work strategies before proceeding to the completion of the task. Similarly attentive to changes over time, Gersick and Hackman (1990) addressed the development of habits and routines in teams. Ancona (1987; Ancona & Chong, 1996) developed the notion of entrainment as it affects the behavior of people in teams embedded in organizations. The tasks of organizations often have time cycles that are predictable. Universities on semester systems experience heightened activity in late August or early September and again in January; they are also relatively slow over the winter holidays. Shopping malls also cycle on an annual basis. The people who work in them also have personal life cycles that vary by the time of day, their family responsibilities, and so on. As Ancona points out, time and tempo are not only important within persons or tasks or organizations but the simultaneous match among all three is critical for the total system to function effectively. Entrainment is the integration of all three, and it is this matching that Ancona and others (e.g., McGrath, 1990) point out is necessary to understand the impact of time on team performance when teams are embedded in organizations.

Team Composition Few contextual characteristics more clearly differentiate the earlier approaches toward studying stand-alone teams 135

versus teams embedded in organizations than composition. The realities of the workplace rather than theory drive team-composition research in organizations. Social, legal, and economic forces have converged to focus attention on diversity among organizational members. In the United States, Title VII of the Civil Rights Act of 1964 was the single most important force toward creating a diversified workforce. The act demanded that decisions about employment and placement in organizations not adversely affect protected classes for reasons that were not related to job performance. Since that time, physical disabilities and age have been added to this list. Concurrent with that legal mandate is the growing acceptance of the position that diversity across these dimensions is a desirable state for organizational membership, although there is less consensus on how that state should be reached. Finally, from an economic perspective, the rapid internationalizing of markets has led to greater numbers of multinational corporations. These organizations functioning across national boundaries often find it necessary to create teams composed of members drawn from different nations. We are just beginning to explore the implications of creating work teams whose members are heterogeneous as to natural origin (Ilgen, LePine, & Hollenbeck, 1997). Argote and McGrath (1993) framed composition as a dilemma. On the one hand, heterogeneity (diversity) increases the knowledge pool important for addressing team tasks. Also diversity is a legal and social imperative if we are going to have organizations that are open to all. On the other hand, individuals, if left to their own devices, generally prefer to be with others like themselves. Too much heterogeneity can make it difficult for communication and coordination among team members. Thus, neither extreme homogeneity or heterogeneity is likely to be optimal. There needs to be a series of compromises considered as groups are composed to meet multiple objectives, some of which are incompatible. Could this conclusion about small groups have been reached without a sojourn into organizations? Of course. But, the path through teams in organizations has identified a number of specific compositional issues that, more than likely, would have been ignored had that journey not been undertaken.

Research Paradigms and Methods Simulations and Case Studies Focusing attention on teams embedded in organizations has led to increased use of two research methods—simulations and case studies—although neither method is new to team research. Siegfried Streufert, for example, conducted a number of studies using a tactical and negotiations game simulation that was very involved when introduced 30 years ago (Streufert, 1968). The difference today is that computer technology has allowed for a great deal of flexibility in configuring simulations. Computers have also simplified data collection and management. Simulations with high degrees of fidelity are currently being used for both team training and research (Stout et al., 1997). Other simulations with relatively low fidelity, such as our own 136

TIDE2 simulation of team decision making, use the simulated scenarios to capture team decision making demands and study them under controlled conditions (Hollenbeck et al., 1995). The attractiveness of simulations goes back to a point made earlier about the emphasis on critical organizational inputs and outputs along with the normative bias among those who study these teams. The realism that is possible without sacrificing control often makes simulation methods popular. The presumed positive transfer to field settings of knowledge gained from research and training on simulations is attractive to researchers whose ultimate aim is to improve the functioning of teams in organizations. The use of case studies is driven by similar desires to attain transfer of knowledge to field conditions. Trade-offs are made between control and realism. Case studies often lead to a decrease in confidence regarding inferences to causal explanations of the behaviors that are observed. However, when case data are considered as only one source of information and combined with data from other types of studies, knowledge is gained when the diverse methods converge on general principles.

Teams of Team Researchers One final methodological change in team research that has surfaced with the increased emphasis on the behavior of teams embedded in organizations is a change in the way research is conducted. There is an increase in the use of teams of investigators working together as a team themselves to address team issues. In one case, that reported by Hackman (1990) in an edited book entitled Team That Work (and Some That Don't), researchers met together and shared experiences so that they could learn from each other. The individually authored chapters of the book benefited because the authors were informed by the works of others with whom they were interacting as their own research was being conducted. A far more integrated and complex team of team researchers is that of the Tactical Decision Making Under Stress (TADMUS) Project (see Cannon-Bowers & Salas, 1998, for a detailed report of the project). A major multiyear and multiinvestigator project was undertaken to apply principles of cognitive and social psychology to the development of training for command and control teams embedded in organizations and working under stressful conditions. From the start, the eventual users of the system were integrated into the research team along with the behavioral scientists through the use of advisory panels and "customef'-focused efforts to solicit inputs about what should be done and to evaluate what was produced by the team of researchers (Howell, 1998). Within the larger research team were smaller teams with very focused responsibilities, such as those of building decision aids and others addressing communication structures. Here, too, the project leaders demanded high levels of teamwork among participants throughout the project. In sum, the complex and potentially conflicting demands of multiple constituencies that are typically encountered by teams embedded in organizations were confronted in the development of the research team in February 1999 • American Psychologist

an attempt to recognize and address them throughout the project.

Implications Declining interest and stagnation were persistent themes among those who reflected on the state of much of the traditional research on small groups and teams in the 1980s (e.g., Levine & Moreland, 1990; McGrath, 1997; Steiner, 1986). Yet, both Levine and Moreland and McGrath recognized that research on teams in organizations was an exception to this trend. For Levine and Moreland to conclude in 1990 that research on teams embedded in organizations was alive and well was reasonable, and it is even more reasonable today. It was suggested here that impact of research on teams embedded in organizations has been twofold. First, necessary conditions present in organizational settings in which the knowledge about the teams of interest should generalize modified the emphasis placed on inputs, outputs, and the nature of processes addressed in teams. These modifications were discussed with respect to five critical issues— team performance, team training, levels of analysis, time, and team composition. Illustrations of research on each of these presented earlier point to the conclusion that indeed this research has been and is alive and well; the substantive issues addressed in the literature have piqued the interest of both scholars and practitioners. But, how sustainable is this interest? What are the perspectives for long-term health? When McGrath (1997) reflected on the past of small group research, he observed a literature with a "diffuse array of theory and research" (p. 8) that clustered into schools, each of which failed to recognize its limitations, due, in part, to myopic tendencies to look primarily inward (to others within the same school) for answers. Ilgen et al. (1993) reached similar conclusions from a different perspective. They argued that small group or team research typically is stimulated by events that draw public attention to teams and create a strong desire to "fix" them or make them work better. Failures of cockpit crews or command and military command and control units or the use of management teams to develop economic strategies for remaining competitive in the world market are recent examples of environmental impetuses for team research; successes and failures of military squads in World War II or post hoc analyses of the Bay of Pigs decision-making disaster represent such events in the past. Ilgen et al. suggested that there were two types of responses to these problems. One was an attempt to design solutions to the problem by developing applied programs and practices. These approaches rarely sought out critical evaluation and were implemented in ways that allowed for little learning about why or if they worked. The second approach was to identify a single process, label it, then study it over and over under more and more experimental control (e.g., risky-shift, free loading, group think). Researchers within each approach sought inputs primarily from like-minded colleagues. With two research communities spinning in February 1999 • American Psychologist

separate domains, neither informed the other, and often neither theory nor practice was well served. The second implication of the recent research on teams embedded in organizations is a cautiously optimistic one—that the research will continue to avoid following along the two trajectories just mentioned. There are a number of reasons to hope that some of the pitfalls may be avoided. Researchers recognize that such teams exist in environments that are adaptive, dynamic, and structurally complex, with multiple variables from more than one level present at any given time. Thus, to understand these teams, the complexities of the situations cannot be ignored for long. Because no one paradigm is able to deal with all or most of the problems, researchers are more open to many different approaches from the field to the laboratory. Simulations that are able to capture some of the dynamic complexities of team tasks play a bigger role in the research. There has also been more tolerance for trading off control and also statistical power with case studies. In a sense, the set of research on teams embedded in organizations represents more of a portfolio model than a neatly bound paradigm one. It progresses by filling gaps rather than following a more linear path from one set of questions to another. The practice of addressing problems with teams of researchers and advisors also militates against premature convergence within a narrow framework. Building research teams with members who have multiple objectives may make coordination and consensus more difficult, but it provides critics within the team who may recognize problems as the project unfolds rather than at its conclusion. For example, much of the success of the TADMUS project was attributed to the heterogeneity of the research team (Howell, 1998). Finally, statistical and computational methods are evolving that are better able to deal with the dynamic conditions present in organizational teams (Weingart, 1997). These provide means to deal with some of the problems of dynamic change and structural complexity. The study of teams in organizations is utilizing many new methods and ways of conducting research. All of these factors imply that a number of conditions are right for both the long life and continued health of work in this area. REFERENCES Ancona, D. G. (1987). Groups in organizations: Extending laboratory models. In C. Hendrick (Ed.), Group processes and intergroup relations (pp. 178-202). Newbury Park, CA: Sage. Ancona, D. G., & Chong, C. L. (1996). Entrainment: Pace, cycle and rhythm in organizational behavior. In L. L. Cummings & B. M. Staw (Ed.), Research in organizational behavior (Vol. 18, pp. 251-284). Greenwich, CT: JAI Press. Argote, L., & McGrath, J. E. (1993). Group process in organizations: Continuity and change. In C. L. Cooper & I. T. Robertson (Eds.), International review of industrial and organizational psychology (Vol. 8, pp. 333-389). London: Wiley. Baker, D. P., & Salas, E. (1992). Principles for measuring teamwork skills. Human Factors, 34, 469-475. Baker, D. P., & Salas, E. (1997). Principles for measuring teamwork: A summary and a look toward the future. In M. T. Brannick, E. Salas, &

137

C. Prince (Eds.), Team performance and measurement (pp. 331-355). Mahwah, NJ: Erlbaum. Blake, R. R., & Mouton, J. S. (1969). Building a dynamic corporation through grid organization development. Reading, MA: Addison-Wesley. Campbell, J. P., Dunnette, M. D., Arvey, R. D., & Hellervik, L. V. (1973). The development and evaluation of behaviorally based rating scales. Journal of Applied Psychology, 57, 15-22. Cannon-Bowers, J., & Salas, E. (1998). Decision making under stress: Implications for training and simulation. Washington, DC: American Psychological Association. Cannon-Bowers, J., Salas, E., & Converse, S. (1993). Shared mental models in expert team decision making. In N. J. Castellan, Jr. (Ed.), Individual and group decision making: Current issues (pp. 221-246). Hillsdale, NJ: Erlbaum. Cannon-Bowers, J., Tannenbaum, S. I., Salas, E., & Volpe, C. E. (1995). Denning team competencies and establishing team training requirements. In R. Guzzo, E. Salas, et al. (Eds.), Team effectiveness and decision making in organizations (pp. 333-380). San Francisco: JosseyBass. Cooper, M., Shiflett, S. C , Korotkin, A. L., & Fleishman, E. A. (1984). Command and control teams: Techniques for assessing team performance. Washington, DC: ARRO. Fleishman, E. A., & Quaintance, M. K. (1984). Taxonomies of human performance. New York: Academic Press. Fleishman, E. A., & Zaccaro, S. J. (1992). Toward a taxonomy of team performance functions. In R. W. Sweezey & E. Salas (Eds.), Teams: Their training and performance (pp. 31-56). Norwood, NJ: Ablex. Ford, J. K., Kozlowski, S. W. J., Kraiger, K., Salas, E., & Teachout, M. S. (Eds.). (1997). Improving training effectiveness in work organizations. Mahwah, NJ: Erlbaum. Foushee, H. C. (1984). Dyads and tryads at 35,000 feet: Factors affecting group process and aircrew performance. American Psychologist, 39, 885-893. Gaba, D. B., Howard, S. K., & Small, S. D. (1995). Situation awareness in anesthesiology. Human Factors, 37, 20—31. Gersick, C. J. (1988). Time and transition in work teams: Toward a new model of group development. Academy of Management Journal, 31, 9-41. Gersick, C. J. (1989). Marking time: Predictable transitions in task groups. Academy of Management Journal, 32, 274-309. Gersick, C. J., & Hackman, J. R. (1990). Habitual routines in task performing groups. Organizational Behavior and Human Decision Processes, 47, 65-97. Graen, G. B. (1976). Role making processes within complex organizations. In M. D. Dunnette (Ed.), Handbook of industrial and organizational psychology (pp. 1201-1245). Chicago: Rand McNally. Graen, G. B., & Cashman, J. (1975). A role making model of leadership in formal organizations: A developmental approach. In J. G. Hunt & L. I. Larson (Eds.), Leadership frontiers (pp. 143-165). Kent, OH: Kent University Press. Graen, G. B., & Scandura, T. A. (1987). Toward a psychology of dyadic organizing. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 9, pp. 175-208). Greenwich, CT: JAI Press. Gregorich, S. E., Helmreich, R. L., & Wilhelm, J. A. (1990). The structure of cockpit management attitudes. Journal of Applied Psychology, 75, 682-690. Hackman, J. R. (1987). The design of work teams. In J. W. Lorsch (Ed.), Handbook of organizational behavior (pp. 315-342). Englewood Cliffs, NJ: Prentice-Hall. Hackman, J. R. (1990). Teams that work (and some that don't). Englewood Cliffs, NJ: Prentice-Hall. Hackman, J. R. (1992). Group influences in organizations. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psychology (2nd ed., pp. 199-267). Palo Alto, CA: Consulting Psychologists Press. Helmreich, R. L. (1984). Cockpit management attitudes. Human Factors, 26, 583-589. Hollenbeck, J. R., Ilgen, D. R., Sego, D. J., Hedlund, J., Major, D. A., & Phillips, J. (1995). The multilevel theory of team decision making:

138

Decision performance in teams incorporating distributed expertise. Journal of Applied Psychology, 80, 292-316. Howell, W. C. (1998). When applied research works: Lessons from the TADMUS Project. In J. A. Cannon-Bowers & E. Salas (Eds.), Decision making under stress: Implications for training and simulation (pp. 415-425). Washington, DC: American Psychological Association. Hulin, C. L. (1991). Adaptation, persistence, and commitment in organizations. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psychology (2nd ed., Vol. 2, pp. 445505). Palo Alto, CA: Consulting Psychologists Press. Hulin, C. L., Henry, R. A., & Moon, S. L. (1989). Adding a dimension: Time as a factor in the generalizability of predictive relationships. Psychological Bulletin, 107, 328-340. Ilgen, D. R. (1994). Jobs and roles: Accepting and coping with the changing structure of organizations. In M. G. Rumsey, C. B. Walker, & J. H. Harris (Eds.), Personnel selection and classification (pp. 13-22). Hillsdale, NJ: Erlbaum. Ilgen, D. R., & Hollenbeck, J. R. (1991). Job design and roles. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psychology (2nd ed., Vol. 2, pp. 165-208). Palo Alto, CA: Consulting Psychologists Press. Ilgen, D. R., LePine, J. A., & Hollenbeck, J. R. (1997). Effective decision making in multinational teams. In C. Earley & M. Erez (Eds.), New perspectives in international industrial/organizational psychology (pp. 377-409). San Francisco: Jossey-Bass. Ilgen, D. R., Major, D. A., Hollenbeck, J. R., & Sego, D. J. (1993). Team research in the 1990s. In M. M. Chemers & R. Aymon (Eds.), Leadership theory and research: Perspectives and directions (pp. 245-271). San Diego, CA: Academic Press. Johnson-Laird, P. N. (1980). Mental models in cognitive sciences. Cognitive Sciences, 4, 71-115. Katz, D., & Kahn, R. L. (1978). The social psychology of organizing (2nd ed.). New York: Wiley. Kleinman, D. L., Serfaty, D., & Luh, P. B. (1984). A research paradigm for multi-human decision-making. In Proceedings of the 1984 American Control Conference (pp. 6-11). New York: Institute for Electrical and Electronics Engineers. Klimoski, R., & Mohammed, S. (1994). Team mental model: Construct or metaphor? Journal of Management, 20, 403-437. Kozlowski, S. W. J., Gully, S. M., McHugh, P. P., Salas, E., & CannonBowers, J. A. (1996). A dynamic theory of leadership and team effectiveness: Developmental and task contingent leader roles. In G. R. Ferris (Ed.), Research in personnel and human resource management (Vol. 14, pp. 253-305). Greenwich, CT: JAI Press. Kozlowski, S. W. J., Gully, S. M., Nason, E. R., & Smith, E. M. (in press). Team compilation: Development, performance, and effectiveness across time and levels. In. D. R. Ilgen & E. D. Pulakos (Eds.), The changing nature of work and performance: Implications for staffing personnel actions and development. San Francisco: Jossey-Bass. Kozlowski, S. W. J., & Salas, E. (1997). A multilevel organizational systems approach for the implementation and transfer of training. In J. K. Ford, S. W. J. Kozlowski, K. Kraiger, E. Salas, & M. S. Teachout (Eds.), Improving training effectiveness in work organizations (pp. 247-287). Mahwah, NJ: Erlbaum. Levine, J. M., & Moreland, R. L. (1990). Progress in small group research. Annual Review of Psychology, 41, 585-634. McGrath, J. E. (1984). Groups: Interaction and performance. Englewood Cliffs, NJ: Prentice-Hall. McGrath, J. E. (1990). Time matters in groups. In J. Galegher, R. Kraut, & C. C. Egido (Eds.), Intellectual teamwork (pp. 23-61). Hillsdale, NJ: Erlbaum. McGrath, J. E. (1991). Time, interaction, and performance (TIP): A theory of groups. Small Groups Research, 22, 147-174. McGrath, J. E. (1997). Small group research, that once and future field: An interpretation of the past with an eye to the future. Group Dynamics, 1, 7-27. McGrath, J. E., & Gruenfeld, D. H. (1993). Toward a dynamic and systemic theory of groups: An integration of six temporally enriched perspectives. In M. M. Chemers & R. Ayman (Eds.), The future of leadership research: Promise and perspective (pp. 2-46). Orlando, FL: Academic Press. Mclntyre, R. M., & Salas, E. (1995). Measuring and managing for team

February 1999 • American Psychologist

performance: Emerging principles from complex environments. In R. A. Guzzo & E. Salas (Eds.), Team effectiveness and decision making in organizations (pp. 9-45). San Francisco, CA: Jossey-Bass. Morgan, B. B., Jr., Glickman, A. S., Woodward, E. A., Blaiwes, A., & Salas, E. (1986). Measurement of team behaviors in a Navy environment (NTSC Report No. 86-014). Orlando, FL: Naval Training System Center. Naylor, J. C , Pritchard, R. D., & Ilgen, D. R. (1980). A theory of behavior in organizations. New York: Academic Press. Nieva, V. F., Fleishman, E. A., & Rieck, A. M. (1978). Team dimensions: Their identity, their measurement, and their relationships (Tech. Rep.). Washington, DC: ARRO. Orasanu, J. (1990). Shared mental models and crew performance. Paper presented at the 34th annual meeting of the Human Factors Society, Orlando, FL. Orasanu, J., & Salas, E. (1993). Team decision making in complex environments. In G. A. Klein, I. Orasanu, R. Calderwood, & C. Zsambok (Eds.), Decision making in action: Models and methods (pp. 327-345). Norwood, NJ: Ablex. Pritchard, R. D. (1995). Productivity measurement and improvement: Organizational case studies. Westport, CT: Praeger. Pritchard, R. D., Jones, S. D., Roth, P. L., Stuebing, K. K., & Ekeberg, S. E. (1988). Effects of group feedback, goal setting, and incentives on organizational productivity. Journal of Applied Psychology, 73, 337358. Rogers, W., Maurer, T., Salas, E., & Fisk, A. (1997). Task analysis and cognitive theory: Controlled and automatic processing task analytic methodology. In J. K. Ford, S. W. J. Kozlowski, K. Kraiger, E. Salas, & M. S. Teachout (Eds.), Improving training effectiveness in work organizations (pp. 19-46). Mahwah, NJ: Erlbaum. Rousseau, D. M. (1985). Issues of level in organizational research: Multilevel and cross-level perspectives. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 7, pp. 1-38). Greenwich, CT: JAI Press. Salas, E., Dickinson, T., Converse, S. A., & Tannenbaum, S. I. (1992). Toward an understanding of team performance and training. In R. W.

February 1999 • American Psychologist

Swezey & E. Salas (Eds.), Teams: Their training and performance (pp. 3-29). Norwood, NJ: Ablex. Salas, E., Fowlkes, J. E., Stout, R. J., Milanovich, D. M., & Prince, C. (1998). Does team training improve teamwork skills in the cockpit? Manuscript in preparation. Seashore, S. E., & Bowers, D. G. (1970). The durability of organizational change. American Psychologist, 25, 227-233. Shiflett, S. C , Eisner, E. J., Price, S. J., & Schemmer, F. M. (1982). The definition and measurement of team functions. Bethesda, MD: ARRO. Simon, H. (1973). The organization of complex systems. In H. H. Pattee (Ed.), Hierarchy theory: The challenge of complex systems (pp. 1-28). New York: Braziller. Smith, P. C , & Kendall, L. M. (1963). Retranslation of expectations: An approach to the construction of unambiguous anchors for rating scales. Journal of Applied Psychology, 47, 149-155. Steiner, I. D. (1972). Group process and productivity. New York: Academic Press. Steiner, I. D. (1974). Whatever happened to the group in social psychology. Journal of Experimental Social Psychology, 10, 94-108. Steiner, I. D. (1986). Paradigms and groups. Advances in Experimental Social Psychology, 19, 251-289. Stogdill, R. M. (1974). Handbook of leadership. New York: Free Press. Stout, R. J., Salas, E., & Fowlkes, J. E. (1997). Enhancing teamwork in complex environments through team training. Group and Organizational Studies, 1, 169-182. Streufert, S. (1968). The components of simulations of local conflict: An analysis of the tactical and negotiations game (ARPA Tech. Rep., Projects SD260). Evanston, IL: Northwestern University. Sundstrom, E., DeMeuse, K. P., & Futrell, D. (1990). Work teams: Applications and effectiveness. American Psychologist, 45, 120-123. Swezey, R. W., & Salas, E. (1992). Teams: Their training and performance. Norwood, NJ: Ablex. Weingart, L. R. (1997). How did they do that? The ways and means of studying group process. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 19, pp. 189-239). Greenwich, CT: JAI Press.

139