Human Factors: The Journal of the Human Factors and Ergonomics Society http://hfs.sagepub.com/ Training Adaptive Teams Jamie C. Gorman, Nancy J. Cooke and Polemnia G. Amazeen Human Factors: The Journal of the Human Factors and Ergonomics Society 2010 52: 295 originally published online 23 July 2010 DOI: 10.1177/0018720810371689 The online version of this article can be found at: http://hfs.sagepub.com/content/52/2/295 Published by: http://www.sagepublications.com On behalf of: Human Factors and Ergonomics Society Additional services and information for Human Factors: The Journal of the Human Factors and Ergonomics Society can be found at: Email Alerts: http://hfs.sagepub.com/cgi/alerts Subscriptions: http://hfs.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://hfs.sagepub.com/content/52/2/295.refs.html Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 >> Version of Record - Sep 14, 2010 OnlineFirst Version of Record - Jul 23, 2010 What is This? Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 SPECIAL ISSUE Training Adaptive Teams Jamie C. Gorman and Nancy J. Cooke, Arizona State University–Polytechnic, Mesa, Arizona, and Polemnia G. Amazeen, Arizona State University, Tempe, Arizona Objective: We report an experiment in which three training approaches are compared with the goal of training adaptive teams. Background: Cross-training is an established method in which team members are trained with the goal of building shared knowledge. Perturbation training is a new method in which team interactions are constrained to provide new coordination experiences during task acquisition. These two approaches, and a more traditional procedural approach, are compared. Method: Assigned to three training conditions were 26 teams. Teams flew nine simulated uninhabited air vehicle missions; three were critical tests of the team’s ability to adapt to novel situations. Team performance, response time to novel events, and shared knowledge were measured. Results: Perturbation-trained teams significantly outperformed teams in the other conditions in two out of three critical test missions. Cross-training resulted in significant increases in shared teamwork knowledge and highest mean performance in one critical test. Procedural training led to the least adaptive teams. Conclusion: Perturbation training allows teams to match coordination variability during training to demands for coordination variability during posttraining performance. Although cross-training has adaptive benefits, it is suggested that process-oriented approaches, such as perturbation training, can lead to more adaptive teams. Application: Perturbation training is amenable to simulation-based training, where perturbations provide interaction experiences that teams can transfer to novel, real-world situations. INTRODUCTION In settings ranging from business and manufacturing to military and medical operations, there are many tasks that are too cognitively demanding to be performed by individuals working alone. An example is a surgical task, which requires a set of highly trained individuals, including two surgeons, an anesthesiologist and two nurses, each of whom brings different cognitive capabilities to the team. But it is not enough to bring together a set of highly trained individuals. To function as a team, individuals must coordinate their activities. Adaptive teams have the ability to coordinate their activities not only under routine conditions but also under novel conditions for which they have not been explicitly trained. Adaptation is the altering of structure in accordance with changes in the environment. Because they have the ability to change their interactions to match the changing demands of the environment, adaptive teams can perform at a high level under novel task conditions. A number of relatively recent tragic system failures can be at least partially attributed to poor coordination of a team-level response to environmental uncertainty. System failures attributable to poor team skills at Three Mile Island and Chernobyl (Gaddy & Wachtel, 1992), social pathogens behind the 1986 launch decision of the space shuttle Challenger (Vaughan, 1996), and lack of communication in the Operation Provide Comfort friendly fire incident (Gorman, Cooke, & Winner, 2006; Snook, 2002) each implicate, in different ways, deficiencies in interaction and coordination that result in a failure to adapt to changes in the task environment. These incidents, and others like them, highlight the need for training that addresses limitations and Address correspondence to Jamie C. Gorman, Cognitive Engineering Research Institute, 5810 S. Sossaman Rd., Ste. 106, Mesa, AZ 85212; jgorman@cerici.org. HUMAN FACTORS, Vol. 52, No. 2, April 2010, pp. 295–307. DOI: 10.1177/0018720810371689. Copyright © 2010, Human Factors and Ergonomics Society. Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 296 April 2010 - Human Factors deficiencies at the team level in responding to novel patterns of events and threats. Approaches to Training Adaptive Teams A challenging problem for training team cognition (i.e., training teams to decide, plan, think, and act as an integrated unit; Cooke, Gorman, & Winner, 2007) is how to balance training for high performance under routine task conditions with training to adapt to novel task demands (Marks, Zaccaro, & Mathieu, 2000; Stachowski, Kaplan, & Waller, 2009). These training goals can be approached with varying theoretical motives. In this article, we report an experiment in which three training approaches, each with a different underlying theoretical motive, were investigated with the goal of training teams that perform at a high level under novel task conditions. The training approaches include cross-training, procedural training, and perturbation training. Cross-training. In cross-training, team members are trained on each other’s roles and responsibilities (e.g., Blickensderfer, Cannon-Bowers, & Salas, 1998). Cross-training is theoretically aligned with the idea that team cognition is the shared knowledge of the team members and is found widely in the team training literature (see Salas et al., 2008, and Salas, Nichols, & Driskell, 2007, for recent meta-analyses). The goal of cross-training is the development of shared, or interpositional, knowledge (Cannon-Bowers, Salas, Blickensderfer, & Bowers, 1998; Cooke et al., 2003; Volpe, Cannon-Bowers, Salas, & Spector, 1996). Positional clarification (receiving information on other roles), positional modeling (observing other roles), and positional rotation (firsthand experience performing different roles) (Blickensderfer et al., 1998), which are types of cross-training, have been effective in the development of shared knowledge, ultimately improving coordination and team performance (e.g., Marks, Sabella, Burke, & Zacarro, 2002). Cross-training has a firm empirical grounding and a record of success in the team training literature, making it a good point of comparison with the other approaches in this study. One of the potential benefits of cross-training for shared knowledge is a high level of team performance under stress (high workload, time pressure). Drawing on shared knowledge, team members anticipate each other’s needs to communicate efficiently, or coordinate implicitly, under stress (Cannon-Bowers et al., 1998; Entin & Serfaty, 1999; Stout, Cannon-Bowers, Salas, & Milanovich, 1999). It is thought that shared expectations, resulting from the development of shared knowledge, allow team members to generate predictions for appropriate behavior under novel conditions, enabling them to quickly adapt to the changing demands of the task environment (Fiore, Salas, & Cannon-Bowers, 2001). A possible drawback of a shared set of expectations, however, is the habituation of team member interaction, which could result in dysfunctional consequences if the situation is highly novel (e.g., Gersick & Hackman, 1990; Gorman, Cooke, & Winner, 2006). Whereas cross-training is feasible for relatively small, homogeneously skilled teams, it can become impractical as teams grow in diversity and size. For example, it would be impractical to cross-train the surgeon and nurse positions of an emergency room team (Cooke et al., 2003; Marks et al., 2002). Also, cross-training may negatively impact individual-level performance due to the demands of training for multiple team member roles (Cannon-Bowers et al., 1998), which is problematic as teams grow in size. Although there are adaptive benefits of crosstraining, there are practical limitations to its applicability. Procedural training. Procedural training is a form of process training in which operators in complex systems are positively reinforced (through feedback) to follow a standard sequence of actions (a procedure) each time a particular stimulus is encountered. The assumption behind procedural training is that if the procedure is always followed, then errors resulting from human interaction will be reduced and performance will be enhanced, particularly under conditions of stress and high workload (e.g., Hockey, Sauer, & Wastell, 2007; Sauer, Burkholter, Kluge, Ritzmann, & Schuler, 2008). Procedural training is widely used in aviation, military, medical, manufacturing, and business settings, in which deviations from complicated procedures can be catastrophic. The prevalence of procedural training for coordination in highly critical team tasks Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Training Adaptive Teams 297 (e.g., emergency response; Ford & Schmidt, 2000; Stachowski et al., 2009) make it a good point of comparison for the other training methods in this study. Procedural training is compatible with the concept of “overlearning”: continuation of practice beyond mastery that leads to automatic responding. Drilling a standard team interaction pattern, for a specific class of event, over the entirety of training can lead to an automatic response that a team can rely on under stress. The goal of procedural training, as operationalized in the current study, is to overlearn a team coordination procedure. Ideally, due to overlearning, procedurally trained teams perform under stress by automatically (reflexively) reacting with an a priori coordinated response. Procedural training does not impose the practical limitations of cross-training but may limit a team’s ability to transfer training to novel situations. Similar to the concept of a “set effect” (Luchins, 1942), procedural training may set teams up to coordinate in a routine fashion under a novel condition. We argue, therefore, that like habituation (Gersick & Hackman, 1990), rigid proceedure-following during task acquisition can lead to poor performance when posttraining conditions do not match training conditions. Perturbation training. Perturbation training is a form of process training introduced in this study. Adopted from the dynamic systems literature, a perturbation is an extrinsic application of force that briefly disrupts a dynamic process, forcing the reacquisition of a new stable trajectory, and is typically used to probe the stability of that process (Gorman, Amazeen, & Cooke, in press). The concept of perturbation can be applied to team training by disrupting standard coordination procedures multiple times during task acquisition, forcing teams to coordinate in novel ways to achieve their objective. Unlike training in which the situation or objectives are varied (e.g., training for low- vs. high-frequency circumstances), in perturbation training, critical coordination links are disrupted while the team objective remains constant. The goal of perturbation training is to counteract habituation and procedural rigidity associated with team interactions—possible outcomes of cross-training and procedural training, respectively—allowing teams to acquire flexible interaction processes that will transfer to novel task conditions. Perturbation training is theoretically inspired by findings in the motor- and verbal-learning literatures that suggest that introducing difficulties for the learner, such as practice condition variability, facilitates performance under novel posttraining conditions (Schmidt & Bjork, 1992). Perturbation training thus shares some features of motor schema theory (Schmidt, 1975) and desirable difficulties (Bjork, 1994) but for coordination and for teams. According to motor schema theory, varying the conditions of practice during motor skill acquisition enhances the “rules” that relate movements to external task demands. In verbal learning, desirable difficulties are unpredictable and variable conditions of practice that cause difficulty for the learner but ultimately enhance the transfer of concepts to new contexts. Whereas those approaches employ equally probable but randomly varying training conditions to introduce practice condition variability, perturbation training employs abrupt but focused disruptions to team coordination. Bjork (1994) argued that varying practice conditions exercises more elaborate encoding and retrieval processes needed in the posttraining environment. Perturbation training extends this idea to team process: When coordination is perturbed, all team member interactions (not just those directly affected by the perturbation) must readjust to accommodate the perturbation in such a manner that the team objective is nevertheless met (Turvey, 1990). We suggest that similar to the effects of practice condition variability, perturbations exercise the team processes needed to adapt in the posttraining environment. A major limitation of perturbation training is that it has not previously been applied and its effectiveness is unknown. An experiment described by Gorman, Cooke, Pedersen, et al. (2006) provided some empirical grounding for perturbation training. Teams were initially trained and performed a repetitive command-and-control task during a 3-hr experimental session. Participants returned for a second session after a retention interval, after which they were either intact (kept the same team members) or mixed (same role on the team but different team members). Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 298 April 2010 - Human Factors As expected, intact teams outperformed mixed teams under routine conditions, but the effect was short-lived. Mixed teams, however, performed better on tests of situation awareness, had higher process ratings (Gorman, Cooke, Pedersen, et al., 2006), and had more flexible coordination dynamics (Gorman et al., in press). Those benefits were not attributable to increased shared knowledge or procedural rigidity but to increased variation in interaction experience. In some sense, mixing up the team members perturbed rigid coordination patterns, ultimately leading to a more flexible and adaptive team. Perturbation training, as operationalized in the current study, does not involve mixing team members but was designed have the same effect. By disrupting standard coordination procedures during task acquisition, perturbation training increases interaction experience in intact command-and-control teams. The Current Study We compared the three different training approaches in an uninhabited air vehicle (UAV) simulator with the goal of producing teams that perform at a high level under novel task conditions and that respond rapidly to novel events. In the UAV task, three team members (navigator, photographer, and pilot) coordinate to take pictures of stationary ground targets. Three training protocols were developed for cross-training, perturbation training, and procedural training of UAV teams. The following hypotheses are based on prior results and existing literature. Hypothesis 1: By focusing on introducing varied interaction experiences during task acquisition, perturbation training will result in performance scores and response times to novel events that are as good as or better than cross-training and superior to procedural training. Hypothesis 2: Because of its focus on training team members to know each other’s roles and responsibilities, cross-training will result in higher levels of shared knowledge compared with both procedural and perturbation training. Hypothesis 3: By training teams to rigidly follow a procedure, procedural training will result in the least adaptive teams (i.e., poor performance and slow response to novel events) compared with both perturbation and cross-training. METHOD Participants We recruited 32 three-person teams (96 participants) for participation from Mesa, Arizona, and surrounding areas. The team members had no prior experience working together. Participants ranged in age from 18 to 54 (M = 28), and 71 were male. The experiment occurred during two 3- to 4-hr sessions. Because of scheduling conflicts for Session 2, a total of 26 teams (78 participants) completed both experimental sessions. Participants were paid $10 per hour, and each member of the highest-performing team received a $100 bonus. Materials and Apparatus The experiment was conducted in a UAV synthetic task environment (UAV-STE) for teams (Cooke & Shope, 2005). Each of the three team members was seated at a workstation in front of three computer monitors with a keyboard and a mouse. To interact, team members wore aviation-quality headsets and communicated by holding down push-to-talk buttons. The workstations were located in the same room, configured in a U shape with team members backs to each other. With the team members donning headsets, the UAV-STE did not afford face-to-face interaction. The team’s task was to take reconnaissance photographs of stationary ground targets during a series of nine 40-min missions divided across two experimental sessions. There were 11 to 12 targets per mission except for one highworkload mission that had 20 targets. The three team member roles—navigator, photographer, and pilot—were each associated with different, yet interdependent tasks, information resources, and needs. Measures Team performance. Performance was measured for each UAV-STE mission as the weighted composite of several team-level mission parameters, including number of missed targets, time to process targets, and time spent with unaddressed warnings and alarms. Cooke, Gorman, Pedersen, et al. (2007) report the parameter weights, which were established in previous experiments to maximize score sensitivity. Teams started each Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Training Adaptive Teams 299 mission with a performance score of 1,000, and points were subtracted on the basis of final values of the mission parameters. This team performance score has been validated against other measures of team process and performance (Cooke, Gorman, Duran, & Taylor, 2007; Cooke, Gorman, Pedersen, et al., 2007). Response time to novel events. Novel events were introduced within UAV missions by introducing roadblocks. Roadblocks are novel changes in the task environment that have to be jointly recognized by two or more team members who take action to overcome them (e.g., a new target is introduced, equipment fails, an enemy threat appears; Cooke, Gorman, & Rowe, 2009; Gorman, Cooke, & Winner, 2006). Time to overcome roadblocks, defined as the time from the initiation of the roadblock to the time that action is taken that overcomes the roadblock, was the measure of response time to novel events. Interpositional taskwork knowledge. This measure assessed a team’s average knowledge of the taskwork associated with the other two roles. To measure taskwork knowledge, relatedness ratings (1 = completely unrelated to 5 = completely related) were elicited for 55 pairs of concepts from the UAV task (e.g., airspeed, altitude). Individual team member ratings were analyzed using the Pathfinder algorithm (Schvaneveldt, 1990), which translates relatedness ratings across pairs of concepts into a graphical network representation of conceptual interrelatedness. Individual networks were compared with expert role referent networks. The referents were derived empirically from the top five individual performers at each role in previous UAV-STE experiments (Cooke, Gorman, Pedersen, et al., 2007). Each team member was scored against the other two role referents on the basis of the proportion of shared links (0 = no similarity to 1 = exactly similar). Team-level interpositional taskwork knowledge was taken as the average of these two scores across each of the three team members. Scores closer to 1 indicated a higher level of interpositional taskwork knowledge across team members. Interpositional teamwork knowledge. This measure assessed a team’s average knowledge of the teamwork associated with the other two roles. Interpositional teamwork knowledge was elicited with the use of a questionnaire that consisted of 16 items related to which communications were necessary to achieve a given scenario goal (e.g., “For a priority target, must the photographer communicate camera settings to the navigator, the pilot, or both?”). Items that were necessary had to be indicated by individual team members using check marks. To calculate teamwork knowledge, individual responses were compared with role-specific answer keys that were generated by experimenters familiar with the task, and points were awarded for correct answers (Cooke, Gorman, Pedersen, et al., 2007). To measure interpositional teamwork knowledge, each team member was scored on the basis of the proportion correct relative to the answer key for each of the other two roles. Interpositional teamwork knowledge was calculated as the average number of these two scores across the three team members. Scores closer to 1 indicated a higher level of interpositional teamwork knowledge across team members. Procedure When participants arrived for the first session, they were randomly assigned to a team member role and the team was assigned to one of the three training conditions. Participants received approximately 1 hr 45 min of training via three PowerPoint training modules and a hands-on training mission. The first two PowerPoint modules were identical for all training conditions and covered the general task and interface. The third module and the hands-on training mission differed on the basis of training condition. (Procedures for each training condition are described in the following section.) Teams then completed Missions 1 through 5. Knowledge measures were taken after Mission 1. Missions 2 through 4 were condition-specific training missions (Table 1). The first roadblock was introduced during Mission 5, the first posttraining mission. The roadblock consisted of cutting communication from the navigator to the pilot for 5 min, during which teams had to reroute navigator-to-pilot communications through the photographer to overcome the roadblock. Mission 5 was the first of three critical missions that tested the teams’ ability to perform under novel conditions. The completion of Mission 5 concluded the first session. Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 300 April 2010 - Human Factors TABLE 1: Experimental Procedure Session 1 Session 2 Initial participant training Mission 1 Knowledge measures Mission 2T Mission 3T Mission 4T Mission 5 (first roadblock)C Refresher training Mission 6 (retention test and second roadblock)C Knowledge measures Mission 7 Mission 8 Mission 9 (high workload and two-part roadblock)C Debriefing Note. T = condition-specific training mission; C = critical test mission. Teams returned after 8 to 10 weeks for the second session. All participants received refresher training on the software interfaces, after which teams completed four additional UAV-STE missions. Roadblocks were introduced during each mission. Mission 6 was the second critical mission, which tested retention of team skill after the break and included the second roadblock (a disguised target was hidden on the navigator map and photographer target list; teams had to recognize and photograph the target to overcome the roadblock). Knowledge was measured for a second time after Mission 6 as a test of knowledge retention. Teams then completed their final three missions (Missions 7 through 9). Mission 9 was the high-workload mission, in which the rate of targets per minute was almost doubled from 0.28 to 0.5 and teams were exposed to a two-part roadblock (communication channel cut for 5 min from pilot to navigator and from navigator to pilot; teams had to reroute their communications through open channels to overcome the roadblock). Mission 9 was the third of the three critical test missions. Training Procedure Cross-training. For the third PowerPoint training module, team members in the cross-training condition received training on the other two roles (positional clarification). Teams then completed a short training mission followed by approximately 15 min of hands-on experience performing all team member roles (positional rotation). After Missions 2 through 4, teams in the cross-training condition were prompted to discuss how well they performed and to plan for the next mission. Procedural training. For the third PowerPoint training module, teams in the procedural training condition received training on the standard UAV-STE target photographing procedure: (a) The navigator provides target information to the pilot, (b) the pilot and photographer negotiate altitude and airspeed for that target, and (c) the photographer provides feedback on the status of the target photograph (Figure 1). Teams then completed a short training mission followed by approximately 15 min of hands-on training using the target photographing procedure. After Missions 2 through 4, teams in the procedural condition received experimenter feedback on deviations from the standard procedure. During training, team members in the procedural condition were provided with a hard copy of the target photographing procedure. Perturbation training. Teams in the perturbation training condition received filler PowerPoint training on the history and current uses of UAVs. Teams then completed a short training mission followed by approximately 15 min of communications system testing in which they identified the source of static in the UAV-STE communication system (i.e., which push-to-talk button was emanating static). This training exercise provided experience on the use of multiple communication paths. During Missions 2 through 4, teams in the perturbation condition received perturbations to the target-photographing procedure (Table 2) as they attempted to photograph targets. Perturbations were less general than roadblocks and forced teams to adjust specific interactions Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Training Adaptive Teams 301 Figure 1. Standard photographing procedure for the uninhabited air vehicle synthetic task environment. TABLE 2: Perturbations to the Standard Uninhabited Air Vehicle Synthetic Task Environment TargetPhotographing Procedure Used for Perturbation Training During Missions 2 Through 4 Link in the Procedure Perturbation Method of Introducing Perturbation When Introduced Information Photographer must provide target information to pilot Experimenter calls in new target restrictions to photographer and disables camera until restrictions are communicated to pilot Once in Mission 2 Once in Mission 3 Twice in Mission 4 Negotiation Navigator/pilot must negotiate airspeed/altitude Experimenter calls in new airspeed/altitude of current target to navigator Once in Mission 2 Once in Mission 3 Twice in Mission 4 Feedback Photographer does not provide feedback to navigator and pilot Experimenter calls in status of target photo to navigator and pilot and cuts all photographer communications Once in Mission 2 Twice in Mission 3 Twice in Mission 4 relative to the information-negotiation-feedback procedure (Figure 1). RESULTS Of the 26 teams that completed the experiment, there were 10 teams in the procedural condition and 8 teams each in the cross-training and perturbation conditions. Previous experiments in the UAV-STE exhibited low between-subjects power with a = .05 (M = .11, SD = .05) on tests of team performance due to small sample size. To increase statistical power, a significance level of a = .10 was used. For planned critical test mission comparisons, two conditions were pooled to form a comparison against a single condition. These planned comparisons also served to increase power. Team Performance Team performance results are summarized in Table 3 and graphed in Figure 2. Team performance was analyzed using a 3 (training) × 9 (mission) mixed ANOVA. The main effect of mission was significant, F(8, 184) = 22.14, p < .001, MSE = 3535.25, η2 = .53. No other effects in the omnibus test were significant. A repeated contrast on mission revealed that performance increased significantly during initial performance acquisition until Mission 4. There was a significant drop in performance at Mission 6, after the retention interval. Performance then improved significantly as teams reacquired the task, until Mission 8. Task reacquisition was followed by a significant drop in performance at Mission 9, the high-workload mission. These results reinforce Missions 5, 6, and 9 as critical missions. Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 302 April 2010 - Human Factors TABLE 3: Mean Team Performance by Training Condition Mission Cross-Trained Perturbation Procedural 1 2 3 4 5 6 7 8 9 345.04 (65.80) 383.18 (72.89) 422.58 (74.39) 450.54 (77.71) 446.40 (64.41) 435.99 (54.48) 477.79 (77.32) 513.25 (70.94) 389.13 (76.39) 342.78 (54.23) 409.33 (80.94) 463.39 (80.69) 483.76 (59.83) 500.37 (50.93)* 380.30 (166.10) 471.77 (75.38) 547.06 (47.86) 442.24 (36.83)* 316.92 (78.88) 373.90 (65.65) 439.92 (54.71) 447.83 (54.26) 469.63 (46.92) 383.76 (100.91) 421.38 (86.88) 502.47 (58.60) 372.02 (46.00) Note. Standard deviations in parentheses. *p < .10. Reacquisition 600 Critical Test 1 First Roadblock Acquisition Team Performance 550 Critical Test 3 High Workload Critical Test 2 Retention 500 450 400 Cross-Training Procedural 350 Perturbation 300 1 2 3 4 5 6 7 8 9 Mission Figure 2. Team performance for each training condition across missions. Adapted from Gorman et al. (2007). Planned comparisons were performed for each of the three critical test missions to address which training condition resulted in the highest performance under novel conditions. As shown in Figure 2, perturbation-trained teams exhibited better mean performance in two of the three critical test missions (Missions 5 and 9), whereas cross-trained teams exhibited better mean performance in one of the critical test missions (Mission 6). Performance of the perturbation-trained teams at Mission 5 (M = 500.37, SD = 50.93) was significantly better than the other two conditions (M = 459.31, SD = 54.91), F(1, 24) = 3.23, p = .085, MSE = 2892.21, η2 = .12. Performance of perturbation-trained teams at Mission 9 (M = 442.24, SD = 36.83) was also significantly better than the other two conditions (M = 379.62, SD = 60.00), F(1, 24) = 7.37, p = .012, MSE = 2945.49, η2 = .24. The crosstraining performance advantage at Mission 6 was not significant, F(1, 24) = .76, p = .392, MSE = 8379.75, η2 = .03. Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Time to Overcome Roadblock (seconds) Training Adaptive Teams 303 Interpositional Knowledge 800 Cross-Trained Perturbation Procedural 700 600 500 400 300 200 100 0 5 (First Roadblock) 6 (Retention) 9 (High Workload) Mission Figure 3. Time to overcome roadblocks by training condition across the critical missions (error bars represent 90% confidence intervals). Response Time to Novel Events Time to overcome roadblocks was analyzed with a 3 (training) × 3 (critical mission) mixed ANOVA. One observation was missing from the cross-training condition. There was a significant main effect of mission, F(1.53, 33.59) = 119.82, p < .001, MSE = 16018.63, η2 = .85 (GreenhouseGeisser correction used). The significant mission effect was attributed to differences in the difficulty of roadblocks. Therefore, no further analyses were performed to isolate that effect. The main effect of training was also significant, F(2, 22) = 3.53, p = .047, MSE = 19460.72, η2 = .24 (Figure 3). The planned comparisons at the critical missions revealed that procedural-trained teams were significantly slower to overcome roadblocks (M = 218.70, SD = 94.34) than were teams in the other two conditions (M = 146.67, SD = 103.52) at Mission 5, F(1, 23) = 3.11, p = .091, MSE = 10005.89, η2 = .12. Teams with procedural training were also significantly slower (M = 656.00, SD = 191.01) than those in the other two conditions (M = 533.07, SD = 98.35) at Mission 6, F(1, 23) = 4.50, p = .045, MSE = 20164.71, η2 = .16. The same comparison at Mission 9 was not significant. The Training × Critical Mission interaction was not significant. Analysis of the measure for time to overcome roadblock in the noncritical missions (i.e., Missions 7 and 8) did not reveal any significant differences. Interpositional teamwork and taskwork knowledge results are summarized in Table 4. Interpositional teamwork and taskwork knowledge were separately analyzed with 3 (training) × 2 (session) mixed ANOVAs. The taskwork ANOVA did not reveal any significant differences. There was a significant interaction effect for interpositional teamwork knowledge, F(2, 23) = 2.70, p = .089, MSE = .01, η2 = .19 (Figure 4). Pooled comparisons revealed that cross-training (M = .87, SD = .07) led to significantly higher interpositional teamwork knowledge compared with the other two conditions (M = .78, SD = .08) at Session 2, F(1, 24) = 7.04, p = .014, MSE = .01, η2 = .23. The same comparison for Session 1 was not significant. The main effect of session for interpositional teamwork knowledge was also significant, F(1, 23) = 6.24, p = .02, MSE = .01, η2 = .21. Although teams in all training conditions exhibited some increase in interpositional teamwork knowledge across sessions, teams in the cross-training condition exhibited a significantly greater increase. DISCUSSION Perturbation-trained teams significantly outperformed teams in the other conditions in two out of three critical test missions, and their response times to overcome novel roadblock events were roughly equivalent to cross-trained teams. These results lend support to our first hypothesis that perturbation training leads to high performance under novel conditions. The results suggest that something similar to the effects of practice condition variability (Schmidt & Bjork, 1992) contributed to transfer at the team level: Perturbation training allowed teams to generalize performance to novel conditions by forcing the teams to coordinate in new ways during task acquisition. However, whereas practice condition variability provides a range of task conditions for the individual learner, perturbation training induced coordination variability across team members during repetitions of the same task. By training teams to formulate and test new solutions to the problem of coordinating ground targets during task acquisition, perturbation training actively engaged Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 304 April 2010 - Human Factors TABLE 4: Mean Interpositional Teamwork and Taskwork Knowledge by Training Condition Measure Teamwork Session 1 Session 2 Taskwork Session 1 Session 2 Cross-Trained Perturbation Procedural .73 (.12) .87 (.07)* .78 (.12) .79 (.06) .74 (.11) .77 (.09) .47 (.04) .47 (.01) .46 (.05) .48 (.04) .48 (.05) .48 (.03) Note. Standard deviations in parentheses. *p < .10. Interpositional Teamwork Knowledge 1 0.9 Cross-Trained Perturbation Procedural 0.8 0.7 0.6 0.5 Session 1 Session 2 Figure 4. Interpositional teamwork knowledge by training condition across knowledge measurement sessions (error bars represent 90% confidence intervals). team processes that are needed to adapt to novel, but related, coordination problems in the posttraining environment. Because perturbation training builds on prior novelty, it may also have allowed teams to develop within a rich experiential learning environment (Kolb, 1984). Our second hypothesis was that cross-training would result in the highest levels of shared knowledge but that this would not necessarily result in the best performance under novel task conditions or the fastest response times to novel events. Support for that hypothesis was mixed. Cross-training resulted in greater shared teamwork knowledge in the second session but not in the first session. This is not a surprise given that the majority of condition-specific training took place after the first knowledge measurement session. However, shared taskwork knowledge did not change across experimental sessions, regardless of training condition. This may suggest a ceiling effect, such that taskrelated concept relatedness (e.g., the association between airspeed and altitude) does not change after initial participant training. Cross-training also resulted in highest mean performance at one of the critical missions (the retention test) and faster response times for overcoming roadblocks, although those differences were not significant. It is possible that with a larger sample size, or a less variable task environment, cross-training would have resulted in significant advantages. Marks et al. (2000) found that development of a shared mental model predicted performance under novel task conditions better than under routine task conditions. The retention and roadblock tests are novel conditions unique to our experiment, however, and further empirical work is needed to better understand the benefits of cross-training and shared knowledge under these conditions. The results support our third hypothesis that procedural training should result in the least adaptive teams. Procedural training is arguably the most prevalent form of training for coordinating highly critical team tasks, but its utility for training adaptive teams has been increasingly called into question (e.g., Ford & Schmidt, 2000; Grote, Kolbe, Zala-Mezö, Bienefeld-Seall, & Künzle, 2010; Stachowski et al., 2009). Procedural training need not be limited to a single, standardized coordination process; assuming that the space of possible future events is finite, procedures can be scripted for a variety of foreseeable contingencies. Nevertheless, given the current results, we argue that teams trained to automatically follow a standardized coordination procedure become rigid and slow to adapt to novel changes in highly dynamic task environments. Whereas proceduralization may be good for unchanging and foreseeable aspects of a task, in training adaptive teams, Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Training Adaptive Teams 305 there should be a match between interaction variability and the changing dynamics of the task environment. We examined team performance under three critical situations: adaptation to a novel event (roadblock), after a retention interval, and under high workload. This is not an exhaustive list of possibilities. There are many forms of adaptation for which a team could be trained (e.g., role structure adaptation; Lepine, 2005). The current results are intuitively plausible, however, given the nature of mechanisms of team adaptation currently found in the adaptive team literature, and extend the idea of process-based adaptability training. The building and maintenance of shared mental models are thought to support team adaptation (Burke, Stagl, Salas, Pierce, & Kendall, 2006; Stout et al., 1999; Waller, Gupta, & Giambatista, 2004). Indeed, cross-training was successful in building shared teamwork knowledge, and crosstrained teams exhibited potential for high performance under novel conditions. Parallel to the motivation for perturbation training, however, teams directly adapt via flexible interaction processes (Gorman et al., in press; Manser, Harrison, Gaba, & Howard, 2009; Stachowski et al., 2009; Waller, 1999). Kozlowski, Gully, Nason, and Smith (1999) suggested that teams adapt by selecting an appropriate form of interaction from a preexisting repertoire or by creating a new form. Approaches like perturbation training have the potential to broaden a team’s interaction repertoire not by prescribing preexisting forms of coordination but by allowing teams to exercise bottom-up organization of new coordination links. What are the practical implications for training adaptive teams, and how can we apply perturbation training? Simulation-based team training (Dorsey et al., 2009) would allow for the design of perturbations that focus on specific events, times, or interactions (Gorman, Cooke, & Duran, 2009). Simulation-based training can emphasize physical (equipment) fidelity or cognitive fidelity (how well the simulation exercises psychological processes required for that task; Goettle, Ashworth, & Chaiken, 2007). For perturbation training, cognitive fidelity should be emphasized in order to exercise the team interaction processes needed for the real-world task (Bowers & Jentsch, 2001). Another concern is the specifics of introducing perturbations: when, how many, what kind, and how often? Simulation-based training would be the ideal venue for perturbation training, and although approaches such as crew resource management (see Salas, Wilson, Burke, & Wightman, 2006) may use simulators to train for rare or novel events, our results suggest that more thought and research should go into identifying the types of team interaction experiences needed and the ideal timing of those experiences. What are the implications of the varying theoretical training motives—shared knowledge, proceduralization, flexible interactions—for team cognition? Prevalent in the team cognition literature is a distinction between knowledge and process and which contributes most to team effectiveness (e.g., Cooke, Gorman, & Winner, 2007). We submit that cross-training most directly impacts knowledge, that perturbation training most directly impacts process, and that procedural training may have little impact on either. The current study is not an unequivocal test of knowledge versus process accounts of team cognition, nor is it an exhaustive sampling of variations on procedural, perturbation, or cross-training in a variety of contexts. Nonetheless, the results do suggest that training focused on process may contribute something to team effectiveness that a knowledge-focused approach does not. CONCLUSION The details of team adaptation are not specified at the outset of a novel event. The details accrue gradually, during the process of adaptation, and there lies the problem for training adaptive teams: They must be able to decide, plan, think, and act under conditions never experienced. Adaptation is the altering of structure in accordance with environmental change and, under many circumstances, is not a purely topdown, knowledge-driven process. Teams should be provided opportunities to exercise adaptive competency using not only top-down (knowledgefocused) training but also bottom-up (processoriented) training. Perturbing coordination as team members interact is one means of eliciting the bottom-up, process-oriented flexibility that Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 306 April 2010 - Human Factors teams need in order to adapt. Future research should continue to explore mechanisms of flexible team interaction and how teams use them to adapt to the pressures of highly dynamic, high-stakes work environments. ACKNOWLEDGMENTs This research was funded by Air Force Office of Scientific Research Grant FA9550-04-1-0234 and Air Force Research Laboratory Grant FA8650-04-6442; additional support came from National Science Foundation Grant BCS 0447039. The authors would like to thank the following individuals who contributed to this research: Dee Andrews, Christy Caballero, Olena Connor, Jasmine Duran, Preston Kiekel, Harry Pedersen, Steven Shope, Amanda Taylor, and Jennifer Winner. Some team performance results were previously reported in a technical report (Cooke, Gorman, Pederson, et al., 2007) and a conference proceeding paper (Gorman et al., 2007). REFERENCES Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 185–205). Cambridge, MA: MIT Press. Blickensderfer, E., Cannon-Bowers, J. A., & Salas, E. (1998). Cross-training and team performance. In J. A. Cannon-Bowers & E. Salas (Eds.), Making decisions under stress: Implications for individual and team training (pp. 299–311). Washington, DC: American Psychological Association. Bowers, C. A. & Jentsch, F. (2001). Use of commercial, off-theshelf, simulations for team research. In E. Salas (Ed.), Advances in human performance and cognitive engineering research (Vol. 1., pp. 293–317). Amsterdam, Netherlands: Elsevier Science. Burke, C. S., Stagl, K. C., Salas, E., Pierce, L., & Kendall, D. (2006). Understanding team adaptation: A conceptual analysis and model. Journal of Applied Psychology, 91, 1189–1207. Cannon-Bowers, J. A., Salas, E., Blickensderfer, E., & Bowers, C. A. (1998). The impact of cross-training and workload on team functioning: A replication and extension of initial findings. Human Factors, 40, 92–101. Cannon-Bowers, J. A., Salas, E., & Converse, S. (1993). Shared mental models in expert team decision making. In N. J. Castellan (Ed.), Individual and group decision making (pp. 221–246). Hillsdale, NJ: Erlbaum. Cooke, N. J., Gorman, J. C., Duran, J. L., & Taylor, A. R. (2007). Team cognition in experienced command-and-control teams. Journal of Experimental Psychology: Applied, 13, 146–157. Cooke, N. J., Gorman, J. C., Pedersen, H. K., Winner, J., Duran, J., Taylor, A., Amazeen P. G., Andrews, D., & Rowe, L. (2007). Acquisition and retention of team coordination in commandand-control (Technical Report for AFOSR Grant FA9550-041-0234 and AFRL Award No. FA8650-04-6442). Washington, DC: Department of Defense. Cooke, N. J., Gorman, J. C., & Rowe, L. J. (2009). An ecological perspective on team cognition. In E. Salas, J. Goodwin, & C. S. Burke (Eds.), Team effectiveness in complex organizations: Cross-disciplinary perspectives and approaches (pp. 157–182). New York, NY: Taylor and Francis. Cooke, N. J., Gorman, J. C., & Winner, J. L. (2007). Team cognition. In F. Durso, R. Nickerson, S. Dumais, S. Lewandowsky, & T. Perfect (Eds.), Handbook of applied cognition (2nd ed., pp. 239–268). New York, NY: Wiley. Cooke, N. J., Kiekel, P. A., Salas, E., Stout, R. J., Bowers, C., & Cannon-Bowers, J. (2003). Measuring team knowledge: A window to the cognitive underpinnings of team performance. Group Dynamics: Theory, Research and Practice, 7, 179–199. Cooke, N. J., & Shope, S. M. (2005). Synthetic task environments for teams: CERTT’s UAV-STE. In N. Stanton, A. Hedge, K. Brookhuis, E. Salas & H. Hendrick (Eds.), Handbook of Human Factors and Ergonomics Methods (pp. 46-41-46-46). Boca Raton, FL: CRC Press. Dorsey, D., Russell, S., Keil, C., Campbell, G., Van Buskirk, W., & Schuck, P. (2009). Measuring teams in action: Automated performance measurement and feedback in simulation-based training. In E. Salas, G. F. Goodwin, & C. S. Burke (Eds.), Team effectiveness in complex organizations: Cross-disciplinary perspectives and approaches (pp. 351–381). New York, NY: Taylor and Francis. Entin, E. E. & Serfaty, D. (1999). Adaptive team coordination. Human Factors, 41, 312-325. Fiore, S. M., Salas, E., & Cannon-Bowers, J. A. (2001). Group dynamics and shared mental model development. In M. London (Ed.), How people evaluate others in organizations (pp. 309– 336). Mahwah, NJ: Erlbaum. Ford, J. K., & Schmidt, A. M. (2000). Emergency response training: Strategies for enhancing real-world performance. Journal of Hazardous Materials, 75, 195–215. Gaddy, C. D., & Wachtel, J. A. (1992). Team skills training in nuclear power plant operations. In R. W. Swezey & E. Salas (Eds.), Teams: Their training and performance (pp. 379–396). Norwood, NJ: Ablex. Gersick, C. J. G., & Hackman, J. R. (1990). Habitual routines in task-performing groups. Organizational Behavior and Human Decision Processes, 47, 65–97. Goettle, B. P., Ashworth, A. R. S., III, & Chaiken, S. R. (2007). Advanced distributed learning for team training in command and control applications. In S. M. Fiore & E. Salas (Eds.), Toward a science of distributed learning (pp. 93–117). Washington, DC: American Psychological Association. Gorman, J. C., Amazeen, P. G., & Cooke, N. J. (in press). Team coordination dynamics. In Nonlinear dynamics, psychology and life sciences. Gorman, J. C., Cooke, N. J., Amazeen, P. G., Winner, J. L., Duran, J. L., Pedersen, H. K., & Taylor, A. R. (2007). Knowledge training versus process training: The effects of training protocol on team coordination and performance. In Proceedings of the Human Factors and Ergonomics Society 50th Annual Meeting (pp. 382–387). Santa Monica, CA: Human Factors and Ergonomics Society. Gorman, J. C., Cooke, N. J., & Duran, J. L. (2009). Development of simulated team environments for measuring team cognition and performance. In D. Schmorrow, J. Cohen, & D. Nicholson (Eds.), PSI handbook of virtual environments for training and education (pp. 347–361). Westport, CT: Praeger Security International. Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Training Adaptive Teams 307 Gorman, J. C., Cooke, N. J., Pedersen, H. K., Winner, J. L., Andrews, D., & Amazeen, P. G. (2006). Changes in team composition after a break: Building adaptive command-and-control teams. In Proceedings of the Human Factors and Ergonomics Society 49th Annual Meeting (pp. 487–491). Santa Monica, CA: Human Factors and Ergonomics Society. Gorman, J. C., Cooke, N. J., & Winner, J. L. (2006). Measuring team situation awareness in decentralized command and control environments. Ergonomics, 49, 1312–1325. Grote, G., Kolbe, M., Zala-Mezö, E., Bienefeld-Seall, N., & Künzle, B. (2010). Adaptive coordination and heedfulness make better cockpit crews. Ergonomics, 53, 211–228. Hockey, G. R. J., Sauer, J., & Wastell, D. G. (2007). Adaptability of training in simulated process control: Knowledge- versus rulebased guidance under task changes and environmental stress. Human Factors, 49, 158–174. Kolb, D. A. (1984). Experiential learning: Experience as the source of learning and development. Englewood Cliffs, NJ: Prentice Hall. Kozlowski, S. W. J., Gully, S. M., Nason, E. R., & Smith, E. M. (1999). Developing adaptive teams: A theory of compilation and performance across levels and time. In D. R. Ilgen & E. D. Pulakos (Eds.), The changing nature of work and performance: Implications for staffing, personnel actions, and development (pp. 240–292). San Francisco, CA: Jossey-Bass. Lepine, J. A. (2005). Adaptation of teams in response to unforeseen change: Effects of goal difficulty and team composition in terms of cognitive ability and goal orientation. Journal of Applied Psychology, 90, 1153–1167. Luchins, A. S. (1942). Mechanization in problem solving. Psychological Monographs, 54(248). Manser, T., Harrison, T. K., Gaba, D. M., & Howard, S. K. (2009). Coordination patterns related to high clinical performance in a simulated anesthetic crisis. International Anesthesia Research Society, 108, 1606–1615. Marks, M. A., Sabella, M. J., Burke, C. S., & Zaccaro, S. J. (2002). The impact of cross-training on team effectiveness. Journal of Applied Psychology, 87, 3–13. Marks, M. A., Zaccaro, S. J., & Mathieu, J. E. (2000). Performance implications of leader briefings and team-interaction training for team adaptation to novel environments. Journal of Applied Psychology, 85, 971–986. Sauer, J., Burkolter, D., Kluge, A., Ritzmann, S., & Schuler, K. (2008). The effects of heuristic rule training on operator performance in a simulated process control environment. Ergonomics, 51, 953–967. Salas, E., DiazGranados, D., Klein, C., Burke, C. S., Stagl, K. C., Goodwin, G. F., & Halpin, S. M. (2008). Does team training improve team performance? A meta-analysis. Human Factors, 50, 903–933. Salas, E., Nichols, D., & Driskell, J. E. (2007). Testing three team training strategies in intact teams: A meta-analysis. Small Group Research, 38, 471–488. Salas, E., Wilson, K. A., Burke, C. S., & Wightman, D. C. (2006). Does crew resource management training work? An update, an extension, and some critical needs. Human Factors, 48, 392–412. Schmidt, R. A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225–260. Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3, 207–217. Schvaneveldt, R. W. (1990). Pathfinder associative networks: Studies in knowledge organization. Norwood, NJ: Ablex. Snook, S. A. (2002). Friendly fire: The accidental shootdown of U.S. black hawks over northern Iraq. Princeton, NJ: Princeton University Press. Stachowski, A. A., Kaplan, S. A., & Waller, M. J. (2009). The benefits of flexible team interaction during crises. Journal of Applied Psychology, 94, 1536–1543. Stout, R. J., Cannon-Bowers, J. A., Salas, E., & Milanovich, D. M. (1999). Planning, shared mental models, and coordinated performance: An empirical link is established. Human Factors, 41, 61–71. Turvey, M. T. (1990). Coordination. American Psychologist, 45, 938–953. Vaughan, D. (1996). The Challenger launch decision: Risky technology, culture, and deviance at NASA. Chicago, IL: University of Chicago Press. Volpe, C. E., Cannon-Bowers, J. A., Salas, E., & Spector, P. E. (1996). The impact of cross-training on team functioning: An empirical investigation. Human Factors, 38, 87-100. Waller, M. J. (1999). The timing of adaptive group responses to nonroutine events. Academy of Management Journal, 42, 127–137. Waller, M. J., Gupta, N., & Giambatista, R. C. (2004). Effects of adaptive behaviors and shared mental models on control crew performance. Management Science, 50, 1534–1544. Jamie C. Gorman received his PhD in cognitive psychology from New Mexico State University, Las Cruces, in 2006 and is a postdoctoral research associate at Arizona State University–Polytechnic and the Cognitive Engineering Research Institute in Mesa, Arizona. Beginning August 2010, he will be an assistant professor in the Human Factors Program at Texas Tech University. Nancy J. Cooke received her PhD in cognitive psychology from New Mexico State University, Las Cruces, in 1987 and is a professor in applied psychology at Arizona State University–Polytechnic and science director of the Cognitive Engineering Research Institute in Mesa, Arizona. Polemnia G. Amazeen received her PhD in experimental psychology from the University of Connecticut, Storrs, in 1996 and is an associate professor in psychology at Arizona State University, Tempe, and a faculty research associate at the Cognitive Engineering Research Institute in Mesa, Arizona. Date received: May 13, 2009 Date accepted: March 29, 2010 Downloaded from hfs.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Small http://sgr.sagepub.com/ Group Research Does Team Building Work? Cameron Klein, Deborah DiazGranados, Eduardo Salas, Huy Le, C. Shawn Burke, Rebecca Lyons and Gerald F. Goodwin Small Group Research 2009 40: 181 originally published online 6 January 2009 DOI: 10.1177/1046496408328821 The online version of this article can be found at: http://sgr.sagepub.com/content/40/2/181 Published by: http://www.sagepublications.com Additional services and information for Small Group Research can be found at: Email Alerts: http://sgr.sagepub.com/cgi/alerts Subscriptions: http://sgr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://sgr.sagepub.com/content/40/2/181.refs.html >> Version of Record - Mar 10, 2009 OnlineFirst Version of Record - Jan 6, 2009 What is This? Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Does Team Building Work? Cameron Klein Deborah DiazGranados Eduardo Salas Huy Le C. Shawn Burke Rebecca Lyons Small Group Research Volume 40 Number 2 April 2009 181-222 © 2009 SAGE Publications 10.1177/1046496408328821 http://sgr.sagepub.com hosted at http://online.sagepub.com University of Central Florida Gerald F. Goodwin Army Research Institute This research reports the results of a comprehensive investigation into the effectiveness of team building. The article serves to update and extend Salas, Rozell, Mullen, and Driskell’s (1999) team-building meta-analysis by assessing a larger database and examining a broader set of outcomes. Our study considers the impact of four specific team-building components (goal setting, interpersonal relations, problem solving, and role clarification) on cognitive, affective, process, and performance outcomes. Results (based on 60 correlations) suggest that team building has a positive moderate effect across all team outcomes. In terms of specific outcomes, team building was most strongly related to affective and process outcomes. Results are also presented on the differential effectiveness of team building based upon the team size. Keywords: team building; team performance; team development Teams of people working together for a common cause touch all our lives. From everyday activities like air travel, fire fighting, and running the United Way drive to amazing feats of human accomplishment like climbing Mt. Everest and reaching for the stars, teams are at the center of how work gets done in modern life. Kozlowski & Ilgen, 2006, p. 78 This quote, from a recent review of work-team effectiveness, exemplifies the central role that teams play in our lives. Although many labels have been applied to team-based forms of organizing (i.e., crews, teams, groups, and collectives), these entities are essential to the accomplishment of organizational goals. Indeed, there is ample support in the literature for the 181 Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 182 Small Group Research contention that team-based forms of organizing are beneficial both to organizations and to individuals. For example, Applebaum and Batt (1994) reviewed 12 large-scale surveys and 185 case studies of managerial practices and concluded that team-based work leads to improvements in organizational performance on measures of both efficiency and quality. They argued that team-based systems benefit workers because of the higher likelihood of job enhancement, autonomy, and skill development associated with these systems. However, the simple existence of a team-based organizing structure is not enough to ensure that positive outcomes will result. Teams must be nurtured, supported, and developed. The Motivation for Understanding the Efficacy of Team Building There are three motivations for understanding the efficacy of teambuilding interventions in organizations. First, team building is one of the most commonly applied group development interventions in organizations today. It is widely used and comes in many forms, including outdoor experiential activities and indoor group process discussions. However, no one is quite sure how and why these interventions work, or if they even work at all. Considering the vast sum of money directed toward the development of teams in organizations, it is important that practitioners (and researchers) gain a better understanding of the effectiveness and boundary conditions of team building. Second, as there are many options available to organizations in the pursuit of improved teamwork, it is important to determine whether team building is a worthy choice. Some of these interventions are organizational Authors’ Note: This work was partially supported by funding from the U.S. Army Research Institute for the Behavioral and Social Sciences (Contract W74V8H-04-C-0025) and grant SES0527675 from the National Science Foundation, awarded to Glenn Harrison, Stephen M. Fiore, Charlie Hughes, and Eduardo Salas. All opinions expressed in this article are those of the authors and do not necessarily reflect the official opinion or position of the University of Central Florida, the Department of the Army, the Department of Defense, or the National Science Foundation. Portions of the article were presented at the 2nd Annual Conference of the Interdisciplinary Network for Group Research (INGRoup) held at Michigan State University, in Lansing, Michigan. Correspondence concerning this article should be addressed to: Dr. Eduardo Salas, Institute for Simulation & Training, University of Central Florida, 3100 Technology Parkway, Suite 132, Orlando, FL 32826, USA. Tel: (407) 882-1325; fax: (407) 882-1550; e-mail: esalas@ist.ucf.edu. Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 183 or structural in nature and do not specifically target team member interactions (e.g., job redesign, selection systems, and group incentive and performance management programs). In contrast, team-development interventions—those whose quest is to directly impact the functioning and effectiveness of work teams—provide the focus for the current research integration. These interventions, when properly conducted, can have a positive impact on organizations. Consider a research study conducted by Macy and Izumi (1993), who analyzed 131 studies of organizational change. They found that interventions with the largest effects upon financial measures of organizational performance were team-development interventions. That is, of all organizational interventions, those that focus on team development had the largest effect on measures of financial performance. Third, beyond financial performance, it is widely understood that team developmental interventions are key mechanisms that may be used to facilitate team effectiveness (Noe, 2002). Therefore, it is important to understand how these interventions are most effective. At a general level, team-development interventions may include some form of team-training or team-building activities. Although both types of team-development interventions are designed to improve team functioning and effectiveness (especially when the science of individual and team training is utilized; Salas & Cannon-Bowers, 1997), team training and team building differ in important ways (Tannenbaum, Beard, & Salas, 1992). Team training is skill-focused (i.e., it is focused on gaining specific competencies), typically includes a practice component, and is done in context. It is generally formal and systematic. Team building, on the other hand, does not target skill-based competencies, is not systematic in nature, and is typically done in settings that do not approximate the actual performance environment. For our purposes, we define team building as a class of formal and informal team-level interventions that focus on improving social relations and clarifying roles, as well as solving task and interpersonal problems that affect team functioning. Team building works by assisting individuals and groups to examine, diagnose, and act upon their behavior and interpersonal relationships (Schein, 1969, 1999). In light of the above, this article investigates the efficacy of team building. We begin with a brief review of conceptual issues and methodological issues in team building. Next, the hypotheses for this research are presented. Finally, the results of several meta-analytic integrations are discussed. First, however, is the rationale for the present research. Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 184 Small Group Research Understanding the Need for an Update It is an unfortunate indictment of the literature and practice in this area that we are still searching for answers to questions posed by Beer (1976) and Salas, Rozell, Mullen, and Driskell (1999). Namely, does team building result in positive outcomes? Why? Under what conditions? Upon a careful review of the extent literature, it is clear that these questions need to be examined more closely to clarify our understanding of the effectiveness and boundary conditions of team building. Thus, this article reports the findings from a series of meta-analytic investigations of research on the efficacy of team-building interventions to update and extend the current state of knowledge in the team-building domain. In total, this research will examine three moderator variables in addition to making an overall assessment of the efficacy of team-building interventions. The current replication is meant to be systematic, rather than direct in nature (e.g., Aronson, Ellsworth, Carlsmith, & Gonzales, 1990). That is, rather than replicating the exact analyses from the study by Salas and colleagues, the current research will replicate and extend the earlier study in an effort to resolve ambiguities and provide new insights for organizational stakeholders and academicians alike concerning the effectiveness of team building. Although the empirical evidence on team-building interventions is limited, a critical investigation of the available literature is warranted for several reasons. First, since the 1990s there has been an increasing incursion of team-building interventions in organizations. Some of the most recent trends in team building have taken these interventions into the kitchen and even the wilderness. Second, many practitioners feel that these interventions are useful. However, a more careful exploration of the utilities and strengths of these interventions would benefit practitioners now and would benefit the development of these interventions in the long run. Finally, there are enough data available to form estimates regarding some of the variables that may moderate the impact a team-building intervention may have on team outcomes. There is a general consensus for the idea that the science of team training can be applied to enhance team functioning across organizations (e.g., Kozlowski & Ilgen, 2006; Salas & Cannon-Bowers, 1997). At the same time, reviews of team effectiveness have noted the inconsistent findings for the effectiveness of team-building interventions. Despite these inconsistent findings, researchers such as Kozlowski and Ilgen have acknowledged the potential that these interventions may have on shaping team development and on improving team effectiveness. Given the increasing frequency of Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 185 team-building interventions, if researchers do not catch up and start doing this research, an important window might be missed to constructively shape the important practice in this area. Thus, examining team building further— a topic that has not received recent sufficient attention in the literature—is critical. In the next section, a number of conceptual and methodological issues in team building are discussed. Team Building Conceptual Issues in Team Building Originally designed as a group process intervention (e.g., Schein, 1969, 1999) for improving interpersonal relations and social interactions, team building has evolved to also include a concern for achieving results, meeting goals, and accomplishing tasks (Payne, 2001). In the late 1990s, Salas and colleagues (1999) described team-building interventions as extremely popular and common. According to Beer (1976), there are four basic approaches to team building, including: (a) a goal-setting, problem-solving model; (b) an interpersonal model; (c) a role model; and (d) the Managerial Grid (Blake & Mouton, 1964) model. However, this initial conceptualization has since been reconsidered. Refinements to this four-pronged system began with the Managerial Grid model being dropped as a distinct team-building approach. In addition to dropping the Managerial Grid model, modern conceptualizations have separated the goal-setting and problem-solving approaches, using Buller’s (1986) problem-solving component as a distinct approach. As discussed by Buller, team-building models rarely exist in pure form. That is, the interventions reported in the literature usually involve elements from several or all of the models. As a result, he proposed a general problem-solving model that follows Dyer’s (1977) problem-solving framework. This model incorporates a focus on task or interpersonal issues, goal setting, and role clarification, depending on the nature of the specific problems identified for the group under investigation. Moreover, the problem-solving approach to team building is said to subsume each of Beer’s (1976) components, and, as perhaps evident by its title, emphasizes the identification of major problems in the team. All told, these modifications have added clarity to the investigation and implementation of team building. Unfortunately, this newfound conceptual clarity may not have come soon enough for many investigators who had previously sought to assess the efficacy of team building. Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 186 Small Group Research Perhaps because of the conceptual confusion that existed in this area, many reviews of team building did not include the same articles. For example two noteworthy reviews (DeMeuse & Liebowitz, 1981; Woodman & Sherwood, 1980) had only 14 studies in common out of a total of 66 studies reviewed. A subsequent review by Buller (1986) included only 9 studies, 3 in common with Woodman and Sherwood (1980) and 6 in common with DeMeuse and Liebowitz. One reason for the apparent lack of consistency in the articles chosen for reviews is that team building has been an ill-defined concept (Buller, 1986). At the time of these early reviews there was no agreed-upon operational definition of the intervention. Taken together, there is now a consensus position that there exist four distinct models of team building. Although combinations of these approaches are common, the models include goal-setting, developing interpersonal relations, clarifying roles, and creating additional capacity for problem solving (Beer, 1976; Buller, 1986; Dyer, 1987; Salas et al., 1999). Table 1 describes each of the four models or components of team building in more detail. Despite the recently established consensus on team-building components, there have been other problematic issues that have persisted for researchers and practitioners of this topic. Specifically, many early efforts were plagued by a number of methodological issues. A few of these are discussed in the next section. Methodological Issues in Team Building Early reviews of team building described both a lack of extensive research on the issue and trepidation concerning the methodological rigor of published studies (Buller, 1986; DeMeuse & Liebowitz, 1981; Tannenbaum et al., 1992; Woodman & Sherwood, 1980). That is, even if one could get beyond the disagreement concerning operational definitions of team building, much of the previous team-building research is characterized by methodological flaws (i.e., study design and measurement issues). For example, Buller (1982) found that more than half of the reported team-building studies employed pre-experimental designs—designs that do not allow for causal inferences. Also problematic, there was often no attempt to disentangle the effects of team building from other interventions that may be in process within an organization. Although this can be a common problem for field research in general, it has been particularly problematic in the team-building domain. Tannenbaum and colleagues (1992) highlighted another methodological flaw in the team-building research. Specifically, they pointed out that there Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 187 Table 1 Models/Components of Team Building Component Goal setting Salas, Rozell, Mullen, & Driskell, 1999 Salas, Priest, & DeRouin, 2005 Emphasis: Setting objectives and development of individual and team goals. Team members: Become involved in action planning to identify ways to achieve goals. Designed to strengthen team member motivation to achieve team goals and objectives. By identifying specific outcome levels, teams can determine what future resources are needed. Individual characteristics (e.g., team member motivation) can also be altered by use of this intervention. Interpersonal Emphasis: Increasing teamwork Based on the assumption that teams relations skills (i.e., mutual supportiveness, with fewer interpersonal conflicts communication, and sharing function more effectively than of feelings). teams with greater numbers of Team members: Develop trust in interpersonal conflicts. one another and confidence Requires the use of a facilitator to in the team. develop mutual trust and open communication between team members. As team members achieve higher levels of trust, cooperation, and cohesiveness, team characteristics can be changed as well. Role Emphasis: Increasing communication Defines the team as comprising a set clarification among team members regarding of overlapping roles. their respective roles within These overlapping roles are the team. characterized as the behaviors that Team members: Improve are expected of each individual understanding of their own and team member. others’ respective roles and duties Can be used to improve team and within the team. individual characteristics (i.e., by reducing role ambiguity) and work structure by negotiating, defining, and adjusting team member roles. Problem Emphasis: Identifying Buller’s (1986) problem-solving solving major task-related problems component subsumes aspects from within the team. all of the components described by Team members: Become involved Beer (1976). in action planning, implement Team members practice setting goals, solutions to identify problems develop interpersonal relations, clarify team roles, and work to (continued) Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 188 Small Group Research Table 1 (continued) Component Salas, Rozell, Mullen, & Driskell, 1999 and to evaluate those solutions. Salas, Priest, & DeRouin, 2005 improve organizational characteristics through problem-solving tasks. Can have the added benefit of enhancing critical-thinking skills. has been a reliance on measuring the effectiveness of team-building interventions with process measures. Although implicitly appealing, improvements in processes can not always be linked to improvements in team performance (e.g., Porras & Wilkens, 1980). For example, team performance is typically fashioned by additional environmental and/or organizational characteristics and contingencies that are out of the volitional control of team members. Stated differently, team members are active participants in the enactment of team processes, but must also interact within the larger system to produce more distal performance outputs. Team processes may also be impacted by the larger organizational environment, but not likely to the same degree as team performance outputs. As a final methodological concern with previous team-building research, there has been an overreliance on subjective indicators of group or organizational performance criteria as dependent measures (Tannenbaum et al., 1992). Though this type of information may be interesting and relevant to measuring participant satisfaction and other affective outcomes that may be impacted by team-building interventions, it is often not concrete enough to allow for accurate predictions of the performance outcomes of team building. A Recent Advancement in Team-Building Research Despite the conceptual and methodological issues associated with evaluations of team building, one recent effort has represented advancement over previous reviews by empirically investigating the effectiveness of these interventions. Specifically, Salas and colleagues (1999) responded to a call by Buller (1986) to use the primary focus of the intervention (i.e., goal setting, interpersonal relations, role clarification, and problem solving) as a potential moderating variable. The results of their study failed Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 189 to indicate a relationship between the combined set of team-building interventions and team performance (r = .01, k = 16 effect sizes). Moreover, of the four components of team building, only role clarification proved to be effective, as judged by both objective (r = .71) and subjective (r = .75) accounts of performance. Interestingly, for objective and subjective measures there was a nonsignificant effect of goal setting (r = −.06 and −.11, respectively), interpersonal relations (r = −.38 and −.04, respectively), and problem solving (r = −.31 and .09, respectively). Also examined in this study were the potential moderating influences of the source of the criterion measurement of performance (i.e., objective vs. subjective), team size, and training duration. For objective measures of performance, there was no evidence of a relationship between team building and performance (r = −.04, k = 8); for subjective measures, there was a small positive relationship between team building and performance (r = .14, k = 8). Concerning team size, the results of their study suggested that the effects of team building on performance decreased as a function of the size of the team (r = −.34). Finally, there was a slight tendency for the effects of team building to decrease as a function of the duration of the intervention (r = −.20). Taken together, this study enhanced our understanding of the efficacy of team building. However, there were a number of limitations associated with it—limitations that now necessitate the need for additional inquiry. For example, the amount of data analyzed in this study was relatively modest. Specifically, the research findings presented were based on only 16 effect sizes, and thus, it is difficult to discern, with any degree of confidence, the actual effectiveness of these interventions. Moreover, the findings derived from the Salas and colleagues’ (1999) study did not necessarily reflect the findings from existing narrative reviews. Finally, their study left many questions unanswered regarding the potential moderating impact of other relevant variables. As an organizing tool provided to summarize the literature in this area, Table 2 provides a summary of five previous investigations into the efficacy of team building, and includes the number of articles reviewed, the years spanned, and other noteworthy features. In summary the current research was initiated to provide an updated meta-analysis of the team-building literature. Phrased in the form of questions, an examination of these issues will help clarify our understanding of the effectiveness of team building and lead to hypothesized relationships involving the effectiveness of team building, including an investigation of specific moderators. A model that serves to graphically illustrate these hypotheses is presented in Figure 1. Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 190 Small Group Research Table 2 Summary of Previous Team-Building Reviews Review Study Details Woodman & 30 articles Sherwood (1980) Qualitative data Summary of Findings TB elicits positive affective reactions. The linkage between TB and work group performance remains largely unsubstantiated. Years spanned: TB is more commonly 1964–1978 conducted with management teams than groups lower in organizational hierarchies. TB is more commonly conducted with intact and established work teams than new groups. Affective reactions as dependent measures are used more often than objective performance data. DeMeuse & 36 articles TB is described as having great Liebowitz promise for improving employee (1981) Qualitative attitudes, perceptions, behaviors, data and organizational effectiveness. Eighty-seven percent of the 68 Years spanned: evaluations indicated positive results.a 1962–1980 Due to the lack of rigorous research designs, firm conclusions concerning the effectiveness of these interventions could not be made. Buller 9 articles TB must be more carefully defined. (1986) More rigorous experimental Qualitative designs should be used. data There are numerous methodological flaws in previous TB studies. Years spanned: Therefore a clear assessment of the 1964–1981 TB and task performance relationship had yet to emerge. Noteworthy Features Distinguished between team development and T-group or sensitivity training. Made subjective assessments of the internal validity of the studies reviewed. Coded studies according to research design, sample size, multiple dependent variables, and duration of the TB intervention. Presented and described a general problem solving approachto TB that added to Beer’s (1976) four-component model. Argued that Beer’s classification is difficult to use in practice because TB programs usually involve elements from each of the models. Tannenbaum, 17 articles The quantity of TB research decreased; Presented a comprehensive Beard, & however, the quality had improved. model of team effectiveness Salas Qualitative Most studies used multiple components that continues to influence (1992) data in their TB interventions. research and theorizing More researchers began using in this field. Years spanned: behavioral and objective measures. 1980–1988 Presented evidence to cast doubt the connection between process and performance; TB is effective, but for only perceptions and attitudes. (continued) Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 191 Table 2 (continued) Review Study Details Summary of Findings Noteworthy Features Post-intervention strategies may be a key mechanism to ensure the long-term effectiveness of these interventions. Salas, Rozell, 11 articles No significant effect of TB on First known attempt to Mullen, & performance. empirically summarize the Driskell Quantitative A nonsignificant tendency for TB to effectiveness of TB. (1999) data result in lower performance when Assessed a number of measured objectively, but increase moderators, including team Years spanned: performance when subjective building component, team 1965–1990 measures were used. size, training duration, and The role clarification component was type of performance measure used (i.e., objective more likely to increase performance.b The effects of TB decreased as the size versus subjective). of the team increased. The effects of TB decreased as the duration of the intervention increased. Note: TB = team building. a. Many of the 36 studies reported evaluations of multiple dependent variables. b. This result should be interpreted with caution as it was based on a small number of effect sizes (it’s difficult to determine, but likely only three or four effect sizes were used in this calculation). Hypotheses Is Team Building Effective? We agree with Salas and colleagues (1999) that previous narrative reviews have frequently expressed the benefits that can result from team building (e.g., Buller, 1986; DeMeuse & Liebowitz, 1981; Sundstrom, DeMeuse, & Futrell, 1990, Tannenbaum et al., 1992; Woodman & Sherwood, 1980); however, there has been a lack of definitive, compelling evidence concerning the positive effects of team building on team performance. Specifically, previous qualitative reviews of the team-building domain have concluded that evidence of an effect of team building on performance was “inconclusive” (Buller, 1986), “unsubstantiated” (Woodman & Sherwood, 1980), “equivocal” (Tannenbaum et al., 1992), and “mixed” (Sundstrom et al., 1990). Meta-analytic results from one study have suggested there is no overall effect of team building on team performance (Salas et al., 1999). However, there was support for the role-clarification component of team building. Theoretically, one would assume that an intervention focused on Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 192 Small Group Research Figure 1 Theoretical Model Depicting Study Hypotheses improving team functioning would result in positive (as opposed to negative) outcomes. In addition, the moderate support suggested by several narrative reviews (e.g., Tannenbaum et al.) leads us to predict a positive overall effect of team building on team functioning. Thus, we present our first hypothesis: Hypothesis 1: Team building interventions will result in enhanced team outcomes. The result from testing this hypothesis will provide for a baseline judgment concerning the efficacy of this particular form of team-development intervention. The omnibus test will therefore allow for an overall assessment of the efficacy of team building, with all independent outcomes from primary studies combined for this analysis. Is Team Building More Effective for Some Outcomes Than Others? This framing question asks whether the combined set of team-building interventions is shown to be more useful for improving certain team outcomes than others. Specifically, does team building work better for cognitive outcomes (e.g., declarative knowledge of teamwork competencies), team Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 193 member affective outcomes (e.g., trust, team potency), team processes (e.g., coordination, communication), or team performance outcomes (e.g., volume of sales, productivity measures)? This division of team-building outcomes is similar to the commonly discussed cognitive, affective, and skill-based breakdown of general training outcomes (e.g., Kraiger, Ford, & Salas, 1993). However, for the current research, skill-based outcomes are further divided into two additional categories—team processes and more performance-related or productivity-related outcomes. The division and examination of team-building effectiveness based on specific outcomes is intended to help clarify the often disparate results seen in the literature on these interventions. For example, Woodman and Sherwood’s (1980) qualitative review of team building concluded that it was only useful for facilitating affective outcomes, not team performance. Equally pejorative to the position that team building results in improved team performance was the conclusion provided by Buller (1986), who upon reviewing team-building studies conducted through 1980, suggested that the relationship between team building and performance was inconclusive. In yet another review that added to the opaque nature of the efficacy of team building, DeMeuse and Liebowitz (1981) accurately noted that most of the early research on team building relied almost exclusively on perceptual ratings of the dependent variables being studied. More recent reviewers of the efficacy of team building have reported somewhat different (i.e., more positive) conclusions. For example, the integration of team-building research reported by Salas and colleagues (1999) has served, in many ways, as the focal point for the current research. And, although there was no significant overall effect of team building in their research, there was a small, yet significant, tendency for team building to increase performance when criteria were assessed with subjective measures. In another investigation into the efficacy of team building, Tannenbaum and colleagues (1992) reviewed team-building research conducted in the 1980s and found support for these interventions, especially when the outcome of interest was limited to team member perceptions or attitudes. Finally, Svyantek, Goodman, Benz, and Gard (1999) found that team building positively impacted work-group productivity. Specifically, the largest impact of team building was on productivity measures of cost-effectiveness, with a smaller influence on quantity and quality measures. In general, previous research that has investigated the effectiveness of team building for improving specific outcomes has been equivocal at best, especially considering performance or productivity measures. At the same time, there has been perhaps the greatest support for the efficacy of team Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 194 Small Group Research building for improving affective or attitudinal outcomes. Finally, to our knowledge there have been no reviews citing the correlation between teambuilding interventions and improvements in cognitive outcomes. The mixed results found in the literature concerning the effect of teambuilding interventions on team outcomes lead us to believe that moderators may exist. Therefore, team building impacts certain outcomes more than others. Based on the general characteristics of team building, (e.g., that team building develops interpersonal relations, mutual trust, and open communication between team members), it is likely that team building will have a greater effect on affective outcomes than any on other type of outcomes. As a process intervention, it makes theoretical sense that team building would result in enhanced team member affective outcomes. Finally, although the findings of more distal reviews of team building were taken into consideration, a decision was made to place more credence on recent research, which more often reported positive findings. These findings, along with a complete consideration of the expected benefits of team building, led us to Hypotheses 2a and 2b: Hypothesis 2a: Team building interventions will result in improved outcomes across each of the four outcome types. Hypothesis 2b: Team building will be most effective for improving affective outcomes. Does the Focus of Team Building Moderate Its Effectiveness? It is certainly important to be able to identify whether team building results in enhanced team functioning, but it is perhaps more informative to know which forms or components of team building are most effective. Advancements in theory in the last quarter century (e.g., Buller, 1986; Salas et al., 1999; Salas, Priest, & DeRouin, 2005) have allowed for the parceling of team-building interventions into four distinct foci (see Table 1). Unfortunately, the single existing empirical integration on the relative efficacy of different components of team building (i.e., Salas et al.) did not provide encouraging results. Combining subjective and objective estimates, there was a slight (nonsignificant) tendency for goal-setting (r = −.16), interpersonal relations (r = −.06), and problem-solving (r = −.05) models of team building to result in decreased performance. It was only the role-clarification component that appeared to be effective for improving performance (r = .76). However, upon examination of their data set, it Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 195 appears this positive finding for role clarification was supported by the results from only three studies and three effect sizes. Thus, caution and further investigation are warranted before concluding that the role-clarification component of team building is superior to the other forms. Nonetheless, because the role-clarification component of team building is designed to relieve role stress as created by role ambiguity or role conflict, it is reasonable to anticipate that a significant improvement in team functioning should result. In addition, the role-clarification component of team building emphasizes communication among team members, and thus it is likely that an increase in the level and quality of communication between team members will impact their effectiveness. Unlike the other forms of team building, improvements in role clarity and communication are expected to produce more lasting benefits in terms of team functioning (as assessed through an analysis of the combined set of team outcomes). Combining this theoretical rationale with the preliminary findings of Salas and colleagues (1999) concerning the role-clarification component of team building, it is our belief that a team-building intervention that utilizes a role-clarification focus will provide the most benefit to team functioning. At the same time, it is our view that any carefully thought out team-building intervention should have at least some positive impact on team members. Thus, we also expect that team building that focuses on interpersonal relations, goal setting, or problem solving will also prove useful, at least for the short-term benefits that are typically assessed. Taking these dual considerations into perspective, we present our next set of hypotheses: Hypothesis 3a: Each of the four components of team building will demonstrate a moderate level of effectiveness for improving team functioning. Hypothesis 3b: The role clarification component of team building will be most effective for improving team functioning. Is the Effectiveness of Team Building Moderated by Team Size? This research will also investigate whether the effectiveness of team building is moderated by the size of the team. At a basic level, “the resources available on a team result from how many people are on it” (Hambrick & D’Aveni, 1992, p. 1449). Sundstrom, McIntyre, Halfhill, and Richards’s (2000) review of 83 field studies and experiments conducted with work Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 196 Small Group Research groups suggested that the average group size was 11 members. Moreover, Halebian and Finkelstein (1993) have suggested that team size is synonymous with cognitive capability. Providing support for this assertion, Bantel and Jackson (1989) found that larger teams generally have a greater reservoir of cognitive resources than smaller teams. There is little doubt that there are benefits to having medium- to large-sized teams available to perform work tasks rather than smaller teams. However, research has also found that larger teams can facilitate the enactment of other, less desirable, group phenomenon, including the participation leadership effect, the in-group bias effect, the cohesiveness performance effect, and the well-known groupthink effect (cf. Mullen, Anthony, Salas, & Driskell, 1994; Mullen, Brown, & Smith, 1992; Mullen & Copper, 1994; Mullen, Salas, & Driskell, 1989). From other research, we know that information pooling is critical to teambased decision making. That is, the size of a team may impact the effectiveness with which a team pools common and unique information. For example, Stasser, Vaughan, and Stewart (2000) discussed the tendency for group members to discuss shared information rather than the unique information that is held by team members—an issue that becomes more problematic as teams increase in size. This lack of attention to unique information held by individual team members can lead to mal-informed decisions and occasionally even detrimental results. Salas and colleagues (1999) presented the only existing research integration that has assessed the relative efficacy of team building for teams of different sizes. They found that the effects of team building on performance decreased as a function of the size of the team, both in objective measures and in subjective measures of performance. These authors concluded that any positive effect of team building is most likely to prevail only in small teams. Similarly, others have argued that as group size increases, members’ liking for the group (Indik, 1965) and performance (Mullen, 1987) tend to decrease. Taking these findings into consideration, the current research examined team size as a moderator of the effectiveness of team-building efforts on team performance. In this study, we seek to replicate Salas and colleagues’ findings. However, rather than simply correlating the database of effect sizes with their associated team sizes, the current research will examine the efficacy of team building for three distinct subgroup classifications of team size: small teams (i.e., less than 5 members); medium-sized teams (i.e., 5 to 10 members); and large teams (i.e., greater than 10 members). In addition, although there is some reason to expect that larger teams have an increased cognitive capacity with which to perform tasks (e.g., Bantel & Jackson, 1989), the other negative issues often associated with increased group size are expected to be the overwhelming influences on the performance of teams. Therefore, it is reasonable to suggest that larger teams Downloaded from sgr.sagepub.com at LIBERTY UNIV LIBRARY on August 17, 2012 Klein et al. / Team Building 197 are already performing at a lower level than medium or small teams, and would therefore exhibit enhanced benefits from team-building interventions. Stated differently, as teams increase in size, it is more likely they will show substantial benefits from team building. This expectation, although supported by theory, is in direct contradiction to the findings reported by Salas and colleagues. The following hypothesis is proposed to investigate this assertion. Hypothesis 4: Large teams will show greater benefits from team building than small- or medium-sized teams. Method Literature Search A comprehensive literature review was conducted to identify published and unpublished studies relevant to the effects of team building on the four team outcomes. As a starting point for our literature review, we conducted an online search of the Defense Technical Information Center (DTIC), Google Scholar ©, and the most common academic search databases (e.g., PsycINFO). The particular keywords that were used included, but were not limited to, the following terms: team building, team development, team goal setting, interpersonal relations, problem solving, role clarification, group building, and group development. In addition to these search techniques, key articles relevant to team building were inspected by hand for additional, potentially useful primary studies (e.g., Buller, 1986; DeMeuse & Liebowitz, 1981; Salas et al., 1999; Tannenbaum et al., 1992; Woodman & Sherwood, 1980). This ancestry approach was later extended to each of the articles that were included in the database in an effort to ensure that no fugitive studies had been overlooked. In the end, the literature search process resulted in 103 articles being identified for potential inclusion in the database. Criteria for inclusion. Specific limits were placed on the search to include articles published from 1950 to 2007. This date range was selected because previous reviews had not uncovered any published or unpublished teambuilding evaluations prior to 1950. Moreover, studies were include...
Purchase answer to see full attachment