Enhancing the Effectiveness of Team Science (2015)

1

Introduction

The past half-century has witnessed a dramatic increase in the scale and complexity of scientific research that has yielded exciting discoveries about natural phenomena and an array of practical applications, improving human health and the quality of life while fueling the growth of dynamic industries, such as pharmaceuticals, biotechnology, personal computing, advanced manufacturing, and software development.

The growing scale of science has been accompanied by a dramatic shift toward collaborative research referred to as “team science” defined further below. Studying the corpus of 19.9 million research articles across the fields of science and engineering, social science, and arts and humanities (Web of Science) and 2.1 million patent records (National Bureau of Economic Research) for more than five decades, Wuchty, Jones, and Uzzi (2007) discovered that the propensity for teamwork is greatest in the life and physical sciences but is also rapidly increasing in the social sciences. The authors found that 80 percent of all science and engineering publications were written by teams of two or more authors in 2000. The Committee on the Science of Team Science updated the database and trend analysis to find that the share of all papers written by two or more authors increased to 90 percent by the year 2013 (see Figure 1-1).

Wuchty, Jones, and Uzzi (2007) also found that the size of science and engineering authoring teams consistently expanded over the period, from a mean of less than 2 members in 1960 to 3.5 members in 2000. In a follow-up study, Jones, Wuchty, and Uzzi (2008) found that the rapid growth in team-based publications was due to an increase in publications by authors from multiple institutions, showing that team-based research increasingly crosses institutional and geographic boundaries.

FIGURE 1-1 Percentage of publications authored by more than one individual, 1960–2013.

PURPOSE OF THIS REPORT

Although team science is growing rapidly, it can be more challenging than solo science. For example, the increasing size of research teams and groups (Wuchty, Jones, and Uzzi, 2007) brings greater scientific expertise and more advanced instrumentation to a research question but also increases the amount of time required for communication and coordination of work among a greater number of individuals (see further discussion below). Given the growth of team science, evidence-based guidance is needed for addressing the challenges associated with these approaches and achieving their potential to more rapidly solve scientific and societal problems. To provide such guidance, the National Science Foundation (NSF) requested the National Research Council (NRC) to convene an expert committee and address the charge presented in Box 1-1. The study is also supported by Elsevier.

To address this charge, the committee identified, assembled, and re-

viewed many sources of relevant scientific research. When focusing on individual- and team-level factors, the committee drew primarily on the robust evidence on teams in contexts outside of science, supplemented by the emerging evidence from the new interdisciplinary field of the science of team science. When focusing on organizational- and institutional-level factors, it drew on leadership literature, case studies of geographically distributed teams and larger groups of scientists and other professionals, business management literature, sociology, economics, and science policy

studies. The committee’s analysis of organizational and institutional factors was also supplemented by the emerging evidence from the science of team science, which focuses not only on the team level, but also on the organizational, institutional, and policy levels. This report is the culmination of an intensive study conducted to determine what is currently known about the processes and products of team science, and the circumstances under which investments in team-based research are most likely to yield intellectually novel discoveries and demonstrable improvements in contemporary social, environmental, and public health problems.

DEFINING KEY TERMS

To create a framework for this study, the committee first defined the activity of team science and the groups that carry it out (see Box 1-2). The committee’s definitions reflect prior research that has defined a “team” as two or more individuals with different roles and responsibilities, who interact socially and interdependently within an organizational system to perform tasks and accomplish common goals. Because this prior research

has focused on small teams typically including 10 or fewer members, similar in size to most science teams, we refer to a group of 10 or fewer scientists as a “science team.” Recognizing that what is important for successful collaboration changes dramatically as the number of participants grows, we refer to groups of more than 10 scientists as “larger groups of scientists” or simply “larger groups.”

Although an individual investigator can master and integrate knowledge from diverse disciplines—for example, physicist Albert Einstein used mathematics, specifically Riemann geometry to create his new General Theory of Relativity—this process has become more difficult over the past four decades, because of the rapid growth of specialized knowledge in all fields of science and engineering (Jones, 2009). A scientist interested in investigating questions that require knowledge beyond her or his narrow specialization may prefer to team up with colleagues to obtain complementary expertise, rather than spending years mastering another discipline.

Science teams and larger groups vary in the extent to which they include or integrate the knowledge of experts from different disciplines or professions to achieve their scientific and, when relevant, translational goals. These varying degrees of integration have been classified as unidisciplinary, multidisciplinary, interdisciplinary, and transdisciplinary research approaches (see Figure 1-2). Unidisciplinary research relies on the methods, concepts, and approaches of a single discipline. In multidisciplinary research, each discipline makes separate contributions in an additive way. Interdisciplinary research integrates “information, data, techniques, tools, perspectives, concepts, and/or theories from two or more disciplines . . . to advance fundamental understanding or to solve problems” (National Academy of Sciences, National Academy of Engineering, and Institute of Medicine, 2005, p. 26). Interdisciplinary research has grown over the past three decades (Frickel and Jacobs, 2009; Porter and Rafels, 2009), reflecting the need for multiple disciplinary perspectives to address complex scientific and societal problems. Transdisciplinary research integrates but also transcends disciplinary approaches, as follows (Stokols, Hall, and Vogel, 2013, p. 5):

Some, but not all transdisciplinary research projects emphasize translation of research findings into practical solutions to social problems and include societal stakeholders (e.g., health professionals, business representatives) to facilitate this translation.

FIGURE 1-2 Levels of cross-disciplinary integration.
SOURCE: Hall (2014).

Since the 1980s, some parts of the scientific community have increased their use of transdisciplinary research approaches as a promising way to gain new scientific and technical insights on complex phenomena and speed application of these insights. For example, “convergence” integrates expertise from the life, physical, computational, and other sciences within a network of academic, industry, clinical, and funding partnerships to address scientific and societal challenges (National Research Council, 2014). In another example, the field of transdisciplinary sustainability studies brings together environmental scientists, policy makers, citizens, and industry representatives to frame and address multifaceted environmental challenges (Huutoniemi and Tapio, 2014). To illustrate these varying approaches to disciplinary integration, Box 1-3 provides examples from archaeology.

KEY FEATURES THAT CREATE CHALLENGES FOR TEAM SCIENCE

Based on its review of the research evidence, information from team science practitioners, and its own expert judgment, the committee identified seven features that create challenges for team science. A given team or group may need to incorporate one or more of these features to address

its particular research goals, but the features also pose challenges that are important to carefully manage. They include (1) high diversity of membership; (2) deep knowledge integration; (3) large size; (4) goal misalignment with other teams; (5) permeable team and group boundaries; (6) geographic dispersion; and (7) high task interdependence.

These features are based on levels or degrees within the team science dimensions shown in Table 1-1. The dimensions reflect variations in composition, size, and other facets of team science and do not necessarily introduce significant challenges for a team science project. However, we

TABLE 1-1 Dimensions of Team Science

characterize certain levels or degrees along each dimension (e.g., large size) as key features that create challenges for team science, increasing the need for strategies to mitigate such challenges.

Although each team science project is unique in the extent to which it is characterized by these features, as a given project incorporates more features—for instance, the inclusion of more disciplines or large size—so do the accompanying challenges, and the imperative for better understanding how the interacting features influence research processes and outcomes, to enhance the success of the project. Science teams and larger groups are increasingly likely to incorporate one or more of these seven features because they are needed to address complex scientific and societal problems. For example, greater diversity of membership may be needed to answer particularly complex scientific questions or a large group of scientists may be needed to maximize the benefits of an investment in large instrumentation. However, these features may not always be necessary. Therefore, scientists and funders engaged in designing team science projects need not necessarily include highly diverse members or large numbers of participants (Vermeulen et al., 2010), as the costs may outweigh the benefits (Cummings et al., 2013). Rather, strategically considering the nature of the scientific problem, the readiness of the science, and other factors are important to determine the best approach and scale of a research activity.

Next, we discuss each of the seven features in more detail, with an example and more general discussion. The examples are summarized in Table 1-2.

TABLE 1-2 Key Features, Goals, and Potential Challenges of Team Science

High Diversity of Membership

The members of a science team or group may come from different disciplines, research institutions, or nations. When relevant, the members may include community or industry stakeholders (e.g., doctors or product development specialists) to facilitate the research and/or its translation into practical applications. The members may be diverse in age, gender, culture, and other demographic characteristics. For example, the Social Environment, Stress, and Health project supported by the National Institutes of Health used a community-based participatory research approach to understand relationships among neighborhood and community factors, behavioral and biological responses, and breast cancer among women living on Chicago’s South Side (Hall et al., 2012a). The investigators, including natural and social scientists, conducted focus groups to learn about the beliefs, attitudes, and concerns of community members regarding breast cancer. Focus group members who were particularly committed to the research were invited to form a community advisory board as an active partner in the project. The newly evolved group, including scientists and stakeholders, worked with the community to share the research findings and identify and rank translational “action steps” to address them. Developing messages about wellness for 12- to 16-year-olds on the South Side was ranked as the most

important action, a translational focus that would not have occurred to the investigators working by themselves.

A key assumption underlying the formation of interdisciplinary and transdisciplinary team science projects is that the inclusion of individuals with diverse knowledge, perspectives, and research methods will lead to scientific or translational breakthroughs that might not be achieved by a more homogenous group of individuals (e.g., National Academy of Sciences, National Academy of Engineering, and Institute of Medicine, 2005; Fiore, 2008). Research on work groups and teams provides some support for this assumption, suggesting that including individuals with diverse knowledge, expertise, and experience can increase group creativity and effectiveness but only if group members draw on each other’s diverse expertise (Ancona and Caldwell, 1992; Stasser, Stewart, and Wittenbaum, 1995; Homan et al., 2008). However, encouraging members to draw on each other’s diverse expertise can be challenging.

Diversity in membership—whether in terms of expertise or demographic factors—influences the group’s effectiveness through its impact on group processes, such as decision making and conflict management (Bezrukova, 2013). Hence, greater diversity of membership increases the challenges facing a group by influencing these processes. High levels of diversity bring benefits, but differences among members can weaken identification with the group (Cummings et al., 2013). Members may differ in their values and motivations, shaped by their unique areas of expertise, organizational contexts, or life experiences. For example, when universities form research partnerships with private companies, the academic scientists who are rewarded for publications may have very different motivations than the industry scientists, who are rewarded for achieving specific business benchmarks (Bozeman and Boardman, 2013).

In highly diverse team science projects, communication problems can occur because of members’ use of technical or scientific language that is unique to their area of expertise and therefore unfamiliar to other members. The unique languages of the disciplines reflect deeper differences in underlying assumptions, epistemologies (ways of knowing), philosophies, and approaches to science and societal problems (Eigenbrode et al., 2007). For example, laboratories in molecular biology and those in high-energy physics have very different “epistemic cultures”—the practices and beliefs that constitute each discipline’s attitude toward knowledge and its way of justifying knowledge claims (Knorr-Cetina, 1999). When teams or groups fail to identify, discuss, and clarify these differences among their members, confusion and conflict can arise.

Chapter 3 highlights empirical evidence related to the team processes that underlie these challenges, which emerge from increasing diversity. Chapters 4, 5, and 6 introduce strategies for addressing them.

Deep Knowledge Integration

Knowledge integration occurs in some form within all scientific collaborations, as team or group members apply their unique knowledge and skills to the shared research problem. The process of knowledge integration can be challenging, and this challenge increases when scientific and societal questions require not only the combination, but also the deep integration of a broad set of disciplinary and, when relevant, stakeholder perspectives. Such deeper integration is fostered by interdisciplinary and transdisciplinary research approaches (Misra et al., 2011b; Salazar et al., 2012). For example, the National Cancer Institute’s Transdisciplinary Research on Energetics and Cancer (TREC) initiative to integrate social, behavioral, and biological sciences to address obesity and overweight, physical inactivity, and poor diet with the goal of preventing and controlling cancer. The integrative approach led to many novel discoveries; for example, one study found that participation in a 12-month exercise program decreased oxidative stress, which is closely linked to inflammation and cancer (Vogel et al., 2014).

To achieve the goals of interdisciplinary and transdisciplinary research, it is essential to understand and address the challenges associated with the deeper levels of disciplinary integration they entail. These challenges can emerge in efforts to integrate the knowledge of members from different disciplines with different cultures, languages, and research practices (Knorr-Cetina, 1999). Participating scientists may feel uncomfortable crossing the boundaries of their home disciplines—both the physical boundaries of their disciplinary department, laboratory, or office, and the cultural boundaries that guide and focus their research activities (Klein, 2010). Not all collaborators will be ready or willing to engage in the same level of integrative work. As the degree of integration increases, individuals may face challenges with feeling the loss of disciplinary “identity” or fear of becoming a “generalist” (Salazar et al., 2012). In molecular biology, for example, scientists’ identities are closely linked with the materials, techniques, instruments, and enabling theories of their research groups or laboratories, which Hackett (2005) refers to as “ensembles of technologies.” These challenges can be instigated and perpetuated by organizational cultures and incentive systems (e.g., promotion and tenure policies) that reward work within a single laboratory or a single discipline (Fiore, 2008; Stokols, Hall, and Vogel, 2013).

Strategies to address these challenges and foster successful knowledge integration in science teams and larger groups are discussed in Chapters 4 through 9 of this report.

Large Size

Science and engineering teams and larger groups, as reflected in publications, have consistently expanded in size over the past five decades (Adams et al., 2005; Baker, Day, and Salas, 2006; Wuchty, Jones, and Uzzi, 2007). This trend is illustrated in Figure 1-3, which shows the frequency of papers published in each year by single authors and groups of various sizes from 1960 to 2013, based on authorship of published papers recorded in the Web of Science. Across all science and engineering fields, the number of papers written by solo authors has remained relatively constant in absolute numbers but declined in terms of relative share of all papers written. By contrast, the size of authoring groups has increased each year. Pairs and trios were most frequent in the 1990–2000 period, while teams of 6 to 10 authors have been most common since 2000. Publications by very large groups of 100–1,000 authors first appeared in the 1980s, and publications by even larger groups of 1,000 or more authors appeared in the 2000s. The committee updated and analyzed the database, finding that, in 2013, about 95 percent of all papers were authored by 10 or fewer individuals, 5 percent were authored by 11 to 100 individuals, and less than 1 percent were authored by groups of more than 100 individuals.

Large numbers of participants can bring many benefits, yet also generate challenges, especially when the members are geographically dispersed.

FIGURE 1-3 Frequency of author team sizes in science and engineering, 1960–2013.

For example, Stokols et al. (2008b) found that large multi-institutional team science projects are highly labor intensive, prone to conflict, and require substantial preparation and trust among team members to even partially achieve their scientific and translational objectives.

One large project that overcame such challenges was the U.S. effort to build an atomic bomb, known as the Manhattan Project. Initially, in 1941, small groups of physicists and engineers worked at their home universities. After Enrico Fermi demonstrated the first controlled nuclear reaction in 1942, the government built nuclear facilities at Oak Ridge, Tennessee; Hanford, Washington; and Los Alamos, New Mexico, ultimately employing 130,000 people. The scientific and military leaders overseeing the project faced the challenge of coordinating the tasks of thousands of production workers and motivating them to rapidly produce nuclear materials, while maintaining the secrecy of the project goal. In July 1945, scientists successfully detonated the world’s first atomic bomb.

Although size is an increasingly important variable in the study of teams and groups, traditional research has rarely considered size to be a prime focus of analysis (Hackman and Vidmar, 1970; Stewart, 2006). Steiner (1972) identified the importance of team size as a determinant of a team’s division of labor. By increasing team size, a problem is divisible into smaller parts along the line of “more hands make light work.” Also, increasing team size could permit teams to effectively address larger-scale problems or more complex problems. For example, team size has been found to be positively related to the team-level recall of unique information, a driver of final performance of the team (Gallupe et al., 1992).

At the same time, larger team and group sizes are associated with process losses that can offset these potential benefits. As the number of members increases, the division of labor can become more inequitable (Liden et al., 2004) because of relational complexity and opportunities for “social loafing” if some members do little work (Latané, Williams, and Harkins, 1979). More generally, increases in group size require diverting time and resources from more productive activities to troubleshooting task interdependence, overcoming the tendency of individuals to “hoard” their unique knowledge, maintaining cooperative relationships, addressing incentive problems, and avoiding turnover (Jackson et al., 1991; Chompalov, Genuth, and Shrum, 2002; Okhuysena and Bechky, 2009).

Questions about the optimal size of groups remain open in part because the study of groups over time is difficult and in part because group size can have opposing effects on outcomes (e.g., a positive effect on productivity and a negative effect on cooperation). Recent work suggests that the effect of group size on productivity is moderated by the heterogeneity of the members. Observing the productivity of more than 549 information technology research teams and groups funded by NSF, it was found that larger

groups were more productive 5 to 9 years later. Nevertheless, the marginal productivity declined as member heterogeneity rose, measured by increases in the number of disciplines included or the number of institutional affiliations (Cummings et al., 2013). This result reflects decades of research in the social sciences illuminating the challenge of dealing with the “other” and suggests that traditional forces such as ethnocentrism (whether applied to ethnic backgrounds or disciplinary perspectives) will continue to be a major challenge (Levine and Campbell, 1972).

As well as varying based on the degree of heterogeneity, it is likely that the challenges of large group size vary with the disciplinary context or cultural norms in which the team or larger group is embedded. For instance, the physics and genomics communities increasingly work in very large groups and publish with hundreds or even thousands of co-authors (Knorr-Cetina, 1999; Incandela, 2013). These scientific fields have developed infrastructures to support collaboration, including shared scientific instruments, data-sharing platforms, and publication guidelines and tools for large groups of collaborators (see Box 6-1 later in this report).

Goal Misalignment with Other Teams

Large groups of scientists, such as research centers and institutes, typically include multiple science teams engaged in research projects that are relevant to the higher-level goals of the center or institute. Each individual team brings valuable insights, methods, and perspectives and may have its own distinct goals. If the goals of these teams are not aligned, then this can generate conflict, requiring careful management.

Winter and Berente (2012) observed that research centers and other large groups are often composed of science teams from different organizations (e.g., disciplinary departments or medical centers) that may have conflicting or only partially overlapping goals. To some extent, goal misalignment with other teams is a natural consequence of including teams with diverse expertise and research or translational agendas. This problem is particularly salient in translational projects that involve teams of community stakeholders, such as policy makers and citizens, along with science teams. In such projects, it can be difficult for the component teams to formulate and reach consensus on shared, overarching goals, and the goals may change over time as the project evolves and the participants change (Cash, 2003; Hall et al., 2012a; Huutoniemi and Tapio, 2014).

The new concept of a “multiteam system,” a complex system of interconnected teams, is beginning to increase understanding of the challenge of goal misalignment with other teams (Asencio et al., 2012). Such systems face the danger of “countervailing forces” that may advance goals at one level of the system, but slow collaboration at another level. One such

force—strong cohesion within an individual team—may benefit that team’s performance but may also discourage that team from sharing information with other teams that would benefit the system as a whole (DeChurch and Zaccaro, 2013). Furthermore, team members must balance devotion to the goals of their immediate team with the more distant goals of the broader organization or entity; the strong identification of members with a team can increase the success of the team, yet too strong an association with a proximal team can be at the expense of the higher order goals (DeChurch and Marks, 2006). For example, the James Webb Space Telescope, originally authorized in 1999, was expected in 2012 to cost nine times as much and to take a decade longer to complete than originally planned. The delays and cost overruns were attributed to inadequate budgeting for the inherent challenges of new technology development and weaknesses in managing and aligning the multiple academic and industry research and development teams engaged in the project (James Webb Space Telescope Independent Comprehensive Review Panel, 2010; U.S. Government Accountability Office, 2012).

Permeable Team and Group Boundaries

The boundaries of science teams and larger groups are often permeable, reflecting changes in the project goals and needs over time. The membership of a group or team may change as the project moves from one phase, requiring a certain type of expertise, to another that may require different expertise. Although these changes have the benefit of matching expertise to scientific or translational problems as they arise, they can also create challenges for effective team or group interaction.

Changes over time in the membership of a team or group may reflect the career stage and training needs of members as well as the research goals of the team or group. For example, studies of life sciences (Hackett, 2005) and physics laboratories (Traweek, 1988) have found that personnel turnover is ongoing, as students, postdoctoral fellows, and junior scientists are employed for a few years before moving on to other positions. However, unlike business employees who are typically assigned to work teams, scientists often voluntarily join science teams or groups. Therefore, scientists tend to have autonomy and operate like “free agents.” A single scientist may belong to multiple teams at any one time, devoting more or less time to each one, depending on the level of funding available, the scientist’s teaching and other research commitments, the potential for rewards, and other factors, including the scientist’s personal interest in each particular project.

Permeable boundaries are central to transdisciplinary projects that blur not only disciplinary differences, but also the distinctions between scientists and lay people. It can be challenging to elicit lay knowledge in a form that

can be integrated with formal expertise and applied to problems. Such challenges were overcome by the International Maize and Wheat Improvement Center in Mexico (Cash et al., 2003). Before the 1990s, center scientists had conducted research in laboratories or greenhouses to assure scientific rigor before transferring the resulting new crop breeds to the farmers. However, because the new crop breeds had sometimes lacked qualities desired by farmers and did not fit with existing crop management regimes, they were not widely adopted. By bringing the farmers directly into the project and working with them to find the most effective ways to integrate their unique knowledge, the center fostered more productive, sustainable farming practices on a wide scale (Cash, 2003).

The composition and number of team science projects with which a scientist is working can be in constant flux, creating potential challenges, as he or she juggles the conflicting time demands. One factor affecting a scientist’s degree of involvement and allegiance to a particular team may be the level of knowledge integration involved. For instance, if a multidisciplinary project engages an expert briefly in a consulting role, the expert may not feel invested in, or identify with, the team. In contrast, an interdisciplinary or transdisciplinary research project may require all participants to engage more fully over a sustained period in order to integrate knowledge at a deeper level, fostering feelings of identity and investment in the shared work. Cummings and Haas (2012) found that teams whose members devoted a higher percentage of their time to the focal team performed more successfully than did teams whose members devoted a smaller percentage of their time to the focal team.

Teams in other contexts, including emergency response, the military teams, and business, also have permeable boundaries, with attendant benefits and challenges. For example, business teams engaged in new product development have permeable boundaries and changing membership, making it difficult to build trust and cohesion (Edmondson and Nembhard, 2009).

Geographic Dispersion

Most science teams and groups today are geographically dispersed. The dramatic increase in team-based research for more than four decades is due to collaborations that cross university boundaries. Jones, Wuchty, and Uzzi (2008) compared publications produced by solo authors, within-university groups, and multi-university groups each year from 1960 to 2005 across all U.S. institutions of higher learning. They found that while the publications by faculty from the same university remained flat since the 1950s, the increase in co-authored publications was due to the growth of groups from more than one institution.

Currently most scientists work virtually, even with co-located colleagues, but the further geographically dispersed participants are across locations and institutions, the greater the possibility for coordination and communication challenges to emerge. Working across long distances introduces challenges such as a limited number of overlapping work hours among members located in different time zones and differences in incentives structures for members in different organizations. As noted above, some scientists’ identity is closely related to the particular work styles, technologies, and routines of their particular laboratories (Knorr-Cetina, 1999; Hackett, 2005).

Science teams or groups including scientists from different institutions and perhaps different countries may find it difficult to foster shared identification with the project, and to develop common work styles. Additionally, questions regarding access to technology and data can generate challenges. For example, the Thirty Meter Telescope is currently being developed by a large scientific organization including the California Institute of Technology, University of California System, National Astronomical Observatories of the Chinese Academy of Sciences, and National Institute of Natural Sciences/National Astronomical Observatory of Japan. The involvement of scientists from nations with different languages, cultures, politics, and economies could potentially lead to misunderstandings or conflicts.

Teams in business, the military, and other sectors are also increasingly geographically dispersed (Kirkman, Gibson, and Kim, 2012), gaining the benefits of global expertise and encountering similar challenges. Chapter 7 discusses the benefits and challenges of geographically distributed work and provides strategies and recommendations for remediating the challenges.

High Task Interdependence

One of the defining features of a team is that the members are dependent on each other to accomplish a shared task, and science teams are no exception (Kozlowski and Ilgen, 2006; Fiore, 2008). All team science projects, regardless of size or level of disciplinary integration, face challenges related to effectively developing and conducting a shared research agenda. The process of designing and conducting interdependent tasks that draw on and integrate the unique talents of the individual members is challenging, but such interdependence is the norm among the very large groups of physicists who conduct research at the Large Hadron Collider in Geneva, Switzerland. Knorr-Cetina (1999) suggested that the interdependence is inherent in the nature of research that can only be conducted at a few very large sites, leading to a unique “communitarian” culture in high-energy physics (see Box 6-1 for further discussion).

Fiore (2008) proposed that scientists engaged in interdisciplinary and

transdisciplinary research projects are more interdependent than those involved in team science projects that do not require a high degree of knowledge integration. He noted that some scientists avoid interdisciplinary research because they believe they must master multiple disciplines, yet teams in organizations are brought together to achieve shared goals with the recognition that the team members will only be able to develop partially overlapping knowledge.

Greater task interdependence among team or group members can lead to more opportunities for conflicts. Furthermore, when geographically dispersed colleagues must perform highly interdependent tasks, greater coordination and communication efforts may be required to bridge boundaries and facilitate task completion. The challenges of task interdependence and research-based strategies to address these challenges are discussed in Chapters 3 and 4. The unique challenges of task interdependence in dispersed teams are addressed in Chapter 7.

LEARNING FROM RESEARCH ON TEAMS IN OTHER CONTEXTS

Research on teams in contexts outside of science provides a rich foundation of knowledge about team processes and effectiveness. Because teams in science share features and processes with teams in other contexts, and based on the history of generalization of team research across contexts, the committee assumes that this knowledge can inform strategies for improving the effectiveness of science teams and larger groups. Below, we elaborate on these points.

Similar Features

Much of the existing scientific literature about teams has focused on teams in contexts outside of science, such as the military, factories, intelligence analysis, medicine, and emergency response. These teams in other contexts increasingly share the seven features that can create challenges for team science.

In medicine, for example, patient care is carried out by teams of physicians, nurses, and technicians with diverse specialties, who experience the benefits and challenges of high diversity in team membership as they seek to combine their knowledge to effectively solve problems. Intelligence analysts filter and fuse information to make decisions, solve problems, or create new knowledge, as do project teams and research and development teams (Heuer, 1999; Kerr and Tindale, 2004). All of these teams in other contexts seek to deeply integrate their knowledge, as do interdisciplinary and transdisciplinary science teams. In terms of size, teams in these other contexts are similar to science teams, typically including 10 or fewer members.

In the military, corporations, and health care, leaders are replacing traditional departments and divisions with agile teams that have permeable boundaries, adding members when particular skills or expertise are needed, while losing members who are assigned to different teams (Tannenbaum et al., 2012). Corporations once divided into departments specializing in research and development, sales, and production are creating temporary new product development teams that combine all of these functions. Edmondson and Nembhard (2009) identified five features of new product development teams that simultaneously offer the potential for innovation and pose challenges; several of these features also create challenges for team science. They include (1) project complexity, (2) cross-functionality, (3) temporary membership, (4) fluid team boundaries, and (5) embeddedness in organizational structures. The authors emphasized that successfully managing these features yields both organization-level benefits and team-level benefits, in the form of new capabilities and team member resilience.

Businesses with multiple, agile teams face the challenge of goal misalignment with other teams, and their employees and executives face the challenge of juggling the demands of the multiple teams (Espinosa et al., 2003; O’Leary, Mortenson, and Woolley, 2011; Maynard et al., 2012). Teams in business, governmental organizations, and in many other contexts are increasingly geographically dispersed, relying more heavily than in the past on technology to support their communication (Kirkman, Gibson, and Kim, 2012).

All of these features (highly diverse membership, deep knowledge integration, large size, goal misalignment, permeable boundaries, geographic dispersion, and high task interdependence) create challenges for science teams and teams in other contexts.

Similar Processes

Research in other contexts has demonstrated that certain interpersonal processes within teams, such as conflict, cohesion, and shared understanding of goals, are related to achieving team goals (Kozlowski and Ilgen, 2006; see Chapter 3). This research has also illuminated approaches that can be used by team leaders and members to influence these processes in positive ways, thereby increasing team effectiveness (i.e., performance). Recent research focusing specifically on science teams and groups has begun to yield similar findings about the importance of interpersonal processes. For example, intellectual conflicts and disagreements are important processes for advancing knowledge in science and other fields (Collins, 1998). Bennett and Gadlin (2012) analyzed in-depth interviews with members of successful science teams and others that ended because of conflict or did not meet their goals. They found that the more successful teams promoted

intellectual disagreement and discussion—which brought such benefits as continuing the dialogue, working through issues, and keeping problems or issues from accumulating—while also containing conflict and developing trust. In another example, the research on teams in non-science contexts has demonstrated that leadership styles and behaviors can positively influence interpersonal team processes, thereby improving team performance (Kozlowski and Ilgen, 2006). Similarly, a study of research laboratories in Europe found that the quality of laboratory directors’ supervision was positively related to the working climate and research productivity of the laboratories they directed (Knorr et al., 1979).

Generalizing the Research Across Contexts

Teams have been studied in a variety of organizational contexts, and findings in one context have often been generalized to other contexts. For instance, guided team self-correction, also known as team dimensional training, is a research-based approach that helps a team reflect on its teamwork during a past performance episode, identify errors, and develop solutions (Smith-Jentsch et al., 2008; see Chapter 5). It has been shown to improve performance in Navy attack center and shipboard teams and has been generalized to augment teamwork simulation exercises for Navy aircrews, engineering, seamanship, damage control, and combat systems teams, as well as civilian firefighting teams, law enforcement teams, and teams of corrections officers. Finally, it has been used as a tool to support on-the-job performance improvement through accident investigations within the nuclear power industry and to debrief one organization’s response to the terrorist attacks of 9/11 (Smith-Jentsch et al., 2008). Because guided team self-correction is based on a model of expert teamwork behaviors within a particular organizational context, the approach was translated to each new context based on analysis of the components of expert teamwork in that context.

Another example, Crew Resource Management training, was developed in the aviation industry to improve air travel safety by increasing teamwork and communication and reducing human error in the cockpit. The approach is widely used in the airline industry, has gained acceptance from airline crews, and has been shown to change crew behaviors (Helmreich, Merritt, and Wilhelm, 1999; Pizzi, Goldfarb, and Nash, 2001). Crew Resource Management provided the basis for guided team self-correction training described above and has also been translated for health care in TeamSTEPPS training. TeamSTEPPS is designed to improve patient safety by increasing communication and decreasing medical errors within patient care teams (King et al., 2008).

Therefore, based on the similarities in features and processes between

teams in science and those in other contexts and the history of generalization of team research across contexts, the committee assumes that research on teams in other contexts provides a rich foundation of knowledge about team processes and effectiveness that can inform strategies for improving the effectiveness of science teams and larger groups.

THE VALUE OF MULTIPLE APPROACHES AND THE PROMISE OF TEAM SCIENCE

Although team science is growing rapidly, individual scientists continue to make critical contributions and important discoveries, as exemplified by Stephen Hawking’s stream of new insights into the nature of the universe. Public and private funders with finite budgets must make decisions about whether to develop individual investigator or team approaches, and, if a team approach is selected, the scale and scope of the project. Similarly, individual scientists must make decisions about whether to invest time and energy in collaborative projects or to focus on individual investigations. It is important for scientists and other stakeholders to strategically consider the particular research question, subject matter, and intended scientific and/ or policy goals when determining whether a team science approach is appropriate, and if so, the suitable size, duration, and structure of the project or projects (Westfall, 2003).

Several strands of research and data suggest that team science can rapidly advance scientific and technological innovation by increasing research impact, novelty, productivity, and reach. First, group publications are more highly cited than publications by individuals, an indicator of their impact. Wuchty, Jones, and Uzzi (2007) found that teams and groups typically produce more highly cited publications and patents than do individuals (even eliminating self-citations), and that this advantage is increasing over time. Second, Uzzi and colleagues (2013) found evidence of both impact and novelty in team science: Compared with solo authors, teams and groups across disciplines were more likely to put novel combinations of prior work together, and to develop work that assimilated novel ideas into high-impact publications. Third, in a quasi-experimental comparative study, Hall et al. (2012b) found that transdisciplinary tobacco use research centers (large science groups) had higher overall publication rates and published findings from funded projects more consistently than did individuals or small teams investigating tobacco use, highlighting benefits for research productivity and dissemination. Fourth, Stipelman and colleagues (2014) compared the structure and disciplinary topical coverage of publications over time of transdisciplinary research centers with those of two comparison groups consisting of individuals and small teams. An overlay of the resulting publication data on a base map of

science revealed that the publications from the transdisciplinary research centers spread across the disciplinary topics in the map of science more rapidly and more comprehensively than both comparison groups, suggesting that the transdisciplinary team science approach broadens the reach of research findings across areas of science. Finally, the rapid growth of co-authored publications since 1960 documented by Wuchty, Jones, and Uzzi (2007) reflects the expert judgment of scientists in research funding agencies and peer review panels that teams or larger groups were best suited to address important research questions and that the results were worthy of publication.

In light of both the rapid growth and promise of team science, and the seven features that can create challenges, funding agencies and policy makers need to identify the most effective strategies for ensuring that taxpayer investments in team science yield valuable returns (Croyle, 2008, 2012). Scientists and leaders of teams and groups also need information on how to effectively manage these projects. The first step toward increased effectiveness is to gain understanding of the factors that facilitate or hinder team science and how these factors can be leveraged to improve the management, administration, and funding of team science. Although research is emerging from the science of team science, the research on teams, and from many other fields, it is fragmented, and team science practitioners may find it difficult to access or to understand and apply. This report integrates and translates the relevant research to support conclusions and recommendation for practice and identify areas requiring further research.

STUDY APPROACH

The NRC convened a Planning Meeting on Interdisciplinary Science Teams in January 2013 to raise awareness of this study, begin to explore the relevant literature, and solicit input from federal agencies, individual investigators, team science researchers, directors of research institutions, and other stakeholders (see http://tvworldwide.com/events/nas/130111/# [April 2015]).

The NRC then convened this committee, which met for the first time in April 2013. At its April meeting, the committee heard presentations from current and former NSF officials about the need for the study and from psychologist Gregory Feist, who focused on scientific creativity. Most of the meeting was spent in closed session discussing the study charge and how to approach it. The committee’s second meeting, in July 2013, included a Workshop on Team Dynamics and Effectiveness, which explored many individual-level and team-level factors that influence the processes and outcomes of team science (see http://www.tvworldwide.com/events/nas/130701/

[April 2015]). The committee’s third meeting, in October 2013, included a Workshop on Organizational and Institutional Supports for Team Science. Speakers at this workshop included researchers who study organizational factors and university leaders with practical knowledge of how to support team science. The committee’s fourth meeting, a virtual meeting, focused primarily on draft chapters, conclusions, and recommendations of the consensus report and also included a brief discussion with the NSF study sponsors. The committee met for its fifth and final time in March 2014. At this meeting, the committee reached consensus on its conclusions and recommendations and discussed finalizing this report.

ORGANIZATION OF THE REPORT

This report is designed to address the guiding questions in the committee charge (Box 1-1). It is organized into four parts, as follows:

• Part I: Setting the Stage. Chapters 1 and 2 provide the key definitions and conceptual framework for the research review in Parts II and III.
• Part II: The Individual and Team Levels. Chapter 3 provides an overview of the research on team effectiveness. It identifies team process factors at the individual and team levels and ways to manipulate three aspects of a science team or larger group to enhance effectiveness—its composition, professional development, and leadership. The following three chapters address each aspect, focusing in turn on team composition (Chapter 4), professional development and education (Chapter 5), and team and organizational leadership (Chapter 6).
• Part III: The Institutional and Organizational Level. Chapter 7 discusses the challenges of geographically distributed science teams and larger groups, and the role of organizations, leaders, and cyber infrastructure in addressing these challenges. Chapter 8 discusses organizational support for team science, focusing particularly on research universities. Chapter 9 considers the role of funding organizations that provide financial and other supports for team science.
• Part IV: A Path Forward. Chapter 10 provides a research agenda to advance research on team science effectiveness.

Reflecting the complex, multifaceted nature of team science and the multiple levels of analysis required to begin to understand it, many questions in the study charge are addressed in more than one chapter. For example, the role of individual characteristics in science team effectiveness is

introduced in Chapter 3 and discussed in greater detail in Chapter 4. Similarly, leadership influences team science not only at the level of the team, but also at the level of the research organization and the funding agency, often expressed in the development of “structures, policies, practices, and resources.” Hence, issues related to management and leadership are introduced in Chapter 3, elaborated upon in Chapter 6, and also discussed in Chapter 8. Table 1-3 depicts the coverage of the questions in the committee’s charge in the report chapters.

TABLE 1-3 Coverage of the Charge in the Report

This page intentionally left blank.