A growing body of research

shows that groups can

systematically enhance

The
their performance

Science
of Team
Success
By Steve W. J. Kozlowski and Daniel R. Ilgen

“Houston, we’ve had a problem,”

were the famous words that an-
nounced a crisis onboard Apollo 13.
Halfway through Apollo’s mission
to the moon, one of the spacecraft’s
oxygen tanks exploded, putting the
lives of the crew in grave jeopardy. A
group of engineers from NASA was
hastily assembled. Their mission: in-
The right stuff: The dra-
matic rescue of Apollo vent a way for the crew to survive
13’s astronauts would and to pilot their damaged vessel
have been impossible
without the coordinated back to Earth. The engineers were
efforts of NASA engi- successful, transforming a potential
neers. Research is re-
vealing why some groups
disaster into a legend of effective
NASA

work so well together. teamwork.

54 SCIENTIFIC AMERICAN MIND J u n e /J u l y 2 0 07

w w w. s c i a m m i n d . c o m SCIENTIFIC AMERICAN MIND 55
 As experienced
members leave
and new people
join a group, cru-
cial bits of collec-
tive knowledge
can be lost.

Human history is largely the story of people ing the U.S. government’s sluggish response to
working together in groups to explore, achieve Hurricane Katrina, the failure to prevent the
and conquer— and in our modern world the role tragedy of 9/11 and the explosion of NASA’s space
of teams is only growing, spurred by globaliza- shuttle Columbia.
tion and the enabling factor of communications Given the centrality of work teams, it is more
technology. Teams do not always play the role of than a bit remarkable how much our society’s
hero, however. They have also been implicated in perspective is focused on the individual. We
many political and military catastrophes, includ- school our children as individuals. We hire, train
and reward employees as individuals. Yet we
have great faith that individuals thrown into a
FAST FACTS team that has been put together with little
Building Better Teams thought devoted to its composition, training, de-
velopment and leadership will be effective and

1>> An effective work group should be designed well from

the start, bringing together people who can contribute
to the right mix of knowledge, skills, tools and other resources
successful. Science strongly suggests otherwise.
We recently reviewed the past 50 years of re-
search literature on teams and identified factors
necessary to succeed. that characterize the best collaborations. It turns
out that what team members think, feel and do

2>> Face-to-face meetings, social interaction among mem-

bers and a leader who establishes a good relationship
with every worker help a team make the best use of its exper-
provide strong predictors of team success — and
these factors also suggest ways to design, train
and lead teams to help them work even better.
DAV I D T R O O D G e t t y I m a g e s

tise and create a cohesive mission. Unfortunately, although society places a great
value on teamwork, the way organizations make

3>> Generic teamwork skills such as setting goals, adapt-

ing to change, resolving conflict and providing feed-
back allow teams to learn from each challenge and continu-
use of teams often runs against known evidence
for what works — and even against common
sense. For example, it seems obvious that teams
ally improve their performance. need sufficient resources to enable members to
accomplish their goals. Still, in this era of down-

56 SCIENTIFIC AMERICAN MIND J u n e /J u l y 2 0 07

( Organizations reward individuals based on individual
performance rather than team performance. )
sizing and cutbacks, one has to question the wis- sibility, even when the assignment is something
dom of many managers who believe that more that could be done easily by an individual work-
can always be accomplished with less. ing independently. The result is a team that is
Consider, too, that organizations typically re- more likely to impede performance than enhance
ward people with salaries, bonuses and promo- it. Another question is, What type of team struc-
tions based on individual performance rather ture is required? The task of some teams is such
than team performance. These rewards can often that their employees can function independently
inhibit team members’ willingness to work to- for long stretches and occasionally confer and
gether and help one another, even when the suc- pool their results, as with a team of salespeople
cess of the team depends on it. Such success re- working in different geographic regions. Others,
quires a delicate balance between meeting the such as surgical teams, require a high and con-
goals of the team as well as those of the individu- stant degree of coordination.
als who populate it. Research on goal setting, co- The job assigned to a team also determines
operation, competition, conflict and negotiation the primary focus of activities, and how well the
contributes to a better understanding of how peo- individual members complete their related duties
ple remain in teams and work together. determines the team’s effi ciency. That is why
An effective group
Indeed, a crucial question that should be team studies have turned to an approach known brings together
asked before putting a team together is whether as organizational psychology, which focuses on people with the
you need one at all. Some businesses recognize the task as central to understanding the dynam- necessary knowl-
the importance of teams and promptly restruc- ics of teamwork and team performance. (In con- edge, skills and
ture every task so that it becomes a group respon- trast, a traditional social psychology perspective tools to do the job.
SCHLEGELMILCH Corbis

w w w. s c i a m m i n d . c o m SCIENTIFIC AMERICAN MIND 57

 focuses more on interactions among peers, and Mellon University, nicely demonstrated how team
the work merely serves as the context for those members benefit from their collective knowledge
exchanges.) As mentioned before, the task sets when they learn together. These researchers
minimum requirements for the resource pool — trained college students to assemble transistor ra-
the constellation of knowledge, skills, abilities dios either alone or in groups of three. A week
and other characteristics (such as personality, later the subjects were tested with their original
values) — that is available across team members. group or, for people who received solo training,
in newly formed groups. Members of groups that
The Collective Mind had trained together remembered more details,
One of the most important things a team built better-quality radios and showed greater
brings to a task is what its members think, the trust in fellow members’ expertise. People in new-
relevant information they carry in their heads. ly formed groups were less likely to have the right
This knowledge can include a mastery of the mix of skills to complete the task efficiently and
tools they use and an understanding of the task knew less about one another’s strengths.
at hand, its goals, performance requirements and With a different group of collaborators, Ar-
problems. Some knowledge may be shared by all gote studied the effect of individual turnover on
workers, whereas particular members might another chore, making origami birds. Again,
have specialized skills or know-how. The ability groups of three trained together and were given
to access and use this distributed expertise effi- six time periods to make as many paper products
Successful
teams must deal
ciently is one characteristic of successful teams. as possible. The groups with turnover produced
with parasitic A 1995 experiment by psychologist Diane Wei significantly fewer folded creations than groups
members — those Liang, then at the University of Minnesota, psy- whose members stayed constant, suggesting as-
who do not con- chologist Richard L. Moreland of the University pects of group knowledge were being lost when
tribute anything of Pittsburgh and Linda Argote, professor of or- people were replaced.
to the group. ganizational behavior and theory at Carnegie In an interesting twist, organizational behav-

R A N DY FA R I S C o r b i s

58 SCIENTIFIC AMERICAN MIND J u n e /J u l y 2 0 07

 ior expert Kyle Lewis of the McCombs School of less is understood about how emotional state af-
Business at the University of Texas at Austin found fects team performance than about cognitive in-
that the development of a team’s ability to access fluences, it is clear that how teams feel can drag
distributed knowledge required face-to-face inter- down productivity or boost it up — or otherwise
action. In groups that communicated exclusively complicate it. For example, a shared positive at-
by phone or e-mail, this skill did not emerge— an titude can reduce the number of absences in
observation of increasing importance, given the teams and lower the likelihood that people will
rise of teams that operate remotely and coordinate leave the group.
sometimes only through computer interactions. It But there are hints that good moods do not
should prompt concerted efforts to understand the always lead to good outcomes. Social psycholo-

( One person’s behavior leads to group-level changes

in emotion, both negatively and positively. )
reasons for such barriers and explore whether web- gist Joseph P. Forgas of the University of New
cams, videoconferencing or other technologies South Wales in Australia, for example, asked
that allow people to interact will help overcome teams to hold a discussion after they watched
this problem. For now, the best solution may be to happy or sad videos and found that greater divi-
guarantee some face time for team members sions arose in the groups that were given a prior
throughout their project. “feel good” stimulus.
Beyond an understanding of the nuts and It also appears that team members tend to
bolts of any given project, another cognitive in- change their moods in concert. Social psycholo-
fluence on team effectiveness is the emergence gist Peter Totterdell of the University of Sheffield
of an overall objective, mission or strategic im- in England and his colleagues had nurses record
perative of the group — something psychologists their moods each day at work over a period of
call the team climate. The powerful effect of cli- three weeks. They found that the mood of differ-
mate on the real-world impact of teams is well ent teams shifted together over time. Totterdell
established. For example, one of our groups (Koz- has measured a similar convergence in the affect
lowski’s) showed that high-tech businesses whose of teams of accountants and cricket players.
engineers agreed on the objective to stay techno- The fact that emotions move in this lockstep
logically up-to-date showed improved perfor- way has led to a concept of emotional contagion,
mance and had more employees pursuing con- the idea that emotions within teams are trans-
tinuing education and displaying positive job at- ferred from one person to others close by. In a
titudes. Several studies across many industries well-controlled laboratory study, professor of
have shown that when a team has absorbed a management Sigal Barsade of the Wharton School
mission statement that values customer service, of the University of Pennsylvania investigated the
this attribute predicts customer satisfaction. effect of emotional contagion on team process and
Likewise, when a team agrees that the objective performance. The research involved using a drama
is safety, the result is more safety-conscious be- student posing as a research subject whom Bar-
havior by team members and a reduction in the sade trained to participate with a happy, optimis-
rate of accidents. tic attitude or an unpleasant, pessimistic one. She
found that this one person’s behavior did lead to
Ties That Bind group-level changes in emotion, both for positive
Climate emerges in groups with strong ties and negative affect. Although the scientific study
among their members. For example, team mem- of how mood influences performance of the indi-
bers who have a good relationship with their lead- vidual and the team as a whole is still in its infan-
er tend to share climate perceptions with their cy, this area promises to yield important insights.
boss and co-workers. Teams that have frequent
informal social interactions also show greater
(The Authors)
consensus on climate than those that do not.
Part of the glue that binds people to their STEVE W. J. KOZLOWSKI and DANIEL R. ILGEN study the dynamics of
bosses or to one another is emotional. Although teams at Michigan State University.

w w w. s c i a m m i n d . c o m SCIENTIFIC AMERICAN MIND 59

( If teamwork skills were ubiquitous, there would be
enormous benefits to students and society. )
Works Well with Others team over time — and team leaders can play a ma-
Finally, whatever the task, the way people jor role. Prior to action, for example, the leader
perform, or do, the work as a team makes a pro- can help set team learning goals commensurate
found difference. The important elements here with current team capabilities. During action,
appear to be general teamwork skills that are not the leader monitors team performance (and in-
specific to any particular task. Some of the re- tervenes as necessary). As the team disengages
search in this area centers on bad behaviors that from action, the leader diagnoses performance
degrade team performance and spirit— dealing deficiencies and guides process feedback. This
with “free riders,” for example, who rely on cycle repeats, and the complexity of learning
other team members to do their job and thus goals increases incrementally as team skills ac-
contribute less than their fair share. This type cumulate and develop. This kind of feedback
of disruptive behavior can be limited by requir- loop has been shown to reliably improve team
ing that contributions be visible and members thinking and performance.
accountable. Work from Kozlowski’s group, however, has
There are also many positive ways in which found a trade-off in the type of feedback provided
the best teams act that give them an advantage: and the resulting performance. Feedback directed
On-the-job train-
individuals are aware of one another’s perfor- to individuals yields higher individual perfor-
ing: every task a
mance, provide backup coverage for members, mance at the expense of team performance; team
team undertakes
is a chance to set goals, coordinate their actions, communicate feedback yields better team performance at the
learn new skills effectively, make decisions, resolve confl icts, and expense of individual performance. If both types
and to learn how adapt to changing circumstances and new ideas. of feedback are provided, both levels of perfor-
to work together A key point is that this learning process can be a mance cannot be maximized. The findings indi-
more effectively. dynamic one that helps to shape and improve the cate that team designers need to be mindful of

ASHLE Y COOPER Corbis

60 SCIENTIFIC AMERICAN MIND J u n e /J u l y 2 0 07

 precisely what they want to
be salient to team members
and should design supporting
goal and feedback systems ac-
cordingly. Such systems may
need to be adaptive, shifting
the balance depending on
current needs.
One reason that achieving
the right level of feedback is
so important is that teams
learn best while doing. In
some cases, notably in the
military and in aviation, this
on-the-job training can be
supplemented with sophisti-
cated and realistic simula-
tions of combat missions or
of takeoffs and landings. This
virtual training approach is
starting to fi nd applications
elsewhere, such as in medi-
cine, although in most cases the best place teamwork and leadership behaviors, attention is Leaders play a
to develop team skills is on the job itself. Gen- usually on the group’s output — a report, for crucial role in
eral teamwork proficiency turns out to be one example — with little or no attention placed on developing group
area where classroom training appears to make guiding the nature and effectiveness of the team skills by setting
a strong difference, perhaps because these are process. goals, monitoring
performance and
generic skills not related to a specific job. Ac- If teamwork were taught along with reading,
giving feedback.
cordingly, semester-long college-level programs writing and mathematics, and if these skills were
that significantly improve students’ knowledge ubiquitous, there would be enormous benefits to
of generic teamwork competencies have been de- students and society alike. For now, though, it is
veloped. Nevertheless, encouraging work by one often only after a great triumph or tragedy that
of our teams (Ilgen’s) has demonstrated that the importance of teamwork is drawn into the
knowledge of these team competencies can im- spotlight. Ironically, these occasions focus large-
prove significantly with only 30 minutes of indi- ly on singling out individuals for reward or to
vidual training. assign blame, as the case may be. Despite liter-
ally thousands of studies that show much can be
Missed Opportunities done to design teams properly and to ensure they
Although these skills can be taught, they do their jobs well and get better as time goes on,
rarely are — and few formal experiences to im- the question rarely turns to how the successes can
part generic team-process and leadership experi- be replicated or problems avoided the next time
ences are available. If such courses are provided around. We think it is just a matter of applying
at all, they tend to be very late in the educational the science. M
process — in college courses or in professional
programs such as business school, for example — (Further Reading)
and these courses are usually geared toward im-
◆ A Multiple-Goal, Multilevel Model of Feedback Effects on the Regulation
parting factual knowledge rather than building
of Individual and Team Performance. Richard P. DeShon, Steve W. J. Koz-
skills. We sampled a number of well-known lowski, Aaron M. Schmidt, Karen A. Milner and Darin Wiechmann in Journal
M.B.A. programs and found that fewer than of Applied Psychology, Vol. 89, No. 6, pages 1035–1056; December 2004.
E R I N R YA N z e f a / C o r b i s

half listed a course devoted primarily to leader- ◆ An Evaluation of Generic Teamwork Skills Training with Action Teams:
ship or teams. Effects on Cognitive and Skill-based Outcomes. Aleksander P. J. Ellis,
Bradford S. Bell, Robert E. Ployhart, John R. Hollenbeck and Daniel R. Ilgen
Furthermore, although it is not uncommon
in Personnel Psychology, Vol. 58, No. 3, pages 641–672; Autumn 2005.
for educators from elementary school through ◆ Enhancing the Effectiveness of Work Groups and Teams. Steve W. J. Koz-
college to include assignments organized around lowski and Daniel R. Ilgen in Psychological Science in the Public Interest,
group projects in which students may display Vol. 7, No. 3, pages 77–124; December 2006.

w w w. s c i a m m i n d . c o m SCIENTIFIC AMERICAN MIND 61