Before returning to academia for a Ph.D., Angus Hildreth spent a decade in the corporate world as a consultant and manager. "I had the great privilege early in my career to be exposed to very powerful individuals, executives of a multi-billion dollar organization," he says. Hildreth would sit in rooms and watch these executives, each essentially the CEO of his own unit, as they planned the future of the 150,000-person company.

"Perhaps naively, I assumed that you put the most capable, most effective people, who day-to-day make decisions that drive their organizations to success—you put all of these guys in the same room and you get this platonic idea of leadership," he says. "And I was, uh"—he pauses to laugh—"surprised, if you like, by how much these teams struggled to reach decisions."

They bickered, they withheld information, they digressed. Hildreth had some ideas why, which led him to think about researching the topic. "And I met Cameron Anderson"—a professor of organizational behavior at Berkeley—"who's world-renowned in the areas of power and status, and, day one, he said, 'Drop everything, we need to start studying this.'"

Anderson is now Hildreth's doctoral advisor, and the two have just published a paper, in the Journal of Personality and Social Psychology, showing that when teams contain several powerful people they actually become less effective—and it details why. The work is important for understanding the kinds of failures we see in everything from business and politics to sports teams and jury deliberations.

An all-star team may never shine

A similar pattern to what Hildreth observed has been shown in other recent research. Boris Groysberg and colleagues studied investment teams and found that their client-rated effectiveness peaked when groups contained only 65 percent "star analysts" (highlighted as top performers by Institutional Investor magazine); any more standouts hurt the teams. And when their expertise overlapped, even that many was too many cooks.

Roderick Swaab and colleagues found that people believe that more top talent in an organization or on a soccer team is always better. But looking at dozens of World Cup soccer teams and NBA squads, they found that the soccer teams performed best when only about 70 percent of players were stars (playing for elite club teams), and the basketball teams performed best when only about 50 percent were stars (playing in the top third of the league); any more stars hurt coordination and thus performance.

And Lindred Greer and colleagues studied work teams from a financial company in the Netherlands and found that the more power the team had in the organization, the worse they were at a decision-making task, thanks to internal conflict over who should do what. Power was not quite as harmful when members agreed with each other over how much control individuals had within the group.

Hildreth and Anderson hoped to build on this work by looking at what happens if they manipulated power by assigning people roles, rather than observing people who had obtained high-power roles in the real world and measuring correlations. They also set out to use larger samples in their study, to test additional tasks, and to look for mechanisms beyond conflict over group processes.

In the first experiment, college students performed a tower-building task in pairs. In some pairs, one student was told to assume control and make all decisions while the other was instructed to follow, an assignment supposedly based on a questionnaire they'd completed but actually by random selection. All pairs were then split and rearranged into triads for a creativity task. Each team had three high-power members (those given control in the tower task), or three low-power members, or three neutral members whose power hadn't been manipulated. They were videotaped inventing a new organization, and later judges rated the creativity of their idea as well as several aspects of their interaction. These factors included status conflict ("Members of the group competed for control over the group and its decisions"), task conflict ("The group had frequent disagreements about the tasks they were working on"), information sharing ("The group shared all of their information with each other"), and task focus ("Overall, how focused was the group on accomplishing the task?").

Although previous research had shown that power increases creativity in individuals when they work alone, the groups of powerful students were less creative than the other groups. Their failings resulted from increased status conflict, reduced information sharing, and reduced task focus. In other words, they spent their time fighting for control instead of doing their job. A companion study was similar except students did the creativity task alone, and here power increased creativity. So in the first study power hindered group processes, not individual ingenuity.

In another experiment, executives from an organization worked in groups of three or four to decide on hiring a hypothetical chief financial officer. The four most powerful execs worked together, and so on, down to the least powerful. The more powerful the group members, the more their arguments showed knowledge and expertise. And yet, the less likely they were to reach agreement. Their gridlock resulted, again, from increased status conflict, reduced information sharing, and reduced task focus, plus increased task conflict.

Is it all about coordination?

A final experiment resembled the first, except students did three creativity tasks, followed by a persistence task. Two creativity tasks didn't require coordination: Students individually jotted down unusual uses for a cardboard box and for a brick. In the third creativity task, the teams decided together on the most creative use for a cardboard box. Finally, teams were asked to solve four anagrams, the last two impossible, and the time they spent was recorded.

In the low-coordination tasks, power didn't affect quality of ideas and even led to producing more ideas. And in the anagram task, power increased persistence. Yet in the high-coordination task, powerful groups presented less creative ideas than did powerless groups. Individual mental horsepower doesn't matter when everyone's running in different directions, or butting heads.

Swaab and colleagues found a similar effect of coordination in their look at athletes. While too many stars hurt soccer and basketball teams, on baseball teams there was no such thing as too many stars. That's because, in the words of Barack Obama, basketball is "the quintessential team sport," whereas baseball, in the words of Bill Simmons, is "an individual sport masquerading as a team sport." Pitching and hitting are largely solitary endeavors.

The less-noticed cost of empowering

According to Greer, who performed the work on teams in the Netherlands and is now a business school professor at Stanford, "This line of work"—that she and Hildreth and others are pursuing—"is surprising, in that, for the longest time, social psychology brought forward findings that power was great for individuals—that it made individuals more creative, improved their executive functioning, enabled them to pursue their goals and take action, etc. It's a really interesting twist now to see that when you bring multiple high-power people together, such as in management teams or meetings of world-leaders, power can actually lead to negative outcomes at the team-level."

Why might there be status conflict when team members share the same level of power when that level is high, but not when it's low? It could be that while power enhances focus and creativity and persistence, it's also been shown to, in many cases, reduce empathy and politeness and humility, which can make it harder to get along with others. What's more, power and prestige often go hand in hand, and at last month's meeting of the Society for Personality and Social Psychology, Greer presented findings suggesting that the more prestige someone has, the more that person craves additional prestige. This, too, could lead to battles over status at the top of the hierarchy.

The consequences can be severe. "I'm reminded of these issues every time I watch leaders of both parties in congress try to reach bipartisan agreement on issues of national importance," Hildreth says. In their paper, he and Anderson write that powerful people need to overcome differences "whether they are trying to build an international treaty to restrict greenhouse gases, generate a new strategy for their firm, or agree on a fiscal budget for their state."

Researchers have suggested fixes, however. Groups may benefit when everyone's status is made clear, or when decision processes are formalized, or when meeting time is more structured. During his time in the corporate world, Hildreth also found it helpful just to let fussy, power-drunk executives vent before getting down to business. "If day-to-day you're the 800-pound gorilla sitting at the top of your organizational chart, when you say 'Listen to me' people will listen," he says. "But when you come together with your peers it's more of an uncertain environment: What is your status in this group? I think part of this acting out is to reestablish: 'Yes I deserve to be here, I'm important, My voice counts.'" Whether you're at bottom of the food chain or the top, everyone just wants a little respect.