Everywhere you look, from business to science to government, teams of people are set to work solving problems. You might think the trick to getting the smartest team would be to get the smartest people together, but a new study says that might not always be right.
Researchers at Carnegie Mellon University found that collaborative groups who conversed easily with equal participation were more efficient at completing sets of given tasks — and produced better results — than groups dominated by individuals.
Anita Woolley, an assistant professor of organizational behavior and theory, has been studying what it means to say a group is "intelligent." So she created teams of two to five people, drawn from 700 volunteers, and asked the teams to solve various kinds of problems.
"We had some brainstorming tasks, where the idea is to get as many ideas and as creative ideas as possible," she said. The teams performed other tasks where there was only one right answer, and still others where the teams had to come up with innovative solutions to a problem.
Here's an excerpt from one group's conversation:
Participant 1 (male): The problem is that we have to decide whether this star basketball player should be kept on the team despite cheating on the exam, right?
Participant 2 (female): Yeah, I think they're trying to see how we'll punish him.
Participant 3 (female): It looks like keeping a player like that would mean violating school policies.
Participant 1: Yeah, I would keep him out of the next game, unless it's a playoff. At the very minimum.
Participant 2: Yeah, I think that's the best option. He deserves at least that.
The participants in this group took turns discussing the problem, coming up with various angles and issues.
Another group attacked the same problem differently:
Participant 1 (male): I say he should not be allowed to play.
Participant 2 (female): He should not be allowed to play for the semester, at least.
Participant 3 (female): Well ...
Participant 1: I think he should not be allowed to play for his college career at all. It looks like it may just be a status thing. He just cheated on an exam. He should at least have been creative and tried to cheat effectively or something. I hate people who are treated nicely for athletics. They're just stupid.
No one else in the second group got a word in, and there were no alternatives to what the one noisy participant was proposing.
Woolley says this was an example of a moral reasoning task. She says that to do well, a group needed to consider multiple perspectives.
"In groups where the conversation was more evenly distributed, where you had better participation — and more equal participation among all of the group members — the groups were more collectively intelligent," Woolley says.
Not only was it annoying, but groups where one person dominated tended not to come up with as balanced and thoughtful a result — it wasn't as intelligent as the first group's effort. When Woolley looked for the qualities that made successful groups successful, she found that the individual intelligence of group members was unrelated to the outcome.
"A few things that were related however, were surprising," Woolley says. "One was the proportion of females in the group."
As she reports in the journal Science, the more females, the higher the group intelligence, although Woolley thinks it's not so much gender as a quality of social sensitivity that women on average have more of than men.
Thomas Malone, one of Woolley's collaborators and head of the Center for Collective Intelligence at MIT, hopes this research will lead to great things.
"Imagine you could go to a top management team in a company and give them a collective intelligence test that would then predict how well that team would respond to a very wide range of challenges they might face," Malone says. That could be something a board of directors might be very interested to learn.
But Steve Kozlowski, a psychologist at Michigan State University, says such a test is a ways off. He says that Woolley's study only looked at a small set of tasks.
"And it's really very, very difficult to generalize from the small set of tasks that were examined in these studies, using college students, ad hoc teams and very short periods of measurement," he says.
He's also not convinced that Woolley and her colleagues are measuring intelligence. He says they're just measuring how well groups do on a limited number of problems.
"Likening it to intelligence — I wouldn't call it controversial, I just don't see any evidence to support it," Kozlowski says.
Still, he says there may well be such a thing as collective intelligence, but he thinks it will take a lot more work to define it.