Groups produce collective intelligence, study says
January 10, 2011 5:00 AM
Pam Panchak / Post-Gazette
A study by Anita Williams Woolley and her colleagues found "collective intelligence" valuable.
By Mark Roth Pittsburgh Post-Gazette
America is a cowboy nation.
We love the lone hero who rides in to save the town, whether he is a charismatic CEO, a dynamic political leader or a brilliant scientist.
But in today's society, says Carnegie Mellon University professor Anita Williams Woolley, the most valuable kind of smarts may be group IQ, or what she calls "collective intelligence."
In an October study published in Science magazine, Dr. Woolley and her colleagues showed that when they randomly assembled groups of two to five people in Boston and Pittsburgh and gave them a set of intellectually challenging tasks, the groups showed markedly different levels of collective intelligence.
The groups were given a wide range of tests -- brainstorming how many uses a brick could have, for instance, or solving a tough puzzle or collaborating to jointly type a large chunk of text into a shared document.
The group members were also given individual IQ tests, which showed they represented the same range as a typical adult population.
Surprisingly to Dr. Woolley, the average IQ of a group had almost no impact on its collective intelligence. It also didn't matter whether a group had one high-IQ individual.
There were three factors that did make a difference, though.
One was the social sensitivity of group members -- how much they paid attention to each other and asked questions.
The second was turn-taking. Groups that shared the floor had much better results. "When you had someone really dominating the conversations in these groups, the group did not perform well," she said.
Finally, in general, the more women in a group, the smarter it was.
As they analyzed that result, she said, it didn't mean the women had higher IQs than the men, but that they were more socially sensitive and less likely to dominate discussions.
Asked whether a simple way of boosting group performance would be to set a minimum quota for female members, Dr. Woolley laughed.
"I wouldn't advocate that because all of us know counterexamples to the general rule that women are more socially sensitive."
Dr. Woolley speculated that one particularly important behavior that many women display is the willingness to admit what they don't know, and ask if anyone else in the group has better expertise.
"Men seem less likely to reveal such deficits."
The study, along with other group performance research, suggests some practical ways for boosting the effectiveness of groups in workplaces, schools, churches and nonprofits.
The first priority, Dr. Woolley said, is to make sure "the right people are in the room," meaning that group members need to have a good mix of skills and thinking styles.
The other is to help group members learn how to do a better job of communicating with each other. One critical step, she said, is to make sure group members take some time to discuss how they can work together to achieve their goal.
Many people don't like to discuss such "process" issues, she said, but the research has shown that "groups that pay more attention to that gets things done more efficiently."
The study hasn't met with universal applause.
Among some of the online comments about the study was one that asked: "Whoever heard of a great sonnet being composed by a group?"
"But a lot of times when you look at individuals who did great things," she said, "you can often trace that to a whole support system of other individuals behind them, so to what degree was that achievement really the product of an individual, or of a system?"
The October study, carried out with Massachusetts Institute of Technology researchers Thomas Malone, Alex Pentland and Nada Hashmi and intelligence researcher Christopher Chabris of Union College in New York, is part of a surge in research on group performance.
In Massachusetts, for instance, Tufts University psychology professor Samuel Sommers has studied the impact of racial diversity on group performance.
He has found that when groups are dealing with racially sensitive issues, the more racially diverse they are, the better job they do.
In one experiment, Dr. Sommers asked mixed groups of white and African-American college students to prepare to debate either a racially neutral controversy, such as same-sex marriage, or a racially charged one, such as affirmative action enrollment in universities. The students were given reading materials on the topics, and the groups' racial makeup varied.
The students were then asked not to debate the topic, but to take an SAT-style comprehension test on the material they had read.
When the subject was racially sensitive, he found, the white students in racially diverse groups were able to do a much better job of remembering and analyzing the reading material.
He believes two things might have been going on. The white students might have anticipated disagreements on the topic, and therefore prepared themselves better for the debate. They also might have wanted not to appear biased, and mastered the material better for that reason.
Carnegie Mellon's Dr. Woolley believes some of the techniques used by groups with strong collective intelligence can be taught.
At MIT, for instance, Dr. Pentland has used personal devices known as "sociometric sensors" and a cell phone application to help groups do a better job at taking turns in discussions.
The meters can register how often and how long someone speaks. The cell phone app has a display with an icon for each group member and a ball in the center. If someone begins to dominate the conversation, the ball moves toward that person, and everyone in the group can see it.
Researchers found that people in the groups that used the devices tended to speak for shorter periods -- seven seconds versus 10 seconds on average -- than in groups that didn't. The metered groups' members also tended not to talk over top of each other nearly as much as the control groups' members.
Group achievement is increasingly important today, Dr. Woolley said.
The number of inventors listed on each patent application in the United States has grown steadily over the past 40 years. Today, nearly 70 percent of patent applications list more than one inventor, and 13 percent list five or more.
As the world becomes more complex and employees' skills more specialized, she said, "more and more of our problems will have solutions that lie at the intersection of individuals with specialties, so they need to be able to collaborate."
"While it's simpler and in some ways more gratifying to some people to think of accomplishments as residing in one individual," she said, "in reality they just don't -- or not as often as we'd like to think they do."