What Predicts the Success of People Working Together to Solve a Problem? It’s Not What You May Think
Having smart people in groups doesn’t necessarily make groups smart
Posted November 4, 2010
Teams of people working together - both face-to-face and virtually - is becoming an increasingly popular method for problem solving in today's workforce. Whether it is a sales team working collectively to land a client, a management team restructuring a company to cut costs, or a medical research team working to find a cure for a life-threatening disease, groups of people are often brought together to tackle big problems. So, what predicts how successful these groups will be? It turns out that the whole is often greater than the sum of its parts.
It's been known for some time that when individuals solve problem on their own, a single factor - termed general intelligence - can predict success. The term general intelligence or "g" came in vogue in the early 1900s when psychologists showed that people who did well on one mental test tended to do well on other mental tests, despite large variation in the content of the tests themselves. The idea of "g" was exciting because it suggested that, in less than an hour, one's general intelligence could be measured and used to reliably predict performance in many facets of life ranging from school grades to occupational success. Of course, the idea of a general intelligence is a contentious topic and many folks have argued about how wide a net "g" can cast or how reliable and important a predictor it is. Nonetheless, despite this controversy, there is a good deal of evidence that people posses some general cognitive ability that can be measured and used to predict future performance on a variety of tasks.
But, in today's workforce, people seldom work on big problems on their own. Rather, they are often brought together - either physically or virtually - when difficult tasks arise. So, it's important to understand what predicts the success of a group of people working together. So, what does?
On the one hand, it's easy to assume that since "g" is a fairly reliable predictor of an individual's problem solving success, the average or maximum "g" of individuals in a group would be the best predictor of group success as well. The more smart people you have in a group, the better the group performs. On the other hand, as anyone who has been part of a group has probably experienced, success depends on more than just each person's knowledge. A paper published this month in Science confirms this latter view by showing that having a bunch of smart people in a group doesn't necessarily make a group smart.
Researchers from Carnegie Mellon University, Union College, and MIT got together and observed the performance of roughly 700 people, working in groups of two to five, for up to five hours, to solve simple tasks ranging from visual puzzles, to brainstorming, to negotiating over limited resources. The groups also solved more complex tasks, such as an architectural design task modeled after a complex research and development problem.
What the researchers found was that a group's performance on one simple task predicted their performance on other simple tasks. They also found that when people performed, for example, the architecture design task alone, their individual intelligence predicted performance. However, when people performed the same task as a group, it wasn't the average or maximum intelligence of individual group members that predicted performance, but a collective measure of intelligence or "c" of the group itself.
So what is this "c" factor? Both the average and maximum intelligence of individual team members did somewhat relate to "c." But, importantly, "c" did a much better job at predicting a group's performance than either of these individual measures alone. So, "c" seems to be more than the sum of its parts. Interestingly, "c" was highly related to the social sensitivity of group members and the equality of conversational turn-taking. In other words, the more cooperation among a group, the higher its "c".
These findings suggest that, just as in individual intelligence testing, a short collective intelligence test may be able to predict a sales or management team's long-term effectiveness. And, given that it may be easier to change a groups' collective intelligence, relative to an individual's "g" (although both are most likely possible), there may be easy ways to get the most out of a problem solving team. By the way, the researchers found that another factor predicting "c" was simply the number of females in the group. The more women, the higher a group's "c" - most likely because the women in the groups tended to be higher in social sensitivity than the men.
The take home point? The way group members interact when they are assembled matters. Indeed, these interactions are more important for success than the intelligence of each group member.
Follow me on Twitter
Woolley et al. (2010). Evidence for a Collective Intelligence Factor in the Performance of Human Groups. Science, 330, 686.