A plane crashes in the desert and six survivors are left to choose what items to salvage. Should they take a gun? Salt tablets? Heavy overcoats? How about a cosmetics mirror? With 15 possibilities and limited time to make their choices, the group must quickly prioritize.
You might think that a high degree of initial consensus about what to take would indicate the survivors chose the “correct” items, the ones experts deemed would best help them stay alive. After all, five out of six people aren’t likely to be wrong, right? As it turns out, an experiment that presented the above scenario to business-school students revealed that groups with more initial dissent were more likely to pick the right items, ultimately choosing wisely almost twice as often as the other groups. In teams where there was little disagreement, says Richard Larrick, a professor at Duke University’s Fuqua School of Business who ran the experiment with Ph.D. candidate Al Mannes, “no one asked, ‘Where might we be wrong?’”
This experiment is part of a growing body of research on groupthink. First analyzed by Yale psychologist Irving Janis in 1972, groupthink is hardly a new problem. Yet it continues to sway all sorts of important decisions. Groupthink has been blamed for any number of recent fiascoes, from the Columbia space-shuttle disaster to the implosion of Enron to the prolonged war in Iraq. Corporate boards and committees are said to be particularly vulnerable to the phenomenon. In a 2003 working paper, law professors Stephen J. Choi and Adam C. Pritchard posited that the Securities and Exchange Commission may itself be prone to groupthink, given its strong organizational culture and the presumed self-selection of like-minded regulators.
While averaging individually generated answers may produce astonishingly accurate answers, strange things start to happen when you put a handful of otherwise intelligent people in a room together. One, for example, is “the fear that everyone else knows more, so I’ll just go along,” says Larrick. “Another is the fear that the boss has already really decided, so why bother to stick my neck out?”
But groups can be more intelligent than individuals working alone, say professors who study them, if groupthink (a form of “process loss” in academese) can be avoided. To that end, “group processes have to be actively and repeatedly managed,” says John W. Payne, a professor at Fuqua. “You’ve got to make sure that people with unique information share it, or else you’re likely to focus only on what is held in common.”
Managing a group for optimal performance begins by gathering people with diverse perspectives. A bigger group is not necessarily better. “You’re better off with a smaller group representing diverse viewpoints than a larger group that is similar,” says Payne.
Then, to get the appropriate amount of dissent, “people have to feel psychologically safe to make mistakes and say stupid things,” says Paul Paulus, a professor at the University of Texas at Arlington who has studied group interactions for almost 20 years. That means coming up with a process to allow all voices to be heard and perhaps appointing a devil’s advocate to make sure all possibilities are considered. It’s also important for the leader of the group to appear genuinely impartial, to avoid influencing would-be sycophants.