Your independent source for Harvard news since 1898 | SUBSCRIBE

Your independent source for Harvard news since 1898

Right Now | Obtuse Organizations

Secret Errors Kill


There is a story told about the early years of IBM, when an executive made a drastic error in judgment that cost the company $30,000, a sizable sum in those days. IBM president Thomas J. Watson asked to see the guilty party, who was sure that he would be fired. In their meeting, Watson walked the executive through his mistake in detail, asking what led to the error and how it could have been avoided. To the unhappy employee, this seemed like a form of slow torture. Finally he asked why Watson didn't just get it over with and fire him. "Fire you?" Watson replied. "We've just spent $30,000 educating you!"

"That's the right attitude," says assistant professor of business administration Amy Edmondson '81, Ph.D. '96, a specialist in organizational behavior. "But learning from mistakes is anything but automatic. Errors bring up feelings of shame, and we would rather not confront the bad feelings associated with our failures as individuals. In organizations, however, another factor emerges. It's important to surface errors--to make them known so they can be corrected and learned from--but because of the social pressure of needing to look good, be accepted, or get promoted, a group can have a way of magnifying the negative consequences of surfacing error."

For the last few years, Edmondson and colleagues have explored factors that encourage--or inhibit--surfacing errors in small work groups. She was invited by Lucien Leape, M.D. '59, adjunct professor of health policy, and David Bates, M.P.M. '90, associate professor of medicine, to join a study of hospital errors in drug administration. "Lots of mistakes are made in hospitals, in part due to the number of transactions and the complexity of the tasks," Edmondson says. "But there are built-in checks. For example, suppose a doctor prescribes quadruple the correct dosage of a drug. That's a mistake. But it doesn't kill the patient. It is a fatal error only if the patient actually receives that dosage--or, perhaps, receives it for several days running. Before that happens, the pharmacist could inquire about the order, the nurse in charge could question it, another nurse might ask a colleague if it seems right. There is a sequence; before the mistake harmed the patient, the entire working group would have to participate by not intercepting the original error. And when you study different work groups, you find that there are significant differences in the interception rates. In some groups, people are willing to say, 'Hey, that doesn't look right,' while in others they are not."

Interestingly, these differences show up among different units or wards in the same hospital. "It's not an 'organizational culture' effect," says Edmondson, "but a characteristic of a small work group [typically, four to 20 people]. It's very much a local phenomenon that has a lot to do with particular doctors and nurse managers." Edmondson analyzed "detected error rates" in eight hospital units whose management styles varied widely, from authoritarian and punitive to open and supportive. In the former, the nurse managers tended to dress in business suits and regard mistakes in a manner reflected by one erring nurse's comment: "I was made to feel like a two-year-old who had been bad." In more open units, nurse managers often wore scrubs, discussed their own mistakes with staff, and fostered an atmosphere that Edmondson describes as one of "psychological safety." Staff in such units said things like, "When I screw something up, I trust the people here not to make me feel worse than I do already," or "It's easy to take a risk in this group." Detected error rates had almost a direct linear relationship to the openness of the work group.

"The team leaders' behavior and attitude determine a lot," Edmondson explains. "Are they comfortable with a certain level of imperfection? Are they open and honest about their own vulnerability? Do they say things like, 'I don't have all the answers,' and invite input? If that's an inauthentic invitation--what amounts to a charade of participation by the staff--it can close the door even more powerfully than out-and-out authoritarianism. And how does the leader frame an error--is it seen in a performance context, or a learning context?"

Leadership that encourages the surfacing of errors can actually reduce mistakes significantly in the long run. If errors surface and are publicly discussed, team members can help each other avoid the booby traps hidden in their work environment. "You don't find yourself repeating the same mistake--since you didn't know about it--that a colleague made last week," Edmondson explains. "When errors aren't surfaced, an individual may be learning, but the group doesn't learn. Obtuse organizations can make the same mistake again and again."