What's wrong with that?

For one thing, you're ignoring people who might know a lot that would be useful in making the decision. You're accepting the idea that because you're in the CEO seat, you somehow know more or you're smarter than everyone else. But what you're really doing is cutting yourself off from hearing options or ideas that might be better.

How do you create the kind of atmosphere where in­formation flows freely?

You have to recognize that your position can be a hindrance to getting the best information. And so can your personality. My own greatest enemy is my personality—I can convince the people on my team of a point of view. I'm older than they are. I've done more research than they have. I know more than they do. I can influence them perhaps too much and therefore not get the best answers. So when we were "doing the research for Good to Great, I built a culture that began with disagreements, that set people up to disagree with each other and disagree with me.

I tried to increase what I call my questions-to-statements ratio. I learned this from the Good to Great leaders we were studying. They were just marvelous at igniting dialogue and debate with Socratic questions. And I tried to make heroes out of those on my team who identified flaws in my thinking. At the next meeting I might say, "I really want to give Leigh or Brian or Stefanie credit. She really pushed my thinking, and I wasn't looking at this right."

I looked for people with a streak of irreverence and independent thought. One of my favorite researchers is a young man who went to Princeton, majored in medieval literature, and then joined the Marine Corps. Now, that's independent thinking. I wanted him on my team because he's not going to care what I think.

The really critical part came in designing the re­search so that for every piece of the puzzle—for every case, every analysis—someone on the team knows that piece as well as I do or better. This was a key mechanism to reduce the odds that my au­thority and strong personality would override the evidence.

Does having that kind of team make it harder to reach consensus?

I really want to underscore something. This is not about consensus.