A dangerous form of group-think

You probably know the old fable about the three blind men asked to describe an elephant. The first, stationed near a leg, concludes, "The elephant is round like a tree." The second, who happens to feel the trunk, exclaims, "No! The elephant is long and thin like a snake!" The third, with his hands placed on the side of the elephant, scoffs, "You two are nuts. The elephant is large and flat like a house." The implication of the fable is that each of us has a limited perspective and we should learn from one another before forming conclusions.

Not so for marketing researchers. We want to understand what people think and do as individuals in the marketplace. Our approach is that despite the disagreement, each blind man is 100% right. For us, consensus generates error. You say snake, I say tree, it's all good.

But, well-known fact, people in focus groups tend to try to form a consensus. A typical solution is to ensure that each of the group participants has a chance to form an individual opinion before the group discusses it; for example, they may read quietly to themselves and jot down a few notes before the discussion, or fill out a rating form. I've found that to be helpful, but a recent blog post (about the financial crisis, as it happens) makes me wonder if it's really enough.

Naked Capitalism cited a story about "social validation". Apparently, people in group settings will hesitate to assess the nature of what they perceive until they have checked in with those around them. The story concerns a lifeguard who nearly let someone drown because the the other lifeguards seemed unconcerned. What's disturbing is that he was unable to perceive that the swimmer was struggling because nobody else perceived it. This phenomenon has also been cited in the infamous 1964 killing of Kitty Genovese, in which 38 people witnessed the murder and did nothing to stop it, each concluding that because of a seeming lack of concern on the part of the other 37, the situation must somehow be OK!

I have always assumed that group-think was a kind of self-censorship due to social pressures. But in fact, people in group settings may postpone forming any perceptions until they check in with the group. No self-censorship is needed. Even disagreement with the group may just mean that a participant has an oppositional bias, and is reacting to the social consensus. It's still an artifact of group-think.

At long last, this explains a result I once saw in groups of technologically-sophisticated business people. They were asked to develop categories for a set of product claims. They could come up with any categories they chose, but "service", "reliability" and so on would be obvious ones.

The exercise turned out to be largely pointless. Groups assigned claims to categories very idiosyncratically; there was little inter-group consensus. Frequently it seemed that as soon as any member of a group "perceived" a connection between ideas, the rest of the group also perceived it. ("Yep!" "Yes!" "That makes sense." Even if it obviously didn't.) We in the back room alternated between humor and frustration.

I think on that occasion someone cited the old joke: the IQ of a group is the IQ of its stupidest member divided by the total number of members. That's harsh, but we need to at least keep in mind the possibility that participating in a group dulls the perceptual abilities of its members. Ouch.

What's a researcher to do? Groups will always have social validation issues, no matter how they are managed. Even if you make them take notes prior to discussion, participants are exposed to facial expressions, body language, and so on, which may influence their perceptions. There's no way out.

One solution? Think small. It may seem paradoxical, but in my experience, the opinions generated by smaller groups (three or four) are more diverse than those of larger groups (six or more). It may be that the smaller the group, the easier it is for the individual to feel that his or her perceptions are "equal" to everyone else's. (Would Kitty Genovese have been saved if there had been just a couple of neighbors looking out their windows that night?)

We probably should also address social validation issues more agressively in our research approach. Thoughts:

1) Spend less energy on generating gut-level perceptions, which are questionable, and more time on the internal experience of perceiving. You may even want to avoid having the group articulate base perceptions because once these are formed, they're hard to change. Focus instead on the components of the decision-making process; e.g., emotions, thoughts, what other people say, what you say.

2) Role play. Create scenarios where group members act out operating as individuals, for example, telling a sales person what they want or need.

3) An experiment I'd like to try: instead of the moderator playing "devil's advocate" to a consensus position, challenge group think by asking a group member to take this role. When the moderator does this, the group tends to close ranks around the consensus. It might encourage the group to reconsider consensus perceptions if a group member is the challenger.

You may have better ideas, and I'd love to hear them!