Hidden Profiles

Consider a study of serious errors within working groups.³ The purpose of the study was to see how groups collaborate to make personnel decisions.

Résumés for three candidates applying for a position of marketing manager were placed before group members. The experimenters rigged the attributes of the candidates so that one applicant was clearly the best for the job described. Group members would learn who was best if they were able to see, and to consider, all available information. (The experimenters verified this by forming some control groups in which all subjects were given full information. These groups invariably chose the best candidate.) The subjects were given information packets, each containing a subset of information from the résumés, so that each group member had only part of the relevant information. The groups consisted of three people, some operating face-to-face, some operating online.

Almost none of the deliberating groups made what was conspicuously the right choice! The reason is simple: people failed to share their information in a way that would permit the group to make that choice. Members tended to share positive information about the winning candidate and negative information about the losers. They suppressed negative information about the winner and positive information about the losers. Hence, their statements served to “reinforce the march toward group consensus rather than add complications and fuel debate.”⁴

Or consider a simulation of political elections, in which information was parceled out to individual members about three candidates for political office. In this simulation, properly pooled information could have led to what was clearly the best choice, candidate A.⁵ In the first condition, each member of the four-person groups was given most of the relevant information (66 percent of the information about each candidate). In that condition, 67 percent of group members favored candidate A before discussion, and 85 percent did so after discussion.⁶ This is a clear example of sensible aggregation of information. Groups significantly outperformed individuals, apparently because of the exchange of information and reasons. Here, then, is a clear illustration of the possibility that groups can aggregate what members know in a way that produces sensible outcomes.

In the second condition, by contrast, the information that favored candidate A was parceled out to various members of the group so that only 33 percent of the information about each candidate was shared. Furthermore, as the condition was designed, the shared information favored two unambiguously inferior candidates, B and C. In that condition, fewer than 25 percent of group members favored candidate A before discussion—a natural and rational product of the initial distribution of information. But if the unshared information were to emerge through group discussion and were to be taken seriously, the group would end up selecting candidate A. The bad news is that the number of people favoring candidate A actually fell after discussion, simply because the shared information had disproportionate influence on group members.⁷

In other words, groups did worse, not better, than individuals when the information was distributed so that the key material was unshared and could emerge only from discussion. Under those circumstances, the shared information was far more influential than the unshared information, to the detriment of the group’s ultimate decision.

From this and many similar studies, the general conclusion is that shared information has a far larger impact than unshared information. If everyone in a six-person group has the same information, and if two members have information of their own that everyone else lacks, there is a good chance that the unshared information will never be discussed and shared information will dominate. More specifically, when most of the unshared information is opposed to the position that is initially the most popular, that unshared information will be omitted from the discussion—and will have little influence on what groups end up choosing.⁸ Garold Stasser and William Titus, who conducted many relevant experiments, conclude that “group decisions and postgroup preferences reflect the initial preferences of group members even when the exchange of unshared information should have resulted in substantial shifts in opinion.”⁹

Nor does discussion increase the recall of unshared information. On the contrary, its major effect is to increase people’s recall of the attributes of the initially most popular candidate or position.¹⁰ The most disturbing conclusion is that when key information is unshared, groups are more likely to choose an inferior option after discussion than their individual members might before any discussion occurs.¹¹

The Common-Knowledge Effect

These results are best understood as a consequence of the common-knowledge effect, by which information held by all group members has a lot more influence on group judgments than information held by one member or a few.¹² More technically, Stasser and Titus find that the “influence of a particular item of information is directly and positively related to the number of group members who have knowledge of that item before the group discussion and judgment.”¹³ When information is unshared, group judgments have been found to be no more accurate than the average of the individual judgments, even though—and this is the central point—the groups have possession of more information than do any of the individuals.¹⁴

The more that information is shared, the more impact it will have as discussion proceeds, precisely because it is more likely to be discussed. In a representative study, deliberating groups failed to elicit and use unshared information, and they would have lost nothing in terms of accuracy if they had simply averaged the original judgments of the people involved—a clear finding that deliberation may not improve on the judgments that come from taking statistical averages.¹⁵ (As we have seen, deliberation may also produce worse judgments.)

Cognitively Central, Cognitively Peripheral

Many of the hidden-profile experiments involve volunteer participants from college courses. It is natural to ask: Would the same results be found in the real world?

They would indeed. Consider a study of hiring by high-level executives.¹⁶ In that study, no experimenter controlled the information that the executives had about the various candidates. Instead, the executives’ knowledge arose naturally from their own information searches. As a result of those searches, some information was known to all, some was shared by some but not by all, and some was held by only one person.

The result? Even among high-level executives, common information had a disproportionately large impact on discussions and conclusions. The executives gave disproportionately little weight to valuable information held by one person or a few. As a result, the executives made bad decisions, because they did not use the information that their colleagues had.

The same study offers another finding of considerable importance. Some group members are cognitively central, in the sense that their personal knowledge is also held by many other group members.¹⁷ What the cognitively central members know, other people know as well. A cognitively central group member is thus defined as one who possesses information in common with all or most group members. By contrast, other group members are cognitively peripheral; their own information is uniquely held. What they know is known by no one else, and what they may know might be really important.

For that very reason, well-functioning groups need to take advantage of cognitively peripheral people. These people are especially significant. But in most groups, cognitively central people have a disproportionate influence in discussion and participate more in group deliberations. By contrast, cognitively peripheral people end up having little influence and participate less, often to the group’s detriment.

A simple explanation for these results is that group members prefer to hear information that is commonly held—and prefer to hear people who have such information. Cognitively central people also have high levels of credibility, while cognitively peripheral people have correspondingly low levels. And in the particular experiment just described, the executives who were cognitively central ended up having a greater influence on the final judgments. The general conclusion is that when some group members count more than others, it is often because members to whom people most listen know only what everyone else knows. Unfortunately, people who are more peripheral, with unique knowledge, are often the ones group members most need to hear.

As might be expected, a group’s focus on shared information increases with the size of the group.¹⁸ For this reason, groups can go badly wrong even if they are fairly large, and would seem to be well informed—a real problem for the wisdom of crowds. In a study designed to test judgments about candidates for office, involving both three-person and six-person groups, all the group discussions focused far more on shared information than on unshared information—but the effect was significantly greater for six-person groups. Most remarkably, the researchers reported, “it was almost as likely for a shared item to be mentioned twice as it was for an unshared item to be mentioned at all.”¹⁹

It follows that for very large groups—twelve people, thirty people, two hundred people—the effect of shared information will be compounded. And despite the failures of their deliberations, group members tend to be significantly more confident about their judgments after discussion—and we have seen that confidence and error are a bad combination.

Why Hidden Profiles?

Why do hidden profiles remain hidden? The principal explanations build on the informational and social influence accounts emphasized thus far. When information is held by all or most group members, it is especially likely, as a statistical matter, to be repeated in group discussion and hence more likely to be influential than information that is held by one person or a few people. There are two points here. First, shared information, simply because it is shared, is more likely to be mentioned and explored during group discussion. Second, information held by all or most group members is more likely to influence individual judgments, and those judgments will in turn affect the judgments of the group.

Suppose that a team of five people is advising the president whether to embark on military action to combat a perceived threat. If each of the five people has information indicating that the use of military force would be successful, that information is likely to be discussed by the group—far more so than are separate parcels of information, individually held by each adviser, suggesting that the use of force would run into trouble. If the team of advisers stresses the information that is held in advance by each, that information will have a disproportionate influence on its ultimate decision. This is a simple statistical point.

But hidden profiles remain a lot more hidden than would be predicted by statistics alone. To understand the additional element, consider the finding that low-status members of groups are “increasingly reluctant over the course of discussion to repeat unique information.”²⁰ Those in the group who are inexperienced or are thought to be low in the hierarchy are particularly loath to emphasize their privately held information as discussion proceeds. In a business, then, leaders are unlikely to receive the views of those who do not have much experience or respect, even if these less powerful people know something important. The same can happen in the White House and in the boardroom at General Motors.

This important finding suggests that group members, and especially lower-status ones, are nervous about emphasizing information that most group members lack. Indeed, lower-status members might well shut up. They do so partly because of the difficulty of establishing the credibility and relevance of the information they have, and partly because they risk the group’s disapproval if they press a line of argument that others reject. We have seen that cognitively peripheral members have little influence on the group. Those who have uniquely held information end up participating less than those who have shared information, and what the cognitively peripheral members have to say is undervalued. In many deliberating groups, people who emphasize uniquely held information take an obvious social risk. Unfortunately for the group, they understand this risk.

Hidden profiles thus create a big problem for many groups, which fail to get information that they need and are therefore led to error. Group members typically underestimate the performance of low-status members and typically overestimate the performance of high-status members, in a way that gives high-status members a degree of deference that is not warranted by reality.²¹

Liking

Here is related problem. Those who discuss shared information tend to get a nice social reward: after a group member discusses shared information, other group members end up thinking that he or she is more competent and likable. Maybe that finding is not so surprising. It may be built into human nature to respect and like people more once they have told you something that you already know. (“If they tell me that, they must be smart, and nice too!”) But perhaps more surprisingly, and certainly more tellingly, a person who is provided with information that she already has ends up feeling more competent and likable in her own eyes as well.²² Strange but true (and important): if someone tells you something you already know, you will probably like not only that person, but also yourself, a bit better as a result. (“If they tell me that, I must be smart, and nice too!”)

In face-to-face discussions and in purely written exchanges, people give higher ratings (in terms of knowledge, competence, and credibility) both to themselves and to others after receiving information that they knew already. This finding is full of implications. Within groups, those who reject the apparent consensus risk their own reputation along important dimensions. They also endanger their own self-image. It is no wonder, in this light, that common knowledge is emphasized and that people self-censor even when they could reveal a hidden profile.

Here’s a suggestion: suppose that you have valuable but unshared information. You may do best to earn credibility and trust first. You can do that “by telling others what they already know before telling them what they do not already know.”²³ That strategy might work, helping to furnish information that could steer groups in the right directions.

A Very Quick Recap of Part 1

Our goal in part 1 has been to catalog four common failures of deliberating groups. (1) Sometimes individual errors are amplified, not merely propagated, as a result of deliberation. (2) Groups fall victim to cascade effects, as the early speakers or actors ensure that people do not learn what is known by their successors. (3) Because of group polarization, members of deliberating groups often end up in a more extreme position in line with their predeliberation tendencies. (4) In deliberating groups, shared information often dominates or crowds out unshared information, ensuring that groups do not learn everything that their members know.

We have seen that informational signals and reputational pressure help to explain all four errors. A disturbing result is that groups often do not improve on, and sometimes do worse than, their average or median member. For many deliberating groups, a crucially important task is to devise strategies that will overcome these problems and increase the likelihood that group judgments will take advantage of the knowledge held by group members or perhaps by those who are not in the group at all. How can that be done?

Let’s answer that question.