Group Decision-Making

In the previous section, we explored some of the important ways that being in a group affects individual members' behavior and, in turn, influences the group's overall performance. Beyond achieving high performance, another important task of groups is to make decisions. We often entrust groups, rather than individuals, with key decisions in our societies—for example, those made by juries and political parties. An important question to ask is whether we are right to trust groups more than individuals to reach sound decisions. Are many heads really better than one?

It turns out this can be a hard question to answer. For one thing, studying decision making is difficult, because it is hard to assess the quality of a decision based on what was known at the time, independent of its outcome. This is especially challenging because we naturally tend to look too much at the outcome when we evaluate decision making—a phenomenon known as the outcome bias. Moreover, studying decision making in laboratory environments has generally involved giving group members more information than they would typically have in the real world, so the results may not always generalize.

Nevertheless, with these caveats in mind, it is possible to draw some tentative conclusions about when and why groups make better decisions than individuals—and when and why they may end up making worse ones.

Process Gains in Group versus Individual Decision Making

One important factor that helps groups outperform individuals on decision-making tasks is the type of interdependence they have. In general, positively interdependent (cooperative) groups tend to make better decisions than both negatively interdependent (competitive) groups and individuals, particularly on complex tasks. These process gains come from several factors. When group members interact, they often generate new ideas and solutions they would not have reached individually. Group members are also more likely than individuals to notice and correct mistakes. They have better collective memory—many minds hold more relevant information than one—and superior transactive memory, in which interactions between members facilitate recall of important material. And when individual members share information that is unique to them, they increase the total data the group can draw on. Given these advantages, are there ever times when groups make less optimal decisions than individuals? If you have ever sat in a group where, in hindsight, a foolhardy decision was reached, you probably already have your answer. The more interesting question is: why are many heads sometimes worse than one? Let's explore some of the most dramatic reasons.

Process Losses Due to Group Conformity Pressures: Groupthink

Groups make effective decisions only when they can use the advantages of group membership outlined above. These conditions are not always met in real groups. One group process that can lead to very poor decisions is groupthink. Groupthink occurs when a group made up of members who may actually be very competent—and thus quite capable of making excellent decisions—nevertheless makes a poor one as a result of a flawed group process and strong conformity pressures.

Groupthink is more likely in groups where members feel a strong social identity—for instance, when there is a powerful, directive leader who creates a positive group feeling, and in times of stress and crisis when the group must rise to the occasion. The problem is that groups suffering from groupthink become unwilling to seek out or discuss discrepant or unsettling information, and members do not voice contradictory opinions. Because members are afraid to express ideas that contradict the leader, or to bring in outsiders with other information, the group is prevented from making a fully informed decision. Figure 10.9 summarizes the basic causes and outcomes of groupthink.

Although some scholars are skeptical of the importance of groupthink in real group decisions, many others suggest it was involved in a number of well-known but very poor decisions made by government and business groups. Historical decisions analyzed in terms of groupthink include the decision to invade Iraq made by President George Bush and his advisors, with the support of other governments (including the United Kingdom, Spain, Italy, South Korea, Japan, Singapore, and Australia); President John F. Kennedy and his advisors' 1962 decision to commit U.S. forces to help invade Cuba to overthrow Fidel Castro; and the policy of appeasement of Nazi Germany pursued by many European leaders in the 1930s. Groupthink has also been applied to less well-known but important domains, including pack journalism—and even to account for perceived anti-right-wing political biases among social psychologists.

Careful analyses of decision making in these historical cases have documented the role of conformity pressures. In fact, the group process often seems arranged to maximize conformity rather than foster free and open discussion. In meetings of the Bay of Pigs advisory committee, for instance, President Kennedy sometimes demanded a voice vote of individual opinions before the group discussed the pros and cons of a new idea. The result is a general unwillingness to express ideas that do not match the group norm.

Pressures for conformity also lead to a situation in which only a few members are involved in conversation while the others express no opinions. Because little or no dissent is expressed, members come to believe they are in complete agreement. In some cases, the leader may even select individuals (known as mindguards) whose job is to help quash dissent and increase conformity to the leader's opinions.

An outcome of high conformity is that the group begins to see itself as extremely valuable, highly capable, and invulnerable. Members develop very high levels of conformity and social identity. Although this social identity can have some positive outcomes—commitment to group goals, and members feeling good about themselves—it also tends to produce illusions of invulnerability, leading members to feel superior and to believe they do not need outside information. Such a situation is often conducive to poor decision making, sometimes with tragic consequences.

Interestingly, the composition of the group itself can affect the likelihood of groupthink. More diverse groups can ensure a wider range of views is available, reducing the risk. The more homogeneous a group is—in internal characteristics such as beliefs, and external characteristics such as gender—the more at risk it may be. Perhaps, then, mixed-gender corporate boards are more successful partly because they are better able to avoid the dangerous phenomenon of groupthink.

Cognitive Process Losses: Lack of Information Sharing

Although group discussion generally improves decision quality, this holds only if the group discusses the information most useful to the decision. One difficulty is that groups tend to discuss some types of information more than others. In addition to pressures to focus on information from leaders and consistent with group norms, discussion is influenced by how the relevant information is originally shared among members. The problem is that group members tend to discuss information they all have access to while ignoring equally important information available to only a few members—a tendency known as the shared information bias.

Research Focus

Poor Information Sharing in Groups

In one demonstration of the shared information bias, Stasser and Titus (1985) used an experimental design based on the hidden-profile task, as shown in Table 10.1. Students read descriptions of two candidates for a hypothetical student-body presidential election, then met in groups to discuss and pick the best candidate. The information was arranged so that one candidate (Candidate A) had more positive qualities overall than the other (Candidate B). Reflecting this, in groups where all members were given all the information about both candidates, members chose Candidate A 83% of the time after discussion.

However, in some cases the experimenters made the task harder by creating a "hidden profile," in which each member received only part of the information. All the information was potentially available to the group, but it needed to be properly shared to make the correct choice. When the information favoring Candidate B was shared but the information favoring Candidate A was not, only 18% of groups chose A; the rest chose the inferior candidate. This happened because, although members collectively had access to all the positive information, the information not originally shared among all members was never discussed. This bias occurred even among participants explicitly instructed to avoid expressing their initial preferences and to review all available facts.

Although poor information sharing seems to occur frequently—at least in experimentally created groups—it does not occur equally under all conditions. Groups share information better when members believe there is a correct answer that can be found through sufficient discussion, and when they are forced to continue discussing even after they believe they have covered everything. These findings suggest that an important job of the group leader is to continue discussion until convinced all relevant information has been addressed.

The structure of the group also influences information sharing. Groups whose members are physically separated, and thus have difficulty communicating, may need to reorganize to improve communication. The status of members matters too: lower-status members may have less confidence and be unlikely to express opinions. Wittenbaum found that higher-status members were more likely to share new information—though they may also dominate the discussion even when their information is no more valid or important. Groups are also more likely to share unique information when members do not initially know the alternatives or the preferences of other members.

Findings that groups neither share nor discuss originally unshared information have disconcerting implications, because they suggest group discussion can lead to poor judgments. Not only is unshared information left off the table, but because shared information is discussed repeatedly, it is likely to be seen as more valid and to have greater influence due to its high cognitive accessibility. Members of a working group often arrive with different types of information that need to be presented. In a design team for a new building, for example, the architects, engineers, and customer representatives will have different and potentially incompatible information. Leaders of working groups must be aware of this and work hard to foster open climates that encourage information sharing and discussion.

Given its pitfalls, why is the shared information bias so pervasive? Recalling the confirmation bias discussed in the chapter on social cognition, perhaps it reflects that tendency played out at the group level, where members collaborate to provide confirmatory evidence for each other's positions. It could also reflect a tendency to use groups to reinforce one's own views—groups becoming places where people mutually validate shared perspectives at the expense of searching for alternatives. If so, and given that we often associate with similar others, it may be important to seek out the views of members likely to be most different from our own, in order to weaken the damaging effects of the shared information bias.

Cognitive Process Losses: Ineffective Brainstorming

One technique frequently used to produce creative decisions in working groups is brainstorming. It was first developed by Osborn (1953) to increase the effectiveness of group sessions at his advertising agency. Osborn believed people could effectively "storm" a problem by sharing ideas in groups, and that creative solutions would increase when members generated many ideas and deferred judgments about quality until later. Brainstorming was therefore based on these rules:

• Each member should create as many ideas as possible, no matter how silly, unimportant, or unworkable they seemed.
• As many ideas as possible should be generated by the group.
• No one was allowed to offer opinions about the quality of an idea (even one's own).
• Members were encouraged and expected to modify and expand upon others' ideas.

Researchers have devoted considerable effort to testing brainstorming, yet—despite the creativeness of the idea itself—there is very little evidence that it works. Virtually all individual studies, as well as meta-analyses, find that regardless of the exact instructions, brainstorming groups do not generate as many ideas as expected, and the ideas they do generate are usually of lesser quality than those produced by an equal number of individuals working alone and then sharing results. Brainstorming is thus another case where, despite the expectation of a process gain, a process loss is observed instead.

Several explanations have been proposed and found important. One obvious problem is social loafing, and research suggests it does cause part of the problem. Paulus and Dzindolet found that social loafing in brainstorming groups occurred partly because individuals perceived that others were not working very hard and matched their own behavior to this perceived norm. To test this more directly, Diehl and Stroebe compared face-to-face brainstorming groups with equal numbers of individuals working alone, and found the face-to-face groups generated fewer and less creative solutions. For some face-to-face groups, the researchers set up a television camera to make individual contributions identifiable; being identifiable reduced social loafing and increased productivity—but the face-to-face groups still did not match the individuals.

Even though members are told that no evaluation of idea quality will be made, individuals may still be unwilling to state some ideas for fear of being negatively evaluated by others. When told that other members are more knowledgeable, individuals reduce their contributions; when convinced they are experts, their contributions increase.

Although social loafing and evaluation apprehension matter, the most important difficulty is that being in a group hinders idea production and expression. In a group, only one person can speak at a time, which can cause people to forget their ideas because they are listening to others, or to miss what others say because they are thinking of their own ideas—a problem known as production blocking. In other words, while individuals working alone can spend all the available time generating ideas, participants in face-to-face groups must perform other tasks as well, reducing their creativity.

Diehl and Stroebe demonstrated the importance of production blocking in another experiment. Rather than changing the real group, they created production blocking in the individual condition through a turn-taking procedure: individuals working in separate cubicles had to speak their ideas into a microphone, but only when no one else was speaking. Having to coordinate this way decreased individuals' performance so much that they were no longer better than the face-to-face groups.

Follow-up research showed that the main factor in productivity loss is that face-to-face group members cannot make good use of the time spent waiting for others. While waiting, they forget their ideas because they must concentrate on negotiating when it will be their turn to speak. Even when researchers gave face-to-face groups extra time to compensate for waiting, they still did not reach the productivity of individuals. Monitoring others' behavior and the delay in waiting to speak reduce the ability to think creatively.

Although brainstorming is a classic example of group process loss, there are ways to make it more effective. One variation is the nominal group technique, which uses individual sessions to generate initial ideas, followed by face-to-face meetings to discuss and build on them. Participants first work alone to generate and write down ideas before discussion begins; the group then records the ideas, using a round-robin procedure to ensure everyone gets a chance to contribute. Other similar approaches include the Delphi technique.

Advances in technology let individuals work together on creativity tasks via computer. These group support systems are used in many businesses and organizations. Each individual works at their own computer; as they write suggestions, the ideas are passed to other members via the network, so everyone can see all suggestions, including their own.

Research has found electronic brainstorming more effective than face-to-face brainstorming, largely because it reduces production blocking. Groups that work virtually rather than face-to-face are also more likely to share unique information. Each individual has all members' comments at hand and can read them when convenient, alternating between reading and writing without waiting to express ideas. Electronic brainstorming can also reduce evaluation apprehension, particularly when contributions are anonymous.

In summary, the most important conclusion from the brainstorming literature is that the technique is less effective than expected because members must do other things in addition to being creative. This does not mean brainstorming is useless overall—modifications of the original procedure have proven quite effective at producing creative thinking. Techniques that begin with individual thought, followed by group discussion, represent the best approaches. When you are in a group that needs to make a decision, use this knowledge: ask members to think about and write down their own ideas before the discussion begins.

Group Polarization

One common decision-making task is for a group to reach consensus on a judgment—where to hold a party, whether a defendant is innocent or guilty, or how much a corporation should invest in a new product. Whenever a majority of members favors a given opinion, even by a slim margin, the group is likely to adopt that opinion. This is expected, since conformity pressures should make the group's final judgment reflect the average of members' initial opinions.

Although groups generally show pressures toward conformity, the tendency to side with the majority after discussion is even stronger than this. Groups often make more extreme decisions, in the direction of the existing norm, than we would predict from members' initial opinions. Group polarization is said to occur when, after discussion, the attitudes of individual members become more extreme than they were before the group began discussing the topic. This may seem surprising, given the widespread belief that groups push people toward consensus and the middle ground; in fact, they may lead to more extreme decisions than individuals would make alone.

Group polarization was initially observed using problems in which members indicated how an individual should choose between a risky but very positive outcome and a certain but less desirable one. Consider the following:

Frederica has a secure job with a large bank. Her salary is adequate but unlikely to increase. However, she has been offered a job with a relatively unknown startup, where the likelihood of failure is high and the salary depends on the company's success. What is the minimum probability of the startup's success you would find acceptable to make it worthwhile for Frederica to take the job?

Research finds group polarization on these decisions: the group's recommendation is riskier (requiring a lower probability of the startup's success) than the average of members' initial opinions. This can be explained partly by diffusion of responsibility. Because the group as a whole takes responsibility, individuals may be willing to take a more extreme stand, since they can share the blame if the risky decision fails.

But group polarization is not limited to decisions involving risk. In an experiment by Myers and Kaplan, groups of students assessed the guilt or innocence of defendants in traffic cases. The researchers manipulated the strength of the evidence so that in some groups it was strong and in others weak. This produced two kinds of juries—some in which the majority initially favored conviction (strong evidence) and others in which the majority initially favored acquittal (weak evidence). Participants expressed their opinions both before and after deliberation.

As Figure 10.10 shows, opinions about guilt or innocence were more extreme after discussion than they were, on average, before. Members of juries that initially favored conviction became more likely to believe the defendant was guilty, and members of juries that initially favored acquittal became more likely to believe the defendant was innocent. Similarly, Myers and Bishop (1970) found that groups of college students with initially racist attitudes became more racist after discussion, whereas those with initially antiracist attitudes became less racist. Similar findings appear across a wide variety of topics and many cultures.

The juries in this research received either strong or weak evidence about a defendant's guilt and were then either allowed or not allowed to discuss it before deciding. Demonstrating group polarization, juries that discussed the case made significantly more extreme decisions than those that did not. Data are from Myers and Kaplan.

Group polarization does not occur in all groups and settings; it tends to happen when two conditions are present. First, members must have an initial leaning toward a given opinion. If members generally support liberal policies, their opinions are likely to become more liberal after discussion—but if the group is evenly split between liberals and conservatives, polarization would not be expected. Second, polarization is strengthened by discussion of the topic. In Myers and Kaplan's research, conditions in which members expressed opinions but did not discuss the issue showed less polarization than conditions that included discussion.

Group polarization has also been observed in important real-world contexts, including financial decision making in corporate boardrooms. It has been argued that recent political polarization in many countries—for example, between "blue" Democratic and "red" Republican states in the United States—occurs largely because each group spends time communicating with like-minded members, leading to more extreme opinions. Some have argued that terrorist groups develop extreme positions and engage in violence as a result of group polarization in their everyday interactions: as members, all of whom initially hold some radical beliefs, meet and discuss their concerns, their opinions polarize and become progressively more extreme. Removed from any moderating influences, they may eventually become mass killers.

Group polarization results from both cognitive and affective factors. The persuasive arguments approach is cognitive in orientation. It assumes there is a set of potential arguments supporting any opinion and another set refuting it, and that an individual's current opinion is based on the arguments they are aware of. During discussion, each member presents arguments supporting their own opinion. Because members are initially leaning in one direction, many arguments will support that leaning, exposing each member to new supporting arguments and polarizing the group's opinions. Supporting this, research shows the number of novel arguments mentioned is related to the amount of polarization, and that there is little polarization without discussion. Note the parallels between this approach and the concept of informational conformity.

Group polarization is also based partly on members' affective responses—particularly the social identity they receive from being good group members. In their desire to create a positive social identity, members try to differentiate their group from other groups by adopting extreme beliefs. So polarization is determined not only by ingroup norms but also by movement away from the norms of relevant outgroups. Groups with well-defined (extreme) beliefs are better able to produce social identity than groups with more moderate (and potentially less clear) beliefs. Note the similarity of this account to the notion of normative conformity.

Polarization effects are stronger when members have high social identity. Diane Mackie had participants listen to three people discussing a topic, supposedly to become familiar with the issue. The speakers were described as members of a group the participants would join in the upcoming session, members of a group they were not expecting to join, or simply individuals who were not a group at all. Mackie found that the perceived norms of the (future) ingroup were seen as more extreme than those of the other group or the individuals, and that participants were more likely to agree with the ingroup's arguments. This supports the idea that group norms are perceived as more extreme for groups people identify with. Another experiment by Mackie supported the social-identity prediction that the existence of a rival outgroup increases polarization, as members try to differentiate themselves by adopting more extreme positions.

Taken together, the research reveals another potential problem with group decision making: it can be polarized. These shifts toward more extreme positions have a variety of causes and occur more under some conditions than others, but they must be kept in mind whenever groups come together to make important decisions.

Social Psychology in the Public Interest

Decision Making by a Jury

Although many countries rely on the decisions of judges in civil and criminal trials, the jury is the foundation of the legal system in many others. The notion of a trial by one's peers rests on the assumption that average individuals can make informed, fair decisions when working together in groups. But given all the problems facing groups, social psychologists and others frequently wonder whether juries are the best way to make these decisions, and whether a jury's particular composition influences the outcome of its deliberation.

As small working groups, juries have the potential to produce good or poor decisions, depending on many of the factors discussed in this chapter—both person characteristics and group process. In terms of person variables, there is some evidence that juror characteristics matter. Individuals who have served on juries before are more likely to be seen as experts, more likely to be chosen as foreperson, and give more input during deliberation. Status also matters: jurors with higher-status occupations and education, males rather than females, and those who talk first are more likely to be chosen as foreperson and contribute more. As in other small groups, a minority of members generally dominates the jury discussion, and there is often a tendency toward social loafing—so relevant information or opinions may remain unshared because some individuals rarely participate.

Perhaps the strongest evidence for the importance of member characteristics concerns the selection of death-qualified juries in trials where a potential sentence includes the death penalty. To be selected, potential members must indicate they would, in principle, be willing to recommend the death penalty. In some countries, potential jurors opposed to the death penalty cannot serve. This creates a potential bias, because individuals who say they would never vote for the death penalty are also more likely to be rigid and punitive—and thus more likely to find defendants guilty—increasing the chances of a conviction.

Although some member characteristics influence jury decisions, group process plays a more important role in the outcome than member characteristics do. Like any group, juries develop their own norms, which can profoundly affect how they reach decisions. Different juries take very different approaches: some spend a lot of time in initial planning, while others jump straight into deliberation. Some base discussion on a review and reorganization of the evidence, waiting to vote until everything has been considered; others first determine the group's preference by taking a poll, then organize discussion around those opinions. These two approaches are used about equally often but may sometimes lead to different decisions.

Perhaps most important, conformity pressures strongly affect jury decision making. As Figure 10.11 shows, when more jury members hold the majority position, it becomes increasingly certain their opinion will prevail. This is not to say minorities can never be persuasive, but it is very difficult. The strong influence of the majority is probably due to both informational conformity (more arguments support the favored position) and normative conformity (people are less likely to want to be seen disagreeing with the majority).

This figure shows decisions of six-member mock juries that made "majority rules" decisions. When the majority initially favored guilty, the jury almost always voted guilty; when the majority initially favored innocent, the jury almost always voted innocent. Juries were frequently hung (unable to decide) when the initial split was three to three. Data are from Stasser, Kerr, and Bray.

Research has also found that evenly split juries (three to three, or six to six) tend to show a leniency bias, voting toward acquittal more often than conviction, all else equal. This is partly because juries are usually instructed to assume innocence unless there is sufficient evidence of guilt—applying a burden of proof "beyond a reasonable doubt." The leniency bias does not always occur, but it is more likely when the potential penalty is more severe.

Given the difficulties groups face, you might worry that jury verdicts may not be particularly effective, accurate, or fair. Yet, despite these concerns, the evidence suggests juries may not do as badly as we would expect. The deliberation process seems to cancel out many individual juror biases, and the importance of the decision leads members to consider the evidence carefully.

Key Takeaways

• Under certain situations, groups can show significant process gains in decision making compared with individuals. However, a number of social forces can hinder effective group decision making, sometimes leading to process losses.
• Some process losses result from groupthink—when a group, as a result of a flawed process and strong conformity pressures, makes a poor judgment.
• Process losses may result from the tendency to discuss information all members have access to while ignoring equally important information available to only a few members.
• Brainstorming is a technique designed to foster group creativity. Although it often leads to process losses, alternative approaches—including group support systems—may be more effective.
• Group decisions can be influenced by group polarization—when members' attitudes become more extreme than they were before discussion.
• Understanding group processes helps us understand the factors that lead juries to make better or worse decisions.

Exercises and Critical Thinking

1. Consider a time when a group you belonged to experienced a process gain, and another when it showed a process loss in decision making. Which factors discussed in this section help explain these two outcomes?
2. Describe a current social or political issue where you have seen groupthink in action. Which features of groupthink were particularly evident? When have you been in a group situation where groupthink was evident? What decision was reached, and what was the outcome for you?
3. When have you been in a group that did not share information effectively? Why do you think this happened, and what were the consequences?
4. Outline two situations: one where brainstorming helped your group's decision making, and another where it was a hindrance. Why do you think brainstorming had these opposite effects?
5. What examples of group polarization have you seen in the media recently? How well do the ideas of normative and informational conformity explain it? What other factors might have been at work?
6. If you or someone you knew had a choice to be tried by a judge or a jury, which would you choose, given the research in this section, and why?

Source: Rajiv Jhangiani and Hammond Tarry. View original source.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License.