Group Communication

Three early studies of the forces at work in small groups will identify the themes of this essay. The three studies are by Solomon Asch concerning pressures toward conformity with a majority, by Deutsch and Gerard on two types of conformity, and by Stoner and others on group polarization. We can learn from these studies that both the influences upon group behavior, and the decision processes of group members themselves, are of two kinds--one at the task level of group activity, dealing with the substance of the group's duties, and one at the social level of group activity, dealing with the interpersonal relations developed while the group is at work on their task.

Asch. Asch's experiments placed naive subjects in a situation where their perceptions about matters of fact were brought into conflict with their relationship with a group of peers. A panel of individuals, usually eight in number, was shown three lines of unequal length and was asked to state which of the three lines was equal in length to a fourth line, the standard. The figure below replicates the stimulus object:



Sense perception says that Comparison Line #1 is equal to the Standard Line. But on several "critical" trials the naive subject heard other group members say that Line #2 or Line #3 was the matching line. In some experimental trials, the naive subject was the only person who was not instructed to give false answers. In other trials, the naive subject was permitted to give written answers. And in other trials, the number of "confederates" who were instructed to give false answers was reduced from eight to a lower number (1-4), or was raised to sixteen. In about two-thirds of the critical trials subjects did not yield, but gave correct answers which contradicted the group majority.

The purpose of the experiments was to discover how people respond in this conflict situation to various interpersonal pressures. We will return to Asch's experiments later in this essay, to discuss individually some of the interpersonal pressures at work in small groups, but for now we will note that those group members who did yield to group pressure did so in three distinct ways. Two of the yielding responses dealt with the substance of the task, that is, the actual judgments of the lengths of lines. A small number of naive subjects experienced a change in their perceptions--they came to see the lines as matching the way the majority said they matched; and others experienced a change in their judgments--they came to believe that their perceptions were mistaken, that the majority must be correct. The third yielding response by naive subjects dealt with interpersonal relations; subjects continued privately to trust their own perceptions and judgments, but they changed their public behavior, and gave incorrect answers, in order to avoid disapproval from the other group members.

Deutsch and Gerard. In a variant of Asch's matching-lines technique, Deutsch and Gerard created pressures more directly resembling those of a task-oriented small group. Their criticism of Asch's work was that the social pressures, which Deutsch and Gerard call normative influences, were present in the Asch experiments only incidentally--that the "group" did not interact or cooperate, and had no stake as a group in the outcome of the voting. In their version, Deutsch and Gerard told some groups that they would be rewarded if effective:

This group is one of twenty similar groups who are participating in this experiment. We want to see how accurately you can make judgments. We are going to give a reward to the five best groups--the five groups that make the fewest errors on the series of judgments that you are given. The reward will be a pair of tickets to a Broadway play of your own choosing for each member of the winning group. An error will be counted any time one of you makes an incorrect judgment.

Additionally, experimental subjects were distinguished as to those in face-to-face contact with their fellow group members as opposed to those visually separated, and as to those who made some commitment to an answer as opposed to those who made no commitment.

The instruction quoted above was sufficient to induce a sense of obligation to the group--over half of the "group" members mentioned such an obligation, whereas none of the experimental subjects in the "nongroup" condition did--and there were more than twice as many errors in the group than in the nongroup condition. Deutsch and Gerard argue that the errors observed by Asch can be explained as informational social influence, since it is not irrational to accept the testimony of others as to matters of fact. But the considerably greater errors observed in the 1955 study can only be explained as normative social influence, or as influence "to conform with the positive expectations of another" in order to gain a desired response.

Polarization Studies. A Master's study at M.I.T. in 1961 has led to an impressive line of research involving hundreds of journal articles (Isenberg). Stoner discovered that an effect of group discussion is to generate a group decision that involves more risk than any individual in the group would accept if deciding alone. The phenomenon was called the "risky shift," and was explained as the result of diffusion of responsibility among group members. A typical testing situation is this:

A chess player must decide whether to take a maneuver that might bring victory if successful or a defeat if unsuccessful. (Stoner, "Risky")

Group members individually would be somewhat favorable to the risky choice (try the move), but following discussion would adopt an even stronger preference for risk.

It was quickly discovered that on some occasions the effect of group discussion is to generate a group decision that involves more caution than any individual in the group would feel necessary if deciding alone. A typical cautious situation is this:

A couple must choose between allowing a complicated pregnancy to continue, with danger to the mother's life, or having the pregnancy terminated.

Group members before discussion would be somewhat favorable to the cautious choice (terminate the pregnancy), but after discussion would adopt as a group an even stronger preference for caution. This phenomenon was called a "cautious shift," but it and the risky shift have since been combined as group polarization effects, the tendency of group discussion to lead to a decision which exaggerates the pre-discussion preferences of the group members.

One effect of the addition of a cautious shift was to undermine the "diffusion of responsibility" explanation for the risky shift. Several other explanations of polarization have been proposed during the three decades of research into this phenomenon, but there are two dominant ones: persuasive arguments theory and social comparison theory. According to persuasive arguments theory, a group member in a moderately polarized group will hear more arguments favoring that side of the issue than favoring the other side. If those arguments are perceived by the group member to be persuasive--that is, to be novel and valid--then the member will change position in the direction of those arguments. As each group member does the same thing, the group as a whole tends to believe more strongly its initial opinions, or to move further toward the pole it favored before discussion. According to social comparison theory, a group member in a moderately polarized group will understand the values and opinions of the other group members, and will change positions in the direction favored by the group in order to win their approval. Both explanations are based on assumptions: the persuasive arguments theory assumes that individuals desire cognitive consistency, so that positions will change to reflect a new understanding of ideas, and the social comparison theory assumes that individuals desire approval from others, so that positions will change to win a desired response from fellow members of the group.


There is enough similarity among the studies of Asch, Deutsch and Gerard, and the writers on polarization to permit the general observation that the life of the group has two dimensions, one task-related and one social. The task dimension of the group would present to each group member various influences bearing on the substance of the group's decision. These influences include information presented by group members, their arguments interpreting and evaluating that information, and the very structure of the group's decision-making process. Success in collecting and considering information and arguments will lead to high productivity, productivity being a measure of the group's work at the task level. The social dimension of the group would present to each group member various influences bearing on the group member's relationship to others in the group. These influences include the unanimity and size of the group, surveillance by the group, and the group's prestige and cohesiveness. Success in adapting to the expectations of other group members will lead to high cohesiveness, cohesiveness being a measure of the group's work at the social level.

A variety of other bodies of material discussing small group communication supports the judgment that the life of the group is conducted on two levels, task and social. These other materials include distinctions among types of "tension" in a group, types of "roles" played by group members, types of statements made during the group's discussion, phases of group decision making, and levels of meaning present in any speech act. There is not space in the present essay to discuss this additional evidence concerning task and social levels in the group, but the ideas are summarized in a separate table. For our purposes, the importance of the distinction is that it offers a useful way to understand the entire process of group decision making, which involves individuals choosing among response options in consideration of various influences.

The understanding of the relationship between forces of influence in a small group and decision processes that is developed in this essay is represented in the diagram below.

Roughly, the diagram shows on the right side a variety of decision choices available to group members, and indicates that those choices will be influenced by some combination of the informational and normative forces listed at the left of the diagram. Three preliminary observations concerning the diagram must be made:

• The diagram is divided into the same task and social categories discussed above. Reading across the diagram, informational influences and internalization and independence responses are task-related categories. Normative influences and compliance and anti-conformity responses are social-related categories. (We will discuss identification as sharing features of internalization and compliance, rather than as a separate decision process.)
  
  
• The diagram is incomplete, for example in that it omits various mediating factors which help to determine which of the forces of influence a group member considers. Factors such as one's level of communication anxiety, gender, age, experience in groups, the work norms of the group, the cohesiveness of group members, and the nature of the task will encourage an interest at some times in informational or task-related concerns, at other times in normative or social-related concerns. The diagram also omits such contextual factors as stress, institutional demands, and time constraints, and such "outcomes" as productivity, member satisfaction, and cohesiveness.
  
  
• The diagram is a model, an abstraction, and therefore is a partial view of the group process. In this regard, it is most important to remember that the group process is actional and not causal; group members choose to attend to some influences and not others, choose to react with one behavior and not another. The model is laid out in a linear fashion, but people are not like machines responding to an electrical signal; we decode and interpret messages, engage our understanding and feelings, and consider some combination of selfish and benevolent interests as we decide to act. The diagram is intended to show some of the principal ways in which group processes can be discussed, but in emphasizing those it may seem to leave out the essentially human character of co-acting with others.

The remainder of this essay is devoted to an explanation of the diagram. We will discuss first the informational influences, then the normative influences, and finally the various decision processes.

Structure. The role of structure in group discussions, represented by an agenda or by a prescriptive discussion pattern such as the "reflective thinking" pattern, is contentious, but a strong body of opinion exists that groups perform better if their discussions are structured. The issue is how structure should arise.

In a representative early study, Larson compared four discussion methods for solving a problem. There was a "correct" answer to the problem, so the results of the groups could easily be measured. Control groups were given "no pattern," were left to their own lights to discover a path to solving the problem, thus, according to Larson, "theoretically duplicating conditions which exist when groups deliberate with no systematic analysis form governing their deliberations." Other groups followed a "single question form," others an "ideal solution form," and finally others the "reflective thinking form." For each analysis form, there were 32 trial discussions. Larson obtained these overall results, showing the number of correct answers among the 32 trials:



The "no pattern" groups arrived at correct answers in fewer than half the trials, whereas the "ideal solution" groups arrived at correct answers in over 80% of the trials. Larson concludes that "... successful group outcomes ... can be facilitated" by "prescriptive patterns for group problem-solving."

Larson's results have been challenged by Fisher (1980), who argues that Larson's groups did not meet long enough to develop a group "culture," and that the scores of the control group would have improved had normal group structuring been allowed to emerge. Of course, that analysis explicitly recognizes the productive value of structure; Fisher himself states, "I do not discourage the use of an agenda. Indeed, it is often quite beneficial in assisting a group's decision-making efforts" (133). And the argument that proficiency in using a discussion structure is a function of time spent together as a group suggests the possibility that more experience with the three prescribed agendas would have led to still greater productivity on the part of the groups in the other three conditions.

Probably the most popular prescriptive model for problem solving is the "reflective thinking" model, derived from John Dewey. In one version, this model would ask group members to discuss these questions:

1. What is the problem?
2. How serious is the problem?
3. What are the causes of the problem?
4. What are the available solutions?
5. What are the strengths and weaknesses of each solution?
6. How should the best solution(s) be implemented?

But while this model is serviceable, its popularity should not obscure the point that some method of structuring discussion and focusing analysis is better than none at all. Any decision-making agenda which forces group members to distinguish between information and evaluation, which forces a systematic comparison of decision choices, and which includes a reconsideration of initial decisions is likely to yield better decisions than a process with no structure.


Vigilance. A structured discussion is likely to be "vigilant," in that group members who are acutely conscious of the point at issue will be able to address that point directly and to notice remarks that are irrelevant. Vigilance is in effect rational conflict management. It assumes that group decisions are the products of competing pools of information and arguments, and that superior decisions result from adequate testing of the different points of view. Discussion and debate among group members are the tools available for that testing.

Vigilant decision making was the focus of two experiments in the early 1970s by Hall and Watson and by Nemiroff and King. The authors used a common instrument, which asks group members to rank 15 survival items, and compared the results achieved by "uninstructed" groups and by "instructed" groups. An uninstructed group was simply given the ranking instrument and was told to arrive at a group ranking of the 15 items. An instructed group was to make decisions based on the strength of the best arguments. Specifically, instructed groups were advised (1) to support their preferences with relevant reasons, (2) to share evidence which supports preferences, (3) to avoid non-rational decision strategies such as majority vote, averaging, and trading, and (4) to encourage disagreement and debate. Among the research questions asked by the two studies was one concerning the role of vigilance in discussion: will an "instructed" group, which presumably is thinking critically about its decisions, perform better than an "uninstructed" group which works without advice as to procedure? The studies arrived at these results:



The numbers in the column labeled "Individual Decisions" are mean averages for the scores of group members acting alone, without any exchanges among group members. The scores measure "errors," so that a lower score is better. The slight differences among the four entries in the "Individual Decisions" column simply reflect the accidental composition of the groups; if one or two more knowledgeable individuals are assigned to a given group, their scores and thus the average group score will be lower.

Three conclusions can be drawn from the results of these experiments. First, on a test of this kind--pooling information to arrive at a ranking of the usefulness of survival items--group results are superior to average individual results obtained before discussion. Second, turning to the matter of vigilance, groups which presumably look carefully at the basis for their decisions will outperform groups which do not. And third, reinforcing the point about vigilance, the "instructed" groups will achieve "synergy" more often than will "uninstructed" groups.

Synergy--also called the assembly effect, the bonus effect, or even the principle of non-summativity--is a discussion outcome in which the group result is superior to the result obtained before discussion by the best-informed individual in the group; in effect, synergy means that the group has utilized what the best-informed group member could contribute, but has gone beyond that member's contribution by incorporating some of the best ideas of other group members as well. A group whose scores before discussion are 40, 44, 48, 52, and 56 will have an average Individual Decisions score of 48; if that group's Group Decision score is lower than 40, then the group achieved "synergy." Hall-Watson and Nemiroff-King demonstrate that groups engaging in critical thinking achieve that superior group outcome more often than do "uninstructed" groups.

Vigilance can be contrasted with "fantasy chaining," a concept formulated by Robert Bales and used extensively by Bormann to discuss group dynamics. Groups develop a shared culture, consisting of symbols, values, roles, and attitudes toward subject matter. Part of the shared culture, often, is a sequence of narratives, initiated by one group member and then continued, or "chained out," by the others. These narratives, or "fantasies," foster a sense of group identity and promote a shared understanding of the subjects of the stories.

In other words, fantasy chains have both a social and a task aspect. But a principal theme of this essay is that a tension exists between the task and social dimensions of group activity, such that interests of one kind seem at times incompatible with interests of the other kind. In a study of the inner circle at the White House during the Watergate scandal, Laurinda Porter discovered from the transcripts of Oval Office tapes that Nixon and his advisors constructed narratives concerning image control, public relations, communication management, and media personalities. It is the case that almost all groups experience this "symbolic convergence" (Bormann), but when the fantasy becomes more important than reality, group decision making suffers. As Porter concluded, "The press repeatedly attempted to discover the reality of Nixon's deeds, not the illusion or the appearance he presented. The press reality differed from the Nixon group's reality, which consisted of their beliefs and fantasies about themselves and their position."

In 1987, Hirokawa studied student groups as they completed a survival-items decision-making task. These groups were "informed," in that they had available to them the information necessary to make a high-quality decision. But some groups performed better than others. Groups that had access to information but nonetheless performed poorly, Hirokawa found, often engaged in fantasy-chaining; groups that arrived at superior decisions avoided fantasy chains and were vigilant in basing their decisions on appropriate evidence.


Participation. Vigilance, or the practice of making decisions according to the strength of the best arguments, is one characteristic of "ideal" groups in the theory of group discussion developed by Habermas from "ordinary language" philosophy. Another characteristic of such groups is active participation from all group members, and an equal opportunity to participate, such that turn-taking will not be influenced by age, or status, or argumentativeness, or some other personal trait. Habermas' theory is consistent with common-sense notions about the basic rationale for group as opposed to individual decision making: as we saw in the Hall-Watson and Nemiroff-King experiments, groups can outperform individuals when a discussion is managed so as to pool thoughtfully the information each individual can contribute to the decision. For the procedure to work, individuals must contribute, and the decisions must reflect what is contributed, not the source of the contribution. Habermas calls for equal opportunity to participate, not equal participation, but it is fair to ask whether even that is a realistic goal; Habermas himself calls his preferred group an "ideal speech situation," and one might expect that high-status members, for example, will always exert a disproportionate influence on decisions.

One spectacularly productive group--as judged in terms of the needs of the time--did include members of radically different power and prestige, who worked in a climate of freedom to contribute. This was the group of physicists, chemists, mathematicians, and engineers who designed the first atomic bomb at Los Alamos during World WarÊII. The group was led by J. Robert Oppenheimer, who by all accounts was a superb manager of people and their decision making (Stern, Kunetka). Oppenheimer was able to generate an atmosphere resembling a college classroom, in which even a lowly laboratory technician felt free to challenge the ideas of a Nobel Prize winner such as Enrico Fermi or Neils Bohr, or those of the somewhat daunting Edward Teller. There had been some thought given to making Los Alamos a military installation; but that idea was opposed by physicists I. I. Rabi and Robert Bacher on the ground that the hierarchical and authoritarian character of the military would inhibit free communication. Oppenheimer's vision, which he implemented at Los Alamos in a way which changed history, was of a non-hierarchical, egalitarian flow of ideas. As he put it after the war, in a speech on December 11, 1948, in Rochester, N.Y.,

... an indispensable, perhaps in some ways the indispensable, element in giving meaning to the dignity of man, and in making possible the taking of decision on the basis of honest conviction, is the openness of men's minds, and the openness of whatever media there are for communion between men, free of restraint, free of repression, and free even of that most pervasive of all restraints, that of status and hierarchy. (Oppenheimer 51)

Special Techniques. To conclude this discussion of the "management" of a discussion, we will note briefly that several special techniques are available to promote the goals of structure, vigilance, and participation. Brainstorming is a structuring procedure, which encourages the group to distinguish between the informative task of idea generation and the evaluative task of appraisal and decision making. In brainstorming, the quantity rather than the quality of ideas is emphasized, and group members are asked to avoid making evaluative or even explanatory comments. To employ brainstorming involves the group in a structured sequence of questions: what can be discussed, according to what standards, with what resulting judgments?

Outside evaluators and devil's advocates can test ideas and prevent the group from falling into an easy consensus. President Kennedy adopted both procedures for his planning groups during the Cuban Missile Crisis, having learned during the Bay of Pigs disaster the year before what can happen when group members do not raise and vigorously debate relevant issues (Janis).

And the buzz group is a procedure which encourages participation, by dividing a large group into smaller groups of around five members each, with each small group assigned to discuss some question and then report its conclusions to the larger body. The intent is to promote more involvement on the part of each individual than would be possible in a larger "meeting" format, to develop close working relationships among groups of participants, and to stimulate active rather than passive learning among the group members.

The division of "content of discussion" into the number and the persuasiveness of arguments is based on the line of research discussed earlier as "group polarization" studies. The general theory of the Persuasive Arguments explanation of polarization is that choice shifts in a small group occur because group members hear more arguments favoring one position than favoring another, and that those arguments are perceived to be persuasive.

Number of Arguments. A first consideration in examining the content of a group discussion is the relative number of arguments involved. In the simple alternatives posed by the group polarization studies, where group members are offered a risky choice and a cautious choice, one would expect that a preponderance of arguments favoring risk would lead to more risky decisions, while a preponderance of arguments favoring caution would lead to more cautious decisions.

This expectation was tested by Ebbesen and Bowers. Their experimental subjects listened to tape-recorded discussions in which the proportion of risky to cautious arguments was systematically varied. And then the change of opinions of the subjects, from the pretest measure to the measure following their listening to the tape, was calculated. The figure below shows the results Ebbesen and Bowers obtained.



As you can see, when subjects heard a discussion in which 90% of the arguments favored risk, there was considerable shift in opinion toward the risky choice. Where the arguments favoring risk and those favoring caution were evenly balanced (0.50), there was no shift in the opinions of the experimental subjects. And when subjects heard a discussion in which 90% of the arguments favored caution, there was a shift of opinion toward the cautious choice. As Ebbesen and Bowers state, these results offer "convincing support for the conclusion that shifts in risk level following group discussions are largely the result of the group members observing each other generate a specific proportion of risky to conservative arguments" (324).

Most polarization studies examine those conditions in which group discussion will tend to exaggerate pre-discussion opinions of the group members. But the idea just mentioned, that choice shifts result from the number of arguments favoring an alternative, leads to the possibility that discussion may in some cases have a depolarizing effect. That is, if in a discussion group there are two opposed factions, each of which presents a volume of arguments, then the result might be that each faction would move toward a more centrist position rather than toward one of the decision poles. Vinokur and Burnstein studied this possibility. They created six-member discussion groups consisting of three members who favored risk and three members who favored caution. Depolarization occured if the difference or "gap" between the two subgroups was smaller following discussion than it was before discussion. As the table below indicates,



the mean gap--measuring the absolute difference between the subgroups--did narrow substantially following discussion. As Vinokur and Burnstein summarize their results, "The critical finding in this experiment is that both subgroups shift markedly, and the shift is toward rather than away from each other" (878).

Persuasiveness of Arguments. In addition to an effect from the number of arguments presented, shifts in group opinion will result from the persuasiveness of the arguments advanced during discussion. As developed in the literature on polarization effects, persuasiveness is not a particularly rich concept. It has only two values, novelty and perceived validity. Group members are more likely to shift their opinions in the direction of arguments which they had not previously considered, and which they consider to be cogent. As an example, Madsen compared the decision shifts of Iowa college students following discussion of a salient topic (sex education for Iowa students) with the results of discussion of a less important topic (sex education for Florida students). The somewhat paradoxical result is that discussion of the more important topic produced less opinion shift; Madsen's explanation is that group members in the Iowa condition heard fewer "new" arguments than did students in the Florida condition.

We observed earlier, in presenting the model of influences and decision processes which outlines this essay, that the relationship between influences and processes is actional and not causal. People choose to adopt an opinion or to engage in a behavior; those opinions and behaviors are not mechanically compelled. To the extent, then, that persuasion occurs as a result of group discussion, it is self-persuasion; group members hear or generate themselves reasons for an opinion or an action which they choose to regard as persuasive. In Madsen's experiment, half of the experimental subjects were in a non-discussion condition--they generated arguments privately, but did not participate in group discussion about the issues. Madsen found that these experimental conditions produced "relatively large 'residual' shifts" in opinion, which can only be the result of dialogues the experimental subjects had with themselves. As Madsen notes, other studies (e.g., Burnstein and Vinokur; Tesser and Conlee) have also found that private consideration of arguments on an issue can produce "discussionlike shifts" (1124).

We should also note here a result of Madsen's experiments which will anticipate our discussion of "normative influences." In some cases, the shift in opinion that resulted from group discussion did not match the predictions, and Madsen explains the discrepancy by observing that pre-discussion familiarity with an argument does not guarantee that that argument will be used during discussion. The condition of private expression of arguments is simply different from the condition of public expression of arguments. As Madsen says, "... groups tend to pursue a few limited lines of argumentation in discussion rather than to survey the whole range of arguments available.... [I]t is probably easier to generate a large variety of ideas in private (i.e., under the argument-listing procedure) than it is during the give and take of group discussion" (1125).

Finally, the persuasiveness of arguments presented in a discussion is determined by the perceived validity of the arguments. This concept has received a full treatment in the literature on rhetorical theory and argumentation; textbooks teach a set of stock considerations, such as whether the argument is supported by credible (that is, competent and unbiased) evidence, by recent evidence, whether the argument is relevant to the issue being decided, whether the argument is consistent with other known evidence, and so forth. Little of this tradition for judging the validity of arguments has found its way into the literature which explains group polarization effects as the result of persuasive arguments.

To take a single example of the study of the merits of an argument, Borgida and Nisbett compared the effect of college course evaluations consisting of statistical summaries with evaluations consisting of live face-to-face comments from previous students. The authors arrived at the conclusion--which will not startle students of rhetorical theory--that concrete evidence is more persuasive than abstract evidence. More interesting than that finding are the authors' explanations. Concrete information may be more powerful because it has a sensory and therefore memorable quality that abstract information lacks; because concrete information provides a more direct access to the "scripts" by which we interpret the world (that is, we more easily link eyewitness testimony about courses to familiar "solid course" and "boring wipeout" judgments of courses); and because people generally lack the competence to judge the reliability of statistical evidence.

Asch compared the responses of group members who were confronted by unanimous group opposition with the responses of group members who had an ally or a partner. In some cases the partner was a confederate, who had been instructed to give correct answers. In other cases the partner was a second naive subject, who would also be weighing his own sense perceptions against the judgments of the group. Asch discovered that when the group was otherwise unanimous, 32% of the answers by naive subjects were incorrect--that is, represented a yielding to the group position despite the evidence of the senses. It is important not to exaggerate the occurrence of yielding or conformity; in 68% of the critical trials the answers were correct, despite the opposition of the group, and furthermore two-thirds of the naive subjects resisted group influence on over half of the critical trials, one-fourth of them giving consistently correct answers on all critical trials. But even the 32% conformity effect could be reduced substantially by the presence of a partner. When the partner gave consistently correct answers--that is, answers matching the evidence presented by the senses of the naive subject, and contradicting the judgment of the other members of the group--conformity with the majority dropped to 5.5%.

Asch demonstrated, then, that an otherwise unanimous group can exert considerable pressure on a dissenter to adopt the position of the group, and that the pressure felt by a naive subject to conform to the group is almost eliminated if an ally is present. But Asch's experiment did not concern anything of substance, and it did not involve discussion among group members. One might wonder, therefore, whether the influence of unanimity would appear also in the case of a typical group problem-solving or decision-making task. Among the sources of evidence proving the influence of unanimity in the group communication setting is the literature on how juries make decisions. We will look briefly at some experimental evidence and at some historical evidence concerning jury decision making.

A good example of the experimental study of jury decision making is a journal article published in 1986. Tanford and Penrod were interested in answering these questions:

• How closely do final verdicts resemble the initial votes of the juries? We would expect, based on the Asch experiment, that a strong majority position among jurors would win "converts," so that the initial votes of the juries would predict accurately the final outcome of decision making.
  
  
• Are decisions of jurors influenced more by the weight of the evidence, or by factors unrelated to the substance of the trial? Those extraneous factors include the pressure on a minority member exerted by an otherwise unanimous jury.

To answer these questions, Tanford and Penrod designed an experiment in which six-person juries saw a videotape of a realistic trial reenactment. Jurors voted individually immediately after the trial, and then deliberated until a group verdict was reached. Their deliberations were videotaped and were studied to determine what considerations were most important in their decision making. The following table shows the relationship between initial votes and jury decisions for the first and most important of the three charges being tried in the reenactment.



As you can see, in none of those juries in which there was initially a lone dissenter--a single vote for conviction in 22 cases, a single vote for acquittal in 4 cases--did the dissenter's position become the jury's verdict. Furthermore, excluding the 21 cases in which the jury was at first evenly divided, the initial majority position held in 78% of the decisions. In a few cases (7.6%) an initial minority position did become the jury verdict, but that result always required a minority of more than one person.

But one might still wonder, as we did with the Asch experiment, whether behavior in the psychologist's laboratory resembles that of real juries. As it turns out, there is considerable evidence about how juries actually reach decisions. Some of the most interesting evidence comes from the Chicago Jury Project, since it included tape recordings of actual jury deliberations, as well as interviews with jurors and other methods of collecting information (Kalven and Zeisel). Juries studied for the Chicago Jury Project did differ in some ways from the experimental juries reported by Tanford and Penrod; for example, in 225 actual trials there was a unanimous jury on the first ballot in over 30% of the cases as compared with 9% of the experimental juries, and about 5% of the actual juries "hung" as compared with 18% of the experimental juries (487). But the historical evidence is strikingly similar to the experimental evidence on the present point, concerning the influence of majority, even unanimous, opposition to a dissenter. The Chicago Jury Project found that an initial minority prevailed in only 5% of the cases--and then only if the minority began with at least three members of the jury. As Kalven and Zeisel say,

Nevertheless, for one or two jurors to hold out to the end, it would appear necessary that they had companionship at the beginning of the deliberations. The juror psychology recalls a famous series of experiments by the psychologist Asch and others which showed that in an ambiguous situation a member of a group will doubt and finally disbelieve his own correct observation if all other members of the group claim that he must have been mistaken. To maintain his original position, not only before others but even before himself, it is necessary for him to have at least one ally. (463)

These results do not mean that an articulate and determined juror--or a member of some other decision-making group--can never persuade enough group members to make her position the majority one. The dramatic example of the Henry Fonda character in the film Twelve Angry Men, in which one man's intelligent queries turn an 11-1 majority favoring conviction to a 12-0 jury acquitting the defendant, is almost entirely absent from the historical record. A rare exception is the Watergate trial of John Mitchell and Maurice Stans, in which a jury which had begun its deliberations divided 11-1 for conviction was persuaded to return a verdict of acquittal. An irony in this case is that the persuasive juror was at first an alternate--one Andrew Choa, a vice president at the First National City Bank in New York--who during the trial took the place of a juror who had become sick (Zeisel and Diamond).

But far more typical are the experiences of two recently-concluded trials. The jury in the case of former Panama leader Manuel Noriega was reportedly divided 11-1 for conviction, and told the trial judge they thought they were deadlocked; but upon returning to the jury room the lone dissenter finally switched positions, and on April 9, 1992, the jury returned a unanimous verdict of guilty. And see the account of the social pressure felt by the lone juror favoring conviction in the trial of the Los Angeles police officers accused of assaulting Rodney King.

Asch included a "control group" in his experiments, in which the naive subjects were permitted to give private, written answers about the matching lines, rather than oral answers that could be observed by the group. The results were as dramatic as the introduction of a partner who also gave correct answers. When the group could not observe the answers given by a naive subject, conformity to the group's wrong answers dropped from 68% to 7.4% of the critical trials; the mean average for the occurrence of errors per subject dropped from 2.3 to 0.5.

Asch demonstrated, then, that group members are more likely to conform to the position of the majority if the majority group members can observe their behavior. How does that finding apply to the work of a problem-solving or a decision-making group?

There have been several proposals for decision-making methods which would shield group members from surveillance. One of these is the "nominal" group--a group "in name only," in that there is no oral interaction among group members and contributions by group members are anonymous. In one implementation of the nominal group procedure, a moderator would pose a problem to isolated group members, would collect from them written solutions, and would post--or share with the other group members--the initial answers. Group members would see what others had said (though would not be told the source of the statements) and would then be asked in a second round of comments to revise their initial statements and respond to the statements of others. These second-round answers would also be collected and posted. Then group members would be given a third round to state positions and respond to others. And so forth, until a consensus was reached or until the moderator sensed that the issues had been fully explored.

The nominal group procedure is intended to break the connection between normative influences and the lone decision maker, since all of those influences--unanimity, surveillance, size, and prestige--lose their force in the absence of oral, face-to-face interaction. If I do not know that an idea comes from a college dean, then I will judge the idea on its merits, rather than being influenced by the prestige of the source. If I know that the "group" cannot observe me or connect me with my statements, then I will act according to my convictions, rather than being influenced by my relationship with the group.

One interesting implementation of the nominal group technique involves computers. A procedure developed at the University of Arizona in the mid-1980s, which has now been adopted by IBM as "TeamFocus," places individuals at widely-separated computer terminals, under the guidance of a facilitator and software which controls steps such as brainstorming, evaluation of alternatives, and voting (Nunamaker). Contributions typed by individuals at their terminals appear anonymously on a large screen visible to all. A recent review of TeamFocus in an IBM promotional publication makes these claims for it:

TeamFocus offers educators numerous advantages over the status quo. Because comments are entered online, participants are able to "talk" at the same time, greatly speeding the process. The TeamFocus process democratizes meetings by ensuring that senior faculty or administrative staff don't dominate. Thanks to its software and workstation design, TeamFocus provides complete anonymity for all participants. This encourages the free flow of ideas--especially from less assertive staff members who might others remain silent. ("Groupware" 1992)

That, at least, is the theory. How well does it work?

The nominal group technique generally, and the TeamFocus implementation in particular, are appealing. Everyone has had unhappy group experiences, resulting from domination, conflict, apprehension, and other causes. The thought that a decision-making technique might let people pool their information and test ideas, while avoiding the disadvantages that seem to come with oral, face-to-face communication, is intriguing. But the crucial question is not whether a nominal group works, but whether it works better than an interacting group. If the goal is not just a comfortable procedure but a successful one, then we should ask which method--the anonymous, nominal group technique or the traditional method of oral interaction--is more productive.

That question was asked in 1984 by Burleson and his colleagues. In an experiment designed to answer the question, Burleson used a common instrument, which asks group members to rank 15 survival items, and compared the results achieved by nominal groups and by interacting groups, with these results:



You will recall from our discussion earlier of the Hall-Watson and Nemiroff-King studies that the numbers in the column labeled "Individual Decisions" are mean averages for the scores of group members acting alone, without any exchanges among group members. The scores measure "errors," so that a lower score is better. The nominal groups, which did not use oral interaction to reach a group decision, did improve slightly over the average individual decisions of the group members. But the interacting groups realized a much greater improvement.

The Burleson experiment, then, offers some reason to be sceptical of a computer-mediated decision-making procedure. Despite the disadvantages some people have found with oral interaction, the immediacy and spontaneity of direct contact may yield superior decisions. A procedure without oral, face-to-face interaction may eliminate more than surveillance. Consider, as a related matter, the difference between reading a legal deposition and watching a witness testify under cross-examination. The sincerity and credibility of the witness can be judged better if the testimony is accompanied by the sights and sounds of posture, eye contact, and vocal intonation.

But from the point of view of the individual group member, the important lesson to be learned from the Asch experiment and from innovations such as TeamFocus is that candor is valuable. The pressure to conform to the position of the group is stronger if the group can observe our actions, and so it is important that the individual expects and understands that pressure, and at the same time recognizes that the interests of the group will be served best if the eventual decision is based not on pressures to conform but on the weight of the evidence.

The first experiments by Asch pitted confederate majorities of seven against lone naive subjects, with the result that the lone individual conformed to the position of the group in nearly one-third of the critical trials. Asch wanted to discover how the pressure to conform would vary if the size of the confederate majority were reduced to one, two, three, or four, or were increased to sixteen. The table below reports the results Asch obtained; recall that in the case of the Control group, the naive subject was shielded from group influence by giving private, written responses.



If someone were matching the lines solely on the evidence of the senses, without any disturbing testimony from others, we would expect the mean number of errors to approach 0. In the presence of a group which answers incorrectly, with the naive subject protected from group pressure by giving private, written answers, the mean number of errors increases but only slightly, to a 0.7% error rate. When the naive subject must answer while being observed by a confederate who has answered wrongly, the number of errors jumps considerably--though it is still low, an error rate of 2.8%. The addition of a second confederate creates another jump in the rate of conformity, to 12.8%, and the addition of a third confederate still another jump, to 33.3%. At that point, the rate of conformity appears to remain almost level.

One can conclude from Asch's data, then, that the pressure to conform varies, up to a point, with the size of the group.

As a final social comparison influence, we will consider the personal relationship of the individual to the other group members. Among the factors which would lead an individual to defer to the judgment of others are a personal relationship which would be threatened by disagreement, and a halo effect which would lend credibility to the judgments of the others. It is easy to imagine variants of the Asch experiment in which the naive subjects (a) would either have a cohesive relationship with the confederate panelists (perhaps all would be members of the same fraternity), or would be matched with strangers; and (b) would either be told that the other panelists were college seniors or had been selected for their visual acuity, or that they were college freshmen or might see the panelists wearing eye glasses. What would be the effect on the judgments or behavior of the naive subjects? There are good reasons to believe that attitude toward group members will influence judgments and actions.

Among the factors which can deflect attention from the substance of a group's task is the prestige of the source of comments made during discussion. We have known since Aristotle that the credibility or "ethos" of an individual will affect our judgments of the truth or importance of her remarks. In the small group, judgments of credibility might be based on such factors as age, ethnic background, religious belief, political persuasion, physical attractiveness, personality, and so forth; none of these factors has any necessary relationship to the wisdom of the contributions of a group member. The factors which we will consider briefly in this examination of credibility are gender, the attribution to a group member of expertise, and nonverbal behavior.

The first of the studies that we will examine dealing with credibility and small group communication, by Brown and Geis, used the same "Lost on the Moon" exercise that Burleson had used in comparing nominal and interacting groups. In this case, however, several versions of a videotape were prepared in which actors made decisions about the survival items; the experimental subjects were students who viewed the tapes and made judgments about the actors. The videotapes varied in part according to the prestige of the actors. In half of the videotapes the leader was "Chris," a male, and in the other half the leader was "Chris," a female. And in half of the videotapes the tape was preceded by a message from the group's "professor" praising Chris as an able leader, whereas in the other videotapes the professor neither commended nor denigrated Chris; these were the "legitimated" and "unlegitimated" conditions of the experiment. In each videotape, however, Chris read the same script; therefore, there was no substantive difference between the contributions of the leaders. Subjects viewing the tapes were asked to judge the effectiveness of the leader and of the other group members--two males and two females--according to such measures as amount of leadership, quality of contributions, desirability for hiring, salary, promotion, and the like. Additionally, the leader was evaluated separately according to strengths and weaknesses:



The figure below shows the judgments of the leaders and of the other group members:



It is clear that the leader was judged to be better if introduced as able by an authority figure. There was no difference in the rankings of male and female leaders. But male members were ranked higher than female members on the leadership scale in both the legitimated and unlegitimated conditions. Furthermore, when asked to identify the weaknesses of the leader, subjects "found" in the same performances different qualities depending on the gender of the leader. The "weaknesses" for male and female leaders seem to reflect stereotypes: since males are stereotyped as "strong," a poor male leader was judged as "too weak," and since females are stereotyped as "sensitive to others," a poor female leader was judged as "Too cold and insensitive" and as "Too dominating." Recall that exactly the same performance is being judged.

The second study dealing with credibility and the small group, by Natalie Porter and others, asked experimental subjects to make judgments of the leadership, dominance, talkativeness, intelligence, and warmth of individuals known only from a photograph of them sitting at a conference table. In two typical photographs, those seated were either all males or all females



and experimental subjects, based on this meager evidence, identified the person at the head of the table as the "leader" of the group, or as the person "contributing most to [the] group," about half of the time. The rule they were employing in making that judgment was apparently Leader = person at the head of the table. The same result occurred with the following photograph,



in which a mixed-sex group was shown with a male at the head of the table.

But something very different happened with the next photograph:



Shown a mixed-sex group with a female at the head of the table, the attribution of leadership to the person at the head of the table dropped from 50% to 18%. Subjects who made such judgments--and there was no difference among viewers who were female as opposed to male, or among viewers who were sympathetic or unsympathetic to the goals of the women's movement--apparently employed a new rule in determining leadership: In a mixed-sex group, Leader = one of the males. Where this rule conflicts with the earlier one about position at the table, as it does in the case of the fourth photograph, the second rule is more powerful. We should note in passing that subjects who viewed the second photograph judged the female at the head of the table to be less "warm" than her colleagues, as though one counter-stereotypical quality, leadership, needed to be explained by a second counter-stereotypical quality, absence of warmth.

The important consideration for our purposes is the bearing of these findings on group decision making. There is obviously no necessary relationship between the quality of an individual's contribution to the group and either that individual's seating position or gender. But our perceptions of leadership and competence seem to be affected by such considerations--and by attributions of expertise coming from some outside authority, as Brown and Geis demonstrated--and those perceptions will in turn affect our responses to the ideas of group members, and eventually the decision of the group.

"Cohesiveness" is closely related to prestige, in that both are measures of the attraction toward the group felt by group members. The principal study of the impact of cohesiveness on decision making is the work of Irving Janis on "groupthink." Janis' central conception is that high cohesiveness will tend to suppress critical thinking, since group members who place a high value on their relationships with colleagues will not be willing to risk those relationships by challenging dubious ideas during group discussions. Of course, others have noted--and Janis has subsequently agreed--that cohesiveness is not the cause of groupthink; to the contrary, cohesiveness is a prerequisite for critical thinking in the group (Longley and Pruitt). The conflict required to test and refine ideas involves risk, and only groups that have developed trust among group members will assume that risk. To avoid groupthink, a group must have both high cohesiveness and a work norm encouraging debate. In the absence of that work norm, a highly cohesive group may avoid disagreement, reach an easy and premature consensus, and possibly arrive at a faulty decision.

Janis' eight "symptoms" of groupthink have become standard terms for discussing the phenomenon. There are two ways by which group members can overestimate the group: (1) they can adopt an illusion of invulnerability, as the group of military planners responsible for the defense of Pearl Harbor did in 1941, and (2) they can convince themselves that their motives are more noble than those of their competitors. There are two forms of group closedmindedness: (3) group members can rationalize their decisions, assigning good reasons rather than the real reasons for their actions, and (4) members can adopt shared stereotypes of competitors or regulators. And finally, there are four pressures toward group uniformity: (5) group members can suppress their own doubts, or engage in self-censorship; (6) group members can act as mindguards, protecting the group from information which would disturb the group consensus; (7) group members who do speak against the group consensus can receive direct pressure to conform; and (8) group members who therefore hear little or no dissent may develop an illusion of unanimity, the mistaken belief that everyone in the group is in agreement.

Less familiar than the eight symptoms of groupthink, but probably more useful in concrete efforts to avoid groupthink, are Janis' "symptoms of defective decision making." Groups which experience groupthink are likely to make some of the following mistakes in arriving at a decision:

1. Incomplete survey of alternatives
2. Incomplete survey of objectives
3. Failure to examine risks of preferred choice
4. Failure to reappraise initially rejected alternatives
5. Poor information search
6. Selective bias in processing information at hand
7. Failure to work out contingency plans

Examples of cohesive groups which experienced groupthink are, according to Janis, President Kennedy's advisors who planned the Bay of Pigs invasion, President Truman's advisors who failed to anticipate China's entry into the Korean War, the military leaders who were responsible for the defense of Pearl Harbor in 1941, President Johnson's advisors who managed the escalation of the Vietnam War, and President Nixon and his advisors who became entangled in the Watergate crimes. On the other hand, President Truman's advisors who developed the Marshall Plan for the reconstruction of Europe after World War II, and President Kennedy's advisors who responded to the Cuban Missile Crisis, avoided groupthink. As more recent examples, there is evidence that groupthink was involved in the fatal decision to launch the shuttle Challenger (Cline and Lane; Renz and Greg). And you should be able to find evidence, in the account of the launching of the Hubble space telescope, that the engineers who built the telescope and the scientists and administrators who decided to launch it into orbit failed to examine the risks of their decision, performed a poor information search, and evaluated the available evidence in a biased manner.

Informational Conformity. You may recall from the discussion of the Asch experiments on matching lines with which we began this essay that some of the subjects who reported matches different from their initial sense perceptions were making task judgments. A small number of naive subjects experienced a change in their perceptions--they came to see the lines as matching the way the majority said they matched; and others experienced a change in their judgments--they came to believe that their perceptions were mistaken, that the majority must be correct.

Another term for the same phenomenon, in which someone adopts another's position based on task considerations, is "internalization." The new position can be accepted because it is believed to be true; it can be "internalized" because it is consistent with beliefs and values already held. An example from popular culture of the phenomenon of informational conformity or internalization is the stockbroker character, played by E. G. Marshall, in the film Twelve Angry Men. Marshall argued for conviction of the defendant until near the end of the film; he finally changed his mind, and adopted the position of the new group majority favoring acquittal, when he became convinced that the testimony of the eyewitness to the crime was not credible.

Normative Conformity. On the other hand, sometimes the position of the group is adopted not because it is believed to be true--a task consideration--but in response to the people involved, in order to gain from them some desired reaction. Recall that in the Asch experiments the third yielding response by naive subjects dealt with interpersonal relations; subjects continued privately to trust their own perceptions and judgments, but they changed their public behavior, and gave incorrect answers, in order to avoid disapproval from the other group members.

Another term for the same phenomenon, in which someone adopts another's position based on social considerations, is "compliance." Compliance is a matter of attitude expression, not attitude change; the compliant individual continues to believe that the group is wrong, but supports the group's position in order to gain some reward. A current example of the phenomenon of normative conformity or compliance is reported in the news article reprinted in the news story concerning juror Virginia Loya in the Rodney King trial. Loya plainly believed that the jury's non-guilty verdict was mistaken on at least some of the charges; she told the reporter that if she could repeat the jury experience she "would have held out longer" for at least one conviction. Why did she agree to a verdict of acquittal on all but one charge, and a compromise position of no verdict on the final charge? As Ms. Loya reported, she had been "mocked" by the other jurors, she felt "enormous pressure," even to the point that "tears were shed."


Several points need to be made about this discussion of conformity responses as consisting of informational conformity (or internalization) and normative conformity (or compliance).

First, and most important, if the life of every group consists of task and social considerations, then no pure case will arise of informational or normative conformity. One is always responding in some way both to what is said and to the people saying it. Therefore, to characterize a decision response as informational conformity is simply to say that the person was more influenced by information than by interpersonal pressures.

An aspect of this co-presence of task and social influences is the inherent ambiguity of the distinction between normative and informational conformity. Take as an example the results of Willis' study (1963), in which conformity in a matching lines experiment was affected by manipulations of credibility. When the naive subject was led to believe that the other person was experienced and competent, conformity with that person's judgments increased significantly. Credibility could be interpreted as prestige, with the naive subject desiring approval from a high-status person. In that case, conformity would be normative. Or credibility could be interpreted as expertise, with the naive subject taking the other's testimony as evidence about reality. In that case, conformity would be informational.

Second, where decision responses can be distinguished they often concern different types of tasks. As an illustration, let us return to the experiment by Tanford and Penrod concerning jury decision making. The results they obtained, in which no lone juror was able to prevail against unanimous opposition, do not explain why the decisions were made. It is possible that the interpersonal pressure of the majority led some jury members to change their votes, but it is also possible that those jurors who switched positions heard persuasive arguments. After all, if in a six-member jury five of the jurors are convinced that the defendant is innocent (or guilty), it is likely that they heard convincing evidence and arguments during the trial--and so would repeat those persuasive arguments during the jury deliberations.

To discover how the juries arrived at their decisions, Tanford and Penrod studied the videotapes of the jury deliberations. Those videotapes revealed that jury decision making was of two kinds, depending on the seriousness of the charge being discussed and depending on the stage of the deliberation. There were three charges against the defendant; the table reprinted above shows the vote patterns for the first of those three charges, which was also the most serious of the three. Tanford and Penrod found that informational conformity, or internalization, characterized the debates on the first and most important charge; but juries were influenced by social comparison processes on the later and less important charges--decision making on those charges was characterized by normative conformity, or compliance. As the authors explain,

The present research provides evidence for two qualitatively different judgment processes within a jury trial. The first type of decision is one that is based on the facts of the case and affects individual cognitive processes as well as verdicts. This type of process operates on the first or strongest in a series of charges; it is a decision based largely on informational influence processes. The second type of decision operates on weaker or later charges, and is affected primarily by the group vote distribution as well as decisions on other charges. It is also the type of decision in which jurors change positions as a result of majority pressure, but do not change individual cognitions. This type of decision involves normative influence processes. (344)

And third, a compliant response can become an internalization response. That is, a group member can adopt the position of a group out of concern for relationships with other group members, but then in order to justify to herself having adopted an untrue position will come to believe that the position is valid. There is a large body of research in the "cognitive dissonance" literature dealing with the phenomenon of "forced compliance," or counterattitudinal advocacy, in which in some cases people who are induced to express a belief that is contrary to their true convictions will come to "internalize" that belief. There is broad agreement that the moderation of racial attitudes in the South following the adoption of civil rights laws in the mid-Sixties followed this pattern. Initially, white Southerners complied with the laws in order to avoid legal punishments (compliance), but they then changed their attitudes and beliefs (internalization) so as to justify and incorporate the new behaviors.


Identification. The terms "internalization" and "compliance" come from some well-known writings by Kelman, who says that attitudes can be changed, or at least expressed, for different reasons. In addition to internalization and compliance as grounds for forming or stating attitudes, Kelman says that some attitudes are based on "identification." Identification is a form of self-persuasion, in which an individual models the behavior of someone she finds "attractive." We will not discuss identification here, except to observe its similarities with the other two terms. Like compliance, the influence leading to identification is personal; as Kelman says, the goal in identification is "to establish or maintain a satisfying self-defining relationship to another person or group." And like internalization, identification is a process which does not require surveillance (as compliance does); in both cases the motivation to believe or act is internal--the belief or action is consistent with other beliefs and values.

Parallel to the two classes of conformity responses are two classes of resistance responses. From Tarde to Willis (1965), resistance to group influence has been described as "independent" behavior and "anti-conformist" behavior.

Independence. The Henry Fonda character in Twelve Angry Men acted "independently" in resisting the other eleven jurors who on the first ballot favored conviction. His position was grounded in the task of the group, and in his own beliefs and values. By structuring the jury's discussion around the various items of evidence, by insisting on a vigilant examination of each piece of evidence, and by making the other jurors active participants in the discussion Fonda persuaded them to change their judgments, so that finally he represented rather than resisted the group.

Independence has this in common with internalization, that both are based on acceptance of arguments, irrespective of the reactions of other individuals. In the case of internalization, that consideration of arguments happens to lead to a position consistent with what others believe; in the case of independence, that consideration happens to lead to a position different from what others believe.

Anti-conformity. Behavior can sometimes seem to defy group norms, while still being controlled by those norms. One popular scheme for classifying roles of group members includes a class of "individual roles," adopted by members who are pursuing personal rather than group goals (Benne and Sheats). Individual roles include the "aggressor" and the "blocker," whose positions will be negative but specifically will negate what the other group members are saying. Such "anti-conformity" gives the surface appearance of independence, but in fact it is specifically the ideas and values of others which determine the position the aggressor or blocker takes.

Anti-conformity has this in common with compliance, that both are based on the reactions of other individuals, irrespective of other values. In the case of compliance, that consideration of the reactions of the group happens to lead to the expression of a position consistent with what others believe; in the case of anti-conformity, that consideration happens to lead to the expression of a position different from what others believe.

Task vs. social dimensions of the group

Decision making in Rodney King trial

Decision making in Hubble telescope launching