An important issue in the study of groups concerns the determinants of poor versus good decision making. One influential theory of group decision making is Janis’ (1972, 1982) groupthink theory. Groupthink theory (Janis, 1982; Janis & Mann, 1977) specifies several factors (antecedents of groupthink) which, in combination, facilitate the poor decision making called groupthink. These antecedents include high group cohesiveness, a stressful situation, and a variety of structural or administrative factors such as insulation of the group, promotional leadership, lack of methodical decision-making procedures, and lack of variety among members’ values and perspectives. In the present study the framework of groupthink theory was used to investigate the influence of two factors on group decision making: (1) promotional leadership, an antecedent included by Janis, and (2) group members’ conformity predisposition, a personality factor not considered by Janis.

The term groupthink refers to a premature striving by group members for unanimous agreement on a course of action. Groupthink is undesirable because it cuts off necessary consideration of the pros and cons of the various decision options. Groupthink is identifiable by eight symptoms which represent different ways that group members attempt to avoid, or reduce, the anxiety associated with decision making and protect their self-esteem (Janis, 1982). The symptoms are of three types. Type I symptoms involve overestimating the group; group members share the illusion that the group is invulnerable and the unquestioned belief that the group’s cause is just. Type II symptoms involve closed-mindedness. Group members collectively rationalize away information which is inconsistent with their preferred position and stereotype enemies as weak or stupid. Type III symptoms involve pressures toward uniformity. Members tend to censor their own misgivings about the group’s position, contributing to a shared illusion that the group is unanimous in its acceptance of the majority position. The group also puts direct pressure on any member who does express a dissenting view. Finally, members may take on the role of mindguards to protect the group from outside information which could threaten the members’ confidence in the correctness of the group’s position.

When most, or all, of the symptoms of groupthink are present, the group’s decision-making process is likely to be seriously flawed. Janis (1972) listed seven symptoms of the defective decision-making process which groupthink can produce: (1) incomplete survey of alternatives, (2) incomplete survey of objectives, (3) failure to examine risks of the preferred choice, (4) failure to reappraise initially rejected alternatives, (5) poor information search, (6) selective bias in processing available information, and (7) failure to develop contingency plans. The end result of this causal chain is a greater probability that a poor quality decision will be made.

Several researchers, in their reviews of the groupthink literature (Aldag & Fuller, 1993; Esser, 1998; Hart, 1991; Neck & Moorhead, 1995; Park, 1990) have attempted to assess the validity of groupthink theory. However, at present, the best general conclusion may be that although empirical support for predictions derived from groupthink theory has been mixed, at best, too little empirical research has been reported to allow a complete, point-by-point assessment of the validity of groupthink theory.

Among Janis’ antecedents of groupthink the leader’s promotional stance has received perhaps the most empirical attention. In the first laboratory test of groupthink, Flowers (1977) trained confederates to lead their groups either by promoting their own preferred solution or by encouraging members to express all possible viewpoints. As predicted by groupthink, it was found that groups whose leaders promoted their personal preferences discussed fewer facts and proposed fewer solutions than groups whose leaders encouraged expression of various viewpoints. However, the Flowers study did not examine the full constellation of dependent measures characterizing the groupthink syndrome (i.e., all eight symptoms of groupthink and all seven symptoms of a defective decision process), nor was the quality of the decision examined.

Two additional laboratory experiments have produced some similar leadership effects in support of groupthink predictions. Leana (1985) trained participants to lead their groups either by promoting an assigned preference or by encouraging discussion of divergent solutions. As predicted, groups with leaders who promoted their assigned solutions proposed and discussed fewer solutions than groups whose leaders encouraged discussion. Similarly, Fodor and Smith (1982) assigned group leaders on the basis of their need for power. They found that groups whose leaders had high power needs (and who, therefore, were likely to behave in ways which promoted their own points of view) discussed fewer facts, proposed fewer alternatives, and tended to express fewer moral concerns than did groups whose leaders had low power needs. As with the Flowers (1977) study, however, these experiments did not involve an examination of most of the measures which identify groupthink.

A final laboratory study (Moorhead & Montanari, 1986) included an attempt to measure the full set of antecedents and consequences of groupthink which Janis (1982) listed. Preestablished teams of business students worked on two decision-making tasks, then completed a questionnaire (Moorhead & Montanari, 1982; Montanari & Moorhead, 1989) designed to measure all antecedents and consequences of groupthink. Factor analyses of the questionnaire reduced the set of measures to three antecedents, four symptoms of groupthink, and two symptoms of a poor decision process. A path analysis revealed that, as predicted by groupthink theory, groups with more promotional leaders were more likely to discourage dissent and adopt an illusion of morality than were groups with less promotional leaders. However, contrary to the groupthink hypothesis, groups with more promotional leaders considered more alternatives than did groups with less promotional leaders.

Unlike most laboratory studies of groupthink, research based on the analysis of historical cases involving important group decisions has typically involved an examination of a larger set of variables (i.e., the symptoms of groupthink and the symptoms of a poor decision process). After reviewing this literature, Esser (1998) concluded that there is support for the groupthink hypothesis that promotional leadership is related to the presence of groupthink symptoms, symptoms of poor decision process, and poor decision quality (cf., McCauley, 1989; Neck & Moorhead, 1992; Tetlock, Peterson, McGuire, Chang, & Feld, 1992).

To prescreen for conformity predisposition the CFA was administered to students in large groups (n = 20-30). The CFA is composed of 20 items, each of which is scored on a 0-5 scale. A higher score was interpreted as indicating a greater predisposition to conform. For the present sample of 459 students the CFA had good internal consistency reliability, Cronbah’s alpha = .89. Based on their CFA scores, students were classified as either high (upper 45 percent) or low (lower 45 percent) in conformity predisposition.

Homogeneous groups of high or low conformers were scheduled for each experimental session. Each group comprised four or five members and groups were randomly assigned to leadership conditions.

When participants arrived for their experimental session, they were asked to sign informed consent forms which included the information that their group would be video-recorded. Then the experimenter selected the participant who had arrived first to be the group leader. The group was asked to wait quietly while the experimenter trained the leader. The experimenter then took the leader to a separate room and gave leadership instructions which were adapted from those used by Flowers (1977).

Nonpromotional leaders were instructed to (1) begin the group decision making with a statement (and later remind the group) that the most important thing was to weigh all viewpoints in order to arrive at the best possible decision; (2) refrain from stating the leader’s own preferred solution until all group members had stated their preferred solutions; and (3) encourage discussion of all possible solutions.

Promotional leaders were instructed to (1) begin the group decision making with a statement (and later remind the group) that the most important thing was for the members to be in total agreement with the group decision; (2) state the leader’s preferred solution immediately after the leader’s opening statement and before any group members had an opportunity to speak; and (3) discourage discussion of possible alternative solutions.

Next, the leader was returned to the group. The group task was assigned, and a copy of the task materials and a stopwatch were given to the leader. The leader was told to use the stopwatch to keep track of how much time the group took to reach its decision. However, the experimenter kept the official time. After the group completed the decision task, the experimenter collected these materials and distributed two questionnaires to all participants. Finally, participants were thanked for their participation, debriefed, and dismissed.

The leadership manipulation was assessed by having each group member complete a 4-item questionnaire about the group leader’s behavior. These data were analyzed by first finding the mean for each group for each question. Next, these scores were analyzed using a 2 (leadership: promotional or nonpromotional) × 2 (conformity predisposition: low or high) analysis of variance (ANOVA) for each question.

The first question assessed the degree to which the leader suggested his/her own solution to the problem. The ANOVA yielded a significant main effect of leadership, F(1,60) = 17.77, p < .001, and no other significant effects. As intended, promotional leaders (M = 3.88) were seen as suggesting their own solutions to the problem to a greater degree than were nonpromotional leaders (M = 3.24).

The remaining three questions assessed the degree to which the leader encouraged discussion of every suggested solution, stressed the importance of weighing all viewpoints to reach a good decision, and stressed that the most important thing was for the group to agree on its decision. ANOVAs for these items revealed no significant effects.

The Groupthink Index yielded a total score and subscale scores corresponding to each of the eight symptoms of groupthink for each participant. Internal reliabilities for the total scale and for the subscales were estimated by calculating coefficient alphas, using data from the 284 participants who completed the Groupthink Index. Next, means for each decision-making group were obtained. These group means were used in 2 (leadership) × 2 (conformity) ANOVAs performed on the total scale and on the eight subscales.

The internal consistency of the total Groupthink Index was acceptable, Cronbach’s alpha = .70. The ANOVA revealed a significant main effect of leadership, F(1, 60) = 4.52, p < .05. As hypothesized, groups with promotional leaders (M = 106.42) gained higher scores on the Groupthink Index than did groups with nonpromotional leaders (M = 102.28). No other significant effects were found.

The internal consistency reliabilities of the subscales ranged from clearly unacceptable (Cronbach’s alphas = .02, .11, and .25 for the belief in the group’s morality, rationalization, and pressuring dissenters symptoms, respectively) to marginal (Cronbach’s alphas = .32, .36, .40, .45, and .46 for mindguarding, self-censorship, stereotyping, unanimity, and invulnerability respectively). ANOVAs performed on the subscales revealed two significant leadership main effects. As hypothesized, members of groups with promotional leaders reported more self-censorship (M = 13.27) than did members of groups with nonpromotional leaders (M = 12.15), F(1,60) = 11.82, p < .01. Also as hypothesized, members of groups with promotional leaders reported more mindguarding (M = 12.38) than did members of groups with nonpromotional leaders, (M = 11.62), F(1,60) = 4.81, p < .05.

Because the internal consistency reliabilities of the individual groupthink symptom subscales were low, the entire 40-item Groupthink Index was subjected to a principal components analysis (n = 284). This analysis with a varimax rotation yielded three interpretable factors, which accounted for 28.4% of the variance. The first factor accounted for 15.4% of the variance and was composed of six items with loadings ranging from .49 to .61. This factor involved an objecttive attitude, avoiding a closed-minded, self-righteous orientation. The second factor accounted for an additional 7.7% of the variance. It was composed of four items with loadings ranging from .49 to .69 and focused on a procedure which encourages intellectual conflict and debate. The third factor accounted for 5.2% of the variance and was composed of four items with loadings ranging from .55 to .61. This factor reflected the belief that the majority is correct, and involved pressure on dissenters.

In the present study two hypotheses were investigated: (1) that groups whose leaders promoted their own preferred solutions would exhibit symptoms of groupthink, symptoms of poor decision-making process, and poor quality decisions, and (2) that groups whose members were predisposed to conform would exhibit symptoms of groupthink, symptoms of a poor decision-making process, and poor quality decisions. The first hypothesis received partial support, while the second hypothesis received no support.

The present results dovetail with previous laboratory studies of the effects of promotional leadership on groupthink. The authors found that groups with promotional leaders produced more symptoms of groupthink, discussed fewer facts, and reached a decision more quickly than did groups with nonpromotional leadership. Because it is the first experiment in which the effects of promotional leadership on the full range of groupthink indicators were investigated, the present study represents an attempt to extend understanding of the leadership–groupthink relationship. However, these results should be interpreted cautiously. Only five analyses of the measures of groupthink consequences (symptoms of groupthink, symptoms of a poor decision-making process, and decision quality) yielded statistically significant effects. Perhaps the authors’ manipulation of promotional leadership was not as effective as had been intended. Only one of four manipulation check items indicated that the manipulation was successful.

Another problem in the present study, and in groupthink research generally, concerns the measurement of groupthink indicators. For example, in the present study the Groupthink Index (R. Glaser, 1993) was used to assess the presence of symptoms of groupthink. The estimate of the internal consistency reliability of the overall, 40-item scale was acceptable, but the reliability estimates for the 5-item scales for the individual groupthink symptoms were questionable, at best. A similar concern can be raised regarding the only other groupthink symptom questionnaire which has been reported in the groupthink literature, a 24-item scale developed by Moorhead and Montanari (1982; Montanari & Moorhead, 1989).

The second hypothesis, that groups composed of members who were predisposed to conform would be candidates for groupthink, was based on McCauley’s (1989) interpretation of groupthink as sometimes involving compliance – public conformity to the groupís preferred position, but without private acceptance. The question remains, however, under what conditions does compliance in groupthink occur? McCauley (1989, p. 258) offered only the “tantalizing hint” that compliance may be most likely when the groupthink antecedent of group homogeneity is absent. The authors believe that, in the present experiment, groups were homogeneous; all participants were students enrolled in introductory psychology classes at a state-supported, regional university. Most grew up and attended high schools within a 60 mile radius of the university campus. Hence, their values and perspectives could be expected to be similar. Thus, they may have been testing for the compliance process in circumstances where it was not to be expected.

In conclusion, the present experiment provides additional support for the groupthink hypothesis regarding the effects of promotional leadership.

However, no support was obtained for the hypothesis that groups of people who are predisposed to conform would be more susceptible to groupthink than groups of less conforming people. It is suggested that the further development of better ways to measure groupthink indicators (viz., groupthink symptoms and symptoms of poor decision process) is perhaps the greatest need for future laboratory research on groupthink.

Aldag, R. J., & Fuller, S. R. (1993). Beyond fiasco: A reappraisal of the groupthink phenomenon and a new model of group decision processes. Psychological Bulletin, 113, 533-552.

Callaway, M. R., Marriott, R. G., & Esser, J. K. (1985). Effects of dominance on group decision making: Toward a stress reduction explanation of groupthink. Journal of Personality and Social Psychology, 49, 949-952.

Esser, J. K. (1998). Alive and well after 25 years: A review of groupthink research. Organizational Behavior and Human Decision Processes, 73, 116-141.

Flowers, M. L. (1977). A laboratory test of some implications of Janis’ groupthink hypothesis. Journal of Personality and Social Psychology, 35, 888-896.

Fodor, E. M., & Smith, T. (1982). The power motive as an influence on group decision making. Journal of Personality and Social Psychology, 42, 178-185.

Glaser, B. (1993). Black bear! King of Prussia, PA: Organization Design and Development.

Glaser, R. (1993). Groupthink index. King of Prussia, PA: Organization Design and Development. Hart, P. (1991). Irving L. Janis’ Victims of groupthink. Political Psychology, 12, 247-278.

Janis, I. L. (1972). Victims of groupthink. Boston: Houghton-Mifflin.

Janis, I. L. (1982). Groupthink. (2nd ed.). Boston: Houghton-Mifflin.

Janis, I. L. & Mann, L. (1977). Decision making: A psychological analysis of conflict, choice, and commitment. New York: Free Press.

Johnson, M. A. (1989). Concern for Appropriateness Scale and behavioral conformity. Journal of Personality Assessment, 53, 567-574.

Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of attitude Change. Journal of Conflict Resolution, 2, 51-60.

Leana, C. R. (1985). A partial test of Janis’ groupthink model: Effects of group cohesiveness and leader behavior on defective decision making. Journal of Management, 11, 5-17.

Lennox, R. D., & Wolfe, R. N. (1984). Revision of the Self-Monitoring Scale. Journal of Personality and Social Psychology, 46, 1349-1364.

McCauley, C. (1989). The nature of social influence in groupthink: Compliance and internalization. Journal of Personality and Social Psychology, 57, 250-260.

Montanari, J. R., & Moorhead, G. (1989). Development of the Groupthink Assessment Inventory. Educational and Psychological Measurement, 49, 209-219.

Moorhead, G., & Montanari, J. R. (1982). Groupthink: A research methodology. In G. P. White (Ed.), Proceedings of the 14th Annual Meeting of the American Institute for Decision Sciences (Vol. 1, pp. 380-382).

Moorhead, G., & Montanari, J. R. (1986). An empirical investigation of the groupthink phenomenon. Human Relations, 39, 399-410.

Neck, C. P., & Moorhead, G. (1992). Jury deliberations in the trial of U.S. v. John DeLorean: A case analysis of groupthink avoidance and an enhanced framework. Human Relations, 48, 1077-1091.

Neck, C. P., & Moorhead, G. (1995). Groupthink remodeled: The importance of leadership, time pressure, and methodical decision-making procedures. Human Relations, 48, 537-557.

Park, W. (1990). A review of research on groupthink. Journal of Behavioral Decision Making, 3, 229-245.

Tetlock, P. E., Peterson, R. S., McGuire, C., Chang, S., & Feld, P. (1992). Assessing political group dynamics: A test of the groupthink model. Journal of Personality and Social Psychology, 63, 403-425.

Wolfe, R. N., Lennox, R. D., & Cutler, B. L. (1986). Getting along and getting ahead: Empirical support for a theory of protective and acquisitive self-presentation. Journal of Personality and Social Psychology, 50, 356-361.

Wolfe, R. N., Welch, L. K., Lennox, R. D., & Cutler, B. L. (1985). Concern for appropriateness as a moderator variable in the statistical explanation of the self reported use of alcohol and marijuana. Journal of Personality, 53, 1-16.