Journal of Management

Structured conflict and consensus outcomes in group decision making

Structured conflict and consensus outcomes in group decision making

Richard L. Priem

Pressures for early consensus during group decision processes often lead to poor choices. However, consensus as an outcome of group decision processes is often desirable for implementing choices. We propose and test hypotheses that structured decision making techniques designed to enhance the expression of cognitive conflict will, paradoxically, (1) strengthen group consensus about and individual acceptance of the group’s eventual choices, and (2) increase member satisfaction with the group. Faced with a realistic managerial scenario, nineteen groups in this study deliberated using the structured, conflict-enhancing dialectical inquiry (DI) approach; nineteen used the consensus (C) approach. Group consensus on the decision, individual acceptance of the decision, and member satisfaction with the group were higher in the DI than in the C conditions. We discuss implications for group decision aids and for future laboratory and field studies of group consensus on a course of action.

Recent evidence suggests increases in both the use and the study of small groups in organizations (Bettenhausen, 1991; Levine & Moreland, 1990; Peters, 1988; Shea & Guzzo, 1987; Sundstrom, De Meuse & Futrell, 1990). This renewed emphasis on groups may stem from two widely shared perceptions about what groups do well. First, small groups are seen as effective vehicles for decision making and task performance (David, Pearce & Randolph, 1989; Hackman, 1990; Semler, 1989). They plan, choose, and execute a variety of organizational functions (McGrath, 1984), and may make better decisions than individuals when facing complicated problems (Schweiger & Sandberg, 1989). Second, small groups are seen as effective vehicles for influence and socialization. Research indicates that groups can reduce variance in member behaviors through the development and maintenance of explicit and implicit group norms (Feldman, 1984; George, 1990; Hayes, 1976; McGrath, 1984; Tuckman, 1965).

These ideas–that groups are effective for task performance and for member socialization–may be at odds in certain contexts. High quality decisions by groups facing complex, ambiguous situations often require multiple perspectives (Hoffman & Maier, 1961; Shaw, 1981; Triandis, Hall & Ewen, 1965), the expression of contrary viewpoints (Nemeth, 1986), and the evaluation of multiple alternatives (Schweiger, Sandberg & Ragan, 1986). Member socialization may, however, compromise decision quality by leading to groupthink (Janis, 1972, 1982), to pressures toward concurrence-seeking (Gero, 1985) and to satisficing behaviors (Simon, 1957). Historian Arthur Schlesinger, aide to President Kennedy and a member of Kennedy’s top decision making team, offers an anecdote in A Thousand Days (1965) that exemplifies pressures toward premature consensus. Reflecting on the planning of the Bay of Pigs invasion, he admonishes himself:

for having kept so silent during those crucial discussions in the Cabinet

Room, though my feelings of guilt were tempered by the knowledge

that a course of objection would have accomplished little save to gain

me a name as a nuisance. I can only explain my failure to do more

than raise a few timid questions by reporting that one’s impulse to

blow the whistle on this nonsense was simply undone by the

circumstances of the discussion (p. 255, emphasis added).

Researchers have suggested techniques for reducing the risk of unexpressed disagreement and resulting suboptimal decisions in groups facing complex, ambiguous problems. The dialectical inquiry (DI) technique, for example, introduces a subgroup structure into group interaction to increase the otherwise limited expression of cognitive conflict during decision making (Cosier & Schwenk, 1990; Mason, 1969). In DI, the decision making group is divided into two subgroups. The first subgroup independently develops recommendations for the problem, lists the assumptions on which these recommendations are based, and presents their recommendations and assumptions to the second subgroup. The second ,subgroup is then mandated to develop new assumptions that are counter to those of the first subgroup and, from these assumptions, to develop a new set of recommendations for the problem. The counter assumptions and new recommendations are then presented to the first subgroup, and a debate ensues until both subgroups can agree upon a set of assumptions. These final assumptions are then used by the entire group in developing the group’s final recommendations.

The frequently used consensus (C) technique of group decision making encourages the expression of cognitive conflict among group members without providing an explicit structure for group interaction. In C decision making, subgroups are not formed. The entire group is instructed to treat early agreement as suspect, to encourage each member to fully express his or her opinions, and to treat disagreement as a positive part of the decision making process. The group is further instructed to strive for consensus among all members rather than resorting to majority rule, coin flips, or other techniques to speed reaching the final recommendations. Schweiger et al. (1986) provide more detailed descriptions of the DI and C techniques.

Recent research suggests that DI produces higher quality decisions than does the C technique (Schweiger et al., 1986; Schweiger, Sandberg & Rechner, 1989; Schweiger & Sandberg, 1989). There has been relatively little evaluation of other important outcomes of structured decision techniques (e.g., member satisfaction, decision acceptance–for exceptions, see: Schweiger et al., 1986; Wall & Nolan, 1987; Watson, De Sanctis & Poole, 1988). Group task performance, however, frequently involves both the making of choices and their effective implementation. The studies of conflict-inducing techniques performed to date have generally equated subjectively-evaluated decision quality with group performance. The focus has been on the decision itself, with little theoretical or empirical attention to other outcomes that may influence how effectively the decision may be implemented by the group (McGrath, 1984). Future research on group decision aids must begin to consider multiple outcomes of the decision process, and their potential to influence “as implemented” group effectiveness.

In this paper, we take an initial step toward that consideration. First, we review theory and empirical results suggesting that increasing the generally low level of cognitive conflict expressed during group decision making produces higher quality decisions for complex problems. Then, we extend the theory by relating the expression of cognitive conflict during decision making to other key outcomes of the decision process that may influence effective implementation of a group’s choices. Next, we test the theoretical extensions in a laboratory study, using two different techniques that have been advocated as beneficial for groups making complex decisions. Finally, we discuss implications of the results for future laboratory and field research on group techniques for solving complex problems.

Consensus Pressures and Cognitive Conflict

Problem-solving groups have a strong tendency to converge on a single solution (Steiner, 1982). Once a position appears to have gained the support of a majority, no matter how slight the majority is, other positions are unlikely to be adopted even if they are demonstrably better. Janis (1972, 1982) and others have long warned against the danger of this overzealous concurrenceseeking, wherein the conformity pressures and ensuing self-censorship exemplified by Schlesinger’s anecdote can lead to premature consensus in group decisions. Premature consensus occurs when consensus develops on a position before all viable alternative positions have been thoroughly evaluated, rather than after. Frequently cited group decision disasters that may have been influenced by pressures toward premature consensus include the bombing of Pearl Harbor, the Bay of Pigs Invasion, the escalation of the Vietnam War, the launching of the space shuttle Challenger, and the Iran-Contra affair (Whyte, 1989).

Decision making techniques that add structure to group interaction have been proposed as useful for minimizing the problem of premature consensus in groups solving complex, ambiguous problems (e.g., Cosier & Schwenk, 1990). These techniques seek to facilitate adoption of the best solution by introducing subgroup structure designed to increase the level of cognitive conflict expressed during group discussion. Cognitive conflict is depersonalized, task-related conflict, involving the amount of disagreement over the implications of a common set of facts, or over the proper courses of action toward reaching a common goal (Cosier & Rose, 1977; Pace, 1990). Because cognitive conflict is debate over substantive rather than socioemotional issues (Hoffman, Harburg & Maier, 1962), opinion changes resulting from the resolution of cognitive conflict are more likely to be due to informational (fact-based) than interpersonal (affect-based) influences on group members (Stasser, Kerr & Davis, 1980).

Socio-emotional conflict, on the other hand, can involve conflicts of individual motives, competition to obtain individual-level rewards or avoid costs, or personal disagreements on dimensions that are unrelated or tangentially related to the group’s task (Cosier & Rose, 1977; Putnam, 1986; Torrance, 1957). Socio-emotional conflict has been theorized to have negative effects on several short-and long-term group outcomes (McGrath, 1984; Putnam, 1986). Although few empirical studies have dealt with it directly, Jehn (1991) reported socio-emotional conflict to be negatively related to participant satisfaction with the group.

The DI technique uses subgroup structure to increase the expression of cognitive conflict during decision making, but contains no explicit instructions for supressing socio-emotional conflict. Schweiger and Finger’s (1984) review and Schwenk’s (1990) recent meta-analysis of research examining the efficacy of decision making techniques each indicate only equivocal support for the relative superiority of DI when compared to a similar structured technique called devil’s advocacy. However, research has been fairly consistent in finding that the structured techniques lead to more effective decisions for well-structured tasks than expert-based techniques (where expert knowledge is made available to the group: see Schwenk, 1988) and, to a lesser extent, more effective than the C technique for complex, ambiguous tasks (Schweiger et al., 1986; Schweiger et al., 1989). The structure introduced by the DI technique may serve to legitimize the expression of cognitive conflict in groups making complex decisions, thereby encouraging the formation and discussion of minority viewpoints (Nemeth, 1986). The C technique also encourages cognitive conflict during decision making, but does so in an unstructured way by simply asking participants to fully express their ideas and opinions. Schweiger et al. (1986) found, however, that C groups reported greater acceptance of the group decision and greater satisfaction with the group process than did DI groups.

Cognitive Conflict and Group Consensus, Decision Acceptance, and Member Satisfaction with the Group

A number of outcomes of the decision making process have been tied to implementation effectiveness. For example, consensus among group members, individual acceptance of the group decision, and member satisfaction have all been argued as influencing solution implementation (Bettenhausen, 1991; Dess, 1987; Tjosvold, 1987; Vroom & Yetton, 1973; Watson et al., 1988). The following sections suggest that the level of cognitive conflict expressed and resolved during the decision making process is an important antecedent to group consensus and individual acceptance of the eventual group decision, and to member satisfaction with the group.

Group Consensus

Consensus, defined as general agreement among all or most group members, has long been considered a desired outcome of group decision processes. As noted by Whyte, for example, “the task, after all, of a decision making group is to produce consensus from the initial preferences of its members” (1989, p. 41). Most group research, however, has concentrated on pre-decision agreement among the initial preferences of group members (e.g., Castore & Murnighan, 1978), rather than on post-decision consensus. Why is such outcome consensus important? Although there are few data, some have suggested that consensus on a group decision may be necessary for successful implementation of that decision (Brodwin & Bourgeois, 1984; Quinn, 1980; Torrance, 1957; Watson et al., 1988). Quinn, for example, notes the need for “crystallizing” consensus before taking action. Others have suggested indirectly that consensus is necessary for the intermediate or long-term productivity and survival of the group (Bennis & Shepard, 1956; Hare, 1976; Tuckman, 1965). In the strategic management literature, top management team consensus about strategic choices has been proposed as a precondition of effective firm performance (Bourgeois, 1980, 1985; Dess, 1987).

Fisher (1980) clarifies the important distinction between the consensus process and the consensus outcome of decisions. His four-phase decision-emergence model (orientation, conflict, emergence, reinforcement) leads to the consensus outcome, but even the consensus process must pass through the cognitive conflict phase. Researchers have suggested a link between the expression of cognitive conflict during the decision process and the consensus decision outcome (Fisher, 1970, 1980; Priem, 1990; Torrance, 1957). Fisher (1980), for example, argues that greater consensus results when groups go through a greater amount of disagreement during their interactive decision making. This somewhat counterintuitive assertion is based on the rationale that cognitive conflict, when expressed and resolved during group decision making, will produce a strong consensus on the decision outcome, while premature consensus due to concurrence-seeking “smooths over” considerable latent disagreement. One would therefore expect that the DI technique, structured to encourage expression of substantive disagreement during the decision making process, would produce higher levels of outcome consensus than would the C technique. These arguments lead to our first hypothesis:

H1: Group decision making techniques that induce greater

expression of cognitive conflict during the decision making process

will result in stronger group consensus concerning the decision

outcome than will techniques that induce lesser expression of cognitive

conflict during decision making.

Individual Acceptance

In examining social influence in groups, Kelman (1958, 1961) made the critical distinction between public compliance with a group decision (i.e., conformity in manifest behaviors) and private acceptance of that decision (i.e., a true change in internal attitude level). One might expect that public compliance (but covert disagreement) is most likely in the C technique because of the well- documented pressure for uniformity of opinion among group members, even on important and ambiguous issues. The teeth of this pressure is the implied threat of social ostracism for not going along with the group’s opinion (e.g., as described in Schlesinger’s anecdote). The higher-conflict DI technique legitimizes the expression of dissent, reducing the threat of punishment for initial non-acceptance by mandating disagreement in the structured discussion. This leads to our second hypothesis

H2: Group decision making techniques that induce greater

expression of cognitive conflict during the group decision making

process will result in greater acceptance of the group decision by

individual group members than will techniques that induce lesser

expression of cognitive conflict during decision making.

Member Satisfaction

Lott and Lott (1965) reviewed evidence suggesting that a member’s attraction or liking toward the group and its members is a fundamental component of a group’s effectiveness. Such member satisfaction with the group may be predicted by hypotheses from Folger’s (1977; Folger & Greenberg, 1985; Folger, Rosenfield, Grove & Corkran, 1979) work on “voice” processes in procedural justice, and from the voluminous research on participation in decision-making (e.g., Tjosvold, 1987). A basic assertion and finding of these research areas is that individuals receive greater satisfaction from social groups and systems that contain voice-giving, participatory, or high-input procedures (Greenberg & Folger, 1983; Miller & Monge, 1986). Consensual decision making tends to stifle active participation and voice among some group members. The DI technique introduces a subgroup structure and procedures that ensure all group members have an unimpeded voice in a group’s decision–first, by assigning group members a position to champion (a positive voice), and second, by assigning group members a position to criticize (a negative voice). These opportunities for voice are contrived to take place at an abstract, conceptual level rather than an interpersonal level. These ideas lead to our final hypothesis:

H3: Group decision making techniques that induce greater

expression of cognitive conflict during the group decision making

process will result in greater member satisfaction (i.e., affect) with the

group than will techniques that induce lesser expression of cognitive

conflict during decision making.



Eighty male and 72 female business students at two large, urban universities in the southwestern U.S. took part in the investigation. All subjects participated during the final weeks of a 16-week, senior-level capstone course in strategic or international management. Thus, subjects were knowledgeable about the type of problem to be solved in their group discussions. The subjects were diverse in their respective functional backgrounds or courses of study (accounting, marketing, engineering, finance, real estate, information systems, health care, nursing, urban affairs, and management). The mean age of the subjects was 26.4 years (SD = 6.3). Eighty students (52%) had managerial experience. The average amount of this managerial experience was 3.5 years (SD = 1.8). The average amount of full-time work experience was 5.9 years (SD = 6.9).


During a class period one week before the experiment, the instructor handed out the “Harvey Wallbanger” case exercise (Cateora, 1990). Students were instructed that the exercise would be used in a graded group decision making activity during the next class period. The case exercise involves a U.S. gourmet popcorn manufacturer deciding how to enter the savory snack market in the United Kingdom. The students were assigned to individually develop (prior to the next class meeting) four prioritized strategies for entering the market, and four actions recommended to implement their preferred strategy.

The task and procedures had several strengths not commonly found in laboratory research on groups (McGrath, 1984; Shea & Guzzo, 1987). First, the case had a non-trivial level of realism for the students debating it. It contained vivid information about members of existing organizations making actual strategic decisions. Second, the students did not perform the task “cold.” Each subject had a week to go over the details of the case and was instructed to develop his or her own solutions to it. Thus, the case and its potential solutions were not completely novel; subjects had some degree of historical familiarity with it before they came to the laboratory. Third, the case dealt with principles that the subjects had studied in their course, so the problem-solving process and its parameters also were not novel. Fourth, the solution to the case exercise was ambiguous (it had no demonstrably or existentially “right” answer, which was important for generating differences of opinion in the group discussion). Fifth, the quality of the group’s final solution had real, valued consequences (course points) for the group members.

At the next class meeting, students were assigned to four-person groups using a random number table, with each group subsequently assigned to either a DI or C treatment condition. The groups were separated by treatment, and each group was provided with a written description of the decision making technique they were to use. The experimenter stressed to each group the important factors associated with using their decision making technique, and fielded questions from group members. During the experiment each group was asked to develop, using their assigned technique, four identifiably different strategies for their case study firm to enter the U.K. market (e.g., franchising, forming a joint venture with a British firm, or establishing a wholly-owned subsidiary in the U. K.), and then to rank those entry strategies from most to least desirable. For their top-ranked strategy, each group was next instructed to develop four different implementation actions (e.g., one group supported their joint venture recommendation by suggesting: that the firm reorganize to include an international division; that this division identify possible joint venture partners based on complementary products and distribution systems; that the product introduction stress consumer education through samples in markets and pubs; and that a small London office be established for quality control and local interface). Lastly, the groups were instructed to develop a group rating of the relative importance of each implementation action to the success of their top-ranked strategy, by responding (as a group) to a 10-point Likert-type scale.

After their group work was finished, group members were asked to individually respond to questionnaire items designed to evaluate their acceptance of the group’s entry strategy rankings and their acceptance of the group’s importance weightings for implementation actions. Other questions evaluated their satisfaction with the group. Finally, the group members were also asked to indicate their own, individual preferences for entry strategy rankings and their own, individual ratings for relative importance of the group’s implementation actions.

The experimenter provided time guidelines prior to and during the decision exercise, and coordinated separation of the subgroups when required by the DI technique. All groups completed the exercise within the allotted 80-minute time period. The experimenter debriefed the subjects after all group members had completed their discussion and rating tasks.

Manipulation of the Independent Construct

The independent construct in our hypotheses was the degree of cognitive conflict expressed in group discussion. We created two levels of this construct by using two opposing group decision techniques: Dialectical Inquiry (DI) and Consensus (C). Each subject read detailed instructions for their group’s assigned technique, approximately one single-spaced page in length, from Schweiger et al. (1986). The DI instructions first divided the group into two subgroups. Then, one of the subgroups independently developed recommendations for the problem under consideration, and listed both their recommendations and the assumptions on which they were based. Next, the second subgroup was given this information and instructed to independently develop plausible assumptions opposite to those of the first subgroup. The second subgroup then developed recommendations based on these counter assumptions. The entire group finally convened and debated the two sets of assumptions and recommendations. The assumptions that survived this debate were the foundation for final recommendations developed and agreed to by the entire group. The C instructions did not provide a subgroup structure for the decision making process, but did provide guidance designed to encourage the expression of cognitive conflict. For example, the C instructions stated that the group should view disagreement as a positive part of the process, should be suspicious of early agreement, and should avoid coin flips, majority rules, and other conflict-limiting methods. Both the DI and the C instructions stress that full expression of cognitive conflict by all group members during decision making is necessary for sound decisions.

We used the more highly-structured DI technique to create higher levels of expressed cognitive conflict during decision making. We used the C technique a control because it is frequently used in industry (Gero, 1985; Priem, 1990), and because it has no subgroup structure. However, the C technique is not a laissez-faire technique, since it provides specific instructions for decision making. The experimenter spent as much time instructing and guiding subjects to use this technique as to use DI.

Manipulation Checks

Success of the manipulation of expressed cognitive conflict was gauged by the critical evaluation scale from Schweiger et al. (1986). This 2-item scale, using a 5-point Liken response format, is designed to measure how much critical evaluation of assumptions and recommendations occurred during decision making. The items were: “The group decision process made me critically reevaluate the validity of the assumptions and recommendations I made in my individual analysis,” and “The group decision process uncovered valid recommendations and assumptions that I did not consider in my individual analysis.” The estimated reliability of this scale was [Alpha] = .76.

The DI manipulation was checked with an additional 2-item scale from Schweiger et al. (1986) containing 5-point, Liken response formats. The individual items were: “Our group considered two conflicting sets of strategies and implementation methods based on opposite sets of assumptions,” and “On the basis of a debate over two sets of recommended strategies and implementation actions (based on opposing assumptions) developed by each subgroup, our group developed a pool of valid assumptions and based its recommended strategy and implementation methods on these assumptions.” The estimated reliability of this additional scale was [Alpha] = .66.

Agreement with the critical evaluation scale and DI manipulation check differed significantly in the expected direction by treatment condition, tested at both the individual and group levels ([t.sub.150]=3.4,p [is less than] .01; [t.sub.150]=2.7, p [is less than] .01; [t.sub.36]=3.2, p [is less than] .01; [t.sub.36]=2.6, p [is less than] .01, respectively). The mean values for critical evaluation were 8.2 (s.d.= 1.2) for DI and 7.6 (s.d.= 1.8) for C. The DI manipulation check means were 8.1 (s.d.=1.3) for DI and 7.1 (s.d.=1.9) for C.

Measurement of Dependent Constructs

Group Consensus: Entry Strategies. Groups and individuals ranked their preferences for U.K. market entry strategies that might solve the case exercise, Group consensus on strategy preference was determined by first calculating the absolute value of the differences in ranks between each individual member’s (post-discussion) ranking and the group’s rankings. This score reflects group members’ variation around the group’s stated priorities. The absolute values of these individual disagreement measures were then summed to produce a group-level value. Larger summed deviations, indicate less consensus about the group s stated priorities. This operationalization has been used by past consensus researchers (Wooldridge & Floyd, 1990), and a similar form has been used to assess disagreement in a number of group decision making exercises (e.g., Hall, 1971).

Group Consensus: Implementation Actions. Individual group members rated the importance of each of their group’s four chosen implementation actions on a Likert-type scale. Group consensus on implementation actions was determined by the standard deviation of individual members’ preference ratings for each recommended action, summed over the four recommended actions. Once again, higher summed standard deviations within the group indicate lower levels of group consensus. This operationalization of group consensus has been used frequently in prior work by consensus researchers (Bourgeois, 1980, 1985; Doss, 1987; Dess & Keats, 1987).

The separate sets of researchers mentioned above use separate operationalizations of consensus. To bring both of these methodological streams together, we used both ratings (for deriving consensus on implementation actions) and rankings (for deriving consensus on entry strategies). This provides multiple, convergent operations of a key dependent construct in our study.

Individual Acceptance, Individual acceptance of the group’s strategic choice or implementation actions was measured via n straightforward approach based on past research. A post-task questionnaire included two, 2-item acceptance scales. The response format for each item was a 5-point Likert scale (ranging from strongly disagree to strongly agree). The items were: “I am committed to my group’s recommended strategy,” and “I am satisfied with my group’s recommended strategy.” These two items were repeated for acceptance of the group’s rated importance of implementation actions by substituting the word “strategy” with “implementation actions.” Coefficient [Alpha] for acceptance of the group’s strategic choice was .92. Coefficient [Alpha] for acceptance of the group’s implementation actions was also .92. These two-item scales were used previously in Schweiger et al. (1986) and Tjosvold and Field (1985).

Member Satisfaction. Individual satisfaction with the group was measured through a two-item scale on the post-task questionnaire. The two items, employing the same S-point Likert scale, were: “Working with my group was an enjoyable experience,” and “1 would like to work with this group for the additional group exercise that will be performed next clan period, rather than switching to another group.” Coefficient alpha for this scale, modified from Schweiger et al. (1986) and Tjosvold and Field (1985), was .81.

Decision Quality. The primary foci of this research were consensus, acceptance, and satisfaction, since each of these variables may influence a group’s effectiveness at implementing a chosen course of action. We did not develop an explicit hypothesis concerning the subjectively-evaluated quality of the decision itself. However, a noteworthy body of prior research relates the use of DI and C techniques to decision quality, suggesting that the DI technique leads to higher quality decisions (Schweiger et al, 1986; Schweiger et al., 1988). Therefore, we compared the quality of our groups’ solutions across DI and C conditions. The Harvey Wallbanger case was initially chosen for this research because solutions are neither obvious nor easily agreed upon. That is, the “correct solution” to the problem outlined in the case is ambiguous. This makes the measurement of decision quality somewhat difficult. In the interest of replicating previous studies and providing additional evidence, however, we scored each group’s set of solutions for quality.

The quality of each group’s entry strategies and recommended implementation actions was judged by two strategic management professors who were unaware of the assigned treatment conditions, using holistic scoring techniques. All group recommendations were read separately by each judge and ordered by subjectively-judged quality based on that judge’s overall (i.e., holistic), rather than criterion-specific, evaluation of the work. This reading-ordering was repeated several times until the judge felt confident in assigning ratings to each group ranging from 1 (low quality) to 5 (high quality). The sum of the judges’ ratings represents each group’s decision quality score. The interrater reliability estimate for this decision quality measure was r = .95.


This experiment used a one-factor, two-group design. However, the dependent constructs of group consensus and individual acceptance each had multiple operations: consensus on strategy and consensus on implementation actions were operationalized differently, and 2-item scales were used for both individual acceptance of the group’s strategy choice and individual acceptance of the group’s implementation actions. These operations were correlated, but not enough to warrant combining them into single scales since they did not share more than 50% of their variance with one another. Group consensus on strategy priorities was positively correlated with consensus on recommended implementation actions: [r.sub.36]=.33 (p [is less than] .05). Individual acceptance of the group’s strategy priorities was positively related to individual acceptance of the group’s recommended implementation actions: [r.sub.152]=.68 (p [is less than] .01).

The use of two operations for each construct, and the moderate correlations between these operations, prompted the use of MANOVA techniques to compare DI and C conditions on group consensus and individual acceptance. Member satisfaction and decision quality were compared using simple t tests. Table 1 presents the means, standard deviations, and results for each of these dependent measures. Correlations and descriptive statistics for these measures, at both group and individual levels, are given in Table 2.

Table 1. Statistical Comparisons of Group Consensus, Individual Acceptance, Satisfaction, and Decision Quality under Dialectical Inquiry (DI) and Consensus (C) Conditions


Dependent Measure Mean (SD)

Group Consensus(a):

Strategies 2.8 (3.0)

Implementation Plans 2.1 (1.3)

Individual Acceptance:

Strategies 8.8 (0.7)

Implementation Plans 8.8 (0.7)

Member Satisfaction: 9.1 (0.7)

Decision Quality: 5.9 (2.7)


Dependent Measure Mean (SD)

Group Consensus(a):

Strategies 3.9 (2.9) [F.sub.2,35] = 4.02(*)

Implementation Plans 3.7 (1.5)

Individual Acceptance: [F.sub.2,35] = 2.48(*)

Strategies 8.6 (0.6)

Implementation Plans 8.2 (0.7)

Member Satisfaction: 8.3 (1.3) [F.sub.1,36] = 4.36(*)

Decision Quality: 6.2 (2.0) [F.sub.1,36] = .11(*)


(a) Smaller numbers indicate less dispersion, and hence, more consensus. p < .10;

(*) p < .05.

Table 2. Descriptive Statistics and Correlations Among Dependent Measures

Group-Level Individual-Level

Dependent Measure Mean (SD) Mean (SD)

1. Cognitive Conflict (man. chk) 7.9 (0.9) 7.9 (1.6)

2. GC: Strategies 3.0 (2.9)

3. GC: Implementation Plans 2.8 (1.5)

4. IA: Strategies 8.6 (0.7) 8.6 (1.3)

5. IA: Implementation Actions 8.5 (0.7) 8.5 (1.3)

6. Member Satisfaction 8.7 (1.1) 8.7 (1.5)

7. Decision Quality 6.0 (2.4)

Dependent Measure 1 2 3 4

1. Cognitive Conflict (man. chk) .76 03 -13 .47(**)

2. GC: Strategies .01 — .33(*) -.33(*)

3. GC: Implementation Plans -.08 — — -.29(*)

4. IA: Strategies .18(*) — — .92

5. IA: Implementation Actions .19(*) — — .68(**)

6. Member Satisfaction .25(**) — — .38(**)

7. Decision Quality .03 — — —

Dependent Measure 5 6 7

1. Cognitive Conflict (man. chk) .36(*) .37(*) .07

2. GC: Strategies -.15 -.19 -.10

3. GC: Implementation Plans -.39(*) -.07 -.16

4. IA: Strategies .71(**) .56(**) .26

5. IA: Implementation Actions .92 .47(**) .14

6. Member Satisfaction .36(**) .81 -.11

7. Decision Quality — — .95


GC = Group Consensus (reverse scaled): IA = Individual Acceptance; Correlations above the diagonal summarize group-level relations, those below the diagonal summarize individual-level relations. Diagonal entries are reliability estimates.

p < 0.1;

(*) p < 05;

(**) p < .01.

Group Consensus

H1 predicts stronger consensus in groups that express more cognitive conflict during decision making. The direction and strength of this prediction is borne out by the pattern of means in Table 1 and the significant main effect in the MANOVA: [F.sub.2,35]=4.02 (p [is less than] .05; all uni- and multivariate distributional assumptions held). However, univariate tests indicated that more of the multivariate effect was concentrated in consensus about the group’s recommended implementation actions ([t.sub.36]=2.83, p [is less than] .01) than in consensus about strategies ([t.sub.36]=1.26, p=.21). This may be due, in part, to the greater power provided by metric (rating) rather than ordinal (ranking) information in the raw data. For the implementation actions, the manipulation of structured cognitive conflict explained [[Eta].sup.2]=.18 (18%) of the variance in group consensus. However, the manipulation explained a smaller, [[Eta].sup.2]=.05 proportion of the variance in consensus about the group’s strategic priorities.

Individual Acceptance

H2 predicts stronger individual acceptance of a group’s position for members of groups whose decision making process involve greater expression of cognitive conflict. The direction and strength of this second prediction is also reflected in the results from Table 1. Means for the acceptance measures were both higher in the DI condition, producing a significant multivariate effect: [F.sub.2,35]=2.48 (p [is less than] .10; there were no observable violations of uni- or multivariate distributional assumptions). Once again, univariate tests showed that more of the effect was concentrated in acceptance of the group’s recommended implementation actions ([t.sub.36]=2.15, p [is less than] .05) than in acceptance of strategic priorities ([t.sub.36]=.95, p=.34).(1)

Member Satisfaction

According to H3, group members who make decisions under conditions of greater expressed cognitive conflict will be more satisfied with their groups than will those persons in groups that express less cognitive conflict. Table 1 shows that DI group members did report greater satisfaction with their group than did C group members ([t.sub.36]=2.10, p [is less than] .05), even in the presence of a somewhat restricted range of satisfaction (no one expressed a clear dissatisfaction with their assigned group). The conflict-inducing manipulation explained [[Eta].sup.2]=.06 (6%) of the variance in individual member satisfaction. Member satisfaction may have flowed from acceptance of (the correctness of) the group’s solutions. Correlations between satisfaction and individual acceptance were [r.sub.36]=.56 (p [is less than] .01) for entry strategies, and [r.sub.36]=.47 (p [is less than] .01) for implementation plans.

Decision Quality

Although we were prepared to adjust course points to equalize any group performance (decision quality) differences due to the experimental conditions, it was unnecessary. There were no reliable or even marginally reliable differences between the DI and C groups. In fact, Table 1 shows that mean decision quality for the C groups was slightly but not significantly higher than for DI groups ([t.sub.36]=-.33, p [is greater than] .5).

Discussion and Implications


Consensus among decision makers has long been accepted as a desirable outcome of group decision making (Quinn, 1980; Whyte, 1989), yet it has been relatively ignored in research on decision making groups. The current study found that post-decision consensus among group members is greater when the DI group decision making technique is used than when the C technique is used. This suggests that techniques with subgroup structure that allows legitimizing, expressing and resolving cognitive conflict during group decision making produce strong agreement among group members on the resulting decision. This finding may be an important first step in introducing the consensus construct into the group decision making literature. Our study also found that the group consensus was stronger on actions for implementing strategies than it was for the strategies themselves. This second finding must be viewed cautiously, however, since consensus on strategies was measured using less powerful scaling than consensus on implementation actions.

Further research incorporating consensus may serve to clarify important relationships (Rosenberg, 1968) among group decision processes, decision quality and implementation effectiveness. Case studies by Reicken (1952), for example, suggest that higher levels of conflict in the decision making process are related to a willingness to expend greater effort to implement the decision outcome. Similarly, combined laboratory and field studies by White, Dittrich and Lang (1980) indicate that structure in group decision making processes is positively related to the number of implementation attempts undertaken. Thus, one might expect higher levels of consensus following the decision process to be related to the amount of individual and group resources allocated to implement the decision outcome, although this remains an empirical question.

Acceptance and Satisfaction

Also as hypothesized, use of the DI technique yielded greater decision acceptance and greater satisfaction with other group members than did the C technique. These results are contrary to the findings of Schweiger et al. (1986), but are consistent with predictions developed from social influence (Kelman, 1958, 1961) and procedural justice (Folger, 1977) theories. Schweiger et al. (1989) found that differences in acceptance between DI and C diminish with subject experience in using the techniques. It may be that the repeated use of each technique in our study (first for group rankings of strategies, then for rating implementation actions) contributed to the higher acceptance of and satisfaction with the DI approach when compared to Schweiger et al.’s (1986) findings.

Decision Quality

This study found no differences in decision quality between the DI and C techniques, contrary to the earlier work of Schweiger and his colleagues (1986, 1989). Characteristics of our subjects and tasks may partially explain the differing results. The subjects’ classroom work in strategy development and evaluation may have overwhelmed the manipulations’ effects regarding the decision quality construct. That is, subjects may have been working harder to please their instructors and demonstrate their use of concepts discussed in class, whereas the managers in the Schweiger et al. (1989) study may have been more focussed on making decisions as if they were actually running the firm in their case. In that regard, the subjects in this study may have had less of a vested interest in seeing their own preferences adopted by the group under the C condition, because those preferences were less likely to be based on direct experience in making such decisions in business settings. In Schweiger et al. (1989), those stronger preferences among practicing managers may have been more likely to generate poorer solutions in the C condition, because the conflict that those preferences generated was not as structured or legitimized as in the DI condition.

In addition, although the Harvey Wallbangcr case addresses important strategic issues, it may lack the length, complexity, or “richness” of the cases used in the Schweiger et al. studies, which may be necessary for performance differences to become evident. Other studies using different tasks have also found no quality differences between DI and C decision making (e.g., Innami, 1992). These results suggest that task dimensions and aspens of the problem solving context may be important characteristics for future research on decision quality.

Limitations and Research Directions

This study has a number of limitations. First, it involved ad hoc rather than on-going groups. Thus, although each group used their assigned technique first for the strategy problem and then for the implementation problem, the study does not address the effects that continued use of conflict-inducing decision making techniques may have on the effectiveness of on-going groups. Second, the case study used made all information available on an equal basis to all decision makers in the group. This is probably not the case in on-going groups of decision makers. In this study, power could be assumed to be equally distributed within the student groups, which again is probably not the case in on-going management groups. Also, practicing managers may be more willing (due to confidence and experience) or less willing (due to political or social pressures) to express cognitive conflict during decision making than were the subjects of this study.

Finally, an important limitation involves the measurement of cognitive conflict. Consistent with the theoretical foundations of the DI technique, this study found that more critical evaluation of assumptions and recommendations was reported in the DI groups than in the C groups. However, this study and other research in group decision making techniques has yet to directly measure the level of cognitive conflict that occurs during decision making using the different techniques. Such measurement could be undertaken through verbal protocol techniques and/or group-level conflict scales (e.g., Rahim, 1983). Future laboratory research might incorporate longitudinal designs and attempt to measure directly the levels of cognitive and social/emotional conflict (Cosier & Rose, 1977; Evan, 1965; Jehn, 1991), consensus, and decision quality resulting from the DI and C interventions.

Assuming that two groups develop solutions of equivalent quality, the performance achieved by the groups still may differ due to differences in implementation effectiveness. Field studies may be useful for examining links between decision making characteristics and the performance actually achieved following decision implementation by the group. For example, a simple field test of the conflict-performance hypothesis underlying the DI versus C lab work might involve measures of cognitive conflict during decision making, consensus outcomes, and achieved group performance for on-going decision making teams. Such designs would be a first step in addressing questions concerning the long-term effects of conflict-inducing techniques on the achieved performance of on-going groups. Consideration of factors involved in the implementation of group choices in future field and laboratory research may allow more reliable prescriptions about the use of DI and C techniques in particular decision making contexts.

Although extremely high or extremely low levels of cognitive conflict and consensus may be counterproductive (Hedberg, Nystrom & Starbuck, 1976; Priem, 1990), Gero (1985) notes that the frequently-practiced consensual decision process involves consensus-seeking behavior that suppresses or prohibits conflicts and disagreements. Given the low level of existing or expressed conflict in consensual decision making, the DI technique may either generate a higher (i.e., more optimum) level of conflict, provide a framework for the expression of such conflict where it already exists, or both.

Practical Implications

Practitioners face a number of problems when attempting to evaluate the usefulness of different group decision making techniques for their particular situations. First, neither strong consensus among decision makers nor strong commitment to a course of action may be appropriate in all situations (see, e.g., Grinyer & Norburn, 1975; Hedberg, Nystrom & Starbuck, 1976; Staw, 1976; Staw, Sandelands & Dutton, 1981). Thus, selection of a technique for group decision making should involve, in addition to decision quality factors, situational factors related to the problem, the decision context, and the group itself. Second, a switch to a different decision making approach is not without problems. The structured DI technique is somewhat more complicated than the more frequently used C approach. Individuals considering use of different techniques have been found to “expect less social harmony, lower group affect, and less confidence in the decision outcomes in the DI and DA techniques than in the C technique” (Priem & Price, 1991, p. 220). Thus, participants may initially have less enthusiasm for the structured approaches than for the C approach. The current results suggest, however, that following use of the higher-conflict DI approach decision maker acceptance of and consensus on the group choice, and satisfaction with other group members, can be high. Such factors may contribute considerably to the on-going effectiveness of the group.

Acknowledgements: We wish to thank Greg Dess, Ken Price and Jerry Wofford for their helpful comments on earlier drafts of this article.


(1). We report statistics based on a group-level analysis, with 38 data points per variable, because of the non-independence of subject in the same 4-person group. However, the acceptance and satisfaction data were also re-analyzed at the individual level (as 152 individual scores) for consistency with prior research (e.g., Schweiger et al., 1986). The individual-level analysis yielded patterns of results identical with those of the group-level analyses (Individual Acceptance [F.sub.2,149] = 4.19, p [is less than] .05; Individual Satisfaction [F.sub.1,150] = 10.16, p [is less than] .01).


Bennis, W G. & Shepherd, H.H. (1956). A theory of group development. Human Relations, 9: 415-437.

Bettenhausen, K.L. (1991). Five years of groups research: What we have learned and what needs to be addressed. Journal of Management, 1712): 345-381.

Bourgeois, L.J. III. (1980). Performance and consensus. Strategic Management Journal, 1: 227-248.

–. (1985). Strategic goals, perceived uncertainty, and economic performance in volatile environments. Academy of Management Journal, 28: 548-573.

Brodwin, D.R. & Bourgeois, L.J. III. (1984). Five steps to strategic action. California Management Review. 26(Spring): 176-190.

Castore, C.H. & Murnighan, J.K. (1978). Determinants of support for group decisions. Organizational Behavior and Human Performance, 22: 75-92.

Cateora, P.R. (1990). Harvey Wallbanger popcorn. Pp. 730-733 in P.R. Cateora, international marketing, 7th ed. Homewood, IL: Richard D. Irwin, Inc.

Cosier, R.A. & Rose, R.L. (1977). Cognitive conflict and goal conflict

effects on task performance. Organizational Behavior and Human Performance, 19: 378-391.

Cosier, R.A. & Schwenk, C.R. (1990). Agreement and thinking alike: Ingredients for poor decisions. Academy of Management Executive, 4: 69-74.

David, F.R., Pearce, J.A. & Randolph, W.A. (1989). Linking technology and structure to enhance group performance. Journal of Applied Psychology, 74: 233-241.

Dess, G.G. (1987). Consensus on strategy formulation and organizational performance: Competitor’, in a fragmented industry. Strategic Management Journal, 8: 259-277.

Dess, G.G. & Keats, B.W. (1987). Environmental assessment and organizational performance: An exploratory field study. Academy of Management Proceedings. 21-25.

Evan, W.M. (1965). Conflict and performance in R&D organizations. Industrial Management Review, 7: 37-46.

Feldman, D.C. (1984). The development and enforcement of group norms. Academy of Management Review, 9:. 47-53.

Fisher, B.A. (1970). Decision emergence: Phases in group decision making. Speech Monographs, 37: 53-66.

–. (1980) Small group decision making, 2nd ed. New York: McGraw-Hill.

Folger, R. (1977). Distributive and procedural justice: Combined impact of “voice” and improvement on experienced inequity. Journal of Personality and Social Psychology, 35: 108-119.

Folger, R. & Greenberg, J. (1985). Procedural justice: An interpretive analysis of personnel systems Pp. 141-183 in K.R. Rowland & G.R. Ferris (Eds.), Research in personnel and human resources management, Vol. 3. Greenwich, CT: JAI Press.

Folger, R., Rosenfield, D., Grove, J. & Corkran, L. (1979). Effects of “voice” and peer opinions on responses to inequity. Journal of Personality and Social Psychology, 37: 2253-2261.

George, J.M. (1990). Personality, affect and behavior in groups, Journal of Applied Psychology. 75: 107-116.

Gero, A. (1985). Conflict avoidance in consensual decision processes. Small Group Behavior, 16: 487-499.

Greenberg, J. & Folger, R. (1983). Procedural justice, participation, and the fair process effect in groups and organizations. Pp. 235-256 in P.B. Paulus (Ed.), Basic group processes. New York: Springer-Verlag.

Grinyer, P. & Norburn, D. (1975). Planning for existing markets: Perceptions of executives. Journal of the Royal Statistical Society, 138: 70-97.

Hackman, J.R. (1990). Creating more effective work groups in organizations. In J. R. Hackman (Ed.), Groups that work [and those that don’t]: Creating conditions for effective teamwork. San Francisco: Jossey-Bass.

Hall, J. (1971). Decisions, decisions, decisions. Psychology Today, (November): 51-54, 86, 88.

Hare, A.P. (1976) Handbook of small group research. New York: Free Press.

Hayes, L. (1976). The use of group contingencies for behavioral control: A review. Psychological Bulletin. 83:628.648

Hedberg, B.L.T., Nystrom, P.C. & Starbuck, W.H. (1976). Camping on seesaws: Prescriptions for a self-designing organization. Administrative Science Quarterly, 21:41-65.

Hoffman, L. R., Harburg, E. & Maier, N.R.F. (1962). Differences and disagreement as factors in creative group problem-solving. Journal of Abnormal and Social Psychology, 64:206-214.

Hoffman, L & Maier, N. R. F. (1961). Quality and acceptance of problem solving by mere bets of homogeneous and heterogeneous groups. Journal of Abnormal and Social Psychology, 62: 401-407.

Innami, I. (1992). Comparative effectiveness and working mechanism of the consensual conflict resolution, devil’s advocacy and dialectical inquiry interventions. Paper presented at the Academy of Management meetings, Las Vegas.

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

–. (1982). Groupthink, 2nd ed. Boston: Houghton-Mifflin.

Jehn, K.A. (1991). Benefits and detriments of group conflict. Paper presented at the Academy of Management meetings, Miami.

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

–. (1961). Processes of opinion change. Public Opinion Quarterly, 25: 57-78.

Levine, J.M. & Moreland, R. (1990). Progress in small group research. Annual Review of Psychology, 41: 585-634.

Lott, A.J. & Lott, B.E. (1965). Group cohesiveness as interpersonal attraction: A review of relationships with antecedent and consequent variables. Psychological Bulletin, 64: 259-309.

Mason, R.O. (1969). A dialectical approach to strategic planning. Management Science, 15: 403-414.

McGrath, J.E. (1984). Groups: Interaction and performance. Englewood Cliffs, NJ: Prentice-Hall.

Miller, K.I. & Monge, P.E. (1986). Participation, satisfaction, and productivity: A meta-analytic review. Academy of Management Journal, 29: 727-753.

Nemeth, C. (1986). Differential contributions of majority and minority influence. Psychological Review, 93: 23-32.

Pace, R.C. (1990). Personalized and depersonalized conflict in small group discussions: An examination of differentiation. Small Group Research, 21: 79-96.

Peters, T.J. (1988). Thriving on chaos. New York: Knopf.

Priem, R.L. (1990). Top management team group factors, consensus, and firm performance Strategic Management Journal, 11: 469-478.

Priem, R.L. & Price, K.H. (1991), Process and outcome expectations for the dialectical inquiry, devil’s advocacy, and consensus techniques of strategic decision making. Group and Organization Studies, 16: 206-225.

Putnam, L.L. (1986). Conflict in group decision making. Pp 175-196 in R.Y. Hirokowa & M.S. Poole (Eds.), Communication and group decision-making. Beverly Hills, CA: Sage.

Quinn, J.B. (1980. Strategies for change; Logical incrementalism. Homewood, IL: Irwin.

Rahim, M.A. (1983). Measurement of organizational conflict. Journal of General Psychology, 109: 188-199.

Reicken, H.W. (1952). Some problems of consensus development. Rural Sociology, 17: 245-252.

Rosenberg, M. (1968). The logic of survey analysis. New York: Basic Books.

Schweiger, D.M. & Finger, P.A. (1984), The comparative effectiveness of dialectical inquiry and devil’s advocacy: The impact of task biases on previous research findings. Strategic Management Journal

Schweiger, D.M. & Sandberg, W.R. (1989). The utilization of individual capabilities in group approaches to strategic decision making. Strategic Management Journal, 10: 31-43.

Schweiger, D.M., Sandberg, W.R. & Rasan, J.W. (1986). Group approaches for improving strategic decision making: A comparative analysis of dialectical inquiry, devil’s advocacy, and consensus. Academy of Management Journal, fig: 51-71.

Schweiger, D.M., Sandberg, W.R. & Rechner, P. (1989). Experiential effects of dialectical inquiry, devil’s advocacy, and consensus approaches to strategic decision making. Academy of Management Journal 32: 745-772.

Schwenk, C. R. (1988). The essence of strategic decision making. Lexington, MA: Lexington Books.

–. (1990). Effects of devil’s advocacy and dialectical inquiry on decision making: A meta-analysis. Organizational Behavior and Human Decision Processes, 47: 161-176.

Semler, R. (1989). Managing without managers: How an unorthodox company makes money by avoiding decisions, rules, and executive authority. Harvard Business Review: 76-84.

Shaw, M.E. (1981). Group dynamics: The psychology of small group behavior, 3rd ed. New York: McGraw-Hill.

Shea, G.P. & Guzzo, R.A. (1987), Groups as human resources. Pp. 323-356 in K. Rowland & G. Ferns (Eds.), Research in personnel and human resources management, Vol. 5. Greenwich, CT: JAI Press. Simon, H.A. (1957). Models of man. New York: Wiley.

Stasser, G., Kerr, N.L. & Davis, J.H. (1980). Influence processes in decision-making groups–a modeling approach. Pp. 431-477 in P.B. Paulus (Ed.), Psychology of group influence. Hillsdale, NJ: Erlbaum. Staw, B.M. (1976). Knee deep in the big muddy: A study of escalating commitment to a chosen course of action, Organizational Behavior and Human Performance, 16: 2744.

Staw, B.M., Sandelands, L.E. & Dutton, J.E. (1981). Threat-rigidity effects in organizational behavior: A multilevel analysis. Administrative Science Quarterly, 26: 501-524.

Steiner, I.D. (1982). Heuristic models of groupthink. Pp. 503-524 in M. Brandstatter, J.H Davis & G. Stocker-Kreichgauer (Eds.), Group decision making. New York: Academic Press. Sundstrom, E., De Meuse, K.P. & Futrell, D. (1990). Work teams: Application and effectiveness. American Psychologist, 45: 120-133.

Tjosvold, D. (1987). Participation: A close look at its dynamics. Journal of Management, 13: 739-750.

Tjosvold, D. & Field, R.H.G. (1985). Effect of concurrence, controversy, and consensus on group decision making. Journal of Social Psychology, 125(3): 355-363.

Torrance, E.P. (1957). Group decision making and disagreement. Social Forces, 33: 314-318. Triandis, H.C., Hall, E.R. & Ewen, R.B. (1965). Member homogeneity and dyadic creativity. Human Relations, 18: 33-54.

Tuckman, B.W. (1965). Developmental sequence in small groups. Psychological Bulletin, 63: 384-389.

Vroom, V.H. & Yetton, P.W. (1973). Leadership and decision making. Pittsburgh: University of Pittsburg Press.

Wall, V.D. & Nolan, L.L. (1987). Small group conflict: A look at equity, satisfaction, and styles of conflict management. Small Group Behavior, 18:188-211.

Watson, R., De Sanctis, G. & Poole, M.S. (1988). Using G.D.S.S. to facilitate group consensus: Some intended and unintended consequences. MIS Quarterly. 12: 463-478.

White, S., Dittrich, J. & Lang J. (1980). The effects of group decision making process and problem situation complexity. Administrative Science Quarterly, 25: 428.440.

Whyte, G. (1989). Groupthink reconsidered. Academy of Management Review, 14: 40-56.

Wooldridge, B. & Floyd, S.W. (1990). The strategy process, middle management involvement, and organizational performance. Strategic Management Journal. 10: 231-241.

Direct all correspondence to: Richard L. Priem, The University of Texas at Arlington, Department of Management, UTA Box 19467, Arlington, TX 76019-0467.

COPYRIGHT 1995 JAI Press, Inc.

COPYRIGHT 2004 Gale Group