JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY
ARTICLE NO.
34, 202–226 (1998)
JS971349
Cognitive Flexibility, Communication Strategy, and Integrative
Complexity in Groups: Public versus Private Reactions
to Majority and Minority Status
Deborah H Gruenfeld
Northwestern University
and
Melissa C. Thomas-Hunt and Peter H. Kim
Washington University
Received: December 16, 1996; revised: September 15, 1997; accepted: December 4, 1997
Archival studies of political decision-making groups show that the public statements of
policy makers in the majority are higher in integrative complexity than those of minorityfaction or unanimous group members. However, whether these differences reflect policy
makers’ private thoughts, or their public impression management strategies, cannot be
inferred using only data from the public record. The experiment reported here established
that in freely interacting groups composed of majorities and minorities, this pattern is
obtained under private communication conditions as well as in public statements. Results
suggest that cognitive flexibility in response to influence from insiders, rather than
communication strategies designed to influence outsiders, underlies the differences
observed. r 1998 Academic Press
INTRODUCTION
Researchers of majority and minority influence have traditionally been preoccupied with changes in targets’ opinions. In the typical study, the source or content
of a message is manipulated and its impact on the target is assessed. To the extent
This article is based on the first author’s doctoral dissertation. She is grateful to her thesis committee
for their important contributions to this research: Bob Wyer, Joe McGrath, Ulf Bockenholdt, Peter
Carnevale, and Janet Sniezek at the University of Illinois, Urbana-Champaign; and Larry Baum at The
Ohio State University. She also benefited from a National Science Foundation grant to the Mershon
Center at Ohio State University for their Research Training Group on the role of cognition in collective
political decision making. We are indebted to Max Bazerman, Jeanne Brett, John Levine, Maggie
Neale, Randall Peterson, Scott Tindale, and three anonymous reviewers for feedback on earlier
versions of this manuscript.
Address correspondence and reprint requests to Deborah Gruenfeld, J. L. Kellogg Graduate School
of Management, Northwestern University, Evanston, IL 60208-2011.
202
0022-1031/98 $25.00
Copyright r 1998 by Academic Press
All rights of reproduction in any form reserved.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01 swei
No. of Pages—25 First page no.—202 Last page no.—226
203
COMPLEXITY IN GROUPS
that a shift in the target’s position occurs, evidence of influence is obtained. Over
time, the findings of such studies have become increasingly clear: targets of
majority influence tend to comply in public (Tanford & Penrod, 1984), but don’t
necessarily change their private views (Moscovici, Lage & Naffrechoux, 1969).
In contrast, targets of minority influence don’t usually conform in public, but are
often privately swayed, either toward the minority position (Moscovici, 1980;
Mugny, 1982; Wood, Lundgren, Ouellette, Busceme, Blackstone, 1994), or
toward a broader range of alternatives than they would have considered beforehand (Doise & Mugny, 1984; Nemeth, 1986; Smith, Tindale & Dugoni, 1996).
Although changes in targets’ opinions are the most commonly studied consequences of social influence, there are other effects as well. Messages conveyed by
majorities and minorities are also processed in different ways (for a review see
Moskowitz, 1996). For example, cognitive responses to majority messages are
generally more effortful (Mackie, 1987), and more focused (Nemeth, 1986), than
responses to minority messages. Targets of majority influence experience pressure
to conform, and tend to focus exclusively on the majority message as a result (De
Dreu & De Vries, 1993; Mackie, 1987; Nemeth & Wachtler, 1983; Trost, Maass &
Kenrick, 1992). Responses to minority messages, in contrast, are generally
characterized by greater cognitive flexibility than responses to majority messages
(Moscovici, 1980; Nemeth & Kwan, 1987; Peterson & Nemeth, 1996), and may
involve trying on multiple perspectives (Moscovici, 1980), searching for new
information (Nemeth & Rogers, 1996), and discovering new solutions (Nemeth &
Kwan, 1987; Van Dyne & Saavedra, 1996). As a result, whereas the impact of
majority influence is most likely to be direct and publicly visible, the changes
incited by minority influence are more likely to be private, and only indirectly
related to the content of the minority position (Wood et al., 1994).
This evidence indicates that the targets of minority influence generate more
ideas, and more classes of ideas, than the targets of majority influence (Moskowitz, 1996). However, recent findings suggest that majority and minority influence also differentially affect how targets reason about alternatives. For example,
Peterson and Nemeth (1996) showed that targets of both majority and minority
influence can learn to recognize multiple perspectives, but that the former are less
able to keep both in mind simultaneously than the latter. Hence, minority
influence might also facilitate integrative reasoning by majority members.
This possibility is supported by research on ‘‘integrative complexity,’’ which
reflects decision makers’ tendencies to exhibit (a) conceptual differentiation (i.e.,
recognition of multiple perspectives), and (b) conceptual integration (i.e., recognition of trade-offs among them). In several studies now, majority members have
exhibited greater integrative complexity than minority members in their public
statements (Gruenfeld, 1995). However, in contrast to the cognitive-flexibility
literature, which is based primarily on laboratory experiments, research on the
integrative complexity of majorities and minorities has been purely archival.
Specifically, differences in the integrative complexity of majorities and minorities
have been observed in the U.S. Congress (Tetlock, 1983; Tetlock, Hannum &
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
204
GRUENFELD, THOMAS-HUNT, AND KIM
Micheletti, 1984), the British House of Commons (Tetlock, 1984), the Senate
Budget Committee (Staub, 1991), and the U.S. Supreme Court (Gruenfeld, 1995;
Tetlock, Bernzweig & Gallant, 1985).
Although this finding is robust, its precise theoretical nature is not yet well
understood. In contrast to the subjects of minority influence experiments, who
generally receive information without interacting, members of policy making
groups are both the receivers and transmitters of persuasive appeals; they are both
targets and sources of influence. Hence, while the integrative complexity of their
public statements might reflect their private cognitive responses to others’
influence attempts, it might also reflect their public communication strategies for
attempting to influence others.
Indeed, there is considerable evidence that preparing to transmit messages can
have independent effects on how information is both privately organized and
publicly conveyed (for reviews see Wyer & Gruenfeld, 1994; 1995). For example,
research on ‘‘cognitive tuning’’ shows that when subjects prepare to receive
information from others they develop simple and loosely structured, hence more
flexible, organizing schemas for that information than when they prepare to
transmit information to others (for a review see Levine, Bogart & Zdaniuk, 1996).
Specifically, ‘‘transmitter’s’’ cognitive structures are both more highly differentiated (i.e., contain a greater number of more specific categories), and more
integrated (i.e., show greater interdependence among categories and greater
organization around a single unifying aspect) than those created by ‘‘receivers’’
(Zajonc, 1960).
In policy-making groups, message receptions and message transmissions occur
in tandem, and group members are likely to shift their focus from one to the other
during different phases of the process. During the predecision period, when policy
positions are being formulated, group members engage in dialogue that is
ostensibly orchestrated to facilitate learning about policy options and their
relative trade-offs. This should orient them toward receiving information. If group
members attempt to influence one another and build coalitions around their
preferred alternatives, they will also be focused on transmitting information.
However, once a single policy is formally endorsed, and group members are
committed to a particular outcome, the focus and tenor of these activities should
shift (Tetlock, Skitka & Boettger, 1989). During this postdecision period, policy
makers should be focused on transmitting information about the policy to external
constituents who will be affected by its consequences. Hence, relative to predecision activities, during which majority and minority members are internally
focused and interacting, postdecision activities should be characterized by a more
exclusive external focus on communication targets outside the group.
The postdecision messages designed to influence outsiders might differ substantially in both content and structure from the predecision policy cognitions of those
who transmit them. Such effects are routinely observed in studies of ‘‘accountability’’ (Tetlock, 1983), which show that decision makers who feel compelled to
justify their views to others often publicly acknowledge the validity of perspec-
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
205
COMPLEXITY IN GROUPS
tives that they do not actually believe are valid, in order to make a positive
impression on their audience (for reviews see Tetlock, 1991; 1992).
For example, the authors of Supreme Court opinions might couch their views in
strategic rhetoric that allows them to anticipate and address the potential legal,
social, economic or political consequences of their decision, even if those
consequences were not an explicit reason for supporting it during group deliberations. The authors of majority opinions might concern themselves with specifying
all imaginable contingencies under which the law should and should not apply to
insure the longevity of their precedent. The authors of minority opinions, in
contrast, might focus exclusively on arguments that could eventually facilitate the
precedent’s overruling. In this way, the responsibility of policy making might lead
to greater integrative complexity than the responsibility of policy opposition (cf.,
Tetlock, et al., 1985). If this were the case, the integrative complexity of published
majority and dissenting opinions might differ from that of the cognitions on which
justices’ policy preferences are based.
These considerations suggest that the integrative complexity differences between majority and minority members could be due to their mutual predecision
influence on one another’s private cognitions, or their unique postdecision
attempts to influence others through public communication strategies. Since prior
research on these factors has relied exclusively on data from the public record, it
has generally not been possible to distinguish between these psychological
alternatives (cf., Tetlock et al., 1984; Gruenfeld, 1995).
The experiment reported here was designed to examine these alternatives under
conditions in which their independent and joint effects could be evaluated.
Toward that end, majority and minority members of non-unanimous groups and
members of unanimous groups engaged in a judicial decision task and were asked
to explain their preferences under either public or private communication conditions. In both conditions, subjects were personally anonymous. However, under
public conditions, they were asked to consider the implications of their group’s
decision for an external audience. Under private conditions, they were asked to
respond as they would if they didn’t have to worry about how others would react
to their views.
Alternative Hypotheses
As noted above, the difference in majority and minority members’ integrative
complexity levels could occur via two psychological paths. The first path assumes
that the effects of majority and minority status on integrative complexity occur
prior to the group decision and that changes in cognitive flexibility occur after
group members receive one another’s influence attempts. The second path
assumes that the effects of majority and minority status on integrative complexity
occur after the group decision and reflect group members’ unique reactions to
their communication strategies for attempting to transmit the group’s outcome to
outsiders. Evidence and arguments in support of each path are discussed in turn.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
206
GRUENFELD, THOMAS-HUNT, AND KIM
Cognitive-flexibility hypothesis. Theoretically, there are several reasons that the
greater integrative complexity exhibited by majority members might reflect their
underlying beliefs. As noted earlier, minority influence incites open mindedness
among majority members. This is thought to occur because group members want
to be accurate on the one hand, and to reject the minority on the other (Wood et al.,
1994; Wood, Pool, Leck & Purvis, 1996). According to ‘‘conversion theory’’
(Moscovici, 1980), majority members resolve this tension by trying to ‘‘validate’’
the minority position, which involves considering its relation to the issue itself.
Attempts at validation lead majority members to discover new dimensions of the
problem and to understand the issue in new ways. Ultimately, these unintended
cognitive consequences render the majority more susceptible to the minority
position than they were beforehand, and leads to their private conversion (i.e.,
acceptance of the minority position).
Research on ‘‘divergent thinking’’ suggests that validation increases the majority’s tolerance for new alternatives in general, rather than the minority’s position
per se (Nemeth, 1986). This work shows that subjects exposed to minority
influence initially reject the minority proposal, and begin to search actively for
new alternatives (Nemeth & Rogers, 1996; Nemeth & Wachtler, 1983), performance strategies (Nemeth & Kwan, 1987), and more creative solutions (MucchiFaina, Maass & Volpato, 1989; Van Dyke & Saavedra, 1996). To the extent that
these activities increase awareness of multiple perspectives, they should increase
cognitive differentiation by majority members. In addition, the fact that minority
influence targets are better at keeping two perspectives in mind simultaneously
suggests that they should also be better at conceptual integration than majority
influence targets (Peterson & Nemeth, 1996). Thus, minority influence should
increase integrative complexity in majority members (Gruenfeld, 1995).
In contrast, majority influence should decrease integrative complexity in
minority members because it leads to an exclusive, ‘‘convergent’’ focus on the
majority position. Theoretically, this occurs because alignment with the majority
seems advantageous for both normative (social acceptance) and informational
(accuracy) reasons (Asch, 1951; Deutsch & Gerard, 1955). A convergent focus
precludes consideration of any alternatives, as well as continued exploration of
the issue (Nemeth, 1986), thereby reducing the potential for both differentiation
and integration.
It is important to note that majority members become cognitively flexible in
response to minority influence. Hence, it is not only the case that majority
members should exhibit higher integrative complexity than minority members,
they should also exhibit higher integrative complexity than members of unanimous groups who, in the absence of minority influence, lack the catalyst for
cognitive restructuring (Nemeth & Kwan, 1987). This pattern has already been
observed in a content analysis of Supreme Court opinions, which are prepared
specifically for the public record and are, by definition, written with certain
audiences in mind. In that study, majority opinions written when the Court was
not unanimous were more integratively complex than either minority (i.e.,
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
207
COMPLEXITY IN GROUPS
dissenting) opinions, or majority opinions written when the Court was unanimous
(Gruenfeld, 1995).
To the extent that cognitive flexibility is responsible for the complexity levels
exhibited by the authors of these documents, the same pattern of differences
obtained in these published legal documents should occur when group members
disclose their private views. That is, the integrative complexity exhibited by
representatives of the majority should be greater than that exhibited by representatives of the minority, or that by representatives of unanimous groups, whether
the authors are required to transmit their rationale for the group’s decision to an
external audience (under public communication conditions), or are asked to
informally disclose their thoughts (under private communication conditions).
Communication-strategy hypothesis. In contrast to the cognitive-flexibility
hypothesis, the communication-strategy hypothesis suggests that group members’
integrative complexity levels reflect their public communication strategies, rather
than their private thoughts. In this view, group members who are responsible for
making policies that will be implemented (e.g., the majority), and are therefore
responsible for explaining them to beneficiaries outside of the group, should
exhibit greater integrative complexity when explaining those policies to an
external audience than group members who personally opposed the adopted
policy (e.g., the minority) and are therefore not accountable for either its meaning
or its consequences.
As noted earlier, one reason this might occur is that policy makers in the
majority are accountable to the voters who elect them for the consequences of
policies that are actually implemented, and might display integrative complexity
in attempt to anticipate and appease their constituents’ concerns (Tetlock, 1981;
Tetlock et al., 1984). Accountability, defined as pressure to ‘‘maintain the positive
regard of important constituencies’’ (Tetlock, 1992; p. 332), emphasizes concerns
with social acceptance. However, majority and minority members are simultaneously susceptible to both normative and informational concerns (Deutusch &
Gerard, 1955). While normative concerns might outweigh informational concerns
in political bodies, the opposite is likely to be true in other kinds of groups.
This account was supported in a study of U.S. Senators (Tetlock et al., 1984).
However, while it provides a sensible explanation for the behavior of elected
officials, it is unclear whether accountability to constituents per se is the key
driver for majority and minority influence sources more generally. Supreme Court
justices for example, who are appointed for life by the President and are
purposefully insulated from public opinion (Baum, 1989; Brenner & Spaeth,
1995), are decidedly not accountable in this purely political manner. In this type
of group, accountability to external constituents is likely to invoke the informational pressures associated with policy making more generally, such as constructing a policy statement that accurately and effectively conveys the nature of new
rules and the conditions under which they should be applied. On the Court, as in
many policy groups, majority members, who are responsible for the consequences
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
208
GRUENFELD, THOMAS-HUNT, AND KIM
of their policies on those affected by them, are also responsible for providing this
kind of guidance.
To the extent that this leads to more carefully and thoroughly articulated
postdecision messages, it should also be associated with greater integrative
complexity among majority members than minority members. This argument is
consistent with Zajonc’s (1960) finding that message ‘‘transmitters,’’ who expected to explain things to others, constructed more highly differentiated and
integrated organizing structures for the information they had to convey than
message ‘‘receivers,’’ who expected others to explain things to them.
However, contrary to the predictions based on cognitive flexibility arguments,
this integrative complexity ‘‘advantage’’ should occur whether the majority
represents a group that is unanimous, or one that is non-unanimous. The presence
of a vocal minority is not necessary for postdecisional accountability to have the
predicted effects. Rather, the transmitter orientation described above depends on
(a) the presence of an external constituency, and (b) a sense of responsibility for
group outcome. Majority and unanimous group members should be equally
susceptible to these psychological states.
Furthermore, when considering the implications of this hypothesis, it is
important to note that public communications often occur independently of
decision makers’ private belief structures (Moscovici, Lage & Naffrechoux, 1969;
Tetlock & Kim, 1987). For example, when accountability is manipulated after
decision makers’ are committed to a particular outcome, subjects typically design
their public statements to maximize audience acceptance, but remain true to their
initial beliefs when asked to disclose them afterward in private (for a review see
Tetlock, 1992). Such behaviors should not be observed when subjects express
their views under private communication conditions in which no identifiable
audience exists.
Correspondingly, in the groups studied here, integrative complexity is expected
to be higher among majority members and people in unanimous groups, who are
accountable for the group’s decision, than among minority members, who are not.
However, to the extent that the effect is driven by postdecisional accountability
and communication strategies alone, differences between majority and unanimous
group members on the one hand, and minority members on the other, should only
appear under public communication conditions. Under private communication
conditions, transmitters of majority and unanimous group messages should not
adjust their integrative complexity levels in response to the pressure of external
accountability.
Additive effects hypothesis. A third theoretical possibility is that the effects of
cognitive flexibility on integrative complexity and the effects of communication
strategy on integrative complexity are additive. For example, it might be the case
that the cognitive changes in response to majority and minority influence occur
prior to strategic adjustments in the integrative complexity of public communication.
If both processes are operant, the following results can be expected. Under
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
209
COMPLEXITY IN GROUPS
private communication conditions, the cognitive-flexibility hypothesis should
receive support. Specifically, majority members should express their views with
greater integrative complexity than either minority members or members of
unanimous groups. Under public communication conditions, however, the communication-strategy hypothesis should be supported. That is, majority and unanimous group members, who are equally accountable for the consequences of the
group’s decision, should exhibit greater integrative complexity than minority
members, who are less likely to feel accountable for those consequences.
Study Overview
These alternatives were examined under experimentally controlled conditions.
Subjects in a laboratory experiment reported their individual preference for the
outcome of a hypothetical legal case. This preference was used to assign the
subjects to either unanimous or non-unanimous decision-making groups. Participants wrote rationales for their preferences both (a) before they were informed of
their group membership, and (b) after the group task. In addition, two sets of
rationale-writing instructions were administered. Subjects under private communication conditions were assured that their decision rationales would be anonymous and were told that the study’s success depended on their candor. Subjects
under public communication conditions were urged to consider the impact of their
arguments on hypothetical others who might draw on them as a basis for future
decisions. The rationales were coded for integrative complexity.
METHOD
Overview
Participants reported their opinion concerning the policy of busing to achieve racial desegregation
in public schools. They also wrote a rationale for this preference. Their position (pro vs. anti) was then
used as a basis for assigning them to small decision-making groups that were either (a) unanimous, in
favor of the busing policy, (b) unanimous, opposed to the busing policy, (c) non-unanimous, with a
pro-busing majority and anti-busing minority, or (d) non-unanimous, with an anti-busing majority and
pro-busing minority.
Group members were informed of one another’s positions spontaneously, through free-form
interaction, and were asked to report the group’s final decision using a majority-wins rule. Afterward,
each member wrote a second (post-decision) rationale for his or her position, taking into account what
had transpired during the group interaction. Half of those rationales were written under public
communication conditions and half were written under private communication conditions. A postdiscussion questionnaire was then completed.
Rationales were coded for integrative complexity. The post-task questionnaire included measures of
political attitudes and expertise, and attitudes toward busing.
Subjects and Design
Subjects were 280 introductory psychology students at a large midwestern university who
participated in the study for course credit. They were assigned to one of 16 conditions in a 2 (public vs.
private communication condition) 3 2 (unanimous vs. non-unanimous distribution) 3 2 (majority
preference to affirm vs. overturn) 3 2 (male vs. female sex) design. They were divided into 58 groups
containing 3 to 7 members of the same sex that varied in terms of the distribution of members’ initial
preferences.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
210
GRUENFELD, THOMAS-HUNT, AND KIM
Procedure
Instructions to individuals. Subjects were given an identification number when they entered the
laboratory. When all subjects had arrived, they were told that the purpose of the study was to examine
the psychological underpinnings of judicial decision making. They were instructed in single-sex
groups of 18–20.
Next, a brief description of a controversial (hypothetical) legal decision was distributed. (The topic
was chosen on the basis of pre-testing because it invoked a wide range of opinions among college
undergraduates). The ruling described a Supreme Court decision that made busing mandatory for the
purposes of racial desegregation where no alternative means was available. Subjects were asked to
consider the ruling, and were given a sheet of paper asking for their ‘‘private’’ reactions to it.
The instructions (adapted from Tetlock et al, 1989) said:
Please describe your reactions to what you just read—that is, your thoughts and feelings
about the case—in the space provided. In order for this project to succeed, we need to know
what you think when you don’t have to worry about how other people will react to your
views. Therefore, please respond with complete candor and honesty. To ensure complete
confidentiality, write your subject number at the top of the page, not your name. Your data
will be analyzed by another group of researchers who are not present today, and who
specialize in analyzing these types of free response materials. Please write down all the
thoughts and feelings you have about the issue.
After they had completed this private rationale, subjects were then given a judgment form with the
following question at the top of the page:
If you were a member of the Supreme Court today, and you had the opportunity either to
re-affirm or overturn the ruling that made busing mandatory in segregated school districts,
how would you vote?
Subjects responded by checking either ‘‘re-affirm’’ or ‘‘overturn.’’ This preference was used as a basis
for constructing decision-making groups with the desired distributions. While group assignments were
being made in another room, participants completed a measure of general knowledge about the
Supreme Court, and reported their attitudes toward and personal experience with busing as a means of
desegregation.
Instructions to groups. All of these activities occurred prior to group assignments, and in the
absence of any indication that group interaction would be involved. After individual judgments about
the ruling had been made, subjects were told that they would participate in a group task to simulate
how judicial bodies such as the Supreme Court reach a collective decision. Group assignments were
made by reading off subject identification numbers. Non-unanimous groups contained 4–7 members,
so that meaningful majority and minority factions could be constructed. Unanimous groups contained
3–5 members so that they could be compared directly to majorities in non-unanimous groups.
Each group was directed to a private room and its members seated around a table. Subjects were
reminded that they would have 10 minutes to reach a decision, and to record this decision on a ‘‘group
judgment’’ form. They were advised that only a majority was necessary to reach a final decision, and
that the identification numbers of any dissenters should be recorded on the group judgment form.
Groups were told that if they finished early, they could knock on the door to signal the experimenter. A
timer was set and placed on the table, and the group was left to deliberate. At the end of 10 minutes (or
when the experimenter was signaled), the experimenter returned to collect the ‘‘group judgment’’
form.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
211
COMPLEXITY IN GROUPS
Group members were then given instructions for completing their ‘‘final rationale.’’ Groups under
private communication conditions were instructed as follows:
To help us better understand the effects of group interaction on private thoughts about
judicial issues, we would now like you to write a brief explanation of your personal thoughts
and feelings about the case, taking into account what was discussed in your group. To do
this, imagine that you are a Supreme Court justice and are alone in your chambers, reflecting
about a discussion session that has just concluded. If you agree with the majority of group
members, describe your private thoughts about why you agree. If you disagree with the
majority of group members, describe your private thoughts about why you disagree. Again,
we need to know what you think when you don’t have to worry about how other people will
react to your views, so please respond with complete candor and honesty, and provide
justification for your views. Your responses will be completely confidential. When you are
finished writing, place your form in the envelope provided by the experimenter.
Subjects in public communication conditions received identical instructions, except for the following
change (adapted from Tetlock, et al, 1989):
. . . imagine that you are a Supreme Court justice and have been asked to prepare this
rationale for the public record. If you agree with the majority of group members, write the
equivalent of a ‘‘majority’’ opinion, explaining the reasoning behind the majority view. If
you disagree with the majority of group members, write the equivalent of a ‘‘minority’’
opinion, explaining the reasons that you believe the majority view is not satisfactory.
Remember that your task is to write as convincing a rationale as you can, of a sort that would
go into the public record as the best case possible for the view you represent. Please respond
honestly, and provide justification for your views. . . .
Content Analyses
Each rationale was scored for integrative complexity by two independent, trained raters who were
unaware of the nature of hypotheses and the experimental conditions under which the rationales were
generated. Six groups were selected for reliability coding: one from each combination of member
status (unanimity, majority and minority) and communication condition (public and private). Interrater agreement (Kendall’s W) was .85 ( p , .002). Discrepancies were resolved by an expert rater
who was aware of the hypotheses and independent variables but was blind to the condition in which
the materials being evaluated were generated.
Integrative complexity. The integrative complexity coding system (Suedfeld et al, 1992) was used to
infer levels of differentiation (recognition of multiple dimensions or aspects of an issue) and
integration (recognition of conceptual relations among differentiated dimensions). The following
illustrations of different integrative complexity levels are paraphrased subject responses.
Score of 1 (no differentiation, no integration). Busing students to an environment that is
unfamiliar to them away from their friends and own culture is unfair, no matter what greater
purpose it is meant to serve. People should not have to switch schools against their will. The
consequences could be worse than staying in an inferior school.
Score of 3 (differentiation, no integration). Busing students out of their own community is a
heavy burden for the students to bear. The alternative—lack of equal opportunity—is also
unconscionable. I can see why the Supreme Court decided the way they did, but I’m sure the
people affected didn’t like the idea.
Score of 5 (differentiation, simple integration). Busing students from deprived school
districts into richer ones may be great for the deprived, but it usually brings down the quality
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
212
GRUENFELD, THOMAS-HUNT, AND KIM
TABLE 1
DISTRIBUTION OF NON-UNANIMOUS GROUPS IN WHICH STATUS-CONDITIONS WERE ALTERED
Majority preference
Affirm
Public Condition
Male
Female
Private Condition
Male
Female
Overturn
PC
TC
CC
PC
TC
CC
2
1
0
1
0
0
1
1
2
0
1
1
0
1
1
0
1
1
0
1
1
2
0
0
Note. PC, partial conformity; TC, total conformity; CC, conformity 1 conversion.
of the better school. Choosing which way to vote requires deciding whether equality or
excellence is a more important goal.
Score of 7 (differentiation, complex integration). There are both advantages and disadvantages associated with the busing policy, and they vary depending on whose perspective you
take. Busing was a good short-term solution from a political standpoint, but it had harsh
consequences for the people affected. In the long view, those sacrifices should be weighed
against the social, economic and legal consequences of desegregation.
Scores of 2, 4, and 6 represent transition points between adjacent levels. They are assigned
when there is evidence of implicit differentiation (information seeking, qualification to an
absolute rule) or implicit integration (hints of recognition of interactions and trade-offs).
RESULTS
The responses of individual members within each group were averaged to
provide one set of scores for each of the 20 unanimous groups and two sets of
scores (one for majorities, one for minorities) for each of the 20 non-unanimous
groups retained for primary analyses. This resulted in a total of 60 data points per
dependent variable. Three sets of analyses were performed in which minority vs.
majority comparisons were treated as within-group, and comparisons of each of
these conditions with unanimous group scores were treated as between-group.
Preliminary Analyses
Prior to statistical analyses, examination of the data revealed that in 18 of the
non-unanimous groups, individual members changed their preferences during
group discussion, thereby altering the group’s status conditions. The distribution
of these groups across cells of the experimental design is displayed in Table 1.
Seven groups experienced an increase in majority size and a decrease in minority
size due to partial conformity (PC) by some but not all minority members.
Another seven groups were initially non-unanimous but became unanimous as a
result of total conformity (TC) by all minority members. In four additional groups,
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
213
COMPLEXITY IN GROUPS
both conformity by minorities plus conversion by majorities (CC) occurred, such
that faction membership changed but faction size was not affected.
Data for the 18 changed groups were examined separately for possible
differences from included groups. The pre- versus post-task integrative complexity difference scores of majorities and minorities who experienced total conformity, partial conformity and conformity 1 conversion did not differ significantly
from that displayed by majorities and minorities in the groups retained.1 Since
small sample size precluded statistical analyses of the effects of different types of
change, and those changes were not central to the theoretical objectives of this
study, these groups were excluded from hypothesis tests (see ‘‘Supplementary
Analyses’’ for a description of how inclusion of these groups affects overall
results).
One hundred eighty nine subjects participated in the 40 groups retained for this
purpose. The final sample distribution is shown in Table 2. The size of unanimous
groups (M 5 3.9) was significantly smaller than the size of non-unanimous
groups (M 5 5.5; F(1, 24) 5 29.87, p , .0001). However, this was planned so
that majority size would not be confounded with unanimous and non-unanimous
distribution conditions. Consistent with this objective, majorities in unanimous
groups were no larger (M 5 3.9) than majorities in non-unanimous groups
(M 5 3.8; F(1, 24) 5 .15, p . .10). Minority size did not differ across conditions
(Fs , 1.0, ns). There were some unintentional size differences in the pro-busing
versus anti-busing majority-preference conditions. Group size in the pro-busing
condition (M 5 5.0) was greater than group size in the anti-busing condition
(M 5 4.4; F(1, 24) 5 4.47, p , .05), because majority size was greater in the
pro-busing (M 5 4.2) than in the anti-busing condition (M 5 3.5; F(1, 24) 5 5.20,
p , .03).
To insure that none of the random factors (integrative complexity, attitudes
toward busing, experience with busing, political activities, knowledge of Supreme Court justices, or group size) fortuitously varied as a function of control
variables (sex and majority preference) or experimental manipulations (majority
versus minority status and communication condition), two sets of multivariate
analyses were performed. First, the data from non-unanimous groups were
analyzed as a function of majority and minority status using a 2 (majority
preference) 3 2 (communication condition) 3 2 (sex) 3 12 repeated measures (6
random variables 3 2 status types) multivariate analysis of variance (MANOVA).
1 Although small sample sizes preclude drawing conclusions from these data, findings show that
minorities experienced a decrease in integrative complexity under all conditions (Ms 5 2.35, 2.30
and 2.14 in no-change, total-conformity and partial-conformity conditions, respectively) except for
the rare condition in which both conformity by minorities and conversion by majorities occurred. In
such cases, minorities exhibited an increase in integrative complexity (M 5 1.00). The integrative
complexity of majority members increased under all conditions (Ms 5 .23, .38 and .17 in no-change,
partial-conformity and conformity 1 conversion conditions, respectively), except when total conformity occurred. When all minority members conformed, making non-unanimous majorities unanimous,
majorities exhibited a decrease in integrative complexity (M 5 2.11).
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
214
GRUENFELD, THOMAS-HUNT, AND KIM
TABLE 2
NUMBER OF GROUPS PER EXPERIMENTAL CONDITION AND AVERAGE FACTION SIZE (N) A
Group distribution
Non-unanimous
Communication Condition
Public
Male
Maj size
Min size
Female
Maj size
Min size
Private
Male
Maj size
Min size
Female
Maj size
Min size
Unanimous
Affirm
Overturn
Affirm
Overturn
3
(4.3)
(1.7)
2
(4.0)
(1.0)
2
(3.0)
(1.5)
3
(3.7)
(1.3)
3
(4.0)
—
2
(4.5)
—
2
(3.0)
—
3
(4.3)
—
3
(4.7)
(2.0)
2
(4.0)
(2.0)
2
(3.5)
(1.5)
3
(3.3)
(1.7)
3
(3.3)
—
2
(4.5)
—
2
(4.0)
—
3
(3.7)
—
a Numbers in parentheses correspond to the average number of members per designated faction.
Note that non-unanimous group size corresponds to the sum of majority and minority sizes.
Unanimous group size corresponds to ‘‘Maj size,’’ since minority factions do not exist within
unanimous groups.
Second, for between-group analyses, the responses of unanimous groups were
compared with those of majority and minority non-unanimous members separately, using a 2 (majority preference) 3 2 (communication condition) 3 2
(sex) 3 2 (unanimous vs. non-unanimous distribution) 3 6 (repeated measures)
MANOVA each time.
None of the factors were confounded with the variables of primary concern in
the study. It is particularly important to note that subjects assigned to majorities,
minorities and unanimous groups did not differ in integrative complexity prior to
participating in the group task (all multivariate F’s , 1.0, ns). Since preliminary
analyses revealed no significant differences between all-male and all-female
groups, nor between groups whose majorities favored affirming versus overturning the busing policy, the sex and majority-preference variables were collapsed
over for the purpose of hypothesis testing.
Effects of Distribution, Status, and Communication Conditions
Data pertaining to differences between majority and minority factions in
non-unanimous groups were analyzed using a 2 (communication condition) 3 2
(status) 3 2 (time) MANOVA with one between-group factor (communication
condition) and two within-group factors (status and time of assessment: prediscussion vs. post-discussion). Data pertaining to differences between members
of unanimous groups on the one hand, and (a) majority factions in non-unanimous
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
215
COMPLEXITY IN GROUPS
TABLE 3
MEAN INTEGRATIVE COMPLEXITY AS A FUNCTION OF GROUP DISTRIBUTION,
COMMUNICATION CONDITION, MEMBER STATUS, AND TIME
Member status
Pre Group Task
Private
Public
M
Post Group Task
Private
Public
M
Change
Private
Public
M
Unanimity
Majority
Minority
1.78 (n 5 10)
1.50 (n 5 10)
1.64 (n 5 20)
1.66 (n 5 10)
1.46 (n 5 10)
1.56 (n 5 20)
2.03 (n 5 10)
1.90 (n 5 10)
1.97 (n 5 20)
1.49 (n 5 10)
1.39 (n 5 10)
1.44 (n 5 20)
1.87 (n 5 10)
1.71 (n 5 10)
1.79 (n 5 20)
1.78 (n 5 10)
1.45 (n 5 10)
1.62 (n 5 20)
2.29 (n 5 10)
2.11 (n 5 10)
2.20 (n 5 20)
.21 (n 5 10)
.24 (n 5 10)
.23 (n 5 20)
2.25 (n 5 10)
2.45 (n 5 10)
2.35 (n 5 20)
groups or (b) minority members in non-unanimous groups on the other, were
analyzed using a 2 (communication condition) 3 2 (distribution) 3 2 (time)
repeated measures multivariate analysis of variance (MANOVA) with two betweengroup factors (communication condition and distribution) and one within-group
factor (time).
Since hypotheses pertain to how participating in the group task changes
integrative complexity levels, pre- versus post-group task difference scores are
reported unless otherwise indicated. Results bearing on each hypothesis are
presented in turn.
Cognitive-flexibility hypothesis. After non-unanimous groups participated in
the group task, majority members were expected to exhibit an increase in
integrative complexity, and minority members were expected to exhibit a decrease in integrative complexity, regardless of the communication conditions
under which rationales were generated. Within non-unanimous groups, this
pattern corresponds to a two-way (status 3 time) interaction. In addition, majority members of non-unanimous groups were expected to exhibit higher integrative
complexity than members of unanimous groups. For between-group comparisons,
this prediction corresponds to a two-way (distribution 3 time) interaction involving majorities in unanimous versus non-unanimous groups.
The data in Table 3 indicate that this pattern was obtained. As expected,
participation in the group task led to an increase in integrative complexity among
majority members (M 5 .23), and a decrease in complexity among minority
members (M 5 2.35) in non-unanimous groups. Members of unanimous groups
also exhibited decreased integrative complexity (M 5 2.20). Consistent with
predictions, both the time 3 status interaction for majority versus minority
members within non-unanimous groups (F(1, 18) 5 4.99, p , .04), and the
time 3 distribution interaction for majorities in unanimous versus non-unanimous
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
216
GRUENFELD, THOMAS-HUNT, AND KIM
groups (F(1, 36) 5 4.78, p , .04), were statistically significant. In addition,
comparisons of minorities and unanimous group members revealed a main effect
for time (F(1, 36) 5 4.34, p , .04), indicating that both groups exhibited a
decrease in complexity after the group decision. The decreases in complexity
experienced by minority and unanimous-group members did not differ (F , 1.0,
ns). Furthermore, communication conditions did not enter into any significant
main or interactive effects in any of the analyses performed (F’s , 1.0, ns).
Communication-strategy hypothesis. According to this hypothesis, group members who felt accountable for the consequences of the policy adopted by the group
(i.e., majority and unanimous group members) were expected to exhibit higher
integrative complexity than members who did not feel personally responsible for
this outcome. Hence, majority and unanimous group members were expected to
communicate with greater integrative complexity than minority members, but
only under public communication conditions.
To the extent that these predictions are supported, the MANOVA should reveal
a three-way (status 3 time 3 communication-condition) interaction for comparisons within non-unanimous groups. Between-group comparisons should be
characterized by (a) a two-way (time 3 communication-condition) interaction
involving majorities in unanimous and non-unanimous groups, and (b) a threeway (distribution 3 time 3 communication-condition) interaction involving unanimous groups and minority members.
These predictions were not supported (see Table 3). Contrary to expectations,
majority and unanimous group members did not respond similarly under public
communication conditions. Whereas the complexity exhibited by majority members increased (M 5 .24), the complexity exhibited by unanimous group members
decreased (M 5 2.11) under public communication conditions (F(1, 18) 5 6.38,
p , .03), replicating the pattern observed under private conditions in which
communication strategies were not invoked. Furthermore, as noted in the previous section, majorities exhibited increased integrative complexity, and minorities
exhibited decreased integrative complexity under private, as well as public
communication conditions. Consequently, neither of the predicted interaction
effects were obtained because the impact of status and distribution on group
members’ publicly communicated integrative complexity levels was statistically
indistinguishable from their privately communicated levels. Specifically, integrative complexity increased for majorities under both private (M 5 .21) and public
conditions (M 5 .24), while it decreased for minorities under both private
(M 5 2.25) and public conditions (M 5 2.45; F(1, 18) 5 .21, p . .10). Unanimous group members, who were expected to increase their complexity under
public conditions, decreased their complexity in both public (M 5 2.11) and
private communication conditions (M 5 2.29). Hence the predicted two-way
(time 3 communication-condition) interaction involving majorities in unanimous
and non-unanimous groups was not significant (F(1, 36) 5 .25, p . .10). Finally,
since the integrative complexity of minority members also decreased under
private (M 5 2.25) and public conditions (M 5 2.45), the predicted three-way
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
217
COMPLEXITY IN GROUPS
(distribution 3 time 3 communication-condition) interaction involving unanimous groups and minority members was also non-significant (F(1, 36) 5 .47,
p . .10). The communication-strategy hypothesis is not supported by these
results because neither of the predicted communication-condition interactions
was obtained, and the hypothesis provides no basis for predicting either complexity differences between majority and unanimous group members, or any effects of
status on integrative complexity under private communication conditions.
Additive-effects hypothesis. If both cognitive-flexibility and communicationstrategy effects on integrative complexity occurred, the following pattern of
integrative complexity changes was expected. Under private conditions, majority
members were expected to exhibit higher integrative complexity than both
members of minorities and of unanimous groups. Under public communication
conditions, majority and unanimous group members were expected to exhibit
greater complexity than minority members, but were not expected to differ from
one another in their complexity levels. These predictions would be supported by
the following combination of MANOVA effects: (1) for within-non-unanimous
group comparisons, a two-way (status 3 time) interaction, (2) for between-group
comparisons, a pair of three-way (distribution 3 time 3 communication condition) interactions in which (a) integrative complexity increases for both majorities
and unanimous group members under public conditions, but for majorities alone
under private communication conditions, and (b) integrative complexity does not
increase for either minority or unanimous group members under private conditions but increases under public communication conditions for unanimous group
members.
Although majorities exhibited greater complexity (M 5 .21) than minority
(M 5 2.25) or unanimous group members (M 5 2.29) under private communication conditions, the identical pattern was obtained under public communication
conditions (Ms 5 .24, 245, and 2.11 for majorities, minorities, and unanimous
groups, respectively). Hence, although the a two-way (status 3 time) interaction
for non-unanimous group members was obtained (F(1, 18) 5 4.99, p , .04),
neither the three-way (distribution 3 time 3 communication condition) interaction for majority and unanimous group members (F(1, 36) 5 .10, p . .10), nor
the three-way (distribution 3 time 3 communication) interaction for minority
and unanimous group members (F(1, 36) 5 .47, p . .10) was statistically significant.
Supplementary Analyses
As noted earlier, there were 18 groups in this study in which group members
switched majority and minority status conditions, and/or unanimous and nonunanimous group distribution conditions, after group discussion. Because the
nature of these changes made the effects of assigned experimental conditions
difficult to interpret, and also precluded statistical testing of interaction effects,
these groups were excluded from the hypothesis tests reported above. To determine whether and how the results of those tests might be effected by inclusion of
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
218
GRUENFELD, THOMAS-HUNT, AND KIM
TABLE 4
MEAN INTEGRATIVE COMPLEXITY AS A FUNCTION OF GROUP DISTRIBUTION COMMUNICATION
CONDITION, MEMBER STATUS AND TIME, WHEN CHANGED GROUPS ARE INCLUDED
Member status
Pre Group Task
Private
Public
M
Post Group Task
Private
Public
M
Change
Private
Public
M
Unanimity
Majority
Minority
1.78 (n 5 10)
1.50 (n 5 10)
1.64 (n 5 20)
1.64 (n 5 17)
1.49 (n 5 21)
1.56 (n 5 38)
2.01 (n 5 17)
1.83 (n 5 21)
1.91 (n 5 38)
1.49 (n 5 10)
1.39 (n 5 10)
1.44 (n 5 20)
1.78 (n 5 17)
1.75 (n 5 21)
1.76 (n 5 38)
1.90 (n 5 17)
1.64 (n 5 21)
1.76 (n 5 38)
2.29 (n 5 10)
2.11 (n 5 10)
2.20 (n 5 20)
.14 (n 5 17)
.26 (n 5 21)
.21 (n 5 38)
2.11 (n 5 17)
2.19 (n 5 21)
2.15 (n 5 38)
these groups, the effects of status, group distribution and communication on
integrative complexity were re-assessed with the 18 changed groups included.
When the changed groups were included, the pattern of means is identical to
that obtained without them (see Table 4). That is, majorities exhibited an increase
in integrative complexity (M 5 .21), while minorities exhibited a decrease in
complexity (M 5 2.15) under both public and private communication conditions
(Fs , 1.0 for all communication-condition effects). However, when the 18
changed groups were included, the results of hypothesis tests were altered in two
ways.
First, collapsing over time and communication-condition, minority members
exhibited greater integrative complexity (M 5 1.84) than unanimous group members (M 5 1.54) when the changed groups were included. This group-distribution
main effect (F(1, 54) 5 4.24, p , .04) was not predicted, but it does not affect
hypothesis tests directly. Furthermore, since integrative complexity decreased
among minority members of both changed (M 5 2.15) and unchanged groups
(M 5 2.35), the results of hypothesis tests comparing minorities and unanimous
group members were not altered.
Second, although complexity increased in majorities and decreased in minorities whether the 18 changed groups were included or not, this predicted interaction effect was only marginally significant with the changed groups included
(F(1, 36) 5 3.34, p , .07). When the changed groups were excluded, this
interaction effect was significant at the p , .04 level. The time 3 distribution
interaction effect for majorities in unanimous and non-unanimous groups remained statistically significant (F(1, 54) 5 5.57, p , .02), since the increase in
complexity exhibited by majorities was virtually unaltered when changed groups
were included (Ms 5 .23 and .21 for analyses excluding and including changed
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
219
COMPLEXITY IN GROUPS
groups, respectively). There were no other significant differences in the findings
of hypothesis tests as a consequence of including these 18 groups.
These findings suggest that the primary difference between changed and
unchanged groups lies in the behavior of minority members. Comparison of
means in Table 3 and Table 4 shows that minority members who conformed
exhibited greater integrative complexity after group interaction (M 5 1.76) than
minority members who did not conform (M 5 1.62). Note that prior to status
assignments, the integrative complexity of subjects who would be assigned to
minority status conditions was not altered by including subjects who would later
conform (Ms 5 1.91 and 1.97 for samples including and excluding these subjects,
respectively).
DISCUSSION
The behavior of the freely interacting groups in this study replicates the
findings of archival research on the relation of member status to integrative
complexity in decision-making groups. As observed previously, majority members exhibited greater integrative complexity than members of minorities (Tetlock
et al., 1984; 1985), and members of unanimous groups (Gruenfeld, 1995). After
participating in group discussion, subjects assigned to majority factions experienced an increase in integrative complexity, while subjects assigned to either
minorities or unanimous groups experienced a decrease in integrative complexity.
This occurred under both public and private communication conditions.
This pattern of results calls into question the notion that the effects of member
status on integrative complexity are purely strategic. While the difference between majority and minority members is predictable on the basis of accountability
to external parties, the difference between majorities in unanimous and nonunanimous groups is not. Furthermore, the increase in complexity experienced by
majorities and the decreases experienced by both minorities and unanimous
groups were obtained both under public communication conditions, where
accountability concerns were emphasized, and under private communication
conditions, where they were absent.
These findings can be explained however by the minority influence literature,
which suggests that the increase in integrative complexity exhibited by majorities
in non-unanimous groups is a consequence of minority influence and its impact on
cognitive flexibility. Majority members’ communications should therefore be
characterized by an increase in complexity after exposure to a vocal minority,
whether those communications are delivered under public or private communication conditions. In contrast, integrative complexity should not increase among
unanimous group members, who lack the minority catalyst for either conversion
or divergent thinking, nor among minority members, who should experience
convergent thinking in response to majority influence. In the experiment reported
here, this pattern was obtained under private communication conditions as well as
in formal public statements.
The results of this study address two additional questions raised by archival
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
220
GRUENFELD, THOMAS-HUNT, AND KIM
studies of member status and integrative complexity in policy making groups
such as the U.S. Supreme Court (Gruenfeld, 1995; Tetlock, et al., 1985) and the
U.S. Congress (Tetlock et al., 1984). In these ‘‘real’’ decision-making groups,
member status was chosen by the policy makers who served as subjects. Hence,
the integrative complexity levels of majorities, minorities and unanimities might
have reflected the characteristics of individuals who chose majority versus
minority status, rather than effects of faction membership per se. Subjects in the
laboratory experiment reported here were assigned to status conditions using a
procedure that controlled for the effects of differences in their preferences,
complexity levels, and other personality variables. The same pattern of results
was obtained under these conditions, in which status was assigned, as when status
was chosen in the ‘‘real’’ world.
Status was confounded with two additional factors in the archival studies
mentioned earlier. The effects of both factors were evaluated in the study
presented here. Specifically, the finding that majority members exhibited greater
complexity than either minority members or unanimous group members was first
obtained in a study of the U.S. Supreme Court (Gruenfeld, 1995). This effect
could have been due to (a) the inherent complexity of cases on which the Court
could not agree, and (b) inherent structural differences in the documents written
by majorities and minorities on the Court.
In the present study, members of non-unanimous and unanimous groups in the
laboratory experiment all decided the same case (e.g., whether busing should be
mandatory for the purposes of racial desegregation). Hence, the potential impact
of case content on authors’ complexity levels was eliminated. In addition, group
members were given identical instructions about what to include in their rationales, regardless of which faction they represented. Differences that might have
been associated with the structural requirements of formal ‘‘opinions of the
Court’’ written by majority members, and ‘‘dissenting opinions’’ written by
minority members, were thereby also controlled. The results of the present study
indicate that the effects of member status on integrative complexity in decisionmaking groups occur in the absence of these confounding factors.
Implications for Future Research
As noted above, the fact that spontaneous minority influence exerted during
free discussion led to increased integrative complexity in majority members is
consistent with prior studies of policy making groups where both minority
influence and free discussion are natural components of group interaction. It is
also consistent with experimental research on divergent thinking, which shows
that majorities exhibit increased cognitive flexibility in response to minority
influence (Peterson & Nemeth, 1986). Evidence for this effect is copious and has
established that it is a robust psychological phenomenon (see Nemeth, 1992, for a
review). However, the effects of minority influence on divergent thinking have
only rarely been observed in freely interacting laboratory groups (for an exception
see Smith, Tindale & Dugoni, 1996). Studies of divergent thinking typically use
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
221
COMPLEXITY IN GROUPS
confederate minorities, and often employ nominal groups composed of individuals who work independently, side-by-side, and receive false feedback about one
another’s responses (for a review see Nemeth, 1992). Yet the findings obtained
here are consistent with those observed in more controlled interactions.
Similarly, the fact that majorities exhibited greater integrative complexity than
minorities in this study replicates research on accountability in policy making
groups (Tetlock et al., 1984), which shows that accountability to constituents
outside group boundaries leads to integrative complexity among majority (rather
than minority) members. However, the findings reported here are not consistent
with prior lab research on this phenomenon. Laboratory research on how
accountability affects integrative complexity typically employs subjects who
form judgments as individuals, and anticipate interacting with others but never
actually do (for a review see Tetlock, 1991). In this study, group members
communicated freely, and were interdependent with regard to their task. Therefore, in addition to whatever inter-group accountability conditions group members faced as a result of experimental manipulations, they also faced intra-group
accountability conditions, as they were mutually accountable to one another and
to the group itself.
Research on cognitive responses to multiple accountabilities is rare, and its
findings are inconclusive (Thomas-Hunt, 1996; Thomas-Hunt & Gruenfeld,
1995). However, members of decision-making groups are generally more attentive to the interests of those parties with whom their relationship ties are strong
than they are to the parties from whom they are more psychologically distant
(Hogg & Abrams, 1988; Pruitt, 1981; Roby, 1960). This implies that internal
accountability among freely interacting group members might reduce the impact
of external accountability to hypothetical outsiders with whom subjects never
actually interact. To the extent that this occurred here, it might explain why no
accountability effects were observed.
It is also possible that the public communication condition in this study induced
cognitive processes associated with message transmission to a greater extent than
accountability per se. That is, subjects in the public communication condition
were asked simply to imagine they had to prepare a rationale for the public record,
and to provide the best explanation they could for that purpose. In contrast, most
accountability manipulations lead subjects to believe that they will actually have
to explain themselves to another person later on (cf., Tetlock, 1983; Tetlock,
Skitka & Boettger, 1989). Thus, although subjects in public communication
conditions were given ‘‘transmission’’ instructions, they were not really accountable to any outsiders. Rather, groups in both public and private communication
groups actually discussed their positions, and were therefore accountable to one
another. As noted earlier, this manipulation was designed to replicate findings
from a Supreme Court context in which group members are appointed for life and
are therefore immune to many types of political accountability. However, it can be
argued that it does not provide a strong test of accountability. To the extent that
this is true, the absence of public communication condition effects should not be
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
222
GRUENFELD, THOMAS-HUNT, AND KIM
interpreted as evidence that accountability does not drive integrative complexity
in majorities in minorities.
This possibility has interesting implications for future research on accountability, particularly when moving from the individual- to the group-level of analysis.
In groups where members interact freely with one another and only imagine or
anticipate interacting with outsiders, internal accountability is likely to make
collective interests such as group performance, harmony and fairness more salient
than individual concerns with external communication conditions. If this were the
case, majority members might give more serious consideration to minority
interests and become more integratively complex as a result when group interaction actually occurs. On the other hand, in groups for which size or physical
dispersion precludes face-to-face interaction, accountability to outsiders might
motivate members to maximize individual interests such as dominance, influence
and social acceptance. In this case, group members would be less receptive to
evidence of alternatives and more receptive to evidence that supports the
superiority of their original point of view.
In fact, data from an unpublished study support this possibility. When groups
containing majorities and minorities were made externally accountable via the
traditional manipulation described above, they were more sensitive to accountability conditions than was the case here; however, this only occurred when group
members were prevented from interacting with one another (Gruenfeld, ThomasHunt, & Kim, 1996). This finding suggests that accountability to external
constituents can also affect the integrative complexity of majority members,
particularly when group members don’t discuss their views and are thus more
accountable to outsiders than they are to one another. It is also consistent with a
wide range of studies demonstrating the impact of communication on group
members’ mutual consideration for one another’s interests (for a review see
Orbell & Dawes, 1981). Groups that communicate in mixed-motive situations are
much more likely to cooperate than groups who do not, because face-to-face
communication leads to the development of intra-group trust and commitment
(Messick & Brewer, 1983), and to enhanced group solidarity (Kramer & Brewer,
1984). Communication can also increase familiarity in groups, which heightens
members’ attentiveness to one another’s unique insights (Gruenfeld, Mannix,
Williams & Neale, 1996; Williams, Mannix, Neale & Gruenfeld, 1997). Hence, it
is not surprising that groups in the public communication condition were more
responsive to intra-group dynamics (e.g., minority influence and divergent thinking) than to inter-group dynamics (e.g., message transmission and accountability).
The interactive effects of intra- and inter-group dynamics should be further
explored before the absence of public communication condition effects in this
study can be clearly interpreted.
These considerations suggest that in order for group members to experience the
type of conceptual restructuring associated with increased integrative complexity,
they must be motivated to seriously contemplate the alternatives others present.
Furthermore, they raise the possibility that the effects of both divergent thinking
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
223
COMPLEXITY IN GROUPS
and accountability on integrative complexity might depend on the relative
importance of group members’ relationships with internal versus external constituents. Future research should examine these possible contingencies and assess the
effects of these factors more systematically. Specifically, the accountability
conditions under which majority members will be motivated to seriously consider
minority viewpoints and adjust their integrative complexity as a result should be
more thoroughly investigated.
REFERENCES
Asch, S. E. (1951). Effects of group pressure upon the modification and distortion of judgment. In H.
Guetzkow (Ed.), Groups, leadership, and men. Pittsburgh, PA: Carnegie.
Baker, S. M., & Petty, R. E. (1994). Majority and minority influence: Source-position imbalance as a
determinant of message scrutiny. Journal of Personality and Social Psychology, 67 (1), 5–19.
Baker-Brown, G. Ballard, E. J., Bluck, S. De Vries, B., Suedfeld, P., & Tetlock, P. E. (1992). The
conceptual/integrative complexity scoring manual. In C. Smith (Ed.), Motivation and personality: Handbook of thematic content analysis (pp. 401–418). Cambridge: Cambridge University
Press.
Baum, L. (1989). The Supreme Court (3rd ed.). Washington DC: C.Q. Press.
Bazerman, M. H., & Neale, M. A. (1983). Heuristics in negotiation: Limitations to dispute resolution
effectiveness. In M. H. Bazerman & R. J. Lewicki (Eds.), Negotiating in organizations. Beverly
Hills, CA: Sage.
Brenner, S., & Spaeth, J. (1995). Stare indecisis: The alteration of precedent on the Supreme Court,
1946–1992. Cambridge: Cambridge University Press.
De Dreu, C. K. W., & De Vries, N. K. (1993). Numerical support, information processing and attitude
change. European Journal of Social Psychology, 23, 647–662.
Doise, W., & Mugny, G. (1984). The social development of the intellect. Oxford: Pergamon Press.
Deutsch, M. (1949). A theory of cooperation and competition. Human Relations, 2, 129–152.
Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social influence upon
individual judgment. Journal of Abnormal and Social Psychology, 51, 629–636.
Easterbrook, J. A. (1959). The effect of emotion on cue utilization and the organization of behavior.
Psychological Review, 66, 183–201.
Geen, R. G., & Gagne, J. J. (1977). Drive theory of social facilitation: Twelve years of theory and
research. Psychological Bulletin, 84,(6), 1267–1288.
Gruenfeld, D. H. (1992). A ‘‘status-contingency’’ model of cognition and communication in decision
making groups. Paper presented at the 15th Annual Meeting of the International Society of
Political Psychology, San Francisco.
Gruenfeld, D. H. (1995). Status, ideology and integrative complexity on the U.S. Supreme Court:
Rethinking the politics of political decision making. Journal of Personality and Social Psychology, 68(1), 5–20.
Gruenfeld, D. H., Mannix, E. A., Williams, K. Y., & Neale, M. A. (1996). Group composition and
decision making: How member familiarity and information distribution affect process and
performance, Organizational Behavior and Human Decision Processes, 67(1), 1–15.
Gruenfeld, D. H., Thomas-Hunt, M. C., & Kim, P. H. (1996). Unpublished data, Kellogg Graduate
School of Management, Northwestern University, Evanston, IL.
Habeeb, W. M. (1988). Power and tactics in international negotiation: How weak nations bargain
with strong nations. Baltimore: The Johns Hopkins University Press.
Hermann, M. G., Hermann, C. F., & Hutchins, G. L. (1982). Affect. In P. Callahan, L. P. Brady and
M. G. Hermann (Eds.), Describing foreign policy behavior (pp. 207–222). Beverly Hills: Sage.
Hogg, M. A., & Abrams, D. (1988). Social identifications: A social psychology of intergroup relations
and group processes. London: Routledge.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
224
GRUENFELD, THOMAS-HUNT, AND KIM
Komorita, S. S. (1977). Negotiating from strength and the concept of bargaining strength. Journal of
Theory in Social Behavior. Great Britain.
Kramer, R. M. & Brewer, M. B. (1984). Effects of group identity on resource use in a simulated
commons dilemma. Journal of Personality and Social Psychology, 46, 1044–57.
Kruglanski, A. W., & Mackie, D. M. (1990). Majority and minority influence: A judgmental process
analysis. In W. Stroebe & M. Hewstone (Eds.) European Review of Social Psychology (Vol. 1, pp.
229–261). London: Wiley.
Latane, B., & Wolf, S. (1981). The social impact of majorities and minorities. Psychological Review,
88, 438–453.
Levine, J. M. (1989). Reactions to opinion deviance in small groups. In P. B. Paulus (Ed.), Psychology
of group influence (2nd ed., pp. 187–231). Hillsdale, NJ: Erlbaum.
Levine, J. M., Bogart, L. M., & Zdaniuk, B. (1996). Impact of aniticipated group membership on
cognition. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition
(Vol. 3, pp. 531–569). New York: The Guilford Press.
Levine, J. M., & Russo, E. M. (1987). Majority and minority influence. In C. Hendrick (Ed), Review of
personality and social psychology. Beverly Hills, CA: Sage.
Maass, A., West, S. G., & Cialdini, R. B. (1987). Minority influence and conversion. In C. Hendrick
(Ed.), Review of personality and social psychology (Vol. 8, 9). Newbury Park: Sage.
Mackie, D. M. (1987). Systematic and nonsystematic processing of majority and minority persuasive
communications. Journal of Personality and Social Psychology, 53, 41–52.
Mannix, E. A. (1993). Organizations as resource dilemmas: The effects of power imbalance in group
decision making. Organizational Behavior and Human Decision Processes, 55, 1–22.
Mannix, E. A., & Neale, M. A. (1993). Power imbalance and the pattern of exchange in dyadic
negotiation. Group Decision and Negotiation, 2, 119–133.
McGrath, J. E. (1984). Groups: Interaction and performance. Englewood Cliffs, NJ: Prentice-Hall.
Messick, D. M., & Brewer, M. B. (1983). Solving social dilemmas: A review. In L. Wheeler and P.
Shaver (Eds.), Annual review of personality and social psychology (Vol. 3). Beverly Hills, CA:
Sage.
Moscovici, S. (1980). Toward a theory of conversion behavior. In L. Berkowitz (Ed.), Advances in
experimental social psychology (volnoVol. 13, pp. 209–239). New York: Academic Press.
Moscovici, S. (1985). Social influence and conformity. In G. Lindzey and E. Aronson (Eds.), The
handbook of social psychology (Vol. 2, 3rd ed., pp. 347–412). New York: Random House.
Moscovici, S., Lage, E., & Naffrechoux (1969). Influence of a consistent minority on the responses of
a majority in a colour perception task. Sociometry, 32, 365–379.
Moskowitz, G. B. (1996). The mediational effects of attributions and information processing in
minority social influence. British Journal of Social Psychology, 35(1), 47–66.
Mugny, G. (1982). The power of minorities. New York: Academic Press.
Myers, D. G., & Lamb, H. (1976). The group polarization phenomenon. Psychological Bulletin, 83(4),
602–627.
Neale, M. A., & Bazerman, M. H. (1991). Cognition and rationality in negotiation. New York: The
Free Press.
Nemeth, C. J. (1986). Differential contributions of majority and minority influence. Psychological
Review, 93, 23–32.
Nemeth, C. J. (1992). Minority dissent as a stimulant to group performance. In S. Worchel, W. Wood &
J. A. Simpson (Eds.), Group process and productivity (pp. 95–111). Newbury Park, CA: Sage.
Nemeth, C. J., and Kwan, J. L. (1987). Minority influence, divergent thinking and detection of correct
solutions. Journal of Applied Social Psychology, 17, 786–797.
Nemeth, C. J., Mayseless, O., Sherman, J., & Brown, Y. (1990). Exposure to dissent and recall of
information. Journal of Personality and Social Psychology, 58, 429–437.
Nemeth, C. J., & Rogers, J. (1996). Dissent and the search for information. British Journal of Social
Psychology, 35(1), 67–76.
Nemeth, C. J., & Wachtler, J. (1983). Creative problem solving as a result of majority and minority
influence. European Journal of Social Psychology, 13, 45–55.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
225
COMPLEXITY IN GROUPS
Orbell, M., & Dawes, R. M. (1981). Social dilemmas. In G. Stephenson and J. H. Davis (Eds.),
Progress in applied social psychology (Vol. 1). Chichester: Wiley.
Peterson, R. S., & Nemeth, C. J. (1996). Focus versus flexibility: Majority and minority influence can
both improve performance. Personality and Social Psychology Bulletin, 22(1), 14–23.
Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport’s cognitive analysis of
prejudice. Personality and Social Psychology Bulletin, 5, 461–476.
Pruitt, D. G. (1981). Negotiation behavior. New York: Academic Press.
Pruitt, D. G. (1983). Achieving integrative agreements. In M. H. Bazerman and R. J. Lewicki (Eds.),
Negotiating in organizations. Beverly Hills, CA: Sage.
Pruitt, D., Kimmel, M., Britton, S., Carnevale, P., Magneau, J., Pergallo, J., & Engram, P. (1978). The
effect of accountability and surveillance on integrative bargaining. In H. Sauermann (Ed.),
Contributions to experimental economics (Vol. 7). Turbingen: Mohr.
Pruitt, D. G., & Rubin, J. Z. (1986). Social conflict: Escalation, stalemate and settlement. New York:
Random House.
Roby, T. B. (1960). Commitment. Behavior Science, 5, 253–264.
Schroder, H. M., Driver, M. J., & Streufert, S. (1967). Human information processing. New York:
Holt, Rinehart & Winston.
Smith, C. M., Tindale, R. S., & Dugoni, B. L. (1996). Minority and majority influence in freely
interacting groups: Qualitative versus quantitative differences. British Journal of Social Psychology.
Suedfeld, P., Tetlock, P. E., and Streufert, S. (1992). Conceptual/integrative complexity. In C. Smith
(Ed.), Motivation and personality: Handbook of thematic content analysis (pp. 401–418).
Cambridge: Cambridge University Press.
Staub, A. B. (1991). Individual differences in integrative complexity: A group influence hypothesis.
Unpublished undergraduate honors thesis, Harvard College.
Tanford, S., & Penrod, S. (1984). Social influence model: A formal integration of research on majority
and minority influence processes. Psychological Bulletin, 95, 189–225.
Tetlock, P. E. (1981). Pre- to post-election shifts presidential rhetoric: Impression management or
cognitive adjustment? Journal of Personality and Social Psychology, 41, 207–212.
Tetlock, P. E. (1983). Accountability and complexity of thought. Journal of Personality and Social
Psychology, 45, 74–83.
Tetlock, P. E. (1984). Cognitive style and political belief systems on the British House of Commons.
Journal of Personality and Social Psychology, 46, 365–375.
Tetlock, P. E. (1991). An alternative metaphor in the study of judgment and choice: People as
politicians. Theory and psychology (Vol. 1).
Tetlock, P. E. (1992). The impact of accountability on judgment and choice: Toward a social
contingency model. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 25,
pp. 331– 376).
Tetlock, P. E., Bernzweig, J., & Gallant, J. L. (1985). Supreme Court decision making: Cognitive style
as a predictor of ideological consistency of voting. Journal of Personality and Social Psychology,
48, 1127–1239.
Tetlock, P. E., Hannum, K. A., & Micheletti, P. M. (1984). Stability and change in the complexity of
senatorial debate: Testing the cognitive versus rhetorical style hypotheses. Journal of Personality
and Social Psychology, 46(5), 979–990.
Tetlock, P. E., Skitka, L., Boettger, R. (1989). Social and cognitive strategies of coping with
accountability: Conformity, complexity, and bolstering. Journal of Personality and Social
Psychology, 57, 632–641.
Thomas-Hunt, M. C., & Gruenfeld, D. H. (1995). The effects of multiple group membership on
integrative complexity and performance within a negotiation context. Paper presented at the
annual Academy of Management Conference, Vancouver, British Columbia.
Thomas-Hunt, M. C. (1996). An embeddedness approach to intraorganizational negotiations:
Negotiating amidst multiple group memberships. Ph.D. thesis, Organization Behavior Department, Northwestern University, Evanston, IL.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei
226
GRUENFELD, THOMAS-HUNT, AND KIM
Thompson, L. (1990). Negotiation behavior and outcomes: Empirical evidence and theoretical issues.
Psychological Bulletin, 108, 515–532.
Tindale, R. S., Davis, J. H., Vollrath, D. A., Nagao, D. H., & Hinsz, V. B. (1990). Asymmetrical social
influence in freely interacting groups: A test of three models. Journal of Personality and Social
Psychology, 58, 438–449.
Trost, M., Maass, A., & Kenrick, D. (1992). Minority influence: Ego involvement alters cognitive
processes and reverses private acceptance. Journal of Experimental Social Psychology, 28,
234–254.
Van Dyne, L., and Saavedra, R. (1996). A naturalistic minority influence experiment: Effects on
divergent thinking, conflict and originality in work-groups. British Journal of Social Psychology,
35(1), 151–168.
Williams, K. Y., Mannix, E. A., Neale, M. A., & Gruenfeld, D. H. Sharing unique perspectives: The
process of knoeledge exchange in social and informational coalitions. Annual meeting of the
Academy of Management, Boston, August, 1997.
Wood, W., Lundgren, S., Ouellete, J. A., Busceme, S., & Blackstone, T. (1994). Minority influence: A
meta-analytic review of social influence processes. Psychological Bulletin, 115(3), 323–345.
Wood, W., Pool, G. J., Leck, K., & Pruvis, D. (1996) Self-definition, defensive processing, and
influence: The normative impact of majority and minority groups. Journal of Personality and
Social Psychology, 71(6), 1181–1193.
Woike, B. A., & Aronoff, J. (1992). Antecedents of complex social cognitions. Journal of Personality
and Social Psychology, 63(1), 97–104.
Wyer, R. S., Jr., & Gruenfeld, D. H. (1994). Information processing in interpersonal communication.
In D. E. Hewes (Ed.), The cognitive bases of interpersonal communication. Hillsdale, NJ:
Erlbaum.
Wyer, R. S., Jr., & Gruenfeld, D. H. (1995). Information processing in social contexts: Implications for
social memory and judgment. In M. Zanna (Ed.), Advances in experimental social psychology
(Vol. 27, pp. 49–91). San Diego, CA: Academic Press.
Zajonc, R. B. (1960). The process of cognitive tuning in communication. Journal of Abnormal and
Social Psychology, 61(2), 159–167.
Zajonc, R. B. (1965). Social facilitation. Science, 149, 269–274.
JESP1349
@xyserv2/disk3/CLS_jrnl/GRP_jesp/JOB_jesp34-2/DIV_404a01
swei