Professional Documents
Culture Documents
net/publication/4833755
CITATIONS READS
20 1,055
2 authors, including:
Susan E. Brodt
Queen's University
12 PUBLICATIONS 601 CITATIONS
SEE PROFILE
All content following this page was uploaded by Susan E. Brodt on 16 January 2016.
and
Marla Tuchinsky
INSEAD
We thank Lisa Cavanaugh, Steve Hoeffler, Juan Carlos Duque, and Erich Studer-Ellis for helping
run our experiments. Portions of this paper were written with the help of a Duke University
CIBER grant to the first author. An earlier version of this manuscript was presented at the annual
meetings of the Academy of Management.
Address correspondence and reprint requests to Susan Brodt, Fuqua School of Business, Duke
University, Durham, NC 27708-0120; or to Marla Tuchinsky, INSEAD, Boulevard de Constance,
77305 Fontainebleau Cedex, France. E-mail: susan.brodt@duke.edu or marla.tuchinsky@insead.fr.
155
0749-5978/00 $35.00
Copyright 䉷 2000 by Academic Press
All rights of reproduction in any form reserved.
156 BRODT AND TUCHINSKY
Bazerman, 1991; Northcraft & Neale, 1990; Pinkley, 1990). Recently, research
has extended the social context of negotiation to include theoretically important
variables such as culture (Brodt & Seybolt, 1998; Brodt & Tinsley, 1999; Carnev-
ale, 1996), multiple parties (Polzer, Mannix, & Neale, 1996), various relation-
ships between parties (Sheppard & Tuchinsky, 1996; Valley, Neale, & Mannix,
1995), and negotiating teams (Thompson, Peterson, & Brodt, 1996; O’Connor,
1997; Polzer, 1996).
The purpose of this article is to study negotiating teams and their internal
dynamics, specifically the adoption of different negotiator roles and its effects
on negotiator cognition and on negotiation processes and outcomes. We use
Brodt and Duncan’s (1998) model of strategic role differentiation in negotiating
teams, testing its predictions using the familiar “good-cop/bad-cop” distributive
bargaining tactic (Kamisar, 1980). We examine this tactic as it might appear
in organizational settings (e.g., coalitions that form within work groups), how-
ever, and make specific predictions about cognitive underpinnings, limitations
of the tactic’s success, and implications for subsequent social interaction among
group members.
In a nutshell, we propose that the good-cop/bad-cop tactic, in which negotiat-
ing team members work together but in opposite directions, promotes accep-
tance of a team’s offer because the negotiators choreograph a sequence of
distinct persuasive appeals. Each appeal is directed at a different aspect of a
target individual’s perception of the joint decision task underlying the negotia-
tion. The goal of this choreography is to alter the target’s perception of the
conflict situation such that the individual comes to accept the team’s offer. Not
surprisingly, because individuals may bring different initial perceptions to a
conflict situation (e.g., a proclivity to accept or reject a team’s offer), the order
of persuasive appeals may affect the negotiating team’s success. Moreover, the
choreography can also affect interpersonal relations among members of the
group. We ground this proposal in research on negotiating teams and strategic
role differentiation and then test it by conducting two empirical studies. The
first study examines the two-cop (i.e., two-appeal) tactic, testing the interaction
between a target’s initial inclination (accepting or rejecting) and the ordering
of the persuasive appeals (i.e., bad cop then good cop, good cop then bad cop).
The second study addresses an unanswered question raised by study 1, namely:
are both appeals necessary to change perceptions and enhance acceptance, or
does an all-positive or all-negative appeal have the same effects? In this follow
on study we further identify the tactic’s cognitive underpinnings.
Unlike solo negotiators, negotiation team members may choose for strategic
reasons to play different roles during a negotiation. Such role differentiation
may be based on many factors including individual expertise. For example, if
one team member excels at imagining creative solutions while another can
rapidly calculate the value of various agreements, these team members may
decide to play to their strengths and adopt distinct roles during a negotiation.
GOOD COP/BAD COP NEGOTIATING TEAM TACTIC 157
with substantial potential for gains and losses by either or both sides. Despite
impediments to cooperation, they implicitly believe they are better off continu-
ing the negotiation rather than abandoning it, and they negotiate to forge an
agreement more favorable than impasse.1
Whether used by committee members at a university or by cops in a precinct,
does this familiar good-cop/bad-cop tactic actually promote chances that the
targeted individual will acquiesce and accept an offer? According to both lay
intuition and research, the answer appears to be yes (Hilty & Carnevale, 1993;
Rafaeli & Sutton, 1991). Among Rafaeli and Sutton’s (1991) bill collectors, for
example, good cop/bad cop appeared to be both the most commonly used tactic
and the most successful in engendering cooperation. Qualitative data from
their field study highlighted the tactic’s richness and the conviction with which
these professionals intuitively believe the tactic works. What are not clear,
however, are the reasons for the tactic’s success and under what conditions it
increases the chances that an offer will be accepted. For example, does the
tactic lead to mere compliance without engendering commitment or to altering
a person’s understanding (representation) of a conflict situation?2
Rafaeli and Sutton (1991) have proposed psychological contrast (Cialdini,
1984) as the mechanism responsible for the tactic’s success: a good cop who
follows a bad cop is perceived to be warmer and friendlier than a good cop
who either goes it alone or precedes a bad cop. Hilty and Carnevale (1993)
experimentally tested this proposition, confirming the tactic’s effectiveness and
the importance of psychological contrast. Contrary to lay intuition, however,
their results suggested that two persons might not be necessary to achieve the
good-cop/bad-cop effect; they found similar results when a single negotiator
enacted the two roles. Taken together, these studies show that negotiating
team tactics, particularly the good-cop/bad-cop distributive bargaining tactic,
are widely used and seemingly quite successful in engendering compliance.
Psychological contrast appears to be part of the tactic’s success although the
precise nature of the mechanism and influence process remains unclear.
Brodt and Duncan (1998) have offered a model for studying the tactic’s
psychological underpinnings. They propose a specific cognitive representation
and persuasion process and speculate about the conditions under which the
tactic will be most effective in altering perceptions and enabling agreements.
We briefly describe their model in the next section.
1
Note that there may be instances in which no such agreement exists and the parties are better
served, therefore, by accepting impasse rather than continuing to negotiate.
2
One might ask if level of commitment is important or is it necessary to change another’s
underlying perception of a conflict—as long as a party agrees to one’s demands, does it matter
that his or her “heart isn’t in it”? For implementation beyond the letter of the agreement, perhaps
it is. However, exploring the degree of commitment elicited through the good-cop/bad-cop strategy
is beyond the scope of this article. Thank you to an anonymous reviewer for raising this issue.
GOOD COP/BAD COP NEGOTIATING TEAM TACTIC 159
influence aspects of the good cop/bad cop tactic. They represent in a precursory
decision theoretic model the joint decision task facing the parties, where in
the course of bargaining one can alter another’s perceptions of key decision-
making elements (see Schelling, 1960), particularly through a team approach.
They explain their analysis using the prototypical case of two detectives and
an accused. For our purposes, we discuss their model using the organizationally
relevant and functionally equivalent case of a three-person committee or work
group negotiating task assignments. Imagine that two committee members
have decided how to allocate committee task assignments and must now per-
suade the third member to accept their general view of the situation and their
offer (i.e., task assignment). Team members include bad cop Bart and good cop
Gert, trying to negotiate with and persuade target Tam.
According to the model, a person such as Tam begins to form an understand-
ing of a decision structure’s basic components as negotiations proceed. He
learns from interacting with his teammates about various possible actions he
might take, possible consequences of those actions, and their likelihood. He also
develops some preferences about action/consequence combinations (Clemen,
1996). Negotiation and decision-component specification are simultaneous, in-
teractive, social-cognitive activities. At any point in the negotiation, if forced
to make a decision each party would do so based on its current perception of
the situation, which is likely to change as the negotiation proceeds.
In this simple case, Tam faces a decision problem with two possible actions:
accept the offer or reject it. Initially, he is unsure whether the consequences
of taking either action will be positive or negative. Figure 1 represents the
structure that is evolving under bargaining and confrontation of the issues. In
this decision tree, the “accept” and “reject” branches emerge from the basic
action node.
Sometimes in negotiation, Tam begins in a rejecting, uncooperative mode.
where pa represents the probability of a good result given accepting the offer
(0 ⬍ pa ⬍ 1) and pr represents the probability of a good result given rejecting the
offer (0 ⬍ pr ⬍ 1). The four utility values U are functions of action [acceptance (A)
or rejection (R)] and consequence [bad (B) or good (G)]. Initially, the obvious
constraints are imposed, U(A, G) ⬎ U(A, B) and U(R, G) ⬎ U(R, B); that is,
Tam views good consequences as better than bad. As he gathers, or is fed,
information during negotiation, this relationship between the expected utilities
of accepting and rejecting can change.
How exactly does the good-cop/bad-cop tactic function within this decision
context? Negotiation theorists (e.g., Brodt & Dietz, 1999; Lax & Sebenius, 1986;
Neale & Bazerman, 1991; Raiffa, 1982) suggest that Bart and Gert can work
to change, or even create, Tam’s perception of the negotiation situation and
his aspirations. That is, they can persuade Tam to change his expectations
about the likelihood of the chance events—good versus bad outcomes—as well
as his assessments of the utility associated with various outcomes. Being team-
mates, they can be effective by specializing their appeals: bad cop Bart works
on Tam’s perceptions of the “reject” action branch while good cop Gert focuses
on the “accept” action branch.
Given their roles, how should the two teammates coordinate their activities?
According to the model, if Tam is inclined initially to reject the offer, the
tactic will be more successful if Bart begins the negotiation. Bart’s persuasive
arguments shake Tam’s positional inertia and debunk Tam’s perceptions re-
garding benefits following from rejecting. If, on the other hand, Tam were
initially favorably inclined toward the task and in an “accepting” mode, good
cop Gert alone would be more successful in persuading Tam; Gert’s arguments
further strengthen Tam’s commitment to accept the offer. In fact, Bart’s argu-
ments under these circumstances might prove confusing and even weaken
Tam’s initial resolve to accept the offer. During the process of debunking Tam’s
perceptions, Bart’s articulation of presumed benefits that follow from rejecting
might educate Tam and persuade him to raise his estimates of U(R,G) and pr .
HYPOTHESES
We propose and test two sets of hypotheses. The first set helps us understand
psychological contrast and cognitive mechanisms underlying this negotiating
team tactic. The second set embeds the tactic in the broader social context and
assesses its implications for perceptions of good cops and bad cops and the
relationships that they create in negotiation.
the negotiation (i.e., the opposing party’s decision tree). The bad cop, who
usually goes first, targets the person’s initial tendency to reject an offer and
aims to lower his or her assessment that desirable outcomes will follow from
rejection. The good cop, who usually goes second, targets the person’s slight
proclivity toward acceptance and aims to bolster his or her assessment that
desirable outcomes will follow from acceptance. The two roles work in harmony
to increase the likelihood that a negotiator will accept the offer and comply
with the team’s request. This model specifies the cognitive underpinnings of
psychological contrast (Cialdini, 1984).
As a result of this proposed dynamic, the ordering of roles will matter and
interact with the targeted person’s initial inclination; that is, the bad-cop/
good-cop ordering will work best—result in higher acceptance rates—when
individuals are initially disinclined toward the offer rather than inclined. This
leads to the following hypothesis:
HYPOTHESIS 1: The bad-cop/good-cop ordering will result in greater overall acceptance when
participants are initially rejecting versus accepting; opposite results are expected when the
order of presentation is reversed (good cop followed by bad cop).
HYPOTHESIS 3: For those who are inclined to accept, initial exposure to the bad cop will raise
their judged probability of favorable outcomes resulting from rejecting the offer ( pr).
Perceptions of Good Cops and Bad Cops and the Relationships They Create
Interactions with the good-cop/bad-cop team can influence a person’s percep-
tion of the situation and also create distinct impressions about team members.
These impressions may have long-lasting effects in the context of a work group
about to embark on a joint task. In general, based on the nature of their roles
alone (see role theory, Kahn, Wolfe, Quinn, & Snoek, 1964), we would expect
that good cops will be perceived more positively than bad cops will. Targets
may be more inclined to cooperate with good cops because they find them
likable (Cialdini, 1984). The good cop and bad cop “halos” will result in divergent
perceptions of fairness, openness, trustworthiness, toughness, and a willing-
ness to work together in the future, leading to the following hypothesis:
HYPOTHESIS 4: An encounter with a good cop will be perceived more positively than one with
a bad cop, both overall and in terms of perceived fairness, openness, trustworthiness, toughness
and willingness to work together in the future.
A person’s initial frame of mind influences his or her perception of the other
party (Neale & Bazerman, 1991); therefore, an individual who is inclined to
accept an offer is likely to react more strongly (and negatively) to a bad-
cop approach than will someone who is initially disinclined. Moreover, this
unexpected approach’s distinctiveness suggests that the information imparted
would be processed carefully and would be highly memorable and accessible
162 BRODT AND TUCHINSKY
Similarly, the context created by a team member’s partner also influences the
other party’s perceptions. Researchers (see Hilty & Carnevale, 1993; Rafaeli &
Sutton, 1991) have used this argument as a general explanation for the good-
cop/bad-cop tactic’s effectiveness: the good cop’s effectiveness is enhanced by
the context created by the bad cop. Part of this effect is due to the enhanced
positive emotion and likability associated with the good cop who follows the
bad cop (Rafaeli & Sutton, 1991). As a result of this psychological process, we
expect evaluations of bad cops who follow good cops to be more negative than
evaluations of bad cops who make the first encounter. Similarly, evaluations
of good cops who follow bad cops will be more positive than evaluations of good
cops who make the first encounter.
HYPOTHESIS 6: Impressions of good cops and bad cops will be more extreme when experienced
in the context of the partner’s role versus in isolation.
STUDY 1
Method
Participants. Seventy undergraduate students at Duke University partici-
pated in the study as part of a classroom exercise. They were enrolled in one
of two sections of an introductory management course (36 and 34 students
each) and participation was voluntary.
Design. There were two independent variables in the experiment: the target
person’s inclination (initially accepting or rejecting) and the order in which
the good cop/bad cop tactic was performed (bad cop begins or good cop begins).
Both were between-subject variables, creating four different combinations: (a)
initially accepting target, good cop begins; (b) initially accepting target, bad
cop begins; (c) initially rejecting target, good cop begins; and (d) initially re-
jecting target, bad cop begins. To control for actor effects, two individuals played
each role and two versions of each videotape were made to control for the actor
who played the good cop or bad cop role. This resulted in eight conditions in
the experiment and participants were randomly assigned to one condition.
Materials. We simulated an interaction between a student (target) and two
other members of a hypothetical group working on a term project for a college
course. The participant was cast in the role of the target student and the other
team members, who played the good cop and bad cop roles, were presented via
videotaped recordings. The videotape segments reflected casual encounters
GOOD COP/BAD COP NEGOTIATING TEAM TACTIC 163
between the participant and each team member as if they had run into each
other between classes. Encounters were presented sequentially and, in both
cases, the actor playing the good cop or bad cop role spoke directly to the
participant from the video screen (i.e., spoke directly into the camera).
The interaction centered on delegating tasks for the group project, namely
writing the paper, interviewing industry specialists, or making the presenta-
tion. The videotaped team members wanted the target student to write the
paper. Team members adopted good-cop/bad-cop roles in an attempt to persuade
the participant to accept the task. To add urgency and reality to the situation,
all three team members were presumed to be graduating seniors and poor
performance on the term project could potentially jeopardize their graduation.
Moreover, the group’s presentation and report was to be evaluated by a panel
of business executives, including the CEO of the firm the participant (target
student) was to join after graduation.
Two male doctoral students were used in the videotapes. Neither had had
contact with undergraduate students and hence both were unfamiliar to the
participants in the experiment. Both looked young enough to be undergraduate
seniors. Each doctoral student acted both good- and bad-cop roles and the
scripts were written so that the two segments could be edited to create tapes
including different combinations of actors and their roles. The videotape re-
cordings were shot in a typical classroom and lasted approximately 2 min
apiece. Four sets of videotapes were created, manipulating both the order of
the good-cop/bad-cop arguments and the actor playing the roles. In all cases,
both actors appeared in the videotape; in no case did the same person play
both roles on a single tape.
Results
FIG. 2. Mean assessments of task performance (i.e., expected individual and group project
scores) that would follow from rejection or acceptance of the team’s offer for participants who were
initially accepting or rejecting.
166 BRODT AND TUCHINSKY
FIG. 3. Mean assessments of social evaluations (i.e., expected evaluations by professor, CEO,
and teammates) that would follow from rejection or acceptance of the team’s offer for participants
who were initially accepting or rejecting.
about the target person’s initial inclination. Hypotheses 1, 2, and 3 test the
effects of the tactic on targets’ perceptions of the situation and outcomes.
According to Hypothesis 1, the bad-cop/good-cop ordering would result in
greater overall acceptance among initially rejecting participants than among
initially accepting ones; the opposite results were expected when the ordering
was reversed. To test this hypothesis, we first compared acceptance judgments
made by the two experimental groups (i.e., accepting, rejecting participants)
after being exposed to the bad-cop and good-cop team members, in that order
(i.e., traditional order). We then made the same comparison for groups experi-
encing the tactic in the reverse order. Contrary to our prediction, after their
bad-cop/good-cop encounters, rejecting participants were less likely to accept
relative to their initially accepting peers, although both groups expressed very
high levels of acceptance after this traditional encounter (Mreject ⫽ .80, SD ⫽
.18 vs Maccept ⫽ .90, SD ⫽ .11), F(1, 38) ⫽ 3.99, p ⫽ .05. Consistent with our
prediction, however, after experiencing the tactic in its reverse order rejecting
participants were indeed significantly less likely to accept than were their
accepting peers (Maccept ⫽ .71, SD ⫽ .21 vs Mreject ⫽ .52, SD ⫽ .29), F(1, 28) ⫽
4.30, p ⫽ .04. As well, the two-way ANOVA revealed significant main effects
for both initial inclination and good-cop/bad-cop order, MEincline: F(1, 66) ⫽
8.99, p ⬍ .01, MEorder: F(1, 66) ⫽ 22.89, p ⬍ .01) [interaction term was not
statistically significant, F(1, 66) ⫽ 1.02, p ⫽ .31]. Despite not finding the
anticipated interaction, our hypotheses about the tactics’ effects on rejecting
participants were supported; those about its effects on initially accepting par-
ticipants were not.
GOOD COP/BAD COP NEGOTIATING TEAM TACTIC 167
FIG. 4. Mean probability of accepting the negotiating team’s offer before versus after exposure
to the good-cop/bad-cop team, acting in traditional or reverse-traditional ordering.
Perceptions of good cops and bad cops and the relationships they create. Ex-
posure to the tactic affected more than just perceptions of the decision task;
it also affected impressions of the individuals playing the good cop and bad
cop roles. Because our hypothetical project group was about to embark on its
task, social perceptions and the relationship created among the team members
were very important to assess.
In Hypothesis 4, we predicted that participants would generally view the
good cop more positively than the bad cop. Relative to bad cops, we expected
good cops to be perceived as more fair, open, trustworthy, and less tough,
leading participants to report being more willing to work again with the good
cop than with the bad cop. To test this hypothesis, we combined these five
indices into a single measure representing how positively participants viewed
their teammates. Results strongly supported our hypothesis. The average rat-
ings for good cops and bad cops were 4.26 and 1.94, respectively (using a
REGWQ test, ␣ ⫽ .05, this difference was significant; critical range ⫽ .43).
Next, we predicted that participants initially inclined to accept the writing
task would judge the bad-cop team member more negatively than would their
initially rejecting peers, especially when the bad cop began the negotiation;
the opposite result were predicted when the bad-cop/good-cop order was re-
versed. To test this hypothesis, we averaged three ratings—fairness, trust,
and willingness to work together in the future. When comparing assessments
made by those who encountered the bad cop first, accepting participants rated
him more negatively than did rejecting participants, although this difference
was only marginally significant (Maccept ⫽ 1.93 SD ⫽ .60 vs Mreject ⫽ 2.36,
SD ⫽ .84), F(1, 38) ⫽ 3.41, p ⫽ .06. This was also true when the bad cop was
encountered second (Maccept ⫽ 1.76, SD ⫽ .98 vs Mreject ⫽ 1.93, SD ⫽ 1.20),
F(1, 29) ⫽ 2.93, p ⫽ .09.
We predicted the opposite result for those who encountered the good cop
first, namely accepting participants would rate the good cop more favorably
than would rejecting participants. Although these data were in the predicted
direction (Maccept ⫽ 3.55, SD ⫽ .86 vs Mreject ⫽ 3.42, SD ⫽ 1.02), they were not
GOOD COP/BAD COP NEGOTIATING TEAM TACTIC 169
significantly different [F(1, 29) ⫽ .12, p ⫽ .73]. This was also true when the
good cop was encountered second (Maccept ⫽ 5.10, SD ⫽ 1.38 vs Mreject ⫽ 4.45,
SD ⫽ 1.21), F(1, 39) ⫽ .16, p ⫽ .69. Overall, it seems that these roles influenced
social perception independent of participants’ initial inclination, although en-
countering a bad cop first engendered slightly greater disfavor among accepting
versus rejecting participants.
Hypothesis 6 tested a version of the contrast effect (Cialdini, 1984; Rafaeli &
Sutton, 1991): evaluations of good cops and bad cops were expected to be more
extreme when they were made in the context of the other role. For example,
a good cop following a bad cop would have more positive evaluations than
would a good cop making the first encounter. Results for this hypothesis are
presented in Fig. 5. Bad cops who began the negotiation received an average
evaluation of 2.2, whereas those who followed a good cop received a slightly
lower rating of 1.8, F(1, 69) ⫽ 2.23, p ⫽ .14. On the other hand, good cops who
followed bad cops were rated significantly higher than were those who went
first (M ⫽ 4.78 vs M ⫽ 3.49), F(1, 69) ⫽ 21.19, p ⬍ .01. Hence, the contrast
effect is asymmetric, affecting good cops more than bad cops: a bad cop’s evalua-
tion was only slightly more negative when viewed in light of a good cop partner,
whereas a good cop’s evaluation was significantly enhanced by association with
a bad cop.
Discussion
FIG. 5. Mean evaluations of bad cops and good cops when they were encountered first or
second (i.e., after having encountered the other role).
170 BRODT AND TUCHINSKY
whereas the identical appeals made in the reverse order had no such effect.
Also, the tactic’s success was greatest when the target was initially rejecting
or unsympathetic to the team’s appeal. Using an analytic model proposed by
Brodt and Duncan (1998), we went beyond these outcome effects and identified
some of the cognitive underpinnings or reasons for the tactic’s success. We
assessed respondents’ subjective probabilities about certain events and also
the values of various outcomes, showing specialized influence of each role on
targets’ perceptions of the situation. Specifically, the bad cop role targeted the
reject branch of the party’s decision tree, decreasing targets’ assessments that
good outcomes would follow from rejection; the role had little or no effect on
assessments of good outcomes following from acceptance. As well, the good
cop role enhanced assessments associated with the accept branch (i.e., good
outcomes would follow from acceptance), although this result was only true
for initially accepting participants.
As well, we found that the roles created distinct impressions, powerfully
influencing interpersonal relationships. Relative to the bad cop, the good cop
was perceived to be more fair, open, trustworthy, and less tough, leading partici-
pants to report being more willing to work with this individual than with the
bad cop. Moreover, encountering the bad cop first engendered somewhat greater
disfavor among initially accepting versus initially rejecting participants. In
organizational contexts (e.g., workgroups), these social implications may prove
to be very consequential. Juxtaposing the roles heightened these distinct im-
pressions, especially for good cops. Relative to assessments made in isolation,
participants perceived good cops more positively in the context of their bad
cop partners; this contrast effect did not occur, however, for bad cops. These
results support and extend the contrast effect proposed by Rafaeli and Sutton
(1991) and Hilty and Carnevale (1993), who also point out the asymmetry
between each role’s sensitivity to contrast effects.
Finally, we showed the generalizability of this negotiating team model well
beyond its original use in police interrogation. In our case, it worked success-
fully in describing and analyzing the dynamics of our simulated three-person
workgroups; moreover, this application brought to light new elements to con-
sider (e.g., social implications for ongoing workgroups).
STUDY 2
bad cop). Could a team achieve the same result, for example, without role
differentiation at all? This result would suggest that psychological contrast
might not fully account for our findings. A united front or twofold pressuring
by one’s teammates (i.e., the majority) might prove an even more effective
persuasion strategy than having a mixture of persuasive appeals. In terms of
the analytic model, having two people deliver similar arguments might bolster
one (targeted) branch of the decision tree to such an extent that it dominates
and tips the scales toward acceptance of the team’s offer.
Study 2 was designed to address these questions about a possible majority/
minority effect and clarify this issue. We examine hypotheses from Study 1 in
light of a single appeal or role, again organizing our results in terms of (a) the
tactic’s effects on perceptions and outcomes and (b) the impressions of good
cops and bad cops and the relationships they create.
Method
Participants. Fifty Duke University undergraduate students (7 women, 43
men) participated in the experiment. All were members of a service group and
each received $10 U.S. for participating. (This sample was not significantly
different from that of Study 1.)
Design. The two independent variables were inclination (initially accepting,
rejecting) and role or persuasive appeal (bad cop, good cop). In all cases, a
negotiating team enacted a single role, dividing the persuasive appeal between
them. Both variables were between subject, creating four different combina-
tions: (a) initially accepting target, good-cop/good-cop appeal; (b) initially ac-
cepting target, bad-cop/bad-cop appeal; (c) initially rejecting target, good-cop/
good-cop appeal; and (d) initially rejecting target, bad-cop/bad-cop appeal.
Materials. Materials were identical to those used in Study 1, with one
exception. To create the good-cop/good-cop and bad-cop/bad-cop encounters we
edited the videotapes from Study 1 and divided the teammates’ arguments in
half. So, the first teammate delivered the first part of the good cop or bad cop
argument and the second teammate completed the appeal.
Procedure. The procedure was identical to that used in Study 1, with
one exception. Participants in Study 2 received $10 U.S. before leaving the
experiment.
SD ⫽ .23 vs Mpost ⫽ .49, SD ⫽ .37), t(26) ⫽ ⫺2.65, p ⬍ .01. This basic finding
was also true for accepting and rejecting participants (Mpre-accept ⫽ .68, SD ⫽
.25 vs Mpost-accept ⫽ .60, SD ⫽ .34), t(23) ⫽ ⫺1.44, p ⬍ .16, (Mpre-reject ⫽ .45,
SD ⫽ .19 vs Mpost-reject ⫽ .41, SD ⫽ .27), t(25) ⫽ ⫺0.83, p ⬍ .41. Their initially
accepting or rejecting inclination was unchanged by exposure to the cops. These
results further strengthen our argument that both roles are needed and that
contrast serves to accentuate the persuasiveness of the (final) good-cop appeal.
Recall that in Study 1 using the traditional bad-cop/good-cop ordering, partici-
pants were more likely to comply after the two encounters than they had been
at the start (Mpre ⫽ .60 vs Mpost ⫽ .85), t(38) ⫽ 5.68, p ⬍ .01, and that this
was not true when the ordering was reversed (Mpre ⫽ .61 vs Mpost ⫽ .59),
t(29) ⫽ ⫺ .05, p ⫽ .95. These pre- and postencounter compliance estimates are
shown in Fig. 6.
FIG. 6. Mean probability of accepting the negotiating team’s offer before versus after exposure
to the bad-cop/bad-cop, and good-cop/good-cop teams. (Results for bad-cop/good-cop and good-cop/
bad-cop teams are included for completeness.)
GOOD COP/BAD COP NEGOTIATING TEAM TACTIC 173
That is, exposure to the good cops did not increase estimates associated with
acceptance (M ⫽ ⫺0.02, SD ⫽ .19), t(22) ⫽ ⫺0.63, p ⬍ .52.
Hence, results of Study 2 replicate those of Study 1. We found the same
partial support for our basic conjecture that the two roles have specialized
influence on the target’s perception of the conflict situation. As in Study 1, the
bad cop appeared to target the reject branch of the participant’s decision tree.
After experiencing the bad cop, participants decreased their assessments that
good outcomes would follow from rejection; this exposure had little effect on
their assessments of good outcomes following from acceptance. Also, initial
exposure to the good cop did not have the expected effects on judgments related
to the accept branch.
Perceptions of good cops and bad cops and the relationships they create. Did
exposure to a single tactic, twofold pressuring affect impressions of the individu-
als playing the good cop and bad cop roles? Yes, it did. As in Study 1, distinct
impressions were created, leading participants to view the good cops more
positively than the bad cops in terms of fairness, openness, trustworthiness,
toughness (reverse coded), and willingness to work together in the future (com-
bined score: Mgood ⫽ 3.27, SD ⫽ 1.10 vs Mbad ⫽ 1.99, SD ⫽ 1.11), F(1, 48) ⫽
2.65, p ⬍ .01.
GENERAL DISCUSSION
not been addressed satisfactorily. For individuals who are initially rejecting,
the tactic appears to have the predicted effects on cognition and behavior. For
those who are initially accepting, however, the story is less clear. This finding
has both theoretical and practical implications. On the theoretical side, such
asymmetries have been shown in other areas of cognitive functioning (e.g.,
Tversky & Kahneman, 1974) and it may be fruitful to examine these differences
and integrate them into this growing body of evidence. On the practical side,
we need to understand the tactic’s dynamics when targets are initially accepting
because this proclivity may be quite common in organizational applications.
Although it may be reasonable for police officers to assume an arrestee is
initially rejecting, it may not be reasonable for two faculty members to make
that assumption about a colleague.
As mentioned previously, we showed that the two roles have specialized
influence on perceptions of the decision situation; however, we did not isolate
specific elements of argumentation that produced these effects. Research on
the psycholinguistics of argumentation and persuasion could help understand
these findings. Additionally, it might lend support to our counterintuitive pre-
diction regarding the bad cop’s inadvertent influence on accepting targets.
Our findings about the relationships created by the tactic deserve further
attention. The result of a social interaction like the one simulated in our
experiment extends beyond the agreement to perform a task; individuals must
live up to agreements and deliver on their promises. In our case, they needed
to complete the project, and we found some evidence that participants were
less willing to work with the bad cop again. Given that a target’s negative
feelings may persist for some time, it seems plausible that this tactic might
be reserved for single transactions or for those involving relative strangers.
Hence, do people use this tactic within ongoing relationships, or use it differ-
ently? Also, is a target more receptive within an ongoing work group, especially
as one’s goals and needs become entwined with those of the group? There is
some evidence that people, out of social-relational concerns, often adopt an
influencer’s position “in order to maintain or facilitate a (real or imagined)
relationship” (Chaiken, Wood, & Eagly, 1997, p. 715). It may be worth exploring
how the nature of parties’ relationships affect both the use and efficacy of
this tactic.
Finally, although we showed in Study 2 that both appeals are necessary for
the tactic’s success, the question remains as to whether the two roles (and their
associated effects on different aspects of individuals’ cognitive representations)
require two individuals or whether a single person could produce the same
results. Existing evidence suggests that one person can indeed produce these
effects (e.g., Hilty & Carnevale, 1993; Rafaeli & Sutton, 1991). Acting as a
“reformed sinner,” the person transforms from bad cop to good cop without
seeming incoherent, unstable, or untrustworthy and creates the same results
as a negotiating team. The role of a social label or explanation for this transfor-
mation and the cognitive dynamics shown in our studies may help researchers
understand this phenomenon and its boundary conditions.
176 BRODT AND TUCHINSKY
REFERENCES
Brodt, S., & Dietz, L. (1999). Shared information and information sharing: Understanding negotia-
tion as collective construal. In R. Bies, R. Lewicki, & B. Sheppard (Eds.) Research on negotiation
in organizations (Vol. 9). Greenwich, CT: JAI Press.
Brodt, S., & Duncan, G. (1998). Role-differentiation strategies for negotiating teams: Analysis of
“good cop/bad cop” distributive bargaining tactics. Manuscript under review, Duke University,
Fuqua Business School, Durham, NC.
Brodt, S., & Seybolt, P. (1998). Putting conflict into context: The effects of national culture on
managerial cognition and interpersonal conflict. Manuscript under review, Duke University,
Fuqua Business School, Durham, NC.
Brodt, S., & Tinsley, C. (1999). The role of frames, schemas and scripts in understanding conflict
resolution across cultures. Manuscript under review, Duke University, Fuqua School of Business,
Durham, NC.
Carnevale, P. (1996). Property, culture and negotiation. In R. Kramer & D. Messick (Eds.), Negotia-
tion as a social process (pp. 309–323). Thousand Oaks, CA: Sage.
Chaiken, S., Wood, W., & Eagly, A. (1997). Principles of persuasion. In E. Higgins & A. Kruglanski
(Eds.), Social psychology: Handbook of basic principles (pp. 702–742). New York: Guilford.
Cialdini, R. (1984). Influence: The new psychology of modern persuasion. New York: Quill.
Clemen, R. (1996). Making hard decisions: An introduction to decision analysis. Belmont, CA:
Duxbury.
Hilty, J., & Carnevale, P. (1993). “Black-hat/White-hat” strategy in bilateral negotiation. Organiza-
tional Behavior and Human Decision Processes, 55, 444–469.
Kahn, R., Wolfe, D., Quinn, R., & Snoek, J. (1964). Organizational stress: Studies of role conflict
and ambiguity. New York: Wiley.
Kamisar, Y. (1980). Police interrogation and confession: Essays in law and policy. Ann Arbor:
University of Michigan.
Lax, D., & Sebenius, J. (1986). Manager as negotiator. New York: Free Press.
Neale, M., & Bazerman, M. (1991). Cognition and rationality in negotiation. New York: Free Press.
Northcraft, G., & Neale, M. (1990). Behavioral negotiation theory: A framework for conceptualizing
dyadic bargaining. In L. Cummings & B. Staw (Eds.), Research in organizational behavior
(Vol. 12). Greenwich, CT: JAI Press.
O’Connor, K. (1997). Groups and solos in context: The effects of accountability on team negotiation.
Organizational Behavior and Human Decision Processes, 72, 384–407.
Pinkley, R. (1990). Dimensions of conflict frame: Disputant interpretations of conflict. Journal of
Applied Psychology, 75, 117–126.
Polzer, J. (1996) Intergroup negotiations: The effects of negotiating teams, Journal of Conflict
Resolution, 40, 678–698.
Polzer, J., Mannix, E., & Neale, M. (1996). Multiparty negotiation in its social context. In R.
Kramer & D. Messick (Eds.), Negotiation as a social process (pp. 123–142). Thousand Oaks,
CA: Sage.
Rafaeli, A., & Sutton, R., (1991). Emotional contrast strategies as means of social influence: Lessons
from criminal interrogators and bill collectors. Academy of Management Journal, 34, 749–775.
Raiffa, H., (1982). The art and science of negotiation. Cambridge, MA: Belknap.
Schelling, T. (1960). The strategy of conflict. Cambridge, MA: Harvard Univ. Press.
Sheppard, B., & Tuchinsky, M. (1996). Micro OB and the network organization. In R. Kramer &
T. Tyler (Eds.), Trust in organizations (pp. 140–165). Thousand Oaks: Sage
Stangor, C., & McMillan, D. (1992). Memory for expectancy-congruent and expectancy-incongruent
information: A review of social and social development literatures. Psychological Bulletin,
111, 42–61.
GOOD COP/BAD COP NEGOTIATING TEAM TACTIC 177
Thompson, L. (1990). Negotiation behavior and outcomes: Empirical evidence and theoretical
issues. Psychological Bulletin, 108, 515–532.
Thompson, L., Peterson, E., & Brodt, S. (1996). Team negotiation: Examination of integrative and
distributive bargaining. Journal of Personality and Social Psychology, 70, 66–78.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science,
185, 1124–1131.
Valley, K., Neale, M., & Mannix, E. (1995). Friends, lovers, colleagues, strangers: The effects of
relationships on the process and outcome of dyadic negotiations. In R. Bies, R. Lewicki, & B.
Sheppard (Eds.), Research on negotiation in organizations (Vol. 5, pp. 65–93). Greenwich, CT:
JAI Press.
Walton, R., & McKersie, R. (1965/1991). A behavioral theory of labor negotiations: An analysis of
a social interaction system (2nd ed.). Ithaca, NY: ILR Press.