INTERPERSONAL RELATIONS AND GROUP PROCESSES
The Interpersonal Effects of Anger and Happiness in Negotiations
Gerben A. van Kleef, Carsten K. W. De Dreu, and Antony S. R. Manstead
University of Amsterdam
Three experiments investigated the interpersonal effects of anger and happiness in negotiations. In the
course of a computer-mediated negotiation, participants received information about the emotional state
(anger, happiness, or none) of their opponent. Consistent with a strategic-choice perspective, Experiment
1 showed that participants conceded more to an angry opponent than to a happy one. Experiment 2
showed that this effect was caused by trackingāparticipants used the emotion information to infer the
otherās limit, and they adjusted their demands accordingly. However, this effect was absent when the
other made large concessions. Experiment 3 examined the interplay between experienced and commu-
nicated emotion and showed that angry communications (unlike happy ones) induced fear and thereby
mitigated the effect of the opponentās experienced emotion. These results suggest that negotiators
are especially influenced by their opponentās emotions when they are motivated and able to consider
them.
Conflict is a ubiquitous feature of social life, pervading social
interactions and influencing peopleās behavior throughout all lev-
els of society. Negotiation, defined by Pruitt and Carnevale (1993)
as āa discussion between two or more parties with the apparent aim
of resolving a divergence of interestsā (p. 2), is one of the most
common and constructive ways of dealing with conflict. Despite
decades of research on negotiation, surprisingly little attention has
been given to the role of emotions in negotiation. This is unfor-
tunate, because emotions are inherent to negotiation and social
conflict: If two parties have a difference of opinion but neither has
an emotional reaction, there will be no negotiation (Davidson &
Greenhalgh, 1999). Thus, āas an impetus for and byproduct of
social conflict, emotion is potentially central to understanding how
individuals think about, behave within, and respond to bargaining
situationsā (Barry, 1999, p. 94). This article presents three exper-
iments designed to enhance understanding of the role of emotions
in negotiation. Specifically, we focus on the social consequences
of emotion, that is, the way negotiators respond to their opponentās
emotions.
Emotions in Negotiation
There are probably as many definitions of emotion as there are
researchers interested in the topic. However, most definitions point
to three features of emotion: physiological reactions, action ten-
dencies, and subjective experience (Lazarus, 1991). Emotions dif-
fer from moods in that they are discrete (Russell & Barrett, 1999),
of relatively short duration (Oatley & Jenkins, 1996), and inten-
tionalāthat is, directed at an object or event (Frijda, 1993; Russell
& Barrett, 1999). In the remainder of this article we use the term
emotion in the sense intended above, whereas affect will be used as
a superordinate construct that encompasses both moods and emo-
tions (cf. Barry & Oliver, 1996). In the current research, we chose
to examine the effects of two emotions potentially relevant to
most, if not all, negotiations, namely anger and happiness.
In conceptualizing the role of emotions in negotiation, it is
useful to make a distinction between intrapersonal effects and
interpersonal effects (cf. Morris & Keltner, 2000). Intrapersonal
effects refer to the influence of a negotiatorās emotions on his or
her own negotiation behavior. Interpersonal effects refer to the
influence of one negotiatorās emotions on the other negotiatorās
behavior. Previous studies on emotion in negotiation all focused on
the intrapersonal effects of emotions and affect. The negotiatorās
own positive affect has been shown to increase concession making
(Baron, 1990), stimulate creative problem solving (Isen, Daubman,
& Nowicki, 1987), increase joint gains (Allred, Mallozzi, Matsui,
& Raia, 1997; Carnevale & Isen, 1986), increase preferences for
cooperation (Baron, Fortin, Frei, Hauver, & Shack, 1990), reduce
the use of contentious tactics (Carnevale & Isen, 1986), and
increase the use of cooperative negotiation strategies (Forgas,
1998). Negative affect has been shown to decrease initial offers
(Baron et al., 1990), decrease joint gains (Allred et al., 1997),
Gerben A. van Kleef, Carsten K. W. De Dreu, and Antony S. R.
Manstead, Faculty of Social and Behavioural Sciences, Department of
Psychology, University of Amsterdam, Amsterdam, the Netherlands.
Antony S. R. Manstead is now at the School of Psychology, Cardiff
University, Wales, United Kindgom.
We thank Bernard Nijstad for his valuable comments on previous
versions of this article.
Correspondence concerning this article should be addressed to Gerben
A. van Kleef, Faculty of Social and Behavioural Sciences, Department of
Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB Am-
sterdam, the Netherlands. E-mail: g.a.vankleef@uva.nl
Journal of Personality and Social Psychology, 2004, Vol. 86, No. 1, 57ā76
Copyright 2004 by the American Psychological Association, Inc. 0022-3514/04/$12.00
DOI: 10.1037/0022-3514.86.1.57
57
promote the rejection of ultimatum offers (Pillutla & Murnighan,
1996), increase the use of competitive strategies (Forgas, 1998),
and decrease the desire to work together in the future (Allred et al.,
1997). Thus, these studies have shown quite consistently that
negotiators experiencing positive affect tend to be more coopera-
tive and conciliatory, whereas negotiators who are in a negative
affective state tend to be more competitive and reluctant to make
concessions.
With regard to the interpersonal effects of emotions in negoti-
ations, research is lacking. Recently, however, several scholars
have emphasized the importance of the interpersonal functions of
emotions in negotiations (e.g., Adler, Rosen, & Silverstein, 1998;
Barry, Fulmer, & Van Kleef, in press; Barry & Oliver, 1996;
Davidson & Greenhalgh, 1999; Morris & Keltner, 2000; Thomp-
son, Nadler, & Kim, 1999). It has been argued that emotions have
a number of important social functions and consequences (Frijda
& Mesquita, 1994; Keltner & Haidt, 1999; Keltner & Kring, 1998;
Oatley & Jenkins, 1992). First, emotion displays tend to evoke
complementary or reciprocal emotions in others that help individ-
uals to respond to significant social events (Keltner & Haidt,
1999). For example, rage in one person may evoke fear in the
other, and happiness in one person may induce happiness or
sympathy in the other (Keltner & Kring, 1998).
Second, emotions are communications to both oneself (Schwarz
& Clore, 1983) and other people (Oatley & Johnson-Laird, 1987)
conveying information about how one feels about things (Ekman,
1993; Scherer, 1986), about oneās social intentions (Ekman,
Friesen, & Ellsworth, 1972; Fridlund, 1994), and about oneās
orientation toward other people (Knutson, 1996). Because emo-
tions only arise in response to events that are appraised as relevant
to some concern (Frijda & Mesquita, 1994; Lazarus, 1991), they
provide important information to observing individuals. In the
context of a negotiation, emotions may signal what value one
attaches to the different issues and provide critical feedback about
oneās mood and willingness to agree (Putnam, 1994). For example,
an angry opponent conveys the impression of a hard-to-get, tough
negotiator who will not settle for a suboptimal outcome. By
contrast, happiness detracts from the impression of a tough bar-
gainer (Wall, 1991). Happiness implies that one is satisfied with
the distribution of outcomes and suggests that one does not aspire
to a larger share of the pie.
Third, emotions can serve as incentives or deterrents for other
individualsā social behavior (Klinnert, Campos, Sorce, Emde, &
Svejda, 1983). Anger may signal that some standard of socially
acceptable behavior is violated (Averill, 1982) and that behavioral
adaptation is needed. Thus, in negotiations āanger may be used to
indicate that negative consequences (such as an impasse) may
ensue if an unreasonable tactic or position is pursuedā (Daly, 1991,
p. 36). Furthermore, because anger is often followed by aggression
(Averill, 1982), it may be used to induce fear and compliance. By
contrast, happiness encourages others to pursue their course of
action and may therefore not serve the negotiatorās strategic inter-
ests (Parrott, 1994).
Building on the foregoing discussion, we advance two compet-
ing hypotheses. The first is based on the notion of social contagion
(Levy & Nail, 1993), which can be defined as āthe spread of affect,
attitude, or behavior from Person A (the āinitiatorā) to person B
(āthe recipientā) where the recipient does not perceive an inten-
tional influence attempt on the part of the initiatorā (Levy & Nail,
1993, p. 266). There are a number of subtypes of social contagion,
which occur under different circumstances. Most research has
focused on contagion processes in face-to-face settings (e.g.,
āprimitive emotional contagionā; Hatfield, Cacioppo, & Rapson,
1992, 1994), demonstrating peopleās tendency to unintentionally
and automatically ācatchā other peopleās emotions through their
facial expressions, vocalizations, postures, or movements (e.g.,
Friedman & Riggio, 1981; Hess & Blairy, 2001; Hietanen,
Surakka, & Linnankoski, 1998; Lundqvist & Dimberg, 1995;
Neumann & Strack, 2000; Wild, Erb, & Bartels, 2001). However,
research has suggested that social contagion can also occur in
computer-mediated interaction (De Dreu, Carnevale, Emans, &
Van de Vliert, 1994). For example, Thompson and Nadler (2002)
observed that individuals engaging in computer-mediated commu-
nication unconsciously imitate not only the linguistic structure of
each otherās messages but also the socialā emotional connotations
of the messages. Thus, drawing on social contagion theory and
previous findings regarding the intrapersonal effects of emotions,
it can be predicted that negotiators who are faced with an angry
opponent will themselves become angry and behave competitively,
placing high demands and making small concessions. Conversely,
negotiators who are confronted with a happy opponent can be
expected to become happy and behave more cooperatively, placing
low demands and making large concessions.
A competing hypothesis can be derived from the mixed-motive
nature of negotiations and the social consequences of emotions.
Negotiations involve mixed-motive interdependence (Schelling,
1960). On the one hand, negotiators are motivated to cooperate,
because an agreement is usually more profitable than no agree-
ment. On the other hand, however, they are motivated to compete
in order to maximize their individual profits. This tension between
cooperation and competition becomes manifest in a type of nego-
tiation behavior that is called mismatching (Pruitt & Carnevale,
1993). Mismatching involves placing high demands when the
opponent appears conciliatory and low demands when the oppo-
nent appears tough (Pruitt, 1981). When oneās opponent appears
soft, this leaves scope for placing high demands oneself without
risking impasse. By contrast, the opponentās apparent toughness is
likely to be interpreted as endangering agreement, thereby encour-
aging concession making.
The tendency to mismatch the opponentās negotiation behavior
has been found in a considerable number of studies (e.g., Bartos,
1974; Bateman, 1980; Chertkoff & Conley, 1967; De Dreu et al.,
1994; Druckman & Bonoma, 1976; Druckman, Zechmeister, &
Solomon, 1972; Komorita & Brenner, 1968; Liebert, Smith, Hill,
& Keiffer, 1968; Pruitt & Syna, 1985; Smith, Pruitt, & Carnevale,
1982; Yukl, 1974a, 1974b). It has been shown to be especially
likely to occur when negotiators lack information about the oppo-
nentās outcomes (Liebert et al., 1968; Yukl, 1974b) and limits
(Pruitt & Syna, 1985). When individuals lack such information,
they use other sources of information to locate the opponentās
limits, a behavior that is referred to as tracking (see Pruitt, 1981).
As the research summarized above has shown, one of these types
of information consists of the opponentās demands. However,
there are other sources of information on which negotiators can
base their strategies, one of which may be the opponentās emo-
tions. An angry negotiator cannot be expected to make any (fur-
58
VAN KLEEF, DE DREU, AND MANSTEAD
ther) concessions. The only way to prevent the negotiation from
ending in impasse, then, is by making concessions oneself. By
contrast, a happy opponent does not constitute a threat to agree-
ment and is therefore unlikely to elicit any (more) concessions. On
the basis of these strategic considerations, it can be predicted that
participants with an angry opponent will become more concilia-
tory, making lower demands and larger concessions than partici-
pants with a happy opponent.
Experiment 1
Experiment 1 tested these two competing hypotheses regarding
the effects of the opponentās experienced emotion on the focal
negotiatorās behavior. Participants were led to believe that they
negotiated with either an angry or a happy opponent or, in a control
condition, were given no information about the opponentās emo-
tional state. The social contagion hypothesis posits that partici-
pants with an angry opponent will make higher demands and
smaller concessions than participants with a happy opponent. The
strategic-choice hypothesis predicts that participants with an angry
opponent will make lower demands and larger concessions than
participants with a happy opponent. In a more exploratory fashion,
we also investigated the effects of the opponentās emotions on
participantsā impressions of the opponent.
There are a number of ways in which the opposing negotiatorās
emotions can be manipulated. One is to use scenarios where
participants are asked to indicate how they would behave when
faced with an angry or a happy opponent. However, with scenarios
it is difficult to create experimental realism and to obtain behav-
ioral data. Alternatives are to present pictures of facial expressions
showing anger or happiness (e.g., Dimberg & O
ĀØ hman, 1996; Hess,
Philippot, & Blairy, 1998) or to manipulate the vocal expression of
emotion (e.g., Banse & Scherer, 1996; Neumann & Strack, 2000;
Scherer, 1986). Unfortunately, these methods are difficult to im-
plement in a dynamic context such as a negotiation. Because of the
limitations of the existing methods, we chose to manipulate the
opposing negotiatorās emotion in the context of a computer-
mediated negotiation where parties could not see each other and
communicated via computers (see, e.g., De Dreu & Van Kleef, in
press; De Dreu & Van Lange, 1995; Hilty & Carnevale, 1993;
Moore, Kurtzberg, Thompson, & Morris, 1999). In the past de-
cade, computer-mediated communication has become a ubiquitous
phenomenon. Negotiations are increasingly being conducted by
means of technological media such as the Internet and e-mail
(McKersie & Fonstad, 1997), and as organizations respond to the
globalization of markets by expanding around the world, face-to-
face communications are likely to give way to technologically
mediated communications (Moore et al., 1999).
In the present experiments, we led participants to believe that
the purpose of the study was to find out how knowledge about
oneās opponentās intentions affects negotiation processes and out-
comes. During the negotiation, participants received information
about their opponentās intentions, which contained the manipula-
tion of the opponentās emotion. In this way we were able to create
a relatively dynamic setting with high experimental realism, in
which we had perfect control over the opponentās emotion. We
return to this method, and its advantages and disadvantages, in the
General Discussion section.
Method
Participants and Experimental Design
A total of 128 male and female undergraduate students at the University
of Amsterdam participated in the study for course credit or monetary
compensation (10 Dutch guilders, roughly equivalent to US$4). The one-
factor experimental design included the opponentās emotion (anger vs.
happiness vs. no emotion) as a between-participants variable and demand
level as the main dependent variable. Additional dependent variables were
self-reported emotion, perceptions of the opponent, and manipulation
checks. Participants were randomly assigned to the experimental condi-
tions, and the experimenters were blind to this assignment.
Procedure
On arrival at the laboratory, participants were welcomed to the experi-
ment and were seated in separate cubicles in front of a computer. To
enhance the participantsā awareness of the presence of other participants,
they were told that the experimenter would give them a sign as soon as
everyone had arrived. After 5 min the experimenter told the participants to
press a key, whereupon the computer program started. From that point on,
all instructions were presented on the computer screen. Participants read
that the purpose of the experiment was to study negotiation in a situation
where the negotiating parties could not see each other, and they were led
to believe that they would engage in a computer-mediated negotiation with
another participant (whose behavior was in fact simulated by the
computer).
Negotiation task.
The negotiation task was an adapted version of the
one used by De Dreu and Van Kleef (in press; see also De Dreu & Van
Lange, 1995; Hilty & Carnevale, 1993), which captures the main charac-
teristics of real-life negotiation (i.e., multiple issues differing in utility to
the negotiator, information about oneās own payoffs only, and the typical
offerā counteroffer sequence). In the current version, participants learned
that they would be assigned the role of either buyer or seller of a consign-
ment of mobile phones and that their objective was to negotiate the price,
the warranty period, and the duration of the service contract of the phones.
Participants were then presented with a payoff chart (see Table 1) that
showed them which outcomes were most favorable to them and were told
that their objective was to earn as many points as possible. As can be seen
in Table 1, Level 9 on price ($110) yielded 0 points and Level 1 ($150)
yielded 400 points (i.e., increments of 50 points per level). For warranty
period, Level 9 (9 months) yielded 0 points, and Level 1 (1 month) yielded
120 points (i.e., 15-point increments). Finally, for duration of service
contract, Level 9 (9 months) yielded 0 points, and Level 1 (1 month)
yielded 240 points (i.e., 30-point increments). The corresponding payoff
table for the other party was not shown, and participants were told only that
it differed from their own.
To enhance participantsā involvement in the negotiation task, they were
informed that points would be converted to lottery tickets at the end of the
experiment and that the more points earned, the more lottery tickets one
would obtain and the greater would be oneās chance of winning a prize of
100 Dutch guilders (approximately US$40). To emphasize the mixed-
motive nature of the negotiation, participants were told that only those who
reached an agreement would participate in the lottery. Thus, on the one
hand, there was an incentive to earn as many points as possible, whereas on
the other hand, there was an incentive to reach an agreement.
After a short pause during which the computer supposedly assigned
buyer and seller roles to the participants, all participants were assigned the
role of seller. They were told that the buyer (i.e., the opponent) would make
the first offer and that the negotiation would continue until an agreement
was reached or until time ran out. Just before the negotiation started,
participants learned that an additional goal of the study was to examine the
effects of having versus not having information about the opponentās
59
ANGER AND HAPPINESS IN NEGOTIATIONS
intentions. They read that the computer had randomly determined that they
would receive information about the intentions of the opponent without the
opponent knowing it and that the opponent would not receive information
about their intentions.
After these instructions, the negotiation started and the buyer (i.e., the
computer) made a first offer. Over the negotiation rounds the buyer
proposed the following levels of agreement (for priceāwarrantyāservice):
8 ā7ā 8 (Round 1), 8 ā7ā7 (Round 2), 8 ā 6 ā7 (Round 3), 7ā 6 ā7 (Round 4),
7ā 6 ā 6 (Round 5), and 6 ā 6 ā 6 (Round 6). Past research has shown that this
preprogrammed strategy has face validity and is seen as intermediate in
cooperativeness and competitiveness (De Dreu & Van Lange, 1995). A
demand by the participant was accepted if it equaled or exceeded the offer
the computer was about to make in the next round. Thus, for example, if
the participant demanded 7ā 6 ā 6 in Round 4, this demand was accepted by
the computer because its next offer (in Round 5) would have been 7ā 6 ā 6.
If no agreement was reached after the sixth round, the negotiation was
interrupted, because up to this point few participants suspect that they are
playing against a computer, whereas after the sixth round such suspicion
usually develops rapidly (De Dreu & Van Lange, 1995). Further, we
wished to obtain measures of participantsā emotions and perceptions during
the negotiation, and we suspected that reaching agreement would strongly
color such perceptions. Finally, from past research in our laboratory we
know that participants who reach consensus before the sixth negotiation
round usually do not take the task very seriously. Thus, following Tripp
and Sondak (1992), participants who reached agreement before Round 6
(n
ā«½
10) were excluded from the analyses. (Retaining those participants
did not change the pattern of results reported below.)
Manipulation of the opponentās emotion.
After the first, third, and fifth
negotiation round, participants received information about āthe intentions
of the buyer,ā which contained the manipulation of the buyerās emotion.
Participants had to wait for about a minute and a half while the buyer was
supposedly asked to reveal what he or she intended to offer in the next
round, and why. After this short wait, participants received the answer
supposedly given by the buyer, which was presented in a separate box, in
a different font, and which contained some typing errors in order to
enhance experimental realism. The buyerās intentions were held constant
across conditions and contained the buyerās intended offer for the next
round. That is, after Round 1 the buyer wrote āI think I will offer 8 ā7ā7,ā
which would indeed be the buyerās next offer. The buyerās intention also
contained an emotional statement that constituted the experimental manip-
ulation. It was stressed that the buyer did not know that his or her
āintentionsā were revealed to the participant. This was done in order to lead
participants to believe that they received information about the real emo-
tions of the opponent, and not faked, inhibited, or exaggerated emotions.
Thus, participants were led to believe that the emotion statements they
received reflected the emotions as experienced by the opponent at that
time, and not emotions that were altered for self-presentational or strategic
reasons.
The emotion statements were pretested in a pilot study involving 28
psychology students, none of whom participated in the main study. We
tested eight statements reflecting anger and nine statements reflecting
happiness using a within-participants design. The order of the 17 state-
ments was randomized for each participant, and for each statement partic-
ipants were asked to indicate on a 7-point scale how comprehensible they
found the statement (1
ā«½
very incomprehensible, 7
ā«½
very comprehensi-
ble) and to what extent they felt it reflected anger and happiness (1
ā«½
not
at all, 7
ā«½
to a great extent). We then selected those statements that had
the highest scores on the emotion they were supposed to reflect and the
lowest scores on the emotion that they were not supposed to reflect,
provided that the statements did not differ with respect to comprehensibil-
ity. Ultimately, we selected three statements for each emotion. All selected
statements were rated higher on the emotion they were supposed to express
than on the emotion they were not supposed to express according to
paired-sample t tests (all ts
ā¬
10, all ps
ā¬
.01). Further, one-sample t tests
showed that there was a significant effect of all statements on the rating of
the corresponding emotion (all ts
ā¬
10, all ps
ā¬
.01). Finally, paired-
sample t tests revealed that the statements did not differ with respect to
comprehensibility (all ts
ā¬
1, ns).
After the first negotiation round, participants in the angry opponent
condition received the information āthis offer makes me really angry,ā
followed by the intention statement āI think I will offer 8 ā7ā7,ā which was
the same for all conditions. In the happy opponent condition, participants
read āI am happy with this offer,ā followed by the same intention state-
ment. In the control condition, participants only received the intention
statement. After the third and fifth negotiation rounds participants again
received an emotional statement and an intention. Table 2 displays all
statements used in the experiment. Note that the intended offer always
matched the true offer subsequently made by the opponent.
Dependent measures.
The offers made by participants in each round
were recorded and transformed into an index revealing the negotiatorās
total level of demand for each negotiation round (i.e., the number of points
demanded in that round, summed across the three negotiation issues of
price, warranty, and service; see Table 1). In addition, participants com-
pleted a postnegotiation questionnaire that contained manipulation checks
and a number of items designed to measure participantsā emotions and their
impression of the opponent.
To check the adequacy of the manipulation of the opponentās emotion,
participants were asked to indicate on a 7-point scale how angry, irritated,
Table 1
Participantsā Payoff Chart (Used in Experiments 1 and 3)
Level
Price of phones
Warranty period
Service contract
Price ($)
a
Payoff
Warranty
(in months)
Payoff
Service
(in months)
Payoff
1
150
400
1
120
1
240
2
145
350
2
105
2
210
3
140
300
3
90
3
180
4
135
250
4
75
4
150
5
130
200
5
60
5
120
6
125
150
6
45
6
90
7
120
100
7
30
7
60
8
115
50
8
15
8
30
9
110
0
9
0
9
0
a
Prices in Dutch guilders were converted to U.S. dollars and rounded to the nearest $5.
60
VAN KLEEF, DE DREU, AND MANSTEAD
happy, and satisfied they thought their opponent had been during the
negotiation. The items designed to measure perceived anger and irritation
correlated substantially (r
ā«½
.75) and were averaged into a single index of
perception of the opponentās anger. Similarly, the items pertaining to
happiness and satisfaction were combined into an index of perception of
the opponentās happiness (r
ā«½
.64).
Participantsā own emotions were measured with items pertaining to
anger and happiness: āTo what extent did you experience anger [happiness]
during the negotiation?ā (1
ā«½
not at all, 5
ā«½
to a great extent). Finally, we
assessed participantsā impressions of the opponent: āI have developed a
positive impression of the buyerā (1
ā«½
totally disagree, 5
ā«½
totally agree).
Results
Manipulation Checks
If the manipulation of the opponentās emotion was successful,
we should find an interaction between the opponentās emotion and
the participantās rating of their opponentās emotion, such that
ratings within each emotion condition are higher for the corre-
sponding emotion than for the other emotion (a within-participants
manipulation check) and that ratings between the emotion condi-
tions are higher for the intended emotion than for the other emo-
tion (a between-participants check). To test whether this was the
case, we conducted a 3 (opponentās emotion: angry vs. nonemo-
tional vs. happy)
ā«»
2 (participantās perception of opponentās
emotion: angry vs. happy) analysis of variance (ANOVA), the
latter variable being a within-participants factor.
Results showed the predicted interaction between the oppo-
nentās emotion and the participantsā perception of the opponentās
emotion, F(2, 115)
ā«½
65.68, p
ā¬
.01. From the means in Table 3,
it can be seen that participants in the angry opponent condition
rated their opponents as significantly more angry than did partic-
ipants in the happy opponent or nonemotional opponent condi-
tions. Similarly, participants with a happy opponent rated the
opponent as happier than did participants with an angry or non-
emotional opponent. Further, paired-sample t tests revealed that
ratings within the different emotion conditions were higher for the
intended emotion than for the other emotion: Participants in the
angry opponent condition rated the opponent as more angry than
happy, t(37)
ā«½
11.98, p
ā¬
.01, and those in the happy opponent
condition rated the opponent as more happy than angry, t(39)
ā«½
3.91, p
ā¬
.01. In the nonemotional opponent condition, no signif-
icant difference was found, t(39)
ā¬
1, ns. Together, these results
indicate that the manipulation of the opponentās emotion was
successful.
Demand Level
Demand level in Rounds 1ā 6 was analyzed using ANOVA with
the opponentās emotion (anger vs. happiness vs. no emotion) as a
Table 2
Statements Used for the Manipulation of the Opponentās Experienced Emotion
Opponentās emotion
Statement
After Round 1
Angry
This offer makes me really angry, I think I will offer xāyāz.
Happy
I am happy with this offer, I think I will offer xāyāz.
Nonemotional
I think I will offer xāyāz.
After Round 3
Angry
This is really getting on my nerves. I am going to offer xāyāz.
Happy
This is going pretty well so far. I am going to offer xāyāz.
Nonemotional
I am going to offer xāyāz.
After Round 5
Angry
I am going to offer xāyāz, because this negotiation pisses me off.
Happy
I am going to offer xāyāz, because I feel good about this negotiation.
Nonemotional
I am going to offer xāyāz.
Note.
Statements were pretested and have been translated from Dutch. Deliberate typing errors have been
omitted. In Experiment 2, the opponentās offer was dependent on the concession size condition. xāyāz
ā«½
the
opponentās intended offer corresponded with the actual offer in the next round.
Table 3
Participantās Perception of the Opponentās Emotion and
Participantās Self-Reported Emotion as a Function of the
Opponentās Experienced Emotion in Experiment 1
Emotion
Opponentās emotion
Anger
Happiness
No emotion
Rating of opponentās emotion
Anger
5.82
a
2.64
d
3.65
e
Happiness
1.53
c
4.70
b
3.51
e
Rating of own emotion
Anger
3.16
a
1.95
b
2.48
b
Happiness
2.13
b
2.55
a
2.05
b
Note.
Means not sharing a similar subscript differ at p
ā¬
.05. Ratings of
the opponentās emotion were given on a 7-point scale; ratings of own
emotion were given on a 5-point scale.
61
ANGER AND HAPPINESS IN NEGOTIATIONS
between-participants variable and negotiation round (1ā 6) as a
repeated-measures variable. This analysis revealed a significant
main effect of negotiation round, F(5, 575)
ā«½
222.85, p
ā¬
.01,
indicating that level of demand declined from Round 1 (M
ā«½
629)
to Round 6 (M
ā«½
462). Results also showed a significant effect of
the opponentās emotion, F(2, 115)
ā«½
5.66, p
ā¬
.01, showing that
average demands were lower for participants who negotiated with
an angry opponent (M
ā«½
490) compared with participants who
dealt with a happy opponent (M
ā«½
553). Participants with a
nonemotional opponent occupied an intermediate position that
differed only from the angry opponent condition (M
ā«½
540).
The main effects of the opponentās emotion and negotiation
round were qualified by a significant two-way interaction, F(10,
575)
ā«½
5.99, p
ā¬
.01. As can be seen from Figure 1, the influence
of the opponentās emotion became more apparent as the negotia-
tion progressed. From the second round onward, the different
conditions started to diverge, the effect becoming stronger in each
round. From the fifth round onward, all three conditions differed
significantly at p
ā¬
.05. Participants with an angry opponent
conceded most between the first and sixth negotiation rounds
(distance traveled: M
ā«½
210), followed by those with a nonemo-
tional opponent (distance traveled: M
ā«½
173). Highest demands
and smallest concessions were made by participants with a happy
opponent (distance traveled: M
ā«½
122). This pattern of results is
consistent with the strategic-choice perspective and inconsistent
with the emotional-contagion perspective.
Participantsā Emotions and Impression of the Opponent
ANOVA showed that participantsā own emotions were affected
by their opponentsā emotions. Participants who had negotiated
with an angry opponent reported more anger than did participants
who had negotiated with a happy or nonemotional opponent, F(2,
115)
ā«½
8.01, p
ā¬
.01. Similarly, participants who had negotiated
with a happy opponent reported more happiness than did partici-
pants who had negotiated with an angry or nonemotional oppo-
nent, F(2, 115)
ā«½
3.36, p
ā¬
.05 (all means are displayed in Table
3). These data suggest that in the course of the negotiation, social
contagion occurred. The opponentās emotion also influenced the
participantsā impressions of the opponent. Participants developed
more positive impressions of happy opponents (M
ā«½
3.55) than of
nonemotional opponents (M
ā«½
2.75), who in turn were rated more
positively than angry opponents (M
ā«½
1.68), F(2, 115)
ā«½
43.73,
p
ā¬
.01.
Discussion
The results of Experiment 1 pertaining to demand level and
concession making support the strategic-choice hypothesis. Partic-
ipants with an angry opponent placed lower demands and made
larger concessions than did participants with a happy opponent,
and participants with a nonemotional opponent took an interme-
diate position. Although our data pertaining to self-reported emo-
tions suggest that social contagion occurred (participants with an
angry opponent reported more anger, and participants with a happy
opponent reported more happiness), the participantsā emotions did
not influence their negotiation behavior in the way that was pre-
dicted on the basis of the social contagion hypothesis. That is,
happy participants did not behave more cooperatively than did
angry participants.
There are a number of possible explanations why participantsā
emotions did not influence their negotiation behavior. One is that
participantsā emotions resulted from their own behavior rather than
from the opponentās emotions. With an angry opponent, one
decides to make relatively large concessions so as to avoid im-
passe, but this strategic choice also implies relatively low out-
comes for oneself, which may be a source of frustration and
irritation. Another possibility is that although social contagion
occurred at the level of emotional experience, this did not translate
into demands and concessions. We believe that the best way to
interpret the current findings is that they do not refute the existence
of social contagion processes in negotiation, but rather show that
social contagion, if it occurs, can be overruled by strategic con-
siderations. We return to this issue in more detail in the General
Discussion.
Experiment 2
Although the results of Experiment 1 favor the strategic-choice
hypothesis over the social contagion hypothesis, we felt that more
evidence was needed to firmly establish the effects. Therefore, the
first objective of Experiment 2 was to replicate the major findings
of Experiment 1 by showing that the effect of the opponentās
emotion on participantās demands and concessions is not produced
by the participantās emotions but rather by strategic considerations
that cause participants to mismatch the opponentās anger with low
demands and happiness with high demands. Thus, we predicted
that although social contagion might occur, participantsā emotions
would not influence their negotiation behavior.
The second objective of Experiment 2 was to submit the
strategic-choice reasoning to an additional test by examining
whether the effect of the opponentās emotion would disappear
when strategic considerations played a lesser role. If the effects of
the opponentās emotion on demands and concessions are indeed
caused by a process of strategic decision making on the part of the
Figure 1.
Demand level as a function of the opponentās emotion and
negotiation round in Experiment 1.
62
VAN KLEEF, DE DREU, AND MANSTEAD
focal negotiator, the impact of those emotions should be reduced if
there is no need to consider them. For example, when the opponent
makes large concessions, there is little need to take his or her
emotions into accountāsettlement rather than impasse appears
likely, and one can expect a relatively good deal for oneself. When
the opponent makes small concessions, however, it is harder to get
oneās share. In this case, the strategic information that is provided
by the opponentās emotion is a valuable asset and is more likely to
be scrutinized and acted on. To test this prediction, we manipu-
lated the opponentās emotion and the opponentās concession size
and predicted that negotiators would be less influenced by the
opponentās emotions when the opponent made large rather than
small concessions.
The third goal of Experiment 2 was to examine more closely the
process that underlies the effects of the opponentās emotion on
demands and concessions. Although the results of Experiment 1
are consistent with the strategic-choice hypothesis, they do not
speak directly to the process that causes the effect of the oppo-
nentās emotion. The findings obtained in Experiment 1 suggest that
negotiators engage in trackingāthat is, they use the opponentās
emotion in their attempt to locate the opponentās limits and then
adjust their behavior in accordance with the assumed location of
those limits (cf. Pruitt, 1981). If the opponent is angry and is
thought to have a high limit, negotiators will place less ambitious
demands than when the opponent is happy and is thought to have
a low limit (e.g., Chertkoff & Baird, 1971; Pruitt & Carnevale,
1993; Yukl, 1974a).
The results of Experiment 1 suggest that participants in our
experiment monitored the opponentās emotions, tried to infer the
location of the opponentās limits from these emotions, and used
this information to form their negotiation strategy. Furthermore,
our reasoning regarding the moderating effect of otherās conces-
sion size suggests that the need to engage in tracking, and thus the
attention that is paid to the otherās emotion, is greater when the
other makes small rather than large concessions. Therefore, we
predicted that participants would estimate the otherās limit to be
higher when the other was angry rather than happy, especially
when the other made small rather than large concessions. Finally,
we expected the effects of the otherās emotions on the participantās
demands and concessions to be mediated by the participantās
appraisal of the otherās limits.
Method
Participants and Experimental Design
Participants were 103 male and female undergraduate students at the
University of Amsterdam who received course credit or monetary com-
pensation (10 Dutch guilders, approximately US$4 ) for participating, and
were randomly assigned to the experimental conditions. The 2
ā«»
3 factorial
design included the opponentās emotion (anger vs. happiness) and the
opponentās concession size (small vs. moderate vs. large) as between-
participants variables and demand level as the main dependent variable.
Additional dependent variables were participantsā estimates of the oppo-
nentās limits, impression of the opponent, emotions, and manipulation
checks.
Procedure
The procedure was essentially the same as in Experiment 1. To be able
to manipulate the opponentās concession size symmetrically, we expanded
the original 9-level negotiation task to a 15-level task and made all three
issues equally valuable. Thus, the negotiation consisted of three issues,
each with 15 possible levels of agreement (see Table 4 for details). As in
Experiment 1, participants who reached agreement before Round 6 (n
ā«½
10) were excluded from the analyses. (Including these participants in the
analyses did not change the pattern of results.)
Manipulation of the opponentās emotion.
The manipulation of the
opponentās emotion was the same as in Experiment 1, except that the
intention statements had to be adjusted to match the opponentās offer in the
following round. Thus, after Round 1 the buyer wrote āI think I will offer
[14 ā15ā14 or 14 ā14 ā13 or 13ā14 ā12],ā the intended offer depending on
the concession size condition (cf. Table 5). Likewise, after the third round
Table 4
Participantsā Payoff Table in Experiment 2
Level
Price of phones
Warranty period
Service contract
Price ($)
a
Payoff
Warranty
(in months)
Payoff
Service
(in months)
Payoff
1
175
280
1
280
1
280
2
170
260
2
260
2
260
3
165
240
3
240
3
240
4
160
220
4
220
4
220
5
155
200
5
200
5
200
6
150
180
6
180
6
180
7
145
160
7
160
7
160
8
140
140
8
140
8
140
9
135
120
9
120
9
120
10
130
100
10
100
10
100
11
125
80
11
80
11
80
12
120
60
12
60
12
60
13
115
40
13
40
13
40
14
110
20
14
20
14
20
15
105
0
15
0
15
0
a
Prices in Dutch guilders were converted to U.S. dollars and rounded to the nearest $5.
63
ANGER AND HAPPINESS IN NEGOTIATIONS
the buyer stated āI am going to offer [14 ā14 ā13 or 12ā13ā12 or 11ā12ā
10],ā and after the fifth round the buyer wrote āI am going to offer
[13ā14 ā12 or 11ā12ā10 or 9 ā10 ā 8].ā In this way, the intended offer
always matched the offer subsequently made by the opponent.
Manipulation of the opponentās concession size.
The opponentās con-
cession size was systematically varied using noncontingent prepro-
grammed concession strategies (see Table 5). In the small-concessions
condition, the opponent conceded 1 unit in each round. Thus, in this
condition the opponentās opening offer was 15ā15ā14 (representing a
1-unit concession from the maximum payoff of 15ā15ā15), and the offer in
the sixth and last round was 13ā14 ā12 (i.e., 6 units below 15ā15ā15). In
the moderate-concessions condition the opponent conceded 2 units per
round, offering 14 ā15ā14 (minus 2 units) in the first round, and 11ā12ā10
in the sixth round (minus 12 units). Finally, in the large-concessions
condition the opponent conceded 3 units per round, starting with 14 ā15ā13
(minus 3 units), and finishing with 9 ā10 ā 8 (minus 18 units). Similar
preprogrammed concession patterns have been successfully used in prior
research (see, e.g., De Dreu & Van Lange, 1995; Hilty & Carnevale, 1993;
Smith et al., 1982).
Dependent measures.
Participantsā estimates of the opponentās limits
were measured with three items, one for each issue (āWhat do you think
was the buyerās lowest acceptable level of agreement on [price/warranty/
service]?ā). Responses could range from 1 (indicating an extremely low
limit) to 15 (indicating an extremely high limit; see Table 4). Appraisals of
the opponentās limits on the three issues were highly correlated and were
therefore averaged into a single index of appraisal of the opponentās limit
(
ā£
ā«½
.91).
Impression of the opponent was measured with 11 items pertaining to
the opponentās honesty, trustworthiness, morality, competitiveness, hostil-
ity, cooperativeness, reasonableness, stubbornness, friendliness, and gen-
eral mood and impression (e.g., āDuring the negotiation, the buyer made a
hostile impressionā; āI have developed a positive impression of the buyerā;
1
ā«½
totally disagree, 7
ā«½
totally agree). Negatively framed items were
reverse scored, and the items were combined into an impression scale (
ā£
ā«½
.92). Participantsā own emotions were measured with 8 items, 3 pertaining
to happiness (
ā£
ā«½
.81), and 5 pertaining to anger (
ā£
ā«½
.87).
To check the adequacy of the manipulation of the opponentās emotion,
participants were asked to indicate on a 7-point scale how angry, irritated,
bad-tempered, happy, satisfied, and pleased they thought their opponent
had been during the negotiation. The first three items were combined into
a single index of perception of the opponentās anger (
ā£
ā«½
.94), and the
latter three were combined into an index of the opponentās happiness (
ā£
ā«½
.95). The concession size manipulation was checked with four items (e.g.,
āIn the course of the negotiation, the buyer made large concessionsā;
āDuring the negotiation the buyer stood firmā; 1
ā«½
totally disagree, 7
ā«½
totally agree), which were averaged into a single index of perceived
concessions (
ā£
ā«½
.81).
Results
Manipulation Checks
Opponentās emotion.
The manipulation of the opponentās
emotion was checked in the same way as in Experiment 1. Results
revealed a significant interaction between the opponentās emotion
and the participantās perception of the opponentās emotion, F(1,
91)
ā«½
960.94, p
ā¬
.01. Participants in the angry opponent condi-
tion rated their opponents as significantly more angry (M
ā«½
5.92)
than did participants in the happy opponent condition (M
ā«½
1.69),
and participants with a happy opponent rated the opponent as
happier (M
ā«½
5.21) than did participants with an angry opponent
(M
ā«½
1.80). Paired-sample t tests showed that ratings within the
different emotion conditions were indeed higher for the intended
emotion than for the other emotion: Participants in the angry
opponent condition rated the opponent as more angry than happy
(M
ā«½
5.92 vs. M
ā«½
1.80), t(50)
ā«½
24.85, p
ā¬
.01, and participants
in the happy opponent condition rated the opponent as more happy
than angry (M
ā«½
5.21 vs. M
ā«½
1.69), t(41)
ā«½
19.31, p
ā¬
.01.
Opponentās concession size.
ANOVA revealed a significant
main effect of the opponentās concession size on the perceived
concessions scale, F(2, 90)
ā«½
5.64, p
ā¬
.01, showing that partic-
ipants in the large-concessions condition reported larger conces-
sions (M
ā«½
3.63) than did those in the small-concessions condition
(M
ā«½
2.53). Participants in the moderate-concessions condition
occupied an intermediate position (M
ā«½
2.91), differing from the
large-concessions condition ( p
ā¬
.05) but not from the small-
concessions condition.
Demand Level
Demands in Rounds 1ā 6 were submitted to a 2 (opponentās
emotion: anger vs. happiness)
ā«»
3 (opponentās concession size:
small vs. moderate vs. large) mixed-model ANOVA, with the
opponentās emotion and concession size as between-participants
variables and demands in Rounds 1ā 6 as a repeated-measures
variable. First of all, this analysis yielded a main effect of nego-
tiation round, F(5, 435)
ā«½
154.02, p
ā¬
.01, indicating that partic-
Table 5
Manipulation of the Opponentās Concession Size (Experiment 2)
Round
Small
Moderate
Large
Offer
Concession
Offer
Concession
Offer
Concession
1
15ā15ā14
ā«ŗ
1
14ā15ā14
ā«ŗ
2
14ā15ā13
ā«ŗ
3
2
14ā15ā14
ā«ŗ
1
14ā14ā13
ā«ŗ
2
13ā14ā12
ā«ŗ
3
3
14ā14ā14
ā«ŗ
1
13ā13ā13
ā«ŗ
2
12ā12ā12
ā«ŗ
3
4
14ā14ā13
ā«ŗ
1
12ā13ā12
ā«ŗ
2
11ā12ā10
ā«ŗ
3
5
13ā14ā13
ā«ŗ
1
12ā12ā11
ā«ŗ
2
10ā11ā9
ā«ŗ
3
6
13ā14ā12
ā«ŗ
1
11ā12ā10
ā«ŗ
2
9ā10ā8
ā«ŗ
3
Note.
In the small-concessions conditions the opponent conceded one unit per round, in the moderate-
concessions conditions the opponent conceded two units per round, and in the large-concessions conditions the
opponent conceded three units per round.
64
VAN KLEEF, DE DREU, AND MANSTEAD
ipantsā demands declined over time (the average demands in
Rounds 1 and 6 were 762 and 580, respectively).
Secondly, the analysis revealed a significant multivariate main
effect of the opponentās emotion on demands in Rounds 1ā 6, F(1,
87)
ā«½
5.37, p
ā¬
.03. Replicating the results of Experiment 1,
participants with an angry opponent placed lower average de-
mands (M
ā«½
634) than did participants with a happy opponent
(M
ā«½
682).
Thirdly, we found a significant interaction between the oppo-
nentās emotion and negotiation round, F(5, 435)
ā«½
10.47, p
ā¬
.01.
As in Experiment 1, the effect of the opponentās emotion became
more apparent as the negotiation progressed, causing participants
in the different conditions to diverge more after each consecutive
round. Consistent with the strategic-choice hypothesis, participants
with an angry opponent conceded significantly more between the
first and sixth negotiation rounds (distance traveled: M
ā«½
222)
than did participants who negotiated with a happy opponent (dis-
tance traveled: M
ā«½
131). The interaction between concession size
and negotiation round was not significant, F(10, 435)
ā¬
1, ns.
Fourthly, and most important, we obtained a significant three-
way interaction between the opponentās emotion, the opponentās
concession size, and negotiation round, F(10, 435)
ā«½
2.03, p
ā¬
.03, indicating that the interactive effect of the opponentās emotion
and negotiation round on level of demand was moderated by the
opponentās concession size. This three-way interaction is dis-
played in Figure 2. Simple-effects analysis revealed a highly
significant interaction between emotion and negotiation round for
participants in the small-concessions condition, F(5, 435)
ā«½
10.33,
p
ā¬
.01. For participants in the moderate-concessions condition,
the Emotion
ā«»
Round interaction was weaker but still significant,
F(5, 445)
ā«½
2.22, p
ā¬
.05, whereas in the large-concessions
condition the interaction was truly nonsignificant, F(5, 435)
ā«½
1.49, ns. Thus, as predicted, the effect of the opponentās emotion
is reduced when the opponent makes large concessions.
To facilitate mediation analysis (see below) we also analyzed the
demand level data by calculating a distance-traveled index (demand
in Round 1 minus demand in Round 6; for discussions, see De Dreu,
Koole, & Oldersma, 1999; Pruitt, 1981). This distance-traveled index
yielded results similar to those described above. A main effect of the
opponentās emotion showed that participants with an angry opponent
conceded more between the first and sixth round (M
ā«½
222) than did
participants with a happy opponent (M
ā«½
131), F(1, 87)
ā«½
18.48, p
ā¬
.01, and a marginally significant interaction showed that this effect
was moderated by the opponentās concession size, F(2, 87)
ā«½
2.84,
p
ā¬
.07. Simple-effects analyses revealed that participants whose
opponents made small concessions were strongly influenced by the
opponentās emotions, F(1, 87)
ā«½
18.08, p
ā¬
.01. The effect of
the opponentās emotion was somewhat weaker for participants
whose opponents made moderate concessions, F(1, 87)
ā«½
4.05, p
ā¬
.05, and nonsignificant for those whose opponents made large con-
cessions, F(1, 87)
ā«½
1.21, ns. Cell means pertaining to this interaction
are displayed in Table 6.
Appraisal of the Opponentās Limits
ANOVA revealed a main effect of the opponentās concession
size on appraisal of the opponentās limit, showing that participants
with an opponent who made large concessions judged the oppo-
nentās limit to be lower (M
ā«½
7.03) than did participants with an
opponent who made moderate concessions (M
ā«½
8.21) or small
concessions (M
ā«½
8.23), F(2, 87)
ā«½
4.70, p
ā¬
.02. More impor-
tant, ANOVA showed a significant main effect of the opponentās
emotion on appraisal of the opponentās limit, F(1, 87)
ā«½
8.38, p
ā¬
Figure 2.
Demand level as a function of the opponentās emotion, the
opponentās concession size, and negotiation round in Experiment 2.
65
ANGER AND HAPPINESS IN NEGOTIATIONS
.01. As predicted, angry opponentsā limits were judged to be
higher (M
ā«½
8.27) than were happy opponentsā limits (M
ā«½
7.29).
1
The interaction between the opponentās emotion and concession
size did not reach statistical significance, F(2, 87)
ā«½
1.35, ns. How-
ever, Rosenthal and Rosnow (1984) have argued that an omnibus
interaction test is highly conservative and is not informative when the
experimental design involves more than two levels for one or more
independent variables (as in the current case). We therefore tested the
effects of the opponentās emotion within each of the concessions
conditions using simple-effects analyses.
2
Consistent with our theo-
rizing, these analyses revealed that participantsā estimates of the
opponentās limits were only influenced by the opponentās emotion
when the opponent made small concessions, F(1, 89)
ā«½
7.74, p
ā¬
.01,
and not when the opponent made moderate, F(1, 89)
ā«½
1.63, ns, or
large, F(1, 89)
ā¬
1, ns, concessions. As can be seen in Table 6,
participants estimated the opponentās limit to be higher when the
opponent was angry rather than happy, but only when the opponent
made small rather than moderate or large concessions.
Mediation Analysis
The strategic-choice hypothesis posits that anger induces con-
cessions because it affects negotiatorsā appraisals of the otherās
limits. So far, we have shown that anger elicits lower demands and
larger concessions than does happiness and that anger and happi-
ness influence participantsā estimates of the otherās limits in the
expected direction. Furthermore, and consistent with the strategic-
choice perspective, these effects were especially manifest in the
small-concessions condition and absent in the large-concessions
condition. Thus, to test whether the opponentās emotion indeed
influences participantsā concessions through their effect on the
participantās appraisal of the opponentās limit, we conducted me-
diated regression analyses within the small-concessions condition,
with the opponentās emotion as the independent variable, the
distance-traveled index as the dependent variable, and appraisal of
the opponentās limit as the mediator.
To establish mediation, the following conditions should be
satisfied (see Baron & Kenny, 1986). First, the independent vari-
able (opponentās emotion) should affect the dependent variable
(distance traveled). Second, the independent variable should affect
the proposed mediator (appraisal of opponentās limit). Third, the
mediator should be correlated with the dependent variable. Fourth,
to establish full mediation, the effect of the independent variable
on the dependent variable when controlling for the mediator
should be nonsignificant. If the effect of the independent variable
on the dependent variable at Step 4 continues to be significant but
is substantially reduced, partial mediation is indicated.
Consistent with the mediational model, we found a significant
regression of the opponentās emotion on distance traveled (
ā¤
ā«½
.57, p
ā¬
.01) and a significant regression of the opponentās
emotion on appraisal of the opponentās limit (
ā¤
ā«½
.40, p
ā¬
.03).
Adding appraisal of the otherās limit as a covariate produced a
significant regression of appraisal of the otherās limit on distance
traveled (
ā¤
ā«½
.57, p
ā¬
.01) and reduced the effect of otherās
emotion on distance traveled to
ā¤
ā«½
.34 ( p
ā¬
.05). These data
indicate that the effect of the opponentās emotion on distance
traveled is partially mediated by appraisal of the opponentās limit
(cf. Baron & Kenny, 1986). Furthermore, the reduction of the
direct path from otherās emotion to distance traveled when con-
trolling for appraisal of otherās limit was significant (z
ā«½
2.67, p
ā¬
.01; see Kenny, Kashy, & Bolger, 1998; for an updated version of
the formula, see D. A. Kennyās Web site at http://users.rcn.com/
dakenny/mediate.htm). Thus, in the small-concessions condition,
appraisal of otherās limit mediated the effects of otherās emotions
on participantās demands. In the moderate and large concessions
conditions, mediation was neither predicted nor found.
Participantsā Emotions and Impression of the Opponent
As in Experiment 1, ANOVA showed a significant effect of the
opponentās emotion on participantsā emotions. Suggestive of so-
cial contagion, participants with a happy opponent felt happier
than did those with an angry opponent (M
ā«½
3.26 vs. M
ā«½
2.73),
F(1, 87)
ā«½
10.84, p
ā¬
.01, and participants with an angry opponent
felt angrier than did those with a happy opponent (M
ā«½
2.30 vs.
M
ā«½
1.62), F(1, 87)
ā«½
13.93, p
ā¬
.01.
ANOVA also revealed a main effect of the opponentās emotion
on the participantās impression of the opponent, F(1, 87)
ā«½
1
In line with this pattern, judgment of the opponentās concession magnitude
was not only influenced by the opponentās concession size (see manipulation
check), but also by the opponentās emotion. Participants with a happy oppo-
nent judged the opponentās concessions to be larger (M
ā«½
3.69) than did those
with an angry opponent (M
ā«½
2.48), F(1, 87)
ā«½
27.71, p
ā¬
.01.
2
We used simple-effects analyses to maintain analytical consistency
across experiments. However, a contrast analysis yielded similar results.
Table 6
Appraisal of the Opponentās Limit and Distance Traveled as a Function of the Opponentās
Emotion and Concession Size (Experiment 2)
Measure
Opponentās concession size
Small
Moderate
Large
Anger
Happiness
Anger
Happiness
Anger
Happiness
Appraisal of otherās limit
8.98
a
7.17
b
8.38
a
8.06
a
7.50
a
7.38
a
Distance traveled
240
a
73
b
219
a
146
b
208
a
166
a
Note.
Means within each of the concession conditions with a different subscript differ at p
ā¬
.05. Appraisal
of the opponentās limit ranged from 1 (indicating a low limit) to 15 (indicating a high limit). Distance traveled
represents the amount of points conceded between the first and the sixth negotiation round.
66
VAN KLEEF, DE DREU, AND MANSTEAD
175.68, p
ā¬
.01, showing that participants developed more posi-
tive impressions of happy opponents than of angry opponents
(M
ā«½
5.02 and M
ā«½
2.80, respectively). We also found a main
effect of the opponentās concession size on impression, F(2, 87)
ā«½
3.88, p
ā¬
.03, which revealed a similar pattern. Participants in the
large-concessions condition developed more positive impressions
of the opponent than did those in the small-concessions condition
(M
ā«½
4.06 and M
ā«½
3.51, respectively), and participants in the
moderate-concessions condition took an intermediate position that
did not differ significantly from the other conditions (M
ā«½
3.82).
There was no interaction between the opponentās emotion and
concession size, F(2, 87)
ā¬
1, ns.
Discussion
The findings of Experiment 2 are consistent with those of
Experiment 1 and provide additional support for the strategic-
choice hypothesis in three ways. First, the results of Experiment 1
were replicated: Participants with an angry opponent placed lower
demands and made smaller concessions than did participants with
a happy opponent. Although the data suggest that participants
ācaughtā their opponentās emotions (i.e., social contagion), the
resulting emotions did not influence their negotiation behavior.
Second, the emotion effect was shown to be moderated by the
opponentās concession size. Participants whose opponents made
small concessions were strongly influenced by the opponentās
emotion, whereas those whose opponents made large concessions
were unaffected. This finding shows that the effect of the oppo-
nentās emotion only occurs when it is in the negotiatorās strategic
interest to act on the otherās emotion. If there is no need to take the
otherās emotion into consideration, the emotion effect is absent.
Third, the effect of the opponentās emotion in the small-
concessions condition was mediated by participantsā appraisals of
the opponentās limits, a finding that is strongly indicative of
tracking. In sum, the results of Experiment 2 provide cumulative
support for the strategic-choice hypothesis and render the social
contagion hypothesis less plausible by showing that it is negotia-
torsā strategic thinking, not their emotional reactions, that is re-
sponsible for the effects of the opponentās emotions on demands
and concession making. Finally, Experiment 2 replicated the find-
ings concerning participantsā impressions of the opponent, show-
ing that participants with a happy opponent developed a more
favorable impression of the opponent than did those with an angry
opponent.
Experiment 3
So far we have focused on the effects of knowledge of the
opponentās experienced emotions on the focal negotiatorās behav-
ior. The strong support for the strategic-choice perspective poses
the question of whether participants in the first two experiments
acted on the otherās genuine emotions or on what they suspected to
be strategic considerations. Although we took great effort to ensure
that participants believed that the opponent did not know that his
or her emotion statements would be revealed to the participant, we
did not explicitly contrast this condition with a situation in which
the opponent knowingly communicated emotion statements to-
ward the recipient. The main objective of Experiment 3 was to
explore this issue in more detail.
Besides the methodological issue noted above, studying gen-
uine emotion in conjunction with communicated emotion is of
theoretical importance. In real life, negotiators are often con-
fronted with their opponentsā communicated emotions, which
do not necessarily reflect their true feeling states. In some
situations people try to suppress their emotions for self-
presentational reasons (see DePaulo, 1992) or because they
believe that it is inappropriate to experience (Hochschild, 1983)
or express (Ekman & Friesen, 1969b) these emotions. In nego-
tiations, individuals may even strategically alter or feign emo-
tions in an attempt to influence their opponents (e.g., Barry,
1999; Thompson et al., 1999).
The fact that negotiators may suppress, hide, or fake emotions is
not to say that the opponentās experienced emotions always remain
unknown. In fact, there are a number of ways in which negotiators
become aware of the emotions their opponent experiences. First,
the opponentās experienced emotion may āleakā from his or her
nonverbal (Ekman & Friesen, 1969a) or vocal (Manstead, Wagner,
& MacDonald, 1984; Scherer, Feldstein, Bond, & Rosenthal,
1985) expressions. Second, a negotiator may learn about the op-
ponentās experienced emotions from secondary sources of infor-
mation, such as conversations with a third party, press conferences,
newspaper articles, and the like. Third, an experienced negotiator
should be able to make an educated guess as to which emotions the
opponent is likely to experience given the particularities of the
situation. Thus, given that negotiators often have information
about both the experienced and the communicated emotions of the
opponent, a logical next step is to investigate how experienced and
knowingly communicated emotions interact to influence the focal
negotiatorās behavior.
A key difference between experienced and knowingly com-
municated emotions is that communicated emotions are di-
rected toward the recipient. Experienced and communicated
emotions can therefore be expected to have different effects on
behavior. In a strategic setting like negotiation, the opponentās
true emotion provides strategically useful information about his
or her degree of satisfaction with the present state of affairs. As
was shown in Experiments 1 and 2, this kind of information is
likely to trigger strategic behavior. In contrast, the opponentās
communicated emotion, which is intentionally directed toward
the focal negotiator, conveys information about the opponentās
social intentions and relational orientation (e.g., Fridlund, 1992;
Knutson, 1996). Because the opponentās communicated emo-
tion is addressed to the participant and therefore has direct
bearing on him or her, and because communicated anger is
often followed by aggression directed toward the target of the
communication (Averill, 1982), communicated anger, more so
than communicated happiness, is likely to induce some degree
of fear in the participant. Indeed, research has documented that
displays of anger evoke fear in observers, especially when the
anger display is directed toward the observer (Dimberg &
O
ĀØ hman, 1983, 1996). Thus, because the opponentās expressions
of anger are knowingly directed at the participant, these expres-
sions can be expected to elicit complementary fear in the
participant.
67
ANGER AND HAPPINESS IN NEGOTIATIONS
There is good evidence that fear depletes cognitive resources
and thereby influences information processing. Fear constrains
attention to those features of the environment that are concerned
with safety or danger (Mathews & MacLeod, 1994). Also, threat
and concomitant stress prompt individuals to offer solutions to
problems without considering all the available information (Giora,
1987) and increase cognitive rigidity and black-and-white thinking
(Carnevale & Probst, 1998). For instance, De Dreu, Giebels, and
Van de Vliert (1998) found that the more threats negotiators
exchanged, the less integrative were their agreements (integrative
agreements require high levels of creative and flexible thinking).
As summarized by O
ĀØ hman (2000), āmildly threatening stimuli
capture attention, independently of its current focus. As attention is
then automatically switched to the threat, there is competition for
processing resourcesā (p. 582). In the current context, this means
that the opponentās angry communications should induce more
fear and anxiety than the opponentās happy communications,
thereby reducing the attention given to the opponentās experienced
emotion. We therefore predicted that the effect of the opponentās
experienced emotion that was observed in Experiments 1 and 2
would be moderated by the opponentās communicated emotion,
such that the influence of the opponentās experienced emotion is
greater when the opponent communicates happiness than when the
opponent communicates anger.
Method
Participants and Experimental Design
Participants were 77 male and female undergraduate students of the
University of Amsterdam who participated for course credit or for mone-
tary compensation (10 Dutch guilders, roughly equivalent to US$4). We
used a 2 (opponentās experienced emotion: anger vs. happiness)
ā«»
2 (op-
ponentās communicated emotion: anger vs. happiness) design. In addition
to demand level, dependent variables were the participantsā emotions, their
impressions of the opponent, their evaluations of the negotiation, and
perceived consistency between the opponentās emotions. Participants were
randomly assigned to the experimental conditions.
Procedure
The procedure was the same as in Experiment 1, with one major
exception. In addition to the opponentās experienced emotion, we also
manipulated the opponentās communicated emotion. Participants were told
that during the negotiation there would be a number of opportunities to
send a message to the opponent and that the opponent would also get the
opportunity to send messages. It was made clear to the participants that the
opponent knew that his or her messages would be sent to the participant.
In the first, third, and fifth negotiation round, participants were given the
opportunity to send the opponent a message, which they could select from
a list of prewritten messages. Participants could also choose not to send a
message.
3
In the second, fourth, and sixth negotiation round, the opponentās offer
was accompanied by a message allegedly selected from the list of prewrit-
ten messages. Participants received either three messages reflecting anger
or three messages reflecting happiness. In the angry messages condition,
the following messages were used (in the second, fourth, and sixth rounds,
respectively): āThis is a ridiculous offer, it really pisses me offā; āI am
starting to get really angryā; and āAll this is starting to get really irritating.ā
In the happy messages condition we used the following messages: āThis is
going pretty well, I canāt complainā; āI like the way things are going, I can
only be happy with thisā; and āI am pretty satisfied with this negotiation.ā
The messages were pretested in a way similar to the emotional statements
used in Experiment 1. All messages were rated higher on the emotion they
were supposed to express than on the emotion they were not supposed to
express (all ts
ā¬
14, all ps
ā¬
.01). Further, there was a significant effect of
all messages on the rating of the corresponding emotion (all ts
ā¬
12, all
ps
ā¬
.01), and the messages did not differ with respect to comprehensi-
bility (all ts
ā¬
1, ns).
As in Experiments 1 and 2, the opponent made the first offer. Partici-
pants then responded with a demand and, if they chose to do so, with a
message. Then, the opponentās experienced emotion was revealed, and the
opponent made a new offer. In the second round, the participant posed a
new demand, which was followed by a new offer by the opponent,
accompanied by either an angry or a happy message. In the third and fifth
round, participants again received information about the opponentās expe-
rienced emotion, and in the fourth and sixth round they received an angry
or a happy message. The negotiation was interrupted after six rounds, and
participants who reached consensus prior to the sixth round were dropped
from the analyses (n
ā«½
9; including these participants in the analyses did
not change the pattern of results).
The manipulation check items for the manipulation of the opponentās
experienced emotions were the same as those used in the previous exper-
iments, with the explicit instruction that participants should base their
judgment purely on the opponentās stated intentions (which contained
information about the opponentās experienced emotion). The same manip-
ulation check was also used for the manipulation of the opponentās com-
municated emotion but with the instruction that judgments should be based
purely on the opponentās messages (which included the opponentās com-
municated emotion). In addition, participants answered two questions
designed to check whether they correctly remembered the instructions
pertaining to the two types of information they received about their
opponent. Recall that participants were told that the opponent knew that his
or her messages would be sent to the participant but did not know that his
or her intentions would be revealed to the participant. To check partici-
pantsā recollection of these instructions, they were asked to indicate on a
5-point scale if they thought that the opponent knew that his or her
intentions were revealed to the participant and if they thought that the
opponent knew that his or her messages would be sent to the participant
(for both questions: 1
ā«½
definitely not, 5
ā«½
definitely).
After the manipulation checks, participants completed a questionnaire
that contained items designed to measure participantsā own emotions,
participantsā impressions of the opponent, and participantsā perceptions of
the consistency between the opponentās experienced and communicated
emotions. Participantsā emotions were assessed in the same way as in
Experiment 1. However, we added three items measuring participantsā fear
(e.g., āTo what extent did you experience fear during the negotiation?ā;
1
ā«½
not at all, 5
ā«½
to a great extent), which were combined into a single
index of fear (
ā£
ā«½
.67). Participantsā impressions of the opponent were
assessed by means of six items (e.g., āI have developed a positive impres-
sion of the buyerā; 1
ā«½
totally disagree, 7
ā«½
totally agree), which were
averaged into a single index of impression of the opponent (
ā£
ā«½
.89).
Perceived consistency between the opponentās experienced and communi-
cated emotions was measured with three items (e.g., āThe buyerās mes-
sages were consistent with his or her intentionsā; 1
ā«½
definitely not, 5
ā«½
definitely), which were combined into a single index of consistency (
ā£
ā«½
.92).
3
Analysis of participantsā messages revealed no significant effects and
thus is not discussed any further.
68
VAN KLEEF, DE DREU, AND MANSTEAD
Results
Manipulation Checks
The manipulation of the opponentās experienced emotion was
checked in the same way as in the previous experiments. Again,
ANOVA revealed a significant two-way interaction between the
opponentās experienced emotion and the participantās perception
of the opponentās emotion, F(1, 66)
ā«½
264.02, p
ā¬
.01. As
expected, participants in the angry opponent condition rated the
opponent as more angry (M
ā«½
6.15) than did participants in the
happy opponent condition (M
ā«½
2.23), t(66)
ā«½
12.93, p
ā¬
.01.
Similarly, participants in the happy opponent condition rated the
opponent as happier (M
ā«½
5.32) than did participants in the angry
opponent condition (M
ā«½
1.43), t(66)
ā«½
15.60, p
ā¬
.01. Further,
paired-sample t tests showed that participants in the angry oppo-
nent condition rated the opponent as more angry than happy (M
ā«½
6.15 vs. M
ā«½
1.43), t(33)
ā«½
18.69, p
ā¬
.01, and participants in the
happy opponent condition rated the opponent as more happy than
angry (M
ā«½
5.32 vs. M
ā«½
2.23), t(33)
ā«½
7.54, p
ā¬
.01.
Regarding the manipulation of the opponentās communicated
emotion, ANOVA revealed an interaction between the opponentās
communicated emotion and the participantās perception of the
opponentās communicated emotion, F(1, 66)
ā«½
225.33, p
ā¬
.01.
Follow-up t tests indicated that participants who received angry
messages perceived the opponent as angrier (M
ā«½
5.60) than did
those who received happy messages (M
ā«½
1.72), t(66)
ā«½
12.15,
p
ā¬
.01. Similarly, participants who received happy messages
rated the opponent as happier (M
ā«½
5.63) than did those who
received angry messages (M
ā«½
1.96), t(66)
ā«½
14.53, p
ā¬
.01.
Finally, participants who received angry messages rated these
messages as more angry than happy (M
ā«½
5.60 vs. M
ā«½
1.96),
t(33)
ā«½
9.67, p
ā¬
.01, and participants who received happy
messages rated these messages as more happy than angry (M
ā«½
5.63 vs. M
ā«½
1.72), t(33)
ā«½
11.73, p
ā¬
.01.
Additional analyses were performed to check whether partici-
pants accurately remembered the instructions pertaining to the two
types of information they received about their opponentās emo-
tions. A one-sample t test showed that participants correctly re-
called that their opponent did not know that his or her intentions
(i.e., experienced emotions) would be revealed to the participant
(M
ā«½
1.24), t(67)
ā«½
34.05, p
ā¬
.01. Participants also correctly
remembered that the opponent did know that his or her messages
(i.e., communicated emotions) would be sent to the participant
(M
ā«½
4.54), t(67)
ā«½
14.33, p
ā¬
.01.
Finally, we analyzed the perceived consistency between the
opponentās experienced and communicated emotions. ANOVA
yielded a two-way interaction between the opponentās experienced
emotion and the opponentās communicated emotion, F(1, 64)
ā«½
88.32, p
ā¬
.01. Unsurprisingly, participants whose opponents
communicated emotions that did not correspond with their expe-
rienced emotions reported more inconsistency between the inten-
tions and the messages (M
ā«½
1.26 and M
ā«½
1.75) than did those
whose opponents communicated emotions that did correspond
(M
ā«½
3.56 and M
ā«½
3.96). These data provide additional support
for the effectiveness of the manipulation of the opponentās expe-
rienced and communicated emotions.
Demand Level
Demands in Rounds 1ā 6 were submitted to a 2 (opponentās
experienced emotion: anger vs. happiness)
ā«»
2 (opponentās com-
municated emotion: anger vs. happiness) mixed-model ANOVA
with negotiation round as a repeated-measures variable. First, this
analysis showed a main effect of negotiation round, F(5, 325)
ā«½
136.60, p
ā¬
.01, indicating that demands declined over time
(Round 1, M
ā«½
630 to Round 6, M
ā«½
442). Furthermore, we found
the expected interaction between the opponentās experienced and
communicated emotions, F(1, 64)
ā«½
4.15, p
ā¬
.05. Simple-effects
analyses indicated that as predicted, the opponentās experienced
emotion had no effect when the opponent sent angry messages
(M
ā«½
532 for experienced anger and M
ā«½
522 for experienced
happiness), F(1, 64)
ā¬
1, ns. When the opponent sent happy
messages, however, the opponentās experienced emotion did in-
fluence demands, F(1, 64)
ā«½
6.50, p
ā¬
.02. As in Experiments 1
and 2, participants placed lower demands when the opponent
experienced anger rather than happiness (M
ā«½
465 vs. M
ā«½
541).
Results also revealed an interaction between the opponentās
emotion and negotiation round, F(5, 320)
ā«½
2.83, p
ā¬
.02. Thus,
consistent with the results from the first two experiments, the
effect of the opponentās experienced emotion increased over time.
Although the three-way interaction between the opponentās expe-
rienced emotion, communicated emotion, and negotiation round
did not reach statistical significance, F(5, 320)
ā«½
1.30, ns, simple-
effects analyses yielded results that are fully consistent with those
of Experiments 1 and 2 (see Figure 3). As predicted, we obtained
a significant interaction between the opponentās experienced emo-
tion and negotiation round for participants who received happy
messages, F(5, 320)
ā«½
3.49, p
ā¬
.01, but not for participants who
received angry messages, F(5, 320)
ā¬
1, ns.
Participantsā Emotions and Impression of the Opponent
As predicted, ANOVA showed that participants who received
angry messages from the opponent reported more fear than did
those who received happy messages (M
ā«½
2.91 vs. M
ā«½
1.51,
respectively), F(1, 64)
ā«½
5.77, p
ā¬
.02, whereas there was no
effect of the opponentās experienced emotion, F(1, 64)
ā¬
1, ns. In
contrast to the previous experiments, participantsā self-reported
anger and happiness were not influenced by the opponentās emo-
tions. However, this is not surprising when we take into consid-
eration that half of the participants in Experiment 3 were con-
fronted
with
an
opponent
who
experienced
anger
but
communicated happiness, or vice versa. This explanation is sup-
ported by a marginally significant interaction between experienced
and communicated emotion on participantsā self-reported anger
and happiness, which shows that participantsā own anger and
happiness tend to be influenced by the opponentās emotions when
the opponentās experienced and communicated emotions are con-
sistent but not when they are inconsistent, F(1, 64)
ā«½
2.63, p
ā«½
.08.
In line with the results from Experiments 1 and 2, participantsā
impressions of the opponent were influenced by the opponentās
experienced emotion. Participants who negotiated with a happy
opponent developed a more positive impression of the opponent
69
ANGER AND HAPPINESS IN NEGOTIATIONS
(M
ā«½
4.69) than did those who negotiated with an angry opponent
(M
ā«½
2.07), F(1, 64)
ā«½
147.88, p
ā¬
.01. Impressions of the
opponent were also influenced by the opponentās communicated
emotions. Participants who received happy messages developed a
more favorable impression of the opponent (M
ā«½
3.92) than did
those who received angry messages (M
ā«½
2.84), F(1, 64)
ā«½
37.59,
p
ā¬
.01. Further, an additive interaction between experienced and
communicated emotion showed that the most unfavorable impres-
sions were made by opponents who were angry and also commu-
nicated anger (M
ā«½
1.59), whereas the most favorable impressions
were made by opponents who were happy and also communicated
happiness (M
ā«½
5.77), F(1, 64)
ā«½
5.45, p
ā¬
.03. The other two
cells took intermediate positions that differed from both extremes
(M
ā«½
2.46 for experienced anger/communicated happiness, and
M
ā«½
3.83 for experienced happiness/communicated anger).
Mediation Analysis
We predicted that the moderating effect of the opponentās
communicated emotion would be due to increased fear in partic-
ipants who received angry messages. Thus, we should find that
communicated emotion moderates the effect of experienced emo-
tion on demands through its effect on fear. According to Hull,
Tedlie, and Lehn (1992), if the original predictor (communicated
emotion) is interactively related to the dependent variable (de-
mands), then the proposed mediator (fear) will also be interactively
related to the dependent variable. The proper control is then
formed by taking the ācovariate interactionā between the mediator
(fear) and the independent variable (experienced emotion). To
establish mediation, we should find a main effect of communicated
emotion on fear and an interaction between experienced and com-
municated emotion on demands. Both of these requirements were
met (see above). Further, the interaction between experienced and
communicated emotion should disappear when controlling for the
covariate interaction. To test whether this was the case, we con-
ducted an analysis of covariance in which the interactive effect of
experienced emotion and communicated emotion on level of de-
mand was computed after the covariate interaction (Experienced
Emotion
ā«»
Fear) had been controlled for (see Hull et al., 1992;
Stevens, 1996). As expected, when controlling for this interaction,
the originally significant interaction between experienced emotion
and communicated emotion was reduced to nonsignificance, F(1,
63)
ā«½
1.80, ns, and a significant main effect of experienced
emotion emerged, F(1, 63)
ā«½
5.09, p
ā¬
.03, with means ordered
in the same pattern as in Experiments 1 and 2. These results
support the conclusion that the moderating effect of communicated
emotion is due to increased levels of fear in participants who
received angry rather than happy messages.
Discussion
The results of Experiment 3 corroborate our hypotheses. As
predicted, participants who received happy messages were
strongly influenced by their opponentās experienced emotions,
those whose opponents experienced anger placing lower demands
than those whose opponents experienced happiness. However, this
effect was not obtained for participants who received angry com-
munications. Furthermore, a mediation analysis showed that the
moderating effect of the opponentās communicated emotion was
caused by increased levels of fear in participants who received
angry communications. Although the results pertaining to demands
and concessions are consistent with our predictions, it is important
to consider an alternative account for these findings. Our data
show that participants conceded most to opponents who experi-
enced anger and expressed happiness. The reader might argue that
participants in this condition were exposed to inconsistent emo-
tions and that our effects result from individualsā responses to such
inconsistencies. Below, we consider two alternative explanations
that stem from the notion of inconsistency between emotions, one
pertaining to emotional contrast effects, the other to the strategic
use of emotion in negotiation.
With regard to contrast effects, research has documented that
judgments are relative and that the context in which a stimulus is
presented serves as a reference point against which the stimulus is
Figure 3.
Demand level as a function of the opponentās experienced
emotion, the opponentās communicated emotion, and negotiation round in
Experiment 3.
70
VAN KLEEF, DE DREU, AND MANSTEAD
compared and judged (e.g., Bazerman, 1990; Eiser, 1990). For
example, Helson (1947) showed that weights are judged lighter to
the extent that the stimulus context consists of heavier weights.
Similar contrast effects have been shown in the domain of emo-
tions (Manstead, Wagner, & MacDonald, 1983; Thayer, 1980),
and Rafaeli and Sutton (1991) argued that such emotional contrast
effects may be effectively used as a tool to elicit compliance.
Although at first sight the results of Experiment 3 may seem to
reflect a contrast effect, it is important to note that an emotional
contrast perspective would imply a symmetrical rather than an
asymmetrical effect. That is, if the emotional contrast explanation
were viable, we should have observed lower demands both in the
experienced anger/communicated happiness and in the experi-
enced happiness/communicated anger conditions. However, our
data show that negotiators yielded more to an opponent who felt
angry and expressed happiness than to an opponent who felt happy
and expressed anger.
A second alternative account stems from negotiatorsā possible
responses to the opponentās perceived emotional deception. In
negotiations, permission to mislead is to some degree implied
(Carr, 1968; Ekman, 2001), and people regard the strategic use of
emotion as more ethically acceptable than other forms of deliber-
ate deception, such as misrepresentation of factual information or
the making of false promises (Barry, 1999). Thus, negotiators
dealing with an opponent who is known to be happy but expresses
anger might discount the otherās expressed anger as a strategic
ploy and decide not to give in. In fact, the discovery of deception
in negotiation has been shown to increase disagreements (Roth &
Murnighan, 1983; Schweitzer & Croson, 1999) and to lead nego-
tiators to reject offers that they would have accepted had they not
discovered the deceit (Boles, Croson, & Murnighan, 2000), sug-
gesting that a negotiatorās response to a happy opponent who
expresses anger would consist of higher rather than lower de-
mands. Conversely, the combination of experienced anger and
expressed happiness does not make sense from a strategic view-
point, and as such the expressed happiness may be seen as a
validation of the experienced anger, leading negotiators to concede
to an angry opponent who expresses happiness.
4
Although this explanation can account nicely for the fact that
negotiators do not make large concessions to an opponent who
experiences happiness but expresses anger, it cannot account for
the fact that negotiators concede more to an opponent who expe-
riences anger and expresses happiness than to an opponent who
experiences anger and also expresses anger. An explanation in
terms of fear and information processing can account for this
finding: The opponentās expressions of anger elicit fear in the focal
negotiator, leading him or her to pay less attention to the otherās
experienced emotions and consequently to be less influenced by
them. This explanation is not only compatible with the results
obtained in Experiment 3 but is also consistent with the results of
Experiment 2, which show that negotiators are only affected by
their opponentās emotions if they are motivated to consider them.
In short, we believe that an explanation in terms of fear and
information processing can most parsimoniously account for our
pattern of results. Nevertheless, future research is needed to de-
velop a better understanding of the combined effects of experi-
enced and expressed emotions on negotiation behavior.
General Discussion
Recently researchers have started to address the role of emotions
in negotiation, focusing on the question of how an individualās
affect influences his or her own information processing and be-
havior (i.e., intrapersonal effects) in negotiation. The present re-
search is the first to empirically address the effects of one nego-
tiatorās emotions on the other negotiatorās behavior (i.e.,
interpersonal effects). Experiment 1 showed that anger leads to
lower demands and larger concessions whereas happiness leads to
higher demands and smaller concessions. Experiment 2 demon-
strated that this effect is caused by tracking (cf. Pruitt, 1981) and
that the effect is mitigated if the opponent undermines the moti-
vation to take his or her emotions into consideration by making
large concessions. Finally, Experiment 3 showed that the effect is
removed when the opponent sends angry rather than happy mes-
sages, because angry messages induce fear in the receiver and
distract the receiverās attention from the opponentās experienced
emotion. These findings have a number of important theoretical
and practical implications, which are discussed below.
Implications and Contributions
In the introduction we used two distinct theoretical perspectives
to develop competing hypotheses regarding the interpersonal ef-
fects of anger and happiness in negotiations. The social contagion
hypothesis posits that negotiators with an angry opponent become
angry and, therefore, place high demands, whereas those with a
happy opponent become happy and place low demands. In con-
trast, the strategic-choice hypothesis posits that negotiators mis-
match the opponentās emotions, predicting that those with an angry
counterpart make lower demands than those with a happy coun-
terpart. Although our results provide support for social contagion
(i.e., anger induced anger and happiness induced happiness), this
contagion did not affect negotiation behavior. That is, we did not
replicate previous research showing that happy negotiators place
lower demands and concede more than do angry negotiators (cf.
Baron, 1990; Carnevale & Isen, 1986; Forgas, 1998). Rather, we
obtained strong support for the strategic-choice hypothesis in all
three experiments: Negotiators with an angry opponent made
lower demands and larger concessions than did negotiators with a
happy opponent.
There are several explanations for the apparent inconsistency
between the results for demands and concessions (supporting the
strategic-choice perspective) on the one hand and the results for
self-reported emotion (supporting the social contagion perspec-
tive) on the other hand. For example, one could argue that partic-
ipantsā emotions resulted from self-perception of their own behav-
ior rather than from the opponentās emotions. That is, the anger
that was reported by participants with an angry opponent may have
originated from frustration because of the relatively large conces-
sions they had made during the negotiation (moving them away
from their goal) rather than from the opponentās anger per se.
Similarly, participantsā happiness may have resulted from their
own small concessions (bringing them closer to their goal) rather
4
We thank an anonymous reviewer for suggesting this possibility.
71
ANGER AND HAPPINESS IN NEGOTIATIONS
than from the opponentās happiness. However, multivariate anal-
yses of covariance with the opponentās emotion as the independent
variable, the participantās anger and happiness as the dependent
variables, and the participantās average demands as a covariate
provided no support for such an explanation.
5
Another explanation is that social contagion did occur but was
overruled by strategic considerations. Research has shown that
mismatching is especially likely to occur when negotiators lack
information about the opponentās outcomes (Liebert et al., 1968;
Yukl, 1974b) and limits (Pruitt & Syna, 1985). Because this was
the case in our experiments, the tendency to mismatch the oppo-
nentās perceived toughness (in the case of anger) or weakness (in
the case of happiness) might have been especially strong. Thus,
pending further research, it can be cautiously concluded that social
contagion may affect self-reported emotions in computer-mediated
negotiation settings but that this does not translate into behavior.
Instead, it appears as if two distinct processes emerge when ne-
gotiators receive information about their adversariesā emotion: The
information influences their own emotional state through
contagion-like processes, and at the same time, they strategically
use the information to track the opponentās limit and to make
demands.
Although the specific roles played by contagion and strategic
choice must await further research, the present research adds to the
knowledge of negotiation, emotion, and computer-mediated com-
munication in a number of ways. It enhances understanding of the
negotiation process by showing that one negotiatorās emotions
influence another negotiatorās behavior, thereby underlining once
again the importance of motivational factors in negotiation. It also
contributes to the literature on mismatching. Early research on
mismatching (e.g., Bartos, 1974; Chertkoff & Conley, 1967;
Druckman et al., 1972; Komorita & Brenner, 1968; Liebert et al.,
1968; Pruitt & Syna, 1985; Smith et al., 1982; Yukl, 1974a, 1974b)
specifically focused on the influence of the opponentās demands
and concessions on the focal negotiatorās demands and conces-
sions and showed that they tend to be inversely related: Negotia-
tors respond with high demands to the opponentās low demands
and with low demands to the opponentās high demands. More
recent research has shown that it is not the opponentās concessions
per se that cause mismatching but rather negotiatorsā interpretation
of the opponentās behavior (De Dreu et al., 1994). The present
results are in line with this conclusion and suggest that the emo-
tions that accompany the opponentās demands modify the focal
negotiatorās interpretation of the opponentās behavior. Indeed, the
data of Experiment 2 show that negotiators perceived happy op-
ponents as more conciliatory than angry opponents.
The current findings also extend knowledge of the social effects
of emotions by showing that emotions have interpersonal effects
on negotiation behavior. Furthermore, they show that privately
experienced and knowingly communicated emotions may have
quite different effects on peopleās behavior when they are both
presented at the same time, at least in negotiation settings. It
appears that in negotiations, experienced emotions have an infor-
mational function: They provide background information, which
serves as input for negotiatorsā strategic decision making. Com-
municated emotions, on the other hand, seem to serve a social
function: They are intentionally directed toward the recipient and
hence provide information about the status of the interpersonal
relation and the senderās social intentions (cf. Fridlund, 1992;
Knutson, 1996). Future research might investigate whether this
difference between experienced and communicated emotions also
pertains in situations where people are not dependent on one
another for their outcomes.
The results of Experiment 2 suggest that negotiators only use the
opponentās emotions to track his or her limit when the opponent
uses a tough strategy. Accordingly, their demands and concessions
are affected by the opponentās emotions when the opponent makes
small concessions but not when the opponent makes large conces-
sions. This pattern of results suggests that negotiators consider the
otherās emotions especially when they are motivated to do so, for
example because the otherās behavior is not very conducive to
reaching an agreement that is favorable to oneself. The results of
Experiment 3 suggest that when the opponent sends angry mes-
sages, fear increases, and less attention is given to the otherās
experienced emotions. This pattern of results suggests that the
otherās experienced emotions have an impact especially when
negotiators allocate cognitive resources to consider them. To-
gether, these results seem to indicate that the interpersonal effects
of otherās experienced emotions in negotiation settings occur when
the focal negotiator (a) is motivated to consider the otherās emo-
tion and (b) has the cognitive resources available to consider them.
The current findings also have interesting implications for the-
orizing in the domain of computer-mediated communication. More
and more interactions occur via computer-mediated media such as
the internet or e-mail, and social and organizational decisions are
increasingly being made through negotiations conducted via such
technological media (Moore et al., 1999). Computer-mediated
communication differs from face-to-face communication in a num-
ber of ways. For example, it has been argued that interactions via
communication media such as e-mail are characterized by a higher
frequency of counternormative social behaviors (Kiesler &
Sproull, 1992), and individuals engaged in computer-mediated
communication are more likely to display emotions that in other
settings might be masked in socially appropriate ways (Thompson
& Nadler, 2002). The present research contributes to this literature
by showing that the emotions that arise in computer-mediated
communication can have profound effects on the subsequent in-
teraction and on the decisions that are made.
Finally, the present research has obvious practical implications.
Although professional negotiators and lay people alike tend to
have strong beliefs about how emotions should be managed in
negotiations, these beliefs have never been systematically tested,
and the effects of emotions in negotiations are incompletely un-
derstood. Our findings point to the impact of anger and happiness
on negotiations and have important implications for negotiation
and conflict resolution. Moreover, the relevance of the current
findings is unlikely to be limited to the negotiation setting and may
5
In Experiment 1, the effect of the opponentās emotion on participantsā
happiness was slightly reduced from F(2, 115)
ā«½
3.36, p
ā¬
.04, to F(2,
114)
ā«½
2.85, p
ā«½
.06, when average demand was added as a covariate, and
the effect on participantsā anger was reduced from F(2, 115)
ā«½
8.01, p
ā¬
.01, to F(2, 114)
ā«½
5.53, p
ā¬
.01. However, in Experiments 2 and 3 the
effect of the opponentās emotion on participantsā emotions was unaffected
when average demand was included as a covariate.
72
VAN KLEEF, DE DREU, AND MANSTEAD
well generalize to other domains of social interaction such as
leadership, politics, personal relationships, and, more generally,
persuasion and compliance.
Limitations and Directions for Future Research
There are some limitations to these findings. First and foremost,
there was no face-to-face interaction. As discussed in the intro-
duction to Experiment 1, we made an explicit decision to maintain
experimental control at the expense of mundane realism, and we
chose to use a computer-mediated negotiation paradigm to permit
a carefully controlled manipulation of the opponentās emotion. As
a result, some caution is needed when generalizing the results. At
the very least, our findings pertain to computer-mediated negoti-
ations. Given the pervasiveness of negotiation as a form of social
interaction and the increasing popularity of information technolo-
gies as communication mediums, the question of how individuals
react to each otherās emotions in computer-mediated communica-
tion is of great theoretical and practical importance (cf. McKersie
& Fonstad, 1997; Moore et al., 1999). However, we have no reason
to suspect that our findings are restricted to the domain of
computer-mediated interaction. Future research could shed more
light on this issue by investigating the extent to which the inter-
personal effects of emotions and the underlying processes identi-
fied in the present research generalize across settings.
A related issue concerns the magnitude of the effects obtained in
the present research. One could argue that the fact that participants
in our experiments were only provided with information about the
opponentās emotional state had the effect of focusing their atten-
tion on that information. If so, our results might overestimate the
effects of emotions in real negotiations. As Moore et al. (1999) put
it, āAlthough purely written communication media severely limit
the opportunities for affective displays, opportunities for mutually
reinforcing displays of emotion are not eliminated. Indeed, such
displays, where they occur, may take on heightened importance in
an impoverished medium like e-mailā (p. 26). This possibility
cannot be ruled out on the basis of our data. In fact, the results of
Experiments 2 and 3 suggest that attention and information pro-
cessing play an important role in determining the (magnitude of)
the interpersonal effects of emotions in negotiation, suggesting that
the effects might be smaller if larger quantities of information have
to be attended to and processed. However, we believe that the
possible overestimation of the effect stemming from high levels of
attention to emotional stimuli in the present experiments might be
offset in more naturalistic negotiations by high levels of involve-
ment, which could lead negotiators to be more strongly influenced
by othersā emotions. Furthermore, it is possible that emotions have
stronger effects in face-to-face interaction than they do in
computer-mediated interaction, because in the former case the
emotional information is conveyed nonverbally as well as verbally.
Another issue is whether the opponentās emotion is justified in
the focal negotiatorās mind. In the present experiments, the oppo-
nentās emotion was not made contingent on the participantās
demands, because we wanted to conduct a carefully controlled and
unconfounded test of our hypotheses. Thus, in some cases the
opponentās anger or happiness may have struck participants as
more reasonable than in other cases, depending on participantsā
own demands. This raises the question of how justification of the
otherās emotions influences the focal negotiatorās behavioral reac-
tions to those emotions. For the same methodological reasons, the
opponentās offers and emotions were not contingent on one an-
other. Rather, the opponentās offers were the same in all condi-
tions. Thus, we cannot tell how correspondence (or lack thereof) of
the opponentās offers and emotions influenced the focal negotia-
torās demands. These questions could be addressed in future
studies.
Future research could also investigate the influence of individ-
ual difference variables relating to emotional expressiveness and
attention to emotions. Research by Friedman and Riggio (1981)
has suggested that individual differences in nonverbal emotional
expressiveness influence the spread of emotions. In a somewhat
related vein, Gasper and Clore (2000) showed that individuals high
in attention to emotion tend to perceive their affect as relevant and
to use it in their judgments, whereas individuals low in attention to
emotion tend not to focus on affect and not perceive it as relevant
to their judgments. Both emotional expressiveness and attention to
emotion are related to the concept of emotional intelligenceāāthe
ability to monitor oneās own and othersā feelings and emotions, to
discriminate among them, and to use this information to guide
oneās thinking and actionsā (Salovey & Mayer, 1990, p. 189). We
believe that incorporating individual differences in emotional in-
telligence in future research could enhance understanding of the
role of emotion in negotiation and, more generally, in social
interaction.
Another avenue for future study concerns the long-term conse-
quences of anger. How does anger influence the relationship
between the negotiators? Do the effects of anger persist over time,
or do they diminish or even backfire in the long run? The present
findings show that angry negotiators are evaluated less favorably
than are happy negotiators, suggesting that anger may harm the
interpersonal relationship. These findings point to an interesting
dilemma facing negotiators who anticipate future interaction. On
the one hand, negotiators may be motivated to strategically present
happiness in order to make a good impression and to induce or
maintain a positive interpersonal relationship. On the other hand,
they may choose to use anger to get their opponents to go along
with their preferences. Future research could investigate which of
these strategies is more beneficial in the long run.
Conclusion
Complementing previous research on mismatching and the so-
cial effects of emotions, these three experiments show that anger
on the part of the opponent elicits compliance in the focal nego-
tiator, whereas happiness elicits exploitation. Consistent with a
strategic-choice perspective, this effect was shown to be caused by
tracking. Negotiators concede more to an angry opponent than to
a happy one because they believe the former to have a higher limit
than the latter, which increases the need to make concessions in
order to avoid impasse. This effect disappears when the focal
negotiatorās ability or motivation to consider the information that
is provided by the opponentās emotions is reduced. Together, these
experiments show that emotions have pervasive effects on nego-
tiation behavior and outcomes, especially when negotiators are
both motivated and able to act on the information available in the
opponentās emotions.
73
ANGER AND HAPPINESS IN NEGOTIATIONS
References
Adler, R. S., Rosen, B., & Silverstein, E. M. (1998). Emotions in negoti-
ation: How to manage fear and anger. Negotiation Journal, 14, 161ā180.
Allred, K. G., Mallozzi, J. S., Matsui, F., & Raia, C. P. (1997). The
influence of anger and compassion on negotiation performance. Orga-
nizational Behavior and Human Decision Processes, 70, 175ā187.
Averill, J. R. (1982). Anger and aggression. New York: Springer-Verlag.
Banse, R., & Scherer, K. R. (1996). Acoustic profiles in vocal emotion
expression. Journal of Personality and Social Psychology, 70, 614 ā 636.
Baron, R. A. (1990). Environmentally induced positive affect: Its impact
on self-efficacy, task performance, negotiation, and conflict. Journal of
Applied Social Psychology, 20, 368 ā384.
Baron, R. A., Fortin, S. P., Frei, R. L., Hauver, L. A., & Shack, M. L.
(1990). Reducing organizational conflict: The role of socially-induced
positive affect. International Journal of Conflict Management, 1, 133ā
152.
Baron, R. M., & Kenny, D. A. (1986). The moderatorāmediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173ā1182.
Barry, B. (1999). The tactical use of emotion in negotiation. Research on
Negotiation in Organizations, 7, 93ā121.
Barry, B., Fulmer, I. S., & Van Kleef, G. A. (in press). I laughed, I cried,
I settled: The role of emotion in negotiation. In M. Gelfand & J. Brett
(Eds.), Negotiation: Theoretical advances and cross-cultural perspec-
tives. Stanford, CA: Stanford University Press.
Barry, B., & Oliver, R. L. (1996). Affect in dyadic negotiation: A model
and propositions. Organizational Behavior and Human Decision Pro-
cesses, 67, 127ā143.
Bartos, O. J. (1974). Process and outcome in negotiation. New York:
Columbia University Press.
Bateman, T. S. (1980). Contingent concession strategies in dyadic bargain-
ing. Organizational Behavior and Human Decision Processes, 26, 212ā
221.
Bazerman, M. H. (1990). Judgment in managerial decision making. New
York: Wiley.
Boles, T. L., Croson, R. T. A., & Murnighan, J. K. (2000). Deception and
retribution in repeated ultimatum bargaining. Organizational Behavior
and Human Decision Processes, 83, 235ā259.
Carnevale, P. J., & Isen, A. M. (1986). The influence of positive affect and
visual access on the discovery of integrative solutions in bilateral nego-
tiation. Organizational Behavior and Human Decision Processes, 37,
1ā13.
Carnevale, P. J., & Probst, T. M. (1998). Social values and social conflict
in creative problem solving and categorization. Journal of Personality
and Social Psychology, 74, 1300 ā1309.
Carr, A. Z. (1968, JanuaryāFebruary). Is business bluffing ethical? Har-
vard Business Review, 1, 143ā153.
Chertkoff, J. M., & Baird, S. L. (1971). Applicability of the big lie
technique and the last clear chance doctrine to bargaining. Journal of
Personality and Social Psychology, 20, 298 ā303.
Chertkoff, J. M., & Conley, M. (1967). Opening offer and frequency of
concessions as bargaining strategies. Journal of Personality and Social
Psychology, 7, 181ā185.
Daly, J. P. (1991). The effects of anger on negotiations over mergers and
acquisitions. Negotiation Journal, 7, 31ā39.
Davidson, M. N., & Greenhalgh, L. (1999). The role of emotion in
negotiation: The impact of anger and race. Research on Negotiation in
Organizations, 7, 3ā26.
De Dreu, C. K. W., Carnevale, P. J., Emans, B. J. M., & Van de Vliert, E.
(1994). Effects of gaināloss frames in negotiation: Loss aversion, mis-
matching, and frame adoption. Organizational Behavior and Human
Decision Processes, 60, 90 ā107.
De Dreu, C. K. W., Giebels, E., & Van de Vliert, E. (1998). Social motives
and trust in integrative negotiation: The disruptive effects of punitive
capability. Journal of Applied Psychology, 83, 408 ā 422.
De Dreu, C. K. W., Koole, S. L., & Oldersma, F. L. (1999). On the seizing
and freezing of negotiator inferences: Need for cognitive closure mod-
erates the use of heuristics in negotiation. Personality and Social Psy-
chology Bulletin, 25, 348 ā362.
De Dreu, C. K. W., & Van Kleef, G. A. (in press). The influence of power
on the information search, impression formation, and demands in nego-
tiation. Journal of Experimental Social Psychology.
De Dreu, C. K. W., & Van Lange, P. A. M. (1995). The impact of social
value orientations on negotiator cognition and behavior. Personality and
Social Psychology Bulletin, 21, 1178 ā1188.
DePaulo, B. M. (1992). Nonverbal behavior and self-presentation. Psycho-
logical Bulletin, 111, 203ā243.
Dimberg, U., & O
ĀØ hman, A. (1983). The effects of directional facial cues on
electrodermal conditioning to facial stimuli. Psychophysiology, 20, 160 ā
167.
Dimberg, U., & O
ĀØ hman, A. (1996). Behold the wrath: Psychophysiological
responses to facial stimuli. Motivation and Emotion, 20, 149 ā182.
Druckman, D., & Bonoma, T. V. (1976). Determinants of bargaining
behavior in a bilateral monopoly situation: II. Opponentās concession
rate and similarity. Behavioral Science, 21, 252ā262.
Druckman, D., Zechmeister, K., & Solomon, D. (1972). Determinants of
bargaining behavior in a bilateral monopoly situation: Opponentās con-
cession rate and relative defensibility. Behavioral Science, 17, 514 ā531.
Eiser, J. R. (1990). Social judgment. Buckingham, England: Open Univer-
sity Press.
Ekman, P. (1993). Facial expression and emotion. American Psychologist,
48, 384 ā392.
Ekman, P. (2001). Telling lies: Clues to deceit in the marketplace, politics,
and marriage. New York: Norton.
Ekman, P., & Friesen, W. V. (1969a). Nonverbal leakage and clues to
deception. Psychiatry: Journal for the Study of Interpersonal Processes,
32, 88 ā106.
Ekman, P., & Friesen, W. V. (1969b). The repertoire of nonverbal behav-
iorā categories, origins, usage, and coding. Semiotica, 1, 49 ā98.
Ekman, P., Friesen, W. V., & Ellsworth, P. (1972). Emotion in the human
face: Guidelines for research and an integration of findings. London:
Pergamon Press.
Forgas, J. P. (1998). On feeling good and getting your way: Mood effects
on negotiator cognition and behavior. Journal of Personality and Social
Psychology, 74, 565ā577.
Fridlund, A. J. (1992). The behavioral ecology and sociality of human
faces. Review of Personality and Social Psychology, 13, 90 ā121.
Fridlund, A. J. (1994). Human facial expression: An evolutionary view. San
Diego, CA: Academic Press.
Friedman, H. S., & Riggio, R. E. (1981). Effect of individual differences in
nonverbal expressiveness on transmission of emotion. Journal of Non-
verbal Behavior, 6, 96 ā104.
Frijda, N. H. (1993). Moods, emotion episodes, and emotions. In M. Lewis
& J. M. Haviland (Eds.), Handbook of emotions (pp. 381ā 403). New
York: Guilford Press.
Frijda, N. H., & Mesquita, B. (1994). The social roles and functions of
emotions. In S. Kitayama & H. S. Markus (Eds.), Emotion and culture:
Empirical studies of mutual influence (pp. 51ā 87). Washington, DC:
American Psychological Association.
Gasper, K., & Clore, G. L. (2000). Do you have to pay attention to your
feelings to be influenced by them? Personality and Social Psychology
Bulletin, 26, 698 ā711.
Giora, K. (1987). Decision making under stress: Scanning of alternatives
under controllable and uncontrollable threats. Journal of Personality and
Social Psychology, 52, 639 ā 644.
74
VAN KLEEF, DE DREU, AND MANSTEAD
Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1992). Primitive emotional
contagion. In M. S. Clark (Ed.), Emotion and social behavior (pp.
151ā177). Newbury Park, CA: Sage.
Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1994). Emotional conta-
gion. Cambridge, England: Cambridge University Press.
Helson, H. (1947). Adaptation-level as a frame of reference for prediction
of psychophysical data. American Journal of Psychology, 60, 1ā29.
Hess, U., & Blairy, S. (2001). Facial mimicry and emotional contagion to
dynamic emotional facial expressions and their influence on decoding
accuracy. International Journal of Psychophysiology, 40, 129 ā141.
Hess, U., Philippot, P., & Blairy, S. (1998). Facial reactions to emotional
facial expressions: Affect or cognition? Cognition and Emotion, 12,
509 ā531.
Hietanen, J. K., Surakka, V., & Linnankoski, I. (1998). Facial electromyo-
graphic responses to vocal affect expressions. Psychophysiology, 35,
530 ā536.
Hilty, J. A., & Carnevale, P. J. (1993). Black-hat/white-hat strategy in
bilateral negotiation. Organizational Behavior and Human Decision
Processes, 55, 444 ā 469.
Hochschild, A. R. (1983). The managed heart: Commercialization of
human feeling. University of California Press.
Hull, J. G., Tedlie, J. C., & Lehn, D. A. (1992). Moderator variables in
personality research: The problem of controlling plausible alternatives.
Personality and Social Psychology Bulletin, 18, 115ā117.
Isen, A. M., Daubman, K. A., & Nowicki, G. P. (1987). Positive affect
facilitates creative problem solving. Journal of Personality and Social
Psychology, 52, 1122ā1131.
Keltner, D., & Haidt, J. (1999). Social functions of emotions at four levels
of analysis. Cognition and Emotion, 13, 505ā521.
Keltner, D., & Kring, A. M. (1998). Emotion, social function, and psy-
chopathology. Review of General Psychology, 2, 320 ā342.
Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social
psychology. In D. T. Gilbert, S. T. Fiske, & A. Lindzey (Eds.), The
handbook of social psychology (Vol. 1, pp. 233ā265). Boston: McGraw-
Hill.
Kiesler, S., & Sproull, L. (1992). Group decision making and communi-
cation technology. Organizational Behavior and Human Decision Pro-
cesses, 52, 96 ā123.
Klinnert, M., Campos, J., Sorce, J., Emde, R., & Svejda, M. (1983).
Emotions as behavior regulators: Social referencing in infants. In R.
Plutchik & H. Kellerman (Eds.), Emotion theory, research, and experi-
ence: Vol. 2. Emotions in early development (pp. 57ā 68). New York:
Academic Press.
Knutson, B. (1996). Facial expressions of emotion influence interpersonal
trait inferences. Journal of Nonverbal Behavior, 20, 165ā182.
Komorita, S. S., & Brenner, A. R. (1968). Bargaining and concession
making under bilateral monopoly. Journal of Personality and Social
Psychology, 9, 15ā20.
Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford Uni-
versity Press.
Levy, D. A., & Nail, P. R. (1993). Contagion: A theoretical and empirical
review and reconceptualization. Genetic, Social, and General Psychol-
ogy Monographs, 119, 235ā285.
Liebert, R. M., Smith, W. P., Hill, J. H., & Keiffer, M. (1968). The effects
of information and magnitude of initial offer on interpersonal negotia-
tion. Journal of Experimental Social Psychology, 4, 431ā 441.
Lundqvist, L. O., & Dimberg, U. (1995). Facial expressions are contagious.
Journal of Psychophysiology, 9, 203ā211.
Manstead, A. S. R., Wagner, H. L., & MacDonald, C. J. (1983). A contrast
effect in judgments of own emotional state. Motivation and Emotion, 7,
279 ā290.
Manstead, A. S. R., Wagner, H. L., & MacDonald, C. J. (1984). Face,
body, and speech as channels of communication in the detection of
deception. Basic and Applied Social Psychology, 5, 317ā332.
Mathews, A., & MacLeod, C. (1994). Cognitive approaches to emotion and
emotional disorders. Annual Review of Psychology, 45, 25ā50.
McKersie, R. B., & Fonstad, N. O. (1997). Teaching negotiation theory and
skills over the Internet. Negotiation Journal, 13, 363ā368.
Moore, D. A., Kurtzberg, T. R., Thompson, L. L., & Morris, M. W. (1999).
Long and short routes to success in electronically mediated negotiations:
Group affiliations and good vibrations. Organizational Behavior and
Human Decision Processes, 77, 22ā 43.
Morris, M. W., & Keltner, D. (2000). How emotions work: An analysis of
the social functions of emotional expression in negotiations. Research in
Organizational Behavior, 22, 1ā50.
Neumann, R., & Strack, F. (2000). āMood contagionā: The automatic
transfer of mood between persons. Journal of Personality and Social
Psychology, 79, 211ā223.
Oatley, K., & Jenkins, J. M. (1992). Human emotions: Functions and
dysfunctions. Annual Review of Psychology, 43, 55ā 85.
Oatley, K., & Jenkins, J. M. (1996). Understanding emotions. Cambridge,
MA: Blackwell.
Oatley, K., & Johnson-Laird, P. N. (1987). Towards a cognitive theory of
emotions. Cognition and Emotion, 1, 29 ā50.
O
ĀØ hman, A. (2000). Fear and anxiety: Evolutionary, cognitive, and clinical
perspectives. In M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of
emotions (2nd ed., pp. 573ā593). New York: Guilford Press.
Parrott, W. G. (1994). Beyond hedonism: Motives for inhibiting good
moods and for maintaining bad moods. In D. M. Wegner & J. W.
Pennebaker (Eds.), Handbook of mental control (pp. 278 ā305). Engle-
wood Cliffs, NJ: Prentice-Hall.
Pillutla, M. M., & Murnighan, J. K. (1996). Unfairness, anger, and spite:
Emotional rejections of ultimatum offers. Organizational Behavior and
Human Decision Processes, 68, 208 ā224.
Pruitt, D. G. (1981). Negotiation behavior. New York: Academic Press.
Pruitt, D. G., & Carnevale, P. J. (1993). Negotiation in social conflict.
Buckingham, England: Open University Press.
Pruitt, D. G., & Syna, H. (1985). Mismatching the opponentās offers in
negotiation. Journal of Experimental Social Psychology, 21, 103ā113.
Putnam, L. L. (1994). Challenging the assumptions of traditional ap-
proaches to negotiation. Negotiation Journal, 10, 337ā346.
Rafaeli, A., & Sutton, R. I. (1991). Emotional contrast strategies as means
of social influence: Lessons from criminal interrogators and bill collec-
tors. Academy of Management Journal, 34, 749 ā775.
Rosenthal, R., & Rosnow, R. L. (1984). Essentials of behavioral research:
Methods and data analysis. New York: McGraw-Hill.
Roth, A. E., & Murnighan, J. K. (1983). Information and aspirations in
two-person bargaining. In R. Tietz (Ed.), Aspiration levels in bargaining
and economic decision making (pp. 91ā103). New York: Springer-
Verlag.
Russell, J. A., & Barrett, L. F. (1999). Core affect, prototypical emotional
episodes, and other things called emotion: Dissecting the elephant.
Journal of Personality and Social Psychology, 76, 805ā 819.
Salovey, P., & Mayer, D. (1990). Emotional intelligence. Imagination,
Cognition and Personality, 9, 185ā211.
Schelling, T. (1960). The strategy of conflict. Cambridge, MA: Harvard
University Press.
Scherer, K. R. (1986). Vocal affect expression: A review and a model for
future research. Psychological Bulletin, 99, 143ā165.
Scherer, K. R., Feldstein, S., Bond, R. N., & Rosenthal, R. (1985). Vocal
cues to deception: A comparative channel approach. Journal of Psycho-
linguistic Research, 14, 409 ā 425.
Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgments
of well-being: Informative and directive functions of affective states.
Journal of Personality and Social Psychology, 45, 513ā523.
75
ANGER AND HAPPINESS IN NEGOTIATIONS
Schweitzer, M., & Croson, R. T. A. (1999). Curtailing deception: The
impact of direct questions on lies and omissions. International Journal
of Conflict Management, 10, 225ā248.
Smith, D. L., Pruitt, D. G., & Carnevale, P. J. (1982). Matching and
mismatching: The effect of own limit, otherās toughness, and time
pressure on concession rate in negotiation. Journal of Personality and
Social Psychology, 42, 876 ā 883.
Stevens, J. (1996). Applied multivariate statistics for the social sciences
(3rd ed.). Mahwah, NJ: Erlbaum.
Thayer, S. (1980). The effect of facial expression sequence upon judgments
of emotion. Journal of Social Psychology, 111, 305ā306.
Thompson, L., & Nadler, J. (2002). Negotiating via information technol-
ogy: Theory and application. Journal of Social Issues, 58, 109 ā124.
Thompson, L., Nadler, J., & Kim, P. H. (1999). Some like it hot: The case
for the emotional negotiator. In L. L. Thompson, J. M. Levine, & D. M.
Messick (Eds.), Shared cognition in organizations: The management of
knowledge (pp. 139 ā161). Mahwah, NJ: Erlbaum.
Tripp, T. M., & Sondak, H. (1992). An evaluation of dependent variables
in experimental negotiation studies: Impasse rates and Pareto efficiency.
Organizational Behavior and Human Decision Processes, 51, 273ā295.
Wall, J. A., Jr. (1991). Impression management in negotiations. In R. A.
Giacalone & P. Rosenfeld (Eds.), Applied impression management: How
image-making affects managerial decisions (pp. 133ā156). Thousand
Oaks, CA: Sage.
Wild, B., Erb, M., & Bartels, M. (2001). Are emotions contagious? Evoked
emotions while viewing emotionally expressive faces: Quality, quantity,
time course, and gender differences. Psychiatry Research, 102, 109 ā
124.
Yukl, G. A. (1974a). Effects of the opponentās initial offer, concession
magnitude, and concession frequency on bargaining behavior. Journal of
Personality and Social Psychology, 30, 323ā335.
Yukl, G. A. (1974b). Effects of situational variables and opponent conces-
sions on a bargainerās perception, aspirations, and concessions. Journal
of Personality and Social Psychology, 29, 227ā236.
Received September 29, 2002
Revision received July 14, 2003
Accepted July 18, 2003
䔲
76
VAN KLEEF, DE DREU, AND MANSTEAD