9
Personality and Social Psychology Escalate shamefully, de-escalate angrily or gratefully: The influence of discrete emotions on escalation of commitment JUNHUA DANG, 1 SHANSHAN XIAO 2 and SOPHIE LILJEDAHL 1 1 Lund University, Lund, Sweden 2 Peking University, Beijing, China Dang, J., Xiao, S. & Liljedahl, S. (2014). Escalate shamefully, de-escalate angrily or gratefully: The inuence of discrete emotions on escalation of commitment. Scandinavian Journal of Psychology 55, 380388. Decision makers often tend to escalate their commitment when faced with a dilemma of whether to continue a losing course of action. Researchers recently began to investigate the inuence of discrete emotions on this decision tendency. However, this work has mainly focused on negative emotions and rarely considered positive emotions, to say nothing of comparing the effects of both of them simultaneously. The current study addresses this need by presenting the results of three experiments that examined the effects of four emotions of both positive and negative valences in escalation situations. Experiment 1 investigated the relationships of three trait emotions (hope, shame, and anger) and escalation of commitment. Experiments 2 and 3 examined the effects of three induced emotions (anger, shame, and gratitude) on escalation of commitment in a student sample and an employee sample, respectively. The results revealed that the effects of discrete emotions in escalation situations are mainly due to their associated differences on the appraisal dimension of responsibility that is related to escalation situations rather than their valence. The theoretical and practical implications are discussed. Key words: Escalation of commitment, appraisal-tendency, discrete emotions, responsibility. Shanshan Xiao, Department of Psychology, Peking University, Beijing 100871, China. E-mail: [email protected] INTRODUCTION Decision makers often tend to escalate their commitment when faced with a dilemma of whether to continue a losing course of action. Although such so-called escalation of commitmentphenomenon has long been considered as a maladaptive decision bias, it occurs in various contexts including nancial investment (Ross & Staw, 1986), product development (Arkes & Blumer, 1985; Boulding, Morgan & Staelin, 1997), human resource allo- cation (Staw & Hoang, 1995), performance appraisal (Bazerman, Beekun & Schoorman, 1982; Schoorman, 1988), as well as ordinary events like waiting for the bus (Wong & Kwong, 2007). Although several theories have been proposed to explain the escalating tendency, they mainly concentrate on cognitive processes that result in decision makers investing too much to quit (see Arkes & Blumer, 1985; Brockner, 1992, for reviews). Comparatively, the role of emotion in escalation situations has received little attention until very recently despite evidence showing the powerful inuence of emotions on judgment and decision making (see Forgas, 1995; Loewenstein & Lerner, 2003, for reviews). The rst study investigating the role of emotion in escalation situations found that negative affect was negatively correlated with escalation of commitment when one was responsible for the failing decision (Wong, Yik & Kwong, 2006). In contrast, a subsequent study reported a positive relationship between positive affect and escalation of commitment (Harvey & Victoravich, 2009). These studies followed a valence-based approach that has been criticized for ignoring specicity of different emotions of the same valance (Lerner & Keltner, 2000). More recent research has been further rened by querying the effects of several specic emotions on escalation of commitment. However, this research has principally focused on negative emotions, and rarely considered positive emotions (Ku, 2008; Tsai & Young, 2010; Wong & Kwong, 2007). Therefore, it has been impossible to tell whether a number of positive emotions should be treated as equivalent as the valence-based approach suggested, or whether a different effect on escalation of commitment would be predicted, as recent work in other domains revealed (e.g., Strohminger, Lewis & Meyer, 2011). The examination of both negative and positive emotions simultaneously would add much to the literature, and provide an empirically-based framework for the purpose of reducing the escalating tendency. The current research aims to investigate the inuence of both positively and negatively valenced discrete emotions on decision makersescalating tendencies. Based on appraisal-tendency the- ory (Lerner & Keltner, 2000; 2001), we compared the effects of four emotions in escalation situations through three experiments. Experiment 1 examined the relationships between three trait affects (hope, shame, and anger) and escalation of commitment using a performance appraisal task. Experiment 2 examined the inuence of three induced emotions (shame, anger, and grati- tude) on decision makersescalating tendencies using a resource allocation task in a student sample. Experiment 3 replicated Experiment 2 in an employee sample. The ndings were consis- tent with appraisal-tendency theory, but contrary to the predic- tions of the valence-based approach, suggesting that the effects of discrete emotions, either positive or negative, on escalation of commitment are mainly due to their differences on the appraisal dimension related to escalation situations rather than their valence. © 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd Scandinavian Journal of Psychology, 2014, 55, 380388 DOI: 10.1111/sjop.12128

Escalate shamefully, de-escalate angrily or gratefully: The influence of discrete emotions on escalation of commitment

  • Upload
    sophie

  • View
    215

  • Download
    1

Embed Size (px)

Citation preview

Personality and Social Psychology

Escalate shamefully, de-escalate angrily or gratefully: The influence

of discrete emotions on escalation of commitment

JUNHUA DANG,1 SHANSHAN XIAO2 and SOPHIE LILJEDAHL1

1Lund University, Lund, Sweden2Peking University, Beijing, China

Dang, J., Xiao, S. & Liljedahl, S. (2014). Escalate shamefully, de-escalate angrily or gratefully: The influence of discrete emotions on escalation ofcommitment. Scandinavian Journal of Psychology 55, 380–388.

Decision makers often tend to escalate their commitment when faced with a dilemma of whether to continue a losing course of action. Researchersrecently began to investigate the influence of discrete emotions on this decision tendency. However, this work has mainly focused on negative emotionsand rarely considered positive emotions, to say nothing of comparing the effects of both of them simultaneously. The current study addresses this needby presenting the results of three experiments that examined the effects of four emotions of both positive and negative valences in escalation situations.Experiment 1 investigated the relationships of three trait emotions (hope, shame, and anger) and escalation of commitment. Experiments 2 and 3examined the effects of three induced emotions (anger, shame, and gratitude) on escalation of commitment in a student sample and an employee sample,respectively. The results revealed that the effects of discrete emotions in escalation situations are mainly due to their associated differences on theappraisal dimension of responsibility that is related to escalation situations rather than their valence. The theoretical and practical implications arediscussed.

Key words: Escalation of commitment, appraisal-tendency, discrete emotions, responsibility.

Shanshan Xiao, Department of Psychology, Peking University, Beijing 100871, China. E-mail: [email protected]

INTRODUCTION

Decision makers often tend to escalate their commitment whenfaced with a dilemma of whether to continue a losing course ofaction. Although such so-called “escalation of commitment”phenomenon has long been considered as a maladaptive decisionbias, it occurs in various contexts including financial investment(Ross & Staw, 1986), product development (Arkes & Blumer,1985; Boulding, Morgan & Staelin, 1997), human resource allo-cation (Staw & Hoang, 1995), performance appraisal (Bazerman,Beekun & Schoorman, 1982; Schoorman, 1988), as well asordinary events like waiting for the bus (Wong & Kwong,2007). Although several theories have been proposed to explainthe escalating tendency, they mainly concentrate on cognitiveprocesses that result in decision makers investing too much toquit (see Arkes & Blumer, 1985; Brockner, 1992, for reviews).Comparatively, the role of emotion in escalation situations hasreceived little attention until very recently despite evidenceshowing the powerful influence of emotions on judgment anddecision making (see Forgas, 1995; Loewenstein & Lerner,2003, for reviews).The first study investigating the role of emotion in escalation

situations found that negative affect was negatively correlatedwith escalation of commitment when one was responsible forthe failing decision (Wong, Yik & Kwong, 2006). In contrast,a subsequent study reported a positive relationship betweenpositive affect and escalation of commitment (Harvey &Victoravich, 2009). These studies followed a valence-basedapproach that has been criticized for ignoring specificity ofdifferent emotions of the same valance (Lerner & Keltner,2000). More recent research has been further refined by

querying the effects of several specific emotions on escalationof commitment. However, this research has principally focusedon negative emotions, and rarely considered positive emotions(Ku, 2008; Tsai & Young, 2010; Wong & Kwong, 2007).Therefore, it has been impossible to tell whether a numberof positive emotions should be treated as equivalent as thevalence-based approach suggested, or whether a different effecton escalation of commitment would be predicted, as recentwork in other domains revealed (e.g., Strohminger, Lewis &Meyer, 2011). The examination of both negative and positiveemotions simultaneously would add much to the literature, andprovide an empirically-based framework for the purpose ofreducing the escalating tendency.The current research aims to investigate the influence of both

positively and negatively valenced discrete emotions on decisionmakers’ escalating tendencies. Based on appraisal-tendency the-ory (Lerner & Keltner, 2000; 2001), we compared the effects offour emotions in escalation situations through three experiments.Experiment 1 examined the relationships between three traitaffects (hope, shame, and anger) and escalation of commitmentusing a performance appraisal task. Experiment 2 examined theinfluence of three induced emotions (shame, anger, and grati-tude) on decision makers’ escalating tendencies using a resourceallocation task in a student sample. Experiment 3 replicatedExperiment 2 in an employee sample. The findings were consis-tent with appraisal-tendency theory, but contrary to the predic-tions of the valence-based approach, suggesting that the effectsof discrete emotions, either positive or negative, on escalation ofcommitment are mainly due to their differences on the appraisaldimension related to escalation situations rather than theirvalence.

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

Scandinavian Journal of Psychology, 2014, 55, 380–388 DOI: 10.1111/sjop.12128

ESCALATION OF COMMITMENT

Over the past three decades, ample evidence has demonstrated astrong tendency of decision makers to persevere in situations“where losses have been suffered, where there is an opportunity topersist or withdraw and where the consequences of these actionsare uncertain” (Staw, 1997, p. 192). These situations are character-ized by three features (Brockner, 1992). First, a large number ofresources have been invested toward a particular course of action.Second, this course of action has received negative feedbackshowing that previous expectations are not met. Finally, thedecision maker has to choose between continuing to add to his/herprevious investment in order to redeem loss and withdrawing fromthe failing course of action. Escalation of commitment refers tothe tendency to continue investing, despite evidence of loss.Economically, this tendency is not rational as rationality requiresonly considering future cost and benefit while ignoring actionsthat have already occurred, particularly when those actions areassociated with negative outcomes (Sivanathan, Molden, Galinsky& Ku, 2008). Accordingly, this decision error should be avoidedas much as possible (Brockner, Houser et al., 1986; Staw, 1982).Staw (1976) has identified two preconditions for the occurrence

of escalation of commitment. First, the decision maker must feelpersonally responsible for the initiation of the losing course ofaction, and second, the decision maker must have committedhimself/herself to irrevocable behavioral consequences. In con-trast, if there is no personal responsibility or it is rather easy forthe decision maker to reverse his/her behavior, the decision makeris less likely to be locked into escalation situations (e.g., Arkes &Blumer, 1985; Brockner et al., 1986; Staw, 1976; Staw & Fox,1977). The underlying motive is considered to be the self-justificationmechanism (Brockner, 1992; Staw, 1976, 1997), which suggeststhat if the decision maker is responsible for a project, negativefeedback to previously made decision would threaten the decisionmaker’s self, thereby activating the need to self-justify. This needfor self-justification would then drive the person to continueinvesting in an attempt to prove the correctness of previousdecision. The greater the perceived responsibility to the decisionmaker, the higher escalating tendency he/she would exhibit.

AFFECTIVE INFLUENCE ON ESCALATIONOF COMMITMENT

In the past, most studies took a valence-based approach whenconsidering the influence of affect on judgment and decisionmaking (see Elster, 1998; Forgas, 1995, 2003, for reviews).This approach emphasized that “the only relevant aspect of emo-tion is their valence” (Elster, 1998, p. 64), which suggests afundamental affect-congruent pattern. That is to say, positive-affective traits and states lead to optimistic judgments whereasnegative-affective traits and states lead to pessimistic judgments(e.g., Johnson & Tversky, 1983; Kavanagh & Bower, 1985;Wright & Bower, 1992). The initial research investigating the roleof emotion in escalation situations followed this approach (Harvey& Victoravich, 2009; Wong et al., 2006). As the authorsdescribed, they measured affect as a broad construct representinga variety of emotions (Wong et al., 2006). It was found that nega-tive affect produced a low level of escalation, and positive affect

led to a high level of escalation, when one was personally respon-sible for the failing decision. This pattern remained intact regard-less of whether affect was measured as a trait (i.e., neuroticism;Wong et al., 2006, studies 1 and 2) or as an emotional state(Harvey & Victoravich, 2009; Wong et al., 2006, study 3).The valence-based approach has been criticized, as it may

“sacrifice specificity in the service of parsimony” (Lerner &Keltner, 2000, p. 475). Accordingly, researchers have begun toinvestigate the effects of specific emotions on judgment anddecision making (e.g., Bodenhausen, Sheppard & Kramer, 1994;DeSteno, Dasgupta, Bartlett & Cajdric, 2004; DeSteno, Petty,Wegener & Rucker, 2000; Keltner, Ellsworth & Edwards, 1993).When it comes to escalation of commitment, the following stud-ies have specified the impact of regret and differentiated theinfluence of anger and fear. On one hand, Wong and Kwong(2007) found that escalation of commitment was positively asso-ciated with anticipated regret about withdrawal and negativelyassociated with anticipated regret about persistence, whileKu (2008) demonstrated that post-escalation regret (regretexperienced from an earlier escalation situation) could reducesubsequent escalation in a different context. On the other hand,Tsai and Young (2010) found that fearful compared to angryindividuals perceived higher risk in their initial decision and thusescalated their commitment less.These pioneering researchers have significantly advanced our

understanding of affective influence on escalation of commit-ment. However, they concentrated on negative emotions andpaid little attention to positive emotions. The problem this posesis twofold. First of all, it is unclear whether different positiveemotions would lead to a similar or distinct influence on theescalating tendency. Although the literature historically sug-gested that positive emotions are less differentiated than negativeemotion (e.g., Ellsworth & Smith, 1988; Ekman, 1993), andoften result in comparable behaviors (e.g., Fredrickson, 2004),recent work in other domains has found evidence for differencesamong discrete positive emotions (e.g., Strohminger et al.,2011). Moreover, few studies have been conducted to examineboth negative and positive emotions and compared their effectson judgment and decision making within a unified framework,much less in escalation situations. Such development wouldmake an important contribution to understanding the effects ofdiscrete emotions on decision making (Cavanaugh, Bettman,Luce & Payne, 2007). A clearer understanding of more differen-tiated emotional influence on escalation of commitment may alsogenerate more effective strategies for reducing the escalatingtendency that can have destructive personal and organizationalconsequences (Wong et al., 2006). In sum, the main goal of ourresearch is to contribute to the literature by investigating themechanism through which discrete emotions of both valencesinfluence escalation of commitment.

THE PRESENT STUDIES

In the present research, we examined the impact of discreteemotions on decision makers’ escalating tendencies based on themost influential framework proposed for explaining emotion-specific effect, appraisal-tendency theory (Lerner & Keltner,

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

Emotional influence and escalation of commitment 381Scand J Psychol 55 (2014)

2000, 2001). Several studies guided by this theory have demon-strated that emotions of the same valence had different effectson judgment and decision making, such as risk perception (e.g.,Lerner & Keltner, 2000, 2001), policy preference (e.g., Lerner,Gonzalez, Small & Fischhoff, 2003), and the tendency forselling price to exceed buying price for the same object (i.e.,endowment effect, Lerner, Small & Loewenstein, 2004).According to appraisal-tendency theory, each emotion not

only arises from cognitive appraisals of a specific situation, butalso shapes people’s perception of subsequent unrelated situa-tions in accord with these appraisals. Smith and Ellsworth(1985) differentiated six cognitive appraisal dimensions underly-ing different emotions: pleasantness, certainty, control, responsi-bility, attentional activity, and anticipated effort. If one of thesedimensions was conceptually related to a decision task, emotionsthat differ in their positions on this dimension would influencethe decision differently. For example, certainty and controldimensions are related to risk perception because they map ontothe two cognitive meta-factors that determine risk assessments:“unknown risk” and “dread risk” (e.g., Slovic, 1987; Slovic,Fischhoff & Lichtenstein, 1986). Emotions that fall at oppositeends of the certainty and control dimensions would influencerisk perception differently. Specifically, fear could lead topessimistic risk perception because it arises from appraisals ofuncertainty and situational control, whereas anger could result inoptimistic risk perception since it arises from appraisals ofcertainty and individual control (Lerner & Keltner, 2000, 2001).Moreover, appraisal-tendency theory argues that such appraisal-tendency processes can also apply to dispositional or traitemotions because trait emotions incline people to experience thecorresponding emotional states across time and situations (Lerner& Keltner, 2000, 2001).When considering the influence of discrete emotions in escala-

tion situations, we first had to identify appraisal dimension(s)related to escalation of commitment. Of the six dimensions men-tioned previously, the responsibility dimension is directly associ-ated with escalation situations in that personal responsibility forthe initiation of the losing course of action is the precondition ofescalation of commitment (Staw, 1976). Empirical evidence hasconsistently shown that higher personal responsibility leads tomore escalation (e.g., Boulding et al., 1997; Wong & Kwong,2007; for earlier reviews, see Brockner, 1992). Based on apprai-sal-tendency theory, we proposed that emotions that differ intheir positions on the responsibility dimension would inducedifferent levels of perceived personal responsibility, which inturn lead to different degree of escalation of commitment.Among the emotions that differ in their positions on the

responsibility dimension, we selected four specific emotions thatmerit investigation because of their critical roles in the domainsof decision making and organizational behaviors (Lazarus &Cohon-Charash, 2001; Greene & Haidt, 2002; Lerner & Tiedens,2006). In Experiment 1, we measured two negative traitemotions, shame and anger, and a positive trait emotion, hope.Shame and anger fall at opposite ends of the responsibilitydimension, which implies that shame arises from appraisal ofself-responsibility whereas anger arises from appraisal of other-responsibility. Hope, which is independently elicited in relationto self/other responsibility, lies in the middle of the self-other

responsibility dimension (Smith & Ellsworth, 1985). Thevalence-based approach would predict that trait hope is posi-tively correlated with escalation of commitment whereas traitshame and anger are negatively correlated with this decision ten-dency. However, appraisal-tendency theory predicts that traitshame would lead decision makers to perceive a higher level ofself-responsibility and thus be positively correlated with theescalating tendency, while trait anger would result in a lowerlevel of self-responsibility and thus be negatively correlated withescalation of commitment. Trait hope would have negligibleeffect on decision makers’ responsibility appraisal and thus beinsignificantly correlated with escalation of commitment.In Experiments 2 and 3, we examined the impacts of shame,

anger and another positive emotion, gratitude. Like anger, grati-tude also arises from appraisal of other-responsibility. Instead ofmeasuring trait emotions, we manipulated emotions by inducingpeople’s emotional states in order to avoid the questions aboutcausality and added a control group. We hypothesized thatinduced shame would increase escalation of commitmentwhereas both induced anger and gratitude would decrease it,relative to the control condition. The finding that not only coulda negative emotion (e.g., shame) have a positive effect on esca-lation of commitment but also a positive emotion (e.g., gratitude)could have a negative effect on such decision tendency seems amore robust evidence consistent with appraisal-tendency theorybut contrary to the predictions of the valence-based approach.Finally, in Experiment 3 we replicated Experiment 2 in anemployee sample in order to examine the generalizability of ourfindings to organizational settings.

EXPERIMENT 1

Method

Participants and procedures. Seventy-seven postgraduate stu-dents (61% male; mean age 24.69 years) from Shenzhen GraduateSchool of Peking University were recruited via advertisements oncampus and given 10 RMB (approximately $1.60) for participat-ing. They completed this experiment individually. After signingconsent to participate, they were instructed to complete a decisiontask on a computer by carefully reading a scenario. Participantswere then asked to report their trait emotions and demographicinformation with paper-pencil questionnaires. Finally, they werethanked and debriefed.

Decision task

Similar to escalation of commitment found in product devel-opment (e.g., Arkes & Blumer, 1985) and resource allocation(e.g., Staw, 1976), this tendency has also been demonstratedin performance appraisal (Bazerman, et al., 1982; Schoorman,1988). In the context of performance appraisal, the raters tend toevaluate the person they are evaluating more positively if theevaluated person was previously selected or promoted by the rat-ers, even if the evaluated person demonstrated an exceedinglypoor performance. Based on such findings, Wong et al. (2006)modified Staw’s (1976) resource allocation scenario to a person-nel selection context. We adopted Wong et al.’s (2006) task and

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

382 J. Dang et al. Scand J Psychol 55 (2014)

asked participants to play the role of a multinational company’ssales manager who was responsible for hiring decisions of arecent recruitment. The scenario described that there were onlytwo candidates remaining after many rounds of elimination.Participants were provided with information of these two candi-dates and required to choose one of them to join the sales team.The information given to the participants showed that these twocandidates’ previous performances were very similar. Thenparticipants were told that the computer would run a programbased on historical data and simulate the chosen candidate’s per-formance of the next five years. However, the running resultsshowed that the candidate performed poorly regardless of whichcandidate the participant chose. Finally, participants were askedto evaluate the candidate’s overall performance. Higher evalua-tion would indicate greater escalation of commitment.

Measures

Participants evaluated their chosen candidate’s overall perfor-mance on a seven-point scale ranging from 1 (very poor) to 7(very good). This performance rating served as the index of par-ticipants’ escalation of commitment to their prior decision. Traitanger was measured by a 20-item scale (a = 0.84) which wasused by Lerner and Keltner (2000, 2001). Trait hope was mea-sured by the Hope Scale (a = 0.72; Snyder et al., 1991) and traitshame by the Internalized Shame Scale (ISS) (a = 0.88; Cook,1987; Rybak & Brown, 1996). All measures were translatedfrom English into Chinese and back-translated into English toensure the translation accuracy (Brislin, 1980).

Results

The means, standard deviations, and Pearson correlations (zero-order) are depicted in Table 1. Consistent with the fact that theyshare a negative valence, trait shame and anger were significantlycorrelated (r = 0.41, p < 0.01). In order to examine the indepen-dent influence of each emotion, we used hierarchical regressionanalyses to test our hypotheses with escalation of commitmentas the dependent variable. The regression results are shown inTable 2. In step 1, we entered the demographic variables (i.e., ageand gender) to control their possible effects. In step 2, we simulta-neously entered the three emotions, which jointly contributed asignificant portion of the accounted variance (ΔR2 = 0.12,p < 0.05). Shame was positively correlated with escalation ofcommitment (b = 0.37, p < 0.01), whereas anger was negativelycorrelated with escalation of commitment (b = –0.31, p < 0.05).

The correlation between hope and escalation of commitment wasinsignificant (b = 0.01, p = 0.92). These results were expectedand supported our hypotheses.

Discussion

Although this experiment provided initial evidence for theappraisal-tendency hypotheses in escalation situations, it has twolimitations. First, the measures of trait emotions raised thequestions about causality since there was no experimentalmanipulation. For example, some may argue that the positiverelationship between shame and escalation of commitment canbe explained reversely such that people with higher escalatingtendency are more likely to experience the emotion of shame.Second, it may be argued that the insignificant relationshipbetween hope and escalation of commitment is not convincingevidence against the valence-based approach because hope is justa mildly positive emotion (Smith & Ellsworth, 1985).To solve these methodological and theoretical matters, we

conducted Experiments 2 and 3 in which we manipulated inde-pendent variables (i.e., emotions) and selected another positiveemotion. On one hand, we induced people’s emotional states priorto the decision task. On the other hand, we compared the influ-ence of the two negative emotions examined in Experiment 1 andanother positive emotion, gratitude, on escalation of commitment.Because gratitude arises from appraisal of low personal responsi-bility as anger, we hypothesized that this positive emotion (grati-tude) would lead to a decreased escalating tendency. In addition,we also added a control group as the baseline to examine therelative effect of each emotion. Experiment 2 was conducted in astudent sample and Experiment 3 in an employee sample.

Table 1. Means, standard deviations, and zero-order correlations in Experiment 1

Variables M SD 1 2 3 4 5

1 Age 24.68 2.26 —2 Gender (0 = male, 1 = female) 0.39 0.49 0.14 —3 Hope 3.37 0.48 0.13 �0.28*** —4 Shame 2.72 0.50 �0.12 0.08 �0.08 —5 Anger 2.73 0.48 0.22* 0.31*** �0.33*** 0.41*** —6 Escalation of Commitment 4.05 0.71 0.01 �0.14 0.10 0.22* �0.15

* p < 0.10. ** p < 0.05. *** p < 0.01.

Table 2. Regressions of escalation of commitment on trait shame andanger in Experiment 1

Predictors

Escalation of commitment

Beta DR2

Step 1 0.02Age 0.03Gender –0.14

Step 2 0.12**Shame 0.37**Anger –0.31*Hope –0.01

Note: * p < 0.05. ** p < 0.01.

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

Emotional influence and escalation of commitment 383Scand J Psychol 55 (2014)

EXPERIMENT 2

Method

Participants and procedures. One hundred and thirteen post-graduate students (55% male; mean age 23.41 years) from thesame university as Experiment 1 took part in this experimentand were paid 10 RMB (approximately $1.60) for their partici-pation. They completed this experiment individually. Followingcompletion of informed consent, participants were randomlyassigned to four conditions (a control condition and threeexperimental conditions). All participants were then directed tocomplete two ostensibly independent tasks on a computer. In theexperimental conditions, the first task required participants todescribe five events they experienced that made them mostashamed, grateful, or angry, which acted as a manipulation toinduce participants’ emotional states. The second task askedparticipants to complete a decision task in which they had toread a scenario and make a subsequent decision. Participants inthe control condition first wrote five daily activities and thencompleted the decision task.

Decision task

The decision task was modified from Staw’s (1976) “A & Sfinancial case”. This case described a multinational companywhose profit had declined recently as a result of lackingResearch and Development funding. Participants were instructedto play the role of the financial vice-president with responsibilityfor allocating 10 million dollars to only one of the two biggestdepartments. They were provided with information about thesetwo departments showing that their previous performances werevery similar. Participants were then told that the computer wouldrun a program based on historical data and simulate these twodepartments’ performance over the next five years. The simula-tion results showed that the chosen division performed evenworse than before whereas the unchosen division eventuallyreturned to profitable levels. Subsequently, participants wereasked to allocate additional 20 million dollars of funding andwere allowed to divide money in any way they wished betweenthe two departments. The amount they assigned to the depart-ment that received the prior investment served as the index oftheir escalating tendencies.

Manipulation checks

We did a pilot with 40 participants who were independent fromparticipants in the main experiment to test the effectiveness ofemotion induction. Participants were randomly assigned to acontrol condition and three emotion conditions. Participants in thecontrol condition wrote five daily activities, while participants inthe experimental conditions described five things that made themmost ashamed, grateful, or angry, respectively. All participantswere then required to report their emotional states using 21 adjec-tives. “Angry” and “irritated” were combined to obtain a compos-ite measure of anger (a = 0.73), “grateful” and “thankful” toobtain gratitude (a = 0.88), and “abashed,” “embarrassed” and“shameful” to obtain shame (a = 0.80). A one-way ANOVArevealed significant differences between conditions for all of the

three emotions (for anger, F(3, 36) = 39.83, p < 0.001; for grati-tude, F(3, 36) = 47.47, p < 0.001; for shame, F(3, 36) = 27.78,p < 0.001). Planned comparison showed that participants in theanger condition (M = 5.10, SD = 0.81) felt more angry than didparticipants in all other conditions, t = 9.76, p < 0.001, particu-larly compared to the control condition (M = 3.35, SD = 0.97),t = 5.30, p < 0.001. Similarly, participants in the gratitude condi-tion (M = 5.10, SD = 0.70) felt more grateful than did participantsin all other conditions, t = 11.61, p < 0.001, particularly com-pared to the control condition (M = 2.45, SD = 1.09), t = 7.98,p < 0.001. Participants in the shame condition (M = 4.43, SD =0.82) also felt more ashamed than did participants in all other con-ditions, t = 8.25, p < 0.001, particularly compared to the controlcondition (M = 2.90, SD = 1.09), t = 4.60, p < 0.001.

Results

A one-way ANOVA revealed a significant effect of the emo-tional manipulation, F(3, 109) = 7.59, p < 0.001, g2 = 0.17. Weconducted planned comparisons to examine our hypotheses. Asdisplayed in Figure 1, participants in the shame condition allo-cated more money to the department they initially chose(M = 13.10 million, SD = 4.77) than did participants in all otherconditions, t = 4.20, p < 0.01, and particularly the control condi-tion (M = 11.00 million, SD = 2.95), t = 2.06, p < 0.05. Partici-pants in the anger condition (M = 8.78 million, SD = 3.93) andthe gratitude condition (M = 8.90 million, SD = 3.97) both allo-cated less money to the department they initially chose than didparticipants in the control condition, t = 2.12 and 1.98, respec-tively, ps < 0.05. The difference between the anger conditionand the gratitude condition is insignificant, t = 0.12, ns. Theseresults showed that shame increased decision makers’ escalationof commitment, whereas both anger and gratitude decreased it,which supports the predictions of appraisal-tendency theoryrather than ones of the valence-based approach.

EXPERIMENT 3

Previously, the majority of studies on escalation of commitmentwere conducted in student samples, which potentially limits the

Fig. 1. Amount of money allocated to the department initially-chosen ineach condition of Experiment 2. Error bars indicate standard error.

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

384 J. Dang et al. Scand J Psychol 55 (2014)

generalizability of the findings to other real-world situations,especially organizational settings. Experiments 1 and 2 faced thesame problem. Therefore, in Experiment 3 we replicatedExperiment 2 but used an employee sample to seek evidence forexternal validity.

Method

Participants and procedures. Participants in this experiment were120 Chinese full-time employees (57% male; mean age28.46 years) from 3 private-owned organizations. They were con-tacted through each company’s human resource manager and com-pleted this experiment on the first author’s computer at eachcompany’s meeting room. As in Experiment 2, participants wererandomly assigned to four conditions (a control condition andthree experimental conditions). Participants in the experimentalconditions were directed to complete two ostensibly independenttasks: one for emotion induction and the other for measurement ofthe escalating tendency. Participants in the control condition firstwrote five daily activities and then completed the decision task.

Decision task

The decision task in this experiment was the same as the oneused in Experiment 2.

Results

One-way ANOVA revealed that the emotional induction had asignificant effect on escalation of commitment, F(3,116) =12.39, p < 0.001, g2 = 0.24. Planned comparisons showedthat participants in the anger condition (M = 7.48 million,SD = 4.06) and the gratitude condition (M = 8.63 million,SD = 3.10) both allocated less money to the department theypreviously invested in than did participants in the control condi-tion (M = 11.06 million, SD = 4.59), t = 3.48 (for anger) and2.37 (for gratitude), ps < 0.05. There were no significant differ-ence between the anger condition and the gratitude condition,t = 1.12, ns. Participants in the shame condition allocated moremoney to the department they initially chose (M = 13.20 mil-

lion, SD = 3.98) compared to participants in all other conditions,t = 5.96, p < 0.01, particularly the control condition, t = 2.09,p = 0.04. These results are displayed in Figure 2.

GENERAL DISCUSSION

The current research has designed three experiments to examinethe influence of discrete emotions on escalation of commitment.In Experiment 1, we examined the relationships between threetrait emotions (hope, shame, and anger) and escalation of com-mitment in a performance appraisal situation. Results showedthat while trait shame was positively correlated with escalationof commitment, trait anger was negatively correlated with it.Trait hope was not a significant predictor of this decision ten-dency. In Experiments 2 and 3, we examined the influence ofthree induced emotions (shame, anger, and gratitude) on escala-tion of commitment in a resource allocation situation. These twoexperiments revealed that, relative to the control condition,induced shame increased escalation of commitment, while bothinduced anger and gratitude decreased it. This phenomenon wasfound in both a student sample (Experiment 2) and an employeesample (Experiment 3), which suggests that our findings are reli-able and can be generalized to organizational situations.

Theoretical implications

Previous work investigating the effects of discrete emotions onescalation of commitment focused on negative emotions, yetneglected positive emotions. Our research responded to this needby examining the influence of four emotions of both valences onthe escalating tendency. Consistent with previous studies (e.g.,Lerner & Keltner, 2000, 2001; Lerner et al., 2003), we foundthat two negative emotions (e.g., anger and shame) had oppositeeffects on escalation of commitment. Similarly, our results alsorevealed that two positive emotions (e.g., hope and gratitude)could lead to different influence on such decision tendency,which is the first demonstration of the differentiated effects ofpositive emotions in this field. Further, we compared the effectsof discrete emotions of both valences simultaneously, which gen-erated more robust evidence against the valence-based approach,such that not only could a negative emotion have a positiveeffect on escalation of commitment but also a positive emotioncould result in a reduced impact on such decision tendency.Importantly, these findings revealed the mechanism through

which discrete emotions of both valences influence escalationof commitment. Emotions that arise from appraisal of self-respon-sibility (e.g., shame) would result in the perception of higherself-responsibility in escalation situations, which in turn leads tocontinued investment in their prior choices, regardless of out-comes associated with prior choices. By contrast, emotions thatarise from appraisal of other-responsibility (e.g., anger and grati-tude) would lead participants to perceive lower self-responsibilityand end their failing courses of action. Emotions arising fromappraisal of indifference to self/other responsibility (e.g., hope) arenot significant predictors of decision makers’ escalating tendencies.Further, our research also empirically verified appraisal-

tendency theory on the cognitive appraisal dimension ofresponsibility. Although this theory assumes that emotions are

Fig. 2. Amount of money allocated to the department initially-chosen ineach condition of Experiment 3. Error bars indicate standard error.

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

Emotional influence and escalation of commitment 385Scand J Psychol 55 (2014)

characterized by multiple appraisal dimensions, most studiesguided by this theory examined the effects of emotions that dif-fer in their positions on the control or/and certainty dimensions(e.g., fear and anger; see Han, Lerner, Keltner, 2007, for areview). As some researchers have indicated, the control andcertainty dimensions might not be the most central forunderstanding differences among some discrete emotions, partic-ularly positive emotions (Cavanaugh et al., 2007). The currentstudy adds to the literature by demonstrating that emotions,either positive or negative, differing in their positions on theresponsibility dimension could influence individuals’ decisiontendencies in importantly different ways.

Practical implications

Researchers have tried to develop de-escalation strategies fordecades (e.g., Brockner, Shaw & Rubin, 1979; Simonson &Staw, 1992). However, most of these studies relied on cognitivetheories, and were often demonstrated to be ineffective becauseof two main problems that cannot be easily solved. First, somede-escalation techniques tend to invoke conflicting psychologicalprocesses in escalation situations (Simonson & Staw, 1992). Forexample, increasing decision makers’ self-diagnosticity of out-comes may exert a dual effect on their escalating tendencies. Ifindividuals believe that the outcomes of their decisions reflecttheir ability, they are motivated to simultaneously evaluate theavailable information for decisional accuracy and justify theirdecisions for decisional correctness. Second, some de-escalationtechniques are often difficult to implement in organizations. Forinstance, changing the principal who is in charge of the initiationof the losing project seems an effective way for eliminating thenegative influence of self-justification (Staw, 1982), yet it oftenbrings about even higher loss or cost (e.g., a new competentprincipal is not available).Our findings suggest that emotional states that are induced by

irrelevant events are very important determinants of escalation ofcommitment. An emotion that arises from an appraisal of highresponsibility, such as shame, can lead individuals who areresponsible for initiating a losing course of action to add addi-tional resources to their prior decisions. By contrast, emotionsthat arise from an appraisal of low responsibility, such as angerand gratitude, have the opposite effect. Apparently, these princi-ples are easy to understand and put into practice. Organizationscould design simple strategies in order to influence the emotionalstate of the principal trapped in escalation situations, or teachemotion regulation strategies to reduce the impact of heightenedemotion on decision making. Further, because appraisal-tendencyprocesses can apply to trait emotions, organizations should bemindful of individual differences in dispositional affect, possiblyby administering psychometrically valid measures of disposi-tional affect as part of the hiring process.

Limitations and future research

There are several limitations may warrant further investigation.First, although our results revealed that anger could decreaseescalation of commitment, Tsai and Young (2010) found thatanger insignificantly increased such decision tendency relative

to the control condition. These authors proposed that angryindividuals would perceive lower risk in their initial decisionand thus escalate their commitment more. We suggest that suchinconsistency might arise from the different procedures eachresearch group implemented. In our experiments, participantswere informed that the computer would run a program based onhistorical data and simulate predictive data based on their initialdecisions, which makes their responsibility salient. In Tsai andYoung’s (2010) experiments, however, participants were giventhe results of their decisions by the experimenter after theirchoice, which may lead to a weak sense of responsibilitybecause of little involvement. This supposition was indirectlydemonstrated by the fact that the escalating tendencies of thecontrol conditions in Tsai and Young’s (2010) two studies werelower than the escalating tendencies in the current research, aswell as previous literature (e.g., Simonson & Staw, 1992; Staw,1976; Wong et al., 2006). When decision makers’ senses ofpersonal responsibility are less salient, risk perception maybecome a key factor determining decision makers’ escalatingtendencies. Under these circumstances, according to appraisal-tendency theory, the relevance of the responsibility dimension toescalation situations is blocked and the appraisal dimensionsrelated to risk perception will most directly influence escalationof commitment (Lerner & Keltner, 2001), just as Tsai andYoung (2010) demonstrated. Future studies are needed todirectly examine these decision-making pathways.Second, although we have demonstrated the differentiated

effects of positive emotions (i.e., hope and gratitude) on escala-tion of commitment, future studies are required to determine theinfluence of other positive emotions since little research has beendone in this respect. For example, pride is also relevant to esca-lating situations because it arises from attribution of responsibil-ity to oneself as shame. Thus, a proud decision maker mayeasily get trapped when facing negative feedback from an initialdecision. Future research could also empirically address theunderlying mechanism through which discrete emotions affectescalation of commitment. A practical method would be todirectly measure participants’ perceived responsibility and emo-tional valence in each condition, followed by testing whether theinfluence of emotions is mediated by perceived responsibilityrather than emotional valence. However, this method also has ashortcoming because people are not always aware of their higherorder cognitive processes (Nisbett & Wilson, 1977). Alternativeways can also be used such as orthogonally manipulating emo-tions along the responsibility and the valence dimensions (c.f.,van Steenbergen, Band & Hommel, 2010), which was onlypartly achieved in the current research.Third, although the current research demonstrated that deci-

sion makers’ escalating tendencies were influenced by emotionsarising from irrelevant events, it should also be noted that thereare some boundary conditions for such effects that were nottested in any of our experiments. On the one hand, appraisal-tendency theory addressed two hypotheses concerning conditionsthat might deactivate the impact of emotions on judgment anddecision making (Lerner & Keltner, 2000). This theory assertsthat appraisal tendencies will be deactivated once the eventeliciting a specific emotion is resolved (the goal-attainmenthypothesis) and when people become aware of their own

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

386 J. Dang et al. Scand J Psychol 55 (2014)

judgment processes (the cognitive-awareness hypothesis). On theother hand, the characteristics of escalation situations may alsobe potential moderators. As some researchers have argued, thereare at least two types of escalation situations (Garland &Conlon, 1998; Staw & Hoang, 1995). Unlike the type examinedin this article, the other type of escalation situations is character-ized by a process of project completion. A recent study demon-strated that factors affect escalation of commitment varieddepending on the stage of project completion (He & Mittal,2007). Specifically, need for project information influencesresource commitment at the initial stage of a project, whereasneed for project completion affects it at the terminal stage. Atthe intermediate stage, the decision risk has the strongest effecton escalation of commitment. Accordingly, the effect of aspecific emotion on these three stages of project completion maybe different. Future research may gain further comprehensionabout escalation of commitment and the influence of discreteemotions by examining these moderating effects.In addition, our investigation mainly focused on emotions that

are unrelated to the present judgment or decision (i.e., incidentalemotion), rather than emotions that are normally related to thejudgment or decision at hand (i.e., integral emotion). Future stud-ies should attend to the similarities and differences between thesetwo types of emotion. For example, de Hooge, Breugelmans, andZeelenberg (2008) found that integral shame could promote pro-social behavior for people with pro-self orientations because itthreatens their selves and leads them to alleviate this threat bycomplying with norms for pro-social behavior. Incidental shamehas no such effects, however, as the decision situation is alreadydifferent from the situation that elicited the shameful experienceand thus reduces the threat to self (de Hooge et al., 2008). Suchfindings imply that although incidental shame boosts escalation ofcommitment as our study revealed, integral shame (shame elicitedby escalation situations) may reduce this decision bias.

REFERENCES

Arkes, H. R. & Blumer, C. (1985). The psychology of sunk cost. Orga-nizational Behavior and Human Decision Processes, 35, 124–140.

Bazerman, M. H., Beekun, R. I. & Schoorman, F. D. (1982). Perfor-mance evaluation in a dynamic context: A laboratory study of theimpact of a prior commitment to the ratee. Journal of AppliedPsychology, 67, 873–876.

Bodenhausen, G., Sheppard, L. & Kramer, G. (1994). Negative affectand social judgment: The different impact of anger and sadness.European Journal of Social Psychology, 24, 45–62.

Boulding, W., Morgan, R. & Staelin, R. (1997). Pulling the plug to stopthe new product drain. Journal of Marketing Research, 34, 164–176.

Brislin, R. W. (1980). Translation and content analysis of oral and writ-ten materials. In H. C. Triandis, & H. Berry (Eds.), Handbook ofcross-cultural psychology (Vol 2 pp. 389–444). Boston, MA: Allyn& Bacon.

Brockner, J. (1992). The escalation of commitment to a failing courseof action: Toward theoretical progress. Academy of ManagementReview, 17, 39–61.

Brockner, J., Houser, R., Birnbaum, G., Lloyd, K., Deitcher, J., Nathanson,S. & Rubin, J. Z. (1986). Escalation of commitment to an ineffectivecourse of action: The effect of feedback having negative implicationsfor self–identity. Administrative Science Quarterly, 31, 109–126.

Brockner, J., Shaw, M. C. & Rubin, J. Z. (1979). Factors affectingwithdrawal from an escalating conflict: Quitting before it’s too late.Journal of Experimental Social Psychology, 15, 492–503.

Cavanaugh, L. A., Bettman, J. R., Luce, M. F. & Payne, J. W. (2007).Appraising the appraisal-tendency framework. Journal of ConsumerPsychology, 17, 169–173.

Cook, D. R. (1987). Measuring shame: The internalized shame scale.Alcoholism Treatment Quarterly, 4, 197–215.

De Hooge, I. E., Breugelmans, S. M. & Zeelenberg, M. (2008). Not sougly after all: When shame acts as a commitment device. Journal ofPersonality and Social Psychology, 95, 933–943.

DeSteno, D., Petty, R. E., Wegener, D. T. & Rucker, D. D. (2000).Beyond valence in the perception of likelihood: The role of emotionspecificity. Journal of Personality and Social Psychology, 78,397–416.

DeSteno, D., Dasgupta, N., Bartlett, M. Y. & Cajdric, A. (2004). Preju-dice from thin air: The effect of emotion on automatic intergroupattitudes. Psychological Science, 15, 319–324.

Ekman, P. (1993). Facial expression and emotion. American Psycholo-gist, 48, 384–392.

Ellsworth, P. C. & Smith, C. A. (1988). From appraisal to emotion:Differences among unpleasant feelings. Motivation and Emotion, 12,271–302.

Elster, J. (1998). Emotions and economic theory. Journal of EconomicLiterature, 36, 47–74.

Forgas, J. P. (1995). Mood and judgment: The affect infusion model(AIM). Psychological Bulletin, 117, 39–66.

Forgas, J. P. (2003). Affective influences on attitudes judgments. In R. J.Davidson, K. R. Scherer, & H. H. Goldsmith (Eds.), Handbook ofaffective sciences (pp. 596–618). New York: Oxford UniversityPress.

Fredrickson, B. L. (2004). The broaden-and-build theory of positiveemotions. Philosophical Transactions of the Royal Society of London,359, 1367–1377.

Garland, H. & Conlon, D. E. (1998). Too close to quit: The role of pro-ject completion in maintaining commitment. Journal of AppliedSocial Psychology, 28, 2025–2048.

Greene, J. & Haidt, J. (2002). How (and where) does moral judgmentwork? Trends in Cognitive Sciences, 6, 517–523.

Han, S., Lerner, J. S. & Keltner, D. (2007). Feelings and consumerdecision making: The appraisal-tendency framework. Journal ofConsumer Psychology, 17, 158–168.

Harvey, P. & Victoravich, L. M. (2009). The influence of forward-looking antecedents, uncertainty, and anticipatory emotions on projectescalation. Decision Sciences, 40, 759–782.

He, X. & Mittal, V. (2007). The effect of decision risk and project stageon escalation of commitment. Organizational Behavior and HumanDecision Processes, 103, 225–237.

Johnson, E. J. & Tversky, A. (1983). Affect, generalization, and the per-ception of risk. Journal of Personality and Social Psychology, 45,20–31.

Kavanagh, D. J. & Bower, G. H. (1985). Mood and self-efficacy: Impactof joy and sadness on perceived capabilities. Cognitive Therapy andResearch, 9, 507–525.

Keltner, D., Ellsworth, P. C. & Edwards, K. (1993). Beyond simplepessimism: Effects of sadness and anger on social perception.Journal of Personality and Social Psychology, 64, 740–752.

Ku, G. (2008). Learning to de-escalate: The effects of regret in escalationof commitment. Organizational Behavior and Human DecisionProcesses, 105, 221–232.

Lazarus, R. S. & Cohon-Charash, Y. (2001). Discrete emotions in orga-nizational life. In R. L. Payne, & G. L. Cooper (Eds.), Emotions atwork: Theory, research and applications in management (pp. 45–81).Chichester: Wiley.

Lerner, J. S., Gonzalez, R. M., Small, D. A. & Fischhoff, B. (2003).Effects of fear and anger on perceived risks of terrorism: A nationalfield experiment. Psychological Science, 14, 144–150.

Lerner, J. S. & Keltner, D. (2000). Beyond valence: Toward a model ofemotion-specific influences on judgment and choice. Cognition andEmotion, 14, 473–493.

Lerner, J. S. & Keltner, D. (2001). Fear, anger, and risk. Journal ofPersonality and Social Psychology, 81, 146–159.

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

Emotional influence and escalation of commitment 387Scand J Psychol 55 (2014)

Lerner, J. S., Small, D. A. & Loewenstein, G. (2004). Heart strings andpurse strings: Carryover effects of emotions on economic decisions.Psychological Science, 15, 337–341.

Lerner, J. S. & Tiedens, L. Z. (2006). Portrait of the angry decisionmaker: How appraisal tendencies shape agners’s influence on cogni-tion. Journal of Behavioral Decision Making, 19, 115–137.

Loewenstein, G. & Lerner, J. S. (2003). The role of affect in decisionmaking. In R. Davidson, K. Scherer, & H. Goldsmith (Eds.),Handbook of affective science (pp. 619–642). New York: OxfordUniversity Press.

Nisbett, R. & Wilson, T. (1977). Telling more than we can know: Verbalreports on mental processes. Psychological Review, 84, 231–259.

Ross, J. & Staw, B. M. (1986). Expo 86: An escalation prototype.Administrative Science Quarterly, 31, 274–297.

Rybak, C. J. & Brown, B. M. (1996). Assessment of internalized shame:Validity and reliability of the internalized shame scale. AlcoholismTreatment Quarterly, 14, 71–83.

Schoorman, F. D. (1988). Escalation bias in performance appraisals: Anunintended consequence of supervisor participation in hiring deci-sions. Journal of Applied Psychology, 73, 58–62.

Simonson, I. & Staw, B. M. (1992). Deeacalation strategies: A compari-son of techniques for reducing commitment to losing courses ofaction. Journal of Applied Psychology, 77, 419–426.

Sivanathan, N., Molden, D. C., Galinsky, A. D. & Ku, G. (2008). Thepromise and peril of self-affirmation in de-escalation of commitment.Organizational Behavior and Human Decision Processes, 107, 1–14.

Slovic, P. (1987). Perception of risk. Science, 236, 280–285.Slovic, P., Fischhoff, B. & Lichtenstein, S. (1986). The psychometric

study of risk perception. In V. T. Covello, J. Menkes, &J. Mumpower (Eds.), Risk evaluation and measurement (pp. 3–24).New York: Plenum.

Smith, C. A. & Ellsworth, P. C. (1985). Patterns of cognitive appraisalin emotion. Journal of Personality and Social Psychology, 48,813–838.

Snyder, C. R., Harris, C., Anderson, J. R., Holleran, S. A., Irving, L. M.,Sigmon, S. T., et al. (1991). The will and the ways: Development

and validation of an individual-differences measure of hope. Journalof Personality and Social Psychology, 60, 570–585.

Staw, B. M. (1976). Knee-deep in the big muddy: A study of escalatingcommitment to a chosen course of action. Organizational Behaviorand Human Decision Processes, 16, 27–44.

Staw, B. M. (1982). Counterforces to change. In P. S. Goodman (Ed.),Change in organizations (pp. 87–121). San Francisco, CA: Jossey-Bass.

Staw, B. M. (1997). The escalation of commitment: An update andappraisal. In Z. Shapira (Ed.), Organization decision making (pp.191–215). New York: Cambridge University Press.

Staw, B. M. & Fox, F. V. (1977). Escalation: The determinants ofcommitment to a chosen course of action. Human Relations, 30,431–450.

Staw, B. M. & Hoang, H. (1995). Sunk costs in the NBA: Why draftorder affects playing time and survival in professional basketball.Administrative Science Quarterly, 40, 474–494.

Strohminger, N., Lewis, R. L. & Meyer, D. E. (2011). Divergent effectsof different positive emotions on moral judgment. Cognition, 119,295–300.

Tsai, M. & Young, M. J. (2010). Anger, fear, and escalation of commit-ment. Cognition & Emotion, 24, 962–973.

van Steenbergen, H., Band, G. P. & Hommel, B. (2010). In the moodfor adaptation: How affect regulates conflict-driven control. Psycho-logical Science, 21, 1629–1634.

Wong, K. F. E. & Kwong, J. Y. Y. (2007). The role of anticipated regretin escalation of commitment. Journal of Applied Psychology, 92,545–554.

Wong, K. F. E., Yik, M. & Kwong, J. Y. Y. (2006). Understanding theemotional aspects of escalation of commitment: The role of negativeaffect. Journal of Applied Psychology, 91, 282–297.

Wright, W. F. & Bower, G. H. (1992). Mood effects on subjectiveprobability assessment. Organizational Behavior and HumanDecision Processes, 52, 276–291.

Received 5 September 2013, accepted 10 March 2014

© 2014 Scandinavian Psychological Associations and John Wiley & Sons Ltd

388 J. Dang et al. Scand J Psychol 55 (2014)