Constituents’ blame attributions for ethically ambiguous negotiation tactics of the representative under threat

Constituents’ Blame Attributions for Ethically Ambiguous Negotiation Tactics of the Representative Under Threat

Sander Kammenga (5960355) Master Thesis Psychologie Universiteit van Amsterdam

Supervisor: Özüm Saygi Juli 2014

Abstract

In intergroup negotiations, negotiators can choose to share private information with the opponent group or they can choose a more dishonest tactic by withholding information. In general people perceive dishonesty as negative. The current study contributed to the current theory by adding threat as a contextual factor that can shift the way in which a

representative’s transgression is viewed. A moderated mediation model was proposed where the interactive effect of threat and representative behavior on blame attributions via moral foundations would be different for low versus high levels of threat. Results showed that under low threat dishonest behavior of the in-group representative leads to more perceived harm and unfairness, which in turn leads to increased blame attributions to the representative. The predicted effect of perceived harm and fairness weakened under high threat, honesty and dishonesty of the leader were perceived to be equally unfair and harmful and equally blamed. The moderated mediation model was however not supported for loyalty. When a negotiation is perceived as less threatful representatives should be careful to use a dishonest negotiating tacting.

Constituents’ Blame Attributions for Ethically Ambiguous Negotiation Tactics of the Representative Under Threat

In most bargaining situations negotiators can choose different tactics concerning how to manage private information. On the one hand negotiators can share the private information with the opponent group which can be costly because of a potential loss in the negotiation. On the other hand negotiators could opt for a more dishonest tactic by withholding

information and hereby increasing their chances on winning in the negotiations. Constituents will be in a dilemma when the moral choice on their group is on the one hand and the

potential gain is on the other hand. How do constituents perceive the morality of their representative?

Previous research has repeatedly shown that dishonesty is commonly viewed as negative and that it usually leads to punishment (Wang, Galinsky & Murningham, 2009). Furthermore, in general individuals are willing to punish dishonest others even when punishment is financially costly and leaves both parties worse-off (Boles, Croson & Murnighan, 2000; Brandts & Charness, 2003; Wang & Leung, 2010). Moreover, people exhibit stronger evaluative reactions toward deception than toward honesty (Wang, Galinsky & Murnighan, 2009). By nature people react more adversely to deceitfulness than to any other personality trait (Anderson & Bushman, 2002). Previous research focused on how we perceive an opponent’s unethicality, but in this study we will be looking at how people perceive their own group member’s transgression. Being in a leadership position inflates the experienced importance of the group's goals, which can lead to the attainment of their goals through dishonest means (Hoyt, Prince, & Poatsy, 2013). But how tolerant are the

constituents towards their representative’s dishonest negotiation tactics? People in general prefer leaders who are honest, fair and just (Martin, Keating, Resick, et al., 2013). When a dishonest representative is chosen, honest individuals will find their moral standard clashing

with their dishonest representative, this in turn will lead to stress and even to more stress if it is impossible to elect a new representative (Dewe, 1993). In the organizational context this can result in negative outcomes such as lower job satisfaction, decreased productivity and higher turnover (Marshall, Barnett, & Sayer, 1997). Therefore it is important in a multi-level representative negotiation context, to understand how constituency members will judge the ethically ambiguous behavior of their representative.

In general people blame other individuals who are acting dishonestly without

sufficient justification (Mikula, 2003). ‘Without sufficient justification’ implicates that under certain circumstances dishonesty can be justified and constituents’ tolerance towards

dishonest behaviors may increase. In the current study we examine threat as a contextual factor which can potentially change how much ingroup members blame dishonest negotiation tactics.

Hypothesis 1: Threat moderates the relationship between representative’s behavior and blame attributions. The relation between representative’s dishonesty and constituents’ blame will be stronger under low threat and weaker under high threat. For representative’s costly honesty, constituents’ blame will be stronger under high threat and weaker under low threat.

Haidt and Graham (2007) propose that people use moral foundations to decide whether a behavior is ethical or unethical. They proposed five moral foundations as

psychological preparations for detecting and reacting to moral issues. These foundations are harm, fairness, loyalty, authority and sanctity. Given their importance in the negotiation context we will just focus on Harm/Fairness and Loyalty.

Harm/Fairness

The moral foundation of harm/care is thought to have evolved as an automatic response to symptoms of distress and suffering in one’s surrounding. Most individuals

generally dislike seeing others suffer and therefore developed the potential to feel compassion for them. Because people have sensitivity to cruelty and harm, they feel approval toward those who prevent or relieve harm. Similarly people feel disapproval toward displays of cruelty and aggression. The moral foundation of fairness can be seen as a requiring

reciprocity of pro-social behavior (favors) and punishment of antisocial and norm-violating behavior (cheating). In sum, harm/care and fairness morals demand that we do not harm but care for others, and treat them fairly and justly. Following moral foundations, dishonesty would lead to a harmful and unfair foundation. Honesty would lead to less blame because it would be considered as a fair deal.

Harm and fairness are the most dominant moral foundations that are used by the majority of people across cultures (Haidt & Graham, 2007). Furthermore, the moral foundations theory states that people vary in the degree to which they construct and value each of the five intuitive foundations. For example perceptions of social dangers predict the tendency to emphasize specific moral foundations pertaining loyalty, authority and sanctity (van Leeuwen & Park, 2009).

Loyalty

Loyalty was described as a result of people living in kin-based groups. In this way people developed emotions related to recognizing, trusting and cooperating with members of one’s in-group while being wary and distrustful of members of other groups. People value those who sacrifice for the in-group and despise those who betray or fail to come to the aid of the in-group, especially in times of conflict.

Research showed that intergroup conflicts does not necessarily arise out of a desire among group members to hurt out-groups but, rather, to help their in-group (Havely, Weisel & Bornstein, 2011). This could imply that constituents would see their representative’s deceptive behavior as helping the group and thus use the loyalty foundation. People may shift

in these foundations under certain conditions. When people judge an in-group versus an outgroup member, they shift moral foundations (Leidner & Castano, 2011). We know from previous research that people perceive a transgression enacted by an in-group member as more acceptable than the identical transgression enacted by an out-group member (Valdesolo & DeSteno, 2007). In other words, in-group members justify their representative’s

transgression, while they blame outgroup or opponent’s transgressions. Staub (1990) proposed that conflicts between goals (for example, not condemning in-group violence in order to protect one’s identity) and moral values (e.g., “do no harm”) can be resolved by shifting moral foundations to less stringent values (e.g., loyalty to and protection of the ingroup) that do not conflict with and therefore serve the goal. It seems logical that honesty is something that’s always rewarded, however when people’s values are under threat people may use different moral foundations to judge a course of action and hence react differently (Leidner & Castano, 2012). We propose that under threat people will use a loyal dimension and see dishonest as good and sharing info as bad.

The Moderating Role of Threat

Keeping the value constant we propose that people’s moral foundations shift under high versus low threat situations. Previous research by de Dreu (2012) showed that people choose more threatening leaders under oxytocin, which is associated with increased in-group favoritism, compared to placebo condition. This research suggests that when people feel that they need to be loyal to their group, they want aggressive and threatful looking leaders who can protect their group rather than nice and warm looking leaders (de Dreu, 2012). Also, other research shows that when a person is experiencing threat in a negotiation, it will result in fewer agreements and/or less cooperative agreements (White, Tynan, Galinsky, &

Wildschut, Pinter, Vevea et al., (2003) proposed that individuals can justify unethical behavior toward their rivaling out-group in terms of trying to help their in-group. Selart and Johansen (2011) found that in an organizational crisis people experience more stress and this stress in turn can lead to more unethical behaviors. Also stress was found to narrow cognitive processes which causes attention to fall on a reduced number of perspectives a negotiator can use (Christensen & Kohls, 2003). This in turn could lead negotiators to see things differently and use a more dominant response. Lewicki and Spencer (1990) indicated that negotiators who expected to interact with a competitive other rated more deceptive tactics such as actively misrepresenting information as appropriate and indicated more willingness to use them than did negotiators who did not expect to interact with a competitive other. In other words under threat people feel stressed and they would use a loyal foundation to judge their representative’s behavior. We proposes that threat would moderate the relationship between representative behavior and the moral foundations.

Hypothesis 2. Threat will moderate the relationship between constituent’s perceived harm/unfairness and loyalty of their representative and representative’s behavior. Under low threat, dishonesty is perceived as more harmful and unfair compared to under high threat. Under high threat dishonesty is perceived as more loyal compared to under low threat.

In sum, people become more competitive and aggressive under threat. Hence it is hypothesized that people under high threat would be more likely to support instrumental dishonesty, and perceive this as more loyal compared to under low threat. Under low threat people would blame dishonesty compared to honesty which puts the group in a disadvantaged position in a competitive context, perceiving dishonest behavior as harmful and unfair. Based on the findings mentioned above, we suggest that threatful negotiations may decrease

loyalty of the representative. In this same line, under high threat constituent members would blame an honest representative more because of his disloyal behavior. Threat would moderate the strength of the mediated relationship between representative behavior and blame

attributions through the moral foundations, such that the mediation will shift from harm/fairness to loyalty under high threat. We propose a moderated mediation: The interactive effect of threat and representative behavior on blame attributions via moral foundations will be different for different levels of threat (see Figure 1 for the conceptual representation).

Hypothesis 3a. Under low threat, a dishonest representative will be perceived as acting harmful and unfair and hence lead to higher blame attributions than an honest representative.

Hypothesis 3b. Under high threat, a dishonest representative will be perceived as acting loyal and hence lead to lower blame attributions than an honest representative. Figure 1.

Conceptual model of the moderated mediatio

Method Sample and Design

The data consist of 89 students from the University of Amsterdam ( 34% male, 66% female). The age of the participants ranged from 18 to 35 years with a mean of 22.

Threat

Moral Foundations

Representative

Participants were students from the University of Amsterdam and were recruited by posters stating ‘Petition’ through the University complex. As compensation, participants were offered 0,5 credit or 5 euros for their participation.

Procedure

Participants were told that they would be participating in a study which is related to their opinion about possible changes in the minimum passing grade at Dutch universities. Until now students have passed an exam if they scored 5.5 which is rounded to 6. The experiment simulated parties who either want to heighten the minimum passing grade to 6.5 or who want to keep the present passing grade. Participants signed a consent form and a confidentiality form stating that they would not share the information they gathered in the experiment. Then participants were set in the cubicles behind a computer. Participants were introduced with some universities wanting to heighten the current minimum passing grade. Six Dutch universities were chosen to negotiate with the Ministry of Education. Participants who agreed on heightening the passing grade were excluded from the experiment.

Participants were told that their own university’s representative wanted to support his arguments by referring to some relevant research about whether Dutch universities are viewed as challenging and competent compared to other universities from over the world. Then participants were shown results from the experiment stating that Dutch Universities were perceived less challenging and less competent compared to UK, USA, and German universities. Furthermore, results pointed out that students at Dutch universities find Dutch examinations rather easy and unchallenging compared to students at UK, USA and German universities. Finally, Dutch universities were only found to be more challenging and

competent compared to Turkish and Italian universities. It may be clear that the research findings clearly contradict the views of the participant’s university. Then participants were shown a fake student magazine article manipulating threat to their current grade system and

in-group representative’s negotiation behavior. We manipulated the degree of threat exerted to the values of the participant in the first paragraph of the article. The information tactic was manipulated in the second paragraph of the article.

Manipulation of threat to values

In our experiment, we kept the value constant in all conditions. We manipulated the degree of threat exerted to the educational value of the participant, namely minimum passing grade. In the experiment there are six universities. To manipulate the level of the threat, we used two factors. Given that perceived threat has been shown to be a function of the size of a group (Quillian, 1995), first we manipulated the number of universities who share the same opinion with the participant’s university. In the low threat condition, participant’s university as well as four other Dutch universities wanted to maintain the current minimum passing grade. In the high threat condition it was just the participants’ university versus the rest of the universities. Furthermore to strengthen our manipulation of threat participants either read that their school was chosen as the pilot school to test the new grading system in the next year (high threat) or another Dutch university was chosen as the pilot school (low threat). In sum, there are two versions of the first part of the newspaper article: High threat and Low threat. We manipulated in-group representative’s negotiation behavior in the second part of the newspaper article. The full articles can be found in Appendix I.

Manipulation of ingroup representative negotiation behavior

To manipulate ingroup representative’s negotiation behavior, the representative either shared or withheld the research findings in the negotiations. In the honest condition, the ingroup representative frankly presented the research findings to the Ministry of Education during the negotiations, even though the findings were against their arguments for the sake of being honest. In the dishonest condition, the ingroup representative decided to withhold the findings that damaged their negotiating position. The article was read at once and not at

different stages. Also, we used test questions to make sure the article was understood correctly.

Following this, participants were asked a series of questions with regard to the

mediating variables, namely the moral foundations participants used to judge this behavior. Then, the participants were asked a series of questions with regard to their blame attributions Finally they were asked questions to check if the manipulation had worked.

Materials

Blame. Blame attributions were measured by three items from Wade’s (1989)

Victimization subscale. On a scale from 1(not at all) to 7(very much), participants were asked to what extent they agree with the following statements: “I blamed Dr. de Bakker because of his negotiation behavior.”,”Dr. de Bakker wronged us in the negotiations” and “Dr. de Bakker is guilty because of his negotiation behavior”. The cronbach’s alpha was .89.

Moral Foundations. Participants first read “The way the representative behaved in the negotiations can be considered as?” They then rated 10 moral relevance items (see Appendix II) on 7-point scales anchored by the labels not agree at all and completely agree. The items were written to be face-valid measures of concerns related to the three foundations Harm, Fairness and Loyalty. For example, Fairness item stated “treating people differently.” Cronbach’s alphas for the three-item measures of each foundation were .85 (Harm and Fairness) and .56 (Loyalty).

Threat manipulation check. Participants rated the threat of values in 5 items on a 7-point scale, ranging from 1 (not at all) to 7 (very much). The Threat scale of 5 items had high reliability (α=.81). Items were: “I feel threatened by the other universities.”, “My values are in danger”, “The educational system I value is in danger”, “The other group seems

Representative behavior manipulation check. The representative behavior manipulation check was measured by four items. Participant rated the behavior on a 1 item Morality scale, ranging from 1 (not at all) to 7 (very much) on item “Dr. de Bakker omitted information that Dutch universities are viewed as less competent to universities in UK, VS and Germany.”.

Results Manipulation Check

The results of the analysis of variance showed that in fact participants perceived higher threat in the high threat condition (M = 4.0, SD = 1.15), compared to those in the low threat condition (M = 3.1, SD = 1.16), F(2,86) = 11.87, p = .001. No other effects reached significance. To check for the manipulation of representative negotiation behavior, we ran an analysis of variance which revealed that participants in the honesty condition perceived their representative’s behavior as less dishonest (M = 3.0, SD = 1.73) than participants in the dishonesty condition (M = 5.2, SD = 1.54), F (2,86) = 63.40 , p < .001. No other effects reached significance.

Confirmatory factor analysis

To test whether the measures of the Moral Foundations questionnaire were consistent with the theory, a principal component analysis (PCA) was conducted on the 10 items with orthogonal rotation (varimax). Four items (Harm1, Harm2, Fair3 and Loy3) were deleted for having an extraction value below .7. The Kaiser-Meyer-Olkin measure verified the sampling adequacy for the analysis, KMO = .74 (‘good’ according to Field, 2009). Bartlett’s test of sphericity χ2(15) = 367.28, p<.001, indicated that correlations between items were

sufficiently large for PCA. An initial analysis was run to obtain eigenvalues for each

component in the data. Two components had eigenvalues over Kaiser’s criterion of 1 and in combination explained 84.10% of the variance. Table 1 shows the factor loadings after

rotations. The items that cluster on the same components suggest that component 1 represents the harm/fairness moral foundation and component 2 represents the loyalty moral foundation. Two moral foundation variables Harm/Fairness and Loyalty were created using the mean of the items in component 1 and respectively component 2.

Table 1

Summary of confirmatory factor analysis of Moral Foundation Questionnaire (N= 89) Rotated Factor Loadings

Item Harm/Fairness Loyalty

Cruel to other universities .90 .07

Treating people differently .94 .07

Acting unfair to other group .92 .20

Showing love for own group .02 .91

Being loyal to own group .61 .95

Being protective of own group .49 .71

Eigenvalues 3.29 1.76

% of variance 54.81 29.30

α .92 .85

Main Analyses

As can be seen in Table 2, a simple association was found between the

representative’s behavior and blame attributions (r = .35, p < .01). Furthermore, participants who reported experiencing higher threat also reported that they viewed the representative’s behavior as more harmful and unfair (r = .40, p < .01) and as more loyal (r = .23, p < .05). The higher harm/fairness scores were related to increased blame toward the representative (r = .76, p < .01) and interestingly enough, higher loyalty scores were related to higher blame of the representative (r = .25, p < .05).

Table 2

Descriptive Statistics and Bivariate Correlations

M SD 1 2 3 4 5 1 Blame 2.81 1.06 (.89) 2 Threat 0.47 0.50 .38** - 3 Rep Behavior 0.48 0.50 .35** .03 - 4 Harm/Fairness 3.61 1.74 .76** .40** .45* (.92) 5 Loyalty 4.03 1.37 .25* .23* .27* .33** (.85) Note. Figures in parentheses are Cronbach’s Alphas.

**p<.01. *p<.05

To directly test our proposed model with the Moral Foundations as depicted in Figure 1, we used a regression-based path analysis with the aid of existing computational tools for estimating and probing interactions and conditional indirect effects in moderated mediation models (Hayes & Matthes, 2009; Preacher, Rucker, & Hayes, 2007). This model in path diagram form can be found in Figure 2 and consists of three distinct sub models. The first model (first panel) is used to test whether threat moderates the relationship between

representative’s behavior and blame attributions (Hypothesis 1). Of interest in this model is an estimate and test of the significance of path c3.The second and third models (in the second panel) are pertinent to whether the Moral Foundations Harm/Fairness and Loyalty moderate the effect of representative behavior on constituents’ blame attributions (the estimate and test of path ad3; Hypothesis 2) and the estimation of the conditional indirect effect of

representative’s behavior on blame attributions through the moral foundations, contingent on threat (Hypothesis 3; Preacher et al., 2007), quantified as the product of the function of ad paths and path b1 and b2. For clarity the arrows are summarized as ad, in a full model three a paths would go to the Harm/Fairness dimension and three d paths would point to the Loyalty dimension. The component of the model in second panel is also used for testing mediated

moderation, in which a moderated effect is carried through a mediator, quantified as the product of a3 and b2 for Harm/Fairness and d3 and b1 for Loyalty (Morgan-Lopez & MacKinnon, 2006).

T

Ordinary least squares regression model coefficients (standard errors in parentheses).

Model1 Model 2

Moral Foundations

Outcome → Blame Harm/Fairness Loyalty

Predictor Coefficient p Coefficient p Coefficient p

Intercept 1.920 .251 1.813 <.001 3.267 <.001 (0.182) (0.267) (0.260) Rep Behavior c1 1.080 .003 a1 2.444 <.001 d1 0.991 .011 (0.266) (0.390) (0.380) Threat c2 1.159 .002 a2 2.298 <.001 d2 0.908 .020 (0.269) (0.395) (.384) Behavior x Threat c3 -.763 .052 a3 -1.968 .001 d3 -0.610 .272 .386 (0.568) (0.552) Model R2 .295 <.001 .432 <.001 .133 .007 Interaction ∆R2 .032 .052 .080 .001 .012 .272 Representative’s Behavior Rep Behavior X Threat Threat _Blame c1 c2 c3 Model 1

Figure 2. The conceptual model in Figure 1 represented in the form of a path model and a

visual depicting the three ordinary least squares regressions estimated and reported in Table 3.

First, does the relationship between representative’s behavior and blame depend on the level of threat? As can be seen, the answer leans towards yes. The results of an ordinary least squares regression in which blame is estimated for representative’s behavior, threat, and their product can be found in Table 3 (Model 1). The interaction is only a marginally

significant result and the variance in blame by these three predictors is about 30 percent. Only a small amount of this explained variance (almost 11%, from .032/.295) is attributable to the moderated effect of threat. We depict this interaction graphically in Figure 3, which plots the conditional effect or ‘simple slope’ of representative’s behavior at low and high levels of threat by using the estimated coefficients from the model. Using simple slope analysis it was shown that under low threat dishonesty was blamed more (M = 3.0, SD = 0.99) than honesty (M = 1.9, SD = 0.95, F(2,86) =14.53, p < .001. Under high threat honesty (M = 3.1, SD =

Harm Fairness Representative’s Behavior Rep Behavior X Threat Threat _Blame c’1 c’2 c’3 b1 ad1 ad2 ad3 Model 2 Model 3 Loyalty b2

0.88) was blamed just as much as dishonesty (M = 3.4, SD = 0.79), F(2,86) =1.52, ns. This shows that constituents became more tolerant towards dishonesty under high threat. Another interesting finding is that constituents blame the representative more under high threat (M = 3.24 SD = 0.84) compared to under low threat (M =2.43, SD = 1.10), F(2,86) = 15.01, p < .001. This implies that when stakes are high constituents blame their representatives regardless of their negotiation tactic.

Figure 3. Blame as a function of Representative Behavior and Threat.

Hypothesis 2 predicts a contingent effect of representative behavior on the moral foundations Harm/Fairness and Loyalty, with threat serving as the moderator. The results can be found in Table 3 (Model 2), which shows the relationship between representative’s

behavior and the moral foundation is contingent on threat, with this moderations accounting for 18.5 percent of the explained variance for Harm/Fairness (.080/.432). However, results show that the interaction terms for representative behavior with threat were not significant in predicting Loyalty, hence no further simple effects analysis was conducted. We depicted the effect for Harm/Fairness graphically in Figure 4. As can be seen, the pattern is similar to the

0 0,5 1 1,5 2 2,5 3 3,5 4 Honesty Dishonesty B la me Low Threat High Threat

one reported with blame as the outcome. Under low threat, dishonest behavior is seen as more harmful and unfair (M = 4.3, SD = 1.42) compared to honest behavior (M = 1.8, SD = 1.10), F(2,86) = 43.97, p < .001. Under high threat, honesty (M = 4.1, SD = 1.39) was viewed equally harmful and unfair as dishonesty (M = 4.6, SD = 1.44), F(2,86) =1.19, ns. Taken together, Hypothesis 2 was supported for Harm/Fairness but not for Loyalty. Given that no significant interaction was found between representative behavior and threat on Loyalty, we no longer tested the mediating role of loyalty in our moderated mediation model.

Figure 4. Harm/Fairness as a function of Representative Behavior and Threat.

Hypothesis 3 proposes that the effect of representative’s behavior on blame is carried in part indirectly through the moral foundation Harm/Fairness, with this process being

moderated by threat. In path analysis, an indirect effect is the product of the effect of a causal agent (here, representative’s behavior) on an intervening variable or “mediator”

(Harm/Fairness in this case) and the effect of the mediator on the outcome (blame in this study) holding the proposed cause constant (e.g., Baron & Kenny, 1986; Hayes, 2009; MacKinnon, Fairchild, & Fritz, 2007). If one of these paths is moderated, then so too is the

0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 Honesty Dishonesty H a rm / F a irn es s Low Threat High Threat

indirect effect (Edwards & Lambert, 2007; Muller, Judd, & Yzerbyt, Preacher et al., 2007). The prior analysis establishes that the path from representative’s behavior to Harm/Fairness is indeed moderated – it is contingent on threat. In the estimated conceptual model using path analysis diagrammed in Figure 2, we estimated the conditional indirect effect of

representative’s behavior on blame as (a1 + a3 threat) b2, where b2is the partial effect of Harm/Fairness on blame from regression model summarized in Table 4, Model 3. Table 4.

Ordinary least squares regression model coefficients (standard errors in parentheses). Model3 Outcome → Blame Predictor Coefficient p Intercept 1.133 <.001 (0.175) Rep Behavior c’1 0.020 .937 (0.249) Threat c’2 0.163 .511 (0.246) Harm/Fairness b2 0.434 <.001 (0.057) Behavior x Threat c’3 0.091 .776 (0.320) Model R2 .582 <.001 Interaction ∆R2 .000 .791

Notice that holding representative behavior and threat constant, those who reported higher harm and unfairness also reported more blame (b2=0.434, p<.001). Combining the conditional effect of the representative’s behavior on Harm/Fairness with the effect of Harm/Fairness on constituent’s blame towards the representative results in the conditional indirect effect of representative’s behavior on blame through Harm/Fairness: (2.444-1.968 x threat; 0.434). The effects are stated in the indirect effect column in Table 5. Observe that as threat gets higher the effect gets weaker, meaning that the conditional indirect effect

To further assess the findings of moderated mediation relationships, we examined whether the magnitude of the conditional indirect effect of representative behavior via Harm/Fairness was different for constituents across high and low levels of threat. We used Hayes et al.’s (2013) statistical significance test (model 8) to compute confidence intervals for the conditional indirect effect, as can be seen in Table 5. For Harm/Fairness, Hypothesis 3 was supported for low threat. Under high threat, however, this indirect effect was not

significant. Under low threat dishonest behavior of the in-group representative leads to more perceived harm and unfairness, which in turn leads to increased blame attributions to the representative. In line with the hypothesis the predicted effect of perceived harm and fairness weakened under high threat, honesty and dishonesty of the leader were perceived to be equally unfair and harmful and equally blamed.

Table 5.

Model Coefficients for the Conditional Process Model.

Indirect Effect Direct Effect

Threat Effect 95% Bias-Corrected Bootstrap CI Effect SE p Low 1.065 0.698 to 1.492 0.030 0.254 0.906 High 0.208 -0.209 to 0.535 0.116 0.221 0.602

Discussion

The current study investigated whether or not constituent members would blame a dishonest representative more as opposed to an honest representative. Honesty in this study was associated with a loss in a competitive negotiation. A moderated mediation model was tested which showed that, under low threat dishonest behavior of the in-group representative leads to more perceived harm and unfairness, which in turn leads to increased blame

attributions to the representative. The moderated mediation model was however not

dishonesty: honesty and dishonesty of the representative were equally blamed. An

explanation for this finding could be that in the dishonesty condition, the representative used the argument that following research the university is still seen as better than marginal universities. Even though the information that could damage the in-group’s negotiation position was omitted, the representative could still have been blamed because of using a weak argument. This weak argumentation could also account for the finding that the representative was blamed more under high threat compared to under low threat regardless of his

negotiating behavior. Because constituents can become more critical when the pressure on the negotiation is high (Boin, & Hart, 2003), we can infer that constituents can view both weak argumentation and costly honesty as detrimental to the negotiation and both were therefore blamed more under high threat. No influences of threat and loyalty were found on constituents’ blame towards their representative. In fact, a weak positive association was found between perceived loyalty and blame. A possible explanation for this is that threat is an antecedent of both loyalty and blame. In other words, a threat here can be a confounding variable because higher threat on the one hand leads to in-group favoritism (and thus higher perceived loyalty; Brewer, 1999) and on the other hand is associated with more criticism (and thus more blame attributions; Boin, & Hart, 2003). Again here, both honesty and dishonesty of the representative can be viewed as detrimental to the negotiation and even though the representative is seen as acting loyal, constituents could still disagree with the behavior the representative displayed. What is more, concerning loyalty: in moral foundation theory (Haidt, 2007) virtues of patriotism and self-sacrifice for the group underlie this foundation. In other words, a representative is expected to be viewed as loyal when he chooses the group’s interest over his self-interest. This experiment was designed in such a way that the

representative’s self-interest aligns with that of the group. In future research effects of loyalty may be found when it is made explicit that a dishonest negotiation style would lead to

punishment of the representative on the one hand and would be beneficial for the group on the other hand.

Theoretical and Practical Implications

These findings extend our knowledge how constituents view their representative’s behavior during a negotiation. Under different levels of threat constituents can evaluate the same behavior differently and representatives need therefore be aware of the context of the negotiation when deciding which tactic to use. Where previous research focused mainly on how we perceive an opponent’s unethicality, this study looked at how people perceive their own group member’s transgression. Also for the first time that we know of this study tested constituents’ views of negotiation behavior when the negotiator could choose between costly honesty and dishonesty. Threat was shown as a contextual factor that can shift the way in which a representative’s transgression is viewed. Previous research shows that being in a leadership position inflates the experienced importance of the group's goals and that this can lead to the attainment of their goals through dishonest means (Hoyt et al., 2013). It seems logical that in situations where the other negotiation option is to use a costly honesty negotiation tactic, representatives will even more be inclined to use a dishonest negotiation tactic. The current research on the other hand, tells leaders must be careful to use a dishonest negotiation tactic; when there is less threat perceived, the leaders will be blamed for their dishonesty. Blame in the long run is possibly more detrimental to the negotiation: When constituents blame their representative for their deceitful actions, this could result in a lack of trust (Driedger, Mazur & Mistry, 2013). Distrust in turn could lead to fractions in the

constituency, which leaves the group even worse off concerning the negotiation position. A representative can’t represent a divided constituency, so the other party would have the advantage of playing into the potential unrest. This tactic is as old as Roman times and it is known as divide and conquer. This strategy is used to gain and maintain power by breaking

up larger concentrations of power into pieces that individually have less power than the one implementing the strategy (Belkin, 2006).

Limitations and Future Research

A limitation of the study is that the magazine article can be viewed as being made for the research. Since participants are psychology students in a laboratory setting they might expect that the information was made-up and might therefore feel less threatened than that they would feel when they were really convinced the article was genuine. Future research can address this problem by using actors in a video to create a more real-life setting like a news report. Also a question can be added to check whether the respondents actually believed the issue in question. Another limitation is that no behavioral measures were used. A behavioral measure can influence the external validity in the sense that it gives people the opportunity to actually influence the status quo. Also to say a leader is harmful, disloyal or unfair is less socially acceptable because it can be viewed as betrayal (Moreland & McMinn, 1999) and therefore a more implicit attitude is hard to assess. Future research can address this limitation by using an implicit task such as the lexicon decision task (LTD). An LTD offers the

possibility of a dissociation between information processing and conscious awareness of that processing. Furthermore, future research can address whether under high threat people focus more on what is harmful and unfair to themselves as opposed to others. Because there is more focus on the self, both a bad argument and honesty are blamed to a larger extent under high threat compared to under low threat because it harms the in-group. The expected effect of the constituents’ shift from the foundation harm and fairness to loyalty might occur only under higher perceived threat than that was measured in this study. A stronger manipulation of threat might trigger this shift. One way this can be accomplished is through picking a more core identity value for the negotiation. When certain values such as abortion or a possible loss of payment might invade a sense of privacy or autonomy more, this could lead to more

experienced threat. Also the main effect of threat implies that when stakes are high people engage in more blaming towards their representative regardless the outcome of the

negotiation. Future research can add an option for participants to describe the behavior they would want the representative to display. In this way it can be examined whether constituents blame more because they expected a better behavior or whether they blame any behavior because of higher threat.

Conclusion

In sum, when leaders in negotiations are faced between choosing a tactic which is costly honest and on the other hand deceiving the other party, threat is an important factor which decides how the leaders are viewed after the negotiation. When values are at stake constituents will become more tolerant towards dishonesty. However when the negotiation is less threatful representatives should be careful to use a dishonest negotiating tactic.

Literature

Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of

Psychology, 53, 27-51.

Baron, R, M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of

Personality and Social Psychology, 51, 1173-1182.

Belkin, A. (2006). United we stand? Divide-and-conquer politics. New York: State University of New York Press.

Boin, A., & ‘t Hart, P. (2003). Public leadership in times of crisis: mission impossible?

Public Adminitration Review, 5, 544-554.

Boles, T. L., Croson, R. T. A., & Murnighan, J. K. (2000). Deception and retribution in repeated ultimatum bargaining. Organizational Behavior and Human Decision

Processes, 2, 235-259.

Brandts, J., & Charness, G. (2003). Truth or consequence: an experiment. Management

Science, 49, 116–130.

Brewer, M. B. (1999). The psychology of prejudice: ingroup love or outgroup hate? Journal

of Social Issues, 3, 429-444.

Christensen, S. L., & Kohls, J. (2003). Ethical decision making in times of organizational crisis. Business Society, 3, 328-358.

Dewe, P. (1993). Measuring primary appraisal: scale construction and directions for future research. Journal of Social Behavior and Personality, 8, 331-351.

De Dreu, C. K. W. (2010). Social conflict: The emergence and consequences of struggle and negotiation. In: S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds). Handbook of social psychology, 5th ed., (pp. 983–1023). New York, NY: Wiley.

De Dreu, C. K. W., Shalvi, S., Greer L. L., van Kleef, G. A., & Handgraaf, M. J. J. (2012). Oxytocin motivates non-cooperation in intergroup conflict to protect vulnerable in- group members. Plos One, 11.

Driedger, S. A., Mazura, C., & Mistry, B. (2013). The evolution of blame and trust: an examination of a Canadian drinking water contamination event. Journal of Risk

Research, DOI: 10.1080/13669877.2013.816335

Edwards, J. R., & Lambert, L. S. (2007). Methods for integrating moderation and mediation: A general analytical framework using moderated path analysis. Psychological

Methods, 12, 1-22.

Field, A. (2009). Discovering statistics using SPSS: and sex and drugs and rock ‘n’ roll, 3rd edn. London: Sage Publications.

Haidt, J., & Graham, J. (2007). When morality opposes justice: conservatives have moral intuitions that liberals may not recognize. Social Justice Research, 1, 7-34.

Halevy, N., Weisel, O., & Bornstein, G. (2011). In-group love and out-group hate in repeated interaction between groups. Journal of Behavioral Decision Making, 2, 188-195. Hayes, A. F. (2009). Beyond Baron and Kenny: Stratistical mediation analysis in the new

millennium. Communication Monographs, 76, 408-420.

Hayes, A. F., & Matthes, J. (2009). Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementations. Behavior Research

Methods, 41, 924-236.

Hoyt, C. L., Prince, T. L., & Poatsy, L.(2013) The social role theory of unethical leadership.

Leeuwen van, F., & Park, J.H. (2009). Perceptions of social dangers, moral foundations, and political orientation. Personality and Individual Differences, 3, 169-173.

Leidner, B. & Castano, E. (2011). Morality shifting in the context of intergroup violence.

European Journal of Social Psychology, 1, 82-91.

Lewicki, R. J., & Spencer, G. (1990). Lies and dirty tricks. Paper presented at the meeting of the International Association for Conflict Management, Vancouver, B.C.

MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual Review

of Psychology, 58, 593-614.

Marshall, N. L., Barnett, R. C., & Sayer, A. (1997). The changing workforce, job stress and psychological distress. Journal of Occupational Health Psychology, 2, 99-107.

Martin, G. S., Keating, M. A., Resick, C. J., Szabo, E., Kwam, H. K., & Peng, C. (2013). The meaning of leader integrity: A comparative study across Anglo, Asian and Germanic cultures. The Leadership Quarterly, 24, 445-461.

Moreland, R. L., & McMinn, J. G. (1999). One but not forgotten: Loyalty and betrayal among ex-members of small groups. Personal and Social Psychology Bulletin, 12, 1476-1486.

Morgan-Lopez, A., & MacKinnon, D. P. (2006). Demonstration and evaluation of a method for assessing mediated moderation. Behavior Research Methods, 38, 77-89.

Mikula, G. (2003). Testing an attribution-of-blame model of judgements of injustice.

European Journal of Social Psychology, 6, 793-811.

Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005). When mediation is moderated and moderation is mediated. Journal of Personality and Social Psychology, 89, 852-863. Preacher, K. J., Rucker, D. D., & Hayes, A. F. (2007). Assessing moderated mediation

hypotheses: Theory, methods, and prescriptions, Multivariative Behavioral Research,

42,185-227.

Preacher, K. J., & Hayes, A. F. (2007). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, 40, 717-731. Selart, M.& Johansen, S. T. (2011). Ethical decision making in organizations: the role

of leadership stress. Journal of Business Ethics, 99, 129-143.

Staub, E. (1990). Moral exclusion, personal goal theory and extreme destructiveness. Social

issues, 1, 47-64.

Valdesolo, P., & DeSteno, D.A. (2007). Moral hypocrisy: Social groups and the flexibility of virtue. Psychological Science, 18, 689-690.

Wade, S. H. (1989). The development of a scale to measure forgiveness. Unpublished

doctoral, Fuller Graduate School of Psychology, Pasadena, CA.

Wang, C. S., & Leung, A. K. (2009) The cultural dynamics of rewarding honesty and punishing deception. Personality and Social Psychology Bulletin, 36, 1529–1542. Wang, C. S., Galinsky, A. D., & Murnighan, J. K. (2009). Bad drives psychological reactions

but good propels behavior: Reactions to honesty and deception. Psychological

Science, 20, 634-644.

White, J. B., Tynan, R., Galinsky, A. D., & Thompson, L. Face threat sensitivity in negotiation: roadblock to agreement and joint gain. Organizational Behavior and

Human Decision Processes, 2, 102-124.

Wildschut, T., Pinter, B., Vevea, J.L., Insko, C.A., & Schopler, J. (2003). Beyond the group mind: A quantitive review of the interindividual-intergroup discontinuity effect.

APPENDIX I

The magazine articles are made up of a combination of either high threat or low threat with either honesty or dishonesty. This makes up for four conditions. Below we show the separate paragraphs.

LOW THREAT HIGH THREAT

Will the minimum passing grade be heightened to 6.5?

AMSTERDAM - The minimum passing grade for exams at Universities needs to be heightened to a 6.5, according to the representative of the University of Utrecht. whereas University of Amsterdam,

Groningen, Leiden, Delft and Maastricht would like to maintain the passing grade as 5.5. The board of executives from the Ministry of Education will decide in the following weeks about possible new legislation which will commence in study year 2015-2016 at every Dutch university. Topic in the coming negotiations will also be the controversial pilot test which will

commence the coming study year. By means of a lottery Leiden University has been chosen to test the pilot. From coming September students at this university will already need a 6.5 to pass an exam. Until now student have passed an exam if they score 5.5 which is rounded to 6. Some professors from Universities of Utrecht thought students can get a 5.5 with minimal effort and that the Dutch universities cannot be perceived as competent and challenging internationally. To improve the image of Dutch universities internationally, they believe they need to increase the passing grade to 6.5 so that students can study harder and get higher grades. The representative of the UvA thinks it’d be better to invest in the quality of education rather than making it harder for students. Accordingly, in the coming weeks representatives from the UvA , Groningen, Leiden, Delft and

Maastricht(maintain) versus Utrecht

(heighten) will initiate a negotiation with the ministry of education in the Netherlands in

Dutch students under threat: passing grade to 6.5.

AMSTERDAM - The minimum passing grade for exams at Universities needs to be heightened to a 6.5, according to the

representative of the Universities of Utrecht, Groningen, Leiden, Delft and Maastricht. Only the University of Amsterdam would like to maintain the passing grade as 5.5. The board of executives from the Ministry of Education will decide in the following weeks about possible new legislation which will commence in study year 2015-2016 at every Dutch university. Topic in the coming negotiations will also be the controversial pilot test which will commence the coming study year. By means of a lottery the

University of Amsterdam has been chosen to test the pilot. From coming September

students at this university will already need a 6.5 to pass an exam.

Until now student have passed an exam if they score 5.5 which is rounded to 6. Most professors from Dutch Universities thought students can get a 5.5 with minimal effort and that the Dutch universities cannot be

perceived as competent and challenging internationally. To improve the image of Dutch universities internationally, they believe they need to increase the passing grade to 6.5 so that students can study harder and get higher grades. The representative of the UvA thinks it’d be better to invest in the quality of education rather than making it harder for students. Accordingly, in the coming weeks representatives from the UvA (maintain) versus Utrecht, Groningen, Leiden, Delft and Maastricht (heighten) will initiate a negotiation with the ministry of education in the Netherlands in order to

order to review the passing grade for Dutch examinations at the universities. Because all the other universities are in the same team as the UVA it doesn’t seem hard for the

university to defend itself in order to keep the current minimal grade.

review the passing grade for Dutch

examinations at the universities. Because all the other universities are in the other team as the UVA it seems this university is going to have a hard time to defend its students during in the negotiation to keep the current minimal grade.

HONESTY DISHONESTY

In a press conference, it was reported that the representative of UvA Dr. DE BAKKER, presented some research findings in the negotiations. Dr. De Bakker said: “Although the findings of a research seem contradictory to our argument, I believe I still need to share the results with you”. The UvA

representative admitted that Dutch

Universities are perceived less challenging and less competent compared to UK, USA, and German universities. Furthermore he said that students at Dutch universities find Dutch examinations rather easy and unchallenging compared to students at UK, USA and German universities. Finally he added that Dutch universities were only found to be more challenging and competent compared to Turkish and Italian universities. After what Dr. de Bakker said in the negotiations, both students and teachers are predicting that what he said in the negotiations will put his group in a very disadvantaged position in the negotiations. The negotiations are coming to close in the following months, and only then the Ministry of Education will decide on a new legislation. Ministry of Education purely stated that their decision will be solely based on the negotiation performance rather than how many universities are on either side.

In a press conference, it was reported that the representative of UvA, Dr. DE BAKKER, presented some research findings in the negotiations. Dr. de Bakker said “we have empirical evidence that shows that Dutch universities are already perceived to be challenging and competent internationally. Dutch universities were found to be more challenging and competent compared to Italian en Turkish universities”. After what dr. de Bakker said in the negotiations, both students and teachers are predicting that what he said in the negotiations will put his group in an advantaged position in the negotiations. The negotiations are coming to close in the following months, and only then the Ministry of Education will decide on a new legislation. Ministry of Education purely stated that their decision will be solely based on the

negotiation performance rather than how many universities are on either side.

APPENDIX II

MORAL FOUNDATION QUESTIONNAIRE

The way the representative behaved in the negotiations can be considered as: 1- Causing harm to other people (harm)

2- Caring for the weak and vulnerable (harm) 3- Being cruel other universities (harm) 4- Treating people differently (fairness) 5- Acting unfair to the other group (fairness) 6- Unfair to my group (fairness in-group) 7- Showing love for own group (loyalty) 8- Being loyal to own group (loyalty) 9- Betrayal of own group (loyalty)