Motives for Punishing Powerful vs. Prestigious Offenders

Status can be seen as power over valued resources or as prestige that lies in the eyes of the beholder. In the present research, we examine how power versus prestige influence observers’ punishing motives. Possession of power implies the capacity to harm and elicits threat and therefore should trigger stronger incapacitative motives for punishing an offender. In contrast, prestige signals the observer's admiration of the target and therefore should elicit a strong motivation to help an offender reintegrate into society. Studies 1 and 2 manipulated an offender's status (power vs. prestige vs. control) and group identity (ingroup vs. out-group). Supporting our hypotheses, both studies revealed that observers had stronger incapacitative motivations towards powerful as opposed to prestigious offenders, particularly when the offender came from the ingroup. Study 2 also showed that observers had stronger restorative motives towards a prestigious as opposed to powerful offender. Contrary to expectations, group identity did not moderate the effect of status on observer's restorative motives. Study 3 manipulated power and prestige separately and showed that power

Yet, no study has so far investigated how power vs. prestige of an offender shapes the motives with which third party observers seek punishment.Such motives are people's justifications when assigning punishments, and therefore they have predictive value as they determine how observers eventually treat offenders (Carlsmith et al., 2002).In other words, whether or not observers punish an offender and how harshly they treat them largely depends on their underlying motives.Besides punishing offenders to pay them for their actions (e.g.give them their just deserts; Carlsmith et al., 2002;Goldberg et al., 1999;Kant, 1797;Van Prooijen, 2018;Wenzel & Thielmann, 2006), people may punish an offender to incapacitate them and protect society, or help them restore their relationship with victims and reintegrate into society (incapacitative versus restorative motives).Incapacitative motives are instrumental and seek to protect society by restraining offenders such that they cannot commit more harm (Carlsmith & Darley, 2008;Kahane et al., 2018;Van Prooijen, 2018).Restorative motives (de Beaumont & Tocqueville, 1833;Saleilles, 1898) seek to help offenders learn from their mistakes, improve their behaviour, and eventually reintegrate them into society (Zehr, 1997).How offenders' power or prestige influences observers' punishing motives is yet unknown, however.In the present paper, we investigated people's underlying motives for punishing powerful vs. prestigious offenders, and hypothesized stronger incapacitative motives for punishing a powerful offender yet stronger restorative motives for punishing a prestigious offender.Moreover, we investigate whether these effects are particularly pronounced when the offender is an ingroup (as opposed to outgroup) member.

Status as power versus prestige
People punish high status offenders more harshly as compared to low status ones (Abrams et al., 2014;Hoyt et al., 2013;Kellerman, 2004) because those offenders can influence lay people's welfare to a greater extent and cause more harm (Fousiani & Van Prooijen 2022a, 2022b;Gruenfeld et al., 2008;Kipnis, 1972;Lammers et al., 2010).However, another line of research reveals opposite effects: High status offenders are treated more leniently by third party observers, as compared to their low status counterparts (e.g.Abrams et al., 2008Abrams et al., , 2018)), as people identify more strongly with high status offenders and see their immoralities in a less negative light (Cialdini et al., 1976).How can we reconcile these seemingly contradictory findings?
It is noteworthy that the literature identifies two forms of social status, namely power and prestige (see Blader & Chen, 2012;Blader et al., 2016;Cheng et al., 2013;Kakkar et al., 2020).Power (or dominance) 1 is defined as control over valued resources (Fiske, 2010) and as the ability to produce intended outcomes (Emerson, 1962).Powerful individuals are seen as assertive, controlling and dominant (Maner & Case, 2016), and as striving for gaining superiority and prevalence over others (Cheng et al., 2013, see also Laustsen & Petersen, 2015;Van Vugt, 2006).In contrast, prestige is defined as respect, esteem, recognition and admiration that an individual has in the eyes of an observer because of their personal qualities such as knowledge, expertise and skills (Anderson & Kilduff, 2009;Blader & Chen, 2012;Henrich & Gil-White, 2001;Maner & Case, 2016;Ridgeway & Erickson, 2000).Prestige has its origins externally as it refers to social worth that observers themselves attribute to an individual (Blau, 1964;Homans, 1961;Ridgeway & Erickson, 2000).Differently put, 'prestige is determined by the perceiver and, as such, necessarily lies in the eyes of the beholder' (Maner & Case, 2016: 138).Kakkar et al. (2020) found that powerful (versus prestigious) offenders are punished more harshly, and hence suggest different effects of power versus prestige on third party punishment.Indeed, power is more closely related to the perception of an actor as immoral and intentional, triggering harsher punishment.In contrast, prestige prompts perceivers to give offenders the benefit of the doubt.People's motives for punishing powerful as opposed to prestigious offenders therefore might also vary.We assume that powerful offenders will elicit stronger incapacitative motives, as such offenders are viewed as more harmful and dangerous for society given their access to valued resources (Hypothesis 1a).Conversely, people are more willing to reinclude prestigious offenders into society, triggering stronger restorative motives (Hypothesis 1b).

The moderating role of group identity of an offender
Prior research suggests that the affiliation of an offender plays a core role in people's justice-related decisions.Indeed, research on the 'black sheep effect' (BSE; Marques & Yzerbyt, 1988;Marques et al., 1988) shows that people treat ingroup offenders more negatively than outgroup offenders.More specifically, immoral-doing triggers harsher reactions towards ingroup than outgroup offenders, because ingroup offenders threaten the reputation of the group (Brambilla et al., 2013;Pagliaro et al., 2013), social cohesion (Tyler, 1997;Vidmar, 2000) and group values (Okimoto & Wenzel, 2010;Rullo et al., 2015;Sankaran et al., 2017;Van der Toorn et al., 2015).Consistent with this theorizing, Fousiani et al. (2019) found that observers have stronger incapacitative than restorative or retributive punishment motives towards ingroup as opposed to outgroup offenders.Incapacitative punishment motives aim to send a symbolic message by excluding the offender, thus condemning the offender's actions and publicly validating group values.
Based on the above theorizing, we hypothesize that the effect of status (power vs. prestige) on motives for punishment will vary as a function of an offender's group identity.Immorality enacted by powerful ingroup offenders should be considered as a particularly strong threat to one's group identity, given their capacity to abuse their power towards other group members, exploit valued group resources, and cause harm to the group.We therefore predicted stronger incapacitative motives for punishing a powerful as opposed to prestigious offender especially when that offender comes from the ingroup (see also Brambilla et al., 2013;Pagliaro et al., 2013) (Hypothesis 2a).In contrast, a prestigious (as opposed to powerful) ingroup offender will trigger stronger restorative motives for punishment (Hypothesis 2b), as such an offender is viewed less negatively, and has less control over resources to enable power abuse.

1
The literature uses the terms 'power' (see Blader & Chen, 2012) and 'dominance' (see Kakkar et al., 2020) interchangeably when referring to individuals who are seen as controlling, assertive and striving for superiority through their ability to control resources and withhold them.In fact, the term power is more commonly used in social psychology research, whereas the term dominance is more commonly used in personality psychology and the evolutionary psychology literature (see Cheng et al., 2013 for this distinction).In the current work, we use the term power instead of dominance, although dominance could have been used instead.

Overview of the studies
Three experiments investigated the effect of offender status on punishment motives, moderated by group identity of the offender (ingroup vs. outgroup).Although not a core hypothesis, in Studies 1 and 2, we also added a control condition (power vs. prestige vs. control) to investigate possible differences in people's punishing preferences when no information about an offender's status is available.To further disentangle the relationship between power and prestige, Study 3 manipulated status separately (high vs. low power and high vs. low prestige; cf.Blader & Chen, 2012), and investigated the mediating role of group threat and perceived harm capacity of the offender in these relationships.The offenses focussed on sexual harassment (Study 1) and football vandalism (Studies 2 and 3).Besides punishment motives, Studies 2 and 3 also assessed observers' punishment intent.2Moreover, since powerful people are often seen as more capable of causing serious harm given their extensive resource control (Fiske, 1993) in all three studies, we controlled for the perceived severity of harm3 .Studies 1 and 2 also manipulated group identity (ingroup vs. outgroup), while in Study 3 we kept group identity constant.

Participants
Four hundred and seventy-four participants (320 females; M age = 34.49,SD = 10.10)living in various cities in Greece took part in this study.Of the participants, 73.4% were working as employees or managers, 12.4% were students, and 7.8% were unemployed.Moreover, 13.7% had finished high school and 73.9% had higher education.Of the participants, 6.3% did not indicate their occupation or education.An a priori power analysis revealed that 400 participants were required to achieve 95% power to detect an interaction with a medium effect size of f = .25.

Experimental design and procedure
Graduate students recruited respondents using their social network and work connections.Participants were invited via social media (e.g.Facebook, LinkedIn and Twitter) to take part in this online study.Participation lasted about 8-10 minutes and was not paid.We manipulated the offender's status in vignettes similar to Blader and Chen (2012).Participants read a scenario presenting the offender ('Mr X.') as a powerful division manager in a large company having control over an unusually large amount of resources (power condition) vs. a prestigious person in a large company who had acquired a very positive reputation in his division (prestige condition) vs. a person working in a very large company (control condition) (see Appendix S1 for the complete vignettes).As manipulation checks, participants then answered the following questions (1 = not at all, 7 = to a great extent): 'Mr.X. has control over a lot of resources in his division' (power manipulation check); 'Mr.X. is held in high regard in his division'; 'Mr.X. is widely accepted by others in his division' (prestige manipulation check; Blader & Chen, 2012).
Subsequently, we manipulated the offender's group identity.Participants imagined working together with Mr X. in the same division (ingroup) vs. working in a different division than Mr X.'s (outgroup).As a manipulation check, participants indicated whether they were an employee in the same division with Mr. X ( yes / no).Participants were randomly assigned to one of the six experimental conditions.

Motives for punishment
We measured incapacitative motives with the 3-item homonymous subscale of the motives for punishment scale (Fousiani & Demoulin, 2019;Fousiani & Van Prooijen, 2019;Fousiani et al., 2019;Fousiani & Van Prooijen 2022a, 2022b), (e.g.'Mr X. should be ousted from the division in order for the rest of the employees to be better protected'; α = .80).Similarly, we measured restorative motives with the homonymous subscale of the motives for punishment scale (e.g.'The best way to address this sort of behaviour is to offer Mr. X the chance to learn from his mistakes and become a better person'; α = .60).Finally, we assessed retributive motives (e.g.'Mr.X should be assigned a punishment equivalent to the magnitude of his offense; not more lenient or harsher than that', α = .62). 4 Participants indicated their agreement on a 7-point Likert scale (1 = absolutely disagree, 7 = absolutely agree).

Control variables
Severity of harm was measured with one item: 'The act of Mr X. has serious ramifications for the members of his division' (1 = absolutely disagree, 7 = absolutely agree).

Results
Correlations, means and standard deviations are in Table 1.

Manipulation checks
An ANOVA with offender's status (power / prestige / control) as independent variable revealed a significant status effect on the power manipulation check, F(2,471) = 60.57,p < .001,η 2 = .21.Participants perceived the offender as more powerful in the power as opposed to the control or prestige conditions (Mpower = 6.45,SD = 1.11,Mcontrol = 5.25, SD = 1.48,Mprestige = 4.78, SD = 1.56).Also on the prestige manipulation check, the effect of status was significant F(2,471) = 95.20,p < .001,η 2 = .29.Participants perceived the offender as having more prestige in the prestige as opposed to power or control conditions (Mpower = 4.65, SD = 1.12,Mcontrol = 4.69, SD = 1.05,Mprestige = 6.13,SD = .99).None of the participants failed the manipulation check for the group identity manipulation, and hence, all participants were included in the analyses.We conclude that the manipulations worked as intended.

Discussion
Results showed that observers had stronger incapacitative motives towards powerful as opposed to prestigious offenders.Moreover, in line with the BSE (Marques & Yzerbyt, 1988;Marques et al., 1988), this effect emerged only when the offender was an ingroup member.These findings support Hypotheses 1a and 2a, and are consistent with the underlying argument that people display stronger exclusionary reactions to powerful as opposed to prestigious ingroup offenders because such offenders have a higher potential to cause harm to the group (Thomas & Cage, 1976;Thomas & Foster, 1975).Status and group identity did not influence restorative motives, however, and therefore Hypotheses 1b and 2b were not supported.
A limitation of Study 1 is that the design may have varied differences in self-relevance: Particularly in the powerful ingroup offender condition, participants might have seen themselves as directly impacted by the offender's act.After all, a powerful ingroup offender is more likely to sexually harass another ingroup member.This limitation is addressed in Study 2, which focusses on football hooliganism.Moreover, Study 2 includes an additional measure: punishment intent.Although Kakkar et al. (2020) have already demonstrated people's harsher reaction to powerful as opposed to prestigious offenders, they have not investigated the moderating role of an offender's group identity in this relationship.We hypothesize that observers will display a stronger intention to punish a powerful as opposed to prestigious offender, especially when that offender comes from the ingroup (rather than an outgroup) (Hypothesis 3).
T A B L E 2 Means and standard deviations for the study variables across experimental conditions (Study 1) Note: All ratings were on 7-point scales ranging from 1 = absolutely disagree to 7 = absolutely agree.

ST U DY 2 Method
Participants A total of 224 British participants (160 females; M age = 36.88,SD = 11.46)took part in this study.Of the participants, 68% were working as employees or managers, 20.6% were unemployed, and 11.3% were students.Moreover, 34% had finished high school and 66% had higher education.A sensitivity power analysis revealed that this sample yields 80% power to detect an interaction with an effect size of f = .27.

Experimental design and procedure
We again manipulated offender's status and group membership in vignettes.The offender ('Mr.X') was presented as a member of a famous football team in the United Kingdom.who had instigated a football hooliganism episode.The offender was either a powerful member of the football team managing a large amount of the team's resources (status as power), or a well-reputed and well-accepted member of the team (status as prestige), or his status was undefined (control condition).Moreover, in the ingroup condition, participants were asked to take the perspective of a committed supporter of the same or a different football team (see Appendix S1 for the full vignettes).Manipulation checks were similar as Study 1.The study was conducted online via Prolific and lasted 8-10 minutes approximately.Participants got paid £1.

Punishment intent
We assessed punishment intent with a 1-item bipolar scale (1= Mr X should not be punished, 7= Mr X should be punished).

Control variables
Severity of harm was measured with one item: 'The act of Mr X. is very serious' (1= absolutely disagree, 7= absolutely agree).

Results
Correlations, means and standard deviations are in Table 3.

Manipulation checks
An ANOVA with offender's status (power/prestige/control) as independent variable yielded a significant status effect on the power manipulation check, F(2,219) = 90.16,p < .001,η 2 = .45.Participants perceived the offender as more powerful in the power condition versus the control or prestige conditions (Mpower = 6.45,SD=1.03,Mcontrol = 3.68, SD=1.12,Mprestige=4.13,SD = 1.67).On the prestige manipulation check, the effect of status also was significant F(2,219) = 108.50,p < .001,η 2 = .50.Participants attributed more prestige to the offender in the prestige condition as opposed to the power or control conditions (Mpower=4.69,SD=1.20,Mcontrol = 3.88, SD=1.24,Mprestige=6.47,SD = .83).None of the participants failed the manipulation checks for group identity, and therefore no participants needed to be excluded based on this criterion.These results suggest that the manipulations worked as intended.

M
F I G U R E 2 Punishment intent across experimental conditions (Study 2).Ratings were on a 7-point scale 1= Mr X should not be punished, 7= Mr X should be punished F(2,106) = .78,p = .46,η 2 = .02.Finally, the univariate interaction effects were not significant for restorative and retributive motives, Fs < 1, providing no support for Hypothesis 2b.

Discussion
Replicating Study 1 and supporting Hypotheses 1a and 2a, participants had stronger incapacitative motives towards a powerful versus prestigious offender, and these effects were particularly pronounced for an ingroup offender.Moreover, participants had stronger restorative motives towards prestigious than powerful offenders, supporting Hypothesis 1b.Status did not interact with group identity on restorative motives for punishment, again not supporting Hypothesis 2b.Finally, people indicated a stronger intention to punish powerful ingroup as opposed to prestigious ingroup offenders, supporting Hypothesis 3.

ST U DY 3
In Study 3, we expanded on the previous studies by separately manipulating power and prestige, given that power and prestige are distinct dimensions of status (Blader & Chen, 2012).Furthermore, Study 3 held group membership of the offender constant (i.e.ingroup), and examined the underlying mechanisms that drive the effects of power and prestige on punishment motives.Our line of reasoning suggests that immorality (see Brambilla et al., 2021) by powerful ingroup members is particularly threatening to Note: Motives for punishment scales: All ratings were on 7-point scales ranging from 1 = absolutely disagree to 7 = absolutely agree; Punishment intent scale: Ratings were on a 7-point scale 1 = Mr X should not be punished, 7= Mr X should be punished.

F I G U R E 3
Motives for punishing offenders across experimental conditions (Study 2).All ratings were on 7-point scales ranging from 1 = absolutely disagree to 7 = absolutely agree one's group identity, given that power holders are often seen as group representatives (Tyler & Lind, 1992).Prior research has found that ingroup threat (i.e. a danger to the in-group's well-being and reputation; Stephan et al., 2002) mediates the relationship between immoral doing and incapacitative motives (Fousiani et al., 2019).In Study 3, we therefore hypothesize that a high (as opposed to low) power offender will trigger higher punishment intent and stronger incapacitative motives (Hypothesis 4a) through increased ingroup threat (Hypothesis 4b).In contrast, a high prestige (as opposed to low prestige) offender, who is held in high regard, will trigger lower punishment intent and stronger restorative punishment motives (Hypothesis 5a) through decreased ingroup threat (Hypothesis 5b).
As additional mediator, we considered the offender's capacity to harm the ingroup.Indeed, a high power offender is most able to harm the group given their capacity to abuse group resources (see Fiske, 1993).However, this is not the case for prestigious offenders who do not necessarily control group resources.Accordingly, we hypothesized that increased harm capacity will mediate the positive effects of high (as opposed to low) power on punishment intent and incapacitative motives (Hypothesis 6a), whereas decreased harm capacity will mediate the negative effect of high (as opposed to low) prestige offender on punishment intent and the positive effect on restorative motives (Hypothesis 6b).The study was preregistered (https://osf.io/68vh4/?view_only=d6863 63ca4 4b43e 1a1e9 92787 3c8ad9c).
As recent research (Fousiani & Van Prooijen 2022a, 2022b) found that power of a suspect of immoraldoing has no influence on retributive motives for punishment, we did not measure retributive motives in Study 3.

Participants
A total of 405 British participants (279 females; M age = 35.18,SD = 11.00)took part in this study.Of the participants, 56.9% were working as employees, 17.4% were managers, 12.6% were unemployed, and 9.5% were students.Moreover, 26.2% had finished high school and 70.2% had higher education.3.6% of the participants did not indicate their demographic characteristics.An a priori power analysis revealed that 400 participants were required to achieve 95% power to detect an interaction with a medium effect size of f = .25.

Experimental design and procedure
As in Study 2, the participants were asked to imagine that they were a member of a famous football team (the 'Blue' team).We manipulated the offender's status in vignettes using a 2 (power: high vs. low) × 2 (prestige: high vs. low) experimental design (cf.Blader & Chen, 2012).The vignettes were largely similar to Study 2 (see Appendix S1).Manipulation checks for power, prestige and group identity were similar to the ones in Study 2. The study was conducted online via Prolific, and lasted 8-10 minutes approximately.Participants got paid £1.

Punishment intent
We measured punishment intent with the same item as in Study 2.

Motives for punishment
We developed a 6-item motives for punishment scale that improved the previously used scale by not containing implicit information about punishment severity.Three items measured incapacitative motives (The main reason to punish Mr X would be to… '…protect the "BLUE" team and its supporters from this kind of behaviours', '…incapacitate Mr X and prevent him from harming the "BLUE" team again', '…protect the football team "BLUE" and the fan club from Mr X's actions'; α=.81).Three items measured restorative motives (The main reason to punish Mr X would be to… '…give Mr X the chance to learn from his mistakes and become a better person', '…give Mr X the opportunity to improve himself', '…help Mr X understand that what he did was wrong and improve himself'; α=.94).

Ingroup threat
We assessed ingroup threat with a 5-item scale, based on ingroup identity threat scale (Duckitt, 2006; e.g.'Mr X seems to ignore moral values that might be important to the supporters of the BLUE 1 = absolutely disagree, 7 = absolutely agree; α = .84).

Perception of harm capacity
We developed a 2-item perception of harm capacity scale (e.g.'Due to his position in the team, Mr X. can really damage the "BLUE" team'; 1 = absolutely disagree, 7 = absolutely agree; α=.77).

Control variables
Severity of harm was assessed with one item: 'The act of Mr X has very serious consequences for the "BLUE" supporters' (1 = absolutely disagree, 7 = absolutely agree).

Results
Correlations, means and standard deviations are in Table 6.
In sum, the power and prestige manipulations worked as intended, although powerful offenders were also perceived as more prestigious and vice versa.Nevertheless, the expected main effects had much stronger effect sizes than the unintended effects, and therefore we regard the power and prestige manipulations as satisfactory for the present purposes.None of the participants failed the group identity check.

Mediating role of ingroup threat and harm capacity
We ran a mediation analysis using Model 4 in Process (Hayes, 2013).Power was the independent variable, ingroup threat and perceived harm capacity were two parallel mediators, incapacitative motives were the dependent variable, and severity of punishment was the control variable (covariate).The overall model was significant (R 2 = .19),F(2,403) = 48.18,p < .001.The main effects of power on ingroup threat and Note: Motives for punishment scales: All ratings were on 7-point scales ranging from 1 = absolutely disagree to 7 = absolutely agree; Punishment intent scale: Ratings were on a 7-point scale 1 = Mr X should not be punished, 7 = Mr X should be punished.
perceived harm capacity both were significant (see Table 8 for the relevant statistics).Participants perceived a high power offender as more threatening to the group, and more capable of harming the group, than a low power offender.Furthermore, both ingroup threat and harm capacity had a significant and positive effect on incapacitative motives.The indirect effects were significant, supporting Hypotheses 4b and 6a.Hypotheses 5b and 6b were not tested given the non-significant main effect of prestige on restorative motives.

Discussion
Study 3 aimed to investigate the differential effects of power and prestige on motives for punishment manipulating power and prestige separately.Supporting Hypothesis 4a, results revealed that people recommend more severe punishment, and endorse stronger incapacitative motives, for high as opposed to low power offenders.The effect of power on incapacitative motives was mediated by ingroup threat and harm capacity, providing evidence for the underlying process and supporting Hypotheses 4b and 6a.Unexpectedly, results revealed no effects of prestige on restorative motives.

GEN ER A L DISCUS SION
Previous research revealed that people recommend more severe punishments for offenders who are at the top of the hierarchy (Fragale et al., 2009;Fousiani & Van Prooijen 2022a, 2022b;Karelaia & Keck, 2013) because such offenders are perceived as more capable of doing harm (Blader & Yap, 2016;Boles et al., 2000;Fragale et al., 2009;Galinsky et al., 2015;Lammers et al., 2015).However, high status offenders can be distinguished as powerful (or dominant) versus prestigious (Blader & Chen, 2012;Blader et al., 2016).Kakkar et al. (2020) showed that people treat powerful as opposed to prestigious offenders more harshly, as the former are seen as more intentional and less moral.What is less known, however, is observers' underlying motives for punishing powerful vs. prestigious offenders, which drive people's justice-related decisions.The present study was designed to investigate the motives for which people punish powerful vs. prestigious offenders.Furthermore, we examined the moderating role of an offender's group identity (ingroup vs. outgroup) in these relationships and the mediating mechanisms underlying these effects.
The results of Studies 1 and 2 suggest that observers have stronger incapacitate motives towards powerful than prestigious offenders, and this effect emerges particularly for ingroup offenders.A similar moderating effect of group identity emerges for the effect of status on punitive intent, expanding prior research (Kakkar et al., 2020).These findings are in line with the BSE theory (Marques & Yzerbyt, 1988;Marques et al., 1988) and suggest that when ingroup offenders have access to resources, they are particularly dangerous for the group and therefore incapacitated.Furthermore, Study 2 (but not Study 1) showed that people display stronger restorative motives towards prestigious as opposed to powerful offenders, suggesting that people are more willing to offer prestigious offenders an opportunity to make up for their wrong-doings and reintegrate to the group.Contrary to our expectations, status did not  interact with group identity in the prediction of restorative motives.Preregistered Study 3 manipulated power and prestige to investigate their separable effects on punishment motives.Results revealed that people display stronger incapacitative motives and punitive intent towards high versus low power offenders.Moreover, ingroup identity threat and perceived harm capacity of the offender mediated this effect.Contrary to our predictions, however, prestige had no effect on restorative motives in Study 3. The effects of status on incapacitative punishment motives, as well as the moderating effect of an offender's group identity, were consistent across Studies 1 and 2. Importantly, Study 3 replicated these findings through a different experimental design.This indicates that the effects of power on incapacitative punishing motives are robust, suggesting that observers experience the acts of powerful offenders as a bigger threat than the acts of prestigious offenders.The effects of prestige on restorative motives for punishment were only found in Study 2 but not in Studies 1 and 3. We therefore regard the main effects of prestige on punitive motives as inconclusive at this point.Also, prestige did not interact with group identity on restorative motives in Studies 1 and 2. One possible explanation for this is that observers might identify strongly with prestigious offenders no matter what their group identity is (ingroup or outgroup).In other words, as soon as one holds a person in high esteem, one might easily connect their own identity to that person and, for instance, focus on overarching categorizations (e.g.member of the same company, even when in different divisions).Another possible explanation is that the links between prestige, group identity and punishment motives depend on group identification (Leach et al., 2008).For instance, observers might be willing to reintegrate a prestigious ingroup offender only when they strongly identify with the group.Future research may further investigate the effect of prestige on motives for punishment.

Practical implications
This research has important practical implications.First, our findings are particularly relevant for relatively hierarchical organizations.HR practitioners should be aware of the differential role that status as power vs. prestige plays in order to achieve fair outcomes when immoralities are enacted.More specifically, our findings hold implications for leader-follower relationships where leaders, depending on whether they are predominantly viewed as powerful or prestigious, might evoke differential reactions among followers when breaking the rules.The current findings suggest that followers are negatively predisposed towards a leader on whom they are dependent for gaining access to resources, and thus react more negatively to them when they commit an offense; but they are more likely to excuse an offense of a leader who is held in high regard and is admired.
Second, our findings inform legal decision makers of potential biases when high status harm-doers are involved.Quite regularly people in high-rank positions are accused of criminal behaviours (e.g.money-laundering, tax evasion, etc.).Information about a suspect's access to resources may shape to what extent they are perceived as threatening to society, influencing punitive decisions about them.The current findings also have broader implications for society, as they suggest that people's punishment motives serve to protect ingroup rather than outgroup members.The present findings may raise awareness among governmental institutions of this bias, facilitating interventions to promote fair treatment of offenders independent from their group identity.Finally, this study underscores the role of underlying motives in punishment.Although behavioural punishment is easily observed and assessed, punitive motives justify and shape such punishment decisions (Carlsmith et al., 2002).Therefore, punitive motives deserve more attention in future research.

Strengths and limitations
Our study has both strengths and limitations.A strength of our study is that in three experiments, we confronted participants with different offenses (sexual harassment and football hooliganism).The consistency in results suggests that some of the findings observed here generalize across different types of offense, and different types of victims.Moreover, we manipulated offender's group identity in different contexts (i.e. a corporate setting in Study 1 and a football team in Studies 2).In both studies, group identity of the offender interacted with status in predicting incapacitative motives revealing the robustness of this effect.The findings also replicated across samples drawn from two different cultures (Greece, a collectivistic country and United Kingdom, an individualistic country, Hofstede et al., 2010).Finally, Study 3 identified two underlying mechanisms by showing that ingroup threat and perceived harm capacity mediate the effects of power on incapacitative motives.
This study includes a number of limitations as well.For instance, we focussed on observer's punitive motives, and did not assess behavioural punishment measures.Moreover, we manipulated the offense through hypothetical vignettes.These considerations suggest that there are ample opportunities to further investigate the effects observed here through field studies and lab experiments, to gain a better understanding of how status influences punitive motives and behaviours towards offenders across social settings.

CONC LUSION
The present research was designed to investigate people's motives for punishing powerful vs. prestigious offenders.The primary conclusion is that people have different motives to punish powerful versus prestigious offenders.People are more likely to indicate instrumental (i.e.utilitarian; see Bentham, 1789) decisionmaking when faced with offenders who have access to resources, by seeking to incapacitate them.This is the case especially when offenders are ingroup rather than outgroup members.The evidence that people hold a preference for restorative punishing practices towards prestigious offenders is suggestive but not conclusive.We conclude that offender's status shapes people's incapacitative motives to punish them.

C ON F L IC T OF I N T ER E S T
Authors declare that they have no conflict of interest.

F
Motives for punishing offenders across experimental conditions (Study 1).All ratings were on 7-point scales ranging from 1 = absolutely disagree to 7 = absolutely agree

Note:
Standard errors in parentheses (bootstrap standard errors for the indirect effect estimate); BCA CI: bias-corrected and accelerated bootstrap confidence interval; paths a and b correspond to the prediction coefficients of the independent variable to the mediator (path a) and of the mediator to the dependent variable (path b); **p < .01;***p < .001.
Pearson correlations coefficients between study variables, means and standard deviations (Study 2) T A B L E 3 Means and standard deviations for the study variables across status conditions (Study 2) T A B L E 5 Pearson correlations coefficients between study variables, means and standard deviations(Study 3) T A B L E 6Note: ***p<.001.
Means and standard deviations for the study variables across status conditions (Study 3) Mediation results with power as predictor, ingroup threat and harm capacity as mediators, and incapacitative motives for punishment as dependent variable(Study 3) T A B L E 8