Not under my watch: Hindering power attributions and leadership affordance to
norm-violators with moral observers and sanctions.
Rima-Maria Rahal University of Amsterdam
Abstract
Traditionally, science and society hold that power corrupts. Reversely, we pursue the notion that corrupt people rise to power. Replicating previous findings in two studies, we show that norm-violators were seen as more powerful than norm-abiders, partially because norm-violators were attributed higher volitional capacity. Expanding previous research by comparing norm-abiding, moderately, and severely norm-violating actions in Study 1, we demonstrate that more severe norm-violations yielded higher power attributions. Additionally, we show that, although powerful people were seen to have higher leadership potential, which predicted leadership affordance, powerful norm-violators were
hypothetically afforded less leadership than less powerful norm-abiders. Contrastingly, through behavioral leadership nominations (Study 2), powerful norm-violators were selected as leaders more, indicating norm-violators’ advantage to rise to the top in practice. Finally, we establish boundary conditions of norm-violators’ rise to the top. Supplementary analyses revealed that observer morality moderated the power-leadership relation: powerful norm-violators were afforded more leadership than norm-abiders by low moral observers, but less leadership by highly moral observers. Lastly,
sanctioning reduced power attributions, particularly to severe norm-violators. In sum, we show that corrupt people rise to the top, partially because they are seen as freer to act at will, and as more potent as leaders. However, norm-violators’ rise to the top could be impeded through moral observers’ rejection and sanctioning.
Not under my watch: Hindering power attributions and leadership affordance to
norm-violators with moral observers and sanctions.
Power doesn’t corrupt people, people corrupt power. - William Gaddis
Leaders do not always use their power for good ends (Gini, 2004; Glad, 2002; Judge, Piccolo, & Kosalka, 2009; Sankowsky, 1995), as showcased by countless historical examples from the Nixon administration’s illegal surveillance to the government of the People’s Republic of Korea’s violations of human rights. But why do some leaders use their power for bad ends? Following a traditional line of reasoning, leaders could abuse their power because they were corrupted by it (Kipnis, 1976). In other words, it has traditionally been argued that power corrupts. This corruption is argued to be reflected, for instance, in findings that powerful people have less compassion (van Kleef et al., 2008) and react less to others’ emotions (Anderson, Keltner, & John, 2003; van Kleef, De Dreu, & Manstead, 2004), view others with prejudice (Guinote, Willis, & Martellotta, 2010) and little esteem (Kipnis, 1972; van Kleef, Homan, Finkenauer, Gündemir, & Stamkou, 2011), and use others to their own advantage (Gruenfeld, Inesi, Magee, & Galinsky, 2008). With a new twist to this traditional view, recent research has rattled the notion that power corrupts by reversely arguing that the corrupt rise to power.
Norm-violators’ and rule-breakers’ rise to the top is illustrated in several prominent examples, such as regarding Robert Ford, who was known for substance abuse and would become mayor of Toronto in 2010, or Joschka Fischer, who was known to have attacked police people and would become German Vice Chancellor in 1998. Indeed, people who break the rules are seen as more powerful: for instance, they were judged to have higher influence over others, to be an authority, and to have legitimacy to influence others and the ability to coerce them into fulfilling job requirements (van Kleef et al., 2011). Moreover, norm-violator were even seen as leaders more (van Kleef et al., 2011). In sum, these findings indicate that corrupt people are seen to be more powerful and are afforded more leadership, giving them an edge to rise to the top. As norm-violators’ previous behavior demonstrates that they are willing to break the rules, they might continue to do so once they have risen to the top. Therefore, norm-violators’ rise to the top could ultimately be another explanation for some leaders’ abuse of power.
Although it has been shown that corrupt norm-violators are seen to be more powerful, to date, the scope of the phenomenon and its limiting factors remain relatively unexplored. To investigate further if and how the corrupt rise to the top, we study leadership affordance based on power attributions to norm-violators, addressing three core questions (Figure 1). First, expanding previous findings that norm-violators are attributed power (van Kleef et al., 2011), we establish that more severe transgressions lend more power to norm-violators. Second, we investigate if people whose power
stems from norm-violations are afforded leadership. Third, we study two mechanisms that could hinder norm-violators’ rise to the top. We aim to show that observer morality moderates the
affordance of leadership to powerful people, to reconcile conflicting views whether powerful people are generally afforded leadership (Brunell et al., 2008; Son Hing, Bobocel, Zanna, & McBride, 2007), or only when they have committed "good" norm-violations (van Kleef, Homan, Finkenauer, Blaker, & Heerdink, 2012). Finally, we show that sanctions hinder the attribution of power to norm-violators.
Norm-Violations Increase Power Attributions
When it comes to assessing a target person’s power, that is their ability to influence others’ behavior (e.g., Dahl, 1957; Winter, 1973) and to "control [. . . ] another’s valued outcomes" (Fiske & Berdahl, 2007, p. 680), observers utilize a plethora of cues and rely on vivid power stereotypes (Keltner, van Kleef, Chen, & Kraus, 2008). Beyond judging power from status symbols (Berger, Ridgeway, Fisek, & Norman, 1998; Goffman, 1951), certain behaviors such as speaking before others (for a review, see Hollander, 1985), displaying certain nonverbal behaviors (for a review, see Hall, Coats, & LeBeau, 2005) like touching others (Henley, 1973), or expressing anger (Tiedens, 2001) can signal power (Ridgeway, Berger, & Smith, 1985). Recent research has demonstrated that
norm-violations such as littering, or drinking someone else’s coffee, similarly lend the impression that norm-violators have power and are more powerful than people who stick to the rules (van Kleef et al., 2011). In sum, norm-violations make actors appear powerful in the eyes of observers.
Although it has been shown that norm-violations yield power attributions, the relative impact of norm-violation severity remains unclear. In other words, do people attribute the same amount of power to actors who commit norm-violations of different severity? For instance, do observers judge the power of someone placing their feet on a chair to be the same as the power of someone tampering with finance reports (van Kleef et al., 2011)? Previous research has shown that norm-violations stimulate power attributions through increasing inferred volitional capacity (van Kleef et al., 2011). Put
differently, norm-violators are seen to be free to do as they please without being affected by conformity pressures (Galinsky, Magee, Gruenfeld, Whitson, & Liljenquist, 2008; Magee, 2009), and are
consequently perceived as more powerful than norm-abiders (van Kleef et al., 2011). More severe norm-violations could hold more potential for inferring volitional capacity, and thus for attributing power. For instance, if norm-violators not only drink someone else’s coffee, but steal the coffee set too, they could be seen as even freer in their actions, and thus as even more powerful. Therefore, we suggest that more severe norm-violations are indeed more potent to demonstrate actors’ freedom to act at will, consequently leading to larger increases in power attributions.
H1 Norm-violators are attributed more power than norm-abiders. More severe norm-violations
Power Attributions Afford Leadership
Beyond testing the influence of norm-violation severity on power, we consider if there are other effects indicating that norm-violators are seen differently than norm-abiders. Specifically, if people afford power to norm-violators, consequently, norm-violators could also gain access to leadership positions. Limited previous evidence indeed suggests that norm-violators are also seen more as leaders than norm-abiders (van Kleef et al., 2011). Irrespective of whether their source of power is good or bad, powerful people could be more likely to be afforded leadership, to be endorsed and chosen as leaders (Platow & van Knippenberg, 2001), and to be ranked atop the hierarchy of a group (Keltner et al., 2008). Leadership categorization theory (Eden & Leviatan, 1975; Lord, Foti, & De Vader, 1984) suggests that observers use target people’s characteristics to infer if these targets fit to the category of leaders. For instance, observers’ implicit leadership theories (Offermann, Kennedy, & Wirtz, 1994; Shondrick, Dinh, & Lord, 2010) could indicate that a decisive, intelligent, industrious person would fit the leader prototype well (Lord et al., 1984), therefore leading to the categorization of the target person as a leader. Similarly, observers’ implicit leadership theories could warrant the inference of a leadership position from actors’ power. Indeed, previous research suggests that target peoples’ strength, the extent to which they are seen as bold and powerful, is an attribute used to infer leadership (Offermann et al., 1994). Further indicating that power could be used to infer leadership, traits typical for leaders are often connected with power (Eastwick et al., 2013). For instance, typical leader traits like extraversion (Ensari, Riggio, Christian, & Carslaw, 2011; Judge, Bono, Ilies, & Gerhardt, 2002), assertiveness, dominance (Lord, de Vader, & Alliger, 1986), and control (Heilman, 2001) certainly seem spiced with power implications. Moreover, implicit leadership theories warranting leadership inference based on power attribution would be supported by empirical evidence suggesting that powerful people indeed advance to leadership positions. For instance, the power dimension of narcissism has been shown to predict leader emergence (Brunell et al., 2008), and dominant people are more likely to become leaders (Son Hing et al., 2007). In sum, power is related to leadership
emergence, and leaders often have traits that are related to power. In turn, a target person’s power therefore could function as a signal of leadership: observers’ implicit leadership theories might stipulate that powerful people are more likely to be leaders. Consequently, observers might infer that powerful people are leaders, and afford more leadership to powerful norm-violators.
H2 Increased power increases the leadership afforded to a target person.
Moreover, beyond suggesting that powerful norm-violators could be afforded more leadership, we consider a potential underlying mechanism explaining this relation. Powerful people could be afforded leadership because they are seen to have higher potential to become effective leaders and effectively carry out leadership tasks. Much like a driver needs a steering wheel to direct a vehicle in
the desired direction, leaders need power (Gini, 2004) to influence (Dahl, 1957; Winter, 1973) their followers and to advance effective group-functioning (van Vugt, 2006). Therefore, people with little power could seem ill suited to lead, whereas powerful people could be seen as more capable to lead. Put differently, inferred potential to lead could be an explanation why power affords leadership. More powerful actors who can effectively influence others could be seen as having a higher potential to lead a group of followers. Consequently, actors with higher leadership potential could be afforded more leadership. Thus, leadership potential could be the underlying mechanism explaining the relation between power attribution and leadership affordance.
H3 The relation between power and leadership is mediated by leadership potential.
Hindering Norm-Violators’ Rise to the Top
If people afford power and, consequently, leadership to norm-violators, corrupt people would have easy access to leadership positions. However, countless "good" leaders showcase that it could be possible to circumvent norm-violators’ rise to the top. Here, we suggest two mechanisms hindering norm-violators’ rise to the top: rejection by moral observers and sanctioning.
Moral observers who value fairness and equality (Rai & Fiske, 2011) could impede powerful norm-violators’ path to leadership. Although we argue that powerful norm-violators are generally afforded more leadership, irrespective of whether their power stems from "good" or "bad" actions, conflicting evidence indicates that leadership is only afforded to ”good” norm-violators. Van Kleef et al. (2012) found that leadership was only afforded to norm-violators committing prosocial
norm-violations. Deviating from our argument above, norm-violators who drank someone else’s coffee were not generally afforded more leadership than norm-abiders. Rather, norm-violators were only afforded more leadership when they acted prosocially and shared the stolen coffee with another person (van Kleef et al., 2012). We propose to resolve the inconsistency between the argumentation that more powerful people are generally afforded more leadership and the finding that only "good", prosocial norm-violators were afforded leadership by considering if observer characteristics moderate the relation of power and leadership. As power (Fiske & Berdahl, 2007) and leadership (Grint, 2005) are socially situated and depend on observer perception (DeRue & Ashford, 2010), observer characteristics could influence judgments about norm-violators. A key observer quality changing judgments of norm violators’ leadership qualities may be morality, the degree to which people embrace moral principles, and to which moral principles are accessible (Lapsley & Narvaez, 2004), and part of one’s self-identity (Aquino & Reed, 2002). As previous evidence suggests that only "good", prosocial norm-violators are afforded leadership (van Kleef et al., 2012), morality, the observers’ general tendency to perceive "good" and "bad" and to advocate the "good", could determine whether observers afford leadership to powerful norm-violators who did something "bad". Specifically, highly moral observers could be more
sensitive to norm-violations as instances of moral transgressions, and less tolerant of powerful
norm-violators as leaders because of their moral transgressions. Indeed, people experiencing moralistic anger at a norm-violation aim to vent their anger and to express their indignation (Miller, 2001). For instance, people experiencing moral outrage when observing moral transgressions have the heightened urge to punish offenders (Miller, 2001) and unfair (Singer & Steinbeis, 2009), selfish freeriders (Fehr & Gächter, 2000, 2002). Beyond expressing observers’ indignation, these punitive behaviors are used to secure and promote equality (Fowler, Johnson, & Smirnov, 2005) as a moral principle (Rai & Fiske, 2011). In sum, to ensure equality and to promote equality, or to express their indignation at
norm-violations, highly moral observers might oppose leadership affordance to powerful individuals if their power is based on norm-violations, and thus on moral transgressions. Therefore, more moral observers couls be unlikely to afford leadership because of power achieved from norm-violations. However, less moral observers could be more likely to disregard the source of power, and to thus afford leadership based on power from norm-violations.
H4 The relation between power attribution and leadership affordance is moderated by observer
morality, such that the positive relationship between power attribution and leadership affordance is weakened for relatively highly moral individuals (compared to relatively less moral individuals).
In sum, for the relation of power attributions to leadership affordance, we therefore assess a mediated moderation model, considering leadership potential as a mediator and morality as a moderator.
Moreover, sanctioning may be an effective tool to reduce power attributions to norm-violators. Sanctions and punishment both deter norm-violators from future transgressions (Boyd, Gintis, Bowles, & Richerson, 2003; Sigmund, 2007), and change how others see norm-violators (Uggen, Manza, & Behrens, 2004). Indeed, sanctions for norm-violations may change norm-violators’ image to be free to do as they please, effectively reducing their ascribed volitional capacity. Fines or other sanctions for committing norm-violations create conformity pressures by restricting the actors’ freedom to act at will, and coerce them to obey to certain norms (e.g., Andreoni, Harbaugh, & Vesterlund, 2003; Fehr & Fischbacher, 2004). Such conformity pressures diminish the actors’ volitional capacity (Galinsky et al., 2008; Magee, 2009). Because volitional capacity as the prerequisite for norm-violators’ power (van Kleef et al., 2011) is reduced, sanctioned actors could be attributed less power. Moreover, the imposition of sanctions directly demonstrates that a norm-violator’s power is inferior to the power of the person executing the sanction. Indeed, imposing sanctions could be understood as a source of coercive power, indicating the enforcer’s ability to punish others (French & Raven, 1959). When norm-violators heed the imposition of sanctions, they are subordinated to the coercive power of the enforcer. Thus, norm-violators could be seen as inferior to the power of enforcers, leading to reduced power attributions to norm-violators. In sum, because sanctions reduce norm-violators’ ascribed
volitional capacity, and subordinate them to the enforcer’s power, sanctioning could effectively reduce power attributions to norm-violators.
H5 Sanctioning moderates the relationship between norm-violation and power attributions, such
that sanctioned norm violators are attributed less power than non-sanctioned norm violators.
Exploratory Questions
Although we have some indications why norm-violations yield power attributions, it remains important to understand underlying causes of the phenomenon in more detail. Van Kleef et al. (2011) explained the positive influence of norm-violations on power attributions through a mediation of volitional capacity, norm-violators’ attributed freedom to act at will. We further investigate if the relation between norm-violations and power is also driven by responsibility as an additional mediator. Power is sometimes seen to have two facets (Sassenberg, Ellemers, & Scheepers, 2012): the opportunity to achieve one’s own goals (Fiske & Berdahl, 2007), and the responsibility for the implications of one’s actions for other (Torelli & Shavitt, 2010). Previous research assessing the attribution of power to norm-violators focused on power inferred from the freedom of choosing one’s actions and thus achieving one’s goals (van Kleef et al., 2011), but disregarded the actors’ responsibility to consider their commitments to others. In the light of theorizing that power is indeed two-faceted, we argue that the amount of power attributed to actors will not be only determined by their freedom to do as they please, but also by their perceived responsibility for possible consequences of their actions. Therefore, we explore if responsibility also mediates the relation between norm-violation and power attribution.
EH1 The positive relation between norm-violations and leadership is mediated both by volitional
capacity and by responsibility.
Finally, we investigate the role of several constructs similar to morality (Haidt, 2008) as potential moderators on the power-leadership relation. Specifically, we include prosocial value
orientation (De Dreu & van Lange, 1995; Griesinger & Livingston, 1973), right-wing authoritarianism (Duckitt, Bizumic, Krauss, & Heled, 2010), and agreeableness (Costa & McCrae, 1992).
EH2 The relation between power attribution and leadership affordance is moderated by prosocial
value orientation, authoritarianism and agreeableness, such that the positive relationship between power attribution and leadership affordance is weakened for relatively high values of prosocial value orientation, authoritarianism and agreeableness (compared to relatively low values).
Present Research
In the present research, we investigated leadership affordance based on power from
In Study 1, we compared power attributions and hypothetical leadership affordance to norm-abiders and moderate or severe norm-violators, depending on sanctions and chronic observer morality. In Study 2, we experimentally manipulated observer morality and additionally assessed leadership affordance through leader nominations in fictitious teams.
Pretest
To ensure the viability of the manipulations used in the subsequent studies, we conducted a pretest assessing if the stimulus material was understood correctly on a factual level, if the intended manipulations would indeed lead to predicted differences in rated severity of norm-violating behavior and sanctioning, and if the manipulations would be orthogonal. Moreover, we developed a German version of the adjectives used in the Moral Identity Scale to measure self-importance of moral identity, which was also reused to manipulate morality in Study 2.
Method
Participants. Ninety-seven German-speaking participants (61 female,
Mage = 24.26, SDage= 6.14) who were recruited via German online participant pools took part in this
online study ostensibly focusing on impression formation given limited information availability. They entered a raffle for aAC15 gift voucher as compensation.
Design and Stimulus Material. The study followed a 3 (norm-violating behavior: norm-abiding vs. moderately norm-violating vs. severely norm-violating) × 2 (sanctioning: non-sanctioning vs. sanctioning) between-subjects design.
To manipulate norm-violating behavior and sanctioning, participants read a short story about a traveler entering a train station, whose norm-related behavior was varied. In the norm-abiding
condition, he was portrayed to buy a train ticket worthAC5. However, in both norm-violating
conditions, he only bought coffee and some snacks, which also costAC5. Then he was said to catch his train, where a conductor came to check his ticket. In the norm-abiding condition, the traveler could not find his ticket and explained to the conductor that he bought a ticket before boarding the train. In the moderately norm-violating condition, the traveler could not show a valid ticket and told the conductor that he did not buy a ticket. In the severely norm-violating condition, the traveler could not show a valid ticket. Yet, he told the conductor that his ticket had already been checked.
To manipulate the sanctioning of the norm-violating behavior, in the non-sanctioning condition, the conductor made an exception: the traveler was not fined. However, in the sanctioning condition, the conductor did not make an exception: the traveler was fined.
To check the success of both manipulations, participants answered eight factual questions ("Traveling without a ticket was allowed/prohibited", "The traveler was fined/not fined.", "The traveler bought/did not buy a ticket", and "The traveler was honest/lied to the conductor."), and four questions
where they indicated to what extent the traveler broke or adhered to existing rules, and was punished or not punished for his behavior (1 = not at all, 7 = extremely).
To develop a German version of the adjectives used in the Moral Identity Scale, we assembled 24 German adjectives closely related to the 9 English adjectives used in the Moral Identity Scale instructions and asked participants to indicate on a 7-step scale (1 = not at all, 7 = extremely) how well these adjectives described a moral person.
Procedure. Participants were instructed that they would read a short story, answer some questions about the main actor of the story, and finally answer some general questions and questions about themselves. Then, participants were randomly assigned to read one of the six stories to manipulate norm-violating behavior and sanctioning, and subsequently answered the manipulation check questions. Finally, participants completed the assessment of how descriptive a presented list of adjectives was for a moral person. Following the collection of socio-demographic information, participants were thanked and debriefed.
Results
Manipulation Checks. Regarding the factual understanding of the short stories, the
majority of participants (86.7%) answered all factual manipulation check question correctly, indicating that the materials were well understood. Individual manipulation check questions were predominantly answered correctly (91.6% to 97.6%, respectively). Correct answers did not differ between the
norm-violating (F [2, 77] = 0.52, p = 0.60, η2
p= 0.01) and sanctioning conditions (F [1, 77] = 1.06,
p = 0.31, η2
p= 0.01). Finally, the manipulations did not influence the judgment whether it is generally
allowed (Fnorm-violating[2, 78] = 0.60, p = 0.55 ηp2= 0.02, Fsanctioning[1, 78] = 1.13, p = 0.29, ηp2= 0.01)
or prohibited (Fnorm-violating[2, 77] = 0.74, p = 0.48, ηp2= 0.02, Fsanctioning[1, 77] = 0.70, p = 0.41,
η2
p = 0.01) to travel on a train without a ticket, indicating that traveling without a ticket was
perceived as a norm-violation across groups.
With relation to the effect of the manipulations on the impression of the target person’s adherence to rules and punishment, we conducted a multivariate analysis of variance. Results indicate that the norm-violating manipulation led to the impression of the target person as breaking the rules more (F [2, 78] = 19.411, p < 0.001, η2
p= 0.33) and reduced the impression of adhering to the rules
(F [2, 78] = 59.85, p < 0.001, η2
p= 0.61) in the norm-violating conditions, whereas the sanctioning
manipulation had no effect on these dependent variables (Fbreaking rules[1, 78] = 1.90, p = 0.30,
η2
p = 0.01; Fadhering to rules[1, 78] = 0.28, p = 0.60, η2p= 0.004). We additionally assessed repeated
contrasts, indicating that norm-abiders were seen as breaking the rules more (t[78] = 5.27, p < 0.001) and adhering to the rules less (t[78] = 8.09, p < 0.001) than moderate norm-violators. However, the difference between the perception of moderate norm-violators compared to severe norm-violators was
not significant for rule-breaking (t[78] = 0.01, p = 1.00) and only marginally significant for adherence to the rules (t[78] = 1.93, p = 0.06).
Moreover, the target person was seen as punished more in the sanctioning condition (F [1, 78] = 85.23, p < 0.001, η2
p= 0.52), and was seen to get away without punishment more in the
non-sanctioning condition (F [1, 78] = 174.65, p < 0.001, η2
p = 0.69), whereas the norm-violation
manipulation had no effect on these dependent variables (Fpunished[2, 78] = 2.27, p = 0.11, ηp2= 0.06,
Fnot punished[2, 78] = 0.36, p = 0.70, η2p= 0.01). Finally, we found no significant interactions between
the norm-violating and the sanctioning manipulations on the perceived adherence to rules
(Fbreaking rules[2, 78] = 0.64, p = 0, 53, η2p= 0.02, Fadhering to rules[2, 78] = 0.20, p = 0.82, ηp2= 0.01) or
the perceived punishment (Fpunished[2, 78] = 1.82, p = 0.17, ηp2= 0.05, Fnot punished[2, 78] = 1.99,
p = 0.14; η2
p= 0.06).
Morality Judgments. To find a German adjectives related to morality corresponding to the adjectvies used in the Moral Identity Scale, we assessed the average ratings of the adjectives as descriptive of a moral person, and determined the nine adjectives with the highest mean ratings (Table 1) to be retained in subsequent studies.
Discussion
In the pretest, we tested new stimulus material to manipulate norm-violating behavior and sanctioning, and developed German stimulus material to be used in the intstructions of the Moral Identity Scale and for the morality manipulation.
First, we assessed if the developed manipulations for norm-violating behavior and sanctioning would be understood well, would lead to differences in rated severity of the norm-violating behavior and of punishment, and if the manipulations would be orthogonal. Results indicate that the general understanding of the stimulus material was good. Moreover, the manipulations overall produced the desired results: the rated norm-violation severity and punishment differed as expected between the conditions, and norm-abiders were indeed seen to break the rules less and adhere to the rules more than moderate norm-abiders. Moreover, the manipulations were found to be orthogonal.
Yet, we did not find a reliable difference in rated norm-violation severity between the moderate and severe norm-violation conditions. To achieve the desired difference in norm-violation severity between these conditions, we adapted the manipulation for Study 1 by emphasizing if the actor committed only the norm-violation of not buying a ticket, or would additionally commit a rule violation by lying to the train conductor. In the revised moderate norm-violation condition, the traveler is described as candidly admitting to the conductor that he had not bought a train ticket, whereas the revised severe norm-violation condition states that the traveler lies and deceitfully tells
the conductor that his ticket had already been checked (Appendix A). We otherwise retained the manipulations for the subsequent studies.
Second, we aimed to develop a German set of adjectives to be used in the instructions to the Moral Identity Scale, and as morality cues in Study 2. Based on the results of this pretest, we selected the nine adjectives rated as most descriptive of a moral person and retained them for the subsequent studies.
In sum, the pretest provided stimulus material which we used further in the subsequent studies.
Study 1
In Study 1, we utilized the previously developed manipulations to study the relation between norm-violations and power attributions with sanctioning as a moderator. Moreover, we examined whether the proposed positive relationship between power attributions and hypothetical leadership affordance was mediated by leadership potential and moderated by observer morality. Finally, to understand the underlying mechanisms driving the relation of norm-violations and power attribution, we explored if the relation of norm-violation and power attribution was mediated by volitional capacity and responsibility.
Method
Participants. In this online study, 462 German-speaking participants (240 female,
Mage = 23.67, SDage= 6.51) were recruited from German online participant pools to take part and
were offered to enter a raffle for fourAC15 gift vouchers as compensation.
Design and Stimulus Material. Study 1 followed a 3 (norm-violating behavior:
norm-abiding vs. moderately norm-violating vs. severely norm-violating) × 2 (sanction: no sanction vs. sanction) between-subjects design, where the attribution of power and leadership affordance were the dependent variables. Morality was assessed as a continuous moderator of the power-leadership link. Several other variables were included for an exploratory analysis of their potential moderating role. Moreover, we assessed the mediating role of leadership potential. Sanctioning was assessed as a dichotomous moderator of the relation of norm-violation and power. In addition, we aimed to replicate the mediating role of volitional capacity (van Kleef et al., 2011), and further studied the mediating role of responsibility regarding the relation of norm-violation and power.
To manipulate norm-violation and sanctioning, we use the revised manipulations derived from the pretest (Appendix A). For manipulation checks, we retain the eight factual questions and four assessments of norm-violation severity and sanctioning developed in the pretest.
Procedure. Participants first self-assessed morality on two scales. On the Portrait Values Questionnaire (Schmidt, Bamberg, Davidov, Herrmann, & Schwarz, 2007; Schwarz et al., 2001, retest-reliability 0.66 to 0.88), participants indicated how similar a fictitious person upholding or
violating moral values was to them (1 = not like me at all, to 7 = very much like me). On an adapted version of the Self-Importance of Moral Identity Scale (Aquino & Reed, 2002, retest-reliability 0.49 to 0.70), where we used the nine German adjectives selected in the pretest as most descriptive of moral people in the instructions, participants indicated their agreement with statements describing the symbolization and internalization of moral values (1 = strongly disagree, to 7 = strongly agree). Following this self-assessment, participants completed an unrelated filler task where they sorted German cities by population size to avoid carry-over-effects on the subsequent manipulations of norm-violating behavior and sanctioning.
After manipulating norm-violation and sanctioning by instructing participants to read one out of six short stories, the assessment of the target person was collected. Regarding power attributions, participants indicated on a first-person version (van Kleef et al., 2011, 0.73 ≤ α ≤ 0.81) of the
eight-item Generalized Sense of Power Scale (Anderson & Galinsky, 2006, α = 0.88) how powerful they thought the target person was (”I think this person has a great deal of power.”, 1 = strongly disagree, to 7 = strongly agree), and if they believed the target person could influence others (1 = definitely not, to 7 = definitely) on eight items of an adapted version (van Kleef et al., 2011) of the Coercive and Legitimate Power scales (Hinkin & Schriesheim, 1989, 0.80 ≤ α ≤ 0.86).
Hypothetical leadership affordance was operationalized with leader endorsement, leader acceptance and leader reverence. For leader endorsement, participants indicated with four items if they thought the actor was a suitable leader ("This person is suitable as a leader.", Halevy, Berson, & Galinsky, 2011) and if they would select this person as a leader ("I would vote for this person as a leader.", Platow & van Knippenberg, 2001) on a seven-step scale (1 = strongly disagree, to 7 = strongly agree). Then participants indicated if they would accept ("I would like this person as my boss.”, van Kleef et al., 2012, adapted, α = 0.75) and respect ("I would hold this person in high respect as my boss.”, Conger, Kanungo, & Menon, 2000, four items, adapted) the target person their leader (1 = definitely not, to 7 = definitely).
Next, participants assessed the actor’s volitional capacity by indicating on six items how much volitional capacity they ascribed to the target person (To what extent does this person feel free to do what s/he wants?, 1 = not very much, to 7 = very much, Magee, 2009). To assess the actor’s responsibility, they indicated on five items adapted from the Discursive Responsible Leadership scale (Voegtlin, 2011) if the target person would consider the consequences of his decisions (1 = not at all, to 7 = frequently, if not always). Regarding leadership potential, participants assessed on four items if the target person had the potential to effectively carry out leadership tasks ("This person has the potential to become an effective leader.", 1 = not at all, to 7 = very much, Mueller, Goncalo, & Kamdar, 2011, α = 0.92).
Following the assessment of the actor, participants completed the same manipulation check questions as used in the pretest. Finally, we collected data on participants’ social motivation, authoritarianism and agreeableness. To assess social motivation, participants played the decomposed games measure (De Dreu & van Lange, 1995; Griesinger & Livingston, 1973; van Lange, 1999), where they made nine point-allocation choices between themselves and a hypothetical other person.
Participants were regarded as prosocials if they make at least six prosocial choices, or as proselfs if they made at least six individualistic choices. Then, participants self-assessed authoritarian attitudes on the six-item short-form authoritarianism sub-scale of the ACT questionnaire ("The facts on crime and the recent public disorders show we have to crack down harder on troublemakers if we are going preserve law and order.", 1 = strongly disagree, to 7 = strongly agree, Duckitt et al., 2010, 0.80 ≤ α ≤ 0.87). On the 12-item agreeableness sub-scale of the Revised NEO Personality Inventory-Short Form (Costa & McCrae, 1992, 0.75 ≤ α ≤ 0.82), participants self-assessed agreeableness ("I generally try to be thoughtful and considerate.", 1 = strongly disagree, to 7 = strongly agree).
Finally, we collected participants’ hypotheses regarding the true purpose of the study and the relation of the study’s tasks. Following the collection socio-demographic information, participants were thanked and debriefed.
Results
Data Preparation. We summarized the data for concepts measured with several items in compound scores by calculating the mean of the respective items (0.77 ≤ α ≤ 0.94, Table 2).
Moreover, we dummy-coded norm-violation into two sets, allowing for comparisons to moderate and to severe norm-violations, respectively. Six participants were excluded from further analyses as outliers, because their scores on power attribution and leadership affordance differed more than three standard deviations from the mean. Moreover, we excluded participants from further analyses who failed to answer the factual manipulation check questions correctly, and retained data only from participants who completed the survey (65.8%, nfinal= 304) Participant dropout did not depend on the allocation
to the manipulated conditions (Fnorm-violation[2, 455] = 0.14, p = 0.87, η2p= 0.001,
Fsanctioning[1, 455] = 1.23, p = 0.27, η2p= 0.003, Finteraction[2, 455] = 1.49, p = 0.23, η2p= 0.01). Manipulation Checks. Regarding the understanding of the stimulus material, most participants (91%) answered all factual manipulation check questions correctly, and predominantly correct answers were given for each question, respectively (96.2% to 98.2%). Correct answers did not differ between the norm-violating (F [2, 334] = 1.90, p = 0.15, η2p= 0.01) and sanctioning conditions
(F [1, 334] = 1.58, p = 0.21, η2
p= 0.005) and we found no interaction effect (F [2, 334] = 1.39, p = 0.25,
η2
p = 0.01). Finally, the manipulations did not influence the judgment whether it is generally allowed
Finteraction[2, 298] = 0.75, p = 0.47, η2p= 0.01) or prohibited (Fnorm-violating[2, 298] = 1.96, p = 0.14,
η2
p = 0.01, Fsanctioning[1, 298] = 0.10, p = 0.75, η2p< 0.001, Finteraction[2, 298] = 0.12, p = 0.89,
ηp2= 0.001) to travel on a train without a ticket.
Regarding the success of the manipulations and their orthogonality, we conducted a multivariate analysis of variance to examine the effect of the manipulations on the impression of the target person’s adherence to rules and punishment. Results indicate that in the norm-violating conditions, the target person was seen as breaking the rules more (F [2, 298] = 266.03, p < 0.001, ηp2= 0.64) and as adhering
to the rules less (F [2, 298] = 510.11, p < 0.001, η2
p = 0.77) than norm-abiders. Repeated contrasts
indicated that moderate norm-violators were seen as breaking the rules more (t[298] = 16.99, p < 0.001) and adhering to the rules less (t[298] = 23.22, p < 0.001) than norm-abiders. Severe norm-violators were seen to break rules more (t[298] = 5.17, p < 0.001) and adhere to rules less (t[298] = 7.30, p < 0.001) than moderate norm-violators. However, the assessment of the target person’s adherence to rules was also unexpectedly influenced by the sanctioning manipulation: unsanctioned actors were also seen as breaking the rules more (F [1, 298] = 11.97, p = 0.001, η2
p= 0.04)
and adhering to the rules less (F [1, 298] = 35.60, p < 0.001, η2
p = 0.11) than sanctioned actors.
Regarding the sanctioning manipulation, sanctioned actors were seen as punished more (F [1, 298] = 2267.42, p < 0.001, η2
p = 0.88), and unsanctioned actors were seen to get away without
punishment more (F [1, 298] = 1779.91, p < 0.001, η2
p= 0.86). The norm-violation manipulation did
not influence the impression of the target as punished (Fpunished[2, 298] = 0.33, p = 0.72, η2p= 0.002).
However, norm-violators were unexpectedly seen as being more unpunished (Fnot punished[2, 298] = 3.30, p = 0.04, η2p= 0.02) than norm-abiders.
Finally, we found interactions between the norm-violating and the sanctioning manipulations. Whereas sanctioned norm-abiders and moderate norm-violation were seen to break the rules less than unsanctioned actors, sanctioned severe norm-violators were seen to break the rules more than
unsanctioned actors (Fbreaking[2, 298] = 10.19, p < 0.001, η2p= 0.06). Regarding the perceived
adherence to rules, the difference between sanctioned and unsanctioned actors decreased with more severe norm-violations (Fadhering to rules[2, 298] = 3.65, p = 0.03, ηp2= 0.02). No interactions were found
regarding perceived punishment (Fpunished[2, 298] = 0.11, p = 0.89, ηp2= 0.001,
Fnot punished[2, 298] = 0.32, p = 0.73, ηp2= 0.002.).
Confirmatory Analyses. First, to predict power attribution from norm-violation (H1), considering sanctioning as a moderator (H5), we conducted a hierarchical regression analysis (Table 3). In the first step, we entered participant sex as a control variable, yielding a marginally significant effect (b = −0.22, SE = 0.12, p = 0.06) indicating that women attributed more power than men. In the second step, we entered norm-violation and sanctioning. Supporting hypothesis 1, results indicate that norm-abiders were attributed less power than moderate norm-violators (b = −0.61, SE = 0.11,
p < 0.001) and than severe norm-violators (b = −0.97, SE = 0.11, p < 0.001), and that severe norm-violators were attributed more power than moderate norm-violators (b = 0.36, SE = 0.11, p = 0.001). Moreover, sanctioned actors were seen as less powerful than unsanctioned actors (b = −0.53, SE = 0.09, p < 0.001). In the third step, we entered the interaction terms of
norm-violation and sanctioning. Partially supporting hypothesis 5, this interaction was significant for the comparison between norm-abiders and severe norm-violators (b = 0.48, SE = 0.22, p = 0.03): although sanctioned actors were seen as less powerful in both conditions, the power difference was larger for severe norm-violators than for norm-abiders (Figure 2). Unexpectedly, we found no significant interaction when comparing norm-abiders with moderate norm-violators (b = 0.29, SE = 0.22, p = 0.18), or moderate with severe norm-violators (b = 0.19, SE = 0.22, p = 0.39).
Secondly, to predict hypothetical leadership affordance with power attributions (H2), we conducted a hierarchical regression analysis (Table 4). In a first step, we entered participant sex as a control variable, yielding no effect (b = −0.05, SE = 0.17, p = 0.79). In the second step, we entered power attribution. Contrary to hypothesis 2, results indicate that more powerful targets were generally afforded less hypothetical leadership (b = −0.27, SE = 0.08, p = 0.001).
Thirdly, to assess our mediated moderation hypothesis1, predicting hypothetical leadership affordance with power attributions while leadership potential is considered as a mediator (H4) and morality as a moderator (H5), controlling for participant sex, we conducted a regression-based path analysis (Figure 3) using the PROCESS macro (Hayes, 2013; Preacher & Hayes, 2008; Preacher, Rucker, & Hayes, 2007, Model 5). Participant sex had no effect on leadership potential (b = −0.14, SE = 0.16, p = 0.38) or hypothetical leadership affordance (b = −0.02, SE = 0.13, p = 0.89).
Regarding hypothesis 4, results (Table 5) indicate that more powerful people were seen to have higher leadership potential (a = 0.18, SE = 0.08, p = 0.02), and that people with higher leadership potential were afforded more hypothetical leadership (b = 0.66, SE = 0.05, p < 0.001). The negative relation between power attribution and leadership affordance remained significant (c1= −0.34, SE = 0.08,
p < 0.001). A bias-corrected bootstrap 95% CI showed that the indirect effect through leadership potential was significant (a × b = 0.12, SE = 0.06, 95% CI: [0.02, 0.23]), indicating a partial mediation by leadership potential in partial support of hypothesis 4. However, regarding hypothesis 5, there was no main effect of morality on hypothetical leadership affordance (c2= 0.01, SE = 0.07, p = 0.94) and
unexpectedly no significant interaction effect of morality and power attribution (c3= −0.07,
SE = 0.07, p = 0.33). The conditional direct effect remained significant at three levels of morality. A direct assessment of the effect of norm-violations on leadership affordance, mediated by power attribution and moderated by sanctioning and observer morality, respectively, can be found in
Appendix C.
1Results for other potential moderators in this mediated moderation analysis (EH2) and an alternative analysis
Exploratory Analysis. Regarding the relation between norm-violations and power, to explore the potential mediating role of perceived responsibility in addition to volitional capacity (EH1), we further conduct a regression-based path analysis (Figure 4) with PROCESS (Hayes, 2013, Model 6). Results (Table 6) regarding volitional capacity indicate that norm-abiders were seen to have less volitional capacity than moderate (a1= −1.30, SE = 0.15, p < 0.001) and severe norm-violators
(a1= −1.83, SE = 0.15, p < 0.001). Moreover, moderate norm-violators were seen to have less
volitional capacity than severe norm-violators (a1= −0.533, SE = 0.15, p < 0.001). In turn, higher
volitional capacity lead to higher power attributions (b1= 0.18, SE = 0.04, p < 0.001). Regarding
responsibility, norm-abiders were seen to be more responsible than moderate (a2= 0.83, SE = 0.18,
p < 0.001) and severe norm-violators (a2= 1.41, SE = 0.20, p < 0.001). Moreover, moderate
norm-violators were seen to be more responsible than severe norm-violators (a2= 0.58, SE = 0.17,
p < 0.001). However, we found no main effect of responsibility on power attribution (b2= −0.05,
SE = 0.04, p = 0.24). Overall, the positive relation between norm-violations and power attribution remained significant (see Table 6, Model 3). A bias-corrected bootstrap 95% CI showed that the total indirect effects through volitional capacity and responsibility were significant, but partial indirect effects reached significance only through volitional capacity (Table 7). Implying a partial mediation by volitional capacity, but not by responsibility, these results partially supported exploratory hypothesis 1.
Discussion
In Study 1, we replicated and extended previous findings indicating that norm-violators are attributed more power than norm-abiders by testing the role of norm-violation severity. Moreover, we investigated whether norm-violators’ increased power would yield an advantage in hypothetical leadership affordance because of increased inference of leadership potential. Finally, we considered the moderating roles of observer morality and sanctioning. Results partially confirmed our expectations.
Although our norm-violation manipulation affected perceived norm-violation severity and our sanctioning manipulation affected perceived sanctioning as expected, we found unexpected influences of the norm-violation manipulation on perceived punishment and of the sanctioning manipulation on perceived norm-violation severity, as well as an interaction effect. However, the pretest of the manipulations did not show these unexpected influences, and rather provided good indications that the manipulations would function as expected. Possibly, the comparatively large sample size in Study 1 boosted these previously nonsignificant effects to significance. Finally, these unexpected influences had small effect sizes and can therefore be considered to have only little practical implications for our interpretations.
Supporting hypothesis 1, we replicated previous findings that norm-violators are attributed more power than norm-abiders. Further supporting hypothesis 1, we found that increasing
norm-violation severity led to increased power-attribution. Regarding the underlying mediating mechanism of the relation between norm-violations and power attribution (EH 1), we partially replicated previous findings by showing a partial mediation of volitional capacity. Contrary to our speculations, we did not find responsibility to simultaneously mediate the relation between norm-violations and power attribution.
Contrary to hypothesis 2, we found that more powerful people were generally afforded less hypothetical leadership. Nevertheless, we found an expected partial mediation of the power-leadership relation by leadership potential (H3), indicating that more powerful actors were attributed higher leadership potential, which in turn predicted higher leadership affordance. Moreover, we aimed to provide reconciliatory evidence for the previously inconsistent relation between power attribution and hypothetical leadership affordance by showing that the relation between power and hypothetical leadership depends on observer morality (H4). However, we did not find that observer morality moderated the power-leadership relation.
Finally, we aimed to show that sanctioning would moderate the relation of norm-violations and power attribution (H5). We found that sanctioned actors were seen as less powerful than unsanctioned actors, and that the power difference between sanctioned and unsanctioned actors was larger when a severe norm-violation was committed than when the actor abode by the rules. However, as sanctioning did not moderate the relation of norm-violations and power when comparing norm-abiders with moderate norm-violators, or moderate with severe norm-violators, our expectations were only partially supported.
In sum, results indicating that norm-violators were afforded more power and that sanctioning moderated this relation were largely in accordance with our expectations, whereas evidence indicating that more powerful actors were afforded less leadership ran counter to our expectations. Results regarding the expected moderation of the power-leadership relation by morality were inconclusive. To replicate and substantiate these findings, we conducted a second, consecutive study. As initial results in Study 1 indicate that more severe norm-violations increased power attribution, we condensed the experimental design of Study 2 to compare only two levels of norm-violating behavior: norm-abidance and severe norm-violations violations. To further investigate the power-leadership relation, we studied if experimentally manipulated rather than chronic morality moderated leadership affordance to norm-violators. Finally, to reconcile our expectation that powerful people would be afforded more leadership with the contradicting finding that, hypothetically, participants afforded less leadership to powerful actors, we additionally assessed leadership nominations in fictitious teams reflecting
Study 2
In Study 2, we aimed to replicate and expand the evidence collected in Study 1. Regarding the role of observer characteristics in attributions of power after rule breaking, we experimentally induced varying levels of observer morality. In addition, we assessed whether powerful actors were more likely to be chosen as leaders for fictitious teams, although they were hypothetically afforded less leadership.
Method
Participants. In this online study, 530 German-speaking participants (212 female, Mage = 23.26, SDage= 5.69) recruited from German online participant pools took part, and were
offered to enter a raffle for fourAC15 gift vouchers as compensation
Design and Stimulus Material. In this 2 (morality: high vs. control) × 2 (norm-violating behavior: norm-violating vs. norm-abiding) × 2 (sanctioning: non-sanctioning vs. sanctioning) between-subjects study, power attribution, and hypothetical leadership affordance and leadership nominations were the dependent variables. We again studied the moderating role of sanctioning in the expected relation between norm-violations and power affordance, as well as the role of volitional capacity and responsibility as mediators to this relation. Moreover, we assessed the role of leadership potential in the expected relation between power affordance and leadership nominations, and tested if observer morality moderated this relation. We further included agreeableness as an exploratory moderator to the power-leadership relation.
We manipulated morality with a procedure adapted from the "Handwriting Task" (Aquino, Freeman, Reed, Felps, & Lim, 2009; Aquino, Reed, Thau, & Freeman, 2007; Reed, Aquino, & Levy, 2007). Participants first sorted nine positively valanced adjectives alphabetically and by word length. In the moral prime condition, we used adjectives related to morality (e.g., fair, honest, Table 1) corresponding to the Moral Identity Scale (Aquino & Reed, 2002), which were selected in the pretest. In the control condition, the positively valenced adjectives were unconnected to morality (e.g., carefree, compatible2, Aquino et al., 2007). After the sorting task, participants were asked to
contemplate the nine adjectives, and wrote a short story about themselves, using each word at least once. To check the morality manipulation, participants indicated how much they see themselves as a student, a member of an organization, a moral person and safety conscious (1 = to some extent, to 7 = to a great extent, Aquino et al., 2007), and completed the Moral Identity Scale (Aquino & Reed, 2002). For the norm-violation manipulation (Appendix B), we reused and shortened two texts from the norm-violating behavior manipulations of Study 1 where participants read about a traveler buying a train ticket and honestly admitting to a train conductor that he could not find it when he was being
2Positively valenced adjectives translated into German and used in the control condition were sorglos, verträglich,
checked (norm-abidance), or not buying a train ticket and deceitfully telling a train conductor that his ticket had already been checked (severe norm-violation). Sanctioning (Appendix B) was manipulated as in Study 1. However, we embedded these texts into a collection of three other short stories about people carrying out norm-abiding, unsanctioned everyday behaviors: going to the movies, a conference or to the museum. The manipulation checks where the same as used in Study 1.
Procedure. First, we manipulated morality with the sorting and writing task, and norm-violating behavior as well as sanctioning with four short stories, followed by manipulation checks. Participants completed the power attribution and hypothetical leader affordance items from Study 1, assessing two target persons: the traveler, and a norm-abiding, unsanctioned movie-goer. Then, participants assessed both target persons’ leadership potential, volitional capacity and the traveler’s responsibility. Moreover, participants were asked to assess the hierarchy among the four people they had read about through pairwise leadership nominations (Emery, Daniloski, & Hamby, 2010) by arranging them in decreasing hierarchical order.
Lastly, participants self-assessed social agreeableness as in Study 1, and carried out the morality manipulation check. We then collected participants’ hypotheses regarding the true purpose of the study and the relation of the study’s tasks. Following the collection socio-demographic information, participants were thanked and debriefed.
Results
Data Preparation. As in Study 1, we created compound scores by calculating the mean of the respective items (0.80 ≤ α ≤ 0.95, Table 8). One participant was excluded from further analyses as an outlier. As in Study 1, we excluded participants from further analyses if they failed to answer the factual manipulation check questions correctly, and retained data only from participants who
completed the survey (48%, nfinal= 242). Participant dropout did not depend on the allocation to the
manipulated conditions (Fnorm-violation[1, 501] = 0.54, p = 0.46, η2p= 0.001, Fsanctioning[1, 501] = 0.15,
p = 0.70, η2p< 0.001, Fmorality[1, 501] = 2.79, p = 0.10, ηp2= 0.01,
Fnorm-violation×sanctioning[1, 501] = 3.01, p = 0.08, η2p= 0.01, Fnorm-violation×morality[1, 501] = 4.03,
p = 0.05, η2
p= 0.01, Fsanctioning×morality[1, 501] = 0.05, p = 0.83, ηp2< 0.001).
Manipulation Checks. Regarding the understanding of the stimulus material, most participants (93.7%) answered all factual manipulation check questions correctly, and predominantly correct answers were given for each question, respectively (97.4% to 99.6%). Correct answers did not differ between the norm-violating (F [1, 250] = 0.39, p = 0.53, η2
p= 0.002), sanctioning
(F [1, 250] = 0.29, p = 0.59, ηp2= 0.001), or morality conditions (F [1, 250] = 0.48, p = 0.49, ηp2= 0.002),
and we found no interaction effects (Fnorm-violation×sanctioning[1, 250] = 0.29, p = 0.59, η2p= 0.001,
Fnorm-violation×morality[1, 250] = 1.00, p = 0.32, ηp2= 0.004, Fsanctioning×morality[1, 250] = 0.06, p = 0.81,
η2
(Fnorm-violating[1, 235] = 0.04, p = 0.84, η2p< 0.001, Fsanctioning[1, 235] = 0.03, p = 0.86, η2p< 0.001,
Fmorality[1, 235] = 0.04, p = 0.85, ηp2< 0.001) or prohibited (Fnorm-violating[1, 235] = 0.16, p = 0.83,
ηp2= 0.001, Fsanctioning[1, 235] = 0.05, p = 0.83, ηp2< 0.001, Fmorality[1, 235] = 0.34, p = 0.55,
η2
p = 0.002) to travel on a train without a ticket. Again, we found no interaction effects, both
regarding whether it was allowed (Fnorm-violation×sanctioning[1, 235] = 2.45, p = 0.12, ηp2= 0.01,
Fnorm-violation×morality[1, 235] = 2.42, p = 0.12, ηp2= 0.01, Fsanctioning×morality[1, 235] = 2.46, p = 0.12,
η2
p = 01) or prohibited (Fnorm-violation×sanctioning[1, 235] = 0.12, p = 0.73, ηp2< 0.001,
Fnorm-violation×morality[1, 235] = 1.33, p = 0.25, ηp2= 0.01, Fsanctioning×morality[1, 235] = 3.76, p = 0.05,
η2
p < 0.02) to travel without a ticket.
Regarding the success of the norm-violation and sanctioning manipulations and their orthogonality, we again conducted a multivariate analysis of variance to examine the effect of the manipulations on the impression of the target person’s adherence to rules, and punishment. Results indicate that in the norm-violating condition the target person was seen as breaking the rules more (F [1, 238] = 451.76, p < 0.001, η2
p= 0.66) and as less adhering to the rules (F [1, 238] = 1476.03,
p < 0.001, η2
p= 0.86). However, the assessment of the target person’s adherence to rules was
unexpectedly also influenced by the sanctioning manipulation: unsanctioned actors were also seen as breaking the rules more (Fbreaking rules[1, 238] = 3.68, p = 0.06, η2p= 0.02) and adhering to the rules
less (Fadhering to rules[1, 238] = 21.24, p < 0.001, ηp2= 0.08) than sanctioned actors.
Regarding the sanctioning manipulation, sanctioned actors were seen as punished more (F [1, 238] = 1413.59, p < 0.001, η2
p= 0.87), and unsanctioned actors were seen as getting away without
punishment more (F [1, 238] = 1377.07, p < 0.001, η2
p= 0.85). The norm-violation manipulation did
not influence the impression of the target as getting away without punishment (F [1, 238] = 3.22, p = 0.07, η2
p= 0.01). However, norm-violators were unexpectedly seen as being punished more
(F [1, 238] = 9.59, p = 0.002, η2
p= 0.04). Moreover, we found no significant interactions between the
norm-violating and the sanctioning manipulations on the perceived adherence to rules
(Fbreaking rules[1, 238] = 0.61, p = 0.43, η2p= 0.003, Fadhering to rules[1, 238] = 2.97, p = 0.09, ηp2= 0.01)
or the perceived punishment (Fpunished[1, 238] = 0.31, p = 0.58, η2p= 0.001, Fnot punished[1, 238] = 1.32,
p = 0.25, η2p= 0.01).
Regarding the success of the morality manipulation, we unexpectedly found no difference in the importance of moral identity (F [1, 240] = 1.20, p = 0.27, η2
p= 0.01). Yet we found that participants in
the morality condition produced short stories that related to them as a moral person more than participants in the control condition (F [1, 240] = 13.77, p < 0.001, η2
p = 0.05). Although in the
morality condition participants’ stories did not relate to themselves as students more than in the control condition (F [1, 240] = 0.03, p = 0.88, ηp2< 0.001), the stories in the morality condition unexpectedly related less to the participants as safety-conscious (F [1, 240] = 9.98, p = 0.002,
η2
p = 0.04) and more as members of certain organizations (F [1, 240] = 11.44, p = 0.001, η2p< 0.05).
Therefore, the morality manipulation is considered unreliable, and results are only reported in
Appendix C. However, since the morality manipulation did not affect the importance of moral identity as shown above, we report results of analyses with chronic observer morality below.
Confirmatory Analyses. To predict power attribution from norm-violations (H1), considering sanctioning as a moderator (H5), we first conducted a hierarchical regression analysis (Table 9). In the first step, we entered participant sex as a control variable, yielding no effect of sex (b = −0.25, SE = 0.17, p = 0.14). In the second step, we entered norm-violation and sanctioning. Supporting hypothesis 1 and replicating findings from Study 1, results indicated that more power was attributed to severe norm-violators than to norm-abiders (b = 0.94, SE = 0.11, p < 0.001). Moreover, as in Study 1, sanctioned actors were seen as less powerful than unsanctioned actors (b = −0.69, SE = 0.11, p < 0.001). In the third step, we entered the interaction term of norm-violation and sanctioning regarding hypothesis 5. Unlike in Study 1, the interaction of norm-violation and sanctioning unexpectedly was nonsignificant (b = −0.04, SE = 0.21, p = 0.84).
Secondly, to predict hypothetical leadership affordance and leadership nominations with power attributions (H2), we conducted hierarchical regression analyses (Table 10). In a first step, we entered participant sex as a control variable, showing that women afforded more leadership (b = −0.53, SE = 0.23, p = 0.03) and by trend nominated more leaders (b = −0.40, SE = 0.22, p = 0.07). In the second step, we entered power attribution. Contrary to hypothesis 2 but in support of Study 1, marginally significant results indicate that more powerful targets were afforded less hypothetical leadership (b = −0.15, SE = 0.09, p = 0.09). However, diverging results predicting leadership
nominations from power attribution showed that more powerful actors were nominated as leaders more (b = 0.32, SE = 0.08, p < 0.001), supporting hypothesis 2.
Thirdly, to assess our mediated moderation hypotheses3, predicting hypothetical leadership
affordance (Table 11) and leadership nominations (Table 12) with power attributions, considering leadership potential as a mediator (H3), chronic morality as a moderator (H4) and controlling for participant sex, we conduct two conditional process analysis using PROCESS (Figure 5, Hayes, 2013, Model 5). Participant sex had marginally significant effects on leadership potential (b = −0.41, SE = 0.24, p = 0.09) and on hypothetical leadership affordance (b = −0.30, SE = 0.16, p = 0.05), indicating that women attributed more leadership potential and afforded more hypothetical leadership than men. However, participant sex had no effect on leadership nominations (b = −0.18, SE = 0.20, p = 0.37). Contradicting hypothesis 4 and diverging from Study 1, results indicate no significant effect of power attribution on leadership potential (a = 0.13, SE = 0.09, p = 0.16). However, we found that people with higher leadership potential were afforded more hypothetical leadership (b = 0.73,
3Results for other potential moderators in this mediated moderation analysis (EH2) and alternative analysis accounting
SE = 0.04, p < 0.001)as in Study 1, and nominated as leaders more (b = 0.31, SE = 0.05, p < 0.001). The relation between power attribution and hypothetical leadership affordance remained significant and negative (c1= −0.23, SE = 0.06, p < 0.001) as in Study 1, whereas the relation between power
attribution and leadership nominations remained significant and positive (c1= 0.27, SE = 0.08,
p < 0.001). A bias-corrected bootstrap 95% CI showed that the indirect effect through leadership potential was nonsignificant (hypothetical leadership affordance: a × b = 0.09, SE = 0.07, 95% CI: [−0.05, 0.23], leadership nominations: a × b = 0.04, SE = 0.03, 95% CI: [−0.02, 0.11]), indicating no mediation by leadership potential, contradicting hypothesis 4 and in contrast to Study 1. Regarding hypothesis 5, there was again no main effect of morality on hypothetical leadership affordance (c2= −0.08, SE = 0.06, p = 0.21) or leadership nominations (c2= 0.02, SE = 0.08, p = 0.82) and
again no significant interaction effect of morality and power attribution on hypothetical leadership (c3= −0.08, SE = 0.06, p = 0.21) or leadership nominations (c3= 0.03, SE = 0.08, p = 0.73).
Regarding hypothetical leadership, the conditional direct effect was significant for morality at the mean and one standard deviation above the mean, but only marginally significant at morality one standard deviation below the mean (Table 11). Regarding leadership nominations, the conditional direct effect was significant at three levels of morality (Table 12).
A direct assessment of the effect of norm-violations on leadership affordance, mediated by power attribution and moderated by sanctioning and observer morality, respectively, can be found in
Appendix C.
Exploratory Analysis. Regarding the relation between norm-violations and power, to explore the potential mediating role of perceived responsibility in addition to volitional capacity (EH1), we further conducted a conditional process analysis with PROCESS (Figure 6) (Hayes, 2013, Model 6). Results (Table 13) indicate that norm-abiders were seen to have less volitional capacity than norm-violators (a1= 1.48, SE = 0.13, p < 0.001), and that higher volitional capacity led to higher
power attributions (b1= 0.22, SE = 0.07, p < 0.001). Moreover, norm-abiders were seen to be more
responsible than severe norm-violators (a1= −2.06, SE = 0.18, p < 0.001). However, there was no
main effect of responsibility on power attribution (b2= 0.01, SE = 0.05, p = 0.88). The positive
relation between norm-violations and power attribution remained significant (c = 0.61, SE = 0.17, p < 0.001). A bias-corrected bootstrap 95% CI showed that the total indirect effect through volitional capacity and responsibility was significant (c0= 0.31, SE = 0.13, 95% CI: [0.07, 0.56]). Regarding the partial indirect effects, only the effect through volitional capacity reached significance, indicating a partial mediation by volitional capacity, but not by responsibility. We found no influence of participant sex as a control variable (volitional capacity: b = −0.26, SE = 0.18, p = 0.14; responsibility: b = 0.23, SE = 0.19, p = 0.23; power b = −0.22, SE = 0.15, p = 0.15).
Discussion
In Study 2, we aimed to replicate the findings of Study 1, and to extend previous findings by assessing the role of experimentally manipulated morality in the power-leadership relation. Moreover, we expanded previous findings by adding a behavioral measure of leadership affordance: leadership nominations. Results partially confirmed our original expectations, and were somewhat consistent with the findings of Study 1.
Again, although our norm-violation manipulation affected perceived norm-violation severity and our sanctioning manipulation affected perceived sanctioning as expected, we found additonal
unexpected influences of the norm-violation manipulation on perceived punishment and of the sanctioning manipulation on perceived norm-violation severity. Relying on our pretest of the manipulation material, the large sample size in Study 2 could again have boosted these previously nonsignificant effects to significance. However, these unexpected influences had only small effect sizes, and can therefore again be seen to be of little practical influence for our interpretations.
However, our manipulation of morality showed insufficient fidelity, despite pretesting morality cues and using a previously validated manipulation (Aquino et al., 2009; Aquino & Reed, 2002). The manipulation did not influence the importance of moral identity, which may, however, be attributed to a lack of sensitivity of the Moral Identity Scale to state-changes. Moreover, although participants saw themselves as moral persons more in the morality condition as expected, we found confoundations with need for security and membership of certain organizations. These unexpected effects could have developed because participants worked online, where motivational aspects could have influenced the perception of the stimulus material (Reips, 2002a, 2002b). In sum, our morality manipulation may have been weaker than expected, and may be confounded, indicating that results relating to manipulated morality found in Appendix C should be interpreted with caution. Here, we further describe the results with chronic observer morality.
Supporting hypothesis 1 and as in Study 1, we replicated previous findings, showing again that norm-violators are attributed more power than norm-abiders. Regarding the underlying mediating mechanism of the relation between norm-violations and power attribution (EH1), we partially replicated previous findings by showing again a partial mediation of volitional capacity. Contrary to our speculation, we again did not find responsibility to be a mediator of the relation between norm-violations and power attribution.
Regarding the relation between power and leadership (H2) results were mixed and deviated from the findings of Study 1. Again, rather than finding a positive relation between power and hypothetical leadership affordance, we found a tendency that more powerful actors were generally afforded less hypothetical leadership. Yet, contradicting behavioral evidence with leadership nominations suggests that more powerful actors were selected as leaders more often. Moreover, the
