Self-justification as an explanation for the corrupting effects of power
The effect of power for unethical behavior, and the moderating effects of availability ofself-justifications and moral identity
Master Thesis, MSc Human Resource Management Rijksuniversiteit Groningen, Faculty of Business & Economics
June 2016 Elsard Haanstra Eendrachtskade 10D7 9726 CW Groningen 0614858633 e.g.haanstra@student.rug.nl S2177315
Abstract
There is a large amount of unethical behavior by powerful individuals. Moreover, research has found contradicting results regarding the effect of power for unethical behavior. Some research shows that power increases unethical behavior, while other research shows that power enables people to be more ethical (e.g., Chen,et al., 2001; DeCelles et al., 2012). The present research therefore aims to further develop the understanding of the effect of power for unethical behavior, and does so by investigating the roles of availability of self-justifications and moral identity, respectively. In an experimental setting, power and availability of
self-justifications were manipulated. Results show no significant moderating effect of the availability of self-justifications, and no significant moderating effect of moral identity. Results do show, however, that power increases unethical behavior. Moreover, results show that participants tended to serve their self-interest, but few participants served their self-interest as much as they could have. Implications of the present research findings, as well as future research directions are discussed.
INTRODUCTION
Major cases of ethical misconduct in organizations are piling up one after another (e.g., General Motors, Volkswagen, International Federation of Association Football (FIFA), the International Association of Athletics Federations (IAAF)). Specifically for the cases of FIFA and IAAF, this involved individuals in powerful positions taking bribes on a regular basis. Noteworthy in the FIFA case was not only the frequency with which this unethical behavior occurred, but also the degree to which this behavior was spread amongst the top layers of the organization (Apuzzo, Clifford, & Rashbaum, 2015). This high amount of unethical behavior by people in powerful positions raises the question whether power might have a corrupting effect.
The interest in the corrupting nature of power dates back centuries, and let Lord Acton (1887) to make his famous pronouncement “Power tends to corrupt, and absolute power corrupts absolutely”. In recent years, scholars have picked up this interest as well, and responded with a large number of studies regarding the corrupting effects of power (e.g., Chen, Lee-Chai, & Bargh, 2001; DeCelles, DeRue, Margolis, & Ceranic, 2012; Fiske, 1993; Galinsky, Gruenfeld, & Magee, 2003; Goodwin, Gubin, Fiske, & Yzerbyt, 2000). If anything, however, these studies have shown that the relation between power and unethical behavior is a very complex one, as studies have found a variety of seemingly contradicting and inconsistent results. On the one hand, there is an abundance of studies that show that power leads to more unethical behavior (e.g., Bargh, Raymond, Pryor, & Strack, 1995; Fiske, 1993; Galinsky et al., 2003; Goodwin, et al., 2000). On the other hand, however, there are also studies that show that power can enable people to behave more ethically (e.g., Chen,et al., 2001; DeCelles et al., 2012). Considering the high amount of unethical behavior by powerful individuals in organizations, and the inconsistent results of studies regarding the corrupting nature of power, it is paramount to further investigate the relationship between power and unethical behavior.
A finding that might help explain the contradicting results regarding the effect of power for unethical behavior is that people find it important to see themselves as ethical persons (Fischbacher & Föllmi-Heusi, 2013; Mazar, Amir, & Ariely, 2008; Shalvi, Handgraaf, & De Dreu, 2011a). The notion that people have a desire to see themselves as ethical is particularly interesting because whenever people behave in a way that contradicts their perceptions of themselves (e.g., behaving unethically), they experience cognitive dissonance (Festinger, 1957; Lowell 2012). In other words, if people possess a desire to see themselves as ethical, behaving unethically would induce cognitive dissonance.
Galinsky, Magee, Gruenfeld, Whitson, and Liljenquist (2008) shows that power holders are particularly vulnerable to cognitive dissonance, because they feel less constrained by their environment and therefore perceive more choice. Since cognitive dissonance is such an
unpleasant psychological state (Festinger, 1957), the finding that powerful individuals are more vulnerable to cognitive dissonance thus suggests that it should be harder for powerful individuals to behave unethically, than it should be for low-power individuals. This provides a striking paradox, as it should be harder for powerful individuals to behave unethically, yet both in
practice, as well as in research, power has the tendency to corrupt (e.g., Bargh et al., 1995; Fiske, 1993; Galinsky et al., 2003; Keltner, Gruenfeld, & Anderson, 2003).
Here, I propose that the availability of self-justifications might explain the paradox that theoretically it should be harder for powerful individuals to behave unethically, yet in practice and in research power often tends to corrupt. That is, self-justifications can serve to relieve or prevent cognitive dissonance, because justifications enable individuals to behave unethically while upholding a moral image (Bandura, 1999; Lowell, 2012). In doing so,
self-justifications can nullify the effect of being more vulnerable to cognitive dissonance. Therefore, I propose that the availability of self-justifications moderates the effect of power for unethical behavior, such that high-power individuals behave less unethically than low-power individuals, but only when self-justifications are not available.
In order to accurately depict the moderating effect of availability of self-justifications, however, there is another critical element to consider, namely moral identity. As mentioned before, when people possess a desire to see themselves as moral people, behaving unethically induces cognitive dissonance (Festinger, 1957). Following the same rationale, this also means that if a person does not have a desire to be ethical, behaving unethically will not induce any cognitive dissonance. In research on unethical behavior, the degree to which a person has a desire to be ethical is generally referred to as the moral identity of an individual (see Aquino & Reed, 2002).
Moral identity strongly relates to the moderating effect of the availability of
availability of self-justifications does not have an effect for power holders with a low moral identity. Therefore, I propose that there is a three-way interaction between power, availability of self-justifications, and moral identity. Specifically, I propose that only in the case of high-moral identity does the availability of self-justifications moderate the effect of power for unethical behavior.
To test the propositions mentioned above (visually presented in Figure 1), an
experimental research is conducted in which the effect of power for unethical behavior, and the moderating roles of availability of self-justifications and moral identity are examined. In doing so, the present work provides two theoretical implications. First, a lot of research on power shows that powerful people can more or less do as they please, because they are less constrained by their environment (e.g., Guinote, 2007; Keltner et al., 2003). The current research highlights a
different perspective on power, by showing that power can also have limiting effects on power holders. Specifically, the present research aims to show that in the absence of self-justifications it is harder for powerful individuals to behave unethically, than for powerless individuals.
Second, the present research contributes to much needed research on when and why moral identity exactly induces unethical behavior. As Hardy and Carlo (2011) argue, moral identity is one of the most promising trends in moral psychology in recent years. Despite being such a promising trend, there is still a lot unknown about when and why exactly moral identity affects unethical behavior (DeCelles et al., 2012). The present research intends to contribute to the understanding of moral identity by aiming to show that moral identity determines whether self-justifications moderate the effect of power for unethical behavior.
FIGURE 1 Conceptual Model
THEORETICAL FRAMEWORK Power and Unethical Behavior
Power, in line with other research, can be defined as the degree to which an individual has control over valued resources (e.g., Anderson & Berdahl, 2002; Galinsky et al., 2003). Research on unethical behavior tends to use a wide variety of definitions for unethical behavior, ranging from very specific (e.g., stealing) to very broad (e.g., failing to meet a minimal moral standard; see Jones, 1991; Treviño et al., 2006). A lot of studies and literature reviews examining the corrupting nature of power use the terminology self-interest rather than unethical behavior (e.g., DeCelles et al., 2012; Galinsky et al., 2003; Handgraaf, Van Dijk, Vermunt, Wilke, & De Dreu, 2008). Jones (1991) argues that moral issues stem from situations in which a person’s actions may harm or benefit others. Considering the nature of the experiment in the present research, and to fit with commonly used terminology in both research on unethical behavior, as well as in research on power, unethical behavior is defined as behavior that serves the self-interest and harms others (DeCelles et al., 2012; Galinsky et al., 2003; Handgraaf et al., 2008; Jones, 1991).
As mentioned before, studies examining the relationship between power and unethical behavior show inconsistent results (e.g., Chen et al., 2001; DeCelles et al., 2012; Fiske, 1993; Galinsky et al., 2003; Goodwin, et al., 2000). On the one hand, there is an abundance of studies that show that power increases a variety of unethical behaviors. For example, power increases self-interested behavior (Galinsky et al., 2003), sexual harassment (Bargh et al., 1995), and stereotyping (Fiske, 1993; Goodwin et al., 2000). On the other hand, however, power can also lead to more ethical behavior depending on characteristics of the power holder. For example, power leads to higher levels of generosity for individuals with a communal relationship orientation (Chen et al., 2001), and to less self-interested behavior for individuals with a high moral identity (DeCelles et al., 2012).
the effects of power. The fundamental idea behind the theory is that “power tips the balance of activation between the behavioral approach and inhibition systems” (Anderson & Berdahl, 2002, p. 1362). In other words, experiencing power (i.e., high-power) activates approach related processes, whereas experiencing an absence of power (i.e., low-power) activates the behavioral inhibition system. Individuals with high levels of power, therefore, tend to engage more in behaviors that help the individual in pursuing and obtaining goals and rewards, such as food, achievement, sex, safety, and social attachment. Having low levels of power, on the other hand, tends to activate the behavioral inhibition system, which functions much like an alarm-threat system (Keltner et al., 2003). The behavioral inhibition system triggers affective states, such as anxiety, avoidance and heightened sense of awareness for threats in the environment (Anderson & Berdahl, 2002).
More importantly, the approach inhibition theory of power (Keltner et al., 2003) states that powerful individuals experience less situational constraints, and thus act more in line with their own preferences, with a disregard of (negative) consequences for others. Similarly, Guinote (2007) found that the attention of powerful individuals is more focused on themselves, because they are less dependent on others for their outcomes. Interestingly, because powerful individuals are less influenced and constrained by their environment, and more focused on themselves and their internal states, they are particularly vulnerable to cognitive dissonance (Galinsky et al., 2008). Specifically, Galinsky et al. (2008) argue that “by seeing choice where others see constraint, the powerful become vulnerable to dissonance” (p. 1452).
In the next chapter I will explain that the finding of Galinsky et al. (2008) - that powerful individuals are more vulnerable to cognitive dissonance - implies that it should be harder for powerful individuals to behave unethically. The finding of Galinsky et al. (2008) thus provides a striking paradox, as both in practice and in research power has the tendency to corrupt (e.g., Bargh et al., 1995; Fiske, 1993; Galinsky et al., 2003; Keltner et al., 2003), yet being more vulnerable to cognitive dissonance should make it harder for powerful individuals to behave unethically. Furthermore, I will propose that this paradox can be explained by the availability of self-justifications.
The Moderating Role of Availability of Self-Justifications
As mentioned, in this chapter I will explain why being more vulnerable to cognitive dissonance should make it harder for powerful individuals to behave unethically. In order to do so, I will now first address recent studies suggesting that people possess a desire to see
Abundant research shows that people have a strong desire to appear ethical to others, and are more likely to behave self-interested if they have a seemingly ‘fair’ way of justifying their unethical behavior (for an overview, see Shalvi, Dana, Handgraaf & De Dreu, 2011b). In addition to the desire to appear ethical to others, recent research suggests that people are also motivated to see themselves as moral people. A number of studies show that people have a tendency to not lie as much as they could to benefit their self-interest, even when they cannot be caught (Fischbacher & Föllmi-Heusi, 2013; Shalvi et al., 2011a).
In line with these findings, Mazar et al. (2008) proposed that people will engage in a limited amount of dishonesty if it provides them with a (financial) benefit, but not so much that they cannot maintain a positive view of themselves, and a positive self-concept. Specifically, Mazar et al. (2008) argue that when people encounter the opportunity to benefit financially from being dishonest this provides tension, because people like to think of themselves as moral and honest individuals. To solve this tension people tend to use a small amount of dishonesty. Mazar et al. (2008) suggest that refraining from lying to the largest extent enables people to obtain a small amount of profit, yet still maintain a positive self-view, which includes seeing themselves as moral and honest people. If people did not have a desire to see themselves as honest and moral people, nothing would stop them from lying as much as they can to obtain the highest possible profit.
Taking the above mentioned studies into consideration it could be concluded that people do not only have a strong desire to appear ethical to others, but are also motivated to see
themselves as moral people. In that light it is quite remarkable that there is so much unethical behavior. This is especially so when considering the fact that people have a strong desire to be consistent (see Brown et al., 2011), and that whenever people behave in a way that contradicts their perceptions of themselves (e.g., behaving unethically), they experience cognitive dissonance (Festinger, 1957; Lowell, 2012). Cognitive dissonance is a very unpleasant psychological state, making humans highly motivated to reduce or eliminate it, as well as avoid situations or
information likely to increase this dissonance (Festinger, 1957).
A study that examined both the effect of the availability of self-justifications on unethical behavior, as well as the premise that people have an internal desire to feel ethical, is the study by Shalvi et al. (2011b). Their experiment consisted of a die-under-cup paradigm which allowed participants to lie anonymously and gain financially. The experiment involved participants throwing a die under a cup that only they could observe, and reporting the number that they had thrown. The higher the roll outcome the participants reported, the higher the pay they would receive for participating in the experiment. Participants were randomly assigned to either a self-justification condition or a no-self-justification condition. Participants in the self-self-justification condition were told to throw the die two additional times to verify that it was legitimate. Participants in the no-justification condition, on the other hand, were told to roll the die only once. Observance of these counter-factuals (two additional die rolls) served as the self-justification in their study. They argue that “observing desired counterfactual information justifies lying, as it modifies the extent to which a dishonest act is perceived as unethical, increasing one’s likelihood to lie” (Shalvi et al. 2011b, p. 182). The results showed that participants in the self-justification condition reported higher roll outcomes compared to participants in the no-justification condition. Furthermore, analysis revealed that participants in the self-justification condition reported the largest roll outcome of the three rolls. They did this even though they were aware of the fact these rolls were only meant to verify the legitimacy of the die, and that these rolls were not supposed to determine their pay. Observing a higher number on a roll that did not ‘count’ seemed to justify reporting a higher roll outcome.
The results of the study let Shalvi et al. (2011b) to conclude that “the extent to which people allow themselves to lie depends critically on the availability of self-justifications that no one else knows about” (p. 181). In other words, the availability of a justification for people their own behavior plays a very strong role in unethical behavior, even when this justification cannot possibly be observed by others. The study by Shalvi et al. (2011b) thus further establishes the premise that people have a desire to feel ethical for themselves, as well as emphasizes the great role that the availability of self-justifications plays in eliciting unethical behavior.
to cognitive dissonance thus suggests that it should be harder for powerful individuals to behave unethically, than it should be for low-power individuals. This provides a striking paradox, as being more vulnerable to cognitive dissonance should make it harder for powerful individuals to behave unethically, yet both in practice as well as in research power has the tendency to corrupt (e.g., Bargh et al., 1995; Fiske, 1993; Galinsky et al., 2003; Keltner et al., 2003).
An explanation for this paradox might lie in the availability of self-justifications. The reason for this is that self-justifications enable a person to behave unethically while feeling ethical. In doing so, self-justifications enable a person to behave unethically without the
experience of cognitive dissonance (Bandura, 1999; Lowell 2012). As such can the availability of self-justifications nullify the effect of being more vulnerable to cognitive dissonance. In other words, because self-justifications enable a person to behave unethically without the experience of cognitive dissonance, the availability of self-justifications might explain why there is so much unethical behavior from powerful individuals – despite the fact that being more vulnerable to cognitive dissonance should make it harder for powerful individuals to behave unethically. Therefore, I propose that the availability of self-justifications moderates the effect of power for unethical behavior, such that high-power individuals behave less unethically than low-power individuals, but only when self-justifications are not available.
Hypothesis 1: Power and the availability of self-justifications interact to influence
unethical behavior, such that power decreases unethical behavior when self-justifications are not available, but not when self-justifications are available.
The Moderating Role of Moral-Identity
In order to accurately depict the moderating effect of availability of self-justifications, however, there is another critical element to consider, namely moral identity. As mentioned, when people possess a desire to be ethical, behaving unethically induces cognitive dissonance (Festinger, 1957). Following the same rationale, this also means that if a person does not have a desire to be ethical, behaving unethically will not induce any cognitive dissonance. As outlined in the previous chapter, self-justifications enable a person to behave unethically without
Moral identity consists of two dimensions: internalization and symbolization. Moral identity internalization refers the degree to which being moral is part of a person's self-concept, whereas symbolization refers to the degree to which being moral is expressed publicly through an individual's actions (e.g., wearing certain clothing) (Aquino & Reed, 2002). Although both internalization and symbolization are predictors of (un)ethical behavior, internalization is the most significant predictor of the two (Aquino & Reed, 2002; Reed & Aquino, 2003).
Furthermore, Aquino and Reed (2002) argue that symbolization is a better predictor for actions that show ethicality or support for certain moral principles (e.g., religiosity), as opposed to internalization which better predicts actual ethical behavior (e.g., unobtrusively measured donating). Considering these reasons, internalization is more appropriate in the context of the present research than symbolization. As such, in the present study moral identity refers solely to the internalization dimension of moral identity.
The argued significance of moral identity in determining cognitive dissonance when behaving unethically is also identified by Treviño et al. (2006, p. 962). Specifically, they argue that the reason why moral identity plays such a large role in determining the degree to which a person experiences cognitive dissonance when behaving unethically is because “insofar as morality is central to a person’s self-understanding, failure to act morally creates cognitive dissonance and emotional discomfort (Blasi, 1999; Festinger, 1957).” In other words, this line of research suggests that moral identity plays a major role in the degree to which individuals with power experience cognitive dissonance when behaving unethically, because moral identity determines the degree to which the unethical behavior is incongruent with the power holders’ perceptions of themselves. More specifically, in the case of a power holder with a low moral identity (i.e., being moral is in no way part of the self-concept of a person), the power holder would be able to behave unethically without experiencing any dissonance, guilt, or remorse.
and moral identity, such that only in the case of high-moral identity does the availability of self-justifications moderate the effect of power for unethical behavior.
Hypothesis 2: There is a three-way interaction between power, availability of self-justifications, and moral identity, such that only in the case of high-moral identity does the availability of self-justifications moderate the effect of power for unethical behavior.
METHOD Participants and Design
In total 159 students from either the University of Groningen or the Hanze University of Applied Sciences participated in the research. Three participants were removed from the dataset because they explicitly indicated that they were certain there was no other person. One participant was removed because he or she tried to participate twice. Another participant was removed because the participant thought a trick was being played on him or her, and that the offer would be reversed (i.e., whatever amount participants would choose to keep for themselves would actually be the bonus for the other, and vice versa). Twenty-three participants were removed because they incorrectly answered a question checking whether they had understood what the other person could do after they had made their offer (details of all checks are provided under measures). Twelve participants were removed because they incorrectly answered a question checking whether they had read the explanation of the question. Lastly, one participant was excluded from the dataset because he or she incorrectly answered a question about his or her role in the decision making element of the experiment.
In total 41 participants were deleted, which left a total sample of 118 participants1, of which 50 participants were male (42%), and 68 participants were female (57%). The age of the participants ranged from 17 to 37 years old, with a mean age of 22 years (SD = 3.28).
Furthermore, a total of 24 different nationalities were present in the study. The nationalities most represented were: Dutch (52 participants; 43%), German (11 participants; 9%), and Romanian (10 participants; 8%).
Procedure
The study took place in the research lab in the Duisenberg building of the University of Groningen. Participants were compensated either financially or with course credits and were randomly assigned to one of four conditions, in a 2 (power: high versus low) by 2 (justification: available versus not available) design. When entering the lab, participants were greeted by the
1
experimenter and, after reading and signing the informed consent, were let to a private cubicle where the entirety of the experiment was conducted on a computer. Participants were shown the rules and procedures of the ‘decision making game’, and were told that they were randomly paired with another participant in the lab, with whom they would divide 10 chips, worth €0.50 each. Furthermore, participants were told that the number of their cubicle determined whether they would be allocator or receiver in the decision making game. In reality, all participants were allocators and they were not actually paired with another person in the lab. At the end of the study, the experimenter presented participants with the debriefing of the study, and thanked them for their participation.
Manipulations and Measures
Power was manipulated using a dictator game (high-power) versus ultimatum game
(power) setup, adapted from Handgraaf et al. (2008). Participants in both the high-power and low-power condition were told they could make an offer on how to divide a sum of chips, worth a certain amount of money. Participants in the high-power condition, however, were told their offer could not be rejected by the other participant with whom they supposedly were paired.
Participants in the low-power condition, on the other hand, were told that the other participant could reject their offer, in which case both participants would not receive a bonus payment.
Availability of Self-Justifications was manipulated by altering the default number the
drop-down menu was set to, in which participants had to select how many chips they wanted to keep for themselves. The drop-down menu in the ‘justification available’ condition was set to 8 on default, whereas the drop-down menu in the ‘no justification available’ condition was set to 5 on default
Unethical Behavior. The first measure of unethical behavior was administered by
examining the amount of chips participants allocated to themselves on a scale from 0 - 10. Keeping all 10 chips was considered to be more unethical than keeping 9, keeping 9 to be more unethical than 8, and so on. Additionally, I created a binary measure of unethical behavior where 0 = ethical behavior, and 1 = unethical behavior. Participants allocating 0 up to 5 chips were considered to have behaved ethically, whereas participants allocating 6 up to 10 chips were considered to have behaved unethically.
Moral Identity was measured with the 5-item measure of moral identity internalization
(strongly disagree) to 7 (strongly agree). Example items are: “It would make me feel good to be a person who has these characteristics” and “Being someone who has these characteristics is an important part of who I am”. The Cronbach’s alpha was .73.
Manipulation Check of Power. Four questions were asked to check the manipulation of
power. The items are: “Based on the role assignment, to what extent did you control the
allocation of the chips to yourself and the other person?”, “Based on the role assignment, to what extent did you have power in the decision making game?”, “Based on the role assignment, to what extent did the other person have power over you in the economic game?” and “How powerful did you feel in the economic game?”. The Cronbach’s alpha was .68.
Attention and Understanding of Procedures Checks. To check for attention by
participants, and to make sure they understood the procedure of the decision making game, three questions were asked across the experiment. The first question asked participants what the other person could do after they had made their offer (reject, accept, or provide a counter offer). The second questions asked participants which role they were assigned to based on their cubicle (allocator or receiver). Lastly, the third question checked whether participants read all
instructions fully, by providing a question about which sports participants played and a list of options. In the instruction above the options, however, one sentence stated that they instead should simply select ‘Does not apply’ to indicate that they had read the question fully (adapted from Oppenheimer, Meyvis, & Davidenko, 2009).
Demographic Variables. Finally, participants were asked to report their age, gender, and
nationality.
RESULTS Descriptive Statistics
participants who reported lower power checks. Moreover, power check shows a significant positive correlation with unethical behavior (binary) (r = .23, p < .05), such that participants who reported higher power checks allocated 6 up to 10 chips to themselves more often, than
participants who reported lower power checks. Lastly, moral identity shows a significant negative correlation with unethical behavior (binary) (r = -.22, p < .01), such that participants who
reported a higher moral identity allocated 6 up to 10 chips to themselves less often, than participants who reported a lower moral identity. Moral identity does not show a significant correlation with unethical behavior (r = -.10, p = .26).
Moreover, results show that, interestingly, only a small number of participants allocated the maximum number of chips (10) to themselves. Nine out of the 55 participants in the high-power condition allocated all 10 chips to themselves (16%). In the low-high-power condition, two out of the 63 participants allocated all 10 chips to themselves (3%).
TABLE 1
Spearman Correlation Results
Variables Mean S.D. 1 2 3 4 5 6 7 8 9 1. Gender .58 .50 -2. Age 22.26 3.28 -.09 -3. Level of English 1.98 .52 .07 .03 -4. Power .47 .50 -.16 -.06 -.03 -5. Self-Justification .49 .50 -.02 -.10 .13 -.00 -6. Power Check 5.56 1.06 -.10 -.08 -.09 .60 ** .04 -7. Moral Identity 5.95 .76 .02 .10 .01 -.07 .03 .10 -8. Unethical Behavior (bin.) .50 .50 -.10 .02 -.13 .12 .03 .23* -.22* -9. Unethical Behavior 6.19 1.91 -.13 .03 -.15 .23 * .03 .35** -.10 .92**
-Note. N = 118. * p < .005 ** p < .001. Unethical Behavior (binary) is coded as ethical = 0, unethical = 1. Gender is coded as male = 0, female = 1.
ANOVA Assumptions
There are three assumptions that have to be met in order for an ANOVA to give accurate results. First, there has to be independence of sample groups. Second, the sample groups have to be normally distributed. Third, there has to be homogeneity of variance between the sample groups. Since participants were randomly assigned to one of the conditions, the first assumption (independence of sample groups) is met. I will now test for the second assumption, namely normality. After that I will test for the third assumption, homogeneity of variance.
Normality Testing
To infer whether the dependent variables in the ANOVA tests are normally distributed, first a Shapiro-Wilk test was conducted. The test results are presented in Table 2, and indicate that none of the variables are normally distributed. However, another way to infer whether a variable is normally distributed, is by analyzing the z-scores of skewness and kurtosis. Kim (2013) argues that for medium sized samples (50 < n <300), when the absolute z-values of kurtosis and skewness, respectively, are greater than 3.29, the null hypothesis should be rejected, and the conclusion should be drawn that the distribution of the sample is non-normal. The
respective z-values can be obtained by dividing skewness with the S.E. of skewness, and kurtosis with the S.E. of kurtosis (Kim, 2013). All values have been reported in Table 3. Results show that none of the absolute z-values are greater than 3.29 (see Table 3). Therefore, there is enough evidence to assume that the variables are normally distributed. Moreover, although violations of normality may increase the chance of false positive results when using a test that assumes normally distributed variables, an ANOVA has been found to be quite robust. Moderate
deviations from normality therefore do not affect the false positive rate much. (Glass, Peckham, & Sanders 1972; Harwell, Rubinstein, Hayes, & Olds, 1992; Lix, Keselman, & Keselman, 1996). An ANOVA can thus still be used even if there are moderate violations of normality.
TABLE 2
Shapiro-Wilk Test Results
Variable Statistic Degrees of Freedom Significance
Unethical Behavior .83 118 .00
Power Check .95 118 .00
TABLE 3
Normality Test Results
Variable Skewness S.E. of Skewness Z-value of Skewness Kurtosis S.E. of Kurtosis Z-value of Kurtosis Unethical Behavior 0.28 0.22 1.27 1.03 0.44 2.34 Power Check -0.6 0.22 -2.73 -0.76 0.44 -1.73 Moral Identity -0.4 0.22 -1.82 -0.03 0.44 -0.07
Homogeneity of Variance Testing
To test for homogeneity of variance, two Levene’s tests were conducted. The dependent variable in the power manipulation check, power check, showed a homogeneity of variance (p = .426). For the variable unethical behavior, a Levene’s test showed no homogeneity of variance (p < .001). To give more insight into this absence of homogeneity of variance, the means and standard deviations for each condition are presented in Table 4 below. Moreover, since there is a heteroscedasticity of variance the results of the ANOVA are possibly not fully accurate. As I will explain later, to provide more accurate results a Welch test and a Chi-Square test will be
conducted. Specifically, the Welch test has been found to be a very well performing test under violations of homogeneity (Lix et al., 1996).
TABLE 4
Means and Standard Deviations of Unethical Behavior
Condition N Mean S.D.
1. Low-Power, Self-Justification available 31 5.61 1.67 2. Low-Power, Self- Justification not available 32 5.81 1.26 3. High-Power, Self-Justification available 27 7.00 2.04 4. High-Power, Self-Justification not available 28 6.46 2.37
Power Manipulation Check
To check whether the power manipulation worked, an ANOVA was conducted. The test results showed a significant effect of power on the power check F(1,116) = 58.78, p < .001, R² = .336. Participants in the high-power condition reported a higher power check (M = 6.22; SD = .83), than participants in the low-power condition (M = 4.99; SD = .90), indicating that the power manipulation was successful.
Hypotheses Testing
A 2 (power: low vs. high) x2 (self-justification: not available vs. available) ANOVA tested the effects of power and availability of self-justifications on unethical behavior. Results indicated a significant main effect of power, F(1,114) = 8.89, p = .004, R² = .07. Participants in the high-power condition allocated more chips to themselves (M = 6.73; SD = 1.46), compared to participants in the low-power condition (M = 5.71; SD = 2.21). There was no significant main effect of the availability of self-justifications condition, F(1,114) = .242, p = .624, R² = .002. Moreover, the interaction effect between power and availability of self-justifications was not significant, F(1, 114) = 1.156, p = .284, R² = .010. A visual representation of the interaction effect can be found in Figure 2, which shows that power increases unethical behavior. Moreover, Figure 2 shows that the availability of self-justifications increases unethical behavior in the high-power condition, but not in the low-high-power condition.
Moreover, I examined my hypothesis for participants who participated for money or for research points separately, to see if they provide different results. The reason for this is that participants who participated for money chose to do so voluntarily, whereas participating for research points is an obligation for students of certain courses. Moreover, since a bonus can be obtained by acting unethically, participants participating for money may be more or less
motivated to obtain a high bonus, than participants who participated for research points. For the interaction effect between power and availability of self-justifications for unethical behavior, when excluding participants who were compensated with research points, as opposed to a financial compensation, results indicate a slightly more significant, yet still insignificant effect F(1, 106) = 1.59, p = .211, R²=.015. Visually, the effect looks similar to Figure 2.
As mentioned, a Levene’s test of equality of variances showed that there is no
homogeneity between sample groups (p < .001). This means that the results of the ANOVA are possibly not fully accurate. As mentioned, in the case of a violation of the homogeneity
.005. Furthermore, results showed no significant main effect of availability of self-justifications on unethical behavior F(1, 115) = .162, p = .688.
Since a Welch test cannot be used for interaction effects, to analyze whether power and availability of self-justifications have an interaction effect on unethical behavior, the unethical behavior variable was recoded to a binary variable, where 0-5 chips was considered ethical and 6-10 chips unethical. A Chi-square analysis was then conducted, and showed that there is no significant main effect of power on unethical behavior X²(1) = 1.67, p = .196. Furthermore, a second Chi-square analysis showed no significant main effect of availability of self-justifications on unethical behavior X²(1) = 0.136, p = .713. Lastly, a third Chi-square analysis showed that there is no significant interaction effect of power and availability of self-justifications on unethical behavior X²(1) = 1.15, p = .284.
To summarize, all analyses show no significant interaction effect of power and availability of self-justifications on unethical behavior. This means that for hypothesis 1, we cannot reject the null hypothesis in favor of the alternative hypothesis. Moreover, the data analysis does reveal a significant main effect of power for unethical behavior.
FIGURE 2
Interaction Effect of Power and Availability of Self-Justifications for Unethical Behavior
behavior b = -.05, t(110) = .16, p = .89; and no significant two- way interaction of self-justification and moral identity for unethical behavior b = -.24, t(110) = -.65, p = .52. Most importantly, the results show no significant three-way interaction of power, availability of self-justifications, and moral identity for unethical behavior b = -.86, t(110) = -1.14, p = .26. Therefore, for hypothesis 2, we cannot reject the null hypothesis in favor of the alternative hypothesis. Furthermore, a visual representation of the effect can be found in Figure 3.
TABLE 5
Results Regression Analysis
B t p
Power .53 2.60 .01
Self-Justification .03 .17 .87
Moral Identity .12 .67 .51
Power × Self-Justification .39 .97 .34
Power × Moral Identity -.05 -.16 .89
Self-Justification × Moral Identity -.24 -.65 .52
Power × Self-Justification × Moral Identity -.86 -1.14 .26
Note. N = 118. Power is coded as low-power = 0, high-power = 1. Self-Justification is coded as no justification available = 0, justification
available = 1.
FIGURE 3
DISCUSSION Findings
The goal of the present study was to contribute to the understanding of the effect of power for unethical behavior, and did so by testing two hypotheses. The first hypothesis stated that powerful individuals behave less unethically than low-power individuals, unless
self-justifications are available. The results of the experiment do not support the first hypothesis. The second hypothesis stated that only in the case of high-moral identity does the availability of self-justifications moderate the effect of power for unethical behavior. The results of the experiment do not support the second hypothesis either. Unexpectedly however, results do show that power increases unethical behavior.
Theoretical Implications
The present research has important theoretical implications. Firstly, results show that power leads to more unethical behavior. Since the present research uses an absolute level of power (i.e., full control over resources), results suggest that absolute power corrupts. The finding that absolute power corrupts is particularly interesting because - as stressed earlier in this paper - some studies show that power corrupts (e.g., Fiske, 1993; Galinsky et al., 2003; Goodwin et al., 2000) whereas other studies show that it does not (e.g., Chen et al., 2001; DeCelles et al., 2012).
As mentioned, the present research uses absolute levels of power for participants in the high-power condition. Whereas absolute power can be regarded as having full power and control, relative power refers to the amount of power a person has in relation to another person (see Dahl, 1957; Wolfe & McGinn, 2005). Contrary to the present research, other studies often use lower than absolute levels of power (i.e., relative power). For example, a lot of studies on the effects of power use power primes to manipulate power (e.g., Decelles et al., 2012; Galinsky et al., 2003; Galinsky et al., 2008). A power prime arguably provides participants with a much weaker level of power than the absolute power manipulation used in the present study. As such, it is quite
plausible that in the case of absolute power, power corrupts, whereas in a situation with relative power (i.e., less than absolute power, yet still more powerful than another person), power does not corrupt. This might explain why studies on the effect of power have found varying and contradicting results.
Secondly, the present research also has important theoretical implications for the literature on people their desire to be a moral person. Specifically, the results show that only 9 out of the 55 participants in the high-power condition allocated all 10 chips to themselves (16%). Moreover, as mentioned, results show that power increases unethical behavior. In other words, although
participants actually served their self-interest as much as they could have - despite having full control over resources and full anonymity. The present research thus provides similar results to a number of other recent studies in which participants served their self-interest less than they could have, despite there being no possible repercussions (e.g., Fischbacher & Föllmi-Heusi, 2013; Shalvi et al., 2011a; Mazar et al., 2008).
The finding that people did not serve their self-interest as much as they could have, has interesting implications. This is particularly so, because the finding shows that rather than acting either purely selfishly or purely ethically, participants showed a combination of both selfishness and ethicality. Considering participants were anonymous, it suggests that although people do benefit their self-interest when given the opportunity, they do not want to see their own actions as immoral. Similar to what Mazar et al. (2008) proposed with their theory of self-concept
maintenance, participants seemed to have acted in a way that benefited their self-interest, but not so much as to threaten their positive self-concept or self-worth. The results of the present
research thus further strengthen the premise that people have a desire to see themselves as moral people.
Practical Implications
The present research also has important practical contributions. As mentioned in the introduction, there is a large amount of unethical behavior from powerful individuals in
organizations (e.g., FIFA, IAAF). The results of the present research show that absolute power corrupts. Moreover, the effect of power for unethical behavior was unmitigated by the power holder’s moral identity. Considering the high amount of unethical behavior by powerful individuals, and the results showing that absolute power increases unethical behavior, it is paramount for both organizations, as well as societies, to make sure that power is never centralized too much in one person.
There are two main ways organizations and societies can prevent centralization of power. First, decreasing the amount of control a person has over valued resources will decrease the power of that individual. This could be done by, for example, increasing the amount of
individuals that control a particular set of valued resources. Rather than giving one person full control, a higher number of people should control the valued resources together. In doing so, it is ensured that not one person receives all power, but rather a number of individuals all receive a lower level of power.
combination of the control over resources and the freedom (i.e., absence of constraints) to take action with said resources. Empirical research on accountability indeed shows that decreasing the freedom a powerful person has in abusing power (i.e., by increasing accountability) results in less illegal behavior from powerful individuals (Mitchell, Hopper, Daniels, Falvy, & Ferris, 1998). In other words, in order to reduce unethical and illegal behavior from powerful individuals, it is paramount that the abuse of power is being investigated and prosecuted.
Limitations
The present research has certain limitations which have to be taken into account. The main limitation of the present research is the poor quality of the data. First, the entire sample consisted of students, which hinders the generalizability of the results to the other social groups, such as managers. Second, a high amount of participants failed to meet one or more of the
attention and procedure checks, or who explicitly stated that they did not think they were actually paired with another participant (in total 41 out of 159 participants failed; 26%). Moreover, a third factor that limited the quality of the data of the present research is that the maximum bonus participants could earn in the experiment was relatively small (€5). It is likely that with higher incentives participants are more motivated to act self-interestedly, and higher incentives would have thus likely provided different results.
A limitation with regards to the procedure of the present research, is that a lot of participants indicated that they did not think they were actually paired with another participant during the experiment. Participants, however, indicated this at the end of the experiment, after the actual decision making game. When asked explicitly, most participants seemed to indicate that they figured out there was no other person later in the study (i.e., after the decision making game), mainly because they did not hear anything about whether their offer was accepted or not during the experiment itself. Because participants indicated whether they believed there was another person after the actual experiment, it is hard to really detail which participants did, or did not, believe there was another person during the decision making game itself.
think other participants were actually present. What is especially noteworthy is that some participants indicated that they did not think there were any others present, when in reality there were. In those cases they merely did not actually see any other participants because the other participants had already started, or were yet to arrive. If researchers intend to replicate the present research, making sure participants actually see other participants upon arrival is paramount for participants their belief that they are actually interacting with another person during the decision making game. If at a certain time(slot) only one participant arrives, a confederate could be used.
Lastly, a limitation of the present research is that results indicate that the self-justification manipulation did not work. Moreover, it leads to wonder why the self-justification manipulation did not work. As I will propose in the next chapter ‘Future Research’, it might be that
participants in the high-power condition were not able to accurately interpret the ethicality of the situation correctly because they received absolute power. When people do not interpret a certain situation or behavior as unethical, people could have an extremely strong desire to be ethical, or have all the justifications in the world to justify their behavior, it still would not make any difference: if people feel they are being ethical in the first place, neither the desire to be ethical nor the absence or presence of (self)-justifications is going to have any effect. As such it could be that the reason why the self-justification manipulation did not work, is because participants received absolute power, which, in turn, limited their ability to interpret the morality of the situation accurately.
Future Research
The results of the present research also provide two important directions for future research. First, as mentioned before, the unexpected effect of power for unethical behavior is likely to be the result of the use of absolute power in the present study. Future research should explain why absolute power would corrupt, and relative power would not. A possible explanation for the found effect of power for unethical behavior could be that in the case of absolute power an individual is no longer able to accurately interpret the ethicality of a situation. Since approach tendencies become stronger when power increases (Keltner et al., 2003), perhaps in a situation of absolute power, the tendency to approach takes away too much of the required attention to accurately interpret the morality of a situation. Moreover, since Guinote (2007) found that the attention of powerful individuals is more focused on themselves, it is quite plausible that in the case of absolute power the attention of power holders is simply too focused on themselves to be able to recognize the ethicality of the situation.
whether power increases moral hypocrisy. The results show that “the powerful judged their own moral transgressions as more acceptable than other people’s, but low-power participants did not” (Lammers et al., 2010, p. 742). Although the results do not specifically relate to the effect of absolute power on moral awareness, the results do clearly show that power affects the way individuals interpret the morality of a situation (and their own behavior).
A study that did examine the effect of power for moral awareness is the study of Decelles et al. (2012). They found that in the case of a high moral identity power leads to higher moral awareness, and in the case of a low moral identity power leads to lower moral awareness. Although the finding of Decelles et al. (2012) is interesting, it has to be noted that they used a power prime, rather than an absolute power manipulation. In other words, their study does not signify what absolute power does to the moral awareness of an individual. The effect of absolute power for moral awareness thus remains unexamined.
The results of the present study, and the argumentation provided above, indicate that it is possible that absolute power affects the ability to accurately interpret the morality of a situation. It could be that in the case of absolute power, the tendency to approach becomes so strong that from a cognitive perspective a person with absolute power simply no longer pays enough attention to the situation to be able to interpret the situation's ethicality accurately. Future research is needed to investigate how power affects moral awareness, and if absolute power indeed diminishes a person’s ability to interpret the morality of a situation accurately.
There is a second direction for future research provided by the results of the present research. As argued before, the use of different power manipulations might account for varying results regarding the effects of power for unethical behavior. Indeed, many studies seem to perhaps take a too simplistic, or linear, perspective on the effect of power: in most power research, a low versus high-power approach is used, with the sole distinction being that
participants in the high-power condition experience or receive more power than participants in the low-power condition. There, however, is no distinction made between the actual level of power that participants receive (e.g., absolute-power, high-power, moderate-power, low-power, or even no power at all).
or reject. In case the receiver would reject the offer made by the allocator, the reward participants would receive was determined by multiplying the offer by either 1.0 (i.e., dictator game, highest power), 0.9 (high-power), 0.1 (low-power) or 0.0 (i.e., ultimatum game, no power). Results show that, surprisingly, participants in the highest power manipulation (1.0 multiplier, dictator game) behave less self-interestedly than participants in the high-power condition (0.9 multiplier). In other words, the results of the study by Handgraaf et al. (2008) imply that the effect of power for self-interested behavior is curvilinear, rather than linear, in nature. Moreover, the results of their study signify the importance of carefully assessing different levels of power used, by showing that even slightly different levels of power can have significantly different effects.
The significance in the use of different power manipulations is also stressed by Sturm and Antonakis (2015). They argue that, although most research on power uses power primes, power primes are distinctly different from actual power (i.e., dictator games). Specifically, they argue that such manipulations generally make clear to participants what the purpose of the experiment is, thereby creating expectations for them on how to behave, affecting ecological validity. The difference in ecological validity, in turn, affects the results of studies, making it likely that studies using real power, versus power primes, generate different results.
Overall, I agree with Sturm and Antonakis (2015) their proposition that different power manipulations affect results of power studies. However, I would like to add that this is not only because of ecological validity, but also because (even slightly) different power levels can have very different effects. The common distinction used in research on power (high-power versus low-power) does, in my opinion, not suffice. In order to fully and accurately examine the effects of power it is paramount that in future research more explicit distinctions are made for the different levels of power attributed to participants.
Conclusion
The present research aimed to examine the relation of power and unethical behavior by investigating the roles of availability of self-justifications and moral identity, respectively, in the corrupting effect of power. The results of the study were unexpected: neither of the two
hypotheses were confirmed, and unexpectedly results showed that absolute power increases unethical behavior. The results of the present research signify the corrupting effect absolute power has: not only does absolute power lead to unethical behavior, the effect was also
unmitigated by the power holder’s moral identity. In that sense the results of the present research indicate that Lord Acton was right all along: Power tends to corrupt, and absolute power
REFERENCES
Anderson, C., & Berdahl, J. L. (2002). The experience of power: examining the effects of power on approach and inhibition tendencies. Journal of Personality and Social Psychology, 83(6), 1362.
Antes, A. L., Murphy, S. T., Waples, E. P., Mumford, M. D., Brown, R. P., Connelly, S., & Devenport, L. D. (2009). A meta-analysis of ethics instruction effectiveness in the sciences. Ethics & Behavior, 19(5), 379-402.
Apuzzo, M., Clifford, S., & Rashbaum, W. (2015, May 26). FIFA Officials Arrested on
Corruption Charges; Blatter Isn’t Among Them. The New York Times. Retrieved January 7, 2016, from
http://www.nytimes.com/2015/05/27/sports/soccer/fifa-officials-face-corruption-charges-in-us.html?_r=0
Aquino, K., & Reed II, A. (2002). The self-importance of moral identity. Journal of Personality and Social Psychology, 83(6), 1423.
Bandura, A. (1999). Moral disengagement in the perpetration of inhumanities. Personality and Social Psychology Review, 3(3), 193-209.
Bargh, J. A., Raymond, P., Pryor, J. B., & Strack, F. (1995). Attractiveness of the underling: An automatic power→ sex association and its consequences for sexual harassment and aggression. Journal of Personality and Social Psychology, 68(5), 768.
Blasi, A. (1999). Emotions and moral motivation. Journal for the Theory of Social Behaviour, 29(1), 1-19.
Brown, R. P., Tamborski, M., Wang, X., Barnes, C. D., Mumford, M. D., Connelly, S., & Devenport, L. D. (2011). Moral credentialing and the rationalization of misconduct. Ethics & Behavior, 21(1), 1-12.
Chen, S., Lee-Chai, A. Y., & Bargh, J. A. (2001). Relationship orientation as a moderator of the effects of social power. Journal of Personality and Social Psychology, 80(2), 173.
Dahl, R. A. (1957). The concept of power. Behavioral Science, 2(3), 201-215.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press. Fischbacher, U., & Föllmi-Heusi, F. (2013). Lies in disguise—an experimental study on cheating.
Journal of the European Economic Association, 11(3), 525-547.
Fiske, S. T. (1993). Controlling other people: The impact of power on stereotyping. American Psychologist, 48(6), 621.
Galinsky, A. D., Gruenfeld, D. H., & Magee, J. C. (2003). From power to action. Journal of Personality and Social Psychology, 85(3), 453.
Galinsky, A. D., Magee, J. C., Gruenfeld, D. H., Whitson, J. A., & Liljenquist, K. A. (2008). Power reduces the press of the situation: implications for creativity, conformity, and dissonance. Journal of Personality and Social Psychology, 95(6), 1450.
Glass, G. V., Peckham, P. D., & Sanders, J. R. (1972). Consequences of failure to meet assumptions underlying the fixed effects analyses of variance and covariance. Review of Educational Research, 42(3), 237-288.
Goodwin, S. A., Gubin, A., Fiske, S. T., & Yzerbyt, V. Y. (2000). Power can bias impression processes: Stereotyping subordinates by default and by design. Group Processes & Intergroup Relations, 3(3), 227-256.
Guinote, A. (2007). Behaviour variability and the situated focus theory of power. European Review of Social Psychology, 18(1), 256-295.
Handgraaf, M. J., Van Dijk, E., Vermunt, R. C., Wilke, H. A., & De Dreu, C. K. (2008). Less power or powerless? Egocentric empathy gaps and the irony of having little versus no power in social decision making. Journal of Personality and Social Psychology, 95(5), 1136.
Hardy, S. A., & Carlo, G. (2011). Moral identity: What is it, how does it develop, and is it linked to moral action?. Child Development Perspectives, 5(3), 212-218.
Jones, T. M. (1991). Ethical decision making by individuals in organizations: An issue-contingent model. Academy of Management Review, 16(2), 366-395.
Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110(2), 265.
Kim, H. Y. (2013). Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis. Restorative Dentistry & Endodontics, 38(1), 52-54.
Lix, L. M., Keselman, J. C., & Keselman, H. J. (1996). Consequences of assumption violations revisited: A quantitative review of alternatives to the one-way analysis of variance F test. Review of Educational Research, 66(4), 579-619.
Lord Acton. (1887). Quotes. (n.d.). Quotes.net. Retrieved January 27, 2016, from http://www.quotes.net/quote/37674.
Lowell, J. (2012). Managers and Moral Dissonance: Self Justification as a Big Threat to Ethical Management? Journal of Business Ethics, 105(1), 17-25.
Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: A theory of self-concept maintenance. Journal of Marketing Research, 45(6), 633-644.
Mitchell, T. R., Hopper, H., Daniels, D., Falvy, J. G., & Ferris, G. R. (1998). Power, accountability, and inappropriate actions. Applied Psychology, 47(4), 497-517.
Oppenheimer, D. M., Meyvis, T., & Davidenko, N. (2009). Instructional manipulation checks: Detecting satisficing to increase statistical power. Journal of Experimental Social
Psychology, 45(4), 867-872.
Reed II, A., & Aquino, K. F. (2003). Moral identity and the expanding circle of moral regard toward out-groups. Journal of Personality and Social Psychology, 84(6), 1270.
Shalvi, S., Dana, J., Handgraaf, M. J., & De Dreu, C. K. (2011b). Justified ethicality: Observing desired counterfactuals modifies ethical perceptions and behavior. Organizational Behavior and Human Decision Processes, 115(2), 181-190.
Treviño, L. K., Weaver, G. R., & Reynolds, S. J. (2006). Behavioral ethics in organizations: A review. Journal of Management, 32(6), 951-990.
