Master Thesis, MSc Human Resource Management University of Groningen, Faculty of Economics and Business
June, 2016
Ridwan Y.S. Gumilang Student number: S3036006
Stationsplein 9, F04 9726 AE Groningen
Email: r.y.suryo.gumilang@student.rug.nl
ABSTRACT
Rules can be used to prevent unethical behaviour, and one way to categorize them is by how they are framed: general and specific rules. General rules communicate broad values and affect
multiple behaviours, while specific rules are more detailed and only communicate one particular behaviour. However, a general rule leaves room for moral rationalization while a specific rule do not. Therefore, under which circumstances is a general rule less effective than a specific rule? In this study, I hypothesize that general rules are less effective when the reward for unethical behaviour is high compared to when it is low. Meanwhile, specific rules are effective
INTRODUCTION
Rules are ubiquitous in our daily lives, and we have been introduced to them since our childhood. They take form as laws that govern us as citizens of a nation, ethical codes that regulate professionals and workers in an organization, and even in school code of conducts that directs students. One way to categorize rules is to look at how they are framed, which can be classified as general and specific rules. General rules are more comprehensive and targets more than one behaviour, but they are less enforceable; whereas specific rules are less comprehensive because they target one behaviour, but they are more enforceable (Mulder, Jordan & Rink, 2015).
An example of a general rule would be “do what is morally right”, and a specific rule would be “you are not allowed to copy other people’s work during the exam”. It can be seen in the example that a general rule states a vaguer message compared to specific rules, but the words “do what is morally right” contains a comprehensive value that is not limited to one particular behaviour like the specific rule does. General rules affect multiple behaviours because it contains a broad value and they also generate moral imagination in individuals. General rules can also be said to focus more on values, whereas specific rules emphasize compliance. The value focus is especially intriguing, because according to Wiever and Trevino (1999), a value orientation would generate other outcomes than only compliance, such as commitment to the value itself.
explanations would imply that the abstract nature of general rules could create more room for rationalization for an individual to act unethically.
Thus, there are two sides to general rules, on one hand, general rules are effective in influencing multiple behaviours because they more strongly communicate moral principle than specific rules. On the other hand, the vagueness of general rules makes them prone to moral rationalizations. Because of these two sides of general rules, it is interesting to find out the possible moderators on the effectiveness of general rules in preventing unethical behaviour. In other words, to find out when a general rule has positive effects on ethical behaviour and when it has negative effects. In this study, I propose that the effectiveness of general rules is dependent on how much an individual can gain incentives by being unethical. High incentives motivate individuals to rationalize unethical behaviour (Zhao & Miao, 2015). Since general rules allow for moral rationalizations, it is therefore logical to believe that general rules are less likely to reduce unethical behaviour when incentives are high.
THEORY SECTION
Rules
Rules govern our daily lives, be it formal or informal. Formal rules in particular are very important in regulating people’s daily interaction (Galbiati & Vertova, 2014). A “rule” is defined in the Merriam-Webster Dictionary as “a statement that tells you what is or is not allowed in a particular game, situation, etc.” Whilst Kantorowicz in his book “The Definition of Law”, defined the main function of rules is to prescribe what ought to be done by an individual, with the primary duty of the individual to act in a certain way, and the secondary duty to submit to sanctions if the primary duty is not fulfilled (Woozley, 1960).
General and Specific Rules
A rule or law can be general and leaves room for discretion or it could be specific to reduce discretion (Mahoney & Sanchirico, 2005). Specific rules focus specifically on the targeted behaviour, and general rules communicate the behaviour in an abstract way (Mulder et al., 2015). An example of a general rule would be “do not engage in conflicts of interest”. It is general in the sense that it does not specify the types of behaviour that would specifically be forbidden, but it communicates the general values. Specific rules are more detailed; specifically stating the exact behaviour that an individual should or should not do, like “it is not allowed to take any form of gifts from customers”. In communicating values, general rules could affect multiple behaviours related to the value, rather than one particular behaviour like a specific rule would (Mulder et al., 2015).
conceptualized compliance and value based ethics. In their research, they propose that
compliance-based ethics is focused more on rule compliance and the threat of punishment for noncompliance, while value-based ethics focuses more on ethical values. Value-based ethics usually contain abstract core ideals such as “respect” and “honesty”. Further, they proposed that both value and compliance-based ethics would result in conformity, however, value-based ethics may induce other outcomes such as higher commitment towards organizational values.
Compliance focused ethics emphasizes the “don’t get caught” motivation (Paine 1994). People comply with the rule because the fear of punishment by violating it. In companies using compliance focused ethics, managers set the compliance standards and procedures and they focus on specific actions rather than broad values (Paine, 1994). These specific standards tell individuals what is ethical and what is not. It has been argued that these specific standards limit an ethical behaviour (Tenbrunsel, Wade-Benzoni, Messick, & Bazerman, 2000). It is true, that they will comply and behave by following those standards, mainly because the fear of
punishment, but that’s where it stops. The effect of those standards does not make people more ethical in ambiguous situations not specifically explained in those standards. Furthermore, Gates (2004) argued that ethics based on strictly following only the law does not drive individuals to think more critically about their actions.
In contrast, it is argued that a principle or value based ethics address issues that go
much as compliance focused ethics. Additionally, a value focused ethics empower individuals to think critically how their actions affect other stakeholders (Gates, 2004). Since general rules are more similar to value-based ethics than specific rules, then it is logical to think that the values communicated in a general rule would also be effective beyond the standard behaviours set in a specific rule, which can be seen as an advantage.
Moral Rationalization
The above argument suggests that general rules work better to enhance ethical behaviour than specific rules. However, this may not necessarily be the case. One reason why general rules may be less effective than specific rules is that they allow for moral rationalization. Moral rationalization is the process in which people convince themselves that their behaviour does not violate their moral standards and are still consistent with their moral principle. (Tsang, 2002). Tsang (2002) posits that her definition of moral rationalization is closely related to the theory of moral disengagement which proposes that when doing something immoral, an individual would first need to disengage themselves from their moral self (Bandura 1991, 1999). Moral
rationalization is also present in the concept of motivated reasoning, which describes that an individual’s motivation affect their reasoning through a set of biased cognitive process (Kunda, 1990). According to this theory, an individual motivated to come to a certain conclusion will rationalize that the determined conclusion is logical, even if it is unethical.
First, Individuals engaging in unethical behaviour could reconstruct their behaviour so it is not viewed as immoral. An example of this would be a liar who believes that it is not morally wrong to tell “little white lies”. This example illustrates euphemistic labeling, which involves an individual’s effort to make an unethical conduct look more respectable (Bandura, 1999).
Secondly, individuals who morally disengage could also disregard the consequences of their actions, and they do so by avoiding facing the negative consequence they cause (Bandura, 1999), or try to discredit evidence of harm (Tsang, 2002). An example of this would be to rationalize that a little lie will not hurt anybody. Third, individuals who morally rationalize could blame or dehumanize their victims. Lastly, individuals who morally disengage could displace the
responsibility of their action by blaming the authority who gives the order to act immorally (Bandura, 1999), or by diffusing responsibility, where they deny taking responsibility for harm caused by combined action (Tsang, 2002).
From the discussions above, it is then obvious that moral rationalization is related to unethical behaviour. This is supported by Detert et al. (2008) and Barsky (2011), who proved that moral disengagement is an antecedent of unethical behaviour. Also, Hsee (1995) showed in their study that the opportunity to morally rationalize encourages people to act unethically. In their research, it was discovered that when asked to make a choice between an option that is less rewarding but is relevant to the completion of a task (option A) and another option that is not very relevant to the task but is personally rewarding (option B), individuals would tend to choose option B if there is ambiguity or uncertainty in option A. Ambiguity creates a room for
that is personally satisfying for them even if it is less related to their task, indicating deviant behaviour.
Relationship Between General and Specific Rules with Moral Rationalization
As described above, a specific rule would leave less discretion compared to a general rule (Mahoney and Sanchirico, 2005). This would imply that the little discretion in specific rules would reduce rationalization, and that general rules leave more room for rationalizing. This is supported by Mulder et al. (2015), which proposes that specific rules reduce rationalizations that would conclude an immoral behaviour as being morally permissible compared to general rules. This is because by being specific, there is less room for interpretation. Using the previous example, “do not engage in conflicts of interest” would leave more room for interpretation compared to “if one is offered a gift from a client, then he/she must decline it”. General rules leave room for moral rationalization because they are vague, resulting in more discretion to interpret the communicated values. In the examples, the general rule does not explicitly say that taking gifts are not allowed, but accepting them are still a conflict of interest. However, because it is vague, it does leave room for moral rationalization; for example, an employee accepting a gift can morally rationalize by minimizing the consequence, and say that accepting small gifts does not harm anybody. Meanwhile, the specific rule would not allow such rationalizations because it is perfectly clear that accepting gifts are forbidden.
unethical behaviour. In other words, what the incentives for the unethical behaviour are. This will be argued in the following paragraph.
Incentives and Unethical Behaviour
An Incentive, according to the Merriam-Webster Dictionary, is defined as “something that incites or has a tendency to incite to determination or action”. Going by the definition in the dictionary, incentives is not restricted to money alone like many researches often implies, especially in the field of business. Incentives can also be in the form of social rewards such as reputation (Pascual-Ezama et al., 2015) and prestige (Pascual-Ezama et al., 2013). Therefore, in this research, discussions relating to incentives will include both financial and social rewards.
Further supporting the research of financial incentives’ influence on immoral behaviour, Gino and Mogilner (2014) discovered that priming money increases dishonesty.
Non-monetary rewards may also influence immoral behavior. According to Pascual-Ezama et al., (2015), economic or social rewards can modify an individual’s behaviour. In their research, they found that competition without supervision would increase cheating, and the frequency of cheating increases with social rewards, in this case, reputation. This finding is also supported by a previous research by Pascual-Ezama et al., (2013) that revealed workers would increase their dishonesty when prestige, as a social reward, is at stake.
So, the higher the incentives for unethical behaviour, the more likely people are to be motivated to engage in unethical behaviour. Logically, this would also influence motivated reasoning, which according to Kunda (1990), is a situation when individuals motivated to come to a certain conclusion tries to rationalize and justify their desired conclusion to other people. In the process, individuals could combine information and create a new belief to support their preferred conclusion. In relation to incentives and rationalization, when an unethical act results in personally interesting rewards, people are more motivated to act unethically than when an unethical act results in less interesting rewards. Hence, when rewards are high, motivated reasoning is more likely. Motivated by the reward, the individual would conclude that the unethical behaviour is the appropriate thing to do, and in the process, they will try to rationalize that the unethical act is acceptable. In other words, high rewards will encourage individuals to morally rationalize their unethical behaviour.
the business environment, Baron, Zhao, and Miao’s (2015) research concluded that
entrepreneurs’ motivation for financial gain had an indirect influence on unethical behaviour through the effect of moral disengagement. In their research, one of the examples given was that entrepreneurs firmly focused on financial gains would use advantageous comparison to justify themselves when engaging in unethical behaviour.
Influence of Incentive on the Effectiveness of Rules
This study argues that incentives may determine the effectiveness of general rules, but not so much the success of specific rules. After all, a specific rule leaves no opportunity for an individual to morally rationalize. So, even when incentives and thus motivation to for the unethical behaviour are high, a specific rule will withhold people from engaging in it. This is because specific rules clearly specify what is right or wrong, which makes it harder to rationalize that an unethical act is ethical. In contrast, a general rule does leave room for moral
rationalization. When incentives are low, general rules may succeed in communicating moral values and encouraging individuals to think more critically about their moral behaviour, thus inducing people to act morally. However, when the incentive to act unethically is high,
temptation to engage in it will induce moral rationalization and people may reason in such a way that they conclude their behaviour is not in conflict with any moral values. This may ultimately lead to unethical behaviour.
From the explanations above, I propose the following hypotheses:
H2: The effect of general rules is moderated by incentives. When incentives are high, general rules are less effective in decreasing unethical behaviour than when incentives are low.
METHODS
Design and Participants
This study had a 3 (general rule/specific rule/no rule) x 2 (high incentives/low incentives) design, with unethical behaviour as the dependent variable. The participants in this study were students from the University of Groningen, who participated for research points or for €8. The total number of participants in this study were 221 people (52.5% female, Mage = 21.43, SD = 2.33). The participants were randomly assigned to one of the six conditions, and did the study on a computer in a private cubicle.
Procedure
The experiment was conducted with the ultimatum bargaining game. Before the participants entered the cubicle to start the experiment, they were initially told that they were participating for a research to analyze the behaviour of individuals in a negotiation, and that extra money could be earned based on their choices in the negotiation and whether they are chosen by the computer to win the prize.
randomly chosen pair would receive a prize based on the allocation of tokens in the experiment. In low incentive conditions, participants were told that the value of one token was €0.1. In high incentive conditions, they were told that the value for one token was €1.
All the participants played as the allocators, whose main job was to allocate the given tokens between themselves and the decider. They were told that the decider could accept or decline the proposed allocation, with a declined proposal resulting in the breakdown of the negotiation, and both parties would lose the right to receive the prize. The participants were also explained that the number of tokens are randomly generated by the computer, ranging from 10 to 100, which would be revealed later on in the experiment.
The next step in this experiment was the explanation of the rule of the negotiation. In this experiment, participants had the chance to lie about the number of tokens. This will be explained further on. Participants in the general rule conditions were told to “do what is morally right”, while in the specific rule conditions, they were told “tell the truth about the total number of tokens available”. Participants in the no rules condition were not exposed to any rules. The participants were then informed that only the allocators would know the exact amount, and that the deciders would only know the range of the tokens held (10 to 100). Furthermore, they were explained that they would be given the possibility to, besides making the proposal, inform the decider about the total amount of tokens to divide. Allocators would be given freedom in informing the number of tokens available to divide, or in other words, were given the chance to lie about the number of tokens.
the high incentives condition had the chance to win up to €96, and in the low incentives
condition the potential prize is up to €9.60. After that, they were told to allocate the tokens with the decider. The participants needed to type in a message to the decider containing how many tokens are available to divide and also the proposed amount of tokens to be given to the decider. In the end of the experiment, they were debriefed about the study and were given either research points or €8 for their participation.
Measures
Unethical behaviour. Unethical behavior was operationalized as lying in the participant’s
message to the decider on how many tokens were available to divide. I decided to make this into a continuous and a dichotomous variable. For the continuous variable, the participant’s degree of lying was measured with the difference between the number of tokens that were allocated to the participants, which is 96 tokens, and number of tokens that the participants revealed to the decider. The higher the difference, the more the participants lie. This continuous variable was made with the idea that lying a little (saying that there are 80 tokens to divide) is less unethical than lying a lot (saying that there are 20 tokens to divide).
Manipulation check incentives. A manipulation check was performed to see whether
participants in the high incentive condition perceived that their potential incentive is higher than participants in the low incentive condition. This was measured with 7 point Likert scales (1= Totally disagree and 7 = Totally agree) by four items: “The money I was playing for in this negotiation game was a lot of money”, “In the negotiation game, there was much money at stake”, “There was a lot to gain by lying about the number of tokens to person B”, and “It was very tempting to lie about the number of tokens” (Cronbach’s alpha = 0.78).
Manipulation check rules. A manipulation check was performed to see whether
participants in the specific rule condition perceived that the rule is more specific than participants in the general rule condition, and vice versa. The participants were first shown a statement: “In real life, some rules are more specific than others: they clearly specify what behaviors will not be tolerated. Some rules are more general than others: they provide general guidelines for behavior”. They were then asked to rate the general rule, “do what is morally right”, and the specific rule, “tell the truth about the total number of tokens available”, on a 7 point Likert scale (1 = very specific and 7 = very general).
Moral norms. Moral norms were measured in order to find out whether the general rule
Propensity to Morally disengage. This was measured with 7 point Likert scales (1=
Totally disagree and 7 = Totally agree) by fifteen items based on the measurement of moral judgement by Duffy et al. (2008). Examples of the items in the measure include: “People who are mistreated at work have usually done things to deserve it” and “It's okay to treat someone poorly if they behaved like a 'worm'” (Cronbach alpha = 0.84). This mainly measures an
individual’s tendency to morally disengage. An individual’s tendency to morally disengage was measured because it could be an additional moderator, as according to literature, the higher the moral disengagement score, the more likely an individual would engage in unethical acts.
Moral rationalization. Moral rationalization was measured with 7 point Likert scales (1=
Totally disagree and 7 = Totally agree) by six items: “It is okay to lie about the number of tokens because the other person also gets a share of the tokens anyway”, “Telling a white lie is 'part of the game”, “The experimenters provide an opportunity to lie, which makes it okay to lie about the number of tokens available”, “The experimenters are responsible for persons being lied to in this experiment”, “Person B is not really harmed if person A lies about the total number of tokens”, and “There are much worse things than lying a little about the number of tokens available” (Cronbach’s alpha = 0.79).
Participant suspicion. In the end of the experiment, participants were asked with an open
RESULTS
Participants
In the result analysis, eight participants were excluded because they put a value of zero on their message to the decider. This was done because writing a zero in the message was not an option, as the range of the number of tokens in the experiment was 10-100, which suggests that these people has not understood the game or were trying to communicate that they did not want to reveal the number of tokens at all. Furthermore, 47 participants were marked as suspicious. However, they were included in the analysis considering that the result between including and excluding them in the analysis showed similar patterns1.
Manipulation Checks
Rules. This analysis was done to check whether the general rule was perceived as more
general than the specific rule, and vice versa. To do this, the rule manipulation check scores were compared between participants in the general and specific rule conditions. Participants in the no rule condition were excluded. The score for the manipulation check item depends on the
conditions the participants are assigned to. Participants in the general rule conditions rated the general rule, while participants in the specific rule rated the specific rule. Using one way
1 The analysis excluding suspicious participants had the same pattern as the analysis that includes them. For the
continuous dependent variable, in the 2x3 ANOVA, only rules had a significant effect on lying, F (2, 162) = 3.12, p = 0.05, while the incentives and the interaction variable was not significant. For the dichotomous dependent variable,
the pattern was also the same, as in the general rules analysis, there was a significant three way interaction χ2 (1) =
ANOVA, I found that participants in the general rule conditions perceived that their rule was more general than the specific rule, F (1, 139) = 188.40, p < 0.01.
Incentives. Two-way ANOVA was used to see whether the incentive conditions have an
effect on the manipulation. In the analysis, I found a significant effect of the size of incentives on the incentive manipulation check score F (1, 207) = 110.30, p < 0.01. Participants in the high incentive condition scored significantly higher than their counterparts in the low incentive
condition. This result confirms that the high incentives condition were viewed as more rewarding than the low incentive condition.
Correlations
Table 1 shows the correlation between variables analyzed in this study. The continuous dependent variable, the degree of lying, significantly correlated with the dichotomous dependent variable, which is the decision to lie (r=.82, p<.01). General rules significantly correlated with moral norms (r=.18, p<.01), while specific rules do not. No other manipulations showed any correlations with other variables tested.
Table 1 Correlations Matrix Variables Mean S.D. 1 2 3 4 5 6 7 8 1. Degree of lying 19.28 23.56 .82** 2. Decision to lie - - .82** -.05 3. Specific rule - - -.06 -.05 -.50** 4. General rule - - -.12 -.080 -.50** -.007 5. Incentives - - -.05 -.04 .004 -.007 .01 6. Moral rationalization 4.18 1.29 .58** .65** .07 -.11 .01 -.71** 7. Moral norm 3.70 1.24 -.56** -.66** -.04 .18** -.04 -.71** -.25** 8. Propensity to morally disengage 3.33 .79 .18** .21** -.02 -.11 .03 .37** -.25**
Continuous Dependent Variable Hypothesis Testing
A 2 (incentives) x 3 (rule) ANOVA on the degree of lying and t tests were performed. ANOVA revealed a significant main effect of the type of rules, on the participants’ degree of lying, F (2, 207) = 3.40, p = 0.04, η2= 0.03. Based on the Tukey HSD post-hoc test, it was shown that the participants’ degree of lying was significantly higher when there were no rules compared to when general rules were used, p = 0.01. There was also a significant difference between the no rule and the specific rule condition, p = 0.05, where participants in the specific rule condition generally scored lower. There was no statistically significant difference between the general rule and specific rule conditions. The means for the degree of lying are shown in table 2.
In the same analysis, a significant main effect of incentives on the degree of lying was not found, F (1, 207) = 0.63, p = 0.43, η2= 0.003. Furthermore, the two-way ANOVA did not reveal an interaction between the type of rules and incentives, F (2, 207) = 1.20, p = 0.30, η2= 0.01.
General rules (H1). To test H1, it needs to be proven that the general rule was more
Although the t tests showed that general rules are effective when incentives are low but less so when incentives are high, the ANOVA revealed that the interaction was not significant. For this reason, H1 is rejected.
Specific rules (H2). H2 states that specific rules are effective in both incentive conditions.
To confirm this hypothesis, the specific rule conditions need to have significantly lower means of lying than the no rule groups when incentives are high and low. First, independent sample t tests were conducted. It was found that when incentives were low, subjects in the specific rule condition lied significantly less compared to individuals in the no rule t (72) = 2.39, p = 0.02. In the high incentive condition, the means in the two conditions were not significantly different t (68) = 0.30, p = 0.76, signaling that the specific rule was not effective in decreasing lying. Second, similar to the general rule testing, a 2 (specific and no rules) x 2 (high and low incentives) ANOVA was conducted. The result showed that the interaction between rules and incentives was not significant F (1, 140) = 2.05, p = 0.15, η2= 0.01.
Although the interaction variable was not significant, which was what expected if H2 was
confirmed, the t tests did not show that the specific rules were effective in both incentive conditions. The specific rule was only effective when incentives were low, and for this reason, H2
is rejected.
Mean differences within rule conditions. To further explain the results, independent
Table 2
Mean of Participant’s Degree of Lying
Specific Rule No Rule General Rule
Low Incentive 16.19 29.57 15.58
High Incentive 18.54 20.31 14.91
Dichotomous Dependent Variable Hypothesis Testing
General rules (H1). The effectiveness of the general rule was tested by contrasting the
dependent variable from the general and the no rule conditions. In this test, a log linear model was used to analyze whether there are any interactions between the type of rules and incentives on the participants’ decision to lie. The three way log linear analysis showed that in the final model, all effects were retained. The model had a likelihood ratio of χ2 (0) = 0, p = 1. This result indicates that the highest order interaction (rules x incentive x decision to lie) was significant, χ2 (1) = 4.31, p = 0.04. Using parameter estimates, it was found that the effect size of the three way interaction on the dependent variable was z = -2.03 (p = 0.04).
To look at the effects of each variable, separate chi square tests were conducted. The percentage of participants that lied about the number of tokens in each condition is shown in Table 3. In the high incentive condition, there was no statistically significant difference in the percentage of people who lied between the no rule and general rule conditions, χ2 (1) = 0.06, p = 0.80. In the low incentive condition, there was a significant relationship between the type of rules and participant’s decision to lie, χ2 (1) = 7,25, p = 0.007. Odds ratios showed that the odds that
condition. Thus, the findings confirm the first hypothesis, as it was statistically proven that the general rule was more effective in the low incentive compared to the high incentive condition.
Specific rules (H2). A log linear model was also used to test the second hypothesis. In
this test, only participants who were conditioned with no rules and specific rules were included, because the goal of this hypotheses is to test the effectiveness of specific rules. Using the log linear analysis, it was revealed that there were no significant interactions in the first, second, or third order. To further confirm this, chi square tests in both the low and high incentive condition were carried out. In the low incentive condition, the percentage of lying participants in the specific condition were less compared to the no rule condition, but the difference was marginally significant, χ2 (1) = 3.59, p = 0.06. Odds Ratio suggest that, when offered low incentives, the
odds that participants lie were 2.5 times lower when they were exposed to specific rules
compared to when they were not shown any rules. In the high incentive condition, the percentage of lying participants was not significantly different between the two rule conditions, χ2 (1) = 0.06, p = 0.81.
These results mean that reject the second hypotheses is rejected. Although there were no signs that the type of rules and incentive interact, to be able to confirm this hypothesis, the percentage of people lying in the specific rule condition should be significantly less compared to the no rules group, in both incentive conditions. Instead, the specific rule was only effective in the low incentive condition.
Percentage differences within rule conditions. Chi square tests were done to see
condition χ2 (1) = 0.06, p = 0.80. Secondly, in the general rule condition, there was also no significant difference between the percentages in the low and high incentive condition χ2 (1) =
1.11, p = 0.29. In the control group, the percentage of lying was marginally different between the incentive conditions χ2 (1) = 3.52, p = 0.06, with higher percentages in the low incentive
condition.
Table 3
Percentage of participants that lied in each condition
Specific Rule No Rule General Rule
Low Incentive 48.60% 70.30% 38.90%
High Incentive 45.70% 48.60% 51.50%
Moral Norms
2x3 ANOVA was used to find out whether the rules and incentives affect the
participant’s moral norm. The result showed that there was a significant effect of the type of rules on the participant’s moral norms F (2, 207) = 3.96, p = 0.02. However, there were no significant effects of incentives, F (1, 207) = 0.28, p = 0.60, and the interaction between rules and incentives F (2, 207) = 2.08, p = 0.13. The LSD post hoc test revealed that the mean score of moral norms in the general rule condition (M = 4.03, SE = 0.15) was significantly higher than the no rule condition (M = 3.46, SE = 0.14), p = 0.01 and higher than the specific rule condition (M = 3.63, SE = 0.14), although the latter comparison was marginally significant (p = 0.06). There was no observed difference between the no rule and the specific rule conditions. Moral Disengagement
Continuous dependent variable. A regression analysis was done to see whether there
independent variables (rules and incentives) that affects the degree of lying. The analysis were separated into two parts, based on the hypothesis testing: analysis between the no rule and
general rule condition, and between no rule and specific rule condition. The rules, incentives, and moral disengagement variables were centralized before calculating interaction terms. First, a multiple regression was done using the no rule and general rule condition as the “rules” variable. In the first order, all the main effects were included, which are rules, incentive, and moral
disengagement, in the model, which results in R2 = 0.12, p= 0.00. In the second step, the two
way interactions (rules x incentive, rules x moral disengagement, and incentive x moral
disengagement) were added, and it was found that there was a significant increase of R2, ΔR2 = 0.08, p= 0.00. The interaction variable between moral disengagement and incentives were significant (B= 6.12, SE= 1.81, β= 0.27, p=0.01). The interaction showed that while moral disengagement was positively related the degree of lying when incentives were low, high
incentives seemed to suppress lying in high moral disengagers. The interaction graph can be seen in figure 1. Three way interaction was also tested in the third order, but it was found that the three way interaction variable was not significant, p=0.51.
seen in figure 2. Again, three way interaction was also checked, but it was found that the interaction was not significant p = 0.24.
FIGURE 1
Incentives and Moral Disengagement Interaction (Continous Variable)
FIGURE 2
Specific Rule and Moral Disengagement Interaction (Continuous Variable)
Low Incentive High Incentive
Dichotomous dependent variable. A logistic regression was done to see whether there
are any interactions between the main independent variables and moral disengagement that could affect people’s decision to lie. Similar to the analysis on the continuous variable, the analysis was divided into two parts based on the hypothesis testing. The independent variables was also centralized. In the first analysis, the no rule and general rule conditions were used as the “rules” variable. In the first order, all the main effects variables were included in the regression, which resulted in R2 = 0.13, p = 0.002. In the second order, two ways interaction variables were added,
and found that there was an increase in R2, ΔR2 = 0.22, p = 0.00. It was discovered that the interaction variable between incentives and moral disengagement was significant (B= -1.30, SE= 0.34, p=0.00). The interaction between incentive and moral disengagement is similar to the interaction in the continuous variable analysis, and can be seen in figure 3. There was no
evidence that the three way interactions between incentives, rules, and moral disengagement was significant p = 0.49.
Second, an analysis was done using the specific rule and no rule conditions. In the first order, all the main effects were included, and the model had R2 = 0.09, p = 0.02. In the second
order, two way interactions were added, which resulted in R2 = 0.16, p = 0.007. Then, in the third order, the main effects, the two way interactions and the three way interactions between rules, incentives, and moral disengagement were included, R2 = 0.23, p < 0.001. In the third order regression, the interaction variable between rules and moral disengagement was significant (B= -0.91, SE= 0.34, p=0.01). Furthermore, it was revealed that the three way interaction
in individuals with higher moral disengagement were decreased. The three way interaction can be seen in figure 4.
FIGURE 3
Incentives and Moral Disengagement Interaction (Dichotomous Variable)
FIGURE 4
Incentives, Rules, and Moral Disengagement Interaction (Dichotomous Variable) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Low Moral disengagement High Moral disengagement
P rob ab il ity of L yin g Low incentives High incentives 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Low Moral disengagement High Moral disengagement
P rob ab il ity of L yin g
Moral rationalization
First, a 2 x3 ANOVA was done to see the effects of the manipulations on the moral rationalization score. It was revealed that the rules did not significantly affect rationalization, F (2, 207) = 1.43, p = 0.24, and the same was true for the incentive manipulation F (1, 207) = 0.02, p = 0.88. Next, One way ANOVA was used to see the effectiveness of rules in reducing moral rationalization. In the low incentive condition, there was no indication that there were significant differences in moral rationalization scores between the rule conditions F (2, 107) = 1.89, p = 0.16. However, it is important to note that participants shown the general rule did less moral rationalization than their counterparts in the no rule condition, though the difference was marginal p = 0.07. In the high incentive condition, the differences in moral rationalization between the rule conditions were marginal F (2, 100) = 2.75, p = 0.07. It was revealed that participants in the specific rule condition rationalized more than others in the no rule condition p = 0.04 and general rule condition p = 0.06. There was no significant difference between the general and no rule conditions.
Table 4
Mean of Participant’s Moral Rationalization
Specific Rule No Rule General Rule
Low Incentive 4.05 4.51 3.92
High Incentive 4.57 3.98 4.01
DISCUSSION
In this study, I found that, overall, both the general and specific rule were able to reduce lying. However, the main question in this study is whether the potency of the rules depend on incentives. I argued that general rules would be less effective when the stake for unethical behaviour is high rather than low. In addition, I also stated that the effectiveness of a specific rule is independent of the incentive size. I found evidence that when incentives were low, both the specific and the general rule were effective in decreasing lying. When the reward for lying was low, the general rule “do what is morally right” was equally effective in reducing
participants’ lies as the specific rule “tell the truth about the total number of tokens available”. Although the general rule does not explicitly tell the participants not to lie, telling the truth is a part of doing what is morally right. This is in line with the value based view of Wiever and Trevino (1999), as the general rule communicates a broader moral value compared to the specific rule. Indeed, this was supported by the current result as participants exposed to the general rule had higher moral norms than individuals who were shown the specific rule.
When incentives were high, however, both rules were not able to reduce lying. This would mean that the general rule was less effective when the reward for unethical behaviour was high compared to when it was low, which would support my initial argument for the first
high. This finding would not confirm the second hypotheses, as I claimed that a specific rule would be effective independent of the incentive size. However, after closer inspection, I believe that the effectiveness of both rules in this study when the incentive was high can be explained mainly by the fact that when there was no rule, participants who were offered a higher reward actually lied less than initially expected.
When the experiment was designed, it was assumed that without any rules, participants would lie more when the incentive to do so is high. The basic assumption of this study was based on the argument that higher incentives induce individuals to morally rationalize (Kish-Gephart et al., 2014) and subsequently, act unethically (e.g. Gino & Mogilner, 2014; Kouchaki et al., 2013). Instead, the opposite was true, as participants offered a higher incentive rationalized and lied less than their counterparts offered a lower reward.
Thus, it can be concluded that the principal argument of this research was not supported. Higher incentives did not encourage individuals to morally rationalize, and subsequently,
motivate unethical behaviour. Because of this, I believe that I cannot establish the effect of rules in the high incentive conditions. This implies that I cannot confidently answer whether the general rule’s effectiveness depends on the size of incentives. The same also applies to the specific rule, where I also cannot conclude that its effectiveness is independent from the reward size. However, I do believe that the effect of the rules in low incentive situations still pertains; both rules were able to reduce unethical behaviour.
immoral behaviour through rationalizations, but they will limit the amount of cheating because they still need to retain positive views of themselves, which is the internal reward they receive by being ethical. In their study, they found that when an individual has the opportunity to cheat, they will do so, but only with a low magnitude. This logic can also explain why participants lie less in this study. It can be argued that participants lying in high incentive conditions will have a more negative view of themselves compared to participants who lie in the low incentive
conditions. Although they have a potentially high external reward if they lie, they still need to maintain their self-worth, and to do that, they either lie less or even not lie at all.
Secondly, in addition to the individual self-worth explanation, the design of the
experiment may have contributed to the reason why participants in the high incentive conditions behaved differently from my initial expectation. The experiment is an ultimatum bargaining game where the participants were led to believe that they are playing with another person.
an immoral behaviour, the perpetrator would find it harder to ignore the harm caused, which decreased moral rationalizations related to dismissing harm. Using the same logic that the participants may have viewed that lying is more harmful in high reward situations, this may also explain why subjects did not rationalize more when the incentive was increased.
Earlier, it was established that this study cannot answer whether the effectiveness of general rules depend on incentives because of the behaviour of participants in the high incentive condition. The two issues that may contribute to answering the question why participants lied less when the reward was high are stated above; self-concept maintenance and the experiment design. Therefore, it is logical to imply that future research attempting to answer the same question as this study should address these issues. Of course, it is not possible to alter the mechanism individuals use to maintain positive self-view, as it is out of a researcher’s control. However, one aspect that future studies can improve on is the experimental design.
This study made the harm done to the other party very visible, which other than it is one of the reasons why participants lie and rationalize less, I believe is also a limitation of this research. The reason is because the result of the study does not apply to situations where the consequence of an unethical behaviour is not so noticeable. An example of such situation is a conflict of interest, embezzlement, and fraud in science, where the harm incurred towards the other party is not as clear. Future research should study the effect of incentives on the
unethical. For this reason, I think that reducing the visibility of harm may more likely help answer the question whether the effectiveness of rules depend on incentives.
Moral Disengagement
In the analysis, I discovered one variable that interacts with rules and incentives, namely the participants’ tendency to morally disengage. In the theory section, I focused on moral disengagement as a process that leads to unethical behaviour. However, moral disengagement can also be measured as a personal tendency (e.g. Bandura, 1999, 2002; Moore, 2008). Some people are more likely to morally disengage than others. Individuals with high moral
disengagement have been shown to be more likely to engage in unethical behaviour (Moore, 2008).
Practical Implications
The study has several practical contributions. First, I found that when the stakes of doing an unethical act are low, both general and specific rules are enough to suppress it. Therefore, when aiming to prevent petty unethical acts, choosing either a general or specific rule should equally be able to decrease it. However, when the stakes for unethical behaviour are high, it is still unclear whether general and specific rules are equally effective in reducing them.
Second, I also found that when the consequence of the unethical act is visible to the perpetrator, unethical behaviour is lessened. Thus, to prevent unethical behaviour, the visibility of harm should be increased, and one way of doing this is by communicating the consequence of the crime towards the victim. An example of this is a TV advertisement campaign against speeding that shows an innocent driver and his son being a victim of a car crash caused by a reckless driver. In organizations, explaining the consequences of the unethical behavior instead of just stating the rule that forbids them may make harm more visible. For example, instead of only stating “any form of discrimination is prohibited”, companies can also communicate how discrimination negatively affects the minority.
Limitations and Future Research
for social rewards in influencing rule effectiveness should be similar to financial incentives. Higher social rewards for unethical behaviour should induce moral rationalization, and in this situation, general rules should be less effective compared to when the reward is low. Meanwhile, specific rules should be effective independent of the social reward magnitude.
Second, in the result section, it was argued that suspicious participants were included because the statistical result by excluding them was not very different. However, in hindsight, it does not prove that participants who were not suspicious acted more “naturally” in the
experiment. A known issue for studies concerning ethics, the use of laboratory experiments in particular, is subjects giving socially desirable answers, which may be caused by evaluation apprehension. The subjects in the current study participated knowing that they are in an
academic experiment, and thus may feel that they were being observed. This is important to note since previous studies suggest that, individuals who feel supervised tends to give socially
desirable answers (Trevino, 1992) and cheat less (Covey, Saladin, & Killen, 1989) because the desire not to be seen in negative light. Thus, there is a reason to be cautious that the artificial setting of the experiment may have affected the participants’ decisions to lie. Future research should attempt a field study that reflects more “natural” situations where the supervision of the experimenters is not as obvious as in laboratory studies in order to decrease the likeliness that subjects give socially desirable responses.
REFERENCES
Arjoon, S. (2006). Striking a balance between rules and principles-based approaches for effective governance: a risks-based approach. Journal of Business Ethics, 68(1), 53-82.
Bandura, A. (1991). Social cognitive theory of self-regulation. Organizational Behavior and
Human Decision Processes, 50, 248-287.
Bandura, A. (1999). Moral disengagement in the perpetration of inhumanities. Personality and
Social Psychology Review, 3, 193-209.
Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Mechanisms of moral disengagement in the exercise of moral agency. Journal of Personality and Social
Psychology, 71, 364-374.
Baron, R., Zhao, H., & Miao, Q. (2014). Personal motives, moral disengagement, and unethical decisions by entrepreneurs: Cognitive mechanisms on the “slippery slope”. Journal of
Business Ethics, 128, 107-118.
Barsky, A. (2011). Investigating the effects of moral disengagement and participation on unethical work behavior. Journal of Business Ethics, 104(1), 59-75.
Detert, J., Treviño, L., & Sweitzer, V. (2008). Moral disengagement in ethical decision making: a study of antecedents and outcomes. Journal of Applied Psychology, 93(2), 374-391. Covey, M. K., Saladin, S., & Killen, P. J. (1989). Self-monitoring, surveillance, and incentive
effects on cheating. The Journal of Social Psychology, 129(5), 673-679.
Galbiati, R., & Vertova, P. (2014). How laws affect behavior: Obligations, incentives and cooperative behavior. International Review of Law and Economics, 38, 48-57. Gneezy, U. (2005). Deception: The role of consequences. The American Economic
Review, 95(1), 384-394.
Gates, J. B. (2004). The ethics commitment process: sustainability through value-based ethics.
Business and Society Review, 109(4), 493-505.
Gino, F., & Mogilner, C. (2013). Time, money, and morality. Psychological Science, 25(2), 414-421.
Harris, J., & Bromiley, P. (2007). Incentives to cheat: The influence of executive compensation and firm performance on financial misrepresentation. Organization Science, 18(3), 350-367.
Kish-Gephart, J., Detert, J., Treviño, L., Baker, V., & Martin, S. (2014). Situational moral disengagement: can the effects of self-interest be mitigated? Journal of Business
Ethics, 125(2), 267-285.
Kouchaki, M., Smith-Crowe, K., Brief, A., & Sousa, C. (2013). Seeing green: mere exposure to money triggers a business decision frame and unethical outcomes. Organizational
Behavior and Human Decision Processes, 121, 53-61.
Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108(3), 480-498. Mahoney, P.G., & C. W. Sanchirico (2005). General and specific legal rules. Journal of
Institutional and Theoretical Economics, 161, 329–346.
Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: a theory of self-concept maintenance. Journal of Marketing Research, 45, 633-644.
Merriam-Webster. (2004). Merriam-Webster's collegiate dictionary. Merriam-Webster. Mulder, L. B. (2008). The difference between punishments and rewards in fostering moral
concerns in social decision making. Journal of Experimental Social Psychology, 44(6), 1436-1443.
Mulder, L. B., Jordan, J., & Rink, F. (2015). The effect of specific and general rules and ethical decisions. Organizational Behavior and Human Decision Process, 128, 115-129. Paine, L. S. (1994). Managing for organizational integrity. Harvard Business Review, 72(2):
106-117
Pascual-Ezama, D., Dunfield, D., Gil-Gómez de Liaño, B., & Prelec, D. (2015). Peer effects in unethical behavior: standing or reputation?. PLOS ONE, 10(4).
Pascual-Ezama, D., Prelec, D., & Dunfield, D. (2013). Motivation, money, prestige and cheats.
Journal of Economic Behavior and Organization, 93, 367-373.
Tenbrunsel, A. E., Wade-Benzoni, K. A., Messick, D. M., & Bazerman, M. H. (2000). Understanding the influence of environmental standards on judgments and choices.
Academy of Management Journal, 43(5), 854-866.
Trevino, L. K. (1992). Experimental approaches to studying ethical-unethical behavior in organizations. Business Ethics Quarterly, 2(02), 121-136.
Tsang, J.A. (2002). Moral rationalization and the integration of situational factors and
psychological processes in immoral behavior. Review of General Psychology, 6(1), 25-50. Weaver, G., & Trevino, L. (1999). Compliance and values oriented ethics programs: influences
Woozley, A. D. (1960). [Review of the book The Definition of Law, by H. Kantorowicz]. The Philosophical Quarterly, 10(38), 95–96.
