Accrediting the good to prevent the bad : preventing moral licensing through positive attribution

(1)

Faculty of Economics and Business

Master Thesis

Accrediting the good to prevent the bad

Preventing moral licensing through positive attribution

Titus van Dijk 10087745

MSc Economics

Track: Behavioral Economics & Game Theory

Supervisor: Arthur Schram 26/07/2015

(2)

Introduction

When people act in a way they perceive as ‘good’, they often reward themselves for this good behavior with subsequent more dubious behavior. We are more likely to eat a hamburger after going to the gym, drink beers after a long week of hard work and eat less fruit after taking a vitamin pill.

These are conscious decisions in which people reward themselves with questionable behavior after acting in a way that they perceive as good. Although some of these decisions, where we reward our good behavior, are clearly identifiable and consciously taken, there are numerous moral acts that people perform daily without realizing it. These acts are remembered in our subconscious and possibly influence our following behavior. Moreover, people can use the memory of their moral deeds to rectify subsequent immoral actions. This phenomenon is called moral licensing. Merritt, Effron and Monin (p.344, 2010) define moral licensing as; “Past good deeds can liberate individuals to engage in behaviors that are immoral, unethical, or otherwise problematic, behaviors that they would otherwise avoid for fear of feeling or appearing

immoral.” Gneezy et al. (2014) build on this by arguing that people tend to be inconsistent in their behavior after performing a moral deed, which results in a greater likelihood of acting immorally.

This irrational behavior benefits neither the person in question nor society in general. Society benefits from moral behavior and people prefer to be consistent in their behavior. For these reasons it may be considered valuable to prevent moral licensing. This study will examine whether it is possible to prevent moral licensing by one specific intervention, to wit, positive attribution (to be discussed below). Specifically, this study tests the following research question: Can positive attribution prevent moral licensing?

There seems to be a consensus in the literature that moral licensing exists.

However, not all studies show significant evidence of moral licensing. Blanken, Van de Ven and Zeelenberg (2015) show with their meta-analysis that there is a very small but significant effect of moral licensing. They refer to numerous studies that do not find any effect at all. In addition to this, results from studies that do find an effect of moral licensing have not been replicated,

(4)

intentionally inconsistent in their behavior opposes various studies which suggest people prefer to be consistent in their behavior (Young, Chakroff and Tom, 2012). This contradiction suggests that there are some unresolved issues in our understanding of moral licensing. As Blanken, Van de Ven and Zeelenberg (p.540, 2015) state “the magnitude of the effect and the specific

conditions under which moral licensing is likely to occur remain unclear.” This leaves us with a question as to why in some cases people consistently stick to their behavior and in other cases make use of moral licensing to adjust their behavior.

One factor that could play a role in influencing a change from inconsistent moral

licensing behavior to consistent behavior is positive attribution. Being positively attributed for a previous good deed (e.g., by being told that this deed reflects a positive moral character) could give someone a feeling of acknowledgement, which can lead to the belief that they are moral and should therefore act accordingly. Previous literature has shown that positive attribution can be used as a tool to nudge people into doing something (Miller, Brickman and Bolen, 1975; Kraut, 1973). Positive attribution creates the feeling that you want to do something (because you are moral) instead of the feeling that you have to do something that you naturally would not do (because you need to be moral). This alternative way of motivating people can be more effective than telling someone what to do. It is not only the act of motivating people to be rational that has an effect, but the way people are motivated to act morally also has an effect (Clot, Grolleau and Ibanez, 2014). Studies show that telling people to act morally has a smaller effect than when people are nudged into a situation (Miller, Brickman and Bolen, 1975; Kraut, 1973, Clot, Grolleau and Ibanez, 2014, Thaler and Sunstein, 2003). Through the act of nudging, positive attribution can conceivably be the link that keeps people consistent with their moral behavior, preventing them from falling prey to moral licensing.

There are several reasons why it is important to probe the phenomenon of moral licensing and look at the influence of positive attribution. Firstly, the existing literature is contradictory, as there seems to be evidence of both consistent moral behaviors as well as inconsistent moral licensing behavior. Therefore, already there seems to be an uncomfortable consensus that moral licensing exits. However the parameters of this consensus remain unclear. This leads to an ambiguity in terms of our ability to identify when people are vulnerable to moral licensing. From a behavioral perspective it is important to see how people (subconsciously or consciously) react in certain circumstances. For example, why, and under which circumstances, are people more

(5)

likely to buy ecologically friendly food, donate to charity or do volunteer work, and how can their subsequent behavior remain consistent with their preceding moral deeds. Hence, if positive attribution does rectify moral licensing, it can be used to nudge people into more moral behavior. This would correspond to Thaler and Sunstein’s (2003) idea of ‘libertarian paternalism’,

whereby choice is influenced in such a way as to make people better off without taking away the their freedom to choose. This would imply that positive attribution could be used at all levels of society to improve human behavior.

Literature review

As noted above, the circumstances under which moral licensing occurs is still very much up for discussion and the ongoing debate surrounding this subject has produced a diverse collection of literature. The different studies of moral licensing can be divided into several categories. I will begin by discussing two ways of self-licensing; moral crediting and moral credentialing. These concepts will be explained through findings from past studies. Secondly I will elaborate on the uses of moral licensing and the different domains in which moral licensing occurs. I use these domains to give a better insight into the kind of situations where people license themselves and why it is relevant to study the phenomenon of moral licensing. To summarize the differences in moral licensing, a concise overview of the existing studies and the categories and domains they belong to is displayed in table 1, below. The counterpart of moral licensing is moral cleansing, which is a theory that suggests that doing something morally right after acting morally wrong, ‘cleanses’ your wrongdoings. To clarify and strengthen certain arguments, some examples of moral cleansing will be used. This study therefore aims to explore moral licensing using the points outlined above as routes of exploration.

Moral crediting

Remembering good deeds can cause people to act less morally afterwards (Khan and Dhar, 2006; Jordan, Mullen and Murnighan, 2009). Remembering a good deed allows people to credit that moral action and makes them feel ‘licensed’ to ‘spend’ these moral credits on an immoral act. This concept is called moral crediting and is defined as a moral deed giving credits for doing an

(6)

immoral deed in the future (Merritt, Effron and Monin, 2010). Moral crediting is the illustrative term for situations in which people are aware that they are doing something immoral but feel themselves licensed to do so because of their previously ‘gained credits’. For example, people are more likely to cheat during a test while thinking of an altruistic deed they have previously undertaken (Jordan, Mullen and Murnighan, 2009). Therefore, to continue with this analogy, whilst cheating is unmistakably seen as an immoral deed, the thought underlying this action is linked to moral crediting as people feel they have earned the right to act less morally because of their previous moral actions. It is this sense of awareness within immoral actions that is an essential aspect of moral crediting.

Aside from the experience of morally licensing after a moral act or memory, people also tend to anticipate the need for moral credits before acting immorally. In Merritt et al. (2012), subjects were told that they had to hire a black or white employee in 24 hours. 24 hours prior to the hiring subjects were more likely to give morally correct answers to questions. In this way the subjects ‘protected’ themselves for possible ‘immoral’ actions they had to perform in the future (hiring a white person over a black person). This shows that the human mind creatively

anticipates external expectations and also illustrates how moral crediting is used in different forms. Still, for all the ways people morally credit themselves, circumstance and context still offer spaces for variation. For instance, remembering a moral trait has less effect on moral licensing than the memory of a moral action, and remembering distant moral behavior has less effect than the memory of recent moral behavior (Conway and Peetz, 2012). The reasoning behind this is that moral traits and distant moral behavior make people believe they are moral people and should (consistently) act accordingly. People who remember a recent moral action signal to themselves that they have done something moral (i.e. actually ‘earned’ moral credits). They show themselves they have done something moral which allows them to characterize subsequent self-interested actions as less harmful.

People also morally credit themselves while acting morally. Retrospectively people are aware of having acted immorally, however this feeling is adjusted in relation to the moral credits they have earned. Still, people draw on these ‘moral credits’ to different extents, deciding either to use or not use them, which exposes the possibilities to nudge people into showing consistent behavior where they are not using moral licensing.

(7)

Moral credentialing

A person’s awareness of acting immorally is the main difference between moral crediting and moral credentialing. When performing moral credentialing, people do not experience their following action as immoral (despite it being questionable) because their previous moral action distorts their view of the subsequent action. Therefore the immoral deed is no longer seen as immoral. As with moral crediting, the awareness of a link between acting morally and immorally is absent. For example, people are more likely to make immoral decisions after rejecting immoral decisions (Monin and Miller, 2001). It is at this point that the perspective on good and bad

changes. When, for instance, making a racist comment, people perceive it as bad. However if, prior to making the comment, a person states that he or she voted for Obama, the perception of the racist comment changes in the eyes of this person (Effron, Cameron, and Monin, 2009). Likewise, after rejecting racial comments, people are more likely to hire a white person instead of a black person (Monin and Miller, 2001). These examples show that through moral

credentialing, people subconsciously show inconsistent behavior where they treat all people equally at first but do not treat people equally in their subsequent actions. The immoral deed is reframed as a harmless deed. In contrast, with moral crediting people definitely have the feeling that they are doing something immoral.

As with moral crediting, the extent to which people morally credential themselves depends on the context. A famous example comes from Nisan and Horenczyk (1990). In their study they discovered that when children had to assess whether the actions of other children were good or bad, their assessment depended on previous information they had received. When the children, prior to the assessment, were told that the child they had to assess was ‘well behaving’, they were more likely to assess a bad action as ‘good’. This illustrates that the perception of good and bad actions changed as a result of the information they received in advance of the study. The fact that the children were told who was ‘well behaving’ and who was not, is related to positive attribution. These topics will be discussed further in the following chapters. Nisan and

Horenczyk’s (1990) study shows that a child’s perception changed according to events prior to their assessment of the other children. They did not ‘see’ the bad act as ‘immoral’ because they were told the child performing that act was ‘well behaving’. This finding is striking for moral credentialing, where the perception of a person’s moral values change because of a prior action.

(8)

Additionally, it gives an insight into why and when people tend to show inconsistent behavior and how this can be anticipated.

Grey area

Many situations where moral licensing takes place can be divided into the two categories of moral crediting and moral credentialing. Nevertheless, numerous situations cannot be identified as either one or the other. There is a grey area in which people make use of both categories at the same time meaning that the two are not clearly separable. For example, people who purchased a more efficient, ecologically friendly washing machine went on to wash on average 6% more than before the purchase (Davis, 2008). This can point in two directions. On the one hand, by buying an eco-friendly washing machine, people credited themselves for the ‘immoral’ action of washing more clothes. On the other hand, people do not experience washing more clothes as such a sin because their washing machine is now more efficient. Another example of an indefinable mixture of moral crediting and credentialing comes from a study of Yang and Wan (2011), who suggest that people also license questionable behavior in relation to their own health. In their study, subjects who took vitamin pills were less likely to maintain a healthy lifestyle. They ate less healthily and their amount of exercise decreased. Again, this can suggest that they viewed living healthily as less important because they consumed vitamin pills but it also implies that they credited living unhealthily because of taking these pills. Ultimately, whether through crediting or credentialing, from either perspective these actions are a form of moral licensing.

Besides showing that there is an overlap between moral crediting and moral

credentialing, the effect that these ‘moral’ products have on people provides an understanding as to why people buy certain goods. This is a valuable byproduct of studying moral licensing.

Why use moral licensing

There are many situations where people use moral licensing. Before exploring a possible solution as to how to prevent this inconsistent behavior, I will shed some light on why people are

sensitive to licensing themselves, prior to acting immorally. An obvious answer would be that people are trying to decrease their feeling of guilt. Guilt primarily comes from failing to follow

(9)

through with intended behavior (Keltner and Buswell, 1997). Assuming that most people intend to act morally and adapt their behavior accordingly, failing to do so makes them feel worse. Hence, the fact that people moral license themselves can be partly because the moral act lowers the feeling of guilt. Gneezy et al. (2014) point out that the feeling of guilt that follows an immoral act does have an effect. In their experiment, the subjects donated more to charity when they had previously made an immoral choice. This can be seen as a form of moral cleansing. The authors explain their findings by stating that during the ‘period of guilt’, which immediately follows an immoral act, people become extra morally aware and adapt their social preferences to this awareness. Looking at guilt from the other side of the spectrum, this could imply that, after a moral deed has been done, people are less afraid of feeling guilty, which therefore results in inconsistent behavior. The ‘moral credits’ they just earned would rectify the immoral behavior, leading to a decreased feeling of guilt.

The definition of guilt is ‘a feeling of having committed wrong or failed in an

obligation’1. As with moral credentialing, in some situations a person does not experience an immoral action as wrong. Therefore, additionally, there has to be another reason for moral licensing. One potential reason could be appearance. People not only perform moral licensing to prevent feeling guilty, but also to appear moral (Merritt, Effron and Monin, 2010).

In Merritt et al.’s (2012) study the subjects would probably have hired the best applicant, either white or black, which is the moral thing to do. However, choosing a white person over a black person can make you appear racist. Hence, subjects were trying to protect themselves from appearing immoral (showing bias towards a white person) by giving more moral correct answers prior to the hiring, hereby anticipating a possible immoral deed. As I will explain below, in the so-called ‘consumer domain’ of moral licensing guilt plays a more active role in decision making than appearance, with the guilt a consumer may feel with the purchase of a luxury item cancelled out by a moral act. The concept of social appearance plays amore important role in the so-called ‘social domain’ of moral licensing.

Thus, moral licensing is a way of decreasing feeling or appearing immoral. Nevertheless, appearing moral may be largely influenced by self-image as opposed to by appearance towards others. Monin and Miller (study 3, 2001) show that people who displayed moral behavior, in

(10)

which they created moral credentials, still demonstrated immoral behavior even when the audience did not know about their moral credentials. This suggests that, as long as someone thinks or feels he or she is acting morally, this is enough. This corresponds to the feeling of guilt, where as long as a person does not feel guilty about it, it feels all right to do so. Nonetheless, people can feel that they are moral and appear moral without actually being so. Batson et al. (1999) conducted an experiment where people had to divide two tasks between themselves and a stranger. One task was clearly more pleasant to perform. Some subjects had to flip a coin before dividing the tasks. Despite flipping the coin, they ignored the outcome and gave the nicer task to themselves. Still, they felt they were more moral than those who did not have to flip the coin, even though they chose to disregard the coin’s decision. This shows, both, that people can perceive themselves as moral for totally irrelevant reasons, and that people can be nudged into acting a certain way by a previous action.

These, sometimes questionable, own perceptions of being moral do not detract from the way in which people are affected by the presence of others. Moreover, Mazar et al. (2008) suggest that people are still likely to cheat, as long as they do not get caught, therefore appearing immoral, and experience the cheating as beneficial and not harmful.

In conclusion, people moral license for numerous reasons. Some of those reasons seem irrelevant (e.g. flipping a coin), irrational (overvaluing social appearance) or simply to rectify the action in their own conscience. These reasons boil down to a desire to show, to themselves or others, that they are righteous people.

Different domains

As discussed in various studies, moral licensing occurs in all kinds of situations. For

simplification, these situations can be narrowed down to two domains; the consumer domain and the social domain. The consumer domain represents the situations where people use moral licensing to rectify buying a certain good (Mazar and Zhong, 2010; Davis, 2008) or the kind of moral licensing that the buying of certain goods can create (Yang and Wan, 2011). Studies on the consumer domain provide insights into consumer behavior. This information is valuable for marketers hoping to anticipate consumer behavior by effectively using information on consumer inconsistencies or, more importantly, how to make consumers consistent within particular consumer patterns.

(11)

The social domain represents the effect on social behavior that moral licensing can have. This could reflect, for example, a decrease in socially accepted behavior (Jordan, Mullen, and Murnighan, 2009; Conway and Peetz, 2012) or a decline in the fair treatment of others, such as treating people differently out of a fear of appearing immoral, or discriminating because of earlier ‘gained’ moral credits (Merritt et al., 2012; Monin and Miller, 2001). Literature on the social domain aids the understanding of why and how people act inconsistently in social

environments; research in this area can provide guidance for governmental or private institutions trying to optimize human behavior.

While I have taken the consumer and social domains, as two main categories moral licensing does not restrict itself to any specific domain, it could spillover from one domain to another. In the study of Khan and Dhar (2006) for example, subjects were more likely to choose a less moral alternative after imagining a moral deed. This less moral alternative was choosing a luxury or a necessity good. Choosing a luxury good over a necessity good is not exactly the same as a ‘moral action’ but the results are still relevant in the consumer domain as well as the social domain. The results of research on moral licensing are often useful and interpretable in different domains. After people buy green products, they are more likely to act immorally overall,

irrespective of the domain (Anthes, 2010). Likewise, Mazar and Zhong (2010) show that creating moral credits in the consumer domain can result in ‘spending’ these credits in a social domain. It is precisely this wide interpretation of results in the different domains that makes research on moral licensing so valuable.

(12)

The above discussion is summarized in table 1, which gives an overview the literature on moral licensing while categorizing the various studies on moral crediting and moral credentialing.

Table 1 - Moral licensing categories and domains

Moral crediting Moral credentialing

Social domain

- Giving morally correct answers in anticipation of a possible future immoral decision (Merritt et al., 2012).

- Moral traits and distant moral behavior have less effect on moral licensing than moral actions and recent moral behavior (Conway and Peetz, 2012).

- Remembering a good deed made people more likely to cheat (Jordan, Mullen and Murnighan, 2009).

- Remembering a good deed made people donate less to chariy funds and blood donations (Jordan, Mullen and

Murnighan, 2009; Kahn and Dhar, 2006).

- Children were more likely to assess a bad action as ‘good’ for kids that were labelled as ‘well behaving’ (Nisan and

Horenczyk, 1990).

Consumer domain

- Subjects were more likely to cheat and steal after buying ‘green products’. (Mazar and Zhong, 2010)

- Imagining something altruistic, made subjects more likely to choose a luxury good over a necessity good

(Kahn and Dhar, 2006).

- Subjects were more likely to reject moral decisions after rejecting immoral decisions (Monin and Miller, 2001).

- Subjects were more likely to make racist comments after showing support for Obama (Effron, Cameron, and Monin,

2009).

Grey area - Subjects who purchased a more efficient, ecological friendlier washing machine went on to wash on average 6% more than before the purchase (Davis, 2008).

- Subjects who took vitamin pills were less likely to maintain a healthy lifestyle (Yang and Wan,

2011)

Consistent behavior

The research on moral licensing is plentiful and diverse. Although the results vary, the main theme seems to be that people are inconsistent in their behavior when it comes to being moral or immoral. Yet, the literature also expresses doubt concerning findings on moral licensing. It has been argued that the effects of moral licensing have often been much smaller than suggested (Blanken et al., 2015) and results cannot always be reproduced (Blanken et al., 2014). In addition to these concerns, the most convincing argument challenging the existence of moral licensing are the notable number of studies on consistent behavior. A high proportion of the

(13)

consistent behavior research suggests that people prefer to show consistent behavior. These results contradict the inconsistent preferences indicated by the moral licensing. Where moral licensing would suggest that after recalling a good deed you tend to act less morally (Khan and Dhar, 2006; Jordan, Mullen, and Murnighan, 2009), the literature on consistent behavior

suggests that people are more likely to keep acting morally (Young, Chakroff, and Tom, 2012). Therefore while there is evidence that people prefer to be consistent in their behavior, consistent behavior theory is troubled by the same problems as moral licensing. Specifically, that it only takes place under certain circumstances. Nevertheless, an interesting aspect of consistent behavior is the suggestion that when people recall that they are ‘good people’, they prefer to act accordingly (instead of using it as a moral credit). This provides evidence to suspect that positive attribution may be an instrument to prevent people from moral licensing and to nudge people into behaving morally. The next section will further discuss the literature on positive attribution, focusing on positive attribution as a tool to prevent moral licensing.

Positive attribution

Positive attribution, which is also called positive cueing or social labeling, is defined by Cornelissen et al. (p.279, 2007) as “a persuasion technique that consists of providing a person with a statement about his or her personality or values (i.e., the social label) in an attempt to provoke behavior that is consistent with the label.” This idea of social labeling is a way to nudge people. Nudging is a term used to describe steering someone into an act without forbidding or imposing something but instead letting the individual make an own choice. Positive attribution is a tool to nudge someone into an act, in this case, the act of showing consistent moral behavior. In certain circumstances, nudging can be a more effective way of dictating people’s behavior than directly imposing your will. Telling people to be moral is actually less effective than nudging someone into acting morally (Clot, Grolleau, and Ibanez, 2014).

In recent decades several studies suggest that labeling a person as ‘good’, can have the effect of people acting accordingly. Kraut (1973) was one of the first to suggest that such labeling could nudge people into certain behavior. In his study, subjects were asked to donate to charity. The subjects that, after donating, were labeled as charitable were more likely to donate when they were asked a second time compared to the subjects who were not. A more widely known study on this topic comes from Brickman and Bolen (1975), who conducted an

(14)

experiment where half of a group of children were told that they were tidy and the other half were told that they should be tidy. The children that were told that they were tidy kept the classroom significantly cleaner. Brickman and Bolen (1975) argue that this is because of the positive attribution given to the children. The ‘tidy’ children were explicitly told that they were like that, while other children were implicitly told that they were not tidy, by saying that they ‘should be tidy’.

There are many more examples of positive attribution that show that it can make people behave in more helpful and social ways (Goldman et al., 1982; Strenta and Dejong, 1981; Burger and Caldwell, 2003) or even adopt more ecologically friendly behavior (Cornelissen et al., 2008). These findings suggest that positive attribution can be a powerful tool to nudge people into moral behavior. Moreover, it can be a tool to counteract some of the cognitive mistakes people tend to make while moral licensing.

Conclusion of the previous literature

All in all, the literature shows evidence of moral licensing as well as consistent behavior. Specific circumstances occasionally make people moral license themselves, but not all of the time. Positive attribution could be one plausible way to prevent moral licensing. People tend to be sensitive towards compliments on their well doings. Examples can be found in the literature showing where people are nudged into behaving morally because of the positive attribution they received. This is not strange. People are sensitive to the social pressures that society puts on them to act morally. Moreover, when people become aware that others are concerned about certain actions, they tend to be more concerned as well (White and Plous, 1995). If the appreciation by others of a moral act is outspoken, people are reminded of these social norms and the

appreciation of others. This can create the feeling that it is the correct way to behave.

Still there are some cognitive mistakes people make while trying to or believing they are consistently acting morally. First and foremost, although people like their actions to be

consistently moral they sometimes perceive their actions as moral while instead showing immoral behavior. They use moral credentialing (Merritt, Effron and Monin, 2010), which changes their perception of their moral deeds or make people act immorally (e.g. prejudice towards black people) just to prevent appearing immoral (being seen as a racist) (Merritt et al.,

(15)

2012). Additionally, people can slip into immoral behavior because they think it will not be noticed (Mazar, Amir and Ariely, 2008).

Positive attribution can help prevent such mistakes. Through positive attribution people are reminded that they are moral, and should therefore act accordingly. Besides, confirming previous actions can make a person more conscious that acting morally is definitely noticed and appreciated. Since moral licensing does not restrict itself to one domain (Miller and Effron, 2010), it does not matter if a person receives positive attribution for a moral action in one domain in order for them to keep demonstrating moral behavior in a different domain.

In conclusion, the literature shows that there is uncertainty about what can prevent moral licensing. This, together with the evidence that positive attribution can lead to consistent moral behavior, provides the motivation to address my research question on whether moral licensing can be prevented by positive attribution.

Experimental design

To study the effect of positive attribution on moral self-licensing I have set up an experiment. Specifically, it will test for immoral behavior by looking at whether people who are vocally positively attributed on their moral behavior, become less likely to cheat while taking a quiz. In this experiment all subjects will be subconsciously put into a moral act. The subjects will perform this moral act without knowing it as part of the experiment. For this, subjects needed to be nudged into a moral deed in a way they subconsciously experienced as acting morally. Following this, the subjects were put in a position where they were able to cheat in a conscious way. The domain chosen for this experiment is the social domain.

There have been many variations of experiments on moral licensing. As seen in the literature review, a commonly way of studying the moral licensing effect is by letting subjects imagine or remember certain things, such as moral deeds (Khan and Dhar, 2006), good

characteristics (Sachdeva, Iliev, and Medin, 2009), or previous behavior (Conway and Peetz, 2012). Nevertheless, I argue the results would seem more powerful if subjects actually perform a morally just action and are then presented with the opportunity to do something immoral. In the

(16)

world outside the laboratory, people are also acting (im)morally instead of only imagining so. This is why this study uses a design where subjects actually act morally and immorally.

The experiment consisted of two phases and was conducted on one person at a time. It took place on June 15 and June 16, 2015 in Amsterdam at the UvA University library on the 3rd floor.

Part I

Part I of the experiment started at the time the experimenter met the subject. The subject was not informed that the experiment had already started. The experimenter asked the subject to follow him towards the ‘experiment room’. The experimenter was carrying a stack of papers and books. While walking to the experiment room, the experimenter dropped some of the books. This was done in a clumsy way so that the subject thought it happened by accident. Because the

experimenter held the remaining books and papers, he was not in the position to pick up the books. All the subjects immediately helped the experimenter by picking up the books without being asked to do so. After the subjects picked up the books the two different treatments went into effect. The subjects, who had been randomly selected to be in the Control Group, were only told to bring the books to the room. Note that they were not thanked in any way for picking up the books. When entering the experiment room they were told to put the books on the table. Again, they were not thanked. Those subjects selected to be in the Experimental Group, were clearly thanked for picking up the books and were then asked to bring them to the room. While arriving at the room, the subjects were told to put the books on the table and afterwards were thanked again and specifically told: “you are good person”. This last sentence was spoken out in a natural way without making it seem awkward or suspicious.

Part II

After having entered the experiment room, all the subjects were asked to read the instruction sheet, which was placed on the table (see appendix 1). At the end of the instruction sheet, the subjects had to fill out three control questions to make sure that they understood the experiment. The experimenter remained in the room while this took place. When the subjects finished answering the control questions they were asked to notify the experimenter. The experimenter

(17)

then checked the control questions. Almost all of the subjects answered the control questions correctly. Those who did not answer correctly received an additional explanation from the experimenter until they filled in the control questions correctly.

After the instructions, the subjects were required to undertake the quiz on a laptop. The experimenter left the room and the subjects were told that they should notify the experimenter when finished by opening the door. The quiz consisted of two sections and was created using the online survey software and insight platform Qualtrics. The first part was made up of a short questionnaire with the purpose of obtaining control variables. Immediately after filling out this questionnaire - the subject moved onto the second section - this is where the actual quiz started. This section consisted of six questions. Afterwards one question would be selected for payout. If this question was answered correctly the earnings were 5,00 euro, otherwise 0,00. Each question had six possible answers (A to F) and an option to ‘buy’ a hint (G). By choosing a hint, the subject reduced the available options to just two. Using a hint cost the subject 0,50 euro that was deducted from their earnings (if the earnings were positive). At the end of the experiment, one of the six questions was randomly selected for payout by throwing a die. If the selected question was answered incorrectly the subjects received nothing. If the selected question was answered correctly the subjects received 5,00 euro. The total from the number of hints used was multiplied by 0,50 euro and subtracted from the 5,00 euro. For the instruction sheet and the quiz see

appendix 2 and 3.

The subjects were required to do the test on the laptop, but were also clearly instructed to fill out their answers on a paper answer sheet, placed in front of the laptop. At the end of the paper answer sheet, there was an extra question asking how many Euros should be deducted because of the hints used. The subjects were instructed that the experimenter only looked at the answers on the paper answer sheet and would deduct the number of hints used as reported on the paper answer sheet.

At the end of the experiment, the experimenter checked the answers on the paper answer sheet and, if the selected answer was correct, payed the subjects their earnings.

(18)

Hypotheses

Hypothesis 1: Subjects in the experimental group, who received positive attribution, will be less likely to cheat in the experiment than subjects in the control group.

Receiving a positive attribution should enhance your current state (Miller, Brickman and Bolen, 1975; Kraut, 1973; Goldman et al., 1982; Strenta and Dejong, 198; Burger and Caldwell, 2003) and make you believe that you are such a person (Cornelissen et al., 2008). This suggests that ‘nudging’ someone into acting morally through positive attribution, the person in question should be less likely to use moral self-licensing. Positive attribution has been shown to lead to consistent behavior, which in the case of my experiment would be acting morally by not

cheating. Therefore, in conclusion, I predict that positive attribution decreases the probability of moral licensing (hypothesis 1).

Hypothesis 2a: Subjects who are studying economics are more likely to cheat than subjects who do not study economics.

Another prediction is that economics students are more selfish and rational than other students because they have been educated in rational choice theory or because rationality is associated with a self-selection into the field of economics (implying they were more selfish and rational all along) (Carter and Irons, 1991; Kahneman, Knetsch, and Thaler, 1986). Either way, this would imply that students who are studying economics2 are more likely to act rationally than students who do not study economics. During the experiment cheating is the rational choice for selfish individuals. There is no punishment for cheating and the cheating itself can increase the payoffs significantly. Besides, there is, by design, no immaterial punishment in the form of social peer pressure because the subject is alone in the room and the experimenter only looks at the paper answer sheet.

(19)

Hypothesis 2b: Subjects who are studying economics use more hints than subjects who do not study economics.

Rational choice theory would also imply that it is rational to choose a hint for the questions in the experiment. The questions were chosen with the aim of being extremely difficult and it was therefore very unlikely that the subjects knew the correct answers3. Each individual question had six different answers, and each of these answers could be viewed as an equally plausible answer. By choosing a hint, the subjects reduced the number of possibilities from six to two. By doing this, the subjects reduced their payoff by only 10% (0,50 euro), while (assuming they randomly chose an answer) the likelihood of choosing the correct answer increased from 16,66% to 50%. This would suggest that the more rational people used more hints, and, building on hypothesis 2, this would also suggest that it is the economics students who use more hints.

Subjects

In total 41 subjects participated in the experiment. All subjects were recruited in the corridors of the library of the University of Amsterdam, where they were asked if they had ten minutes of spare time to join an experiment where they could earn 5,00 euro. One subject was excluded as an outlier because she reported more hints than she actually used, an indication that she did not fully understand the experiment4. Table 2 provides an overview of the subjects and their choices. The control group consists of 19 subjects (47,5%) and the experimental group of 21 subjects (52,5%). There were 18 females (45%). Of these 40 subjects, 35 were Dutch (87.5%), 2 South Korean (5%), 2 British (5%) and 1 Swedish (2,5%). The average age was 21.93 years with a minimum of 18 years and a maximum of 26 years. 21 subjects (52,5%) majored in economics, business or finance.

3_{For the questions see appendix 3}

(20)

Table 2- Subjects Total Control group Experimental group Number of cheaters Number of hints withheld Hints withheld per subject Subjects 40 19 (47.5%) 21 (52.5%) 14 (35%) 48 3.4 Males 22 (55%) 9 (47.4%) 13 (61.9%) 8 (36.4%) 30 (62.5%) 3.75 Females 18 (45%) 10 (52.6%) 8 (38.1%) 6 (33.3%) 18 (37.5%) 3.0 Dutch nationality 35 (87.5%) 16 (45.7%) 19 (54.3%) 13 (37.1%) 46 3.54 Foreign nationality 5 (12.5%) 3 (60%) 2 (40%) 1 (20%) 2 2.0 Economics students 21 (52.5%) 10 (47.6%) 11 (52.4) 9 (42.9%) 34 3.78 Non-Economics students 19 (47.5%) 9 (47.4%) 10 (52.6%) 5 (26.3%) 14 2.8 Result 1

35% of the subjects cheated on the quiz by stating fewer hints than actually used. The number of hints withheld was on average 3.4 per cheating subject. As shown in table 3, the control group cheated more often. However, the cheaters in the experimental group showed a greater tendency to cheat by omitting more hints when cheating. To test hypothesis 1, I check whether subjects in the experimental group, who received positive attribution, were less likely to cheat than the control group. To determine whether there is a significant association between cheating and being positively attributed, I use a Fisher’s Exact test5. The result shows no significant effect of positive attribution on cheating (see table 4). Nevertheless, an effect is visible in table 3 that points toward that positive attribution does affect the extent of cheating. Perhaps the results did not turn out to be significant because the low number of observations limits the power of the test.

5_{This test is chosen because of the nominal scales, the binary variables used and the fact that there is only one}

(21)

The correlation between Positive Attribution and the Level of Cheating6 is negative overall but when corrected by only looking at those who cheated, it becomes slightly positive. This is consistent with the higher than average number of hints withheld by the cheaters in the experimental groups (as seen in table 3). The finding that subjects cheated at a higher level when being positively attributed is an unexpected outcome.

Table 3 – Difference in cheating

Total Control group Experimental group Number of cheaters 14 (35%) 9 (47.4%) 5 (23.8%) Number of hints withheld 48 30 (62.5%) 18 (37.5%) Hints withheld per cheating subject 3.4 3.33 3.6

Table 4 – Effect of Positive Attribution on Cheating Exact Sig. (2-sided) Exact Sig. (1-sided) Fisher’s Exact Test .186 .110 Regression model

To look more deeply at the relationship between positive attribution and cheating I set up a logistic regression model.

𝐶ℎ𝑒𝑎𝑡𝑖𝑛𝑔 =0 −1 ∗ 𝑃𝑜𝑠𝑠𝐴𝑡𝑡𝑟−2 ∗ 𝐺𝑒𝑛𝑑𝑒𝑟 +3 ∗ 𝐸𝑑𝑢 −4 ∗ 𝑁𝑜_𝐸𝑐𝑜𝑛_𝑠𝑡𝑢𝑑 −5 ∗ 𝐴𝑔𝑒 +6 ∗ 𝐷𝑢𝑡𝑐ℎ

In the regression model all the control variables are included. For Gender, men are defined as 0 and women as 1, Poss Att is a binary variable that is 1 if the subject is positively attributed and 0 otherwise. Edu is the highest level of education attained7 and No_Econ_stud is the binary

variable to show that the subject is an economics student (0) or not (1). Age is between 17 and 27 and Dutch is the binary variable for being Dutch (1) or not (0).

6_{1 = one hint withheld,……, 6 = six hints withheld.} 7_{2= Intermediate Vocal Education….., 5 = University MSc}

(22)

The only variable that added significantly (at the 10% level) to the model/prediction is Positive Attribution, which has a negative coefficient (see table 5). This is a remarkable result considering the relatively small number of subjects. The significant relation between positive attribution and cheating suggests that, when controlling for the other variables, positive attribution has a negative impact on cheating. According to this model the probability that someone will cheat decreases by more than 30% after someone has been positively attributed (marginal effects are given in de dy/dx column of table 5).

The regression model provides an important piece of evidence in favor of hypothesis 1. It presents significant evidence (at a 10% confidence level) that positive attribution does decrease the likelihood of a person cheating, indicating that positive attribution can be a way of preventing moral licensing. In such a small-scale experiment, the suggested drop-off in the likelihood of a person cheating of 30% suggests that positive attribution may well be an antidote for moral licensing.

Table 5 – Variables in the regression model and marginal effects

B S.E. Sig. dy/dx S.E. Sig. Poss_Attr -1.408 .815 .084* -.303 .167 0.067* Gender -.068 .789 .931 -.015 .173 0.931 Edu .879 .761 .248 0.192 .163 0.237 No_Econ_Stud -.555 .801 .489 -.121 .176 0.490 Age -.305 .337 .365 -.067 .073 0.361 Dutch 1.084 1.305 .406 -.197 .186 0.288 Constant 3.185 5.895 .589

*. Correlation is significant at the 0.10 level

Result 2a

There is a notable difference between economics students and non-economics students (see table 6). Economics students cheated at a higher rate (42,9% versus 26,3%) and the economics

students that did cheat, cheated much more (3,78 versus 2,8). A Fisher’s Exact Test on the effect of the subject’s field of study on cheating (table 7) turns out to be insignificant, however.

Nevertheless, various signs point to the possibility that economics students could be more likely to cheat. There is a negative correlation between field of study and Cheating, Cheat Level8 and

(23)

Number of Hints 9. Although there some signs that economics students may be more likely to cheat, there is not enough evidence to support hypothesis 2a. The lack of significance makes it impossible to state that people who are studying economics are more likely to make rational decisions, which in this experiment was to cheat instead of being honest.

Table 6 – Differences between economic and non- economic students Economic students Non-Economic students Number of cheaters 9 (42.9%) 5 (26.3%) Number of hints withheld 34 14

Hints per subject 3.78 2.8

Table 7 – Effect of studying economics on Cheating Exact Sig. (2-sided) Exact Sig. (1-sided) Fisher’s Exact Test .333 .223 Result 2b

Hypothesis 2b expected economics students to use more hints overall and indeed, economics students did use more cheats (see table 8). Though the effect is in the direction predicted, the differences are statistically insignificant. The questions should not have been any harder for economics students than non-economics students, which suggests that economic students could be more likely to analyze the experiment in a more rational way but no clear effect is found.

Table 8- Difference economic and non-economic students Total Economics

students

Non-economics students

Mean 3.43 3.62 3.21

*Pearson Chi-Square Asymp.Sig. (2-sided): 0.268

(24)

Gender differences

Results show that men cheated slightly more often than women (36,4% versus 33,3%) and when men cheated they did so more frequently than women by withholding 25%-points more hints (see table 9). Gender is negatively correlated with cheating and the level of cheating, but Fisher’s Exact Test shows no significant association between cheating and gender (see table 10). The finding that men cheated slightly more during the experiment could partly be explained by the higher proportion of more male subjects being economics students (64%) compared to female economic students (36%). As explained, subjects who are economics students did use more hints and previous literature suggests that economics students are more likely to make more rational decisions. Nevertheless, the gender differences and the differences between economics and non-economics students are not significant, which makes it impossible to draw conclusions based on these results. Contrary to these results, women did use many more hints on average. Gender is significantly positively correlated at a 5% level to number of hints used (see table 11), indicating that when women are more likely to use more hints. This should also suggest that women are acting more rationally, as explained in hypothesis 2b, by using more hints. These contradicting results leave the difference of the gender effect unanswered and open to further research.

Table 9 – Gender differences Number of hints withheld Hints withheld per subject Total number of hints Average number of hints Males 30 (62.5%) 3,75 59 2,68 Females 18 (37.5%) 3,0 78 4,33

Table 10 – Effect of Gender on Cheating Exact Sig. (2-sided) Exact Sig. (1-sided) Fisher’s Exact Test 1.000 .554 Table 11 – Correlations

Amount of hints used Gender Pearson

Correlation .379

*

Sig. (2-tailed) .016

(25)

Discussion and conclusion

The main focus of this thesis was on studying the effect of positive attribution on moral licensing in order to try and find a method to prevent people from moral licensing their immoral deeds. The research was driven by the contrasting results in the literature where on the one hand it is argued that people tend to moral license themselves, and by doing so act inconsistently, while on the other hand it is argued that people tend to act consistently with their previous behavior. Previous findings argue that human behavior, with respect to moral acts, differs according to situation, context and circumstance. Nevertheless, by focusing on positive attribution, this study has tried to highlight a way that could prevent people from moral licensing.

Based on the numerous studies that provide evidence on positive attribution encouraging consistent behavior, this study set out with the expectation that positive attribution following a moral deed would lead to an increase in consistent subsequent moral behavior. In the experiment people were nudged into a moral deed and directly after this deed proceeded to take a test that provided the opportunity to act immorally. Being positively attributed should, according to the expectations, decrease the opportunity of acting immorally.

The results put forward evidence to show that positive attribution helps to prevent moral licensing. Subjects who were positively attributed cheated less and this effect was found to be statistically significant when controlling for background variables in a regression model.

The data appear to suggest that economics students are more rational than non-economics students. They use more hints and cheat more often, which is, from a Homo-Economicus point of view, more rational. However, none of these results turned out to be significant. Presumably, the main reason for this insignificance is the sample size of 40 subjects. This means the results have a low level of statistical power. A larger sample size would possibly have produced more conclusive results for this research question.

Aside from the low significance due to the small sample size, another problematic aspect that could have been improved upon is the representativeness within the subject group. All subjects were relatively highly educated students, which raises the question whether the results are representative for a broader population. Nonetheless, I see no a-priori reason why the observed results would not be found in other groups (e.g. older/young, lower education,

(26)

rich/poor). Intuitively, the highly educated might even be less influenced by positive attribution, because they are more likely to make a more well thought decision irrespective of the positive attribution.

The results show a slightly different outcome to previous studies on cheating and dishonesty rates. During the experiment ‘only’ 35% of the subjects cheated. In previous experiments on cheating the dishonesty rates were often higher, ranging from 30% up to 80% (Rosenbaum, Billinger and Stieglitz, 2014). There are several possible explanations why my subjects cheated less. Firstly, the experimenter did have some social contact while recruiting the subjects.

Although the contact was minimized after the subjects had agreed to do the experiment, the small talk prior to the experiment may still have created some goodwill by the subjects towards the experimenter, creating some kind of threshold to prevent the subject ‘deceiving’ the

experimenter by cheating.

Additionally, there is the possibility that subjects felt a certain level of social pressure to appear moral. Although it was stated multiple times that only the paper answer sheet was used for payout, which hinted at the possibility of cheating, the program that was used for the quiz is Qualtrics. Students occasionally use Qualtrics, which makes them informed about the data collecting features that the program has. Furthermore, some of the subjects wrote down a sign (e.g. the letter H or a dot) next to the questions where they used a hint. This could have aroused the feeling that the experimenter was able to see where they used a hint, which again could have created a barrier between the subject and the act of cheating.

One way in which cheating could have been stimulated is by raising the payoffs. The chance to win 5 euro was quite low but if the payoff were to be increased notably, the decision to cheat would have been easier. Nonetheless, the effect of positive attribution could also decrease with higher payoffs from cheating.

Besides arguments that some improvements in the design could have made subjects cheat more, a case can also be made that the subjects could have cheated less. For example, there is a

possibility that people were not focused on cheating until the moment they had to fill out the number of hints they used. By the time they reached this question, they may have forgotten how many hints they had used. This may have resulted in accidental cheating. The only way to

(27)

prevent this is by instructing subjects to mark down if they use a hint when filling out the question. Yet, by doing this it would have been possible for the experimenter to see how many hints you used, which needs to remain unknown.

Considering these possible shortcomings of the experiment and comparing the influence they may have had on the outcome, the overall results of the experiment still create the impression that, in future studies, there may be a greater likelihood of findings suggesting the effect of positive attribution can be stronger. In conclusion, I believe that future research on the effect of positive attribution on moral licensing should be stimulated. As with many studies on moral licensing, positive attribution and consistent behavior, one of the main limitations of this study is the context-dependent nature of the results. The way people behave and the times when we act morally and immorally differs from one situation to another. This study has tried to find evidence for a relationship between positive attribution and moral licensing in a social domain. However there are multiple domains where people morally license, therefore highlighting the many

domains where moral licensing can be prevented. Perhaps future research will be able to provide answers on how to prevent moral licensing in all the different situations.

(28)

Bibliography

Anthes, E. (2010). Green and mean. Scientific American Mind, 21(4), 12-12.

Batson, C. D., Thompson, E. R., Seuferling, G., Whitney, H., & Strongman, J. A. (1999). Moral

hypocrisy: Appearing moral to oneself without being so. Journal of Personality and Social

Psychology, 77(3), 525.

Blanken, I., van de Ven, N., Zeelenberg, M., & Meijers, M. H. (2014). Three attempts to replicate the

moral licensing effect. Social Psychology, 45(3), 232.

Blanken, I., van de Ven, N., & Zeelenberg, M. (2015). A meta-analytic review of moral licensing.

Personality & Social Psychology Bulletin, 41(4), 540-558.

Burger, J. M., & Caldwell, D. F. (2003). The effects of monetary incentives and labeling on the

foot-in-the-door effect: Evidence for a self-perception process. Basic and Applied Social Psychology,

25(3), 235-241.

Carter, J. R., & Irons, M. D. (1991). Are economists different, and if so, why? The Journal of

Economic Perspectives, 5(2), 171-177.

Chiou, W. B., Yang, C. C., & Wan, C. S. (2011). Ironic effects of dietary supplementation: Illusory

invulnerability created by taking dietary supplements licenses health-risk behaviors.

Psychological Science, 22(8), 1081-1086.

Clot, S., Grolleau, G., & Ibanez, L. (2011). Do good deeds make bad people? European Journal of

Law and Economics, , 1-23.

Conway, P., & Peetz, J. (2012). When does feeling moral actually make you a better person?

conceptual abstraction moderates whether past moral deeds motivate consistency or

(29)

Cornelissen, G., Pandelaere, M., Warlop, L., & Dewitte, S. (2008). Positive cueing: Promoting

sustainable consumer behavior by cueing common environmental behaviors as environmental.

International Journal of Research in Marketing, 25(1), 46-55.

Davis, L. W. (2008). Durable goods and residential demand for energy and water: Evidence from a

field trial. The Rand Journal of Economics, 39(2), 530-546.

Gneezy, U., Imas, A., & Madarász, K. (2014). Conscience accounting: Emotion dynamics and social

behavior. Management Science, 60(11), 2645-2658.

Goldman, M., Seever, M., & Seever, M. (1982). Social labeling and the foot-in-the-door effect.

Journal of Social Psychology, 117(1), 19-23.

Jordan, J., Mullen, E., & Murnighan, J. K. (2011). Striving for the moral self: The effects of recalling

past moral actions on future moral behavior. Personality & Social Psychology Bulletin, 37(5),

701-713.

Kahneman, D., Knetsch, J. L., & Thaler, R. (1986). Fairness as a constraint on profit seeking:

Entitlements in the market. The American Economic Review, , 728-741.

Keltner, D., & Buswell, B. N. (1997). Embarrassment: Its distinct form and appeasement functions.

Psychological Bulletin, 122(3), 250.

Khan, U., & Dhar, R. (2006). Licensing effect in consumer choice. Journal of Marketing Research,

43(2), 259-266.

Kraut, R. E. (1973). Effects of social labeling on giving to charity. Journal of Experimental Social

Psychology, 9(6), 551-562.

Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: A theory of self-concept

(30)

Mazar, N., & Zhong, C. B. (2010). Do green products make us better people? Psychological Science,

21(4), 494-498.

Merritt, A. C., Effron, D. A., Fein, S., Savitsky, K. K., Tuller, D. M., & Monin, B. (2012). The

strategic pursuit of moral credentials. Journal of Experimental Social Psychology, 48(3),

774-777.

Merritt, A. C., Effron, D. A., & Monin, B. (2010). Moral self‐licensing: When being good frees us to be bad. Social and Personality Psychology Compass, 4(5), 344-357.

Miller, R. L., Brickman, P., & Bolen, D. (1975). Attribution versus persuasion as a means for

modifying behavior. Journal of Personality and Social Psychology, 31(3), 430.

Monin, B., & Miller, D. T. (2001). Moral credentials and the expression of prejudice. Journal of

Personality and Social Psychology, 81(1), 33.

Nisan, M., & Horenczyk, G. (1990). Moral balance: The effect of prior behaviour on decision in

moral conflict. British Journal of Social Psychology, 29(1), 29-42.

Rosenbaum, S. M., Billinger, S., & Stieglitz, N. (2014). Let’s be honest: A review of experimental

evidence of honesty and truth-telling. Journal of Economic Psychology, 45(0), 181-196.

Sachdeva, S., Iliev, R., & Medin, D. L. (2009). Sinning saints and saintly sinners: The paradox of

moral self-regulation. Psychological Science, 20(4), 523-528.

Strenta, A., & DeJong, W. (1981). The effect of a prosocial label on helping behavior. Social

Psychology Quarterly, , 142-147.

Thaler, R. H., & Sunstein, C. R. (2003). Libertarian paternalism. American Economic Review,

175-179.

Young, L., Chakroff, A., & Tom, J. (2012). Doing good leads to more good: The reinforcing power

(31)

Appendix

Appendix 1 - Instruction sheet

During the following experiment you can earn money by answering 6 multiple-choice questions. Your aim is to maximize your earnings. Each question has 6 different options (A to F) where only 1 is correct. You have to fill in the answers both on the computer and on your paper answer sheet. Only the paper answer sheet will be used for payout, not the computer input.

Instead of choosing 1 of the 6 options there is also the possibility to ‘buy’ a hint. To buy a hint you can press the option I want to use a hint (option G). This hint will limit the possible options from 6 to 2 answers. Using a hint will cost you 0,50 euro that is deducted from your earnings. The 0,50 euro is only deducted if you have positive earnings.

At the end of the experiment 1 of the 6 questions will be randomly selected for payout by

throwing a dice. If the selected question is answered incorrectly you will receive 0,00 euro. If the selected question is answered correctly you will receive 5,00 euro. This will be reduced by the amount of hints you used times 0,50 euro.

The payouts will take place immediately after you finished the six questions and will be done by the experimenter. The experimenter will only look at the answer on the paper answer sheet and will deduct the amount of hints used that are filled out at the paper answer sheet. Please

remember the amount of hints you used.

Before starting the experiment please answer the following example questions to make sure you understand the experiment.

1. Only one of the 6 questions is selected for payout. True False

2. All of the used hints are reduced from your earnings. True False

3. When I answered the randomly selected question correctly and used 3 hints during the entire quiz my earnings will be:

A. 0,00 euro B. 2,00 euro C. 3,50 euro D. 5,00 euro

4. When I answered the selected question incorrectly and used 3 hints during the entire quiz my earnings will be:

A. 0,00 euro B. 2,00 euro C. 3,50 euro D. 5,00 euro

(32)

Appendix 2 - Questionnaire 1. What is you gender?

Male Female

2. What is your nationality?

...

3. What is your age?

...

4. What is your highest attained level of education? a. High school

b. Intermediate Vocal Education (MBO) c. College (HBO)

d. University BSc (bachelor universiteit) e. University MSc (master universiteit)

5. What is your current field of study?

a. Economics, business or finance related b. Other

(33)

Appendix 3 - Answer Sheet

Please circle the answer chosen on the computer

Afterwards this paper answer sheet will immediately be used for payout.

1. How many people live in the country Azerbaijan?

A. 2,100,000 B. 4,500,521 C. 6,693,294 D. 7,851,365 E. 9,686210 F. 12,582,377

2. How long is the Great Chinese Wall? A. 20,445 meters B. 21,196 meters C. 22,101 meters D. 22,399 meters E. 23,461 meters F. 25,660 meters

3. How many seats does the Concert Hall of the Sydney Oprah House have?

A. 2,679 B. 2,843 C. 3,005 D. 3,800 E. 4,595 F. 5,678

4. How tall is the Eiffel Tower?

A. 266 meters B. 288 meter C. 299 meters D. 324 meters E. 390 meters F. 403 meters

5. How many McDonalds restaurants were there worldwide in 2014?

A. 27,951 B. 29,103 C. 30,888 D. 34,987 E. 36,258 F. 40,824

(34)

6. For how much dollar did the United States buy Alaska from the Russian Empire in 1867? A. 7.0 million B. 7.1 million C. 7.2 million D. 7.3 million E. 7.4 million F. 7.5 million

How many Euros should be deducted because of the used hints?

The experimenter will use the answer you state below when deducting the payout.

A. 0 B. 0.50 euro C. 1.00 euro D. 1.50 euro E. 2.00 euro F. 2.50 euro G. 3.00 euro

Accrediting the good to prevent the bad : preventing moral licensing through positive attribution