Dishonestly increasing the likelihood of winning

People not only seek to avoid losses or secure gains; they also attempt to create opportunities for obtaining positive outcomes. When distributing money between gambles with equal probabilities, people often invest in turning negative gambles into positive ones, even at a cost of reduced expected value. Results of an experiment revealed that (1) the preference to turn a negative outcome into a positive outcome exists when people’s ability to do so depends on their performance levels (rather than merely on their choice), (2) this preference is amplified when the likelihood to turn negative into positive is high rather than low, and (3) this preference is attenuated when people can lie about their performance levels, allowing them to turn negative into positive not by performing better but rather by lying about how well they performed.

Keywords: lying, decision making, ethics, morality, dishonesty, mixed gambles, choice.

1 Introduction

Losses loom larger than gains (Kahneman & Tversky, 1979), and avoiding losses is more desired than securing gains (Tversky & Kahneman, 1981; Slovic, Fischhoff, & Lichtenstein, 2001), especially when the motivation is in line with the individual’s goals (Elliot & Church, 1997; Elliot & Harackiewicz, 1996; Förster, Higgins, & Idson, 1998; Higgins, 1997). Beyond avoiding losses and securing gains, however, people are motivated to create opportunities for securing positive outcomes (Payne, 2005). People seek to increase the probability to win something, even at the cost of reduced expected value (Payne, Samper, Bettman, & Luce, 2008). This tendency can translate into large monetary losses in real life, as people attempt to minimize the number of debts they have rather than focusing on reducing the total amount they owe (Amar, Ariely, Ayal, Cryder, & Rick, 2011). Specifically, among people holding multiple credit cards Amar et al. (2011) found evidence for a debt account aversion: people consistently paid off small debts first, even though the larger debts had higher interest rates. People’s psychological need to turn a negative-to-positive can be costly.

Payne and colleagues (Payne, 2005; Payne, et al., 2008; Venkatraman, Payne, Bettman, Luce, & Huettel, 2009) studied people’s tendency to attempt increasing their probability of winning a positive outcome using mixed probability gambles. Participants were asked to imagine that they “own a five-outcome gamble with the following payoffs and probabilities: ($100, .20; $50, .20; $0, .20; –$25, .20; –$50, .20)… Now imagine that I tell you that you can change the gamble above in one of two ways. You can add $38 to the outcome that pays $100 or you can add the $38 to the outcome that pays $0.” (Payne, 2005; p. 10). Although the expected value of the gamble remains constant regardless of the outcome participants chose to add money to (as all outcome had the same probability to be chosen), 68% of participants preferred to add money to the outcome that increased their probability of winning a positive outcome (e.g., $0; increasing the probability of winning a positive outcome from 2-out-of-5 to 3-out-of-5). The remaining 32% chose to add the money to the alternative outcome that did not increase their likelihood of winning something (e.g., $100).¹ Venkatraman, et al. (2009) further found that 65% of the people who considered the following gamble ($130, .20; $115, .20; $50, .20; $30, .20; –$10, .20), chose to add their $30 to the –$10 outcome (rather than the $50 outcome), allowing them to turn the one negative outcome into a positive outcome, and thus increase their probability of winning something in the gamble from 4-out-of-5 to 5-out-of-5.

Although people can influence their outcomes by choosing how to allocate funds (selecting gambles, Payne, 2005; repaying debt, Amar et al., 2011), they are often also able to influence their outcomes by choosing a task and working on it. For example, observing multiple tasks piling on their desks, such as emails to respond to and exams to grade, people must choose the task they will attend to first. They may choose the more urgent task of grading exams, or alternatively choose the task that is likely to take the least time to accomplish and reply to (not so urgent) emails. In such, people’s desire to have less unattended tasks on their table, might lead them to procrastinate on the more urgent tasks they are facing.² The current paper focuses on a situation in which people can influence their outcomes by choosing one of multiple outcomes and adding value to it as a function of their task performance: predicting the outcomes of multiple coin tosses. Specifically, the current work asks whether people choose to invest their efforts in turning a negative outcomes into positive outcomes and thus increase their probability of winning a positive outcome. Additional questions addressed here are whether this tendency is amplified when the likelihood to turn the negative-to-positive is high rather than low, that is, when the performance needed to accomplish turning the negative outcome into a positive outcome is relatively feasible (easy to accomplish) rather than relatively less feasible (difficult to accomplish). And finally, I ask whether ambiguous settings in which people may over-report their performance levels (i.e., lie) reduce their sensitivity to the objective feasibility to turn negative-to-positive?

To address these questions, I adapted Payne’s (2005) paradigm and asked participants to add points to one of three outcomes, labeled “Pot A”, “Pot B” and “Pot C”, by choosing one of the pots. Participants then predicted the outcomes of 20 coin tosses, gaining one point per correct prediction. Points were added to the pre-selected pot only. After participants completed the coin tossing task, one of the pots was randomly selected to determine the participant’s final outcome. Points translated to lottery tickets for an alleged lottery with two €10 prizes. Pot A had a positive initial outcome (38 points), Pot C had a negative initial outcome (–30 points), while Pot B had a negative initial value that could potentially turn into a positive outcome, if participants predicted well enough on the coin tossing task. Specifically, I manipulated whether the initial value of Pot B made it likely to turn from negative-to-positive (initial value = –6 points) or unlikely to turn from negative-to-positive (initial value = –12 points). As people seek to maximize the probability of winning something, and thus turn negative outcomes into positive ones, they should be sensitive to the likelihood of turning the negative outcome into a positive one. When the likelihood of successfully turning the negative outcome into a positive outcome is low (i.e., when accomplishing the task is possible but relatively difficult due to the initial value of the pot), people should be less likely to choose modifying this outcome. In contrast, when the likelihood of successfully turning the negative outcome into a positive outcome is high (i.e., when accomplishing the task is possible and also relatively easy due to the initial value of the pot), people should be more likely to choose modifying this outcome. Formally put:

Relative likelihood to accomplish a task however, matters only when performance is evaluated objectively, that is, when people are not able to over-report their performance levels to (unethically) boost their profit. In contrast, when people have the ability to lie about their performance levels, the difficulty of the task becomes less important, as people can lie (rather than perform) in order to achieve a positive outcome. This begs the question of whether situations allowing people to dishonestly over-report their performance levels may lead people to pay less attention to the likelihood of being able to turn a negative-to-positive (for a discussion on the issue in the tax compliance context; Kirchler, 1999; 2007; Kirchler, Hoelzl, & Wahl, 2008). In other words, when people can lie to improve their outcomes, will the likelihood of turning a negative-to-positive (i.e., low vs. high initial pot value) no longer be relevant? Moreover, could the possibility to turn a negative-to-positive justify lying?

1.1 Justified ethicality

Settings that allow people the freedom to report their performance levels themselves (e.g., annual earnings) consequently enable them to exaggerate their reports. Such settings allow people to lie, rather than perform, for profit. The classic economic approach to people’s decision of whether to lie or not is Becker’s cost-benefit model of crime and punishment (Becker, 1968). According to this approach, lying depends on the balance between the potential profits generated by the lie, on the one hand, and the likelihood of getting caught multiplied by the magnitude of subsequent punishment, on the other (Becker, 1968). A growing body of research suggests however, that even in completely anonymous settings, where detection of one’s lie is impossible, people restrict the amount of their lies (Ariely, 2008; Ayal & Gino, 2011; Bazerman & Tenbrunsel, 2011; Gneezy, 2005; Atanasov & Dana, 2011; Gino, Ayal, & Ariely, 2009; Lundquist, Ellingson, & Johannesson, 2009; Schweitzer, Ord’oñez, & Douma, 2004). People seem to strive to balance their competing desires of profiting from lying while maintaining a positive self-view as honest individuals (Mazar, Amir & Ariely, 2008; Shalvi, Handgraaf & De Dreu, 2011), which leads them to lie exactly to the extent that they can self-justify their lies (Shalvi, Dana, Handgraaf, & De Dreu, 2011; Schweitzer & Hsee, 2002).

The idea that people need self-justifications for lying is based on Kunda’s notion that “people are likely to arrive to conclusions that they want to arrive at, but their ability to do so is constrained by their ability to construct seemingly reasonable justifications for these conclusions” (Kunda, 1990, p. 480; see also Shafir, Simonson, Tversky, 1993; Schweitzer & Hsee, 2002). Indeed, recent work suggests that people seek to appear fair and honest not only in the eyes of others but also in their own eyes (Shalvi, Dana, et al., 2011; see also Toure-Tillery & Fishbach, 2011; Dhar & Wertenbroch, 2012; Chance, Norton, Gino, & Ariely, 2011). Shalvi, Dana, Handgraaf and De Dreu (2011) asked participants to roll a die under a paper cup with small hole in the top allowing only them to see the outcome, and earn money according to what they reported rolling (1=$1, 2=$2, etc). As participants’ rolls were truly private, lying was assessed by comparing the reported distribution to the distribution predicted by chance (Fischbacher & Heusi, 2008). Participants were asked to roll three times but report only the outcome of the first roll. While all three rolls were private, the distribution of reported outcomes resembled the distribution of choosing the highest of the three observed rolls. Critically, a condition allowing participants to roll only once, ceteris paribus, revealed less lying (for recent replications of this effect see experiments 3 & 4 in Gino & Ariely, 2012; Shalvi, Eldar, & Bereby-Meyer, in press). Participants clearly found value in being able to justify their lies to themselves.

The theoretical idea behind the finding that more rolls lead to more lying is that observing desired counterfactual outcomes (upward counterfactuals; Roese, 1997; Epstude & Roese, 2008; Markman, Gavanski, Sherman, & McMullen, 1993), in the form of higher values appearing on non-relevant for pay rolls, changes the way participants feel about lying using these specific outcomes. It two additional experiments (Shalvi, Dana et al., 2011; Experiments 3 & 4) participants observed different die roll combinations and reports and were asked to rank the extent to which they considered different combinations to be lies (on a continuous scale ranging from “not at all” to “very much”). Participants ranked justified lies (lies in which the report equaled a value observed on a non relevant for pay roll; 1^st roll < report = 2^nd roll) as less of a lie compared to unjustified lies (lies in which the report did not equal any of the observed values; 1^st roll < report ≠ 2^nd roll). It seems that lying by shuffling observed facts feels more legitimate than lying by inventing new facts.

In the insurance industry it indeed seems to be the case that justified lies are used more than unjustified lies, as more people exaggerate their otherwise legitimate claims (“buildup”) than committing planned or outright fraud by filing claims about events that never occurred (Crocker & Morgan, 1998; Tennyson, 2008). Interestingly, economics students are more likely to use justified lies compared to psychology students, potentially due to their training (or their predisposition) in maximizing profit while maintaining an honest self and public image (Lewis, Bardis, Flint, Mason, Smith, Tickle, & Zinser, 2012). Further support to the idea was gained by the fascinating finding reported by Gino and Ariely (2012) that creative people, who are higher on flexible thinking, and are thus better in creating their own counterfactual reality, do not require extra rolls to justify their lies. While less creative people lied more when they rolled the die multiple times compared to when they rolled only once, creative people were not influenced by the number of rolls they had, and lied extensively in both cases. Evidently, creative people can create alternatives to factual reality using their flexible thinking style.

The question however remains, can the motivation to turn a negative-to-positive justify lying? As in other domains in life, people’s ethical decisions are influenced by systematic cognitive biases in their information perceptions (Bazerman & Banaji, 2004; Bazerman & Tenbrunsel, 2011). Kern and Chung (2009) provide initial support for this idea by asking people to estimate the likelihood that they would lie in a series of hypothetical scenarios. The researchers found that people state they would lie more to avoid a loss than to secure a gain: Students assessed the likelihood that they, in the role of an entrepreneur interested in buying a competing company, would illegally seek inside information about the competing company. Participants learning that lying would result in a 75% chance of not losing the acquisition were more likely to lie compared to participants learning that lying would result in a 25% chance of winning the acquisition. Furthermore, people lie more the closer they are to achieving their goals (Schweitzer, et al., 2004). Together, as people seek to turn negative-to-positive, settings allowing people to over-report performance levels should reduce people’s sensitivity to the likelihood of being able to turn a negative-to-positive honestly, because they may turn the negative-to-positive by lying. Having the private justification to lie in order to have higher probability to win something should lead people to make decisions that would allow them to achieve this goal. That is, as long as their goal of turning a negative-to-positive was not met, people will still be in a loss domain and would be motivated to cheat in order to eliminate their loss.

Focusing on actual rather than hypothetical decisions, I manipulated whether people could (vs. not) lie by over-reporting their performance level, to test:

Finally, the extent to which participants were motivated to avoid losing when making the pot selection decision was assessed.

2 Method

2.1 Participants

Three hundred and sixty-six students participated as part of a course requirement and were randomly assigned to a condition of a 2 (Ability to lie: Yes vs. No) X 2 (Negative-to-Positive Likelihood: Low vs. High) between-participants factorial design.

2.2 Procedure

Participants allocated 20 points into one of three different pots with equal probability of being selected by the computer to determine their outcome. After choosing one of the pots, points were added by predicting the outcome of 20 coin tosses, with every correct prediction adding one point to the pot. Participants were instructed that each point they receive will be converted into a single lottery ticket, and that at the end of the study all participants will enter the lottery with the number of tickets they managed to obtain. It was clarified that more tickets mean a higher chance to win one of two €10 prizes. This procedure was used to motivate participants to perform well on the task (and get as many lottery tickets as they possible can). Critically, I manipulated whether each of the participants’ predictions was typed into the computer before observing the outcome or only kept in their mind before reporting if the predictions was correct (for similar approach to studying dishonesty see Greene & Paxton, 2009; Jiang, 2012; Schurr, Ritov, Kareev & Avrahami, 2012). The latter procedure allowed participants to lie by over-reporting the number of correct predictions they made.

Participants chose whether to add points to Pot A which had +38 points, Pot C which had –30 points, or Pot B which had –6 [vs. –12] points and was thus likely [vs. unlikely] to turn from negative-to-positive (in case participants reported honestly). Once participants chose a pot, each correct coin toss prediction added a point to the chosen pot. Depending on participants’ coin toss predictions, the positive pot thus varied between 38 (=38+0) and 58 (=38+20) points, the negative pot between –30 (=–30+0) and –10 (=–30+20) points, and the negative-to-positive pot between –6 (=–6+0) and 14 (=–6+20) [vs. –12 (=–12+0) and 8 (=–12+20)] points. Choosing the negative-to-positive pot and predicting well enough thus increased the probability that the randomly selected pot would yield a positive outcome (from 1-in-3 to 2-in-3 pots). Importantly, participants’ pot choice did not influence the expected value of the final outcome as all pots were equally likely to be chosen for pay.

Participants read instructions explaining the described procedure, engaged in 5 practice trials familiarizing them with the coin-tossing task, chose to which pot they wanted to add points, engaged in 20 coin-toss predictions, and finally learned which pot was randomly selected for pay. The practice trials ensured that participants in the “ability to lie” condition understood that they were able to over-report the number of coin tosses they correctly predicted, while participants in the “no ability to lie” condition understood that they were not able to over-report the number of coin tosses they correctly predicted. See Figure 1 for a graphical display of the task procedure.

2.3 Dependent variables

The main dependent variable was the pot participants chose to add points to. The aggregate number of correct predictions was measured as an indicator for lying in the condition in which participants were able to lie. While it was not possible to assess whether a specific individual was lying, comparing the distributions of reports in the different conditions to the distribution predicted if participants were reporting honestly (i.e., 50% correct predictions) allowed detecting lying on an aggregate level. After completing the coin-toss predictions, participants’ motivation for choosing their selected pot was assessed by asking about the extent to which their choice was “an attempt to avoid losing points” and “an attempt to get as many points as possible” (1 =not at all to 7 =very much). To verify task comprehension participants indicated from how many pots they had to choose (3 or 5) and whether all pots had the same likelihood to be selected (yes vs. no).

3 Results

3.1 Comprehension check

Most participants (95%) understood that they had to choose between three pots and that all pots had the same likelihood to be selected (99%). The 22 participants who failed to answer those questions correctly were excluded from all analyses.

3.2 Pot selection

Results supported Hypothesis 1, a chi-square analysis revealed a main effect for likelihood on the tendency to attempt turning a negative-to-positive (choose Pot B). People were more likely to attempt turning negative-to-positive when the likelihood to turn a negative-to-positive was high (36.4%, 63 of 173) than when it was low (25.1%, 43 of 171), χ²(1) = 4.61, p = .024.

Hypothesis 2 suggested that the likelihood to turn a negative-to-positive will influence pot choice depending on whether lying about performance is possible or not. Specifically, when lying about performance is not possible, a high likelihood to turn a negative-to-positive will lead more people to choose this option compared to when the likelihood to turn a negative-to-positive is low. In contrast, when lying about performance is possible, the likelihood to turn a negative-to-positive will have less influence on people’s choice of this option. Results provided partial support to Hypothesis 2. Descriptively, when participants were not able to lie, more participants chose the negative-to-positive pot when the likelihood of success was high (41%) than when it was low (21.7%). In contrast, when participants were able to lie, they chose the negative-to-positive pot to similar extents regardless of whether the likelihood of success was high (32.2%) or low (28.4%), see Table 1.

To test the statistical significance of this interaction pattern, a multinominal logistic regression model was used as it allows assessing nominal dependent variables with more than two levels (as in the case of choosing between three pots). A multinominal logistic regression with the likelihood to turn negative-to-positive and the ability to lie as independent variables and the selected pot as dependent variable, revealed a significant interaction effect: χ²(2) = 6.206, p = .045. This interaction effect was driven by the interaction between likelihood and the ability to lie influencing the contrast comparing the choice of the positive pot and the negative-to-positive pot, B = –1.183 (SE = .529), Wald (1) = 5.01, p = .025 (p = .05, after using a Bonferroni correction for the two possible contrasts). The contrast between the negative pot and the negative-to-positive pot was not significant, p = .781.³ In the general discussion I speculate about possible reasons why only the contrast between the positive pot and the negative-to-positive pot was significant.

To further clarify the observed pattern of choices, I assessed whether the proportion of participants choosing to add their points to the negative-to-positive pot differed from choosing one of the three pots at random (i.e., 1 of 3). In the “no ability to lie” condition, the proportion of people choosing the negative-to-positive pot in the low likelihood condition was lower (21.7%) than if they were choosing a pot at random (1/3; binominal probability, p = .014). Furthermore, the proportion of people choosing the negative-to-positive pot in the high likelihood condition was somewhat higher (41.0%) than if they were choosing a pot at random (1/3; binominal probability, p = .088, marginal). In contrast, in the “ability to lie” condition, participants chose the negative-to-positive pot at levels that did not differ than the likelihood predicted by choosing a pot in random (high likelihood, 32.2%; binominal probability, p = .19; low likelihood, 28.4%; binominal probability, p = .46).

3.3 Lying

In the condition enabling participant to lie, participants (ab)used their ability to influence their outcomes and over-reported the number of correct predictions. A binominal test revealed that participants that were able to lie over-reported their outcomes, in both low and high likelihood conditions. The number of successes for a person reporting honestly is distributed according to a binominal distribution with the parameters 20 and 0.5 (B (20, 0.5)). Summing over the N subjects in the low likelihood to turn negative-to-positive condition (N = 88), the number of success these individuals reported (1070) is distributed according to (B (88*20, 0.5)). The likelihood of observing 1070 or more correct predictions out of the 1760 (= 88 * 20) coin tosses is extremely low, p < .000001. Similarly, in the high likelihood to turn negative-to-positive condition (N = 90), the number of success these individuals reported (1063) is distributed according to (B (90*20, 0.5)). The likelihood of observing 1063 or more correct predictions out of the 1800 (= 90 * 20) coin tosses is extremely low, p < .000001. In contrast, in the condition in which participants were not able to lie by over-report their outcomes, they (not surprisingly) predicted at chance levels (820 correct prediction of 1160 in low likelihood condition, p = .64; 805 correct prediction of 1160 in high likelihood condition, p = .23). See Figure 2 for the histograms summarizing (per conditions) the proportion of participants reporting their total correct predictions (out of 20) among participant choosing the negative-to-positive pot (red), and among the remaining participants (gray).

Supporting Hypothesis 3, participants who chose to add points to the negative-to-positive pot managed to raise the pot’s value above zero more often in the condition enabling them to lie compared to the condition not enabling them to lie. When the likelihood to turn a negative-to-positive was high (initial value –6, EV= –6 + 10 = 4; probability for positive outcome based on honest reporting = .94), all but one participant managed to end up with a positive outcome regardless of condition. However, when the likelihood to turn a negative-to-positive was low (initial value –12, EV = –12 + 10 = –2; probability for positive outcome based on honest reporting = .13), participants in the condition enabling them to lie were more likely to turn a negative-to-positive (36%) than participants in the condition not enabling them to lie (5.6%), χ²(1) = 5.45, p = .02.

Interestingly, a Mann-Whitney test comparing the amount of lying between the low and high likelihood conditions did not reveal a significant difference, p = .193. Possibly, this is a result of the fact that, while people usually restrict the amount of their lies, they nevertheless lie a bit. Thus, it might be that, in the high likelihood condition (Pot B = –6), participant were lying (a bit) even though they did not need to in order to turn negative-to-positive. In the low likelihood condition (Pot B = -12), participants might have lied to turn the negative-to-positive which coincide with lying just a bit. The issue is further discussed in the general discussion.

3.4 Avoid losing

I assessed the extent to which participants in the different conditions were motivated by a desire to minimize the number of points they could lose vs. to gain as many points as possible. To do so, I computed an index of the relative desire to avoid losses compared with the desire to secure gains, by adding the avoid loss score and the (reverse coded) maximize gain score (both self-reported ratings measured at the end of the experiment). The index ranged from a desire to maximize gains (–6) to minimize losses (+6). A 2 (Ability to lie: Yes vs. No) X 2 (Likelihood: Low vs. High) between subjects ANOVA predicting the desire to avoid losses revealed no effect for likelihood, F(1, 340) = .03, p = .86. The main effect for the ability to lie was significant, F(1, 340) = 4.01, p < .05. Participants who were able to lie justified their decision by a stronger desire to avoid losses (M = .08, SD = 2.71) than those who were not able to lie (M = –.54, SD = 3.18). The interaction between the ability to lie and the likelihood to turn a negative-to-positive was also significant, F(1, 340) = 8.13, p < .01. A simple effects analysis revealed that the effect of the ability to lie was not significant when the likelihood for turning negative-to-positive was high, F(1, 340) = .37, p = .54, however, when the likelihood for turning negative-to-positive was low, participants who were able to lie justified their decision by a stronger motivation to avoid losses (M =.57, SD =2.48) than those who were not able to lie (M = –.96, SD = 3.10), F(1, 340) = 11.77, p =.001.

4 Discussion

Seeking to increase to probability of winning something makes people attempt turning negative-to-positive even when it comes at a price (Payne, 2005). This tendency influences economic decisions such as choosing which debt to repay first, leading people to seek closure and end up paying higher interest (Amar, et al. 2011). The novel results reported here suggest that (1) people’s desire to turn negative-to-positive occurs even when doing so depends on task performance, (2) this tendency was amplified when the likelihood to turn the negative-to-positive was high, and (3) attenuated when turning negative-to-positive could be obtained by dishonestly exaggerating performance levels.

Past work using mixed probability gambles by Payne and colleagues (Payne, 2005; Payne, et al., 2008; Venkatraman, et al., 2009) typically provided participants with a hypothetical gamble composed by payoff and probabilities (often including 5 options) and provided them with the option to add money to one of two options: (1) an option that increased the probability of winning vs. (2) an option that did not increase the probability of winning. As in the example given in the introduction, participants could choose to add $30 to one of two outcomes in the following gamble: ($130, .20; $115, .20; $50, .20; $30, .20; –$10, .20). Participants could either add the $30 to the outcome that pays -$10 (which will turn it into a positive outcome and increase the likelihood of winning a positive outcome) or to the outcome paying $50 (which will not increase the likelihood winning a positive outcome). A typical result in this line of prior work is that about 60%–70% choose to add value to the option that allows them to turn the negative-to-positive. In the current work, the proportion of participants choosing that option ranged between 21% (when they were not able to lie and had a low likelihood to turn negative-to-positive) to 41% (when they were not able to lie and had a high likelihood to turn negative-to-positive). Importantly, these values differed from the choosing an option at random. When the likelihood to turn negative-to-positive was low [high] participants were less [more] likely to choose this option compared to choosing an option at random.

There are several possible reasons why a different proportion of people chose the option to turn negative-to-positive in the current work compared to the work by Payne and colleagues, all relating to the adaptations of the original task. First, although in prior work participants chose whether to add points to one of two options (enabling vs. not to turn negative-to-positive), in the current work participants were asked to choose to add points (lottery tickets) to any of the three options included in the gamble: (Pot A 38, 1/3; Pot B –6 [vs. –12], 1/3, Pot C –30, 1/3). Thus, participants had more options to choose from. Providing an option to choose any of the available options may be considered a more naturalistic choice than limiting the possibilities to only a few selected values. However, this procedure also creates more variance in choice and reduces the likelihood that a given outcome will be chosen.

Although variations in choices between the positive and negative pots were not the focal point of the current work, they allowed discovering some interesting patterns. For example, although the interaction between likelihood and ability to lie influenced the contrast between the positive pot and the negative-to-positive pot, it did not influence the contrast between the negative pot and the negative-to-positive pot. Specifically, for participants who were not able to lie, having a low likelihood to turn negative-to-positive led more people (53%) to choose investing in the positive pot (Pot A) compared to when the likelihood high (34.9%). No difference on choosing the negative pot (Pot C) were observed (25.3% vs. 24.1%, respectively). In contrast, when participants were able to lie, having a low likelihood to turn negative-to-positive led more people (39.8%) to choose investing in the negative pot (Pot C) compared to when the likelihood high (26.7%). And a less meaningful difference on choosing the positive pot (Pot A) was observed (31.8% vs. 41.1%, respectively). Future work is needed to validate these unpredicted patterns and clarify how and why varying the likelihood to turn the negative outcome into positive outcome, influences people’s choices regarding other options available to them. Such future work may benefit from considering the tradeoff between the more naturalistic choice structure (allowing choosing any of the given options) and the rich and complex predictions this approach entails.

Second, in the current work points were added in a probabilistic rather than a certain way. This was done by predicting the outcome of 20 coin tosses, having each correct prediction adding a point to the selected gamble. Such reduced certainty that the given gamble will indeed turn from negative-to-positive is likely to further contribute to the relative low proportion of participants choosing the negative-to-positive compared to prior work. Critically, participants’ choices of adding points to the negative-to-positive outcome differed between experimental conditions, supporting the proposed theoretical framework.

Third, in the current work a condition in which participants were able to lie about their performance level to boost personal profit was added. A growing body of research provides evidence that, given the possibility to lie without getting caught, people lie (e.g., Barkan, Ayal, Gino, & Ariely, 2012; Mazar, et al. 2008; Vohs & Schooler, 2007; Batson, et a., 1997; Greene & Paxton, 2009; Hao & Hauser, 2010; Lammers et al. 2010). Providing better insight into the conditions which justify such unethical behavior is thus important from theoretical, applied and societal perspectives (Ariely, 2008; Bazerman & Banaji, 2004; Bocchiaro, Zimbardo, Van Lange, 2012; Chugh, et al., 2005; De Cremer, 2009; DePaulo, et al., 1996; Gneezy, 2005; Gunia, Wang, Huang, Wang, & Murnighan, 2012; Haidt, 2007; Jones, 1991; Van Gelder, 2012). Gaining better understanding of people’s “ethical blind spots” (Bazerman & Tenbrunsel, 2011) and the places where they feel legitimate to “stretch the truth” (Schweizer & Hsee, 2002) should allow people within and outside organizations to overcome these undesired behavioral biases (Gibson & Murnighan, 2009). The current work contributes to this attempt by pointing out that, when taking actual (rather than hypothetical) decisions, people feel justified taking decisions that will allow them to turn a negative-to-positive by lying. Holding the restrictions concerning the specific contrast driving the likelihood by ability to lie interaction in mind, the evidence provide here suggest that compared to participants who were not able to lie, participants who were able to lie were more likely to choose the negative-to-positive pot even when the likelihood to turn negative-to-positive was low (that is, if they were to report truthfully).

Another interesting finding concerns the levels of dishonesty identified in the current study. Among participants who could have lied, no differences were observed between the amount of lying among participants in the low and high likelihood conditions. As mentioned in the results section, one possibility for this result is the fact that, although people usually restrict the amount of their lies, they nevertheless lie a bit. In the context of 20 coin tosses, participants are expected (on aggregate) to predict 10 tosses correctly. If they lie a bit, they may claim an average of about 12 correct prediction, as was the case in both studied conditions. Thus, it is possible that in the high likelihood condition (Pot B = –6) participant were lying (a bit) even though they did not need to do so in order to turn negative-to-positive, while in the low likelihood (Pot B = –12), they might have lied to turn negative-to-positive which coincides with lying just a bit. This possibility is supported by the post-decision questions assessing participants’ motivation to avoid losses when making the pot choice. Notwithstanding, manipulating the negative-to-positive pot to hold an even more extreme initial negative value (e.g., –15 or –17) should allow revealing cheating differences between the different likelihood conditions. The idea is that, because the amount of cheating needed to turn negative-to-positive will be larger, it will lead people to lie more than just a bit.

From an applied perspective, the obtained results may be interesting in financial auditing contexts (Bazerman, Loewenstein & Moore, 2002). Tax reports that are filed by individuals or companies are often randomly audited by the tax authorities (Kirchler, 2007; 1999). When reporting taxes, individuals and accountants alike may be tempted to perform “creative book keeping”, interpreting the rules loosely to profit financially. It seems plausible that “creative book keeping” is likely in settings allowing one to move from negative-to-positive, from owing money to the authority to having the authority owing money to the individual. From an applied perspective, addressing this issue will allow assessing which parts of the books people are more likely to handle in such ethically lenient ways, which seems a promising venture for future research. Such research will be valuable in allowing better system designs (Thaler & Sunstein, 2008) that may increase ethical behavior. One possible avenue to consider is algorithms targeted at detecting tax fraud. Such algorithms may benefit from insights relating to people likelihood to try turn a bottom line leading them to owe money to the authorities to a (“winning”) bottom line entitling them with a tax refund. Future research is needed to provide better estimates for such detection algorithms.

The current results point to people’s sensitivity to the likelihood that their actions will turn a negative-to-positive. People take action when they have high rather than low control over the desired outcomes (Keltner, Gruenfeld, & Anderson, 2003; Galinsky, Gruenfeld, & Magee, 2003; De Dreu & Van Kleef, 2004; Van Kleef et al., 2006; Handgraaf et al., 2008) suggesting that people are more likely to attempt turning negative-to-positive when acting rather than avoiding action (omission vs. commission; Ritov & Baron, 1990; 1992; 1995; Baron & Ritov, 1994). An interesting possibility to address in future work is whether people do not only act in an attempt to turn a negative-to-positive but also endorse defaults allowing a negative-to-positive modification without their interference. Such research will allow addressing whether turning negative-to-positive is restricted to acts of commission or whether it also exists in omissions.

Another avenue for future research is to assess whether people use private justifications to dishonestly turn a negative-to-positive in settings influencing not only their own outcomes but also the outcomes of others. Past work suggests that people indeed take into consideration how their (unethical) behavior will influence others (Van Lange, 1999; De Dreu & Van Lange, 1995; Van Dijk, De Cremer & Handgraaf, 2004; Van Dijk & Vermunt, 2000; Koning, Van Dijk, Van Beest & Steinel, 2011; Steinel & De Dreu, 2004) and how these others expect them to behave (Dana, Cain & Dawes, 2006). For example, Gino and Pierce (2009) found that people tend to lie in order to restore equity between themselves and others: Negative inequity (having less than the other) evokes envy and cheating to hurt the other, while positive inequity (having more than the other) evokes guilt and unethical helping behavior. Recently, Xiao (2011) demonstrated that having to justify one’s decision to others makes people more likely to cooperate with those others. Would people also engage in unethical cooperation? And would such unethical cooperation be more likely when people could justify their lies by the fact that they turned another person’s outcome from negative-to-positive? Will such unethical cooperation be restricted to helping the people from the performer’s close surroundings (e.g., family members, friends, colleagues)? Alternatively, will using justified unethical cooperation may extend to settings in which the other is merely singled out of the crowd as a person in need for help (i.e., an identified victim; Kogut 2011a; 2011b, Kogut & Ritov, 2005; 2007; Slovic, 2007; Small & Loewenstein, 2003)? Whether settings in which people’s behavior affects others around them are more likely to push them to use unethical justifications, such as turning a negative-to-positive by bending ethical rules, remains an open question.

4.1 Conclusion

The desire to hold as many “winners” in one’s hands as one possibly can, seems reasonable, but sometimes comes at a cost. Evidence provided here suggests that people are sensitive to their likelihood to turn a negative outcome into a positive one, whether by performing well on a task or by lying to secure desired outcomes. In an attempt to increase the likelihood of getting a positive outcome, people feel justified to pursue all means to turn negative-to-positive, even at the cost of sacrificing their honesty.

References

Amar, M., Ayal, S., Rick, S., Cryder, C., & Ariely, D. (2011). Winning the battle but losing the war: The psychology of debt management. Journal of Marketing Research, 48, S38–S50.

Ariely, D. (2008). Predictably irrational: The hidden forces that shape our decisions. New York: HarperCollins.

Atanasov, P. & Dana, J. (2011). Leveling the playing field: Dishonesty in the face of threat. Journal of Economic Psychology, 32, 809–817.

Ayal, S., & Gino, F. (2011). Honest rationales for dishonest behavior. In M. Mikulincer & P. R. Shaver (Eds.), The Social Psychology of Morality: Exploring the Causes of Good and Evil. Washington, DC: American Psychological Association.

Barkan, R., Ayal, S., Gino, F., & Ariely, D. (2012). The pot calling the kettle black: Distancing response to ethical dissonance. Journal of Experimental Psychology: General, http://dx.doi.org/ 10.1037/a0027588.

Batson, C. D., Kobrynowicz, D., Dinnerstein, J. L., Kampf, H. C., & Wilson, A. D. (1997). In a very different voice: Unmasking moral hypocrisy. Journal of Personality and Social Psychology, 72, 1335–1348.

Baron, J. & Ritov, I. (1994). Reference points and omission bias. Organizational Behavior and Human Decision Processes, 59, 475498.

Bazerman, M. H., & Banaji, M. R. (2004). The social psychology of ordinary unethical behavior. Social Justice Research, 17, 111–115.

Bazerman, M. H., Loewenstein, G., & Moore, D. A. (2002). Why good accountants do bad audits. Harvard Business Review, 80, 87–102.

Bazerman, M. H., & Tenbrunsel, A. E. (2011). Blind spots: Why we fail to do what’s right and what to do about it, NJ: Princeton University Press.

Bocchiaro, P., Zimbardo, P. G., & Van Lange, P. A. M. (2012). To defy or not to defy: An experimental study of the dynamics of disobedience and whistle-blowing. Social Influence, 7, 35–50

Chance, Z., Norton, M. I., Gino, F. & Ariely, D. (2011). Temporal view of the costs and benefits of self-deception. Proceedings of the National Academy of Sciences, 108, 15655–15659.

Chugh, D., Bazerman, M. H., & Banaji, M. R. (2005). Bounded ethicality as a psychological barrier to recognizing conflict of interest, in D. A. Moore, D. M. Cain, G. Loewenstein, G., & M. H. Bazerman (Eds.), Conflict of interest: Challenges and solutions in business, law, medicine, and policy, pp. 74–95, NY, NY: Cambridge University Press.

Becker, G. S. (1968). Crime and Punishment: An Economic Approach. Journal of Political Economy, 76, 169–217.

Crocker, K. J. & Morgan, J. (1998). Is honesty the best policy? Curtailing insurance fraud through optimal incentive contracts. The Journal of Political Economy, 106, 355–375.

Dana, J., Cain, D. M., & Dawes, R. M. (2006). What you don’t know won’t hurt me: Costly (but quiet) exit in dictator games. Organizational Behavior and Human Decision Processes, 100, 193–201.

De Cremer, D. (2009). Psychology and ethics: What it takes to feel ethical when being unethical. In D. De Cremer (Ed.). Psychological Perspectives on Ethical Behavior and Decision Making, Greenwich, CT: Information Age Publishing.

De Dreu, C. K. W., & Van Kleef, G. A. (2004). The influence of power on the information search, impression formation, and demands in negotiation. Journal of Experimental Social Psychology, 40, 303–319.

De Dreu, C. K. W. & Van Lange, P. A. M. (1995). The impact of social value orientations on negotiator cognition and behavior. Personality and Social Psychology Bulletin, 21, 1178–1188.

DePaulo, B. M., Kashy, D. A., Kirkendol, S. E., Wyer, M. M., & Epstein, J. A. (1996). Lying in everyday life. Journal of Personality and Social Psychology, 70, 979–995.

Dhar, R. & Wertenbroch, K. (2012). Self-signaling and the costs and benefits of temptation in consumer choice. Journal of Marketing Research, 49, 15–25.

Epstude, K. & Roese, N. J. (2008). The functional theory of counterfactual thinking. Personality and Social Psychology Review, 12, 168–192.

Elliot, A. J. & Church, M. A. (1997). A hierarchical model of approach and avoidance achievement motivation. Journal of Personality and Social Psychology, 72. 218–232.

Elliot, A. J., & Harackiewicz, J. M. (1996). Approach and avoidance achievement goals and intrinsic motivation: A mediational analysis. Journal of Personality and Social Psychology, 70, 968–980.

Fischbacher, U., & Heusi, F. (2008). Lies in disguise, an experimental study on cheating. TWI Working Paper 40, Thurgau Institute of Economics, University of Konstanz.

Förster, J., Higgins, E. T., & Idson, L. C. (1998). Approach and avoidance strength through goal attainment: Regulatory focus and the “goal looms larger” effect. Journal of Personality and Social Psychology, 75, 1115-1131.

Galinsky A. D., Gruenfeld D. H., & Magee J. C. (2003). From power to action. Journal of Personality and Social Psychology, 85, 453–466.

Gibson, K. W., & Murnighan, J. K. (2009). From theory to practice: Messick and morality, in R. Kramer, A. Tenbrunsel, and M. H. Bazerman (Eds.), Social decision making: Social dilemmas, social values, and ethics, pp. 265–290, Psychology Press.

Gino, F. & Ariely, D. (2012). The dark side of creativity: Original thinkers can be more dishonest. Journal of Personality and Social Psychology,102, 445–459.

Gino, F., Ayal, S., & Ariely, D. (2009). Contagion and differentiation in unethical behavior: The effect of one bad apple on the barrel. Psychological Science, 20, 393–398.

Gino, F., & Pierce, L. (2009). Dishonesty in the name of equity. Psychological Science, 20, 9, 1153–1160.

Gneezy, U. (2005). Deception: The role of consequences. The American Economic Review, 95, 384–394.

Greene, J. D., & Paxton, J. M. (2009). Patterns of neural activity associated with honest and dishonest moral decisions. Proceedings of the National Academy of Sciences USA, 106, 12506–12511.

Gunia, B. C., Wang, L., Huang, L., Wang, J., & Murnighan, J. K. (2012). Contemplation and conversation: subtle Influences on moral decision making. Academy of Management Journal, 55, 13–33.

Haidt, J. (2007). The new synthesis in moral psychology. Science, 316, 998–1002.

Handgraaf, M. J. J., Van Dijk, E., Vermunt, R. C., Wilke, H. A. & De Dreu, C. K. W. (2008). Less power or powerless? Egocentric empathy gaps and the irony of having little versus no power in social decision making. Journal of Personality and Social Psychology, 95, 1136–1149.

Hao, L. & Hauser, D. (2010). Honest lies. GMU Working Paper in Economics, No. 11–16.

Higgins, E. T. (1997). Beyond pleasure and pain. American Psychologist, 52, 1280–1300.

Jiang, T. (2012). The mind game: Invisible cheating and inferable intentions, TIBER working paper.

Jones, T. M. (1991). Ethical decision-making by individuals in organizations: An issue-contingent model. Academy of Management Review, 16, 366–395

Kahneman, D. & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica 47, 263–291.

Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110, 265–284.

Kern, M. C., & Chugh, D. (2009). Bounded Ethicality: The Perils of Loss Framing. Psychological Science, 20, 378–384.

Kirchler, E., Hoelzl, E. & Wahl, I. (2008). Enforced versus voluntary tax compliance: The “slippery slope” framework. Journal of Economic Psychology, 29, 210–225.

Kirchler, E. (2007). The economic psychology of tax behavior, Cambridge University Press, Cambridge.

Kirchler, E. (1999). Reactance to taxation: Employers’ attitudes towards taxes. Journal of Socio-Economics, 28, 131–138.

Kogut, T. (2011a). The role of perspective taking and emotions in punishing identified and unidentified wrongdoers. Cognition and Emotion,25, 1491–1499.

Kogut, T. (2011b). Someone to blame: When identifying a victim decreases helping. Journal of Experimental Social Psychology, 47, 748–755.

Kogut, T. & Ritov, I. (2005). The “Identified Victim Effect”: An identified group, or just a single individual? Journal of Behavioral Decision Making, 18, 157–167.

Kogut, T. & Ritov, I. (2007). Saving one of us: Outstanding willingness to help rescue a single identified compatriot. Organizational Behavior and Human Decision Processes, 104, 150–157

Koning, L., Van Dijk, E., Van Beest, I., & Steinel, W. (2010). An instrumental account of deception and reactions to deceit in bargaining. Business Ethics Quarterly, 20, 57–73.

Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 480–498.

Lammers, J., Stapel, D. A., & Galinsky, A. D. (2010). Power Increases Hypocrisy Moralizing in Reasoning, Immorality in Behavior. Psychological Science, 21, 737–744.

Lewis, A., Bardis, A., Flint, C. Mason, C., Smith, N., Tickle, C., & Zinser, J. (2012). Drawing the line somewhere: An experimental study of moral compromise. Journal of Economic Psychology, 33, 718–725.

Lundquist, T., Ellingson, T., & Johannesson, M. (2009). The Aversion to Lying. Journal of Economic Behavior and Organization, 70, 81–92.

Markman, K. D., Gavanski, I., Sherman, S. J. & McMullen, M. N. (1993). The mental simulation of better and worse possible worlds. Journal of Experimental Social Psychology, 29, 87–109.

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.

Payne, J. W. (2005). It is Whether You Win or Lose: The Importance of the Overall Probabilities of Winning or Losing in Risky Choice. The Journal of Risk and Uncertainty, 30, 5–19.

Payne, J. W., Samper, A., Bettman, J. R., & Luce, M. F. (2008). Boundary conditions on unconscious thought in complex decision making. Psychological Science, 19, 1118–1123.

Ritov, I., & Baron, J. (1990) Reluctance to vaccinate: omission bias and ambiguity. Journal of Behavioral Decision Making, 3, 263–277.

Ritov, I., & Baron, J. (1992) Status-quo and omission biases. Journal of Risk and Uncertainty, 5, 49–61.

Ritov, I., & Baron, J. (1995). Outcome knowledge, regret, and omission bias. Organizational Behavior and Human Decision Processes, 64, 119–127.

Roese, N. (1997). Counterfactual thinking. Psychological Bulletin, 121, 133–148.

Schurr, A., Ritov, I., Kareev, Y. & Avrahami, J. (2012). Is that the answer you had in mind? The effect of perspective on unethical behavior, The Center for the Study of Rationality, Hebrew University of Jerusalem, working paper.

Schweitzer, M. E. & Hsee, C. K. (2002). Stretching the truth: Elastic justification and motivated communication of uncertain information. The Journal of Risk and Uncertainty, 25, 185–201.

Schweitzer, M. E., Ord’o nez, L., & Douma, B. (2004). Goal setting as a motivator of unethical behavior. Academy of Management Journal, 47, 422–432.

Shafir, E., Simonson, I., & Tversky, A. (1993). Reason-based choice. Cognition, 49, 11–36.

Shalvi, S., Dana, J., Handgraaf, M. J. J., & De Dreu, C. K. W. (2011). Justified Ethicality: Observing Desired Counterfactuals Modifies Ethical Perceptions and Behavior. Organizational Behavior and Human Decision Processes, 115, 181–190.

Shalvi, S., Eldar, O. & Bereby-Meyer, Y. (in press). Honesty requires time (and lack of justifications). Psychological Science.

Shalvi, S., Handgraaf, M. J. J., & De Dreu, C. K. W. (2011). Ethical maneuvering: Why people avoid both major and minor lies. British Journal of Management, 22, s16-s27.

Slovic, P. (2007). If I look at the mass I will never act: Psychic numbing and genocide. Judgment and Decision Making, 2, 1–17.

Slovic, P., Fischhoff, B., & Lichtenstein, S. (2001). Facts and fears: Understanding perceived risk. In P. Slovic (Ed.), The Perception of Risk (pp. 137–153). London : Earthscan.

Small, D. A. & Loewenstein, G. (2003). Helping a victim or helping the victim: Altruism and identifiability. Journal of Risk and Uncertainty, 26, 5–16.

Steinel, W., & De Dreu, C. K. W. (2004). Social motives and strategic misrepresentation in social decision making. Journal of Personality and Social Psychology, 86, 419–434.

Tennyson, S. (2008). Moral, social, and economic dimensions of insurance claims fraud. Social Research, 75, 1181- 1204.

Thaler, R.H. & Sunstein, C. R. (2008). Nudge—Improving decisions about health, wealth, and happiness. NY: NY, Penguin Group.

Toure-Tillery, M. & Fishbach, A. (2011). The end justifies the means, but only in the middle. Journal of Experimental Psychology: General. http://dx.doi.org/10.1037/a0025928.

Tversky, A. & Kahneman., D. (1981). The framing of decisions and the psychology of choice. Science, 211, 453–58.

Van Dijk, E., De Cremer, D. & Handgraaf, M.J.J. (2004). Social value orientations and the strategic use of fairness in ultimatum bargaining. Journal of Experimental Social Psychology, 40, 697–707.

Van Dijk E. & Vermunt, R. (2000). Strategy and fairness in social decision making: Sometimes it pays to be powerless. Journal of Experimental Social Psychology, 36, 1–25.

Van Gelder, J. L. (2012). Beyond rational choice: The hot/cool perspective of criminal decision making. Psychology, Crime, & Law, http://dx.doi.org/10.1080/1068316X.2012.660153.

University of Amsterdam, Weesperplein 4, 1018, Amsterdam, The Netherlands. Email: s.shalvi@uva.nl.

I wish to thanks Marieke Roskes, Mariska Kret, and Ozum Saygi for their useful comments on earlier drafts of this paper. I also thank the two anonymous reviewers and the editor for their constructive feedback and guidance.

See Payne (2005) for detailed discussion comparing the alternative predictions by expected utility theory, prospect theory, and cumulative prospect theory.

The example is based on a comment made by Maya Bar-Hillel in response to Moty Amar’s presentation at the 2010 Annual Society of Judgment and Decision Making conference in St. Louis, Missouri, USA.

An additional logistic regression predicting the binary choice between the negative-to-positive pot and the other two pots from the likelihood and ability to lie, did not reveal a significant interaction effect, B = –.738 (SE = .477), Wald (1) = 2.393, p = .06 (one-tailed).

Procedure	Likelihood	Pot selected
		Positive	Neg-to-Pos	Negative
No ability to lie	High (–6)	34.9% (29/83)	41.0% (34/83)	24.1% (20/83)
	Low (–12)	53.0% (44/83)	21.7% (18/83)	25.3% (21/83)
Ability to lie	High (–6)	41.1% (37/90)	32.2% (29/90)	26.7% (24/90)
	Low (–12)	31.8% (28/88)	28.4% (25/88)	39.8% (35/88)