The frame of the game: Loss-framing increases ...

YJESP-03475; No. of pages: 6; 4C: Journal of Experimental Social Psychology xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Journal of Experimental Social Psychology journal homepage: www.elsevier.com/locate/jesp

Case Report

The frame of the game: Loss-framing increases dishonest behavior Simon Schindler a,⁎, Stefan Pfattheicher b a b

Department of Psychology, University of Kassel, Holländische Straße 36-38, 34127 Kassel, Germany Department of Psychology, University of Ulm, Albert-Einstein-Allee 47, 89069 Ulm, Germany

a r t i c l e

i n f o

Article history: Received 9 June 2016 Revised 19 September 2016 Accepted 27 September 2016 Available online xxxx Keywords: Dishonest behavior Loss aversion Framing Gain

a b s t r a c t Occasionally, people trade monetary gains for moral costs and engage in dishonest behavior. Based on research showing that people react more sensitively toward a possible loss compared to a possible gain (i.e., loss aversion), the present contribution examines the idea that people will more likely engage in dishonest behavior to reduce the extent of a loss compared to increasing the extent of a gain. In the two experimental studies, participants could engage in dishonest behavior either to avoid a loss (loss condition) or to approach an equivalent gain (gain condition). To assess dishonest behavior, a die-under-the-cup paradigm (Study 1) and a coin-toss task (Study 2) was applied. Results of both studies demonstrated the predicted effect of framing, supporting the idea that people show more dishonest behavior to avoid a loss compared to approaching an equivalent gain. © 2016 Elsevier Inc. All rights reserved.

“Possession does not make us half as happy as loss makes us unhappy.” [Jean Paul F. Richter] Imagine the yearly tax declaration of an employee: Taxes have been paid in advance, and reporting the declaration offers the possibility to gain some money back. Now imagine the tax declaration of a selfemployed person: not all taxes have been paid in advance, and releasing the declaration means that additional money has to be paid. Which of the two tax declarations is more likely to contain dishonest information (cf. Mazur & Plumley, 2007; Robben et al., 1990)? Previous research by Engström, Nordblom, Ohlsson, and Persson (2015) on taxpaying found that Swedish taxpayers react differently depending on the reference point: If the reference point indicates having to pay additional money (i.e., loss), taxpayers are more likely to claim deductions compared to when the reference point indicates getting money back (i.e., gain). Although Engström and colleagues assume that loss aversion is the driving force behind this effect, they are not able to present direct evidence for this claim. With the present work, we directly address the idea that people show more dishonest behavior to reduce a possible loss compared to increasing an equivalent gain.

⁎ Corresponding author. E-mail addresses: [email protected] (S. Schindler), [email protected] (S. Pfattheicher).

1. Dishonest behavior and its incentives Traditional economic models assume people to cheat according to expected utility, that is, a (monetary) cost-benefit calculation of expected punishment when getting caught versus the possible gain of cheating (e.g., Becker, 1968). Recent literature has extended this perspective, showing that costs of cheating also include the possible erosion of one's positive self-concept (Abeler, Becker, & Falk, 2014; Mazar, Amir, & Ariely, 2008; Mazar & Ariely, 2006; for a review see Rosenbaum, Billinger, & Stieglitz, 2014). Surveys have indicated honesty to be one of the most important values in a person's life (e.g., Geißler, Schöpe, Klewes, Rauh, & von Alemann, 2013). Research further has shown that individuals have a strong general psychological motivation to comply with their value system to maintain a positive self-concept (e.g., Baumeister, 1998; Pyszczynksi, Greenberg, Solomon, Arndt, & Schimel, 2004). Thus, when people are tempted to engage in dishonest behavior, they are simultaneously confronted with personal costs of violating the rule of honesty (e.g., experience of negative emotions; Batigalli, Dufwenberg, & Charness, 2013). Consequently, people strive to find a balance between those opposing motivational forces. In line with this reasoning, research has provided strong empirical evidence that people do not simply cheat as much as they can, even if there is no possibility of getting caught (e.g., Abeler et al., 2014; Fischbacher & Föllmi-Heusi, 2013; Gneezy, 2005; Mazar et al., 2008; Shalvi, Dana, Handgraaf, & De Dreu, 2011; for a recent meta-analysis, see Abeler, Nosenzo, & Raymond, 2016). Moreover, dishonest behavior was found to depend on personality traits (e.g., honesty–humility; Hilbig & Zettler, 2015), as well as on

http://dx.doi.org/10.1016/j.jesp.2016.09.009 0022-1031/© 2016 Elsevier Inc. All rights reserved.

Please cite this article as: Schindler, S., & Pfattheicher, S., The frame of the game: Loss-framing increases dishonest behavior, Journal of Experimental Social Psychology (2016), http://dx.doi.org/10.1016/j.jesp.2016.09.009

2

S. Schindler, S. Pfattheicher / Journal of Experimental Social Psychology xxx (2016) xxx–xxx

contextual cues, such as priming (Mazar et al., 2008) or being treated unfairly (Houser, Vetter, & Winter, 2012). Going back to the tax example above, it is unclear, however, whether people will cheat more to reduce a possible loss compared to increasing a gain. 2. Framing matters: the impact of loss aversion It is a basic psychological principle that people perceive losses as more unattractive than they perceive gains as attractive. Evidence for this asymmetry is provided by psychological research on negativity bias (for reviews, see Baumeister, Bratslavsky, Finkenauer, & Vohs, 2001; Rozin & Royzman, 2001; Vaish, Grossmann, & Woodward, 2008), and by economic research on framing, reference points and the endowment effect (e.g., Kahneman, Knetsch, & Thaler, 1991; Kahneman & Tversky, 1979, 1984). Numerous studies, for example, evidenced higher risk seeking to be related to avoiding possible losses compared to increasing possible gains (e.g., Kahneman & Tversky, 1979; Tversky & Kahneman, 1992). Additionally, McGraw, Larsen, Kahneman, and Schkade (2010) found that participants reported increased distress while thinking about having lost an amount of money compared to the excitement about winning the same amount. The expected pain of losing something can therefore be assumed to surpass the enjoyment of gaining. Whether a situation is framed as involving either a potential loss or a potential gain was further found to affect decision making (Tversky & Kahneman, 1981; for reviews, see e.g., Kühberger, 1998; for a typology of framing effects, see Levin, Schneider, & Gaeth, 1998), supporting the general idea of loss aversion, meaning that people are more motivated to avoid losses than to approach equivalent potential gains.1 3. The current research Research on cheating has mainly studied situations in which people can cheat to gain more money (e.g., Mazar et al., 2008). Some research, however, studied situations in which people can cheat to avoid losing money (e.g., Hilbig & Zettler, 2015, Study 1; Hershfield, Cohen, & Thompson, 2012, Study 4; Yaniv & Siniver, 2016). Based on the assumption that the loss frame induces higher motivation to reduce the loss than does the gain frame to increase the gain, we investigate the following hypothesis: If a situation is framed as one in which a participant is given the opportunity to reduce a possible loss by cheating, it is more likely that cheating will occur compared to a situation in which one is given the opportunity to increase a possible gain by cheating. Regarding the question about how potential gains and losses affect dishonest behavior, several works are worth mentioning. In a series of three experiments, Kern and Chugh (2009) showed that unethical behavior is increased when outcomes are framed as possible losses compared to possible gains, especially under time pressure. However, these studies referred exclusively to hypothetical scenarios to assess unethical behavior. Further, unethical decisions were not monetarily incentivized and did not refer to actual cheating. These limitations are partly addressed by Cameron and Miller (2008, Study 2; also cited in Cameron & Miller, 2009). They found that in providing participants the opportunity to cheat (by allowing them to self-report their performance on an anagram task and to pay themselves accordingly), participants indicated a higher performance when that performance was linked to a reduction of the loss of the previously allotted ten dollars, 1 As there are several conceptualizations of framing (Levin et al., 1998), we want to note that by loss and gain framing we mean that the reference point (e.g., money provided ex ante vs. no money provided) either implies a potential loss or a potential gain (for this approach, see also Grolleau et al., in press). We thus refer to two different situations. As such, we deviate from the original work on framing effects (e.g., the classic study on the Asian disease problem; cf. Tversky & Kahneman, 1981), in which one situation is described either as in a positive way (e.g., 200 out of 600 people are saved) or in a negative way (e.g., 400 out of 600 people will die).

compared to when performance was linked to a gain of money. To rule out the possibility that these effects are merely driven by higher effort in the loss-condition, Grolleau, Kocher, and Sutan (in press) investigated the effect of loss aversion on cheating, also by using a performance-based cheating paradigm (self-reported performance of solved matrix tasks); additionally, performance in the tasks was explicitly monitored or not. Participants were thus given the possibility to cheat only in the latter condition. In the condition in which participants were not monitored, they were more likely to cheat in the loss frame (compared to the gain frame). Performance between loss- and gainconditions was not significantly different when performance was monitored, ruling out the possibility that these effects are driven merely by higher effort in the loss-condition. So far, hypothetical scenarios (Kern & Chugh, 2009) as well as performance tasks (Cameron & Miller, 2008; Grolleau et al., in press) have been used for assessing effects of loss-framing on dishonesty. We extend this line of research in valuable ways. First, we investigate actual dishonest behavior instead of assessing hypothetical behavior. Further, we investigate cheating in a non-performance context. In this way, we can draw conclusions that do not apply only to cheating in performance situations. Additionally, research indicates dishonesty to be higher in a performance compared to a non-performance based cheating paradigm (Gravert, 2013). We extend this line of research showing that cheating even occurs in non-performance situations to a substantial degree, in particular when people can avoid losses. In fact, it is important to investigate the impact of loss-framing on cheating outside the performance context, given that other research shows effects of loss-framing on performance (e.g., Shah, Higgins, & Friedman, 1998). Thus, to further exclude the possibility that actual better performance in a loss frame explains the outlined results, we investigate cheating in a non-performance context. Finally, we consider conceptual replications having an important value in themselves (cf. Brandt et al., 2014; Crandall & Sherman, 2016), especially regarding the ongoing debate about replicability of psychological findings (e.g., Makel, Plucker, & Hegarty, 2012; Open Science Collaboration, 2015).

4. Study 1 To investigate our idea, in Study 1 we relied on a dice task paradigm (Fischbacher & Föllmi-Heusi, 2013). Dice tasks are commonly used in cheating research (Moshagen & Hilbig, in press). The advantage of these tasks is that the expected value serves as a statistical baseline for honest behavior. In most studies, a dice is rolled once. That is, the probability of rolling a certain number is 1/6. With this paradigm it is intended to detect dishonest behavior on the aggregate, because typically full anonymity is provided making it impossible to detect dishonest behavior on an individual level. For the current study, however, we used a multi-round task: participants should report the number of rolled ‘4s’ after having rolled a fair, six-sided die 75 times.2 By assuring full anonymity, they are given the opportunity to cheat by being able to report any number they want (i.e., a continuous cheating range). Due to the known probability of 1/6 to throw a ‘4,’ the expected value of 12.5 (75*1/6) serves as the baseline. With this paradigm it is intended to detect dishonest behavior on the aggregate, that is, by comparing the group means of the reported number of ‘4s’ (also regarding the statistical baseline). Given that in such a multi-round task each participant has the chance to cheat more than once, we suppose this assessment would provide a more reliable estimate of cheating compared to one-shot tasks. 2 In total, we used 18 dice in this experiment. For testing whether the dice were actually fair, six dice were randomly chosen. Each was then rolled a total of 1998 times. Thus, each outcome was expected to occur 333 times (16.67%). This total number of rolls provides optimal power (1 – β N 0.95) to detect even small (w = 0.1) deviations from an equal distribution (cf. Hilbig & Zettler, 2015). As suggested by non-significant χ2-tests (ps N 0.305), all dice used in the following experiments can be considered fair.

Please cite this article as: Schindler, S., & Pfattheicher, S., The frame of the game: Loss-framing increases dishonest behavior, Journal of Experimental Social Psychology (2016), http://dx.doi.org/10.1016/j.jesp.2016.09.009

S. Schindler, S. Pfattheicher / Journal of Experimental Social Psychology xxx (2016) xxx–xxx

4.1. Method 4.1.1. Subjects and design Power was set to 0.80 (Cohen, 1988) and sample size to detect a medium effect (d = 0.5) was calculated. Using G*Power (Faul, Erdfelder, Buchner, & Lang, 2009), a power analysis revealed a required sample size of N = 102 to detect a significant effect (alpha level of 0.05, onetailed) given there is a true effect. In two days, we recruited 90 participants on the campus of a German university (53 females; 84 students; Mage = 23.83, SD = 5.33, range: 18–54). Three participants were excluded because it was strongly suspected that they had not read or understood the instructions of the dice task (false indication of the gained or lost amount of money; see description of the procedure below).3 Furthermore, one participant did not give any indication, leaving a total of 86 participants. Each participant was randomly assigned to either the loss- (n = 39) or the gain-framing condition (n = 47). Participants in the study (also in Study 2) had given written informed consent prior to participating in the study, in line with APA's Ethical Principles of Psychologists and Code of Conduct. 4.1.2. Materials and procedure The entire study took approximately 5 to 10 min. Participation was compensated with one Euro. Participants came into the lab and were assigned to a chair by the experimenter. As dishonest behavior is unlikely when chances of detection are high (e.g., Mazar et al., 2008), participants completed the study in cubicles to increase anonymity. Every 30 min, we switched the experimental condition to which participants were assigned. In both conditions, on the table in front of them, they found three fair, six-sided dice in an opaque cup, and the packet of material containing three sheets: a cover sheet, instructions for the dice task, and a questionnaire for demographic assessments. On the cover sheet, participants in the gain-frame condition read that they can gain up to 7.50 Euro (~8 USD) in a dice-task. In the loss-frame condition, participants were provided with an envelope containing 75 ten-cent coins. On the cover sheet, they read that they would receive 7.50 Euro and that the money belonged to them. They further read, however, that they had to participate in a dice task in which they could lose some of the money. Then, the dice task was introduced. In both conditions, participants were told to roll the three dice 25 times, to count the number of rolled ‘4s,’ and to write down the total number when finished.4 In the gainframe condition, participants read that for every rolled ‘4,’ they would gain 10 cents. In the loss-frame condition, participants read that for every rolled number except ‘4,’ they would lose 10 cents of the 7.50 Euro. To be able to control for accurate understanding of the instructions, we additionally asked participants to write down the amount of money they finally gained or they finally have to give back, respectively (as already mentioned above, three participants were excluded because the number of indicated ‘4s’ did not correspond to the indication of gained or lost amount of money). After the dice task, in the gain-frame condition, they read that they should now take their gained money out of the cashbox positioned in the room behind another dividing wall to exclude the possibility of demanding effects (i.e., through a feeling of being watched by the experimenter; Pfattheicher & Keller, 2015). In the loss-frame condition, they read that they should put their lost amount of money (the respective amount of ten-cent coins) in the cashbox and were told that they could keep the left amount. After having done so, all participants went back to their chair (with their final amount of money) and filled out

3 Analysis including the three participants also revealed the expected significant effect of framing, F(1,87) = 4.31, p = 0.041. 4 The number ‘4’ was chosen to increase the overall likelihood of cheating because research has indicated higher chances of cheating when ‘4’ or ‘3’ are the winning numbers (Hilbig & Hessler, 2013).

3

the demographic measures.5 Finally, in addition to the amount of money, participants received one Euro and were given the opportunity to leave their email address for a full debriefing. All measures, manipulations, and exclusions in this study (and also in Study 2) are reported.

4.2. Results & discussion Overall, the mean reported number of rolled ‘4s’ was 12.70 (SD = 3.28; range: 6 to 21) and did not differ significantly from the expected value of 12.5, Z = −0.49, p = 0.625. To test our hypothesis that a possible loss (vs. gain) increases cheating, we ran an ANOVA with framing as independent variable and the reported number of rolled ‘4s’ as the dependent variable. The analysis yielded a significant effect of framing, F(1,84) = 6.63, p = 0.012, d = 0.56.6 As expected, participants in the loss-frame condition reported a higher number of rolled ‘4 s,’ on average (M = 13.67, SD = 3.22), compared to participants in the gain-frame condition (M = 11.89, SD = 3.15). Additional analysis revealed that the mean reported number in the loss-frame condition differed significantly from the expected value of 12.5, Z = −2.16, p = 0.031, indicating dishonest behavior. The mean reported number in the gain-frame condition was not significantly different from 12.5, Z = −1.31, p = 0.191. Our measurement further allows computing an individual index of dishonesty. We classified subjects as honest and dishonest according to the likelihood that their reported scores were honest (cf. Halevy, Shalvi, & Verschuere, 2014). Taking an alpha of 10% (1.25), 36 subjects (41.9% of the sample) showed a score which was higher than the score predicted by chance (12.5 + 1.25 = 13.75). Using these two groups as a binary dependent variable revealed that there are significant more dishonest participants in the loss-frame condition (56.4%) than in the gain-frame condition (29.8%), χ2 = 6.21, p = 0.013. Analysing the numbers of rolled ‘4s’ of only the dishonest people did not yield a significant framing effect, F b 1. In sum, results provide evidence for the idea that people show higher levels of dishonest behavior to avoid a loss than to approach an equivalent gain. Additionally, and in line with our theoretical reasoning, our analyses suggest that this effect occurred because many participants cheated a bit more in the loss-frame condition, and not because some participants decided to cheat a lot. The mean in the gain-frame condition did not significantly differ from the baseline level. It is thus unclear whether cheating occurred in the gain-frame condition. This finding is somewhat surprising given that people are usually easily tempted to cheat when anonymity is provided (e.g., Mazar et al., 2008). However, Abeler et al. (2014, see also Gächter & Schulz, 2016), for example, found evidence for overall truthful behavior in a German representative sample, suggesting that psychological costs should not be underestimated (see also General discussion). It should be further noted that regarding the repeated dice rolling procedure in our cheating paradigm, one's payoff is linked to the extent of dishonesty. That is, high levels of deception go hand in hand with feelings of self-incrimination, presumably producing effects of social desirability (Hilbig & Zettler, 2015). To exclude such effects, a paradigm is needed in which the responses of any one individual can never be conclusively linked to dishonesty although the degree of dishonesty can still be estimated on the aggregate. In the next study, we applied such a paradigm.

5 Given participants' full anonymity behind the dividing wall, it was theoretically possible for them to take more money out of the cashbox than actually reported (gain condition) or leave more money in the envelope (loss condition). However, the comparison of the sum of all participants' reported final payment (depending on the number of rolled ‘4s’) to the amount of money that was left after the whole experiment did not show any discrepancy. Thus, we can exclude the possibility that participants cheated in this part of the procedure. 6 A non-parametric Wilcoxon test also revealed the expected significant effect of framing, Z = 2.47, p = 0.013.

Please cite this article as: Schindler, S., & Pfattheicher, S., The frame of the game: Loss-framing increases dishonest behavior, Journal of Experimental Social Psychology (2016), http://dx.doi.org/10.1016/j.jesp.2016.09.009

4

S. Schindler, S. Pfattheicher / Journal of Experimental Social Psychology xxx (2016) xxx–xxx

5. Study 2 In this study, we aimed to replicate the findings by applying a common one-shot cheating paradigm (Rosenbaum et al., 2014) in which participants self-report the outcome of tossing a coin (e.g., Abeler et al., 2014). As in Study 1, our setup ensures that the true outcome is only known to the participant. Thereby, participants have both the opportunity and incentive to cheat by claiming to have obtained the target outcome, despite actually having obtained a different outcome. At the same time, the proportion of dishonest individuals can be inferred from these observed responses at the aggregate level, given the probability distribution of 0.5 when tossing a coin (Moshagen, Hilbig, Erdfelder, & Moritz, 2014). 5.1. Method 5.1.1. Subjects and design Using a coin-toss paradigm reduces power, because it adds random noise to the response (Ulrich, Schröter, Striegel, & Simon, 2012). Therefore, compared to standard correlation and regression analyses, larger sample sizes are required to obtain the same level of precision. Power was set to 0.80 (Cohen, 1988) and a proportion of dishonest respondents of 0.2 was assumed (cf. Moshagen & Hilbig, in press). On this basis, sample size for a medium effect (odds ratio = 2.5; Rosenthal, 1996) was calculated. Using the R package RReg (Heck & Moshagen, 2016), the power analysis revealed a required sample size of N = 300 to detect a significant effect (alpha level of 0.05, one-tailed) given there is a true effect. Accordingly, we recruited 300 participants (128 females; Mage = 33.27, SD = 10.57, range: 18–68) from Amazon Mechanical Turk. Each participant was randomly assigned to the loss- (n = 146) or the gain-framing condition (n = 154). 5.1.2. Materials and procedure The entire study took approximately 5 min. Participation was basically compensated with $US 0.25. On the first page, participants in the loss-frame condition read that they would receive $0.50 extra bonus. Participants in the gain-frame condition were not given this information. After having read and agreed to the informed consent, participants were introduced to the coin-toss task. They were asked to make sure that they have a coin and that they should flip it one time. Participants in the loss-frame condition were given the information that if “heads”/ “tails” turns up, they would lose/could keep the $0.50 extra bonus. In the gain-frame condition, they were told that they would receive no $0.50 extra bonus/would receive a $0.50 extra bonus. We randomized the high-payoff outcome (i.e., whether heads or tails wins). Next, they were asked to flip the coin and to indicate whether it turned up heads or tails. Finally, the demographics were assessed.7 5.2. Results & discussion Overall, 76.7% of all participants reported have flipped the high-payoff outcome which is significantly different from the expected outcome of a fair coin, Z = 10.94, p b 0.001. Theoretically, with fair coins and large samples, half of the subjects should flip the high-payoff outcome anyway, having no monetary incentive to cheat. Only among the other half who flip the low-payoff outcome, some might cheat. Following the guidelines of Moshagen and Hilbig (in press), calculations revealed that an estimated 53.4% [(0.767–0.50)/(1–0.50)] in our sample are prepared to cheat if they did not actually win. An estimated 26.7% in our

7

After the task, for exploratory reasons, we included a questionnaire on regulatory focus (prevention and promotion focus; Lockwood, Jordan, & Kunda, 2002). Results of framing are not affected in any regard (in terms of direction and significance levels) when controlling for prevention or promotion focus.

sample actually cheated (0.767–0.50) and the estimated proportion of illegitimate wins amounts to 34.8% [(0.767–0.50)/0.767].8 Moshagen and Hilbig (in press) state that typical analyses of binary cheating paradigms ignore that the observed win-response is contaminated by honest respondents, leading to substantially underestimated effects. Thus, to adequately test our hypothesis that a possible loss (vs. gain) increases cheating, we conducted a logistic regression analysis using the R package RRreg (Heck & Moshagen, 2016). Results show the expected framing effect on the coin-toss outcome (high-payoff outcome: yes vs. no), b = 0.99, SE b = 0.41, Wald's χ2 = 5.88, p = 0.013, odds ratio = 2.70.9 In line with our hypothesis, 82.9% of the participants in the loss-frame condition flipped the high-payoff outcome (proportion of actual cheaters = 32.9%; proportion of illegitimate wins = 41.5%) compared to only 70.8% in the gain-frame condition (proportion of actual cheaters = 20.8%; proportion of illegitimate wins = 35.4%). In line with previous research showing that people are easily tempted to cheat when anonymity is provided (e.g., Mazar et al., 2008), we found a general tendency to cheat. More important however, results provide further strong evidence for the idea that people show high levels of dishonest behavior to avoid a loss than to approach a gain. It is noteworthy, that the probability of cheating was more than twice a large in the loss-frame condition.

6. General discussion The present work addresses the hypothesis of whether dishonest behavior is more likely when a situation implies a potential loss compared to a potential gain. In line with this prediction, in our studies participants reported higher numbers of rolled ‘4s’ and flipped more often the high-payoff outcome to avoid a loss than to approach an equivalent gain. That is, the loss frame increased the likelihood of cheating when it was linked to reducing the loss of one's own endowment. In both studies the effect of framing on cheating had a medium size. With these findings, the present work goes beyond previous research (Cameron & Miller, 2008; Grolleau et al., in press; Kern & Chugh, 2009) by applying a non-performance based cheating paradigm, enabling us to make extended conclusions about the effect of loss-framing. In performancebased situations, the act of having worked on something might provide people with a strong justification for cheating, stronger than in situations outside the performance context (e.g., when reporting the outcome of a die; cf. Gravert, 2013). Nevertheless, evidence obtained in the present work suggests that loss-framing also increases dishonesty outside the performance context. In both studies, participants were paid in advance in the loss-frame condition. According to previous research (e.g., McGraw et al., 2010; Grolleau et al., in press), we assume that loss aversion of one's possession is the primary driving force for the framing effect on cheating. Although one could argue that participants never perceived actual legal endowment of the money, research has indicated that the loss of anticipatory possession (i.e., psychological ownership) can have parallel effects (Ariely & Simonson, 2003; Pierce, Kostova, & Dirks, 2001; Shu & Peck, 2011). Given these findings, it seems an interesting question whether the potential loss of anticipatory (vs. actual) possession induces the same level of motivation to reduce the loss and thus, leads to the same level of dishonesty. Referring to the two opposing factors of cheating (i.e., gains and moral costs; Mazar et al., 2008), loss aversion might induce a feeling of deservingness and being entitled to reduce the loss of endowment (Cameron & Miller, 2009), lowering the moral costs of violating the value of honesty compared to the gain frame (see also Kahneman, 8 By illegitimate wins, we mean the proportion of people who reported a ‘win’ but actually cheated, that is the proportion of people who illegitimately reported a ‘win’. 9 Additional analysis using a simple binary logistic regression analysis yielded a significant effect of framing, b = 0.69, SE b = 0.28, Wald's χ2 = 6.02, p = 0.014, odds ratio = 2.00. This was also the case when using a non-parametric Wilcoxon test, Z = 2.47, p = 0.013.

Please cite this article as: Schindler, S., & Pfattheicher, S., The frame of the game: Loss-framing increases dishonest behavior, Journal of Experimental Social Psychology (2016), http://dx.doi.org/10.1016/j.jesp.2016.09.009

S. Schindler, S. Pfattheicher / Journal of Experimental Social Psychology xxx (2016) xxx–xxx

Knetsch, & Thaler, 1986). Correspondingly, research has suggested ostracism-induced feelings of entitlement to increase cheating (Poon, Chen, & DeWall, 2013). Future research should further address these underlying mechanisms. Assuming the possible losses and gains in our study to be rather small, the found framing effect interestingly contradicts the claim of Harinck, Van Dijk, Van Beest, and Mersmann (2007) that for small amounts of money, gains loom larger than losses. The authors refer to the hedonic principle stating that people are motivated to enjoy positive outcomes and to discount negative outcomes meaning to minimize negative feelings as much as possible. According to the authors' reasoning, this should be more effective when losses are small. However, although people might well be able to anticipate the effect of coping mechanisms for small losses, loss aversion seems to be the stronger force when given the opportunity to reduce the loss by cheating, even when only small losses are at stake. When speaking of ‘small losses’, future research should take into account that the perception of what is ‘small’ is not stable across situations but rather depends on the reference point (see also below). There is growing evidence to support the idea that people avoid lies that are both too big and too small. Major lies imply very high psychological costs (i.e., violating the important value of honesty; Mazar et al., 2008), whereas minor lies imply lower psychological costs but also low benefits. In line with this reasoning, Shalvi, Handgraaf, and De Dreu (2011) found that participants avoided lying to the maximum as well as lying for only minor profit. Notably, in the study by Shalvi and colleagues, the cheating paradigm implied the possibility of gaining instead of losing money. Our results add to these findings. Specifically, we show that people engage in cheating, even for minor reasons, namely when they see the possibility of avoiding losses. Moreover, our result that no cheating occurred on average in the gain-frame condition in Study 1 is in line with the findings of Shalvi et al. (2011), as the incentives to cheat in Study 1 can be considered small. Yet despite the seemingly small incentive of $0.50, cheating nevertheless occurred to a large degree in Study 2, independent of how the task was framed. One could argue that interpreting the incentive relative to the participation payment (Study 1: 1 Euro; Study 2: $0.25) proves that incentives no longer seem that small, especially for Amazon Mturk workers (in our case, an additional amount of $0.50 means twice the amount of the participation payment). Nonetheless, future research should take into account people's perception of incentives. Finally, we want to encourage investigating loss-aversion in the face of existential threat. Literature on terror management theory, for example, suggests that being confronted with one's own death increases the motivation to live up to important salient cultural worldviews (e.g., Schindler, Reinhard, & Stahlberg, 2013). Previous work indicates that this holds for the value of materialism (Arndt, Solomon, Kasser, & Sheldon, 2004). Accordingly, one could expect existential threat to strengthen the effect of loss-aversion on dishonesty to increase material benefits. On the other hand, assuming honesty to be an important cultural value, existential threat can be plausibly expected to decrease dishonesty thus reducing the effect of loss-aversion on dishonesty. In fact, understanding the underlying processes of dishonesty and their dependence on existential threats seems an important question for future research, especially regarding present concerns of European societies about terroristic attacks. 6.1. Conclusion Reconsider the example mentioned at the beginning: if taxes already have been paid (as is usually the case for employees), then tax declaration represents a gain frame because it offers the possibility to get money back. If the declaration, however, means that additional money has to be paid (as is usually the case for self-employed persons in Germany), then it represents a loss frame. The results of the current study clearly suggest that—at least according to this example—the declaration

5

of a self-employed person is more likely to contain dishonest information. Acknowledgements We thank Morten Moshagen for his valuable comments on data analyses, and further Charlotta Völk and Hasan Avcikolu for their help with data collection. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jesp.2016.09.009. References Abeler, J., Becker, A., & Falk, A. (2014). Representative evidence on lying costs. Journal of Public Economics, 113, 96–104. Abeler, J., Nosenzo, D., & Raymond, D. (2016). Preferences for truth-telling. CeDEx working paper no. 2016-13. Ariely, D., & Simonson, I. (2003). Buying, bidding, playing, or competing? Value assessment and decision dynamics in online auctions. Journal of Consumer Psychology, 13, 113–123. Arndt, J., Solomon, S., Kasser, T., & Sheldon, K. M. (2004). The urge to splurge: A terror management account of materialism and consumer behavior. Journal of Consumer Psychology, 14, 198–212. Batigalli, P., Dufwenberg, M., & Charness, G. (2013). Deception: The role of guilt. Journal of Economic Behavior and Organization, 93, 227–232. Baumeister, R. F. (1998). The self. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (pp. 680–740). New York: McGraw-Hill. Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5, 323–370. Becker, G. S. (1968). Crime and punishment: An economic approach. Journal of Political Economy, 76, 169–217. Brandt, M. J., Ijzerman, H., Dijksterhuis, A., Farach, F. J., Geller, J., Giner-Sorolla, R., ... Van't Veer, A. (2014). The replication recipe: What makes for a convincing replication? Journal of Experimental Social Psychology, 50, 217–224. Cameron, J. S., & Miller, D. T. (2008). Deservingness and unethical behavior in loss and gain frames. Unpublished manuscript. Cameron, J. S., & Miller, D. T. (2009). Ethical standards in gain versus loss frames. In D. de Cremer (Ed.), Psychological perspectives on ethical behavior and decision making (pp. 91–106). Charlotte, NC: Information Age Publishing. Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Crandall, C. S., & Sherman, J. W. (2016). On the scientific superiority of conceptual replications for scientific progress. Journal of Experimental Social Psychology, 66, 93–99. Engström, P., Nordblom, K., Ohlsson, H., & Persson, A. (2015). Tax compliance and loss aversion. American Economic Journal: Economic Policy, 7, 132–164. Faul, F., Erdfelder, E., Buchner, A., & Lang, A. -G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149–1160. Fischbacher, U., & Föllmi-Heusi, F. (2013). Lies in disguise—An experimental study on cheating. Journal of the European Economic Association, 11, 525–547. Gächter, S., & Schulz, J. F. (2016). Intrinsic honesty and the prevalence of rule violations across societies. Nature, 531, 496–499. Geißler, H., Schöpe, S., Klewes, J., Rauh, C., & von Alemann, U. (2013). Wertestudie 2013: Wie groß ist die Kluft zwischen dem Volk und seinen Vertretern? Köln: YouGov. Gneezy, U. (2005). Deception: The role of consequences. The American Economic Review, 9, 384–394. Gravert, C. A. (2013). How luck and performance affect stealing. Journal of Economic Behavior and Organization, 93, 301–304. Grolleau, G., Kocher, M. G., & Sutan, A. Cheating and loss aversion: Do people lie more to avoid a loss? Management Science (in press). Halevy, R., Shalvi, S., & Verschuere, B. (2014). Being honest about dishonesty: Correlating self-reports and actual lying. Human Communication Research, 40, 54–72. Harinck, F., Van Dijk, E., Van Beest, I., & Mersmann, P. (2007). When gains loom larger than losses: Reversed loss aversion for small amounts of money. Psychological Science, 18, 1099–1105. Heck, D. W., & Moshagen, M. (2016). RRreg: An R package for correlation and regression analyses of randomized response data. (Manuscript submitted for publication). Hershfield, H. E., Cohen, T. R., & Thompson, L. (2012). Short horizons and tempting situations: Lack of continuity to our future selves leads to unethical decision making and behavior. Organizational Behavior and Human Decision Processes, 117, 298–310. Hilbig, B. E., & Hessler, C. M. (2013). What lies beneath: How the distance between truth and lie drives dishonesty. Journal of Experimental Social Psychology, 49, 263–266. Hilbig, B. E., & Zettler, I. (2015). When the cat's away, some mice will play: A basic trait account of dishonest behavior. Journal of Research in Personality, 57, 72–88. Houser, D., Vetter, S., & Winter, J. (2012). Fairness and cheating. European Economic Review, 56, 1645–1655. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–292.

Please cite this article as: Schindler, S., & Pfattheicher, S., The frame of the game: Loss-framing increases dishonest behavior, Journal of Experimental Social Psychology (2016), http://dx.doi.org/10.1016/j.jesp.2016.09.009

6

S. Schindler, S. Pfattheicher / Journal of Experimental Social Psychology xxx (2016) xxx–xxx

Kahneman, D., & Tversky, A. (1984). Choices, values, and frames. The American Psychologist, 39, 341–350. 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. Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The endowment effect, loss aversion, and status quo bias. The Journal of Economic Perspectives, 5, 193–206. Kern, M. C., & Chugh, D. (2009). Bounded ethicality the perils of loss framing. Psychological Science, 20, 378–384. Kühberger, A. (1998). The influence of framing on risky decisions: A meta-analysis. Organizational Behavior and Human Decision Processes, 75, 23–55. Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Processes, 76, 149–188. Lockwood, P., Jordan, C. H., & Kunda, Z. (2002). Motivation by positive and negative role models: Regulatory focus determines who will best inspire us. Journal of Personality and Social Psychology, 83, 854–864. Makel, M. C., Plucker, J. A., & Hegarty, B. (2012). Replications in psychology research: How often do they really occur? Perspectives on Psychological Science, 7, 537–542. Mazar, N., & Ariely, D. (2006). Dishonesty in everyday life and its policy implications. Journal of Public Policy and Marketing, 25, 117–126. Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: A theory of selfconcept maintenance. Journal of Marketing Research, 45, 633–644. Mazur, M. J., & Plumley, A. H. (2007). Understanding the tax gap. National Tax Journal, 60, 569–576. McGraw, A. P., Larsen, J. T., Kahneman, D., & Schkade, D. (2010). Comparing gains and losses. Psychological Science, 21, 1438–1445. Moshagen, M., & Hilbig, B. E. The statistical analysis of cheating paradigms. Behavior Research Methods (in press). Moshagen, M., Hilbig, B. E., Erdfelder, E., & Moritz, A. (2014). A strategy for validating questioning techniques that assess sensitive issues. Experimental Psychology, 61, 48–54. Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716. Pfattheicher, S., & Keller, J. (2015). The watching eyes phenomenon: The role of a sense of being seen and public self-awareness. European Journal of Social Psychology, 45, 560–566. Pierce, J. L., Kostova, T., & Dirks, K. T. (2001). Towards a theory of psychological ownership in organizations. The Academy of Management Review, 26, 298–310. Poon, K. T., Chen, Z., & DeWall, C. N. (2013). Feeling entitled to more ostracism increases dishonest behavior. Personality and Social Psychology Bulletin, 39, 1227–1239.

Pyszczynksi, T., Greenberg, J., Solomon, S., Arndt, J., & Schimel, J. (2004). Why do people need self-esteem? A theoretical and empirical review. Psychological Bulletin, 130, 435–468. Robben, H. S., Webley, P., Weigel, R. H., Wärneryd, K. E., Kinsey, K. A., Hessing, D. J., ... Long, S. B. (1990). Decision frame and opportunity as determinants of tax cheating: An international experimental study. Journal of Economic Psychology, 11, 341–364. 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, 181–196. Rosenthal, J. A. (1996). Qualitative descriptors of strength of association and effect size. Journal of Social Service Research, 21, 37–59. Rozin, P., & Royzman, E. G. (2001). Negativity bias, negativity dominance, and contagion. Personality and Social Psychology Review, 5, 296–320. Schindler, S., Reinhard, M. -A., & Stahlberg, D. (2013). Tit for tat in the face of death: The effect of mortality salience on reciprocal behavior. Journal of Experimental Social Psychology, 49, 87–92. Shah, J., Higgins, T., & Friedman, R. S. (1998). Performance incentives and means: How regulatory focus influences goal attainment. Journal of Personality and Social Psychology, 74, 285–293. Shalvi, S., Dana, J., Handgraaf, M. J., & De Dreu, C. K. W. (2011a). Justified ethicality: Observing desired counterfactuals modifies ethical perceptions and behaviour. Organizational Behavior and Human Decision Processes, 115, 181–190. Shalvi, S., Handgraaf, M. J., & De Dreu, C. K. (2011b). Ethical manoeuvring: Why people avoid both major and minor lies. British Journal of Management, 22, 16–27. Shu, S. B., & Peck, J. (2011). Psychological ownership and affective reaction: Emotional attachment process variables and the endowment effect. Journal of Consumer Psychology, 21, 439–452. Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211, 453–458. Tversky, A., & Kahneman, D. (1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and Uncertainty, 5, 297–323. Ulrich, R., Schröter, H., Striegel, H., & Simon, P. (2012). Asking sensitive questions: A statistical power analysis of randomized response models. Psychological Methods, 17, 623–641. Vaish, A., Grossmann, T., & Woodward, A. (2008). Not all emotions are created equal: The negativity bias in early social-emotional development. Psychological Bulletin, 134, 383–403. Yaniv, G., & Siniver, E. (2016). The (honest) truth about rational dishonesty. Journal of Economic Psychology, 53, 131–140.

Please cite this article as: Schindler, S., & Pfattheicher, S., The frame of the game: Loss-framing increases dishonest behavior, Journal of Experimental Social Psychology (2016), http://dx.doi.org/10.1016/j.jesp.2016.09.009