فریب افراد بیشتر برای تعداد کمتر: مقایسه اجتماعی اجتماعی صعودی باعث ایجاد انگیزه جبران برای فریب می شود

Abstract Intuitively, people should cheat more when cheating is more lucrative, but we find that the effect of performance-based pay-rates on dishonesty depends on how readily people can compare their pay-rate to that of others. In Experiment 1, participants were paid 5 cents or 25 cents per self-reported point in a trivia task, and half were aware that they could have received the alternative pay-rate. Lower pay-rates increased cheating when the prospect of a higher pay-rate was salient. Experiment 2 illustrates that this effect is driven by the ease with which poorly compensated participants can compare their pay to that of others who earn a higher pay-rate. Our results suggest that low pay-rates are, in and of themselves, unlikely to promote dishonesty. Instead, it is the salience of upward social comparisons that encourages the poorly compensated to cheat.

Introduction Employee dishonesty comes in many forms – from high-powered executives who engage in insider trading to wage workers who over-report hours. While the latter may, superficially, appear less troublesome, widespread low-stakes cheating can add up to substantial losses (Mazar & Ariely, 2006). For example, the phenomenon of “inventory shrinkage” (losses partly attributable to employee dishonesty, such as the misuse of employee discounts) costs retailers billions of dollars annually. And beyond small acts of dishonesty themselves, once people take a step down an unethical road, subsequent steps gradually become easier, and the magnitude of the violations larger (Gino and Bazerman, 2009, Lifton, 1986 and Milgram, 1963). Thus, understanding contextual factors that encourage cheating at low-stakes is important. We examine how economic incentives and fairness concerns interact to influence low-stakes cheating. Some have characterized dishonesty as an economic choice, arguing that it will be more prevalent as its benefits increase, controlling for the probability and costs of getting caught (Becker, 1974). For example, teachers are more likely to inflate students’ grades as the financial incentives for doing so increase (Jacob & Levitt, 2003), and several laboratory studies have observed positive relationships between lying or cheating and the magnitude of incentives (Gneezy, 2005 and Gneezy et al., 2013). Thus, when the benefits of dishonesty are positively correlated with pay-rate, those earning higher wages may be more likely to cheat than those earning less, because they have more to gain. Material gain undoubtedly plays an important role in unethical activity. However, there is mounting evidence that psychological factors also matter, and that dishonesty is not simply the result of economic cost/benefit analysis. For example, fairness concerns can be a better predictor of employee dishonesty than self-interest (Gino & Pierce, 2010b), suggesting that they may be important in determining the relationship between pay-rate and dishonesty. Relatedly, individuals who recall an instance of unfairness or lose a computer game for unfair reasons subsequently behave more selfishly (Zitek, Jordan, Monin, & Leach, 2010), though it is unclear whether this pattern would extend to unethical behavior. One source of workplace unfairness, or at least an indicator of it, is differential pay-rates for similar work. While such differences often exist for reasons that people find justifiable and fair – for example, differences in job tenure – less justifiable wage gaps also exist, such as gender differences attributed to discrimination. A psychological account might therefore predict that low wage earners, upon discovering that others earn more for doing the same work, will feel a sense of unfairness, and may be more likely to behave dishonestly to level the playing field. Thus, given salient interpersonal comparisons, a psychological account of dishonesty might predict increased dishonesty among low wage-earners, even though they have less opportunity to profit from dishonesty than their higher-earning colleagues. In this paper we examine how pay-rate affects dishonesty. We do so in a context in which the material benefits of dishonesty are perfectly positively correlated with pay-rate, pitting economic and psychological predictions against each other and showing when and why each prevails. We hypothesize that the relationship between pay level and dishonesty depends on how readily a person can compare themselves to others who are earning different rates of pay. When this comparison is not salient, consistent with the economic account, we predict greater cheating when it is more lucrative – i.e., at higher pay-rates. However, when people can readily compare their own rate of pay to that of others doing the same work, we predict greater cheating among those earning lower rather than higher pay-rates. The next section discusses the theoretical basis for our predictions. Theoretical framework Organizational behavior scholars have devoted much attention to the role of compensation in employee satisfaction and performance. Much of this work has focused on understanding the antecedents of self-reported pay satisfaction (e.g., whether actual pay level or pay relative to comparable others better predicts pay satisfaction), using hypothetical scenarios or surveys of employees (e.g., Card et al., 2012, Harris et al., 2008, Shore et al., 2006, Sweeney and McFarlin, 2005 and Williams et al., 2006). A number of predictors of pay satisfaction have been identified, with the difference between the amount of pay employees think they should receive and the amount they actually receive being one of the stronger predictors (Williams et al., 2006). In addition, some research has examined the influence of pay (dis)satisfaction on actual workplace performance. Ambrose, Seabright, and Schminke (2002) examined the extent to which perceptions of distributive injustice (largely a function of the extent to which one’s pay is perceived to be fair) helped to explain a wide range of self-reported organizational sabotage behaviors (e.g., aggression, incivility, vandalism). Ambrose et al. (2002, p. 960) found that perceptions of distributive injustice were positively “associated with sabotage behavior aimed at restoring equity.” Pay relative to peers can also help to explain performance by NHL players (Trevor, Reilly, & Gerhart, 2012) and accident rates and on-time deliveries among truck drivers (Kepes, Delery, & Gupta, 2009). Mas (2006) found that in the months after a police union lost final arbitration to management (because the judge selected management’s offer to prevail), arrest rates and average sentence length declined, and crime reports increased, although much more so for property crimes such as burglary and larceny than for more serious crimes such as murder and rape. Perhaps most relevant to the current research is a pair of studies by Greenberg, 1990 and Greenberg, 1993 examining the influence of pay dissatisfaction on employee theft. In a study of manufacturing plant employees who either did or did not suffer a 15% pay cut, Greenberg (1990) found that rates of inventory theft were significantly higher among employees who suffered a pay cut, particularly among those employees who received a sparse, unapologetic explanation for the pay cut. In a laboratory experiment, Greenberg (1993) promised all participants that they would receive $5 for an hour of clerical work, but, once the work was done, either informed them that they would receive the promised $5 or would be under-paid ($3). Participants then paid themselves from a stack of money left on a table. Under-paid participants stole significantly more money than participants who received the promised amount, although under-paid participants still left with less than $5 on average. Both studies suggest that earning less than a salient reference wage (prior or promised earnings) can stimulate theft. This finding is consistent with behavioral decision research suggesting that people are more likely to cheat to recoup losses than they are to achieve gains (Rick & Loewenstein, 2008; cf. Schweitzer, Ordonez, & Douma, 2004). While our discussion thus far has focused on pay compared to promised or expected benchmarks, there are many possible reference points against which one’s pay may be compared (Goodman, 1974 and Ordonez et al., 2000). Some prior work suggests that earning less than comparable others may be even more aversive than earning less than expected. For example, Austin, McGinn, and Susmilch (1980) had participants individually perform a series of tasks for a given pay-rate. Participants were then joined by a confederate, and both independently completed the rest of their tasks. The experimenters manipulated both (non-social) counterfactuals (whether participants earned less, more, or the same per task as they did when working alone), as well as social comparisons (whether participants earned less, more, or the same per task as did the confederate). Ratings of satisfaction with the task were influenced by both social comparisons and counterfactuals, but ratings of anger and fairness were only influenced by social comparisons, with disadvantageous inequality viewed as particularly unfair (cf. Loewenstein, Thompson, & Bazerman, 1989). Similarly, in a survey of pharmaceutical managers, Blau (1994, Table 1) found that pay satisfaction was more closely related to social comparisons (pay level “compared to relevant employees in similar organizations”) than to counterfactuals (pay level “compared to what I earned in previous years” or the difference between what employees thought they should earn and what they actually earn). Thus, prior work suggests that aversive social comparisons may be more likely than aversive (non-social) counterfactuals to influence pay satisfaction. In fact, aversive social comparisons regarding pay may be particularly likely to lead to unethical behavior (e.g., Rick & Loewenstein, 2008). While this precise question has not been investigated, Gino and Pierce, 2009, Gino and Pierce, 2010a and Gino and Pierce, 2010b have found that people are willing to engage in costly dishonesty to reduce wealth-based inequity. In their experiments, participants were randomly paired, and each partner’s initial wealth endowment was orthogonally manipulated. One of the partners was then randomly assigned to solve anagrams; the other was assigned to grade the solver’s work and could behave dishonestly by over- or understating the solver’s score. Wealth-based inequity affected dishonesty such that poor graders dishonestly hurt wealthy solvers (by understating solvers’ scores), even when they incurred a financial cost by doing so (Gino & Pierce, 2009). This work raises the intriguing possibility that aversive (upward) social comparisons based on wealth disparities stimulate retributive dishonesty (cf. Moran & Schweitzer, 2008). Whereas Gino and Pierce (2009) manipulated initial wealth levels, we test the effect of awareness of alternative pay-rates on cheating, holding initial wealth levels constant. We do so because, in an organizational context, cheating is more likely to be a function of differences in pay-rates than of differences in initial wealth levels. Also, we examine how social comparison processes drive unethical behavior that solely benefits the self, rather than, as in Gino and Pierce (2009), unethical behavior that affects both parties. It is important to understand how pay-rates influence unethical behavior that solely benefits oneself, since many acts of dishonesty are intended to solely benefit oneself (DePaulo, Kashy, Kirkendol, Wyer, & Epstein, 1996). Finally, although Gino and Pierce’s (2009) work is consistent with social comparison processes playing an important role in dishonesty, this conclusion cannot be made definitively because dishonesty was not measured in the absence of social comparison information. In all conditions, graders were aware of the wealth level of their solvers – making it, at a minimum, ambiguous whether it is upward or downward social comparisons that affect dishonesty (or both). For example, poor graders tended to overstate the scores of poor solvers and to understate the scores of wealthy solvers, but without knowing the baseline tendency for a poor grader to dishonestly help or hurt his or her solver (i.e., in the absence of information about the solver’s wealth), the direction of the effect is unclear. In the present studies we include conditions in which participants were unaware of pay disparities. Overview of the present research We conducted two incentive-compatible experiments to examine when and why people cheat more at low pay-rates than at high pay-rates. In Experiment 1, we test the basic hypothesis that the effect of performance-based pay-rates on cheating will depend on whether people are aware that others are earning a different pay-rate. When people are aware of an alternative pay-rate, we predict that cheating will be higher with a lower pay-rate, but we predict that this effect will not occur (or reverse) when people have no such awareness. In Experiment 2, we examine two possible mechanisms underlying this pattern. People who are aware that others are earning more for the same work may be driven to cheat by at least two forces: dissatisfaction that they themselves did not earn the higher pay-rate and dissatisfaction that others are earning the higher pay-rate. In Experiment 2, we examine the relative importance of each mechanism by manipulating whether participants were aware that they could have earned a different pay-rate or are aware that others are earning a pay-rate that they themselves could have earned. In reporting the experiments, we follow the standards outlined by Simmons, Nelson, and Simonsohn (2011). No conditions or measures were dropped from any of the experiments reported in this paper. We also provide justification for participants excluded from the analyses (14/132 in Experiment 1 and 11/183 in Experiment 2). Experiment 1 We manipulated the level of performance-based pay, and awareness of an alternative pay-rate, in an experiment in which participants could cheat. Participants received 5 cents or 25 cents per (self-reported) correct response, either with no awareness of the alternative pay-rate (Unaware conditions) or with awareness (Aware conditions) (see Blount & Bazerman, 1996, and Bracha, Gneezy, and Loewenstein (in press), for similar methodologies applied to labor supply). We created awareness of the alternative pay-rate in the Aware conditions by flipping a coin in front of each participant to determine whether they would receive 5 cents or 25 cents per correct response. We hypothesized that participants in the Aware conditions would cheat more at the lower pay-rate than at the higher pay-rate, but that this would not occur (or be reversed) in the Unaware conditions

.