Vicarious Windfalls

 “Windfall (n.) Originally literal, in reference to wood or fruit blown down by the wind, and thus free to all.”1

I.      Introduction

Turing Pharmaceuticals CEO Martin Shkreli made headlines in 2015—and aroused the nation’s collective ire—when he hiked the price of a life-saving HIV treatment drug, Daraprim, by over 5000%, from $13.50 per tablet to $750.00 per tablet.2 Rather than developing its own drugs for the market, Turing targeted out-of-patent medicines—for which there are small markets and few competitors—obtained the licenses for the drugs, and then raised the price of these medicines in pursuit of windfall profits.3 Shkreli pursued this strategy with respect to several pharmaceuticals, including anti-parasitic and toxoplasmosis medications.4

Shkreli’s actions caused a public backlash in which several prominent news organizations published stories critical of his decisions.5 The public outcry against Shkreli’s business model was swift and severe. Several news outlets dubbed Shkreli the “most-hated man in America” and called for congressional investigations in addition to civil and criminal penalties under various state and federal statutes.6 It is uncontroversial that the general public found Turing Phamaceuticals’ actions morally blameworthy—perhaps civilly and criminally so—and that they found Shkreli, in particular, to be the agent who caused the company’s blameworthy actions.7

Now imagine a variation on these facts, which occurred a decade earlier in 2005. Merck & Co., a leading pharmaceutical company, was faced with sluggish sales of two anti-cancer medications.8 Merck then sold the rights to these medications to another company, Ovation, whom Merck may have known would raise the price of the drugs.9 Ovation did, in fact, raise the price of the drugs, such that users’ monthly drug costs skyrocketed; a two-week supply went from $77.50 to $548.01.10 Although print and online journalists reported on the story contemporaneously, the public’s moral outrage did not rise to the level that it did in response to Shkreli’s actions, even though the net public effect remained the same.

To the extent that the general public found Turing to be more blameworthy for its actions than was Merck for similar actions it undertook a decade earlier, what accounts for the different attributions of blameworthiness? The most salient difference involves the directness of Turing’s harm compared to the indirectness of Merck’s harm, which was performed by an intermediary agent. Is attenuated harm, for example, in the scenario in which an employer incurs vicarious liability for the torts of its employee–agent, perceived as less blameworthy than the same harm that lacks attenuation? If so, what implications might that have for litigants in cases involving the respondeat superior doctrine11 or other instantiations of vicarious responsibility for the acts of another?

This Article introduces the concept of the vicarious windfall—based on findings from experimental social psychology—which accounts for the disparity in public perception of harm caused by direct and indirect actors. Construal level theory provides empirical evidence that human beings construe stimuli in our social world differently depending on the distance that we perceive between ourselves and objects in our environment,12 both physically and psychologically.13 To the extent that a social stimulus is construed as abstract (as opposed to concrete), humans tend to associate greater psychological distance between themselves and the stimulus.14

The findings from construal level theory have implications for how we attribute moral culpability. Specifically, humans are more extreme with respect to their appraisals of blameworthiness to proximal actors compared to the behaviors of actors who are distal across psychological space.15 This suggests that, for example, because Shkreli directly raised the price of the HIV medication—and was therefore the concrete actor who caused the harm—his actions were deemed more blameworthy than those of Merck & Co. who, by acting through an intermediary, was a psychologically more distant actor.

Several law and economics theorists, however, subscribe to a starkly different view regarding how fact finders perceive vicarious actors. They agree, to varying extents, that fact finders do not assign the same level of culpability to an actor who causes a direct harm compared to an actor who causes the same harm vicariously (that is, through a legally recognized intermediary).16 They disagree, however, on two important points. First, they believe that jurors are more likely to punish corporate employers vicariously for an employee’s wrongful acts.17 Although they do not all agree on the underlying cause of the disparity, one dominant theory has emerged. Some scholars argue that lay fact finders are likely to take into account the employer’s deeper pockets to pay out claims compared to its employee, which may cause them to become biased against corporate employers.18 So to the extent that corporate employers and their employee–agents are sued for the same activity, law and economics scholars believe that the windfall goes in the opposite direction of that which behavioral research predicts: they believe that the windfall is obtained by the employee.19

Law and economics scholars further argue that, to the extent that employers are burdened by inequitable tort liability under respondeat superior, this will affect their employee selection habits, such that tort victims become undercompensated. Specifically, they argue that employers will limit their tort liability by hiring agents as independent contractors for whose wrongdoing they are not vicariously liable.20 On a grand scale, this may severely limit the ability of claimants to recover damages against tortfeasors, because independent contractors are more likely to file for bankruptcy in the face of a large tort judgment against them compared to a corporate employer.21

Legal and behavioral researchers have diametrically opposed predictions regarding how jurors attribute moral and legal culpability in tort cases that involve the respondeat superior doctrine. Legal scholars bemoan the Robin Hood, anti-corporate jury that unfairly considers the deep pockets of a defendant when assigning tort liability vicariously. Behavioral researchers, however, predict that the inequity is in the opposite direction: defendants who commit harms through intermediary agents are viewed, psychologically, as less culpable than the agents themselves. Determining which scholars are correct will have significant implications for proposed reforms to the vicarious liability doctrine—for the debate over the competency of juries and for the strategies employed by civil attorneys.

This Article is the first to test these predictions in a series of three original experiments. These experiments suggest that, consistent with the theory of moral blameworthiness as a function of the psychological distance between the tortfeasor and the fact finder, jurors are less likely to assign vicarious liability to corporate employers. And although employers (and non-employers alike) apparently believe that jurors will exhibit an anti-corporation bias, the studies reveal no evidence that employers in our sample would meaningfully change their employee selection criteria based on their beliefs about their vicarious liability for the acts of their employees.

This Article proceeds in several parts. Part II outlines the law of vicarious tort liability and the behavioral predictions raised by law and economics scholars. Part III introduces the reader to the response of behavioral researchers, and outlines the literature on construal level theory, psychological distance, and moral psychology. Parts IV, V, and VI report the results from three original laboratory experiments, in which: (1) mock jurors evaluated the tort liability of an employer and its employee in a hypothetical toxic tort lawsuit; and (2) participants’ employee selection habits were examined. Part VII explores the implications of the findings reported in this Article, their limitations, and future directions.

II.     Vicarious Liability and The Corporate Employer

Part II of this Article briefly explains vicarious liability by examining the doctrine’s history and current state. The Article then outlines the law and economics argument that respondeat superior liability for corporate defendants can lead to inefficient legal outcomes.

A.      Respondeat Superior

Vicarious liability is, at its core, a form of strict liability imposed upon a third party for the acts of individuals whom the third party has the right, ability, or duty to control.22 Vicarious liability is often described as a “hybrid” doctrine, insofar as it requires two different forms of liability for a claimant to prevail upon the theory. First, the individual who actually harms the plaintiff must have acted in an intentional or negligent manner, and second, the third party must exhibit sufficient control over the individual such that strict liability for the harm caused by the individual can attach.23

The doctrine originated in 17th century England. Before then, although courts found “masters” liable for harms against others that they expressly directed their “servants” to perpetrate, the law imposed no vicarious duties on a master whose servant perpetrated harms on others that were not at the master’s direction.24 By the end of the 17th century, however, growth in commerce and industry—which required masters to vest more responsibility in their servants—caused courts to reevaluate the state of the law.25 By the 18th century, courts began finding masters liable for the acts of their servants, whether or not the master explicitly directed the servant to commit the harm.26 In Wright v. Wilcox in 1831,27 the first American court to address the issue adopted the broad English rule.28

In its modern state, vicarious liability takes many forms, but it frequently takes the form of respondeat superior in tort litigation. This concept—which translates to “let the superior respond”—holds an employer vicariously liable for certain harmful acts committed by her subordinate employee.29 The employer, however, is not legally responsible for all acts committed by her employee. Instead, respondeat superior liability proceeds on a theory that the employee acts as an agent for the benefit of her employer, and the plaintiff is required to prove an employee–agent relationship at the time of the alleged unlawful act.30 Under the common law, this required agency relationship is proven if the plaintiff can demonstrate, by a preponderance of the evidence, that the employee acted within the general scope of her employment.31 This, in turn, is satisfied in most jurisdictions if three factors are met: (1) the employee is generally doing the business of her employer at the time that the alleged tort occurred; (2) the employee is acting within the geographic and temporal boundaries of her employment; and (3) the employee’s act is motivated, in part, by a desire to serve her employer’s interests.32 If the plaintiff can prove these elements—and prove that the employee’s harmful act was intentional or negligent—the plaintiff can recover against the employer, either in addition to, or instead of, the employee who committed the harm.33 If the plaintiff cannot prove these elements—or if the actor who causes the harm is an independent contractor instead of an employee—the employer is not vicariously liable.34

B.      Law, Economics, and The ‘Anti-Corporate’ Bias

The vicarious liability doctrine—and respondeat superior liability particularly—are hotly debated by legal scholars. The traditional critiques of vicarious liability focus on doctrinal consistency and equity. Scholars note that the vast majority of the tort law—with the notable and narrow exceptions of intentional torts and torts involving dangerous conditions—requires parties to be negligent for a plaintiff to recover damages against them.35 The law of vicarious liability, however, is more complex. Vicarious liability requires an agent to harmfully breach a duty to a plaintiff, but it requires no breach of a duty for a plaintiff to collect against the agent’s employer.36 Rather, if the employee is acting within the scope of her employment, the law deems the employer culpable as a matter of strict liability.37 Many scholars argue that this needlessly complicates the law of negligence. Relatedly, to the extent that the doctrine assumes that the employer is in the best position to control the acts of her employee, it can generate factual scenarios in which well-meaning employers, through no fault of their own, are assigned substantial and disproportionate tort liability.38 These scholars argue that respondeat superior liability therefore is inequitable and leads to unfair outcomes inconsistent with popular notions of moral responsibility.39

A different criticism is leveled against the vicarious liability doctrine by law and economics scholars. These scholars question whether the vicarious liability doctrine actually ensures that deserving tort victims are not undercompensated by insolvent employee tortfeasors.40 Although many law and economics scholars agree with this rationale in the abstract, they point to specific informational asymmetries that lead jurors to be far more likely to find a corporate employer vicariously liable than to find an employee directly liable for the same act. Specifically, these scholars argue that lay fact finders operate under an “anti-corporate bias,” in which they use the tort system to transfer wealth from rich defendants to poor plaintiffs.41 They argue that jurors explicitly take corporate defendants’ financial resources into account when assigning liability and “are committed to running a generous sort of charity” such that “[i]f the new tort system cannot find a careless defendant after an accident, it will often settle for a merely wealthy one.”42 These scholars worry that this anti-business tendency creates an inefficient tort system, insofar as lay fact finders overcompensate tort plaintiffs by looking for reasons—real or imagined—to assign tort liability to a corporate entity with the means to pay the judgment. Thus, they argue that: (1) employees receive a financial windfall compared to their employers; and (2) corporate defendants will hire future employers as independent contractors in an effort to avoid future tort liability.43

These arguments, however, rest on three untested empirical assumptions. First, they assume—without empirical evidence—that jurors do, in fact, exhibit an anti-corporate bias. In actuality, although there is some empirical evidence to suggest that lay fact finders treat corporate defendants differently from individual defendants in some respects,44 empirical researchers have failed to find any evidence of a “deep pockets effect” in real cases or in laboratory simulations. In fact, and as discussed in detail in Part III below, to the extent that inefficiencies exist in the tort system as a result of vicarious liability, the inefficiency is not in the direction that law and economics scholars predict. Rather, it stems from psychological factors that cause lay fact finders to be less likely to assign liability vicariously to corporate defendants.45 The second and third untested assumptions relate to each other. It is unclear that laypeople and businesspeople systematically believe that jurors exhibit an anti-corporation bias, although some anecdotal evidence exists to support the proposition.46 And even if businesspeople do believe that jurors are biased, businesspeople may not consider vicarious liability when selecting employees. This Article seeks to test those claims, and it next provides a psychological framework for understanding how laypeople think about vicarious liability. Part III of this Article explains that framework.

III.     The Social Psychology of Distance

This Article argues that the predictions of torts scholars with respect to the success of vicarious liability claims are exactly backwards. Rather than disproportionately assigning tort liability to the corporate principal (compared to the employee actor who commits the harm), jurors are less likely to assign tort liability to the corporate defendant vicariously, which results in a financial windfall to corporate actors. Until now, however, the support for this theory has been largely anecdotal, evidenced in the lack of moral outrage regarding Merck & Co.’s actions in selling the rights to its medication compared to the moral outrage over the direct actions of Turing Pharmaceuticals, which directly raised the price of its medications.47

To understand the three original experiments that test this hypothesis empirically, this Part proposes a psychological framework for examining the psychological mechanisms that underlie the vicarious windfall. This Part draws on the research of construal level theory to argue that fact finders perceive vicariously liable defendants as more psychologically distant compared to the employees who directly cause the harm. It then draws on research in moral psychology and argues that increased psychological distance causes fact finders to perceive the distant actor as less morally blameworthy—and less legally responsible—for the plaintiff’s harm.

A.     Construal Level Theory

As I have written previously,48 when people think about distance, they most commonly think of physical separation along a spatial plane.49 They do so because the evolutionary need (and ability) to think about objects in one’s environment—both friendly and harmful—as either “near” or “far” occurs very early in human development.50 As cognitive development matures throughout the lifespan, humans begin to apply these concepts of physical distance to abstract, psychological, and philosophical stimuli in their environment.51

Construal level theory—which rests on these psychologically abstract notions of distance—suggests that people make different attributions about stimuli in their environment depending on whether those stimuli are psychologically near to them or far away.52 These different attributions and judgments occur as a result of distance along four separate but related dimensions: physical space, social and emotional space, space across time, and hypothetical distance, which occurs when an individual is asked to imagine the likelihood of certain events occurring.53 As I have written elsewhere:

Construal level theorists contend that thinking about the past or future, a remote location, someone else’s perspective in a social situation, or counterfactual alternatives to a social situation are all different forms of the same underlying process. The psychological distance between an observer and the object of the observer’s attention is rife with social meaning; for example, choosing a seat farther away from another person—or waiting a significant amount of time to return that person’s telephone call—are perceived by others as reflecting social distance. These concepts have implications in myriad psychological domains, including visual perception, categorization, impression formation, heuristic processing, social influence, self-regulation, negotiation, and emotion.54

Construal level theory is so named because of the influence that an object’s psychological distance from an observer has over the observer’s judgments toward, attitudes about, and attributions made about the object.55 Humans tend to construe more psychologically distal objects in high-level, abstract terms, whereas they construe psychologically closer objects in more low-level concrete terms.56 For example, depending on the context, an individual may construe a motor vehicle in low-level terms—such as a Mazda3 sedan—or may construe the vehicle more broadly and abstractly—as a transportation provider.57 A series of studies has examined exactly how these low-level and high-level construals differ from one another:

[L]ow-level construals, which are associated with psychologically close objects, ‘instantiate the present’ and preserve details of the object’s minutiae for immediate use. High-level construals of more distant objects, however, ‘transcend the here and now,’ and conserve for cognitive processing the invariant, essential properties of the object. Thus, although many unimportant features of psychologically close objects are preserved for the purposes of information processing, only the most important features of psychologically distant objects are preserved. Empirical evidence suggests that neither high-level nor low-level construals systematically lead to errors in judgment and decision-making, although some studies suggest that there are cognitive benefits to reasoning using abstract representations.58

These construals occur in our conscious judgments but also outside of conscious awareness. In the legal context, researchers examined whether the psychological distance between a fact finder and a legal actor would subconsciously affect their perceptions of hearsay evidence.59 In the study, mock jurors participated in a hypothetical trial in which the testimony of an out-of-court hearsay declarant was conveyed through an increasing number of hearsay conduits, such that some participants heard from the declarant directly, while others heard his account through ordinary, double, or triple hearsay witnesses.60 The authors found, among other things, that the increase in the psychological distance between the fact finder and the hearsay witness (produced by the different levels of hearsay to which the participants were exposed) was linearly associated with how convincing they found the evidence.61

Does psychological distance exert influence over people’s assessments of moral responsibility? A body of research in moral psychology suggests that laypeople’s moral judgments are affected, to a significant degree, by their subconscious appraisals of the distance between themselves and the actor whose behavior they evaluate.62 The next section in this Article explains how psychological distance affects laypeople’s moral reasoning, the important laboratory studies that support this phenomenon, and explains how they apply to the legal doctrine of vicarious liability.

B.     Distance and Moral Blame

Moral psychology is an interdisciplinary field that blends social psychology and philosophy to determine the evolutionary origins, personality traits, and social cues that predict how laypeople reason about fairness and justice.63 Although early scholars—including Aristotle, Plato, and Socrates—interpreted the field to refer to moral development over the human lifespan,64 modern scholars interpret the field to include an array of topics, including the philosophy of the mind, the creation and development of ethical codes, the notion of free will, moral sensitivity, moral action, moral identity, emotional forecasting (as it relates to appraisals of situational fairness), and, most importantly, moral judgment.65

The most striking feature of morality research involves the counterintuitive nature of moral judgment. The dominant theory strongly suggests that such assessments are not based on deliberative reflection at all; rather they are the result of intuitive, instantaneous, emotional reactions to social stimuli.66 Moreover, although this model of moral decision-making concedes that these initial moral appraisals—at least to some degree—can be altered through deliberation with others, laypeople often do not realize that their moral judgments are the result of subconscious, affective initial appraisals, which are then rationalized through self-serving, post-hoc justifications.67

For example, in his influential article, The Emotional Dog and its Rational Tail: A Social Intuitionist Approach to Moral Judgment, psychologist Jonathan Haidt discusses psychological research demonstrating the primacy of our emotion-based system.68 Discussing the affective system, Haidt states that “[i]t came first in phylogeny, it emerges first in ontogeny, it is triggered more quickly in real-time judgments, and it is more powerful and irrevocable” than our reasoning-based psychological systems.69 Other researchers have demonstrated in clever experiments that our affective system can exert its influence over our moral judgments.70

Psychologists recently have begun examining the role of psychological distance on laypeople’s attributions of morality. The most famous study of psychological directness and indirectness was conducted by Stanley Milgram in the 1960s and 1970s.71 In his classic study of obedience, Milgram had laypeople participate in a “learning” experiment in which they played a memory game with what they believed to be another participant behind a partition.72 (The player was actually just a recorded voice.)73 When the recorded voice made a mistake in the game, the participant was instructed by the experimenter to administer electric shocks to the player of varying degrees of intensity, to which the recorded voice would react negatively.74 Roughly two-thirds of participants administered the highest levels of electric shock in the study at the request of the experimenter.75 In a follow-up study, Milgram repeated the experiment, but this time, participants administered electric shocks through an agent intermediary.76 Milgram found that over 80% of participants administered the highest electric shock when they were not required to do so directly.77

Milgram hypothesized that the presence of an agent intermediary created psychological distance between each participant and the opposing player, which increased the moral acceptability of the participant’s actions.78 Other experimenters extended this research to moral judgments made about others who act through intermediaries and reached similar conclusions. For example, psychologist Neeru Paharia and colleagues conducted four experiments in which individuals acted through indirect agents to harm others by increasing the price of prescription drugs, polluting real property, and paying workers substandard wages.79 The researchers found—across all experimental paradigms—that participants displayed “a moral preference for indirect agency under conditions favoring intuitive judgment” while controlling for alternative explanations for their experimental results.80

Laypeople’s preference for indirect agency—or put another way, their lesser disapproval for harms perpetrated through the acts of others—is the result of the psychological distance that is created when a principal acts through an agent.81 Researchers have found that concrete, detailed information—which is psychologically close—has a significant impact on laypeople’s decision-making and perceptions of probability.82 Others have found that concrete, psychologically close information is more “imageable” and easier to visualize in the mind, which makes it more persuasive.83 Researchers who have explicitly examined the link between distance and blameworthiness have found that people react less negatively to others’ morally ambiguous behaviors when they explicitly take a high-level perspective of that person’s actions.84 Additionally, recent research has found that moral outrage toward the social transgressions of others—as well as moral approval of positive social actions by others—is more extreme when the target is psychologically closer to the decision maker.85

These empirical findings have important implications for vicarious tort liability. To the extent that respondeat superior involves, by definition, civil wrongdoing through the misdeeds of an employee–agent, psychological theory predicts that the presence of an indirect harm causes fact finders to perceive the wrongdoer as more psychologically distant than a wrongdoer who harms a tort victim directly. Moreover, psychological theory also predicts that the distance between the fact finder and an abstract, corporate wrongdoer who acts through an intermediary will cause fact finders to perceive the corporate defendant as less morally culpable—and less legally responsible—for wrongful acts, compared to a concrete actor who causes the same harm. Taken together, these findings suggest that corporate employers may enjoy financial windfalls by escaping liability on vicarious liability theories for recovery in tort. The original experiments reported in this Article examine this hypothesis.

IV.     Study 1: Tort Liability

This Article now reports the results of three original experiments designed to test: (1) whether, and to what degree, corporate defendants are the beneficiaries of “vicarious windfalls,” insofar as fact finders perceive them as less morally responsible for their torts; and (2) employers’ and laypeople’s predictions regarding how jurors perceive corporate defendants, and the extent to which those beliefs affect their employee selection strategies. The original experiments proceed in a progression. They first examine how juries assign tort liability, then they examine the degree to which juries award punitive damages to plaintiffs allegedly harmed by corporate and human defendants. Finally, the experiments examine the extent to which jurors’ business experience moderates both: (1) their judgments of liability; and damages, and (2) the extent to which they consider tort liability in selecting employees.

Study 1 reports the results from a mock trial involving a corporation’s alleged spill of a toxic chemical into a local town’s water supply. The study manipulates the identity of the defendant, such that the defendant is either a company employee or the corporation for which he works. The study examines: (1) whether jurors do assign tort liability differently depending on the identity of the defendant; and (2) whether laypeople believe that jurors will do so.

A priori, the law expects no difference in a jury’s likelihood of finding a corporate employee liable for the harm she directly causes and the likelihood of finding the corporation vicariously liable for the same act. Yet neither torts scholars nor behavioral scholars predict that jurors will actually treat the corporate defendant and the employee defendant the same.86 They differ, however, regarding the direction of the difference that they predict.87 Torts scholars predict that jurors will punish the corporate defendant more than they will punish the employee, because of an anti-corporation bias.88 Moreover, they assert that laypeople and businesspeople will correctly believe that the jury is likely to favor the employee defendant over the corporate defendant.89

Behavioral theorists, however, predict a different outcome. They agree that laypeople are likely to believe that jurors will assign greater tort liability to the corporate defendant (in part because of a general anti-corporation bias). But behavioral theorists assert that laypeople’s prediction is wrong. Rather, jurors will perceive the corporate defendant as more abstract, and thus more psychologically distant from them, compared to the employee defendant. This psychological distance will attenuate jurors’ willingness to assign legal responsibility to the corporate defendant, and they will therefore punish the employee more than they will punish the company for which the employee works.90 The methodology and results from the study appears below.

A.     Participants

Three hundred and two American men and women participated in this study. These participants were recruited through Amazon Mechanical Turk (“mTurk”), an online recruiting tool.91 The sample averaged 38.75 years old (ranging from 20 to 70 years), was 44.40% female, and was 75.30% Caucasian. Sixty-six percent of the sample described themselves as politically liberal, and 57.00% of the sample had completed at least a college degree. The median annual income was between $30,000 and $39,999. A description of the characteristics of the sample appear in Table 1 below.

We also polled participants regarding their experiences with the legal system.92 The poll revealed that 51.20% of the sample reported having spent some time in a courtroom. Their reasons were varied: 21.90% reported spending time in court over minor legal infractions, such as parking tickets; another 21.90% reported having jury duty (of which over half served on the jury), and the remaining participants recounted serving as either a party or a witness in a legal proceeding.


Table 1: Participant Demographics (Study 1)

Table 1 sevier revised


B.     Procedure and Measures

Participants were assigned randomly to one of four different experimental conditions. These four experimental conditions differed along two dimensions. First, participants assumed the role of either a courtroom observer or a mock juror in a civil negligence trial. Second, regardless of whether they were an observer or a mock juror, all participants were exposed to a trial with one of two defendants: either a company employee or the corporation itself. We therefore assigned participants randomly to one of the actor conditions (observer or juror) and one of the defendant conditions (employee or corporation).93

After providing their informed consent to participate in this study, subjects were then asked to imagine that they were attending a local civil trial. Half of the participants were told that they had been called for jury duty in a civil case. They were told to imagine themselves seated in the jury box, surrounded by 11 other local citizens, and that the plaintiff and defendant would present evidence to them. These mock jurors were told that they would be required to determine whether the injury that the plaintiff alleged was caused by the defendant, and whether the defendant was liable for the plaintiff’s injuries.

Other participants were told, instead, to imagine themselves seated in the gallery of their local courthouse. They were told to imagine that they were spectators at the trial, and that they would watch the plaintiff and the defendant present evidence to the judge and to the jury. We further explained that participants would not decide the outcome of the case. Rather, we wanted participants’ impressions of how the jury would decide the case after hearing all of the evidence. Participants were specifically told that their job was to predict how the jury would decide the case, and not to base their judgments on how they would decide the case if they were a juror. All participants acknowledged that they could complete the task assigned to them.

All participants read about a fictional civil negligence action involving a toxic tort. At the outset, participants were told that the case involved a chemical spill at a local power plant that may have sickened local adults and children. Participants were told that several of these adults and children then sued over their alleged injuries.94

All participants read the plaintiffs’ case first. The plaintiffs’ case consisted of witnesses who testified that they habitually had drunk tap water that was funneled from a local stream that runs through the center of town. Plaintiffs also called an expert witness who tested water from the stream and found that it contained high levels of polychlorinated biphenyls (“PCBs”), exceeding the allowable de facto limit (0.5 parts per billion) established by environmental regulators. Other plaintiffs’ witnesses testified that the corporation that owned the power plant used PCBs in its daily operations and that they were stored near the river, which was adjacent to the property on which the power plant operated.

Participants next heard from the defense. Half of the participants—regardless of whether they were observers or mock jurors—believed that the plaintiffs had sued the power plant manager, John Brown, whom they believed polluted the stream. The other participants, however, believed that the plaintiffs had sued the Acme Power Plant under a respondeat superior theory of liability. If participants were in the experimental condition in which the Acme Power Plant was the defendant, they were told not to focus on any alleged negligence on the part of the corporation itself, but to assume that the corporation can be held liable for negligent acts of its employees.

The defense called two witnesses. First, the defense called the power plant manager, John Brown. Mr. Brown discussed an incident in which he knocked over a canister on the outside premises of the power plant. The canister, he testified, rolled into the stream until he was able to retrieve it. He repeatedly stated under cross-examination that although the canister sometimes contained PCBs, the canister was empty at the time it entered the stream. Second, the defense called an expert witness who was familiar with PCB pollution. The expert testified that the amount of PCBs that would typically be in the canister—even if it were full—would not have caused the PCB levels that were found in the stream.

After the defense presented its case, all participants read a jury instruction on the standard of proof for a common law toxic tort action. Further, participants who were in the experimental conditions in which the Acme Power Plant was the defendant received a jury instruction on vicarious liability. Participants were then reminded of their responsibilities as either an observer or a juror: either to (1) determine whether the defendant is liable for the plaintiffs’ injuries; or (2) whether the jury would find that the defendant is liable for the plaintiffs’ injuries.

Participants then answered several questions about the trial and about themselves. Participants were first asked whether the defendant is liable for the harm to the plaintiffs (or whether the defendant is likely to be liable for the harm in the eyes of the jury). Participants were also asked to rate the degree to which the defendant is likely liable to the plaintiffs on a scale from 1 to 7, where 1 is “definitely not liable” and 7 is “definitely liable.” All participants were told that, in the event that the defendant is liable to the plaintiff, the parties had agreed to the amount of compensatory damages that would be owed.

Participants were then asked the extent to which they would award punitive damages to the plaintiff based on the defendant’s behavior. Participants were asked this question in two ways: first, participants were asked to determine the amount of punitive damages along a seven-point scale, where 1 is “no punitive damages” and 7 is “significant punitive damages”; second, participants were asked to provide a dollar amount of punitive damages that they would award in the case.

Next, participants were asked several questions about the defendant, who was either the Acme Power Plant or its employee, John Brown, depending on the experimental condition to which each participant was assigned. Among these questions were two items that are important to the analysis of the current experiment. First, participants were asked, “How close do you feel to the defendant in this case?” This question, which is borrowed from the psychological literature on construal level theory, is designed to gauge each participant’s sense of psychological distance from either John Brown or the Acme Power Plant. Participants answered this question on a seven-point scale, ranging from 1 (“I don’t feel close to the defendant at all”) to 7 (“I feel very close to the defendant”).95 Second, participants were asked, “How moral do you believe the defendant to be?” This question, which was designed to gauge participants’ perceptions of the defendant’s moral blameworthiness, was measured from 1 (“The defendant is extremely immoral”) to 7 (“The defendant is extremely moral”).

Participants then answered several questions designed to gauge their familiarity with the legal system. The questions asked participants whether they had ever been inside a courtroom and in what capacity. Afterward, participants answered several demographic questions, which asked them to report their age, gender, political orientation, race, income, and level of education completed. Participants were then debriefed on the aims of the experiment, and the study was concluded.

C.     Results

This Section proceeds in several subparts. First, it addresses several preliminary issues specifically involving participants’ judgments of liability and damages. Second, it analyzes whether laypeople’s perceptions of the jury’s likelihood to find liability matches how likely the jury actually is to find liability, depending on whether the defendant is the direct injurer or the company (in respondeat superior). Third, it examines whether the jury’s decision to find, or not to find, the defendant liable is associated with the psychological distance between themselves and the defendant.

1.     Preliminary Matters

Before examining the primary hypothesis that underlies Study 1, we examine two preliminary matters. Specifically, we must examine the relationship between different instantiations of our outcome measures: perceptions of the defendant’s liability and the amount of damages, if any, that the defendant owes the plaintiff. This entails two distinct analyses. First, the different ways in which damages were measured were evaluated for consistency. The distribution of participants’ responses on the 1 to 7 scale followed a normal distribution with no skew or kurtosis.96 The distribution of participants’ responses appears below (in Figure 1).

Participants also were required to award damages in the form of a dollar amount. The distribution of awards appears below in the graph titled Figure 2. As is typical in controlled laboratory experiments in which participants must award punitive damages, the distribution of responses was right-skewed with dramatic variance.97 The variance was caused, expectedly, by outlier responses that awarded the plaintiffs punitive damages far in excess of the average award given by the sample participants. Consistent with standard statistical protocol, we performed a logarithmic transformation of the free response data to compress the outlier awards.98 The result of the logarithmic transformation procedure appears below in Figure 3. The logarithmically transformed damages awards were normally distributed with no statistical skewness and minor statistical kurtosis, consistent with the distribution of responses to the “Scaled Damages” measure.99 


Figure 1. Scaled Damages

Fig 1 sevier revised


Figure 2. Free Response Damages

Fig 2 sevier revised


Figure 3. Log Transformed Damages

Fig 3 sevier revised


After performing a statistical transformation on participants’ spontaneous punitive damages awards, we next examined the correlations among participants’ determinations of liability and their damages awards. A bivariate correlation measures the strength of the relationship between variables (which ranges from –1.00 to +1.00), with correlations closer to +1.00 indicating a perfect positive relationship between two variables.100 Table 2 illustrates the bivariate correlations among: (1) our two liability measures (the 1 to 7 Likert Scale and the dichotomous verdict that participants rendered); and (2) our two damages measures (the 1 to 7 Likert Scale and the Log Transformed Damages measure created above). The table reveals high, positive, and statistically meaningful correlations among all variables, ranging from 0.61 to 0.81. To avoid redundancy, because all four measures were highly correlated with one another, Study 1 focuses only on participants’ perceptions of the defendant’s liability.101


Table 2. Correlation Matrix of Depended Measures

Table 2 sevier revised



2.     Main Analysis

We hypothesized in Study 1 that a difference would exist between people’s perceptions of the defendant’s liability depending on whether the defendant was the employee who caused the harm or the corporation that employed her. Moreover, and counterintuitively, we hypothesized that people’s intuitions about the nature of this disconnect—assuming that one exists—would be the opposite of how jurors actually behave. To test this hypothesis, we employed a statistical technique, termed an “analysis of variance,” to examine whether the data that we collected supports our hypothesis. An analysis of variance, in simplest form, examines whether the responses from participants in each experimental condition differ from each other in statistically meaningful ways.102

To test the hypothesis, we examined whether the role that participants undertook in this study (either as an observer or a mock juror) and the identity of the defendant (either employee John Brown or employer Acme Power Plant) affected participants’ judgments of the defendant’s tort liability.103 As hypothesized, the analysis revealed that these variables jointly, and significantly, affected participants’ judgments of the defendant’s liability.104 To determine exactly how the participants’ role in the study and the identity of the defendant affected the participants’ perceptions of the defendant’s liability, we examined the effect of the defendant’s identity separately for observers and mock jurors.

There was an effect of the defendant’s identity on observers’ predictions of whether the defendant would be found liable. As hypothesized, observers expected mock jurors to find the defendant liable more often when the defendant was the Acme corporation than when the defendant was the employee John Brown.105 But interestingly and counterintuitively, there was a different effect of the defendant’s identity on mock jurors’ willingness to find the defendant liable. Consistent with construal level theory, mock jurors were significantly more likely to find the employee liable for his negligence than they were to find the corporation vicariously liable for the employee’s negligence.106 The results from the analysis of variance are reflected in Figure 4.

Figure 4. Likelihood of Tort Liability

Fig 4 sevier revised


We next examined this phenomenon in the context of the actual verdicts that mock jurors rendered. Although this is conceptually the same analysis, we analyzed participants’ verdicts using a logistic regression, which is the statistical technique that must be used when the outcome variable (here, verdicts) is dichotomous.107

As before, and as expected, the logistic regression analysis revealed an interactive effect of both variables on participants’ verdicts. As before, observers believed that jurors would find the defendant liable more often when the defendant was the corporation than when the defendant was the corporation’s employee. But in stark contrast, mock jurors actually found the defendant liable more often when the defendant was the employee himself than when the defendant was the corporation. The logistic regression table, which reports this statistically significant interactive effect, appears below.108


Table 3. Logistic Regression of Participant Role and Defendant Type on Verdicts

Table 3 sevier revised


3.     Serial Mediation Analysis

The analysis of Study 1 has borne out the experimental hypothesis. Laypeople intuit that jurors have different implicit thresholds for finding that a defendant is negligent depending on whether the alleged tortfeasor is a corporation or an individual. Also as expected, participants’ perceptions of who is more likely to be punished is backwards: although observers believe that corporations are more likely to be punished when they are the defendant, mock jurors were more likely to assign tort liability to the individual defendant.

This finding is consistent with construal level theory and the phenomenon of psychological distance. Because the corporate defendant is a more nebulous and abstract entity than is an individual (human) defendant, jurors may, outside of conscious awareness, be less likely to attribute responsibility and blameworthiness to the corporate defendant. This Subpart tests this theory through a statistical technique called a serial mediation analysis of the responses of participants assigned to the ‘mock juror’ experimental condition.

A serial mediation analysis consists of a set of regression analyses that are designed to determine the psychological processes that underlie the effect of a predictor variable on an outcome.109 The psychological process (or processes) that are hypothesized to underlie the effect are termed “mediators” of the effect.110 A mediation analysis is designed to show that the effect of the predictor on the outcome can be explained—either fully or in part—by the psychological mediators.111 A serialized mediation simply involves more than one mediator.112 A serialized mediation analysis tells us that a predictor variable is associated with one psychological mediator, which is associated with another psychological mediator which, in turn, is associated with the outcome.113 An illustration of serialized mediation appears in Figure 5.


Figure 5. Illustration of Serialized Mediation

Fig 5 sevier revised


In the illustration above, the dotted line represents the effect of the predictor on the outcome. The solid black lines illustrate the pathway between the predictor variable and the outcome via the two psychological mediators. The two thick lines in this illustration indicate where there is no relationship between variables: (1) between the predictor variable and the second psychological mediator; and (2) between the first psychological mediator and the outcome. These variables must not be statistically associated with each other for the serialized mediation analysis to be correct.

The serialized mediation analysis applies as follows: (1) the predictor variable is the identity of the alleged tortfeasor (either the individual employee or the corporation); (2) the outcome is mock jurors’ likelihood of finding the defendant negligent; (3) the first mediator is the degree to which the mock jurors felt psychologically close or distant to the defendant; and (4) the second mediator is the extent to which mock jurors believed that the defendant was morally blameworthy. The analysis then proceeds as a series of regression analyses to determine if the effect of the tortfeasor’s identity on mock jurors’ willingness to find the tortfeasor liable is explained as follows: (1) the defendant’s identity affects participants’ perceptions of how psychologically ‘close’ the defendant is to them; (2) this level of psychological distance affects how morally blameworthy the mock jurors believe the defendant to be; and (3) this level of moral blameworthiness predicts mock jurors’ willingness to find the defendant liable for the tort. This hypothesis is tested below.

The identity of the alleged tortfeasor affected mock jurors’ willingness to find the defendant liable, such that the corporate defendant was less likely to be found liable than was the corporation’s employee.114 As predicted, the defendant’s identity was associated with psychological distance, such that mock jurors found the corporation to be significantly more distant from them than was the individual employee.115 Also as predicted, mock jurors’ perceptions of the psychological distance between themselves and the defendant predicted how morally blameworthy they believed the defendant to be, such that defendants who were perceived to be more psychologically distant from mock jurors were deemed more morally culpable for the plaintiff’s illness.116 Finally, perceptions of moral blameworthiness significantly predicted the degree to which mock jurors were willing to assign tort liability to the defendant, such that lower perceptions of the defendant’s morality were associated with a higher likelihood of assigning tort liability to the defendant.117 Moreover, specialized tests for mediation revealed that this pathway significantly accounts for the effect of the tortfeasor’s identity on mock jurors’ willingness to punish the tortfeasor.118 An illustration of this pathway, which includes the beta coefficients from the regression analyses that were performed, appears below.119


Figure 6. Interaction Between Identity, Psychological Distance, Moral Blame, and Tort Liability

Fig 6 sevier revised


D.     Discussion

Study 1 revealed several insights regarding how laypeople understand vicarious liability. First, laypeople appear to understand that jurors will not always act equitably when assigning liability to different tortfeasors. Consistent with the anti-corporation bias, laypeople believe that jurors are more likely to assign liability to a corporate defendant than to an individual employee wrongdoer. But interestingly, Study 1 reveals that laypeople’s intuitions—as well as the intuitions of academic torts scholars—run in the wrong direction. Jurors actually assign greater tort liability to employee defendants than to the corporate employer under the same factual scenario.

Study 1 also examined the psychological processes that account for this counterintuitive result. The serialized mediation analysis revealed that mock jurors perceive corporate defendants as more psychologically distant from themselves (compared to an employee defendant), which causes them to perceive the corporate actors as less morally culpable. The implications from these findings are explored fully at the conclusion of this Article.

This study focused on mock jurors’ willingness to assign liability to a corporate defendant instead of to her individual employee. Study 1 did not examine how the phenomenon of psychological distance affects jurors’ punitive damages awards. We designed a second study to address this question.

V.     Study 2: Punitive Damages

Study 2 serves two purposes. First, it seeks to replicate the counterintuitive finding from Study 1, in which jurors were more likely to assign tort liability to an employee defendant than to his corporate employer in respondeat superior. Second, it extends Study 1 by examining how the identity of the alleged tortfeasor affects jurors’ willingness to award punitive damages. Two specific hypotheses are tested in Study 2. We hypothesize that, similar to Study 1, jurors will assign higher punitive damages to the conduct of a concrete, psychologically close defendant—such as a company employee—than they will to an abstract, psychologically distant defendant—such as the corporation that employs him. Moreover, previous research suggests that when the stakes of moral decision-making are raised—and the vividness of the harm to a victim is made salient—people’s moral judgments will be amplified.120 Therefore, we predict that the differences in the damages that participants award will be magnified when the harm experienced by the plaintiff is high compared to when the harm experienced by the plaintiff is low.

In sum, we predict that the identity of the defendant and the degree of harm will interact with each other to affect the damages awarded by the participants. Further, we predict that this interactive effect of the defendant’s identity and the severity of the harm on participants’ damages awards will be mediated by both the psychological distance between the participants and the defendant and by their perceptions of the defendant’s moral blameworthiness. Each of these hypotheses is tested below.

A.     Participants

Two hundred and one Americans participated in this study and were recruited through mTurk. The sample averaged 35.77 years old (ranging from 20 to 70 years), was 46.30% female, and was 73.90% Caucasian. Sixty-six percent of the sample described themselves as politically liberal and 54.60% of the sample had completed at least a college degree. The median annual income was between $30,000 and $39,999. A complete description of the sample appears in Table 4 below.

We also polled participants regarding their experiences with the legal system. The poll revealed that 48.30% (nearly half) of the sample reported having spent some time in a courtroom. Their reasons were varied: 21.40% reported spending time in court over minor legal infractions, such as parking tickets, another 20.40% reported having served on jury duty (of which approximately half served on the jury), and the remaining participants recounted serving as either a party or a witness in a legal proceeding.

B.     Procedure and Measures

The procedures and measures used in this study are similar to those used in Study 1, although several features of the experiment were tweaked. First, all participants assumed the role of a mock juror and, as in Study 1, everyone was exposed to a trial with one of two defendants: either a company employee or the corporation who hired the employee. Second, participants were assigned to an experimental condition in which the harm alleged by the plaintiffs was either minor or serious.121

After providing their informed consent to participate in this study, subjects were again asked to imagine that they were attending a local civil trial. This time, all of the participants were told that they had been called for jury duty and would be deciding a civil negligence case involving a toxic tort. Participants were told specifically that they would be deciding not only the defendant’s liability but also the amount of damages that the plaintiffs would receive (if the defendant was found liable).

Participants then read the same fictional civil negligence trial as in Study 1. In Study 2, however, the fact pattern was altered in one important respect: participants received more information regarding the harm alleged by the plaintiffs. In the low-harm condition, participants heard evidence suggesting that the plaintiffs suffered severe stomachaches and discomfort after drinking tap water that had been supplied by the stream. In the high-harm condition, participants heard that the plaintiffs had been diagnosed with certain types of cancer associated with PCBs. All other evidence remained the same as it was in Study 1.


Table 4. Participant Demographics (Study 2)

Table 4 sevier revised


Participants also read jury instructions involving the standard of care in a toxic tort action as well as instructions involving punitive damages. Participants who read that the Acme Power Plant was the defendant also read a jury instruction on vicarious liability to ensure that they understood that the employee’s liability—assuming they believed him to be liable—was to be attributed to the corporate defendant.

Participants then answered several dependent measures, as they did in Study 1. These measures asked whether they found the defendant liable and, if so, the punitive damages that the plaintiffs should be awarded. They also answered measures designed to gauge their perceptions of the psychological distance between themselves and the defendant, as well as the defendant’s moral blameworthiness. After answering questions about their background—including demographics, experience with the courts, and any business experience they had—participants were debriefed and the experiment concluded.

C.     Results

This Subpart proceeds in two steps. First, it analyzes whether participants’ punitive damages awards were affected by either the identity of the defendant, the nature of the harm alleged by the plaintiffs, or a combination of those variables. Second, to the extent that the experimental hypotheses are confirmed, it presents a mediation analysis to examine whether the psychological distance between the defendant and the research participants explains the experimental results.

1.     Main Analysis

As a preliminary matter, we again ran tests examining the correlation among our dependent measures. We found that all dependent measures—two liability measures and two damages measures—were highly correlated with each other.122 Moreover, as we did in Study 1, participants’ spontaneous punitive damages awards were log-transformed so that the resulting distribution was normally distributed.123

In Study 2, we hypothesized that the identity of the defendant and the degree of harm alleged would affect participants’ punitive damages awards. Additionally, we made a third prediction: we hypothesized that these variables would jointly affect participants’ punitive damages awards. We tested these hypotheses through an analysis of variance statistical technique.

The analysis revealed several statistically significant effects. Unsurprisingly, and as predicted, the degree of the harm alleged affected participants’ damages awards, such that more severe harm was associated with higher awards.124More importantly, and as predicted, we found an interactive effect of the harm suffered by the defendant and the defendant’s identity on participants’ damages awards.125 Specifically, the degree to which the identity of the defendant affected participants’ damages awards depended on whether the harm was trivial or substantial. A graph of the mean damages awards in each experimental condition is produced below.


Figure 7. Mean Damages Awards

Fig 7 sevier revised


We examined this interactive effect by evaluating the effect of the defendant’s identity on punitive damages awards when the harm alleged by the plaintiffs was low (stomachaches and other ailments) and when it was high (cancers associated with PCB exposure). At low levels of harm, we found a non-significant (and weak) trend to award higher punitive damage awards to the plaintiff when the defendant was an employee.126 At high levels of harm, however, we found a strong, highly significant effect of the defendant’s identity, such that mock jurors awarded significantly higher punitive damages when the employee was the defendant, compared to when the corporation was the defendant.127 In sum, high levels of alleged harm exacerbated mock jurors’ tendency to punish the employee defendant compared to his employer vicariously. To explore the psychological mechanisms that underlie the interactive effect that we observed in Study 2, we performed a second mediation analysis. Those results are reported below.

2.     Mediated Moderation Analysis

The results support the experimental hypotheses. Construal level theory predicts that (contrary to popular wisdom) jurors punish concrete, psychologically close entities for moral wrongdoing to a greater degree than they punish abstract, psychologically distant entities. Moreover, psychological theory predicts that this tendency is exacerbated when participants perceive the stakes of the decision to be greater. This Subpart examines whether the results reported in Study 2 stem from the psychological distance between the defendant and the fact finder. We test this question through a ‘mediated moderation analysis’ on participants’ damages awards.

A mediated moderation analysis, unlike the serialized mediation analysis performed in Study 1, examines whether an interactive effect on a behavioral outcome—called a moderation—is mediated by a separate psychological variable.128 A mediated moderation analysis consists of four variables: (1) the predictor variable; (2) the outcome variable; (3) the variable that interacts with the predictor (called the “moderator”); and (4) the psychological process proposed by the experimental hypothesis, which is called the mediator.129 The predictor and the moderator work together to explain the outcome, but they also work together to explain the psychological mediator. The mediator, in turn, also explains the experimental outcome. Moreover, when the psychological mediator is added to the model, the direct effect of the predictor and the moderator on the outcome disappears (because the effect is truly driven by the psychological mediator).130 A conceptual diagram of a mediated moderation analysis appears below. 


Figure 8. Mediated Moderation Analysis

Fig 8 sevier revised


The mediated moderation analysis takes the following form. The predictor variable is the identity of the alleged tortfeasor (either the individual employee or the more psychologically-distance corporation) and the outcome is the amount of punitive damages awarded. The moderator variable is the degree of the harm alleged by the plaintiffs (either a mild stomach ailment or a cancer diagnosis). The psychological mediator variable is the perceived psychological distance that participants felt toward the defendant in the case. As in Study 1, the analysis proceeds in a series of regression analyses. First, the analysis will show that the identity of the defendant affects participants’ punitive damages awards, but it does so differently, depending on the degree of the harm. The analysis will then show that the defendant’s identity also affects the psychological distance that participants will feel toward the defendant, but at different levels depending on the severity of the harm. Finally, the analysis will show that the degree of psychological distance that participants feel toward the defendant will, in turn, affect the punitive damages that they award the plaintiff. We test this hypothesis below.

The tortfeasor’s identity affected the amount of punitive damages that mock jurors awarded to the plaintiff, but the magnitude of the effect depended on the level of harm alleged by the plaintiff.131 Specifically, participants gave significantly greater damages awards to plaintiffs when the defendant was an individual employee, instead of a corporation, when the harm alleged was severe.132 There was only a non-significant trend in this same direction when the harm alleged was minor.133

As predicted, the analysis found a significant interactive effect of the defendant’s identity and the harm alleged on participants’ perceptions of the psychological distance between themselves and the defendant.134 Specifically, participants felt much greater psychological distance toward the corporation—compared to the employee—when the alleged harm was severe.135 Again, there was only a non-significant trend in this direction when the alleged harm was minor.136 Further, participants’ perceptions of the psychological distance between themselves and the defendant was significantly associated with the amount of damages that they awarded.137 Specifically, higher punitive damages awards were associated with lower perceptions of psychological distance, which suggests that mock jurors punished the concrete actor more severely than they punished the more abstract, distant actor.

Most importantly, when the psychological distance mediator was added to the model (which already included the identity of the defendant, the magnitude of the harm, and mock jurors’ punitive damages awards), the direct effect of defendant identity and harm on participants’ damages awards became non-significant.138 Thus, as predicted, the entire effect found in Study 2 is explained by the phenomenon of psychological distance. The defendant’s identity interacts with the magnitude of the harm to affect participants’ perceptions of the psychological distance between themselves and the defendant. This perception of distance, in turn, affects their punitive damages awards. An illustration of this pathway, which includes the beta coefficients from the regression analyses that we performed, appears below.139 


Figure 9. Interaction Between Identity, Magnitude of Harm, Psychological Distance, and Punitive Damages

Fig 9 sevier revised


D.     Discussion       

This study expanded on the results from Study 1 in several important ways. First, it successfully replicated the main finding from Study 1: contrary to conventional wisdom, mock jurors were less likely to punish a corporate defendant vicariously than they were to punish an employee defendant directly. Study 2 also examined how this tendency manifested itself with respect to punitive damages awards. When the degree of harm alleged by the plaintiffs was minimal, jurors tended to hold the employee more responsible (compared to the corporation defendant), but the difference in punitive damages awards was not statistically reliable. But when the stakes were higher—and the damages alleged by the plaintiffs consisted of serious cancer diagnoses—a much stronger effect emerged: mock jurors’ punitive damages awards were considerably larger when the defendant was the employee who directly caused the harm, compared to a corporation that would be vicariously responsible for the same harm.

Study 2 also examined the psychological processes that underlie the tendency of jurors to attribute less harm vicariously to the corporation. The mediated moderation analysis revealed that mock jurors attribute greater psychological distance between themselves and the corporate defendant, particularly when the harm alleged by the plaintiffs is severe. This greater psychological distance, in turn, makes them less likely to assign moral and legal blame. This results in lower punitive damages awarded to plaintiffs when the defendant is an abstract, psychologically distant corporation compared to a concrete, psychologically close employee. Because blockbuster tort lawsuits often allege a significant harm, jurors’ differential punishment of corporate and employee defendants has several important implications.

One important implication involves employee selection in the private sector and the attitudes of businesspeople who are called for jury duty. The final study reported here presents early-stage pilot data on this question. It is presented below.

VI.     Study 3: Employer Attributes

Torts scholars worry that the vicarious liability doctrine has detrimental effects on employee selection. They argue that a fear of vicarious liability for the acts of its employees may cause employers to seek out workers who function as independent contractors.140 The law of agency allows employers to do so—provided that the hired individual truly acts as an independent contractor—which allows employers to avoid vicarious liability for the contractor’s tortious acts.141

The final study reported in this Article reports pilot data bearing on those predictions and examines how businesspeople in our sample view vicarious liability more generally. These pilot data compare the decision-making styles of business owners who participated in the previous study with non-business owners. The data reported explore two phenomena: (1) whether business owners differ from non-business owners with respect to the punitive damages that they would have awarded in the trial reported in Study 2; and (2) whether vicarious liability affects these business owners’ hiring decisions compared to laypeople. The methodology and results of this pilot study are described below.

A.     Participants, Procedures, and Measures

The data described here were collected from the mock jurors who participated in Study 2. In addition to demographic variables, we asked participants in these studies several questions to gauge their business experience. We asked participants whether they had ever owned a business and whether they had ever made hiring decisions (either as a business owner or in another capacity). We asked participants to characterize their business as small (under ten employees), medium (under 50 employees), or large (50 employees or more), and we asked them to describe the profitability of the business (as either not very profitable, somewhat profitable, or very profitable). We also asked participants questions designed to gauge their understanding of legal terms in business and tort law, including vicarious liability and the distinction between employees and independent contractors.

As expected, a slight majority of the participants in the sample were not business owners, although a substantial minority—over 40%—reported having owned a business. Most business owners in the sample reported owning a small business, although approximately 15% of the sample described their business as either medium or large. Self-reported profitability of participants’ businesses was distributed more evenly, with over half of participants describing the business as at least somewhat profitable and over 15% describing it as very profitable, and a little over one-third of respondents described their business as not very profitable. Although nearly 60% of the sample reported never owning a business, nearly half of all participants had, however, hired or fired employees at their place of employment. As expected, most participants were unfamiliar with the terms “vicarious liability” or “respondeat superior,” but most participants understood the meaning of an independent contractor. A table of descriptive statistics with respect to participants’ business experience, appears in Table 5 below.

All participants completed Study 2.142 Additionally, all participants—regardless of business ownership—were asked several questions to measure their employee selection preferences with respect to the employee presented in Study 2.143


Table 5. Participants Business Experience (Study 3)

Table 5 sevier revised


First, we asked participants, on a seven-point scale ranging from 1 (“not at all”) to 7 (“all the time”), to what degree do you think about your potential liability for an employee’s actions when you decide to hire an employee? Second, we asked participants to think about the employee featured in Study 2, and to tell us whether they would have hired him as an employee or an independent contractor.

Our third question was complex. We asked participants to imagine that they were making an important hiring decision. We presented participants with seven different factors of decision. Participants then ranked these factors from 1 (most important) to 7 (least important) with respect to employee hiring. The considerations were as follows: liability for the employee’s acts, the employee’s intelligence, the employee’s experience, the employee’s level of education completed, the employee’s personality traits, the likelihood that the employee would leave the business to go elsewhere, and the employee’s performance during the hiring interview. After ranking these choices against each other, participants were debriefed and the studies were concluded.

B.     Results

This Subpart proceeds in two parts. First, it examines whether the results of Study 2—which showed that corporate defendants enjoy a vicarious windfall when they are sued vicariously in cases in which damages are severe—were meaningfully different depending on the participants’ business experience. The second section extends this analysis by examining the employee selection strategies of those with business experience in our sample and comparing those strategies with those of laypeople.

1.     Punitive Damages

We first examined whether business experience modified our subjects’ tendency to award lower punitive damages against corporate defendants (compared to their employees). Specifically, we explored whether the interactive effect of the defendant’s identity and the harm alleged on participants’ punitive damages awards remains statistically significant if we hold constant participants’ business experience.

To examine this question, we constructed three different regression models using the data collected in Study 2. The first model examined the effects of demographic variables on the interactive effect that we found in Study 2. This model included the identity of the defendant (corporation vs. employee), the degree of the harm (low vs. high), and the interactive effect of these variables (represented in the model as “defendant x harm”). It also included participants’ gender, age, political orientation (coded as either “more liberal” or “more conservative”), income, and education (coded as “high school,” “some college,” “college,” “Master’s,” or “Ph.D. or professional degree”). Model 2 included all of the variables from Model 1, but it also included a variable indicating whether participants had spent time in a courtroom. Finally, and most importantly, Model 3 included two variables designed to measure two relevant business traits: whether participants had ever owned a business or, in the alternative, whether they have been involved in hiring decisions.

All three models appear in Table 6 below. The first model, which contained participants’ demographics, revealed: (1) a statistically significant interactive effect of the defendant’s identity and the degree of the alleged harm; (2) an effect of each participant’s gender; and (3) an effect of each participant’s political orientation on their willingness to assign tort liability to the defendant. The second model revealed the same results. Finally, and most importantly, the third model—which included participants’ business experience—again revealed the same pattern: a statistically significant interactive effect of the defendant’s identity and the harm, an effect of gender, and an effect of political orientation. There was no meaningful effect of either business ownership or hiring experience on participants’ willingness to award punitive damages to the defendant.


Table 6. Factors Affecting Jurors Punitive Damages Awards (Linear Regression)

Table 6 sevier revised


There are several important takeaways from Table 6. As a preliminary matter, it appears that participants’ gender and their political orientation affected their punitive damages awards. Table 6 reveals that changing a participant’s gender from male to female results in roughly a point reduction of two-thirds-of-a-point in punitive damages awards.144 Similarly, changing a participant’s political orientation from conservative to liberal was associated with nearly a three-quarters-of-a-point increase in their damages awards.145 These preliminary results were obtained across all three regression models and reflect previous research regarding people’s attitudes toward tort reform as a function of their gender.146

There are two critical findings reported in Table 6, both of which support the experimental hypothesis. First, the interactive effect of the defendant’s identity and the alleged harm on mock jurors’ punitive damages awards—which we reported in Study 2—was statistically significant across all three models. Even when we controlled for a host of factors, including the demographics of our sample, their experience with the justice system, or their experience in the business world, participants still viewed corporate defendants as less culpable than their employee counterparts. And whether participants’ business experience was measured through actual ownership or through relevant hiring experience, business experience still had no effect on this outcome.

2.     Employee Selection

We next examined the degree to which vicarious liability informs hiring decisions by businesspeople in our sample, and whether their decisions differ from those of lay people. We first examined the extent to which business owners in our sample considered their potential for vicarious liability for the torts of their employees. Recall that participants answered this question on a seven-point scale ranging from 1 (“not at all”) to 7 (“all the time”). We conducted a one-sample t-test to determine whether business owners’ mean scores on this item differed significantly from 7, which would indicate—as torts theorists suggest—that businesspeople consider vicarious liability frequently in their hiring decisions.147 The test revealed that business owners’ mean scores on this item were significantly below a score of 7, which indicates that they do not consider vicarious liability often when they select employees.148 Moreover, a follow-up test revealed that participants’ responses to this item were significantly lower from even the midpoint of the scale, which suggests that vicarious liability is not even a significant concern among business people in our sample with respect to hiring new employees.149

The graphs that illustrate these results appear below. In each graph, the gray vertical line indicates the expected response from participants (either a “7,” indicating that they think frequently about vicarious liability in their hiring decisions, or a “4,” which indicates that they consider vicarious liability at least to some degree in their hiring decisions). The black vertical line indicates the mean response from the businesspeople in our sample, which, in both tests, was significantly lower than the predicted response. The distribution of the responses from businesspeople in our sample is illustrated by the histograms plotted on each axis.150 


Figure 10. Distribution of Responses from Businesspeople I

Fig 10 sevier revised


Figure 11. Distribution of Responses from Businesspeople II

Fig 11 sevier revised


We next compared these results with those of the non-businesspeople in our sample. We found there to be a significant effect of business experience on participants’ concerns about vicarious liability, such that businesspeople thought more about vicarious liability than did non-businesspeople.151 This effect yields two caveats, however. First, neither the mean score on this item for the businesspeople nor for the non-businesspeople reached the midpoint of the scale,152 which suggests that none of our participants seriously considered vicarious liability when selecting new employees. Moreover, the size of the difference between businesspeople and laypeople in our sample was negligible.153

We next examined the extent to which businesspeople prefer to hire the employee featured in Study 2 as an employee or independent contractor. Recall that torts scholars suggest that savvy businesspeople will seriously consider the implications of vicarious liability for their employees’ actions and will seek independent contractors when they can, to avoid respondeat superior liability.154

We found several counterintuitive results from the data in this study, which are illustrated in Figure 12. First, the vast majority of our participants—both businesspeople and non-businesspeople alike—preferred to hire the manager as an employee rather than as an independent contractor, even though the vast majority of our participants reported that they were aware of the differences between an employee and an independent contractor.155 Moreover, business experience was associated with a lower likelihood of hiring the manager as an independent contractor.156


Figure 12. Likelihood of Hiring a Manager as an Employee vs. as an Independent Contractor

Fig 12 sevier revised


Finally, we asked participants to rank seven different criteria when hiring a new employee: the employee’s intelligence, experience, likelihood of causing vicarious liability on the part of the employer, education, personality, likelihood of leaving the business shortly after being hired, and interview performance. Participants ranked these choices from 1 to 7, and higher mean values reflect greater importance of that item to their hiring decisions. A graph representing participants’ responses—which did not differ based on business experience157—appears below. Figure 13 provides the following information: the 95% confidence interval for each item (represented by a gold box),158 the mean value for each item (represented by the horizontal line within each box), and the range of responses from 1 to 7 (represented by the lines extending from the confidence intervals).


Figure 13. Participants’ Responses Regarding Importance of Hiring Criteria

Fig 13 sevier revised


From lowest mean to highest mean, the criteria are ranked as follows: (1) employee experience; (2) employee intelligence; (3) employee personality; (4) employee education; (5) interview performance; (6) vicarious liability; and (7) threat of employee attrition.159 The data therefore suggest that, with respect to this sample of participants, vicarious liability is not a primary consideration in their employee-selection procedures.

C.     Discussion

The results from Study 3 replicated the findings from Study 2. Contrary to commonly held beliefs, mock jurors punish psychologically distant actors—such as corporate defendants—less severely in respondeat superior than they punish psychologically closer actors directly. Moreover, this tendency is exacerbated—as measured by the punitive damages awarded to plaintiffs—when the stakes are higher and the harm alleged is more vivid. Study 3 also examined whether participants’ business experience alters these results.

Although other demographic variables—including participants’ gender and political orientation—affected participants’ perceptions of the corporate defendant’s tort liability, their perceptions were unchanged by their business experience, measured either by business ownership or experience hiring employees. Not only did we fail to find an effect of business experience, the interactive effect that we found in Study 2 remained statistically meaningful even when participants’ business experience was included in the model.

Moreover, the pilot data reported in Study 3 examined participants’ employee-selection strategies. Although torts scholars suggest that business experience may cause businesspeople to hire exclusively independent contractors—to the detriment of tort victims who may be unable to collect on their tort judgments—the pilot results from Study 3 add a counterweight to that discussion. First, although businesspeople in our sample did consider the implications of vicarious liability more than non-businesspeople did, (1) they did not consider those implications heavily in the daily operations of their business, and (2) the difference between businesspeople and non-business people in this respect was small. Further, when participants were asked to rank several factors that they would consider in determining whether to hire a new employee, businesspeople and non-businesspeople ranked the prospect of vicarious liability for that employee’s acts as a less important consideration, ranking it behind the experience, education, personality, interview performance, and intelligence.

Finally, when participants were asked whether they would have hired the employee featured in Study 2 as an employee or an independent contractor, the vast majority of participants—over 95% of businesspeople and laypeople alike—would have hired the actor as an employee, potentially exposing themselves to vicarious tort liability. And counterintuitively, business experience made participants more likely to hire the actor as an employee compared to non-business people, although the effect was small. There are, of course, significant limitations to the pilot data collected in Study 3, which requires caution. But the data serve as preliminary evidence that some business owners do not focus on issues vicarious liability when hiring new employees.

VII.     Implications, Objections, and Conclusions

Torts scholars have written extensively about legal windfalls160, which are defined as “unexpected or sudden gift[s], gain[s] or advantage[s].”161 A legal windfall can take many forms, including an unexpected inheritance, abundant gambling winnings, larger than expected returns on investments, foreign aid, gains from demutualization, and even employment payroll bonuses.162 Windfall profits are a special subtype of this concept, and they have several legal implications, including taxation, accounting, and corporate governance.163 Few legal scholars, however, have examined windfalls in the context of tort liability.

To the extent that torts scholars opine on the implications of windfall profits in the tort litigation, their conclusions lack empirical support and theoretical coherence. These shortcomings are evident in their scholarship on the harmful effects of vicarious liability doctrine on the efficiency of tort law more generally. These scholars argue that the doctrine of vicarious liability unfairly targets corporate defendants, whom jurors punish to a greater degree than they punish individual defendants for the same harms.164 Thus, they contend, savvy employers will hire independent contractors—whose torts they are not required to indemnify under the respondeat superior doctrine—to counteract the windfall that non-corporate defendants enjoy under the law.165

These scholars correctly note that jurors are inconsistent when they assign legal responsibility in factual situations that implicate the vicarious liability doctrine. To the extent that an employee, acting within the ordinary scope of her employment, has committed a tort, an employer is strictly liable for the employee’s harm. Yet, jurors do not punish the individual wrongdoer to the same degree that they punish the individual’s corporate employer. But as the three original experiments reported in this Article suggest, the direction of the difference is surprising: mock jurors punish corporate actors less than they punish their employees.166

Construal level theory predicts that abstract entities who cause harm indirectly are perceived as more psychologically distant from fact finders, and the greater psychological distance causes fact finders to perceive these actors as less morally culpable.167 This Article is the first to challenge the theoretical framework invoked by torts scholars, integrate behavioral science research, and proffer the results of empirical studies to examine how lay fact finders understand vicarious liability.

This Article investigated how laypeople perceive vicarious liability in three original experiments. These experiments suggest that juries bestow upon corporate actors a vicarious windfall, insofar as jurors are less likely to punish corporate defendants vicariously for the acts of their employees.
Study 1 examined mock jurors’ willingness to assign tort liability. It found that, regardless of how liability was measured (either by a dichotomous verdict measure or as a scaled variable), jurors were significantly more likely to assign tort liability to the defendant who recklessly spilled the toxin into the water supply than they were to assign vicarious liability to that actor’s employer. Moreover, a serial mediation analysis demonstrated that jurors’ reluctance to assign liability to the corporate employer was a result of: (1) the greater psychological distance that they felt toward the corporate defendant, which (2) affected their perceptions of the corporate defendant’s moral (and legal) culpability.

Study 2 examined the punitive damages that mock jurors impose in tort litigation. Specifically, it examined whether the differences that emerge between the liability of employee defendants and corporate defendants varies as a function of the harm alleged by the plaintiff. The results suggest that:
(1) jurors are more punitive toward direct tortfeasors, against whom they assigned significantly greater punitive damages awards; and (2) the degree of harm alleged by the plaintiff affects their judgments as well, insofar as this effect is magnified at high levels of harm.

Finally, Study 3 investigated how participants’ business experience affected the results from these studies. The study suggests that businesspeople (and laypeople) are more likely to punish corporate employers in tort law compared to individual employees. Nonetheless, business experience had no effect on participants’ public legal decisions or their private hiring decisions. These studies, taken together, have several implications for the future direction of the vicarious liability doctrine, the contentious debate over jury competency, and attorneys’ strategic choices in tort litigation. These implications—and the limitations and future directions of this research—are discussed below.

A.     Research and Policy Implications

The respondeat superior doctrine was designed, in part, to prevent tort victims from being undercompensated, insofar as they seek legal redress against civil wrongdoers who are likely to file for bankruptcy.168 On the theory that employers exhibit control over their employees—and are therefore in the best position to mold their employees’ behavior to comply with the law—the respondeat superior doctrine opens another avenue—in which the payee has significantly deeper pockets—for tort victims to collect for wrongs perpetrated against them.169 The law assumes, further, that fact finders will apply the doctrine in an evenhanded manner. The studies reported in this Article suggest, however, that fact finders do not apply the doctrine evenhandedly. Rather, corporate defendants enjoy windfall profits to the extent that fact finders are less willing to assign vicarious tort liability to them, compared to their willingness to assign direct liability against the employee wrongdoer.

The data reported in this Article present a counter-weight to proposals calling for stricter standards in imposing vicarious liability on corporate employers.170 These proposals are based on a mistaken, empirically unsupported assumption that corporate employers are penalized unfairly under the law, resulting in the systematic overcompensation of plaintiffs in the tort system.171 In fact, the studies reported in this Article suggest the opposite: the respondeat superior doctrine should be broadened so that the avenue for recovery against corporate defendants is more accessible to tort victims. One potential solution for policymakers (albeit a highly controversial one) is to consider requiring automatic joinder of the corporate employer in third-party claims against its employees under the Federal Rules of Civil Procedure.172 Automatic joinder under this rule would require the addition of the corporate defendant where “in [the corporate defendant’s] absence, the court cannot accord complete relief among existing parties,” particularly when the employee defendant is, or becomes, insolvent.173 Joining the corporate defendant to the claim against the employee might cause the employer’s actions to appear less psychologically distant to fact-finders, which may cause jurors to recalibrate their assessments of tort liability. This suggestion is far from practical in its current instantiation, of course,174 but it provides at least an initial framework for considering how to ensure that similarly situated plaintiffs do not receive different legal outcomes based solely on the strategic choice of which tortfeasor to sue.

Courts can further counteract the effect of the vicarious windfall by requiring lay fact-finders to complete special verdict forms when deciding questions of fact that implicate the respondeat superior doctrine.175Special verdict forms would require participants, explicitly, to find the employer-defendant strictly liable for the acts of its employee–agent (performed in the ordinary course of employment), if they find that the employee is legally responsible for the harm. Because the employer’s vicarious liability would be predicated, explicitly, on the employee’s direct act, it is possible that the effects of the vicarious windfall would be reduced or obviated, because the fact finder’s attention would be focused on the concrete actor.176 Researchers should continue to examine de-biasing strategies to ensure that jurors apply the vicarious liability doctrine equitably.

The results reported in this Article also have implications for tort litigants. There are always additional costs to joining multiple parties to a lawsuit, including the labor expended to investigate additional parties, give notice to them, take their depositions, evaluate their discovery responses, file motions against them, respond to their cross-claims, and question them at trial. In litigation that implicates the vicarious liability doctrine, failing to join all relevant parties will leave litigants in a troubling quandary: do litigants pursue the individual employee—whom juries are more likely to find liable, but who may become insolvent—or do they pursue the employer instead—who has deeper pockets, but against whom juries are significantly less likely to assign tort liability? Until courts require automatic joinder of the parties—and make the expense of doing so more palatable to tort plaintiffs—a failure to join all parties to the litigation becomes a risky trap for the unwary litigant.177

Finally, behavioral scholars should expand the research reported here to other contexts within the tort system. Specifically, researchers should determine whether other tort actors receive vicarious windfalls by virtue of their psychological distance from the fact finder. For example, we have conducted pilot studies suggesting that this phenomenon applies not only to the corporate defendant, but also to insurance underwriters for the harms of the insured and to police departments for the harms committed by government officers and officials.178 Researchers should also examine state and federal dockets to determine—with appropriate statistical controls—the extent to which a fact finder’s leniency toward psychologically distant actors manifests itself in state and federal jury verdicts.179

B.     Objections and Future Directions

The findings reported in this Article provide new evidence that challenges tort scholars’ assumptions about the vicarious tort liability of corporate defendants. Controlled behavioral experiments are, of course, subject to certain limitations that should be acknowledged by both researchers and policymakers.

The studies reported in this Article are controlled experiments, which differ from studies that use uncontrolled field data. Controlled experiments, unlike field data, are conducted in a uniform setting in an effort to avoid the confounding effects of third variables unassociated with the research question at hand.180 This does not, however, mean that controlled experiments are inferior to field data. Rather, each design has both benefits and trade-offs. Field experiments, for example, are externally more valid than controlled experiments, because they may more accurately reflect behavior in real trials.181 Controlled experiments, which take place outside the real world context, are riskier with respect to external validity, although several studies suggest that this concern may be overstated.182 But controlled experiments are far more advantageous with respect to internal validity, which is the ability of the study to truly measure what it claims to measure, free of the influence of unknown third variables or incorrect inferences of causality.183

Specifically, controlled experiments have greater internal validity than do field studies because the design of the study is kept uniform in all respects with the exception of the variables studied in the experiment. 184 If participants’ behavior differs across the different experimental conditions, researchers in a controlled experiment can state more confidently than a researcher who uses a field methodology that the cause of the differing behavior is the variable that was studied. It is possible for field researchers to control for potential confounding factors in their experimental designs post hoc, but researchers generally agree that a controlled laboratory design remains the gold standard for making causal statements about the effects of environmental variables on human behavior.185

The vignettes employed in the studies reported here provide important information regarding how participants evaluate claims of a corporate defendant’s vicarious tort liability. The trials in each study were the same in nearly every respect—except the identity of the defendant, the role of the participant, and the degree of harm that was alleged—and revealed stark differences among participants with respect to the manner in which they evaluated the tort claims at issue. Although field studies are weaker than laboratory experiments with respect to the causal inferences that readers can draw from them, field studies of vicarious liability determinations in the corporate context—which have not yet been conducted by empirical researchers—should be conducted. Using publicly available data, ambitious researchers could examine real world vicarious liability claims, compare jury verdicts across similarly situated cases involving direct liability and vicarious liability, and code for (and ultimately control for) factors such as the nature and severity of the harm alleged, the demographics of the relevant legal actors, and the complexity of the trial in order to draw conclusions from state and federal cases. The external and ecological validity reported by such cases would supply convergent validity for how laypeople and businesspeople think about vicarious liability.186 At a minimum, other researchers should consider replicating the results reported in this Article in a videotaped trial or a live reenactment.187

The results from Study 3 are worth a separate cautionary note. The sample included a diverse group of past and present business owners who participate in the online recruiting mechanism called mTurk. The clear takeaway message from that study—that vicarious liability does not appear explicitly to be a prominent factor in hiring employees—should be read in the light of the fact that these participants owned and managed relatively small businesses. The hiring practices of large corporate firms with respect to the respondeat superior doctrine therefore likely remains an open question for future researchers.

VIII.     Conclusions

Turing Pharmaceuticals CEO Martin Shkreli repeatedly invoked the Fifth Amendment when he testified before Congress in early 2016, but his impression of the proceedings became clear shortly afterward.188 In a series of exasperated posts on social media, Shkreli derided Congress and expressed surprise over the public backlash regarding his decision to raise the price of an HIV medication by 5,000%.189 He may also have wondered why companies who performed similar acts—but through intermediaries, as in the Merck & Co. case—did not receive similar criticism. Behavioral researchers, however, do not share Shkreli’s surprise. Their research suggests that fact finders assign greater moral culpability to actors who directly cause harm—compared to indirect actors—as a function of the psychological distance between the transgressor and the decision maker.

This phenomenon forms the basis of the vicarious windfall, which suggests that—contrary to the intuitions of laypeople, businesspeople, and even torts scholars—jurors are less likely to assign vicarious liability to third-party employers for the torts of their employees than the law anticipates. The windfall is pernicious precisely because it is so counterintuitive, and the behavioral research to date suggests that proposals to improve the vicarious liability doctrine may unwittingly exacerbate it. Policymakers must call on empirical scholars to continue to research these legal doctrines for unintended consequences. The public’s confidence in the accuracy and equity of their legal institutions—and therefore the legitimacy of these institutions in the eyes of the public—may depend on it.

  1. [1]. Windfall, Online Etymology Dictionary,
    allowed_in_frame=0&search=windfall (last visited Oct. 28, 2016). In medieval England, commoners were forbidden from chopping down trees for fuel. But if a strong wind broke the branches or blew down the trees, the debris was a legitimate find. It was to the commoner, literally, a windfall. See William Morris & Mary Morris, Morris Dictionary of Word and Phrase Origins 605 (1977) (explaining the etymology of the word “windfall”); see also generally Eric Kades, Windfalls, 108 Yale L.J. 1489 (1999) (citing the Morris Dictionary and discussing a varying array of public and private legal windfalls).

  2. [2]. See, e.g., Hazel Sheffield, Martin Shkreli, the CEO that Hiked Price of HIV Treatment Drug, Arrested on Securities Fraud Charges, Independent (Dec. 17, 2015),

  3. [3]. See id.; see also James Surowiecki, Taking on the Drug Profiteers, New Yorker (Oct. 12, 2015),; Arlene Weintraub, Gadfly Pharma Investor Shkreli Starts Anew After Ousting from Retrophin, Forbes (Feb. 27, 2015, 9:10 AM),

  4. [4]. See, e.g., Dave Muoio, Daraprim Price Jump Raises Concerns Among ID Groups, Providers, Healio (Sept. 17, 2015),

  5. [5]. Id.; see also Andrea Mitchell & Phil Helsel, Drug CEO Will Lower Price of Daraprim After Hike Sparked Outrage, NBC News (Sept. 23, 2015, 6:43 AM),; Andrew Pollack, Drug Goes from $13.50 a Tablet to $750, Overnight, N.Y. Times (Sept. 20, 2015),

  6. [6]. Heather Long & Matt Egan, Meet the Guy Behind the $750 AIDS Drug, CNN: Money (Sept. 22, 2015, 7:50 PM), Prominent critics of Shkreli included Hillary Clinton, Donald Trump, and Bernie Sanders. Id.; see also Ariana Eunjung Cha, CEO Who Raised Price of Old Pill More than $700 Calls Journalist a ‘Moron’ for Asking Why, Wash. Post (Sept. 22, 2015),; Brett LoGiurato, Donald Trump Trashes Former Hedge-Fund Guy Who Jacked Up Drug Price: ‘He Looks like a Spoiled Brat’, Bus. Insider (Sept. 23, 2015, 6:04 PM),

  7. [7]. See, e.g., Nathan Bomey, Kevin McCoy & Kim Hjelmgaard, Shkreli, CEO Slammed over Drug Prices, $5M Bond, USA Today (Dec. 18, 2015, 7:48 AM),
    money/2015/12/17/martin-shkreli-arrested/77469752; Jason Tannahill, PR Man Allan Ripp Representing the “Most Hated Man in America, Everything-PR (Oct. 9, 2015), Shkreli was later indicted by federal prosecutors on unrelated charges of securities fraud, stemming from his tenure at MSMB Capital Management and at Retrophin. See Christie Smythe & Keri Geiger, Shkreli, Drug Price Gouger, Denies Fraud and Posts Bail, Bloomberg (Dec. 17, 2015), He was later called upon to give testimony to the United States House of Representatives stemming from his actions at Turing Pharmaceuticals, among other companies. See Carolyn Y. Johnson, ‘Pharma Bro’ Shkreli Stays Silent Before Congress, Calls Lawmakers ‘Imbeciles’ in Tweet, Wash. Post (Feb. 4, 2016),

  8. [8]. See Alex Berenson, A Cancer Drug’s Big Price Rise Is Cause for Concern, N.Y. Times (Mar. 12, 2006), (discussing the effects of sale of Mustargen and Cosmegen on average consumers).

  9. [9]. See id. (“This is not the first time that Ovation has sharply raised the price of a drug it owns. In 2003, the company bought Panhematin, a treatment for a rare enzymatic disease called porphyria, from Abbott Laboratories. While Abbott still produces Panhematin, Ovation raised Panhematin’s price, which had been $230 a dose, to $1,900, according to Desiree Lyon, executive director of the American Porphyria Foundation.”); see also FTC Alleges Price Gouging, Illegal Monopoly for Baby Medicines, Wash. Post (Dec. 17, 2008),
    16/AR2008121602674.html (describing a similar deal between Merck and Ovation in 1985, which resulted in the price increase of a heart defect medication).

  10. [10]. See Berenson, supra note 8.

  11. [11]. See generally Christensen v. Swenson, 874 P.2d 125 (Utah 1994) (finding an employer vicariously liable for the tort of its employee, who collided with a motorcycle rider on her way back from an unscheduled lunch break). A robust discussion of the doctrine of respondeat superior, its history, and scholarly views of its usefulness appears in Part II.

  12. [12]. See generally Yaacov Trope & Nira Liberman, Construal-Level Theory of Psychological Distance, 117 Psychol. Rev. 440 (2010); see also infra Part III.A (discussing construal level theory in detail).

  13. [13]. See generally Yoav Bar-Anan, Nira Liberman & Yaacov Trope, The Association Between Psychological Distance and Construal Level: Evidence From an Implicit Association Test, 135 J. Experimental Psychol. 609 (2006); see also infra Part III.B (discussing the concept of psychological distance).

  14. [14]. See generally Trope & Liberman, supra note 12. Conversely, the more we construe a social stimulus as concrete, the more likely we are to attribute the stimulus as psychologically closer to us. For example, people tend to think of brand name food items, such as a can of Coca-Cola, as psychologically close, whereas they think of soda pop in the abstract as psychologically distant. See generally id.

  15. [15]. See generally Han Gong & Douglas L. Medin, Construal Levels and Moral Judgment: Some Complications, 7 Judgment & Decision Making 628 (2012); Neeru Paharia et al., Dirty Work, Clean Hands: The Moral Psychology of Indirect Agency, 109 Organizational Behav. & Hum. Decision Processes 134 (2009). An in-depth discussion of the relationship between psychological distance and perceptions of moral culpability appears in Part III.B of this Article.

  16. [16]. For a general discussion of the views of these academics, see Yoed Halbersberg & Ehud Guttel, Behavioral Economics and Tort Law, in The Oxford Handbook of Behavioral Economics and the Law 405, 425–32 (Eyal Zamir & Doron Teichman eds., 2014).

  17. [17]. See id.

  18. [18]. See generally Valerie P. Hans, The Contested Role of the Civil Jury in Business Litigation, 79 Judicature 242 (1996) (discussing these arguments and challenging their validity through opinion surveys, interviews, and mock trial experiments).

  19. [19]. See Jennifer H. Arlen & W. Bentley MacLeod, Beyond Master–Servant: A Critique of Vicarious Liability, in Exploring Tort Law 111, 111 (M. Stuart Madden ed., 2005) (“By holding organizations liable for torts committed by employees, but not by independent contractors, vicarious liability discourages organizations from asserting direct control over agents, even when control is the efficient way to induce optimal care.”).

  20. [20]. See Halbersberg & Guttel, supra note 16, at 425–28 (discussing this theory in greater detail).

  21. [21]. See id.; see also Valerie P. Hans, Business on Trial: The Civil Jury & Corporate Responsibility 99–106 (2000) (including anecdotes from various corporate actors suggesting that this may occur, but noting that empirical evidence likely does not support this belief).

  22. [22]. See Christensen v. Swenson, 874 P.2d 125, 129 (Utah 1994); Clark C. Havighurst, Vicarious Liability: Relocating Responsibility for the Quality of Medical Care, 26 Am. J.L. & Med. 7, 8 (2000) (applying the vicarious liability doctrine to the medical context and recommending that a health plan be held vicariously “liable for medical malpractice... committed by health care providers whom it procures to treat its enrollees”); see also generally Jennifer Arlen, The Potentially Perverse Effects of Corporate Criminal Liability, 23 J. Legal Stud. 833 (1994) (criticizing as unjust the vicarious liability doctrine as it is sometimes applied in the criminal context); Martha Chamallas, Vicarious Liability in Torts: The Sex Exception, 48 Val. U. L. Rev. 133 (2013) (critiquing the vicarious liability doctrine in the context of sexual abuse cases); George M. Cohen, Liability of Insurers for Defense Counsel Malpractice, 68 Rutgers U. L. Rev. 119 (2015) (applying the vicarious liability doctrine to the context of legal representation); Paula J. Dalley, All in a Day’s Work: Employers’ Vicarious Liability for Sexual Harassment, 104 W. Va. L. Rev. 517 (2002) (discussing the complications for the vicarious liability doctrine when it is invoked in workplace harassment contexts); Timothy P. Glynn, Beyond “Unlimiting” Shareholder Liability: Vicarious Tort Liability for Corporate Officers, 57 Vand. L. Rev. 329 (2004) (arguing that the underlying policies that govern the vicarious liability doctrine and the law of corporations justifies applying the vicarious liability doctrine to a corporation’s officers for the corporation’s legal obligations); Joseph H. King, Jr., Limiting the Vicarious Liability of Franchisors for the Torts of Their Franchisees, 62 Wash. & Lee L. Rev. 417 (2005) (examining vicarious liability in the franchise context and recommending that the franchisor–franchisee relationship should be treated as a special instantiation of vicarious liability); J.W. Neyers, A Theory of Vicarious Liability, 43 Alta. L. Rev. 287 (2005) (exploring the dominant rationales for the vicarious liability doctrine, arguing that those rationales are not convincing, and proposing an indemnity justification for vicarious liability); Gary T. Schwartz, The Hidden and Fundamental Issue of Employer Vicarious Liability, 69 S. Cal. L. Rev. 1739 (1996) (arguing that an employer’s vicarious liability for its employee’s actions frequently is obscured by courts, law schools, and politicians); Alan O. Sykes, The Boundaries of Vicarious Liability: An Economic Analysis of the Scope of Employment Rule and Related Legal Doctrines, 101 Harv. L. Rev. 563 (1988) (providing an economic justification for the vicarious liability doctrine and discussing its limitations); Alan O. Sykes, The Economics of Vicarious Liability, 93 Yale L.J. 1231 (1984) (describing the vicarious liability doctrine in terms of incentive structures and costs to legal actors); Rebecca Hanner White, Vicarious and Personal Liability for Employment Discrimination, 30 Ga. L. Rev. 509 (1996) (debating the wisdom of expanding the vicarious liability doctrine in workplace disputes other than discrimination claims).

  23. [23]. See Shawn J. Bayern, Closely Held Organizations 49–83 (2014) (discussing the “hybrid” nature of vicarious liability in tort law and raising as an issue its doctrinal coherence).

  24. [24]. See John H. Wigmore, Responsibility for Tortious Acts: Its History, 7 Harv. L. Rev. 315,
    315–18 (1894) (discussing the history of the doctrine in England and its expansion over time).

  25. [25]. See William F. Walsh, A History of Anglo-American Law 322 (2d ed. 1932) (The narrow rule requiring that the master give a specific command in order to impute liability was justified by the immediate control the master had over his servants. With the great expansion of commercial life and the subsequent need for entrusting servants with broader responsibility, this justification no longer truly existed and business demanded that the master be held liable for the acts of his representatives.); Christine W. Young, Respondeat Superior: A Clarification and Broadening of the Current “Scope of Employment” Test, 30 Santa Clara L. Rev. 599, 601 n.16 (quoting Blackstone’s summary of the development of the doctrine).

  26. [26]. See Young, supra note 25, at 601–02.

  27. [27]. Wright v. Wilcox, 19 Wend. 343 (N.Y. Sup. Ct. 1838). Oddly, however, the court refused to find the employer liable, despite declaring that the law required the employer to pay the judgment vicariously. Id. at 346–48.

  28. [28]. Some American jurisdictions, however, sometimes departed from the English rule, or modified it, as occurred in California. See Otis Elevator Co. v. First Nat’l Bank of S.F., 124 P. 704, 711 (Cal. 1912) (declining to find an employer bank liable for the fraudulent activity of its employee, of which the bank had no knowledge); see also Ruppe v. City of Los Angeles, 199 P. 496, 497 (Cal. 1921) (departing from the English rule).

  29. [29]. See, e.g., Bayern, supra note 23, at 49–83 (discussing the doctrine). The doctrine of respondeat superior, a subset of vicarious liability, is discussed in various tort cases. See Lyon v. Carey, 533 F.2d 649, 651 (D.C. Cir. 1976); Ira S. Bushey & Sons, Inc. v. United States, 398 F.2d 167, 170–71 (2d Cir. 1968); Lisa M. v. Henry Mayo Newhall Mem’l Hosp., 907 P.2d 358,
    359–67 (Cal. 1995); Christensen v. Swenson, 874 P.2d 125, 126–27 (Utah 1994); Heims v. Hanke, 93 N.W.2d 455, 457–58 (Wis. 1958); see also Fiocco v. Carver, 137 N.E. 309, 310–11 (N.Y. 1922) (discussing respondeat superior in the historical context of the
    master–servant relationship).

  30. [30]. See Bayern, supra note 23, at 49–53 (discussing the agency underpinnings of the doctrine); see also Heims, 93 N.W.2d at 457–58 (same).

  31. [31]. See Young, supra note 25, at 603–24 (discussing the historical expansion and contraction of the scope requirement).

  32. [32]. See, e.g., Christensen, 874 P.2d at 127–28 n.1 (applying the test flexibly in a vehicular negligence action); Fiocco, 137 N.E. at 310–11 (illustrating the difficulty of returning to the scope of employment once an employee has stepped outside of it); cf. Lyon, 533 F.2d at 651–55 (discussing foreseeability as the test of enterprise liability in intentional tort cases); Ira S. Bushey & Sons, Inc. 398 F.2d at 170–73 (discussing how, even if an employee does an act which is not motivated by serving the interests of his employer, the third prong can be fulfilled based upon enterprise and foreseeability considerations); Lisa M., 907 P.2d at 362–66 (discussing the foreseeability test and its use in determining enterprise liability).

  33. [33]. See Bayern, supra note 23, at 49–50 (discussing the doctrine of respondeat superior); see also generally Sykes, The Boundaries of Vicarious Liability, supra note 22 (developing economic principles to examine the employment rule).

  34. [34]. See Bayern, supra note 23, at 50 (discussing the consequences of failing to prove the elements).

  35. [35]. See generally Neyers, supra note 22; Schwartz, supra note 22.

  36. [36]. Neyers, supra note 22, at 289–91.

  37. [37]. Id.

  38. [38]. See generally Sykes, The Economics of Vicarious Liability, supra note 22 (describing the vicarious liability doctrine in terms of incentive structures and costs to legal actors).

  39. [39]. See, e.g., id.

  40. [40]. Vicarious liability, these supporters argue, allows another avenue from which a tort victim may collect: the corporate employer, who can better absorb the cost and who is likely to have deeper pockets to pay tort judgments than does an individual employee.

  41. [41]. See Hans, supra note 21, at 178–214 (discussing these “Robin Hood” concerns).

  42. [42]. Id. at 178; see also id. at 178–214 (discussing the myth of the “Robin Hood” jury that “steals” from the rich corporation and gives to poor plaintiffs). In one of Professor Hans’s studies reported in her book—which documented business people’s attitudes toward corporate litigation—one executive quipped: “Is it any surprise that many commercial contracts these days have a clause where each party waives its right to a jury trial? Doesn’t that tell you something? That they are not willing to trust twelve peers off the street with the complexity of their business transaction.” Id. at 138 (footnote omitted).

  43. [43]. See Halbersberg & Guttel, supra note 16, at 425–37 (outlining the argument and setting forth the debate). Such a shift in personnel selection will likely undercompensate tort victims because corporations will not be held vicariously liable for the acts of independent contractors.

  44. [44]. See Hans, supra note 21, at 112–37. Hans concludes that, based on empirical research, “[a]t least some jurors.... believe that a different standard of responsibility is appropriate given the asymmetries in resources and potential impact of individual versus corporate action,” but notes that many other jurors do not hold that belief. Id. at 136–37.

  45. [45]. For a robust discussion of this empirical evidence from experimental social psychology, see infra Part III.

  46. [46]. Even if they do believe that jurors exhibit an anti-corporate bias, if jurors do not, in fact exhibit the bias—in accordance with behavioral research—that creates an interesting paradox worth additional scholarly exploration.

  47. [47]. See supra notes 2–10 and accompanying text.

  48. [48]. See Justin Sevier, Testing Tribe’s Triangle: Juries, Hearsay, and Psychological Distance, 103 Geo. L.J. 879, 896–900 (2015) (describing how psychological distance relates to hearsay evidence).

  49. [49]. See, e.g., Lawrence E. Williams & John A. Bargh, Keeping One’s Distance: The Influence of Spatial Distance Cues on Affect and Evaluation, 19 Psychol. Sci. 302, 302–03 (2008).

  50. [50]. See Jean M. Mandler, How to Build a Baby: II. Conceptual Primitives, 99 Psychol. Rev. 587, 589 (1992) (discussing infants’ ability to discriminate objects as a function of distance at three months of age); see also Herbert H. Clark, Space, Time, Semantics, and the Child, in Cognitive Development and the Acquisition of Language 27, 38 (Timothy E. Moore ed., 1973) (describing differences between objects that are perceived to be near or far).

  51. [51]. Mandler, supra note 50, at 595–96. A classic experiment in this area revealed that participants were quicker to categorize positive words on a computer screen when the words appeared at the top of the screen, whereas they were quicker to categorize negative words on the screen when they appeared at the bottom. Brian P. Meier & Michael D. Robinson, Why the Sunny Side Is Up: Associations Between Affect and Vertical Position, 15 Psychol. Sci. 243, 246–47 (2004). The transition from thinking of distance solely in terms of physical space earlier in life to thinking about distance psychologically later in life appears to be the result of increases in reasoning and critical thinking abilities over time. See generally Timothy D. Wilson, Strangers to Ourselves: Discovering the Adaptive Unconscious (2002) (discussing the difficulty that fully developed adults have understanding their internal states); see also Williams & Bargh, supra note 49, at 303 (discussing how “physical-distance cues have adaptive significance”).

  52. [52]. See Trope & Liberman, supra note 12, at 441.

  53. [53]. See id. at 442–44 (discussing the interrelationship between the four dimensions).

  54. [54]. Sevier, supra note 48, at 897­­­­­–98 (footnotes omitted).

  55. [55]. Trope & Liberman, supra note 12, at 440.

  56. [56]. Id. at 441; see also Williams & Bargh, supra note 49, at 302.

  57. [57]. Similarly, a soda can be represented as a “Diet Coke” (a low-level construal) or a “drink” (a high-level construal). Trope & Liberman, supra note 12, at 441, 444–45, 449; see also Sevier, supra note 48, at 898–99 (comparing different representations of a cellular telephone).

  58. [58]. Sevier, supra note 48, at 899. “Because abstract representations necessarily impose one of many alternative interpretations, and because irrelevant or inconsistent details are omitted or assimilated to it, these representations tend to be simpler, less ambiguous, more coherent, more schematic, and more prototypical than concrete representations.” Trope & Liberman, supra note 12, at 441.

  59. [59]. Sevier, supra note 48, at 896–904 (examining how psychological distance affects mock jurors’ perceptions of hearsay declarants and the evidence they produce).

  60. [60]. Id. at 917.

  61. [61]. Id. at 920.

  62. [62]. See, e.g., Paharia et al., supra note 15 (performing an original experiment on and discussing the state of the empirical literature regarding the intersection of psychological distance and moral attributions).

  63. [63]. See generally Shaun Nichols, Sentimental Rules: On the Natural Foundations of Moral Judgment (2004) (outlining these principles); 1 Moral Psychology: The Evolution of Morality: Adaptations and Innateness (Walter Sinnott-Armstrong ed. 2008) (same); Philip E. Tetlock et al., The Psychology of the Unthinkable: Taboo Trade-Offs, Forbidden Base Rates, and Heretical Counterfactuals, 78 J. Personality & Soc. Psychol. 853 (2000) (discussing these themes).

  64. [64]. See, e.g., Daniel K. Lapsley, Moral Psychology (1996) (analyzing moral psychology from a developmental perspective); see also generally 1 Lawrence Kohlberg, Essays on Moral Development: The Philosophy of Moral Development (1981) (discussing, among other topics, his influential theory of moral development).

  65. [65]. See, e.g., Monica Bucciarelli, Sangeet Khemlani & P. N. Johnson-Laird, The Psychology of Moral Reasoning, 3 Judgment & Decision Making 121, 122 (2008) (discussing moral judgment); see also generally Elliott Sober & David Sloan Wilson, Unto Others: The Evolution and Psychology of Unselfish Behavior (1998) (discussing evolution); Augusto Blasi, The Moral Functioning of Mature Adults and the Possibility of Fair Moral Reasoning, in Personality, Identity, and Character: Explorations in Moral Psychology 396–440 (Darcia Narvarez & Daniel K. Lapsley eds., 2009) (discussing ethical codes); see also generally Sam A. Hardy & Gustavo Carlo, Identity as a Source of Moral Motivation, 48 Hum. Dev. 232 (2005) (discussing moral identity).

  66. [66]. See Jonathan Haidt, The Righteous Mind: Why Good People Are Divided by Politics and Religion 52–82 (2012) (outlining the “Social Intuitionist Model” of moral judgment, which challenges the traditional, rational model of moral reasoning proffered by Kohlberg).

  67. [67]. See id.

  68. [68]. See generally Jonathan Haidt, The Emotional Dog and Its Rational Tail: A Social Intuitionist Approach to Moral Judgment, 108 Psychol. Rev. 814 (2001).

  69. [69]. Id. at 819. Haidt reached this conclusion based on his experiments—and those of other researchers—on “moral dumbfounding,” in which people make negative moral attributions toward social stimuli, but, when pressed regarding the reason for their moral condemnation, they are unable to defend or articulate their reasoning. See id. at 822–25; see also generally Simone Schnall et al., Disgust as Embodied Moral Judgment, 34 Personality & Soc. Psychol. Bull. 1096 (2008).

  70. [70]. See, e.g., R. B. Zajonc, On the Primacy of Affect, 39 Am. Psychologist 117 (1984) (reviewing the literature from different areas of psychology and concluding that emotional reactions precede cognitive thought in human judgment).

  71. [71]. See generally Stanley Milgram, Obedience to Authority (1974) (discussing his classic experiment on obedience and reporting the results of several variations on the original experiment).

  72. [72]. Id. at 3.

  73. [73]. Id. at 4.

  74. [74]. Id.

  75. [75]. Id. at 5.

  76. [76]. Id. at 11.

  77. [77]. Id.

  78. [78]. See id.; see also Paharia et al., supra note 15, at 135. Analogous findings have been reported in the legal context. Jurors who sentenced a defendant to death in capital trials were highly likely to highlight the role of an indirect actor—either the trial judge, the state executioner, or even the defendant herself—when describing the cause of the defendant’s death. See Steven J. Sherman, The Capital Jury Project: The Role of Responsibility and How Psychology Can Inform the Law, 70 Ind. L.J. 1241, 1244 (1995).

  79. [79]. See generally Paharia et al., supra note 15.

  80. [80]. Id. at 140.

  81. [81]. See generally Gong & Medin, supra note 15 (observing that over the course of several experiments, when researchers manipulated the construal levels—either abstract or concrete—by which participants evaluated social stimuli, participants’ moral judgments of those social stimuli were consistent with the theory that concrete actors are judged more harshly than are abstract, distant actors).

  82. [82]. See id. at 628; see also generally Eugene Borgida & Richard E. Nisbett, The Differential Impact of Abstract vs. Concrete Information on Decisions, 7 J. Applied Soc. Psychol. 258 (1977); Jochim Hansen & Michaela Wänke, Truth from Language and Truth from Fit: The Impact of Linguistic Concreteness and Level of Construal on Subjective Truth, 36 Personality & Soc. Psychol. Bull. 1576, 1576–78 (2010).

  83. [83]. Gün R. Semin & Eliot R. Smith, Revisiting the Past and Back to the Future: Memory Systems and the Linguistic Representation of Social Events, 76 J. Personality & Soc. Psychol. 877, 877–78 (1999); see also generally Gün R. Semin & Klaus Fiedler, The Cognitive Functions of Linguistic Categories in Describing Persons: Social Cognition and Language, 54 J. Personality & Soc. Psychol. 558 (1988).

  84. [84]. See Gong & Medin, supra note 15, at 628.

  85. [85]. See Joris Lammers, Abstraction Increases Hypocrisy, 48 J. Experimental Soc. Psychol. 475, 479 (2012) (reporting these results in the context of hypocrisy research).

  86. [86]. See supra Part II.B.

  87. [87]. See supra Part II.B.

  88. [88]. See supra Part II.B.

  89. [89]. See supra Part II.B.

  90. [90]. See supra Part III.B.

  91. [91]. Studies have shown that mTurk is an inexpensive mechanism for collecting quality data from a generally representative sample of the internet-using population. See, e.g., Adam J. Berinsky, Gregory A. Huber & Gabriel S. Lenz, Evaluating Online Labor Markets for Experimental Research:’s Mechanical Turk, 20 Pol. Analysis 351, 366 (2012); Michael Buhrmester, Tracy Kwang & Samuel D. Gosling, Amazon’s Mechanical Turk: A New Source of Inexpensive, Yet High-Quality, Data?, 6 Persp. on Psychol. Sci. 3, 5 (2011); Winter Mason & Siddharth Suri, Conducting Behavioral Research on Amazon’s Mechanical Turk, 44 Behav. Res. Methods 1, 4 (2012).

  92. [92]. The arguments and claims made in this Article are entirely the author’s. The phrase “we” is used throughout the Article to acknowledge the dedicated work of the research assistants and research librarians who assisted the author in designing the study. The pronoun is used in that collective spirit.

  93. [93]. The experiment therefore followed a 2 (role: observer vs. mock juror) x 2 (defendant: employee vs. employer) experimental design.

  94. [94]. At this point, participants learned different information regarding the case’s procedural posture. Half of the participants learned that the plaintiffs had sued the company employee whom they allege caused the spill—a man named John Brown who managed the power plant facility. The other participants learned that the plaintiffs had instead sued the corporation itself, the Acme Power Plant Company, on a theory of vicarious liability for the actions of its employee. After receiving this information regarding the case’s procedural posture, participants read about the plaintiffs’ and the defense’s evidence in the case.

  95. [95]. A Likert Scale is a psychometric scale commonly used in questionnaires to record data from participants that takes the form of an ordinal variable (for example, categories from 1 to 7). See Robert M. Lawless et al., Empirical Methods in Law 172 (2010).

  96. [96]. Skewness is a measure of the asymmetry of the probability distribution of a dependent variable, which can be positive or negative, or undefined. Eric W. Weisstein, Skewness, Wolfram MathWorld, (last visited Oct. 28, 2016); see also D. N. Joanes & C. A. Gill, Comparing Measures of Sample Skewness and Kurtosis, 47 J. Royal Stat. Soc’y 183, 183 (1998). Kurtosis is a measure of the “tailedness” of the probability distribution of a dependent variable. See generally Peter H. Westfall, Kurtosis as Peakedness,
    1905–2014. R.I.P.
    , 68 Am. Statistician 191 (2014).

  97. [97]. See, e.g., Theodore Eisenberg & Michael Heise, Judge-Jury Difference in Punitive Damages Awards: Who Listens to the Supreme Court?, 8 J. Empirical Legal Stud. 325, 335 (2011) (“We use logarithmic scales because, as is typical with award amounts, linear scales fail to reveal the relation between the variables.”).

  98. [98]. See id.

  99. [99]. See supra note 96 (defining kurtosis and skewness).

  100. [100]. Earl R. Babbie, The Practice of Social Research 425–29 (14th ed. 2016) (discussing simple correlations).

  101. [101]. Because we examine punitive damages in greater detail in Study 2, we focus in Study 1 solely on measures of liability. The Results reported in Study 1 did, however, replicate to damages awards.

  102. [102]. An analysis of variance (“ANOVA”) provides a statistical test of whether the means of several groups are equal. ANOVA results are represented by an F-statistic, and the sizes of the effects are represented by η2p. Means are denoted by the letter “M” and standard deviations are denoted by the letters “SD.” See Lawless et al., supra note 95, at 277–85 (explaining empirical research methodologies and statistical techniques). Differences are denoted as “statistically significant” in this Article if the statistical tests indicate that the likelihood that the difference observed would occur by chance is 5% or less (as indicated by the p-value as p < 0.05). A difference is “marginally significant” if the likelihood of seeing such a difference by chance is greater than 5% but less than 10%. Jennifer K. Robbennolt, Apologies and Legal Settlement: An Empirical Examination, 102 Mich. L. Rev. 460, 485 n.117 (2003) (citing generally Barbara G. Tabachnick & Linda S. Fidell, Using Multivariate Statistics (2d ed. 1989)).

  103. [103]. In technical terms, we conducted a 2 (role: observer vs. juror) x 2 (identity of the defendant: employee vs. corporation) ANOVA on participants’ willingness to assign tort liability to the defendant.

  104. [104]. F(1, 298) = 7.49, p = .007, η2= .03.

  105. [105]. MCorporation = 5.37 (SD = 1.47), MEmployee = 5.02 (SD = 1.53); F(1, 99) = 1.39, p < .241, η2p = .01.

  106. [106]. MCorporation = 4.31 (SD = 1.85), MEmployee = 5.08 (SD = 1.66); F(1, 199) = 9.56, p < .002, η2p = .05.

  107. [107]. A logistic regression is a regression analysis that examines whether several variables independently predict a binary, dichotomous outcome, such as a guilty or not guilty verdict. See Lawless et al., supra note 95, at 345–50 (discussing logistic regressions).

  108. [108]. The odds ratio on the right side of the table represents the odds that an individual would choose one of the binary options given the effect of the independent variable. Id. at 346. The “B” value indicates the log odds of the effect of the independent variable, and the Wald statistic affects the p-value, which measures the reliability of the effect. See id. at 346–50.

  109. [109]. Mediation analysis detects when predictor variables influence dependent variables “through intervening or mediator variables.” Kristopher J. Preacher & Andrew F. Hayes, Asymptotic and Resampling Strategies for Assessing and Comparing Indirect Effects in Multiple Mediator Models, 40 Behav. Res. Methods 879, 879 (2008). The mediation analysis reported in this Article is performed using a linear regression analysis and reports unstandardized coefficients, “B,” and standard errors, “SE.” See id. at 879–80. It also reports a “t” statistic, which determines whether the coefficients are statistically significant. See id. at 880. A linear regression is a statistical test that estimates the independent effects of several predictor variables on a continuous dependent variable. See Lawless et al., supra note 95, at 29, 300–31.

  110. [110]. Preacher & Hayes, supra note 109, at 879.

  111. [111]. See id.

  112. [112]. See id. (discussing the theoretical and statistical import of mediation analyses).

  113. [113]. See id.

  114. [114]. B = 1.00, SE = 0.32, z = 3.14, p = .002.

  115. [115]. B = ­0.47, SE = 0.22, t = ­2.13, p = .034.

  116. [116]. B = 0.46, SE = 0.06, t = 7.55, p < .001.

  117. [117]. B = ­0.38, SE = 0.13, z = ­2.82, p = .005.

  118. [118]. B = 0.08, SE = 0.05, 95% CI [0.01, 0.21].

  119. [119]. Asterisks in the mediation analysis indicate statistically significant associations; p < 0.05.

  120. [120]. See, e.g., Justin Sevier, The Unintended Consequences of Local Rules, 21 Cornell J.L. & Pub. Pol’y 291, 331 (2011) (examining the effect of vivid information on jury verdicts).

  121. [121]. This study therefore followed a 2 (defendant: employer vs. employee) x 2 (harm: low vs. high) experimental design.

  122. [122]. See supra Table 2 and accompanying text (explaining bivariate correlations).

  123. [123]. See Joanes & Gill, supra note 96, at 183 (explaining skewness and kurtosis).

  124. [124]. MLow-Harm = 3.56 (SD = 1.95), MHigh-Harm = 4.22 (SD = 2.31); F (1, 197) = 5.33, p = .022, η2p = .03.

  125. [125]. F(1, 197) = 3.43, p < .066, η2p = .02.

  126. [126]. MCorporation = 3.35 (SD = 2.07), MEmployee = 3.78 (SD = 1.83); F(1, 101) = 1.30, p < .257, η2p = .01.

  127. [127]. MCorporation = 3.48 (SD = 2.32), MEmployee = 5.00 (SD = 2.04); F(1, 96) = 11.80, p = .001, η2p = .11.

  128. [128]. See Preacher & Hayes, supra note 109, at 887 (discussing the differences among different mediation models).

  129. [129]. See Jeffrey R. Edwards & Lisa Schurer Lambert, Methods For Integrating Moderation and Mediation: A General Analytical Framework Using Moderated Path Analysis, 12 Psychol. Methods 1, 7 (2007).

  130. [130]. See id. at 7, 18.

  131. [131]. B = 1.08, SE = 0.40, t = ­­–2.00, p = .048.

  132. [132]. B = 1.34, SE = 0.50, t = 2.68, p = .008.

  133. [133]. B = 0.58, SE = 0.44, t = 1.31, p = .191.

  134. [134]. B = –0.72, SE = 0.35, t = –2.00, p = .048.

  135. [135]. B = 1.50, SE = 0.40, t = 2.75, p = .001.

  136. [136]. B = 0.42, SE = 0.30, t = 1.40, p = .125.

  137. [137]. B = –0.61, SE = 0.13, t = –3.27, p = .001.

  138. [138]. B = 0.64, SE = 0.67, t = 1.16, p = .249.

  139. [139]. Asterisks in the mediated moderation model indicate statistically significant associations; p < 0.05.

  140. [140]. See supra Part II.B.

  141. [141]. See Bayern, supra note 23, at 49–50 (discussing the doctrine); see also supra Part II.A.

  142. [142]. Because participants were not required to answer all survey questions, there is some “data dropout” evident in some of their responses.

  143. [143]. If the respondent had never owned a business, she was asked to imagine that she owned a business.

  144. [144]. This result was highly reliable and yielded an associated p-value of less than 0.05.

  145. [145]. This result was also highly reliable and yielded an associated p-value of less than 0.05.

  146. [146]. See, e.g., Cass R. Sunstein, Daniel Kahneman & David Schkade, Assessing Punitive Damages (with Notes on Cognition and Valuation in Law), 107 Yale L.J. 2071, 2100 (1998) (finding that women are more likely to “seek more severe punishment of civil defendants” than men).

  147. [147]. A one-sample t-test is used to compare the mean of a sample to either: (1) the mean of the population from which the sample is taken; or (2) a specified value that is theoretically meaningful to the design of the study. See Lawless et al., supra note 95, at 265–69.

  148. [148]. t(61) = –14.27, p < .001.

  149. [149]. t(61) = –6.57, p < .001.

  150. [150]. A histogram is a graphical representation of the distribution of a quantitative variable and can be used to estimate its probability distribution. See Lawless et al., supra note 95, at 194–96.

  151. [151]. F(1, 199) = 4.01, p = .047, η2p = .02.

  152. [152]. MBusiness = 3.53 (SD = 1.91), MLaypeople = 2.97 (SD = 1.80).

  153. [153]. Id.

  154. [154]. See supra Part II.B.

  155. [155]. In total, 98% of businesspeople would have hired John Brown as an employee, whereas 2% would have hired him as an independent contractor. 91% of laypeople would have hired him as an employee, whereas 9% would have hired him as an independent contractor.

  156. [156]. B = 1.49, SE = 1.07, Wald = 1.94, Odds = 4.42, p = .163.

  157. [157]. F(1, 299) = 0.00, p = 1.00, η2p = 0.00.

  158. [158]. A confidence interval provides the range—with a certain level of confidence—within which the true mean of the population exists. See Lawless et al., supra note 95, at 148.

  159. [159]. The descriptive statistics are as follows: MExperience = 4.97 (SD = 1.22); MIntelligence = 4.51 (SD= 1.36); MPersonality = 3.62 (SD = 1.59); MEducation = 2.91 (SD= 1.46); MInterview = 2.52 (SD = 1.65); MVicarious = 1.54 (SD= 1.71); MAttrition = 0.95 (SD= 1.18).

  160. [160]. See generally Eric Kades, Windfalls, 108 Yale L.J. 1489 (1999) (discussing a varying array of public and private windfalls, their benefits and drawbacks, and their implications for the law); see also generally Sykes, The Boundaries of Vicarious Liability, supra note 22 (providing an economic justification for the vicarious liability doctrine and discussing its limitations); Sykes, The Economics of Vicarious Liability, supra note 22 (describing the vicarious liability doctrine in terms of incentive structures and costs to legal actors).

  161. [161]. Windfall, The Merriam-Webster Dictionary (2016).

  162. [162]. Economics A-Z Terms Beginning with W, Economist, (last visited Nov. 6, 2016); Windfall Gains, Economic Times, http://economictimes.indiatimes.
    com/definition/windfall-gains (last visited Nov. 6, 2016).

  163. [163]. Kades, supra note 160, at 1521, 1546, 1556; see also generally Schwartz, supra note 22 (arguing that an employer’s vicarious liability for its employee’s actions frequently is obscured by courts, law schools, and politicians).

  164. [164]. See Halbersberg & Guttel, supra note 16, at 425–37 (synthesizing the views of these scholars and expressing skepticism of their claims).

  165. [165]. See id. at 425–28; see also generally Sykes, The Economics of Vicarious Liability, supra note 22 (outlining the law and economic terms on which the debate over vicarious liability has been framed).

  166. [166]. See, e.g., Paharia et al., supra note 15 (indirect harm is perceived as less blameworthy); Gong & Medin, supra note 15 (indirect harms are considered more abstract and distant); see also generally Borgida & Nisbett, supra note 82; Hansen & Wänke, supra note 82; Lammers, supra note 85; Semin & Fiedler, supra note 83.

  167. [167]. Paharia et al., supra note 15, at 140–41.

  168. [168]. See Neyers, supra note 22, at 292.

  169. [169]. See generally Glynn, supra note 22 (arguing that the underlying policies that govern the vicarious liability doctrine and the law of corporations justifies applying the vicarious liability doctrine to a corporation’s officers for the corporation’s legal obligations); King, supra note 22 (examining vicarious liability in the franchise context and recommending that the
    franchisor–franchisee relationship should be treated as a special instantiation of vicarious liability).

  170. [170]. See, e.g., Glynn, supra note 22, at 432 (arguing that the vicarious liability doctrine should apply to corporate officers and addressing proposals to make it more difficult to impose vicarious liability).

  171. [171]. See Hans, supra note 21, at 216–21; supra note 43 and accompanying text.

  172. [172]. For example, Federal Rule of Civil Procedure 19 reads, in relevant part, “[a] person who is subject to service of process and whose joinder will not deprive the court of subject-matter jurisdiction must be joined as a party if: (A) in that person’s absence, the court cannot accord complete relief among existing parties.” Fed. R. Civ. P. 19(a)(1)(A).

  173. [173]. See id.

  174. [174]. For example, the rule requires, in relevant part, that the joined party be an entity “who is subject to service of process and whose joinder will not deprive the court of subject-matter jurisdiction.” Id. 19(a)(1). The rule therefore may be under-inclusive to the extent that a corporate defendant cannot be joined to the action because the court lacks personal jurisdiction over the defendant or because the defendant’s joinder causes concerns related to subject-matter jurisdiction.

  175. [175]. Special verdict forms are addressed in Federal Rule of Civil Procedure 49, in which “[t]he court may require a jury to return only a special verdict in the form of a special written finding on each issue of fact,” through methods largely given to the discretion of the trial judge. Fed. R. Civ. P. 49(a)(1). Commentators have argued that special verdict forms—which require more detailed factual findings than a general verdict form—may be one way in which juror bias can be combatted. See, e.g., Stuart F. Schaffer, Comment, Informing the Jury of the Legal Effect of Special Verdict Answers in Comparative Negligence Actions, 1981 Duke L.J. 824, 852 (1981).

  176. [176]. See Schaffer, supra note 175, at 829; see also generally Paharia et al., supra note 15.

  177. [177]. See supra notes 165–67 and accompanying text.

  178. [178]. These data are on file with the author and will be published in a future article.

  179. [179]. These data also are also on file with the author. Controlled statistical analysis is ongoing.

  180. [180]. See Lynne ForsterLee & Irwin A. Horowitz, The Effects of Jury-Aid Innovations on Juror Performance in Complex Civil Trials, 86 Judicature 184, 184–85 (2003); Robbennolt, supra note 102, at 483; see also Sevier, supra note 120, at 324–25.

  181. [181]. Sevier, supra note 120, at 324­–25 & n.249.

  182. [182]. See Mark Kelman, Yuval Rottenstreich & Amos Tversky, Context-Dependence in Legal Decision Making, 25 J. Legal Stud. 287, 301–03 (1996).

  183. [183]. See id.

  184. [184]. See Lawless et al., supra note 95, at 93–96.

  185. [185]. Id. at 93–122 (discussing the strengths, weaknesses, and tradeoffs among controlled laboratory experiments, field experiments, quasi-experiments, and natural experiments). Moreover, true threats to external validity involve an interaction between the functional relationship being studied—that is, the effect of the independent variable on the dependent variable, and the setting—a laboratory simulation versus a field test. That the overall level of an effect is higher or lower in one setting compared to another is rarely a matter of concern among scientists; indeed, by definition it is not a concern when the question is whether a hypothesized functional relationship exists or not. External validity concerns arise when an independent variable increases a dependent variable in one setting but decreases it in the other setting. Were an effect found in one setting merely to disappear in the other setting, that might or might not be a concern from a policy perspective, but that would depend on the details of the policy question. Justin Sevier, Popularizing Hearsay, 104 Geo. L.J. 643, 691 (2016).

  186. [186]. Convergent validity is the ability to demonstrate an empirical phenomenon across a variety of populations and experimental designs. See, e.g., Donald T. Campbell & Donald W. Fiske, Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix, 56 Psychol. Bull. 81, 100 (1959).

  187. [187]. Live reenactments add external validity to laboratory findings, but studies strongly suggest that laboratory experiments are highly replicable outside the lab. Moreover, live reenactments are difficult to control experimentally. See Lawless et al., supra note 95, at

  188. [188]. Andrew Pollack & Emmarie Huetteman, Martin Shkreli Invokes the Fifth Amendment During Grilling by Congress, N.Y. Times (Feb. 4, 2016),

  189. [189]. Id.


Charles Ehrhardt Assistant Professor, Florida State University College of Law.

I thank Shawn Bayern, Avlana Eisenberg, Mark Seidenfeld, Mark Spottswood, Kip Viscusi, Joni Hersch, Arden Rowell, Tom R. Tyler, the Florida State University College of Law Faculty and the Yale University Department of Psychology for their guidance and for helpful oral and written comments regarding the contents of this Article. I also thank Vincent Fernandez and Amanda Qadri for their excellent research assistance.