The Psychology of Decision Making: Considerations for a Jury Trial vs. Bench Trial or Arbitration
Litigating disputes to a jury is often a stressful experience for everyone involved. Part of this stress comes from the underlying difficulty in predicting how a group of people, often chosen on the basis of having as little knowledge or experience relevant to the dispute as possible, will make decisions about the case at issue. Attorneys and their clients often worry about whether jurors will be able to make rational decisions based on evidence, especially in complex or emotionally charged cases. These concerns affect both plaintiffs and defendants and sometimes lead to a decision to forgo a jury trial in court in favor of a bench trial or to attempt a resolution through the use of alternative dispute resolution practices.
Are judges likely to be better decision makers than jurors? This is clearly a complex question, and a complete answer would need to consider factors such as group versus individual decision making, cognitive abilities and the ability to understand and apply legal constructs (among many other things). This paper will address the issue of systematic errors that people tend to make when evaluating decision options and interpreting information and examine whether judges tend to make the same sorts of errors as laypersons.
Limitations on Jurors / Cognitive Biases
Psychologists have learned that human beings rely on mental shortcuts, known in the jargon as “heuristics,” to make complex decisions. Reliance on these heuristics facilitates good judgment much of the time, and helps people deal with what would often be overwhelming amounts of information, but it can also produce systematic errors in judgment. Just as certain patterns of visual stimuli can fool people’s eyesight, leading them to see things that are not really there, certain fact patterns can fool people’s judgment, leading them to believe things that are not really true. Reliance on heuristics can sometimes create cognitive illusions that produce erroneous judgments.
Research on juries indicates that cognitive illusions can adversely affect the quality of the decisions jurors make. Researchers have found, for example, that juries believe that litigants should have predicted events that no one could have predicted, allow irrelevant or inadmissible information to influence liability determinations, defer to arbitrary numerical estimates, and rely on incoherent methods to calculate damages. Several of the more important heuristics that have been studied in the legal arena are as follows:
People overstate their own ability to have predicted the past and believe that others should have been able to predict events better than was possible. Psychologists call this tendency for people to overestimate the predictability of past events the “hindsight bias.” It occurs because learning an outcome causes people to update their beliefs about the world. People then rely on these new beliefs to generate estimates of what was predictable, but they ignore the change in their beliefs that learning the outcome inspired. Few judgments in ordinary life require people to assess the predictability of past outcomes, but such judgments are pervasive in the law. This bias frequently operates against defendants because jurors tend to overestimate the likelihood that bad outcomes could have been foreseen (the defendant being blamed for the bad outcome). However, it can also work against plaintiffs in situations where jurors believe that contributory negligence (alleged or not) may have been a factor in a plaintiff’s injury.
When people make numerical estimates (e.g., the fair market value of a house), they commonly rely on the initial value available to them (e.g., the list price). That initial value tends to “anchor” their final estimates. In many situations, reliance on an anchor is reasonable because many anchors convey relevant information about the actual value of an item. The problem, however, is that anchors that do not provide any information about the actual value of an item also influence judgment. Anchors affect judgment by changing the standard of reference that people use when making numeric judgments. Even when people conclude that an anchor provides no useful information, mentally testing the validity of the anchor causes people to adjust their estimates upward or downward toward that anchor. As a consequence, even extreme, wholly absurd anchors can affect judgment. For example, if you ask people whether the average annual temperature of a city like San Francisco is higher or lower than 550 degrees, virtually everyone who understands the concept of temperature will answer that the average temperature is lower. However, when you then ask them to estimate the true average temperature, their estimates will be significantly higher than the estimates given by people who have not first been asked the anchoring question. Another example from the research is to ask an audience to think of the last 4 digits of their social security number, and then to estimate the number of physicians in New York. The correlation (which should be about 0 on a scale from -1 to 1) between an individual’s social security number and their estimate in this situation is around 0.4—far beyond what would be expected by chance. The simple act of thinking of the first number strongly influences the second, even though there is no logical connection between them. Anchoring is relevant in many litigation scenarios, particularly when it comes to parties involved in settlement negotiations and jurors making estimations of damages.
People tend to make judgments about themselves and their abilities that are “egocentric” or self-serving. For example, people routinely estimate that they are above average on a variety of desirable characteristics, including health, driving, professional skills, and likelihood of having a successful marriage. Moreover, people overestimate their contribution to joint activities. For example, after a conversation both parties will estimate that they spoke more than half the time. Similarly, when married couples are asked to estimate the percentage of household tasks they perform, their estimates typically add up to more than 100%.
Egocentric biases occur for several reasons. For one thing, people remember their own actions better than others’ actions. Thus, when asked to recall the percentage of housework they perform, people remember their own contribution more easily and, consequently tend to overestimate it. Egocentric biases can be adaptive, but they can also have misleading influences on the litigation process. Due to egocentric biases, litigants and their lawyers might overestimate their own abilities, the quality of their advocacy and the relative merits of their cases. Research demonstrating this asked participants to evaluate a dispute scenario from the perspective of either a plaintiff or defendant and to judge the potential settlement value of the case. Participants evaluating the case from the perspective of the plaintiff predicted that a judge would award significantly more than the defendant-participants predicted. Similarly, when asked to identify what they perceived to be a fair settlement value, plaintiff-participants selected a higher value than the value selected by defendant-participants. These results suggest that self-serving or egocentric biases could be a factor in bargaining impasses.
When people confront risky decisions such as deciding whether to settle a case or to proceed to trial, they tend to categorize their decision options as potential gains or losses. This categorization, or “framing,” of decision options influences the way people evaluate options and affects their willingness to incur risk. People tend to make risk-averse decisions when choosing between options that appear to represent gains and riskseeking decisions when choosing between options that appear to represent losses. For example, most people prefer a certain $100 gain to a 50% chance of winning $200 but prefer a 50% chance of losing $200 to a certain $100 loss. Litigation produces a natural frame. In most lawsuits, plaintiffs choose either to accept a certain settlement from the defendant or to gamble, hoping that further litigation will produce a larger gain. Most defendants, by contrast, choose either to pay a certain settlement to the plaintiff or to gamble that further litigation will reduce the amount that they must pay. Thus, plaintiffs often choose between options that appear to represent gains, while defendants often choose between options that appear to represent losses. As such, plaintiffs are more likely to prefer settlement, the risk-averse option, while defendants are more likely to prefer trial, the riskseeking option.
When people make categorical judgments (e.g., assessing the likelihood that a criminal defendant is guilty), they tend to base their judgments on the extent to which the evidence being analyzed (e.g., the defendant’s demeanor) is representative of the category. When the evidence appears representative of the category (e.g., defendant is nervous and shifty), people judge the likelihood that the evidence is a product of that category as high (i.e., evidence of guilt). When the evidence being analyzed does not resemble the category (e.g., defendant appears at ease), people judge the likelihood that the evidence is a product of that category as low (i.e., evidence of innocence). Psychologists refer to this phenomenon as the “representativeness heuristic.” Although the representativeness heuristic can be useful, it can lead people to discount or ignore more relevant statistical information. In particular, people undervalue the importance of the frequency with which the underlying category occurs in an underlying population, essentially ignoring the “base-rate” of the phenomenon under consideration.
Impediments inherent in the system
Our legal system is based on assumptions about human decision making, some of which have been demonstrated to be invalid. For example, the system assumes that jurors can disregard inadmissible information when instructed to do so, even though this is psychologically impossible. Indeed, the fact that such information gets pointed out for special attention has been shown to actually increase its impact on juror decision making in some circumstances. A partial listing of aspects of the system, varying wildly from place to place, that make jurors’ jobs difficult (some of which are being addressed in some jurisdictions) include:
- Not being told the law until near the end of the process.
- Not being able to read a written copy of the law.
- Not being given instructions in words that have meaning for laypeople.
- Not being able to ask questions.
- Not being able to take notes.
- Verdict questions that are sometimes constructed with no thought to the difficulty they might pose to laypeople.
- Being told that some of the things they have heard and seen are evidence and others are not.
Distinguishing between questions and statements of attorneys as nonevidence versus answers, admitted exhibits and other stipulations which are evidence.
Simply put, being a juror is a difficult job, on top of which the rules of the task (while created for sound legal reasons) frequently add to its difficulty. Indeed, our interviews with actual jurors post-trial show that they have often been confused about some of the most basic principles, e.g., that the answers given from the witness stand are evidence. We have asked jurors what they made of testimony and whether it was important to their decision only to hear, “They were just saying that.” For such jurors, if there is not written material backing up the testimony, it does not count as evidence, even though those verbal answers are at the heart of what evidence is supposed to be.
The Superiority of Judicial Decision Making
Given cognitive biases and other limitations on juror decision making as discussed above, is it any wonder that parties involved in disputes often wonder whether they would be better off in front of a judge and foregoing a jury trial? One would expect that judges should have an easier time making decisions based on evidence and legal principles and ignoring (or at least minimizing the influence of) tangential or emotional considerations. Judges certainly have a greater amount of knowledge and informational resources available to inform their decision making and are highly motivated in that they are not interested in seeing their decisions overturned due to error.
But are judges more reliable decision makers than jurors? Are they less susceptible to making decisions based on erroneous cognitive processes? Whether or not judges tend to make similar cognitive errors as jurors has not received a great deal of systematic study. Despite the important role that judges play in the courtroom, psychologists have focused most of their research efforts on juries. One study that provides an excellent overview on this topic, as well as experimental data collected from a large population of federal magistrate judges was conducted by Chris Guthrie, J.D., Jeffrey Rachlinski, J.D., Ph.D. and the Honorable Andrew J. Wistrich, J.D., and published in the May 2001 issue of the Cornell Law Review. Their study examined the responses of magistrate judges to scenarios that were manipulated in order to assess the effects, if any, of the cognitive biases discussed above. The sample of 167 judges comprised nearly one-third of all such federal magistrate judges on the bench at the time the data were collected. The results showed that judges did better than lay jurors in some areas but that judges appeared to be just as susceptible to certain cognitive errors as were jurors. A brief summary of the key results is as follows:
Judges in this research were asked to evaluate a hypothetical fact pattern and make decisions about the likelihood of an outcome of an appeal. They were asked to predict which of three potential actions the court of appeals was most likely to have taken (affirm, vacate or remand) after having been informed, on a random basis, that one of the three actions had been taken. After being given all three possible outcomes and asked to predict which of these decisions the court of appeals was most likely to have made, judges were more likely to predict the outcome that was consistent with the outcome that they were told had occurred; that is, judges who were given a particular outcome were more likely to predict that outcome than judges given the other possible outcomes. This demonstrates hindsight bias because the outcome provided to them should have been irrelevant to their own predictions about what the appeals court was most likely to do.
These results may have important implications for courtroom decision making. For example, in a situation where one side succeeds in keeping certain evidence out, the jury may be shielded from that evidence if it was excluded as part of a pre-trial hearing; hence it cannot contribute to a jury finding being based on hindsight. In a bench trial, in contrast, the judge as fact-finder cannot avoid being exposed to such evidence.
To study anchoring effects, judges were asked to assess a damages amount in reaction to a personal injury case where liability was conceded. They were randomly assigned to a “no anchor” condition and an “anchor condition” in which the only additional information was a statement that the defendant had moved for dismissal based on the argument that the claim did not meet a jurisdictional minimum of $75,000 (a clearly merit-less motion given that the damages described were obviously greater than $75,000, including months of hospital bills, loss of the use of plaintiff’s legs, etc.). Although the $75,000 figure in the anchor condition communicated no useful information, it nevertheless had a significant effect on the damages awarded. Judges in the “no anchor” condition made average awards almost 30% higher than judges in the anchor condition, indicating that the mere presence of the $75,000 figure served as a lower anchor and depressed awards in that group.
These data show how easily anchors can affect the way judges think about damage awards. It appears that plaintiffs’ damage requests may have unintended effects on both on juries and judges. The potential for statutory damage caps to anchor judges is another area of potential concern. Although judges might be able to keep juries in the dark about caps on damages, they clearly cannot blind themselves to such caps.
In testing this bias, judges in the study were asked to estimate their own reversal rates on appeal, putting themselves into one of four categories as compared to their peers, equivalent to saying whether they were reversed more than 75% of their peers, more than 50% of their peers, more than 25% of their peers or less than 25% of their peers. Results showed that judges are susceptible to egocentric biases. Over 50% of the judges (56.1%) reported that their appeal rate placed them in the lowest quartile; 31.6% placed themselves in the second-lowest quartile; 7.7% in the second-highest quartile, and 4.5% in the highest quartile. In other words, 87.7% of the judges believed that at least half of their peers had higher reversal rates on appeal. This pattern of results differs significantly from what one would expect if judges were unbiased.
It is possible that egocentric biases might prevent judges from maintaining an awareness of their limitations, and this may make it difficult for litigants to convince judges that they might have been wrong, thereby increasing the hurdles faced by parties in the situation of having to ask a judge to vacate a prior ruling or to set aside a jury’s verdict on the basis of a judicial mistake.
Judges in the study were asked to place themselves in the role of overseeing settlement negotiations. They were given identical economic options as choices, however, for half the decision had been constructed around a “Plaintiff/Gains” perspective whereas the other half received a “Defendant/Losses” perspective. Although the hypothetical litigants in this problem faced settlement offers that were identical in terms of their expected value, the materials created the illusion that the plaintiff faced a choice between potential gains and that the defendant faced a choice between potential losses. The results showed that review from the defense (or riskseeking) perspective was more likely to lead to an evaluation that the case should not be settled (75%) than review from the plaintiff (or riskaverse) view (60%). From the plaintiff’s perspective, settlement seemed relatively more attractive, while from the defendant’s perspective, trial seemed relatively more attractive, even though the two perspectives presented identical economic choices. Although this difference was statistically reliable, it should be noted that the 15% difference shown by this sample of judges was much less than that typically seen in studies of non-expert decisionmakers (often as high as a 40-50% difference). Although judges were affected by framing effects, it appears that they may be less susceptible to this bias than many other people.
Judges were provided with a scenario asking them to estimate the likelihood that an event occurred due to negligence, given a variety of statistical information about the occurrence rates of various steps in a series of events that led to a plaintiff’s injury. Research with most populations evaluating the same scenario shows that the tendency to ignore base-rate information leads a substantial majority of people to overestimate the extent to which negligence was a factor. Here, 40% of the judges (much higher than other populations) made the correct decision taking into account relevant statistical information. Of the 60% who answered incorrectly, however, a large proportion (40% of the total) did appear to have been affected by the representativeness heuristic. Hence it appears that judges are less affected by this cognitive bias than most other people (although they are still somewhat vulnerable to it). Still, the representativeness heuristic might account for some judges’ apparent preference, observed by some researchers, for individuating evidence (e.g., eyewitness testimony) over statistical evidence (e.g., base rates).
So, Jury versus Bench?
This question is too involved to answer thoroughly within the scope of this brief review. Even so, the research on cognitive biases in information processing does provide some information that is applicable to the consideration of whether to forgo trial by jury. If one of the major arguments being put forth in favor of a bench trial is that judges are less affected by extralegal thought processes than jurors, the answer appears to be that this argument does not carry a great deal of weight. Judges, being human after all, are influenced in their decision making by many of the cognitive shortcuts and biases as human jurors.
A point that also needs to be considered is that of individual versus group decision making. Unfortunately, the prior research does not provide a straightforward conclusion about whether groups make better decisions about disputes than individuals. There is evidence that juror deliberations can result in the lessening of the effect of cognitive biases in some instances, improved memory for trial evidence, increased complex reasoning about the evidence and arguments presented, and reduced variability in decisions. However, there is also evidence that group deliberation may exacerbate biases under certain conditions and can result in more extreme judgments in a process known as group polarization. It should be pointed out that there is not even agreement among academic researchers about how to determine whether a jury’s or judge’s decision is the “right” one in the first place, and a number of different ways have been used to attempt to define the measure. In addition, given that every case is at least somewhat different from every other (sometimes wildly so), there is never likely to be a “one-size-fits-all” answer. Fortunately for those of us in the world of applied jury psychology (and for litigants themselves), there is an easier way to deal with this; the “right” decision is the one that provides the best possible result for our client. This is a testable question. Research methodologies exist which allow us to test whether a judge or a jury trial is more likely to lead to a better outcome for the client in any particular situation. Customized studies can help to determine whether judges may be better able than juries to focus on a particular narrow legal issue, for instance. In cases where bench trials or ADR approaches are required, e.g. Markman hearings, some contractual disputes, etc., applied research techniques can also be used to test approaches and refine arguments to make sure that the most persuasive case possible is being presented to the final fact-finder. Hence, parties facing a judge versus jury decision can make it supported by evidence as opposed to hunches; parties obligated to litigate in a non-jury format can take advantage of persuasion techniques that are applicable to both juries and judges.