Not Guilty: Are the Acquitted Innocent?
Authors: Daniel Givelber and Amy Farrell
Publisher: New York: New York University Press, 2012. 209p.
Reviewer: Darryl K. Brown | January 2013
In their terrific if misleadingly titled new book, Not Guilty: Are the Acquitted Innocent?, Daniel Givelber and Amy Farrell ask a practically unanswerable question: how many defendants who are acquitted at trial are actually innocent? Wrongful convictions are surely the greater injustice, though the unknown accuracy of acquittals can impose burdens as well—not only by allowing the guilty to go unpunished but by leaving the innocent-and-acquitted with no public affirmation of their innocence, and often with a widespread assumption of the opposite. Because the standard of proof is designed to produce some acquittals for the guilty, the innocent-acquitted are lumped together with the guilty for whom the state lacked sufficient proof.
As Givelber and Farrell note, perhaps the one unfortunate side effect of the attention to wrongful convictions in recent years is to encourage an improperly narrow conception of innocence as equivalent to cases of mistaken identity. But defendants can be innocent even when they engage in relevant conduct and cause harm, because guilt also depends on fault—on one’s state of mind (mens rea), reasons for acting (e.g., self-defense), and the circumstances in which one acted (e.g., in reaction to an immediate threat). Defendants who are innocent for these sorts of reasons can’t hope for proof in the nature of definitive DNA analysis or a rock-solid alibi.
Givelber and Farrell concede they cannot determine, or even estimate, the portion of the acquitted who are actually innocent. Most of their focus is instead on a primary way by which we assess whether a jury verdict is correct: asking judges whether they would have decided the case the same way. Not Guilty doesn’t get very far in answering the question posed in its subtitle, but it is immensely valuable for the close look it provides at what we know, and can know, about the correctness of jury verdicts. Not Guilty, it turns out, is primarily a study of jury decision-making and of how jury verdicts accord with judges’ decisions, based on a large, recent data set of criminal trials. Its findings raise doubts about the core normative defense of juries that has prevailed for at least half a century, yet also provide alternative reasons—equally strong or stronger—to trust criminal juries, and to wish we used them more often.
This agenda, of course, has been pursued before, in Kalven and Zeisel’s 1966 landmark study, The American Jury. It is hard to overestimate the influence of The American Jury on judicial and scholarly perceptions of juries, yet Not Guilty provides strong reasons to temper reliance on its central findings. Kalven and Zeisel gathered information on 3,576 criminal jury trials from judges in the mid-1950s; their data consisted of judges’ descriptions of jury trials along with judges’ assertions of how they would have decided those same cases themselves (Not Guilty, at p.27). One of the study’s big findings was that judges and juries agreed most of the time (78% of cases) on whether a defendant had been proven guilty or not. In the minority of cases in which they disagreed, judges more often would have convicted but described the trial evidence as “close,” making especially hard cases. These findings had the effect of undermining much skepticism about lay decision-making: except in the hardest cases, where disagreement is most plausible, jury and judge decisions mostly come out the same way.
That agreement is reassuring for those worried about jury competency, but it doesn’t provide a reason to prefer juries over judges. That rationale came from Kalven and Zeisel’s other influential conclusion: the "liberation hypothesis" (pp.28, 71). In the small percentage of cases that juries decided differently from judges, they mostly did so, Kalven and Zeisel inferred, because the trial evidence did not clearly dictate a verdict either way. This "liberated" jurors from the dictates of evidence, and gave them an occasion to allow various forms of "sentiment" to reach their verdicts. This finding, too, had intuitive appeal; it makes sense that conscientious jurors would resist the appeal of sentiment when a given verdict is "easy" on the evidentiary record. It has also turned out to be a finding that bolstered the reasons for having juries at all instead of judges. When the evidence is close, and the case also presents some issue of fairness or extra-legal values, somebody’s sentiments are likely to come in to play. Common law countries prefer those of lay citizens to the government’s judges. Understood this way, the liberation thesis has been taken to confirm a key virtue of juries, not least by the U.S. Supreme Court, which has cited the study for this very point.
If only because of its influence, Kalven and Zeisel’s now-almost-50-year-old-study merits updating, and Not Guilty is in large part a replication of that perhaps-too-influential work, using data on just over 300 jury trials gathered in 2000-01 by the National Center for State Courts (pp.38-39, 74). The sample is smaller but the data in some ways better. For example, The American Jury lacked any survey data from jurors, but the NCSC gathered jurors’ and judges’ responses on whether they viewed evidence in a case as “close” and on several specific "sentiments" likely to affect judgments, such as whether the applicable law seemed fair, whether the legally dictated outcome seemed fair, whether the consequences of the conviction seemed fair, or whether they trusted the police (p.94). An updated study is valuable also because a lot has changed since the 1950s. In the era of The American Jury, women could be constitutionally excluded from jury service, and African Americans effectively were. The fair cross section doctrine was not settled until the late 1960s, and the competing idea of juries relied on the "key man" system that sought not a representative panel but one composed of the "best" men. The judiciary, to whom juries were being compared, was also much more white and male. The context for 1950s trials was different in ways subtle and not so subtle. Fewer exclusionary rules applied, including rape-shield laws. Crime rates were lower and crime-related politics less salient. (Governors routinely pardoned convicted felons.) Sentencing policy was less determinate and kept American incarceration rates in line with European ones, in contrast to current sentencing laws that for the last two decades or so have given the U.S. the world’s highest incarceration rates four or five times higher than those of other advanced democracies. The past, indeed, seems a foreign country. "Sentiments" that came into play in criminal trials of the 1950s might be quite different from today.
Givelber and Farrell generate number of interesting findings from their analysis of the NCSC data which they integrate with other recent sources, notably Bureau of Justice Statistics studies of plea and trial rates and reasons for case dismissals (p.44). I will note only a few. One is that defendants win acquittals more often when they present more of their own evidence. Defendants who testified and presented an additional witness did notably better than those who offered only their own testimony, or only that of another witness, or neither (pp.85, 102-16). Another is that defendants whose reasons for refusing to plead guilty is their insistence that they are truly innocent win acquittals more often—although they won only 44% of the trials in this sample. (pp. 85-87.) Why this might be, Givelber and Farrell note, we can’t say: maybe they really are innocent and so the (misleading) evidence against them is weaker; maybe they motivate their lawyers to work harder; maybe these defendants have a better sense of the state’s weak proof of their guilt. Nonetheless, note the practical use of these findings for defense attorneys: a client who insists he is innocent, can testify, and can produce another witness or two to support him is a client with much better-than-average odds at trial.
Givelber and Farrell’s most important findings, however, relate to the issues of judge-jury agreement that were central to the Kalven and Zeisel study. Some of their findings are consistent: judges would convict more often than juries do; they would convict in most cases in which juries acquitted (pp. 27, 96, 138). (This stands in contrast to data on actual bench trials. Acquittal rates in real bench trials are at least as high as those for real jury trials; p.117). Yet because they have juries’ assessments of "close evidence" cases as well judges, Givelber and Farrell are able to discover what Kalven and Zeisel could not: that judges and juries do not consider the same cases to be "close" on the evidence. Kalven and Zeisel found that the cases that judges deemed close are those in which there are more jury acquittals (and juror "liberation" from the dictates of evidence). But in Givelber and Farrell’s data, judges’ assessments did not predict jury acquittal decisions. Jury judgments about which cases were close, however, did (p.90), although these close cases were not the ones in which jurors seemed to give sentiment or values much role, at least according to their self-reporting about sentiment-related issues (trust in the police, fairness of the law, etc.) in those cases. In fact, sentiment was least likely to play a role in close cases (pp.96-97). The NCSC data provided no support, in other words, for the central premise of the liberation thesis "that factual uncertainty opens a juror’s mind to considering sentiment in determining guilt or innocence"; instead, jurors seem properly to take factual uncertainty is reason to conclude the state hasn’t met its burden of proof (pp.97-98, 137). Givelber and Farrell did, however, find juror sentiment to play a greater role in those cases that judges thought were clear on the evidence. As they explain, this is not as surprising as it may initially seem: something needs to explain why jurors reach a different judgment in those cases that judges think are the easiest. The most plausible candidates are either sentiment or juries’ very different understanding of what the evidence shows.
Givelber and Farrell interpret this data, persuasively to my mind, to reach conclusions that differ from The American Jury, change the empirical bases for normative arguments about juries, and provide tantalizing hints about the relationship of innocence to acquittals. Juries don’t just give sentiment a different or greater role than judges do; they interpret the evidence differently. That seems a big part of why juries depart from judges in some cases that judges see as clear ones for conviction, and why the two groups identify somewhat different cases as close. Evidence requires interpretation—assessments of credibility, reliability, what inferences are plausible, etc. There is nothing in judges’ legal expertise that gives them greater competency in most of that work. It seems that judges as a group interpret evidence somewhat differently than jurors, but not necessarily more accurately; their jobs confront them with large numbers of defendants, most of whom indisputably are guilty; perhaps that (along with fear of gaining reputations for excessive leniency) shades their interpretation of evidence (p.118). Then there is this: the NCSC data shows little variation in rates of jury acquittals for black defendants and all others (29% v. 32%). Judges, however, reported that in those cases they would have acquitted a much lower portion of black defendants (14%) compared to their acquittal rate (26%) for all other defendants (p.125). If judges were getting things right and juries wrong, it means “prosecutors were necessarily bringing stronger criminal cases against Black defendants than they were against defendants of other races” (p.139)—a counter-intuitive policy of relative prosecutorial leniency toward black suspects that goes against much other evidence of racially disparate treatment in the criminal justice system.
All this leaves us little closer to determining how many of the acquitted are actually innocent, but it provides good arguments for common law countries to prefer lay over official fact-finding; juries might be better fact-finders, rather than simply being better at tempering verdicts with sentiment. The stronger conclusion, however, as the authors emphasize (pp.138-40), is that we have good reasons to be skeptical of presuming that judges’ judgments provide the baseline for accuracy from which we infer that jury disagreement indicates jury error, or even greater reliance on values. That suggests that we could use more studies of judicial decision-making, including exploration of how judges disagree with each other in the same cases.
Darryl K. Brown is O.M. Vicars Professor of Law at the University of Virginia School of Law.