INNOCENCE EFFECT FIRST DRAFT – DO NOT CITE WITHOUT PERMISSION 2/12/2016 THE INNOCENCE EFFECT IN PLEA BARGAINING Oren Gazal-Ayal* and Avishalom Tor** ABSTRACT About 95% of all felony convictions in the United States are the result of guilty pleas. To maintain such a high guilty plea rate, plea bargain offers have to be very attractive, compared to trials. Most legal scholars argue that these offers are attractive to guilty and innocent defendants alike. Innocent defendants who cannot signal their innocence to the prosecutor are unable to signal their innocence in a trial too. Hence, like guilty defendants, they will bargain in the shadow of the trial, and accept plea bargains that reflect the probability of conviction and expected jury-trial sentence. Plea bargains, so we are told, lead to wrongful convictions. Many scholars argue that plea bargains should be curtailed because they facilitate wrongful convictions. Others contend that plea offers can only benefit innocent defendants, by offering them an alternative to the risky trial, which may lead to a much harsher sentence. Yet even while drawing contradictory conclusion regarding this practice, both camps in the debate agree that plea bargains often lead innocents to plead guilty. They simply argue about the normative implication of this result. The decades-long plea bargaining debate is based on a combination of scholarly beliefs and theoretical models, with little systematic evidence to support either position. The present article draws on a diverse set of empirical findings to reveal the innocence effect, whereby innocents are significantly less likely to accept plea offers than their guilty counterparts, even when these offers appear objectively attractive in light of the evidence against them and the expected sanction at trial. After substantiating the innocence effect, we examine its implications for the plea bargaining debate, showing it requires both camps to reevaluate their policy prescriptions and offering some new proposals of our own to minimize false convictions, better to protect the innocent, and improve the plea bargaining process. Senior Lecturer, University of Haifa, Faculty of Law. Visiting Professor of Law, Fordham Law School (Spring 2011); Senior Lecturer and CoDirector of the Forum for Law and Markets, University of Haifa Faculty of Law. Earlier versions of this paper benefited from comments and criticisms by Hillel Somer ???. and seminar participants at the University ???. ?? and ?? provided excellent research assistance. * ** 1 INNOCENCE EFFECT 2 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR TABLE OF CONTENTS Introduction ........................................................................................ 2 I. Empirical Evidence ..................................................................... 5 A. Exonerations Database............................................................ 7 B. The Tulia Case ...................................................................... 10 C. Alternative Accounts ............................................................ 11 II. Experimental Studies ............................................................... 15 III. Interviewing Convicts................................................................. 19 IV. Psychological insights ................................................................ 20 A. Evaluating the probability of conviction ................................. 21 B. Fairness-related preferences..................................................... 24 V. Normative Implications ............................................................... 26 A. Miminizng false convictions.................................................... 27 B. “Insuring” the innocent ............................................................ 30 C. Restrictions on plea bargains ................................................... 32 D. Agreements on a Simplified Criminal Process ........................ 34 Conclusion ........................................................................................ 36 INTRODUCTION Plea bargaining dominates the criminal justice landscape in the United States. About 95% of all felony convictions are the result of guilty plea and most guilty pleas follow some sort of plea bargaining. 1 Most legal scholars argue that innocent defendants are also among those who plead guilty. Some argue that innocent defendants sometimes plead guilty because of psychological weakness which leads them to submit to the institutional pressures from the criminal justice system.2 Others suggest, at the same time, that rational innocents may also accept plea offers, with their discounted sanctions, to avoid the risk of a wrongful conviction at trial. Accordingly, innocent (and guilty) defendants will bargain in the shadow of the trial, and accept plea bargains that reflect the probability of conviction and expected jury trial sentence. Plea bargains, it is thus argued, is a major source of wrongful convictions. In an effort to address this innocence problem, many legal scholars argue that the plea bargaining should be curtailed.3 Others contend that plea offers can only benefit innocent defendants, by offering them a potential avenue to escape the risk of a much harsher sentence after a 1 2 3 See, e.g., Alschuler, Bibas INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 3 wrongful conviction at trial.4 Yet even while drawing contradictory conclusion regarding this practice, both camps in the debate agree that plea bargains often lead innocent defendants to plead guilty. They only differ on the normative implications of this result. Thus, numerous articles use economic models and methodology to examine, inter alia, how innocent defendants may rationally prefer to plead guilty, because the risk of conviction at trial, coupled with the harsher sentence expected after such a conviction, leads them to prefer the relative leniency the plea bargain.5 The latter analysis is based on the “shadow of trial” model, which is the dominant prescriptive theory in the plea-bargaining debate.6 In the simplest version of this approach, for instance, defendants calculate whether the anticipated post-trial sentence, multiplied by the probability of conviction, is lower than the plea offer sentence.7 Yet shadow-of-trial models commonly pay little attention to defendants’ culpability per se. To wit, innocents – who may well face a higher average probability of acquittal at trial – are expected to demand a more lenient sentence in return for their guilty plea.8 However, if the plea offer is accurately adjusted to defendants’ probability of conviction, innocence in itself plays no role in the defendant's decision.9 This assumption, which is jointly held by both parties to the plea bargaining debate, is false. Notwithstanding its dramatic importance for the law, the consensus on the likelihood of innocents’ guilty pleas is only based on scholarly beliefs and theoretical models, with little systematic evidence to support either position.10 In contrast, the present 4 5 For a thorough review of the economic literature on plea bargaining see Oren Gazal-Ayal & Limor Riza, Plea Bargaining and Prosecution, in CRIMINAL LAW AND ECONOMICS 145 (Nuno Garoupa ed., 2009). For some of the more prominent Articles see William M. Landes, An Economic Analysis of the Courts, 14 J.L. & ECON. 61, 69 (1971); Gene M. Grossman & Michael L. Katz, Plea Bargaining and Social Welfare, 73 AM. ECON. REV. 749 (1983). 6 Refer to the often use of critics of plea bargaining of this model (schulhofer, alschuler langbein and others) 7 In more sophisticated versions, additional factors like risk aversion and cost of trial are also taken into account. 8 See Grossman & Katz, supra note 5; Robert E. Scott & William J. Stuntz, Plea Bargaining as Contract, 101 YALE L.J. 1909, 1947 (1992). 9 See Stephen J. Schulhofer, Plea Bargaining as Disaster, 101 YALE L.J. 1979, 1984 (1992) ("Innocence by itself (that is, apart from its link to particular evidence) can have only a small impact on the odds of conviction") 10 We found only two studies from the late 1970's early 1980's attempting to examine the influence of innocence in plea bargaining. Larry W. Gregory, John C. Mowen & Darwyn E. Linder, Social psychology and Plea Bargaining: Applications, Methodology, and Theory, 36 JOURNAL OF PERSONALITY & SOCIAL PSYCHOLOGY 1521 (1978) and Kenneth S. Bordens, The Effects of Likelihood of INNOCENCE EFFECT 4 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR article draws on a diverse set of empirical findings to reveal the innocence effect. We show that innocent defendants are significantly less likely to accept plea offers than their guilty counterparts, even when these offers appear objectively attractive in light of the evidence against them and the expected sanction at trial. This innocence effect should turn the plea bargaining debate upside down. While we are not the first to criticize the shadow-of trial model in the criminal arena, our analysis and its policy implications substantially differ from the common critiques of this model. The main line of criticism of that model argues that defendants' decisions are not systematic enough to allow prediction of their bargaining strategies, because defendants are irrational, lack the abilities to make a calculative decision, or suffer from bad representation.11 Yet like other scholars in the field most advocates of this critique rely on intuition or anecdotes alone. Several recent papers, most prominently Stephanos Bibas, theoretically apply extra-legal psychological insights to analyze defendants' decisions, sometimes reaching erroneous results because they like direct evidence on plea behavior.12 Absent empirical basis, none of these studies exposed the innocence effect. Studies of procedural fairness similarly reveal the limitations of the dominant shadow of trial model.13 These studies show empirically that defendants' satisfaction with the criminal process and people's willingness to accept its legitimacy depend not only on the results of the process but also on its perceived fairness. Despite the commonality with our approach, however, our analysis differs from the procedural fairness literature in two important aspects. First, we focus on the result, rather Conviction, Threatened Punishment, and Assumed Role in Mock Plea Bargaining Decisions, 5 BASIC & APPLIED PSYCHOLOGY 59, 65 (1984). A search for these studies in Lexix found that each of them was only mentioned once. We will refer to these studies below. We will also refer to experimental tests we conducted together with Stephen M. Garcia: Avishalom Tor, Oren Gazal-Ayal & Stephen M. Garcia, Fairness and the Willingness to Accept Plea Bargain Offers, 7 JOURNAL OF EMPIRICAL LEGAL STUDIES (2010) [Add full correct reference] 11 Refer to Alschuler, the defense attorney's role – schulhofer, plea bargaining as disaster, more. 12 See Stephanos Bibas, Plea Bargaining Outside the Shadow of Trial, 117 HARV. L. REV. 2463 (2004). See also Rebecca Hollander-Blumoff, Social Psychology, Information Processing, and Plea Bargaining, 91 MARQ. L. REV. 163 (2007); Russell D. Covey Reconsidering the Relationship Between Cognitive Psychology and Plea Bargaining, 91 MARQ. L. REV. 213 (2007); Richard Birke, Reconciling Loss Aversion and Guilty Pleas, 1999 UTAH L. REV. 205 (1999). Say here that Bibas assumes that innocent defendants are more likely to take the plea, especially when they are not informed. We show that the opposite is correct. 13 For review of this literature see Tom R. Tyler, Social Justice: Outcome and Procedure, 35 INTERNATIONAL JOURNAL OF PSYCHOLOGY 117 (2000) INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 5 than process, of the criminal case. Second, we provide empirical foundations for predicting defendants’ behavior instead of their perception. We substantiate the innocence effect with both field and laboratory findings. Part I analyzes field data from several hundreds of wrongfully convicted defendants who were later exonerated, showing that only a fraction of these wrongful convictions resulted from guilty pleas, in sharp contrast with the very high percentage of guilty pleas in the general population of convictions. We further examine alternative accounts for this dramatic difference in plea rates and conclude it should be attributed, at least in part, to the innocence effect. This Part also reviews illustrative anecdotal evidence that further supports our exonerations-based findings. Part II then provides evidence for the innocence effect from the few available experimental studies, both earlier ones and our own, more recent, research. These studies are particularly valuable in corroborating the suggestive field evidence in controlled settings and helping indentify the causes of the innocence, thus laying a foundation for a later analysis. Part III further complements the preceding evidence of the innocence effect, drawing on retrospective studies of convicted defendants, showing that their guilt or innocence plays a central role in their plea decisions. In Part IV we explore the causes of the innocence effect, showing it to result from a combination of fairness-related preferences and a systematic, overoptimistic, bias on the part of innocents regarding their probability of acquittal. Finally, in Part V we critically examine the implications of the innocence effect and its underlying causes for the plea bargaining debate, showing it requires both camps to reevaluate their policy prescriptions and offering some new proposals of our own to minimize false convictions, better to protect the innocent, and improve the plea bargaining process. I. EMPIRICAL EVIDENCE At least at first sight, when plea bargaining is conducted in the shadow of the trial, innocent and guilty defendants are equally likely to resort to plea bargains. Regardless of the defendant's guilt, the prosecutors and defense attorney should estimate the likelihood of conviction and the expected sentence after such conviction, and adjust the plea bargain to these factors. Prosecutors often offer sentences which are much more lenient than the expected post-trial sentences, in order to assure that the vast majority of the defendants agree to the deals. The level of leniency depends on the prosecutors' caseload which INNOCENCE EFFECT 6 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR determines the number of trials they can handle. In most jurisdictions in the United States, the guilty plea sentence is low enough to assure that about 95% of defendants take the plea. Innocence, in this setting, plays only a minor role, if at all. True, one should hope that innocent defendants, on average, face weaker cases (i.e. a lower probability of conviction at trial). Yet, most weak cases are dismissed at earlier stages; the risk to innocent defendants is substantial mainly in the rare situations where the case against them is relatively strong. Moreover, according to the shadow of trial model, plea bargains are just as likely in strong and weak cases. Prosecutors only need to adjust the offer to the probability of conviction in order to reach an agreement. Thus, weaker cases result in more lenient plea bargains, and stronger ones in relative harshness, but both result in an agreement. Admittedly, when the case is weak, the parties must rely on charge bargaining to make the necessary adjustment, in order to avoid judicial rejection of the agreed sentence. But it is hardly an obstacle. Charge bargaining in weak cases is not the exception; it is the norm all around the country.14 Thus, even if the evidence against innocent defendants is, on average, weaker, the likelihood of plea bargains is not dependent on guilt. In fact, plea bargaining should even be more likely in weak cases. There are many indications that prosecutors are willing to go a long way to avoid losing cases.15 Thus, they are less concerned with handling a full-fledged trial when the case is strong than they are when the case is weak. Of course, this usually leads them to dismiss weak cases, but when prosecutors decide to proceed with such weak cases, they are often willing to go a long way to assure that a plea bargain is struck. When a jury trial is very likely to result in conviction, prosecutors are less eager to extract a plea. Hence, guilty plea rate in weak cases is expected to be higher than in strong cases, if defendants make decisions in the shadow of trial. 14 See Ronald F. Wright & Marc L. Miller, The Screening/Bargaining Tradeoff, 55 STAN. L. REV. 29 (2002). 15 See Robert Rabin, Agency Criminal Referrals in the Federal System: An Empirical Study of Prosecutorial Discretion, 24 Stan. L. Rev. 1036, 1045 (1971) (concluding, based on interviews with federal prosecutors, that convictions are the central performance standard and an increase rate of non-convictions raise questions and create anxieties); See also Bibas, supra note 12, at 2472 ("[Prosecutors] may further their careers by racking up good win-loss records, in which every plea bargain counts as a win but trials risk being losses."); Albert W. Alschuler, The Prosecutor's Role in Plea Bargaining, 36 U. CHI. L. REV. 50, 10607 (1968) (noting that prosecutors are often measured by the rate of convictions and thus care much more about conviction than sentencing). INNOCENCE EFFECT 2/12/2016 [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 7 To examine this hypothesis, we need to sample accurate and wrongful convictions and compare the guilty plea rates in both. Obviously, this is impossible without information about the actual guilt of convicted defendants. Yet, this difficulty can be overcome by analyzing known cases of wrongful conviction and comparing them to convictions of other similar cases. We use two different datasets for that purpose: one, a compiled database of known exonerations and, two, a database of defendants wrongfully convicted in the Tulia scandal. If the shadow of trial model is correct, we should expect plea bargaining to be as common in these wrongful conviction cases as it is in all other cases, which are likely to include mostly guilty defendants. Yet, if the innocence effect hypothesis is correct, innocent defendants, i.e. defendants that were later found to be wrongfully convicted, are expected to go to trial much more often than other defendants. After showing that the wrongfully convicted defendants are much less likely to plea bargain, we discuss several alternative explanations for the results, and show that the innocence effect is the most likely explanation. A. Exonerations Database Using several sources16 we compiled a database of 973 exoneration cases, meaning cases which resulted in a conviction which was later overturned when new information revealed that the defendant was factually innocent.17 We used only cases from the 20th and 21st centuries. We categorized the records in the database based 16 Samuel R. Gross, Convicting the innocent, UNIVERSITY OF MICHIGAN PUBLIC LAW WORKING PAPER NO. 103, ANNUAL REVIEW (2007). Available at SSRN: http://ssrn.com/abstract=1100011; Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson, Nicholas Montgomery & Sujata Patil, Exonerations in the United states 1989 through 2003, 95 J CRIM. L. & CRIMINOLOGY 523 (2005); THE INNOCENT PROJECT, http://www.innocenceproject.org (last visited 14.4.2011); NORTHWESTERN CENTER ON WRONGFUL CONVICTIONS, http://www.law.northwestern.edu/wrongfulconvictions/exonerations/usIndex.html (last visited 14.4.2011). FOREJUSTICE: WRONGLY CONVICTED DATABASE INDEX, http://forejustice.org/db/innocents.html (last visited 14.4.2011). INNOCENCE PROJECT NEW ORLEANS, http://ip-no.org/exonerees (last visited 14.4.2011); TRUTH IN JUSTICE, http://www.truthinjustice.org (last visited 14.4.2011). JUSTICE DENIED: THE MAGAZINE FOR THE WRONGLY CONVICTED, http://www.justicedenied.org/ (last visited 14.4.2011). 17 From the sources above we combined a dataset of 1041 exoneration cases. After excluding 54 cases due to lack of sufficient information about the case, and another 14 cases from the 19 century we remained with a subject group of 973 cases. INNOCENCE EFFECT 8 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR on the following variables: the type of felony (mostly rape or murder), the year of the wrongful conviction, the year of exoneration, the sentence, the type of conviction (plea or trial), the causes for the mistaken conviction18 and the type of exonerating evidence.19 In the guilty plea-cases we also added whether the defendant was facing a death sentence if convicted at trial. The information about each case was gathered from the sources mentioned in footnote 16 above and from searches of the cases on the web. An analysis of the categorized data revealed that only 68 of 973 (7%) of the exonerated were convicted following a guilty plea; in the remaining 93% of these cases, an erroneous jury decision led to the conviction. This 7% rate stands in sharp contrast to the common rate of guilty pleas in felony cases; throughout most of the twentieth century, guilty pleas accounted for 75% to 90% of felony convictions (Alschuler 1979).20 In fact, the 99% confidence interval around the 7% rate in our sample data has a lower bound of 4.9% and an upper bound of 9.1%. The dramatic difference between this confidence interval and the general guilty plea rate of felony cases further emphasizes the strong tendency of innocents to opt for a trial notwithstanding conviction odds. Intriguingly, most of the defendants who pleaded guilty to the crime they did not commit did so after wrongfully confessing to the crime during the police investigation (45 of 68 cases, 66.18%). This result indicates that guilty plea which follows a confession is less reliable than other guilty pleas. Yet, it also indicates that the 18 The causes were eye witness misidentification (563 cases), expert testimony (81 cases), police misconduct (167 cases), prosecutors misconduct (144 cases), false confessions (154 cases), false testimony of an informant (173 cases), ineffective or no representation, alleged scientific evidence (93), alleged suspicion statement of the defendant (18), erroneous polygraph results (1), pre-DNA hair analysis (28). In most cases, more than one cause led to the miscarriage of justice. 19 The exonerating evidence were DNA (296 cases), real culprit found (301 cases), solid alibi proved (37 cases), informant reversed testimony (92 cases), the murder victim was found alive (7 cases), fingerprints (7 cases), ballistic tests (13 cases), blood tests (13 cases). In some cases, more than one proof led to the exoneration. 20 All the cases in the database refer to trials in the twentieth and the twenty first centuries. Importantly, guilty plea rates remain at a comparably low 6.5% (42 pleas of 645 cases) even when examining only post-1970 cases. Furthermore, Gross et al. (2005) obtained a similar guilty plea rate upon examining 340 exonerations which took place between 1989 and 2003. They found 20 defendants of the 340 wrongfully convicted defendants (5.9%) pleaded guilty. Thus, this 6% rate of guilty plea seems to represent old and new exoneration cases alike. INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 9 innocence effect is even stronger when the innocent defendant did not confess to the crime. Only 23 defendants of the 812 (2.8%) exonerees who did not confess during the police investigation, pled guilty. The 99% confidence interval around the 2.8% rate in this sample data has a lower bound of 1.31% and an upper bound of 4.29%.21 We also analyzed each of the common offenses in the database separately to get a more nuanced perspective on the data. Although a simple comparison of the raw plea rates for known wrongful convictions versus convictions in general makes the differences between the two groups readily apparent, we conducted further statistical tests to underscore these dramatic differences. However, our database spans over a century of cases, while recordkeeping for U.S. Department of Justice annual felony statistics (published through the Bureau of Justice Statistics) is relatively new. We therefore used 1990, the earliest record we were able to obtain, as a benchmark year for comparison. The data for 1990 was then compared to a subset (N=526) of the 973 wrongful conviction cases in the database. These were the cases whose conviction years spanned from 1980 onwards. Echoing the general pattern found above, only 8% (42 of 526) of the 1980-2000 subset entered guilty pleas, compared to the 91% plea rate among all convicted defendants of violent crimes in the benchmark year 1990. Among the 198 rape and sexual assault exonerations during this timeframe, only 10 followed a guilty plea, which constitute 5.05%; this figure is obviously significantly different from the 95% guilty plea rate among rape and assault convictions in 1990. Among murder and manslaughter cases, we found a guilty plea rate of 7.7% among exonerated defendants (22 of 286). Here too, the difference between this 7.7% rate and the 90% benchmark guilty plea rate among comparable convictions is still dramatic and highly significant. The relatively high number of guilty pleas in these cases may have resulted from the risk of a death sentence at trial; in fact, at least 16 of these 22 defendants explained their guilty plea was driven by their fear of capital punishment. It is therefore possible that when bargaining is conducted in the shadow of the death sentence, even innocents are more likely to plead guilty. 21 Discussing why false confession impact on the defendants' willingness to plea is beyond the scope of this article. INNOCENCE EFFECT 10 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR It is plausible, though, to argue that innocent defendants who plead guilty tend not to challenge their convictions, and hence their misfortune is less likely to be revealed.22 We will discuss this alternative explanation below. Still, to reduce the risk of such selection bias in the database, we examined, separately, the subset of cases in which the exoneration resulted from events which are less likely to depend on the defendants' effort. Seven defendants in the database were exonerated after the alleged murder victim appeared alive. All of these seven defendants pleaded not guilty and convicted by jury trial. Similarly, 235 defendants were exonerated only after the real offender was revealed.23 Even in this subset of exonerations, only 8.9% of the convictions resulted from guilty pleas (21 of 235). This percentage is not significantly different from the remaining 49 of 738 exoneration cases in the database (6.64%) that followed guilty pleas. B. The Tulia Case In the small town of Tulia, Texas, one corrupt, undercover police officer framed about 38 people in allegedly independent cases of drug trafficking.24 Years later, they were exonerated and compensated, and the officer was convicted of perjury. The behavior of these unrelated innocent defendants who knew that a police officer was willing to testify that they sold him cocaine can indicate how innocent defendants behave. The first eight defendants, who were brought to trial consecutively, made no attempt to plea bargain, and, as a consequence, were sentenced to terms ranging from 20 years of imprisonment to 434 years after jury trials. Among them was, for example, Joe Moore, who rejected his lawyers repeated advice to plea bargain on the eve of the trial, even though he was told that they would not be able to follow up on the officer's testimony at the time. With two previous drug offences in his record, with no alibi, and with his word against the word of a white police officer, his lawyer 22 See Huff, Rattner, and Sagarin 1996 In fact, 300 defendants were exonerated after the real offender was revealed however, in 65 cases there were additional evidence leading to the exoneration. Here we refer only to the remaining 235 exonerees. 24 In fact, forty three people were arrested and all but one were charged but charges against four defendants were dropped at later stages when these defendants presented a solid alibi. 23 INNOCENCE EFFECT 2/12/2016 [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 11 was right to argue that he had no chance at trial.25 Similarly, Fred Brookins, who knew that five defendants before him were convicted based on the testimony of the same officer that was about to testify against him, rejected a deal for five years of imprisonment on the eve of the trial, against his lawyer's strong advice, arguing that he shouldn't plead guilty to something he didn't do. After the jury trial, he was sentenced to 20 years of imprisonment.26 Considering the rarity of jury trials in felony cases in Texas (only 4% of the felony convictions in the state result from trials), the independent decisions of these eight defendants support the innocence effect hypothesis. True, following the harsh sentences imposed on these first eight defendants, only three more similarly situated defendants went to trial. The other 27 defendants pleaded guilty, most of them to nonincarceration sentences. The behavior of the later 30 defendants indicates that when the plea concession is very substantial, and when the high likelihood of conviction at trial is very vivid (here, as the result of the conviction of the first eight defendants) even innocent defendants are likely to accept a plea bargain. Still, even when all of the defendants in the case are considered, the trial rate of the Tulia defendants was about 30% (11/38), much higher than the regular rate of jury trials in Texas (about 5%).27 The 99% confidence interval around the 30% rate in this sample data has a lower bound of 10% and an upper bound of 48%. C. Alternative Accounts The empirical result above seems to contradict the shadow of trial model, and indicates that innocence, in itself, has an effect on the willingness to plead guilty. However, before we adopt the innocence effect hypothesis, we should address several alternative 25 See NATE BLAKESLEE, TULIA: RACE, COCAINE AND CORRUPTION IN A SMALL TEXAS TOWN 44-45 (2005). Moore was frustrated by his lawyer's lack of trust and asked to dismiss him, but the court rejected his request, as well as a request for continuance in order to examine the credibility of the officer. Even after his requests were rejected, he refused to accept plea bargain offers and, as expected, was easily convicted and sentenced for 99 years of imprisonment, after a trial that lasted less than a day. 26 Id. 148. 27 Based on the BJS (show based on the table in the web the rate of jury trials in Texas in generally and in small counties specifically [say that the result remains even if all counties are examined) [We used the Bureau of Justice Statistics (2001) to compile the rate of guilty pleas in felony cases in rural counties with fewer than 50,000 inhabitants in Texas, which resemble Tulia's Swisher County.] INNOCENCE EFFECT 12 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR explanations to the relatively low rate of guilty pleas among innocent defendants. First, one can argue that when defendants are innocent, the cases against them are weaker on average, and hence, they are less willing to plea bargain. However, as long as the prosecutors know the strength of the cases, this should not impede a plea bargain. We have shown above that prosecutors are willing and able to adjust the post-plea sentence to the probability of conviction using charge, fact, and sentence bargaining.28 The substantial sentence discounts offered to the Tulia defendants are just another example of such prosecutorial behavior.29 Moreover, want to maintain a high conviction rate.30 Trials in strong cases do not endanger their conviction rate; trials in weak cases do. Thus, they dismiss most of the weak cases, and make a special effort to induce a guilty plea in the weak cases which were not dismissed. This effort comes in the form of very attractive deals to these defendants. On the other hand, prosecutors are less worried about conducting jury trials in strong cases. Hence, it seems it is the innocence, and not the weakness of the evidence, which plays a role in the decision to choose a trial. Second, one might argue that the low rate of plea bargains with innocent defendants results from a rational evaluation of the trial result. Since trials are designed to reveal the truth, Innocent defendants might rationally estimate that their chances are better than guilty defendants. Prosecutors cannot know which defendants are innocent and hence they cannot take innocence into account when adjusting the plea offer to the probability of conviction. Therefore, this information asymmetry hinders the negotiation with innocent defendants.31 While this information asymmetry has some effect, its magnitude should not be overstated. Trials, like plea bargaining, are affected by evidence. To the extent that the innocence is not reflected in the evidence, it has little effect on the trial, and hence it should not affect the bargaining in the shadow of the trial.32 To the extent that 28 Refer to the discussion of charge bargaining Refer to the substantial offers given to the Tulia defendants as mentioned in the text. 30 Reference 31 Refer to grossman and kats , scott and stuntz 32 Stephen J. Schulhofer, Plea Bargaining as Disaster, 101 YALE L.J. 1979, 1984 (1992) ("Innocence by itself (that is, apart from its link to particular evidence) can have only a small impact on the odds of conviction") 29 INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 13 innocent defendants have evidence which can assist them at trial, they can use it during plea bargaining too. True, in few rare cases, innocent defendants might prefer to conceal the exonerating evidence in order to surprise the prosecution at trial, but in most cases, rational innocent defendants would rather use such evidence earlier to convince the prosecutor to dismiss the case, or, at least to agree to a better plea bargain. Thus, it is hard to believe that almost all of the wrongfully convicted defendants in the exoneration database did not settle the case simply because they knew their chances of acquittal at trial were high, and the prosecutor did not. Moreover, from the account we have about the Tulia defendants we know that these defendants could not rationally expect that the trial result would be different than it really was. Furthermore, prosecutors are usually willing to go a long way to secure a plea bargain, especially when there is a real chance that the trial will not result in a conviction. They know that different defendants have different attitudes toward risk, different estimations of trial results, and different characters, and they must offer substantial leniency to assure that almost all defendants will be willing to plea. To assure that plea bargains are accepted by more than 90% of the defendants, prosecutors must be willing to offer deals which are attractive, not only to the average defendant, but to almost all of the defendants. Therefore a small difference in the parties' estimations of the trial result should not prevent them from reaching an agreement. Note that we do not argue that innocent defendants always correctly estimate their chances at trial. To the contrary, as we show below, cognitive biases lead innocent defendants to be overoptimistic about their chances at trial. Here, we only argue that it is unlikely that the huge gap in guilty pleas of innocent and guilty defendants is explained by the rational assessment of innocent defendants that the trial would produce exonerating evidence, even though they are unable to reveal this evidence before the trial. Another limitation of our conclusion lies in the representativeness of the exoneration database. It might be argued that the proportion of innocent defendants who plead guilty is around 90% - as it is with the guilty defendants – but the exoneration database depicts an unrepresentative subgroup of these wrongful convictions in some relevant aspects. INNOCENCE EFFECT 14 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR For instance the exonerations database refers mainly to murder, manslaughter, and sex offences. The Tulia scandal refers to drug trafficking offences. It is possible that in less serious offences, innocent defendants plead guilty more often. However, this does not pose a substantial limitation on our findings, as long as we limit them to these serious offences. Alternatively, there might be some correlation between the defendants' plea and their chances of exoneration. It is possible that many innocent defendants plead guilty and after such a plea, they do not try to overturn their conviction. On the other hand, the few innocent defendants that opt for a trial are also more vigorous in their post conviction attempts to be exonerated. If that is the case, the exoneration database only depicts the innocent defendants who are willing to defend themselves at trial, and later keep on challenging their innocence, and misses the majority who do not. One way of responding to this shortcoming of our database was discussed earlier. As we showed above, the guilty plea rate in the wrongful convictions database remains almost the same even when we analyze the subset of exonerations that resulted from the detection of the offender, and not mainly from the efforts of the defendant. True, some real offenders might be revealed because the wrongfully convicted person kept on challenging his conviction, yet, we believe that it is highly unlikely that this is the case in over 90% of these cases, as the data suggests. Another, and in our view, even more convincing response to the selection bias argument lies in the behavior of the Tulia defendants. These defendants were not exonerated as a result of their efforts. In fact, the efforts to exonerate these defendants were initiated by pro-bono lawyers who heard of the case only after the first eight defendants were convicted. It seems that the refusal of so many defendants in the Tulia scandal to bargain can only be explained by the innocence effect hypothesis. The Tulia database is important in other aspects too. The detailed information we have on the Tulia cases can assure us that there were no unique factors which could explain the high rate of trials other than the innocence effect. We know that the objective probability of conviction was high, that the defendants were told that it was high by their lawyers (and could have known it was high based on the experience of their predecessors), that at least some of them were offered very lenient plea bargains, and that when they INNOCENCE EFFECT 2/12/2016 [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 15 decided to plead not guilty, they did not do it because they planned to surprise the prosecutor with new exonerating evidence at trial.33 These details do not support the alternative explanations to the innocence effect hypothesis. Thus, our analysis of exoneration statistics suggests innocent defendants are more likely to reject plea offers. Though we think these findings are best explained by the innocence effect, we agree that we cannot totally reject other potential explanations based on these findings alone. We thus turn to analyzing experimental studies which, as we show below, also support the innocence effect hypothesis. II. EXPERIMENTAL STUDIES Empirical studies that rely on existing data enjoy the benefit of external validity - i.e., the relevance of their conclusions to the real world is unquestionable - and there lies the secret to their power.34 However, such studies are limited in their ability to control variables that cause the examined phenomenon, because a large number of factors affect the occurrence of phenomena in the real world.35 This limitation is reflected in the analysis of the findings in Part I c. above, when we saw, for example, that the significant difference between the behavior of the innocent to that of the guilty can also be attributed to the characteristics of the sample or to the information gaps between the innocent defendants and the prosecution. Conversely, the experimental methodology can isolate the various factors affecting the behavior of defendants in a plea bargain, as the studies described below will demonstrate. This advantage of the experimental approach stems from its ability to compare the behavior of different groups of subjects while controlling the relevant variables. Typically, the investigator will perform a manipulation on a single variable (“the independent variable") in one group of subjects but not in another group, which will serve as a control group. If we find a systematic difference between groups in a measured variable ("the dependent variable") that would have a 33 See BLAKESLEE, supra note 25. See Avishlom Tor, The Methodology of the Behavioral Analysis of Law, 4 Haifa Law Review, 237, 281-291 (2008). 35 Ibid. 34 INNOCENCE EFFECT 16 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR statistical reliability, it could be attributed to the effect of the independent variable. The obvious advantage of the experimental approach lies in its ability to control variables, thus strengthening the validity of its conclusions regarding the relationship between variables in the environment of the experiment. This advantage also raises an inevitable question about the external validity of the findings of the experiment. We refer to the findings of psychological experiments that examine human behavior, so the external validity depends on the similarities and differences between the experimental environment characteristics and the relevant characteristics of the reality to which one would seek to attribute the findings of the experiments.36 In the context at hand - regarding plea bargains with criminal defendants – there seems to be an embedded difficulty in creating an experimental environment with external validity. First, it is possible that criminal defendants possess different psychological characteristics than the rest of the population, but of course there is an obvious difficulty in using defendants as subjects to an experiment. Second, and mainly, the nature of the dilemma that the criminal defendant is facing - especially when charged with serious crimes involving a long prison term - makes it difficult for a convincing simulation in an experimental environment. Third, and as a continuation of the above, even in minor offences and punishments, cataloging subjects as defendants without their knowledge necessarily involves an element of deception that raises significant ethical issues.37 However, it does not seem that the difficulties involved in using an experimental approach to examine decision-making by defendants justify a complete waiver on attempting to examine the issue using a methodology that has findings with internal validity. Furthermore, this eclectic approach that combines field research with 36 See: Robert Rosenthal & Ralph L. Rosnow, Essentials of Behavioral Research: Methods and Data Analysis (3rd ed. 2007). For Discussion about the benefits and limitations on the use of experimental methodology, see: Tor, ibid, and also: Avishalom Tor, On Contractual Defaults and Experimental Law and Economics, 163 Journal of Institutional and Theoretical Economics 26, 28 (2007) (on the importance of taking the unique characteristics of the judicial system in to account when constructing experimental research 37 For a review on the ethical restrictions on human experimentation, see, for example, Tor, Contractual Defaults, ibid. yet. It should be noted that another study, conducted several decades ago, used a similar simulation, as described below. INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 17 experimental research can provide a more comprehensive and valid empirical picture of the test subject, as each of the methodologies compensates for weaknesses of the others. Indeed, the findings from the field described in part I.c. above clearly show that in the real world there is a systematic difference between the willingness of innocent and guilty defendants to accept plea bargains. These findings therefore provide a significant solution to the problem of external validity, but leave the need for a thorough examination of the variables that affect the willingness of defendants to accept plea bargains. However, an examination of the literature reveals that only a few experimental studies have examined the behavior of defendants in a plea bargain. Moreover, except for research we recently conducted,38 the other relevant studies that are few in number were held in the late seventies and early eighties of the last century, before the shadow of the trial model was established. For this reason, these studies did not make a conscious attempt to compare the findings to the shadow model. In addition, the structure of the experiments and the system in which the findings were reported prevent such comparison in retrospect.39 Despite their limitations, these studies also reinforce the hypothesis that innocence affects the willingness of defendants to plead guilty. In all of these studies the subjects were given information about the charges they were facing and either the evidence in the case or the probability of their conviction. Also, some of the subjects were told that they committed the offense while others were told the opposite. All studies showed much less willingness to plea bargain among defendants who did not commit the offense they were charged with.40 To establish the relevance of the experiment’s findings in the context at hand, one study also examined the behavior of the subjects who were placed in a situation where they were accused of 38 See, Tor, Gazal-Ayal & Garcia, supra note 5. See Ibid. 40 In the study by Gregory, Mowen & Linder, supra note 10, p.1524, 83% of the guilty defendants and 18% of the innocent defendants accepted the offer. In the study by Bordens, supra note 10, 79.6% of the guilty defendants and 20.3% of the innocent defendants accepted the offer. In the first experiment in a research we conducted 67% of the guilty defendants and 20% of the innocent defendants agreed to the offer presented to them. See: Tor, Gazal-Ayal & Garcia, supra note 5. 39 INNOCENCE EFFECT 18 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR committing a disciplinary offense at the university, without them knowing that the accusations they were presented with was a part of an experiment.41 In that case, sixteen psychology students who were waiting in line for a test met, seemingly at random, a man posing as a student. The impostor told half of the subjects that most of the answers to the questions on the test are “B” and to the other half he said nothing meaningful. After the test, all subjects were summoned to a clarification meeting and were told that the test results raised concern that they did not solve the test on their own. Each and every one of them was offered some sort of a plea bargain instead of proceedings before the department’s ethics committee. While six out of eight subjects who received information about the test agreed to the arrangement, the remaining eight who did not receive any information regarding the test refused to do the same. Since the evidence presented to the innocent and guilty students was identical, and no student ("innocent" or "guilty") had access to evidence that could lead to an acquittal, it is reasonable that there were no significant differences in the chances of conviction for all of the students in general (though it may well be, as we shall later show, that the innocent assessed their chances to be better than those of the guilty, although there was no rational objective basis to that assessment). This finding reinforces the conclusion that innocence has an independent influence on the assessment of the chances of an acquittal or on the willingness to agree to a settlement. In addition, the behavior of the innocent and the guilty in this experiment – in which they believed in their innocence or guilt, as the case would be – resembled the behavioral characteristics of subjects in similar experiments who used questionnaires in which innocence and guilt were only hypothetical. This similarity reinforces the assumption that in a plea bargain, as in many other areas of human behavior, experimental studies based on hypothetical surveys generally reflect the existing trends outside of the laboratory and are therefore a valuable source for the study of plea bargains.42 41 See Mowen & Linder, supra note 10, p. 1526-1528 In addition, an exceptional study examined the impact of procedural elements of the criminal procedure on decisions made in the framework of plea bargains. This study compared criminal defendants to typical subjects (university students) and its findings showed that the vast majority of the variables examined had a similar effect on both groups. See Pauline Houlden, Impact of Procedural Modifications on Evaluations of Plea Bargaining, 15 Law & Soc’y Rev. 267 (1980). This finding also contributes to the external validity of the typical experimental methodology in the context at hand. 42 INNOCENCE EFFECT 2/12/2016 [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 19 These studies, therefore, despite their inherent limitations, support the assumption that innocence has an independent influence on the decisions of defendants, contrary to predictions of the “shadow of the trial” model. III. INTERVIEWING CONVICTS A third research group relies on questions addressed to defendants in retrospect, after the criminal proceeding has ended. In this research group, some researchers showed that considerations of expediency on which the shadow model is based upon - are not the first to be taken into account in terms of their importance to formulating a decision in the matter at hand. These studies have shown that many defendants plead guilty simply because they are in fact guilty, and on the other hand, defendants that allege that they are innocent, explain that they plead "not guilty" for the reason that they are not guilty. For example, one study shows that the most common answer to the question "Why did you plead guilty?" is "Because I'm guilty". Answers that relied on utilitarian considerations were far less common (the desire to quickly end the proceedings was a major reason for 10% of the defendants, and the desire to gain lighter penalty was the main reason for 5% of the defendants only).43 Similarly, almost everyone who pled not guilty explained their actions by claiming that they did not commit the offense or that the act committed did not constitute an offense in their opinion. 44 In many cases, these defendants told the researchers that they pleaded not guilty even though they would have decided differently if they would have base their decision on expediency considerations.45 Similar findings were shown in a research carried out by Ami Kobo in Israel.46 Kobo interviewed about 400 defendants immediately after their plea in the Israeli Magistrate Court. When asked "Why did you plead guilty?" or "Why did you plead not guilty?" most of the defendants answered consistently that they did so because they spoke the truth or because they committed the offense (53%). Only a few of the defendants mentioned expediency 43 ANTHONY E .BOTTOMS & JOHN D. MCCLEAN, DEFENDANTS IN THE CRIMINAL PROCESS 111 (1976). 44 ibid, in p.131 For a more detailed analysis of this study See infra note ???. 46 AMI KOBO, INCONSISTENT PLEADERS IN COURT - PLEADING GUILTY AND CLAIMING TO BE INNOCENT (2009) (in Hebrew). 45 INNOCENCE EFFECT 20 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR considerations such as the desire to end the case (10.2%), a plea bargain, or an expected penalty relief pending on admission (11.2%). Another common reason that was given was "I was caught redhanded" or similar versions of the same reason (23.7%). Most of the defendants who pled not guilty stated their very innocence as a top priority consideration in their decision to deny guilt. The main disadvantage of these studies stems from the concern that defendants give a distorted or false version of the truth in an attempt to present themselves in a positive light. However, in order to address this concern studies are usually conducted after the proceedings have ended, when the defendant knows that his or her statements will have no effect on the outcome of the trial. Moreover, as we shall later show, there were times when defendants actually presented themselves in a negative light. For example, the vast majority of defendants who pleaded guilty in trial (that is to say, most of the subjects in these studies) also admitted committing the offense in the survey and even stated that the main reason for their willingness to plead guilty was that they in fact had committed the offence they were accused of. Even if the credibility of those who claimed to be innocent during the survey is questionable, it is hard to believe that respondents who admitted committing the offense did so to present themselves in a positive light. Therefore, one can probably learn from these studies that the subjective guilt of defendants is perceived by them as a significant consideration in their decision to plead guilty. In Conclusion: Although each of the three methodologies presented here suffers from various limitations, the broad range of studies concludes that innocent defendants show less willingness to accept plea bargains than guilty defendants. The question yet to be answer is why. We turn to this question now. IV. PSYCHOLOGICAL INSIGHTS The findings presented in the previous section indicate that innocent defendants are less willing to plead guilty, as required in a plea bargain, than guilty defendants who are facing similar charges and evidence. These findings are not sufficient by themselves in order to establish an effective legal policy for the issue at hand. Doing so would require the examination of the leading causes of the systematic difference between the innocent and the guilty. For example, there may be room to implement a certain legal policy in cases in which the innocent prefer to refuse plea bargains despite INNOCENCE EFFECT 2/12/2016 [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 21 being aware of their objective chances of conviction. But, when handling cases in which innocent defendants’ refusals stem from over-optimism about the outcome of the trial, a different legal policy should be implemented. It is therefore important to try and summarize the findings that may shed light on the factors that can explain the different behavior of the innocent, despite the limited scope of the studies dealing with this issue directly. A. Evaluating the probability of conviction The shadow model acknowledges the fact that litigating parties may not reach a compromise because of differences in their assessments of the trial’s outcome. Therefore, when the defendant evaluates the prospects of an acquittal to be higher than the prosecution’s evaluation, negotiations may fail. In order to minimize the number of such failures, the prosecution that depends on plea bargain increases the proposed "discount" in the charges, and by doing so ensures that the vast majority of defendants will accept the plea bargain offered. Nevertheless, it may be that the few defendants that still refuse to accept plea bargains are those defendants who value their chances at trial to be higher than the prosecution would think. If innocent defendants are over-represented in this group, then this could explain their reluctance to plead guilty. Indeed, the shadow model also shows the possibility that evaluation discrepancies between the prosecution and the defendants are larger when it comes to innocent defendants. As explained above, a defendant's innocence may affect the chances of an acquittal even if the defendant do not possess evidence which he or she can show the prosecutors to support the contention of innocence.47 In such cases, innocent defendants will tend to reject plea bargains that guilty defendants in a similar situation will accept and their decision will be based on the information they hold exclusively but unable to credibly transfer to the prosecution. However, the shadow model only provides a partial picture of these evaluation discrepancies. In addition to private and incomplete information held by one of the parties exclusively, there are other psychological factors that contribute to the over-optimism of innocent defendants in comparison to guilty defendants who are 47 If such evidence is available it can be presented to the prosecutor. This way, the evidence will have a direct affect on both sides’ evaluation of the outcome of the trial, and will not create assessments discrepancies. See text to note 32 supra. INNOCENCE EFFECT 22 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR facing similar evidence. Experiments that examined this issue show that the innocent tend to believe they can convince the court that they are innocent even in the absence of an objective basis for their claim. For example, in one of these studies, subjects were asked to imagine that they are accused of armed robbery. They received information about the event and listened to a recording of their defense lawyer describing the evidence against them. Even though all subjects received identical information and heard the same things from the defense attorney, when asked to evaluate their attorney’s belief in their innocence, innocent subjects’ evaluation was significantly higher. Evaluation of the probability of an acquittal was also higher among the innocent.48 Similarly, in another study in which subjects were told that there were three witnesses to the crime, innocent subjects assessed their chances of conviction to be lower.49 As previously mentioned, the two studies reviewed earlier used an experimental methodology. However, if a casual statement made to the subjects affected their estimation of their lawyers trust in them or their ability to convince the jurors of their innocence, than true guilt or innocence in real cases would probably have a much larger effect on such assessments.50 The Systematic difference in the evaluation of the probability of conviction between the innocent and the guilty is consistent with the "illusion of transparency”; i.e. the exaggerated belief of people in the ability of others to distinguish between truth and lies.51 If defendants believe that the court or the attorney can easily find out if they are lying or telling the truth, they will adjust their assessment of the trial’s outcome to their guilt even though the objective evidence they are facing did not change. 48 Mowen & Linder, supra note 10, p.1522. See Bordens, supra note 10. 50 Indeed, studies in the field of civil litigation and compromise show that determining the role of participants in the experiment, as prosecutors or defendants, is enough to create a bias in the direction of the way they value the same body of evidence. This bias leads to a consistent discrepancy between the parties’ expectations of the trial’s outcome and the diminution of the chances to a compromise in a negotiation. Similar findings were also reported in field studies of civil negotiations, e.g.: Linda Babcock & George Loewenstein, Explaining Bargaining Impasse: The Role of Self-Serving Biases, 11 Journal of Economic Perspective 109 (1997). 51 Thomas Gilovich, Kenneth Savitsky & Victoria H. Medvec, The illusion of transparency: Biased assessments of others Ability to read our emotional states, 75 Journal Personality & Social Psychology, 332 (1998). 49 INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 23 The common belief that the world is fair52 may also contribute to judgmental bias related to private information held by defendants. Thus, the guilty may greatly fear that more evidence may be found to ensure their conviction, while the innocent may believe their lawyers will gather information or witnesses that will prove their innocence. These differences in individuals’ assessments may lead defendants to believe in popular illusions, such as the belief in the judge's ability to expose lies or, alternatively, encourage the innocent into mistakenly thinking that their refusal to plea bargain will serve as evidence of their innocence.53 Moreover, the guilty may overestimate their chances of conviction because of the availability bias.54 This bias leads decision makers to estimate the probability of events based on the ease in which the events are constructed in their mind and are accessible to them. Thus, events that are easier to be remembered or imagined are perceived as having higher probability of occurring. In most cases, common events, for example, are better recalled and their occurrence is easier to imagine. Therefore, the availability is an effective heuristic. However, there are factors that affect memory or the ability to imagine events without changing the probability of their occurrence, thus causing expected errors in probabilistic judgment55. When offered a plea bargain the defendant has to predict the judge’s expected assumption regarding his guilt or innocence in order to decide whether the plea bargain is attractive or not. Guilty defendants are likely to better remember evidence indicating their guilt and will find it easier to imagine how this information, in the 52 The source of the theory is in the righting of lerner. See for example: Melvin J. Lerner, The belief in a just world: A fundamental delusion (1980). For a thorough review of the experimental research about the phenomenon, its scope and limitations see: Carolyn L. Hafer & Laurent Bègue, Experimental Research on Just-World Theory: Problems, Developments, and Future Challenges, 131 Psychcological Bulletin 128 (2005). 53 See Mowen & Linder, supra note 10. 54 See Amos Tversky & Daniel Kahneman, Availability: A Heuristic for Judging Frequency and Probability, 4 COGNITIVE PSYCHOLOGY 207 (1973). An event can influence the subjects estimation of its likelihood even if its availability is only the result of a request to the subjects to imagine its occurrence. See John S. Carroll, The Effect of Imagining an Event on Expectations for the Event: an Interpretation in Terms of the Availability Heuristic, 14 JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY 88 (1978) (finding that subjects who requested to imagine events evaluated the probability of the events as higher than did other subjects. 55 For more samples from experiments and further implications, see Tversky & Kahneman, Ibid, Tor, supra note 34, p. 248-249. INNOCENCE EFFECT 24 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR hands of the judge sitting on the bench, would lead to their conviction. On the other hand, innocent defendants are expected to show the opposite tendency and assess the likelihood of their conviction to be lower. Therefore, it seems there is a string of psychological factors causing innocents’ tendency to overestimate their chances of acquittal - a tendency that leads them to reject plea bargain offers that guilty defendants would accept in similar circumstances. B. Fairness-related preferences Conviction of the innocent is perceived as an unjust or unfair outcome.56 Studies show that people are deterred by consequences they perceive as unfair to them in general,57 particularly in negotiations,58 and are even willing to pay a significant price for their decisions.59 Studies of plea bargains also show that even when the probability of acquittal is similar for innocent and guilty defendants, innocent defendants will tend to reject plea bargain offers that the guilty will tend to accept.60 Similarly, as noted earlier, studies which 56 Thus, for example, it is common to refer to the conviction of the innocent as unfair or immoral, see Ronald Dworkin, Principles, Policy, Procedure, in A MATTER OF PRINCIPLE 72 (1985); HERBERT L. PACKER, THE LIMITS OF CRIMINAL SANCTION 250 (1968). 57 See: Max H. Bazerman, George F. Loewenstein & Sally B. White, Reversals of Preference in .Allocation Decisions: Judging an Alternative Versus Choosing Among Alternative, 37 Administrative Science Quarterly 220 (1992); Daniel Kahneman, Jack L. Kentsch & Richard Thaler, Fairness as a constraint on profit seeking: Entitlements in the market, 126 Am. Econ. Rev. 728 (1986) 58 Colin F. Camerer, Behavioral Game theory: Experiments in Strategic Interaction (2004); Alvin E. Roth, Bargaining Experiments, in The handbook of Experimental Economics (John H. Kagel & Alvin E. Roth eds., 1995) . 59 See: Camerer, Behavioral game theory; Roth, Bargaining experiments (the willingness to suffer a loss in the ultimatum game, even when it comes to large amounts); George F. Loewenstein, Leigh Thompson & Max H. Bazerman, Social Utility and Decision Making in Interpersonal Context, 57 J. Personality & Soc. Psychology, 426 (1989). 60 However, it is important to note that one of the experiments that were conducted in this study indicated the narrowness of the distinct gap between innocent and guilty when the probability of conviction is extremely high. A similar finding was also shown in Bordens, supra note 10, p.67. This finding suggests that the innocent perceive conviction in itself as a relatively heavy penalty they wish to avoid even at the great risk of a formal severe punishment. On the other hand, guilty defendants are not willing to face a similar risk in order to avoid conviction. When the probability that opting for a trial would lead to an acquittal is significantly reduced, the guilt and innocence play a smaller role in defendants' decision. For a INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 25 examine the reasons given by real defendants for pleading either guilty or not guilty, show that actual guilt or innocence is perceived as a relevant consideration, in contrast to the shadow of trial theory.61 As shown above, the vast majority of defendants who plead not guilty reason their decision simply by arguing that they are not guilty, because they believe the act done does not constitute a crime, because they have a legal justification or exemption, or because of other reasons. Only a small number of the guilty pleaders rationalize their decision on some king of cost benefit analysis.62 Moreover, one study found that about 26% of those who pled not guilty and were convicted and 21% of those who pled not guilty and were acquitted told their interviewers that at the time they made their decision they estimated that they would be convicted. It is clear that at least for the latter group, the defendants that were acquitted to their own surprise, this assessment do not reflect hindsight or the impact of cognitive dissonance due to the drastic tension between denying guilt and conviction.63 All of the above support the conclusion that defendants do not only consider the direct costs and benefits of their plea; actual guilt and innocence are an important factor in their decisions too. The fact that only a very small number of respondents base their decision to plead not guilty on considerations of profitability is consistent with the conclusion that fairness does influence guilty defendants to accept plea bargains and innocent defendants to reject them. These findings are also supported by a string of anecdotes on defendants who rejected offers of immediate release from prison in exchange for guilty plea. These defendants were willing to pay the high price of many years of imprisonment for their innocence. Such is the story of Kelly Jarrett, who was convicted at a young age of discussion of this phenomenon, its implications and the limitations of the existing findings See Tor, Gazal-Ayal & Garcia, supra note 5, part II C.2. 61 See: Bottoms & McClean, supra note 43, p. 111. as noted above, their study show that the common answer given to the question “why did you plead guilty?” was “because I am guilty”, while only a few answered that the benefits for a guilty plea had a substantial effect on their decision. 62 Ibid, p 130-131. 63 The Cognitive dissonance theory holds that tension between the two factors that are psychologically important is a driving force that motivates people to change their beliefs or their assessments in ways that reduce this tension. See for example the extended review of the theory in: Cognitive Dissonance: Progress on a Pivotal Theory in Social Psychology (Eddie Harmon-Jones & Judson Mills ed.,1999) INNOCENCE EFFECT 26 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR murder she apparently did not commit, and was sentenced to 25 years to life imprisonment. Jarrett turned down several offers including an offer to plead guilty in exchange for a time-served sentence that would result in her immediate release. Jarrett explained she could not live with herself if she pled guilty to a terrible crime she had not committed. She believed she would be vindicated by the court of appeals. Similarly, Kerry Max Cook was twice convicted and sentenced to death for the murder and mutilation of a woman. After one mistrial, Cook was convicted and sentenced to death a second time. When the U.S. Supreme Court ordered the court of criminal appeals to review the case, the prosecutor offered Cook the chance to plead guilty in exchange for his immediate release. Despite the risk of another possible death sentence, Cook refused to plead guilty to a murder he did not commit. It was only after he had been offered a Nolo contendere settlement – that allowed him to maintain his assertion of innocence – did he agree to the deal that set him free. Two months later, a DNA testing proved that someone else committed the crime and that Cook was innocent64. In summary, the combined data reinforces the conclusion that innocents’ belief that their conviction is unfair affects their willingness to accept a plea bargain. V. NORMATIVE IMPLICATIONS As mentioned above, we showed that defendants’ innocence has a bearing on their willingness to plead guilty in exchange for leniency, and that this phenomenon has rarely been addressed in the innocence problem debate. However, if the innocent do tend to reject plea bargains, then both sides of the debate should reconsider their position. Critics of plea bargains ,who argue that the widespread use of plea bargains encourages the innocent to plead guilty, should conclude that their concern is not as significant as the shadow model predicts. On the other hand, the claim of those supporting plea bargains, whereby a plea bargain is some sort of an insurance policy for the innocent defendant who does not want to risk a trial, should now be reviewed given that this insurance mainly benefit the guilty defendants, not the innocent ones. In this section we shall discuss the 64 For the full stories, see: Confronting the Plea: Four Stories, FRONTLINE, Available at : http://www.pbs.org/wgbh/pages/frontline/shows/plea/. INNOCENCE EFFECT 2/12/2016 [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 27 consequences of the innocence effect, and present initial ideas for remedies. A. Miminizng false convictions Opponents of plea bargaining often refer to the innocence problem and argue that plea bargaining is wrong because it put pressure on innocent defendants to plead guilty and thus lead to wrongful convictions. Even if these defendants are better off pleading guilty, because they want to overcome the risk of trial, society has a moral obligation to prevent them from doing so, and to minimize the risk of false convictions.65 It is better that ten – or a hundred or a thousand – guilty persons escape conviction than that one innocent does not.66 Plea bargains lead to the conviction of innocent defendants some of whom would have been acquitted in court, thus violating this moral principle.67 By this view, there are several reasons why society must not consciously embrace a procedure that result in a guilty plea of so many innocent defendants, even if it is a direct result of the defendant’s voluntary choice. First, conviction of the innocent does not solely harm the defendant; it also has broad and negative implications to the society.68 Second, convicting the innocent is morally wrong, regardless of the direct cost it impose on the innocent and the society at large.69 That is to say, leaving defendants with no choice but to go to trial and face the risk of a heavier penalty may burden them, but this burden is not moral equivalent to assuring a wrongful conviction through guilty 65 In Regard to the importance of the duty to prevent false convictions see: Zuckerman, p.125: “The Protection of the Innocent from conviction is a central theme of the law of criminal evidence… The importance of protecting the innocent from conviction is not justified only on the basis that it will produce the best social results”) ;also see :In re Winship, 397 U.S. 358, 364 (1970) (“It is critical that the moral force of the criminal law not be diluted by a standard of proof that leaves people in doubt whether innocent men are being condemned”) 66 See Blackstone’s famous saying: “Better that ten guilty persons escape, than that one innocent suffer” in: William Blackstone, Commentaries on the Laws of England 352 (1765-1769). The ratio of 1:10 is not acceptable to all. Along the human history, many scholars offered different ratios to better reflect the balance between the interest to convict the guilty and the need to insure the acquittal of the innocent. To review the different opinions see: Alexander Volokh, Guilty Men, 146 U. Pa. L. Rev. 173 (1997). 67 See: Alschuler, The Prosecutor’s Role, p.60 68 See Schulhofer, pp. 1985-1986 69 See Dworkin, supra note 56 INNOCENCE EFFECT 28 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR plea.70 For these reasons, it is argued that the advantage of plea bargaining – ensuring that the innocent will receive penalty relief does not exceed the disadvantage associated with plea bargain – increasing the number of innocent defendants who are convicted. Subscribers to this view place a much higher moral value to the reduction in the rate of wrongful conviction than to the interest in reducing the sentences to those wrongfully convicted. It is morally preferable that a single innocent defendant will be convicted instead of ten innocent, even if the punishment inflicted on that single defendant will be more than ten times higher than that which would apply to the ten.71 Moreover: even if an imperfect trial might end in a wrongful conviction, it is important from a social perspective, that this result, though unjust, would follow a genuine attempt to get to the truth.72 However, if, as we show above, innocent defendants are uniquely averse to pleading guilty, then this moral concern is exaggerated. Prosecutors adjust their plea offers, based on the evidence and the expected trial sentence, to the average, and thus guilty, defendant. Thus most of the guilty defendants will accept the offers while the innocence effect will lead many innocent defendants to reject them. The end result therefore is that the vast majority of innocent defendants would opt for trial and that plea bargains will mainly lead to the conviction of the guilty ones. Moreover, in a system that relies on plea bargains, the rate of wrongful convictions is lower than in a system in which every defendant go through a full-fledged trial, for two reasons. First, most plea bargains ensure that the majority of the guilty defendants plead guilty and therefore are surely convicted, while many innocent at least gain the prospect of an acquittal in a trial. If it were not for plea bargaining some of the guilty defendants that plead guilty would have been acquitted in trial, and so the rate of the innocent among the convicted would have been higher.73 70 See: Schulhofer, p. 1986 See: Alschuler, The Changing Plea Bargaining Debate, p. 715 72 There, p. 686 73 It is important to stress that the principle of false convictions prevention puts the emphasis on the rate of false convictions and not on their absolute numbers. As implied in the statement made by Blackstone, the emphasis is put on the relationship between the interest in the conviction of guilty and the interest of acquitting the innocent, and not on the measurement of the last interest in an 71 INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 29 Second, plea bargains made with most defendants save recourses which are then used in the trials of the few defendants that contest the charges, allowing these trials to be more accurate and more protective to the innocent. If every criminal proceeding would lead to a full trial, resource constraints would force the system to lower the costs of trials in order to maintain a reasonable level of enforcement. In fact, historical studies show that plea bargains have evolved largely due to the construction of defenses’ privileges and rights such as the privilege against self incrimination and the right to representation.74 Even today one can see a distinct connection between expanding the protections granted to defendants and the developing want for plea bargaining in different countries.75 Plea bargains leave the vast majority of defendants out of court and allow the judiciary to ensure a complex and costly trial to those who choose a trial, including most of the innocent defendants. If every case went to trial, the trial process would have to be significantly simplified, and thus less accurate, leading to an increase in the rate of wrongful conviction.76 The critics of plea bargaining, who emphasize its consequences on innocent defendants, should be encouraged. In fact, they should support plea bargaining. Because of the innocence effect, plea bargaining actually reduce the rate of wrongful conviction. To the extent that the rate of wrongful conviction is the yardstick for the quality of the system, plea bargaining improve the criminal justice system. absolute manner. Obviously in larger jurisdictions (i.e. in jurisdictions that prosecute more defendants than others) there will be a higher absolute number of mistakes, but this does not prove that a legal system such as that provides less protection to the innocent than a smaller legal system. It is the rate – not the number – of wrongful convictions that matters. 74 See: John H. Langbein, Understanding the Short History of Plea Bargaining, 13 Law & Soc’y Rev. 261 (1979) 75 Thus, in Italy, the transition from the inquisitorial method to the adversarial method that gave the defendant several new legal rights at trial was accompanied by infiltration of plea bargains as a central component in the judicial system. See William T. Pizzi & Mariangela Montagna, The Battle to Establish an Adversarial Trial System in Italy, 25 Mich. J. Int’l L. 429 (2004). In Germany, the increasing complexity of trials was a significant catalyst in the development of plea bargains. See Máximo Langer, From Legal Transplants to Legal Translations: the Globalization of Plea-Bargaining and the Americanization Thesis in Criminal Procedure, 45 Harv. Int’l L.J. 1, 45 (2004). 76 See: Scott & Stuntz, Plea Bargaining as Contract, p. 1932 INNOCENCE EFFECT 30 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR B. “Insuring” the innocent Unlike the critics of plea bargains, proponents of plea bargaining tend to emphasize the interest of the innocent to minimize their damages.77 Plea bargains provide the innocent with an alternative to the costly and risky trial. These defendants may, therefore, prefer a plea bargain rather than an expensive procedure during which they may be detained or suffer other restrictions, and ultimately may face a harsh penalty. For those defendants, plea bargains are therefore a type of insurance that, in the eyes of many plea bargaining proponents, should not be revoked in the name of protecting innocent defendants. Disregarding the innocence effect, plea bargaining supporters emphasize the interests of innocent defendants in insurance. But if the innocence effect is taken into account, they too should reconsider their position. If innocent defendants are reluctant to plead guilty, then the main beneficiaries from the insurance are actually the guilty ones. The end result is that innocent defendants have a higher average sentence than that of the guilty defendants who are facing similar evidence and are charged with a similar offense. This phenomenon can be explained with this next example: Suppose that in all cases in which the chances of conviction are 80% and the predicted sentence is ten years imprisonment, the prosecution offers defendants a plea bargain resulting in a sentence of five years in prison (which reflects "a discount" of three years in prison in comparison to the expected trial sentence). Suppose that due to the innocence effect, innocence defendants reject that offers and the only guilty ones accept it. Each of the guilty defendants will be sentenced for exactly five years. 80% of the innocent defendants will be convicted at trial and sentenced to ten years in prison, but since 20% of them will be acquitted, the average sentence imposed on an innocent defendant is eight years in prison. In other words, innocent defendants are sentenced, on average, to longer sentences. However, in a legal system without the mechanism of plea bargains, the 77 See: Scott & Stuntz, Imperfect Bargains, p.2013 (explaining the controversy between them and Schulhofer saying “He attacks our analysis because we want innocent defendants to be able to plead, if they so choose, under the best possible terms. In Schulhofer’s view, they are better off going to trial, whether they think so or not”). Also see: Church, pp. 515-516 (critiquing pr. Alschuler’s objection to plea bargains with innocent); Bowers (claiming that plea bargains are especially beneficial to innocent defendants). INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 31 average sentence that would have been imposed on the innocent and the guilty under the circumstances in the example above would be identical.78 Moreover, it is likely that the example above also diminishes the severity of the problem. In order to ensure that 80% to 90% of the defendants agree to a plea bargain, the prosecution must offer discounts of more than 50% to defendants who have a 20% chance of acquittal.79 Perhaps an even more harmful effect occurs in cases where the plea bargain reached changes the type of sentence – from imprisonment sentence to intermediate sanction or probation.80 In these cases, the gap between the punishment imposed on those who 78 Of course one should assume that only few innocent will face 80% prospects of a conviction, seeing that, on average, cases against innocent are likely to be weaker; we will get similar results if we separately analyze each group of cases in which the penalty after a trial and the chances of conviction are similar. The end result is that in all the groups of defendants, the average penalty of the innocent would be higher than that of the guilty in a legal system that relies on plea bargains. 79 If all of the defendants would try to minimize only their punishment expectancy, the prosecution could settle for a sentence discount of slightly more then 20% thus ensuring that all defendants with 20% prospects of an acquittal would accept the offer. In practice, defendants are "averse to losses" and are willing to risk a higher expected sentence in order to preserve the chance of a full acquittal. See Oren Gazal-Ayal, Partial Ban on Plea Bargaining, 27 Cardozo L. Rev. 2295, 2338-2339 (2006). As a result, the prosecution must offer them a larger sentence discount. Also, defendants differ one from the other in their willingness to accept plea bargains, in light of idiosyncratic differences in their success rate assessments and their preferences regarding risk. In a system with budgetary constraints that allows the prosecution to bring to trial only 10% of the cases, the prosecution can not settle for proposals that fit the average defendant and therefore be accepted by about half of the defendants. Instead, the prosecution has to bid offers that the vast majority of defendants will accept. As a result, the prosecution is required to offer much more lenient sentences than the punishment expectancy at trial. In fact, many times the impact a plea bargain has on cases where the chances of an acquittal are high, is the end result of a complete waiver on actual jail sentence as expected after a judicial conviction, and its’ replacement with community service or probation. 80 For example, see the Tolya case brought in the last chapter, where all the defendants who were convicted in a standard procedure were sentenced to prison terms, while most of those who admitted guilt were sentenced to other penalties. Such arrangements are well known in Israel also. it is, of course, difficult to know in case of a plea bargain which penalty would have been imposed without the arrangement. For this reason it is difficult to empirically establish the claim that arrangements that converters prison term sentences to other punishments are common. However, there have been rare cases in which data showing the range of punishment after the trial and the punishment offered in the arrangement itself was revealed. INNOCENCE EFFECT 32 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR plead guilty (most of whom are guilty) and the average penalty imposed on those who plead not guilty (a group that suffers from over-representation of the innocent) is especially large, and accordingly the effect of plea bargains on innocent defendants is larger. One can of course argue that the innocent do not lose from plea bargains even if these arrangements are more profitable to the guilty. It can be argued that if the plea bargains would be canceled, then all defendants would lose, even if the guilty defendants, that are often the main beneficiaries of plea bargains, have more to lose. This argument ignores the effect plea bargains have on the general level of punishment. In legal systems where most of the defendants are convicted through plea bargains, the sentence imposed on these defendants is the standard sentence.81 To ensure congruence between the sentence and the gravity of the offense, the legislature, the prosecutors, and the judges have an interest to inflict harsher sentences on those who do not plead guilty.82 As a result, the sentences imposed on the few who choose to go to trial are heavier than those that would have been imposed if all defendants would have been sentenced after a full-fledged trial.83 Moreover, the overcrowded prison system could not accommodate so many new inmates with longer sentences and the economic pressure of the system probably lead to measures to reduce the number of inmates, and thus, shorten the actual sentences. Therefore, given the impact of innocence on the willingness to plea bargain, it may be that the innocent have an interest that plea bargaining is banned. C. Restrictions on plea bargains The conclusion of this section is that the arguments of both sides plea bargaining controversy are weakened when the innocence effect is taken into account. As a result of the innocence effect, plea bargains probably do not increase the rate of wrongful convictions, as argued by some of the plea bargaining opponents, but also do not promote the interests of innocent defendants as argued by the plea bargaining supporters. However, perhaps an in between solution – between the complete freedom to make plea bargains, as is the case 81 See: Gerard E. Lynch, Our Administrative System of Criminal Justice, 66 Fordham L. Rev. 2117, 2134 (1998) 82 For analysis of the interests of legislators and prosecutors, see: William J. Stuntz, The Pathological Politics of Criminal Law, 100 Mich. L. Rev. 505 (2001) 83 See: Alschuler, The Changing Plea Bargaining Debate, p. 689 INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 33 today, and a complete ban on plea bargaining - can advance both goals at the same time. One way to reduce the negative effect of plea bargaining on the sentences of wrongfully convicted defendants is to limit size of the plea discount. The law can instruct judges not to approve a plea bargain if the post plea sentence is significantly lower than the sentence expected after a jury trial conviction. If the prosecution’s ability to offer significant sentence discount in plea bargaining would be limited, defendants requiring considerable discount in exchange for their guilty plea would not bargain. Since the innocent demand, on average, significantly lighter sentences than guilty defendants, the proportion of innocent defendants among those who plead guilty will be smaller. A similar proposal was offered by one of us84 in order to discourage the prosecution from filing charges in weak cases, by limiting its ability to reach a plea bargain in those cases. The suggestion here emphasize a different end-result: by restricting the guilty plea discounts the law can reduce the number of innocent defendants who plead guilty, and their proportion among all guilty pleaders. Such a restriction would still allow the prosecution to reach plea bargains with most of the guilty defendants. Conversely, as a general rule, the restriction on the sentence discount will usually prevent plea bargains with innocence defendants who require a larger discount because of the innocence effect. Therefore, such a restriction can reduce the risk that innocent defendants would plead guilty, as some probably do today in exchange for very lenient sentences. The fewer innocent defendants pleading guilty will lead to fewer wrongful convictions. Reducing the rate of wrongful conviction is the main normative goal of one side of the debate. But a restricting on the guilty plea discount can also help reducing the sentences imposed on innocent defendants, which is the main concern of others. Restriction of the guilty plea discount will minimize the gap between the sentence imposed on those who opt for a trial, including relatively many innocent, and that of those who plead guilty, most of whom are guilty. The innocent defendants who will be convicted in jury trial will still be sentenced for longer terms than the guilty pleaders, yet, the difference will be smaller. 84 See: Gazal-Ayal, supra note 79 INNOCENCE EFFECT 34 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR How can one impose limits on the practice of plea bargaining? When it comes to sentence bargains, it is not a particularly difficult task. Courts can be instructed to reject plea bargains if the proposed sentence is substantially lower than that imposed in similar circumstances after a trial. Limiting charge bargaining is much harder, but it might still be a possible task.85 D. Agreements on a Simplified Criminal Process Another alternative to plea bargaining can be bargaining for a simplified criminal process. The law can encourage parties to replace plea bargains with agreement on shorter and simplified process in return for a sentence discount. Parties can agree on a bench trial instead of a jury trial,86 agree that the defendants will be the first to testify, before the case for the prosecution is presented, or make other stipulations which allow a cheaper trial for the prosecution. True, in such a simplified process, the risk of mistake might be larger. Yet, innocence defendants in such process can gain a discount for the agreement without pleading guilty – something which they are reluctant to do. Moreover, defendants in such simplified proceedings will maintain the right to contest the case, and, if successful, gain an acquitted, a right they waive in regular plea bargaining. Innocent defendants who may be reluctant to plead guilty, would often prefer a simplifies process in which they are not required to plead guilty, are able to avoid the damaging consequences of prolonged litigation and benefit a lenient sentence in exchange for saving judicial time. True, the defendant in such a simplified trial might have a smaller chance of acquittal. As a result, it is possible that defendants who would have been acquitted in a full trial may be convicted in a simplified trial. Nevertheless, it can be said in confidence that the advantages of the simplified procedure outweigh its disadvantages. First, the proposed procedure will promote the interest of the innocent defendants by allowing them to benefit from a real legal process which is aimed at finding the truth while benefiting from a sentence discount. Indeed, the cost of this procedure is higher than that of a guilty plea, but the gap between the two is not so big when it comes to substantially simplified process. Additionally, most of the 85 Gazal-Ayal, supra note 79, 2340. See Stephen J. Schulhofer, Is Plea Bargaining Inevitable?, 97 HARV. L. REV. 1037. 86 INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 35 guilty defendants who plead guilty today will continue to do so even when offered a simplified process is an option because in order to be acquitted in such a procedure, they would have to lie to the judge. In most cases that involve conclusive evidence, defendants will find it psychologically difficult to lie to a judge; mostly due to the fact that it probably will not help their case. As shown above, guilty defendants tend to plead guilty simply because they are guilty.87 Therefore most of those who plead guilty will continue to do so even when they can get a similar discount in a simplified process. Furthermore, some of the defendants, who in the absence of an alternative procedure would choose a full trial, would now choose the simplified procedure. As a result, judicial resources will be spared and could be channeled towards the implementation of the simplified process. In light of this, the simplified procedure could advance the interests of the innocent without the need for additional resources. Second, the procedure may also reduce wrongful convictions. Despite the reluctance of innocent defendants to plead guilty, some of them succumb to the desire to quickly end the proceedings or to the fear of a wrongful jury trial conviction, and thus plead guilty. The simplified procedure could provide these defendants an opportunity to challenge the charges against them. In light of all of this, we believe that the benefits of the proposed procedure, in view of innocent’s reluctance to plead guilty, outweigh its disadvantages. Moreover, even in cases where a person might be mistakenly convicted in a simplified process, the moral gravity of such conviction will be smaller because it will be reached after an honest judicial attempt to find the truth. Studies on procedural fairness even indicate that defendants who are not satisfied with the outcome of the trial tend to accept it if they were given an opportunity to argue their case before an impartial arbiter before the verdict. Simplified procedure might be better in allowing the defendant a save opportunity to present his story, than the adversarial jury trial.88 87 Reference During a Full trial the defendant may respond only when presenting the defense case after the prosecution rests. The defendant’s response is subjected to the rules of the direct examination and cross examination, and the defendant can not speak directly to the judge. In some cases, following the advice of the defense attorney, the defendant chooses to rest and not exercise his or her right to testify. Once the defendant’s testimony ended, he or she will not be given another chance to speak before the verdict. In general, the criminal law encourages defendants not to tell 88 INNOCENCE EFFECT 36 2/12/2016 OREN GAZAL-AYAL & AVISHALOM TOR CONCLUSION In this study we showed that the assumption in the plea bargaining literature, according to which a defendant's innocence has no direct bearing on the defendant’s decision whether to accept a plea bargain, is incorrect. Even when there are no differences in the likelihood of conviction and in the expected penalty at court in the framework of a plea bargain, innocent defendants are more likely to reject plea offers that guilty defendants would accept. Empirical research on the contributing factors to guilty pleas is still in its early stages, and the research concerning the effect of innocence on the willingness to accept plea bargains also has a long way ahead of it. In this respect, it is necessary to examine whether the type of the defendant’s innocence has an effect on his or her reluctance to enter a guilty plea; it is possible that defendants charged with a crime they did not commit react differently than defendants whose innocence stems from the absence of the required mental component or from the present existence of a legal defense to the charges. Another aspect that should be examined is whether the type of evidence gathered against the defendants has an effect on their willingness to plead guilty. In this context it should be noted that there are indications that defendants who falsely confessed to the police will also demonstrate greater tendency to plead guilty in court compared to defendants facing eye witness identification evidence or other evidence. Similarly, there is room for additional research to examine whether the effect of innocence depends on the strength of the evidence that the defendant is facing, given that there is initial evidence indicating that the gap between the willingness to plea bargain among the innocent and the guilty is reduced when the chances for conviction are high.89 Furthermore, one should examine whether the factors which affect the willingness to plead guilty - their story and actually choose to remain silent. See: Alexandra Natapoff, Speechless: The Silencing of Criminal Defendants, 80 N.Y.U.L. Rev. 1449 (2005) (describing and critiquing the incentives provided by the criminal justice system enabling the silencing of defendants). 89 One experimental study showed that the willingness to accept plea bargains, among the guilty, does not change when the chances for conviction rise from 50% to 90%, whereas among the innocent the difference is much more significant; see: Bordens, supra note 10, p. 67 . In a different study it was established that when the chances for conviction are extremely high, the difference in the willingness to accept a plea bargain, between the innocent and the guilty, vanishes; see: Tor, Gazal-Ayal & Garcia, supra note 5, part II.C.2. INNOCENCE EFFECT [2011] T H E I N N O C E N C E E F F E C T I N P L E A B A R G A I N I N G 2/12/2016 37 such as the time spent in pretrial detention90 or the pressure exerted by the defense attorney upon the defendant91 differently affect the guilty and the innocent. Hopefully, in the future, more studies will be conducted in this field in order to examine the degree of reluctance to plead guilty among innocent defendants as well as the factors that affect the decision whether to accept a plea bargain. The information and findings of such studies would equip the criminal justice system with tools to improve its ability to cope with the innocence problem. However, it is already clear that the extent to which the legal literature relies on the shadow model is excessive. The existing research is enough to show that the innocence of a defendant, as such, affects his or her willingness to accept a plea bargain - beyond the influence of the chances of conviction and the punishment in store for them if convicted at trial. Hence, apart from its direct implications, the research on the impact of innocence illustrates the need to further develop the available empirical database in order to examine the behavior of parties engaged in a plea bargaining process, which is currently the leading process responsible to the vast majority of criminal convictions. 90 Series of studies have shown that detention affects the willingness of defendants to plead guilty. See, for example: William M. Landes Legality and Reality: Some Evidence on Criminal Procedure, 3 J. Legal Stud. 287 (1974); Gail Kellough & Scot Wortley, Remand for Plea: Bail Decisions and Plea-bargaining as Commensurate Decisions, 42 Brit. J. Criminology 186 (2002). When the defendant is in detention, he or she has greater incentives to quickly end the proceedings. In a field study that examined inconsistent admissions among women prisoners, it was shown that the desire to end the pre-trial detention was classified as one of the main reasons for accepting a plea bargain, see: Suzanne Dell, Silent in Court: The Legal Representation of Women who Went to Prison 31-32 (1971). Even without being arrested, many defendants would be willing to admit in the framework of a plea bargain in order to save themselves the ongoing legal proceedings; see: Malcolm M. Feeley, The Process is the Punishment: Handling Cases in a Lower Criminal Court (1992). 91 For a review of the different ways attorneys influence defendants’ decisions and actions, see: Alschuler, The Defense Attorney’s Role. an extensive field study of this issue showed that a substantial number of defendants that refused to admit guilt changed their minds at the last minute following their attorney’s advice; see: John Baldwin & Michael McConville, Negotiated Justice: Pressures To Plead Guilty (1977).