Sentencing and Plea Bargaining: Guilty Pleas versus Trial Verdicts

In the daily work of the criminal justice process, the relationship between plea decision-making and sentencing is one of the most important. Meanwhile, in the academic and policy literatures it is one of the most controversial. This essay examines the moral arguments for and against the practice of altering a sentence as a consequence of a plea of not guilty or guilty. It also appraises the state of international empirical knowledge about the practice and raises questions for future research.

It is widely believed that a defendant who pleads guilty is likely to receive a reduced sentence if she pleads guilty than if she is found guilty, as a result of a trial, of exactly the same charges. Various terms are used to describe this practice (e.g., sentence reduction, guilty plea discount, sentence deduction, allowance in respect of a guilty plea, trial penalty, etc.). However, as discussed below, these terms are value laden and imply different normative positions. Thus, in the interests of neutrality, this essay will generally use the term “plea-dependent sentence differential” (or “sentence differential” for short). The foundation for the sentence differential varies depending on the jurisdiction. In some jurisdictions, statue or case law may permit, or even require, a sentence differential. In others, the basis of the sentence differential is found in guideline schemes. In certain jurisdictions the sentence differential may have little formal basis and is routed in the informal practices of sentencers. However, regardless of its basis, in most criminal justice systems there is a widely held perception that there is a sentence differential. This essay will provide an overview of the main issues and debates resulting from the sentence differential. In doing so, it also raises some provocative questions and suggestions for future research, not least the pressing need for a fuller appreciation of defendants’ perspectives and decision-making.

Yet, before going further, it is worth clarifying some preliminary matters. While procedures vary between jurisdictions, in general there are two broad ways through which a sentence differential may result. The first way is direct. It involves the judge altering the sentence on the basis of the plea entered. Alternatively, a sentence differential can occur indirectly. For example, there may be other plea-bargaining practices such as charge bargaining where an agreement is reached that charges will be amended in return for a guilty plea. These practices are indirect because they do not directly concern the sentence. However, often the defendant hopes that they lead to a reduced (discounted) sentence. For clarity, while there is some crossover (not least since they are both forms of plea bargaining), this essay mainly focuses on the direct ways a sentence differential can occur. Additionally, it is worth clarifying that this essay focuses mainly on adversarial legal systems. This is not to imply that the broad idea of the sentence differential is completely alien to inquisitorial systems. In the past it has been asserted that inquisitorial justice systems, such as those in continental Europe, did not engage in any settlement practices that bore any functional resemblance to plea bargaining (for example, see Langbein 1979). While it is true that typically such systems cannot engage in plea bargaining, as there are no “pleas” and all cases go to trial, defendants can (and do) make confessions or admissions of guilt that are “rewarded.”¹ These practices amount to what some have argued is a functional equivalent to plea bargaining in inquisitorial justice systems (see Thaman 2010 for an overview).

The essay is organized as follows. Section II explores the concern that a sentence differential violates the presumption of innocence. It explores the criticism that the sentence differential operates to penalize those who continue to plead not guilty by imposing (or threatening to impose) a higher sentence than if they plead guilty. Section III explores the criticism that the sentence differential may have disparate impacts on different groups (specifically minorities and those who are socially and economically disadvantaged). Section IV explores why, in light of the dangers to principled sentencing and liberal rule of law values, justice systems apparently continue to persist with guilty plea discounts. Penultimately, section V questions the state of empirical knowledge regarding the sentence differential. Finally, section VI highlights some further questions that future research should explore.

In Anglo-American justice systems the sentence differential is widely considered a “sentence discount” or “reduction” that rewards a guilty plea. However, this view is not universal. For instance, Fiona Leverick acknowledges that there is a “principled objection” (2004, p. 382) to be made against guilty plea discounts as they amount to a punishment for those accused who exercise their right to trial. Similarly, Penny Darbyshire is critical of the argument that guilty plea discounts are indeed “discounts”:

The discount undeniably punishes those who exercise their right to trial then are found guilty, however much the Court of Appeal tries to disguise a Sentence Discount as a reward for remorse. This is stunning hypocrisy in the Anglo-American legal systems, whose rhetoric trumpets the right to trial, especially jury trial, the burden of proof and the presumption of innocence as the hallmarks of the world’s finest democracies.

(Darbyshire 2000, p. 901)

Those who consider the sentence differential as a violation of the right to be presumed innocent sometimes refer to it as a “trial tax” or “trial penalty.” Consequently, terms such as “discount” and “penalty” are value laden and connote a very different opinion (which is why the more neutral term “sentence differential” is used here). The values reflected in these terms are important as they relate to a central tenant of the liberal rule of law: its suspicion of the state, and its desire to protect individuals from its power. It is these values that have helped form an ideological foundation for the presumption of innocence. By potentially inducing guilty pleas, which bypass the trial and arguably undermine the presumption of innocence, a fundamental criticism can be made of the sentence differential. Some have gone further and suggested that large sentence differentials can be coercive (Caldwell 2011; Ashworth 2006, pp. 256–257; McCoy 2005). For example, McCoy (2005, p. 90) argues that the magnitude of the trial penalty is so severe that we must ask “whether it amounts to institutionalized coercion.” She argues:

Implicit plea bargains occur for two reasons: (1) the defendant knows the “going rate” of punishment and can accept it, and (2) for some defendants who believe they have valid legal defences, the threat of trial penalty will cause them to plead guilty anyway. There is nothing wrong with pleading guilty in the expectation of receiving a “going rate” of punishment. But … there is a lot wrong with pleading guilty if that going rate after trial is so huge as to be the reason a defendant will make a pre-emptive guilty plea.

(McCoy 2005, p. 94, emphasis added)²

Thus, two issues can be identified. The first is that the infringement on the presumption of innocence is wrong in principle, even if the defendant is guilty: the guilty are entitled to be presumed innocent and put the state to proof. The second issue is whether the sentence differential can potentially induce innocent defendants to plead guilty, thus resulting in wrongful convictions.

To some the second issue might seem farfetched. One might suppose that the risk of defendants incriminating themselves by pleading guilty to charges of which they are innocent is merely hypothetical. Surely, the supposition runs, no sane person would plead guilty to something they have not done: defendants choose rationally and freely how to plead. Moreover, one may expect that this would not happen because defendants are represented by skilled lawyers fearlessly representing the best interests of their clients and holding steadfastly to cherished values, such as the presumption of innocence. However, while this may be true of some defendants, there is evidence that the sentence differential may contribute to innocent defendants pleading guilty (Garrett 2011). Why this might occur in any individual case is a complex question, but, in general, there are at least five reasons to doubt the supposition that it is implausible that defendants do not plead guilty to charges of which they are not guilty.

First, the literature on the relationship between criminal defense lawyers and their clients shows clearly that while in theory defendants “instruct” their lawyers, the reality is more complex. Empirical research “has consistently highlighted the relative passivity of most clients” (Tata and Stephen 2007, p. 732; see also Carlen 1976; McConville et al. 1994, Ericson and Baranek 1982). While in most contexts the professional-client relationship is characterized by an inequality of power, this is especially acute in the context of most criminal cases. The weak educational, social, and personal resources of most (though not all) defendants can severely limit defendants’ agency. Additionally, these issues can be exacerbated by the immediate stress and anxiety of being subject to criminal charges (especially while held in pretrial detention); the unfamiliar vernacular of the legal system; and the requirements of criminal and court procedures (which can vary by court in some jurisdictions). As a result, it can be difficult for clients to take firm command of their own cases. Indeed, all this means that defendants may not always fully understand the charges against them, and some may plead guilty to charges they do not fully understand. Research is urgently needed to explore more fully what defendants do and do not understand, but there is reason to question assumptions that guilty pleas are always fully informed.

Secondly, research has shown that the criminal process in the lower and intermediate courts tends to be characterized by gearing toward guilty pleas. There is a widespread expectation among court personnel that an admission of guilt by the defendant is inevitable (e.g., McBarnet 1981; Carlen 1976; Feeley 1982; Heumann 1978; Tata 2007; 2010). This expectation can become self-perpetuating in very practical ways. For example, court schedules depend on a high volume of guilty pleas to assist expeditious processing of cases.

Thirdly, while lawyers may endeavor to look after the best interests of their clients, it is now well established that lawyers also have to be cognizant of a range of potentially competing imperatives, including longer term inter-and intraprofessional relationships (e.g., Eisenstein and Jacob 1977; Heumann 1978; Tata 2007) and the structuring of financial interests (Tata 2007). This is not to say that dedicated people such as lawyers simply discard cherished values for career or financial gain: lawyer behavior appears to be more complex than this. It is, however, fair to say that there are a range of competing dynamics that they have to find some way to resolve and which therefore have a bearing on client plea decision-making (Tata 2007).

Fourthly, some defendants, even where they do not accept the prosecution’s case, feel unable to challenge it effectively. For example, defendants may be charged with a crime committed when they were under the influence of alcohol or drugs and may have only a hazy memory of events, thus making them less willing to deny the charges (Tata 2010). Others may not trust the system or feel their denial of guilt is unlikely to be believed.

Fifthly, defendants may make a deliberate choice to plead guilty to charges of which they are not guilty. Prima facia this would seem to conflict with theories of decision-making that posit the defendant as a rational actor who would not wish to incriminate him- or herself. However, some defendants may feel obliged to plead guilty so that another family member, friend, or gang member avoids conviction. Other defendants may distrust the criminal justice system and doubt that the court will believe their denial and so conclude that a plea of guilty is the least bad option. Indeed, pleading guilty, even when innocent, can be considered a better option due to the costs which the criminal process inflicts (Feeley 1982). For example, Albert Alschuler has noted:

A misdemeanor defendant, even if innocent, usually is well advised to … plead guilty at the earliest opportunity … to minimize the painful consequences of criminal proceedings. (1983, p. 953)

This is especially true where the defendant is held in pretrial detention (remand), and the sentence could be back-dated to cover, wholly or in part, time served; or where the defendant is held in remand but the potential sentence is noncustodial. In both cases, the effect of a guilty plea can be immediate liberation. Additionally, the difference in the potential punishment if an accused goes to trial can be exorbitant. For example, an accused may face different charges if going to trial, and these could have a much higher minimum sentence—such as has occurred where going to trial means being charged under a “three-strikes” law. In such circumstances, where there is a perception of a sizable trial tax, an individual may feel coerced and/or choose to plead guilty even if they are innocent and the chances of acquittal are relatively high: paying a small but certain cost to avoid the risk of a larger cost if convicted following a trial. From the perspective of such defendants, this is a meaningful choice. Additionally, when considering what an accused may choose to do it is important to recognize that the decision how to plead is not only shaped by the relative strength of the charges but also by other “extra-legal” factors including pressing social and family considerations.

It might be argued that in practice the distinction between a “discount” for pleading guilty and a “penalty” for conviction at trial is semantic. Both views are based on the same facts, and regardless of whether it is argued to be the former or the latter, the defendant still suffers the same fate (Leverick 2004, p. 383). However, as suggested, the distinction raises issues of principle crucial to any criminal system that claims to be just. Consequently, this difficult question needs to be tackled.

So how should the sentence differential be characterized? Much depends upon what one sees as the “default” or “baseline sentence.” Terms such as “reduction” and “discount” imply that the “baseline sentence” is that which would have been passed if the accused pled not guilty and was convicted at trial of the same charges. Thus, from this perspective, the post-trial sentence is the baseline sentence, and the sentence differential is considered a discount that is deducted from that baseline sentence. Indeed, this perspective seems logical and in line with everyday notions of discounts: the baseline sentence is the equivalent of a “recommended retail price,” and if less is paid, then one has saved money.

However, not all agree with the supposition that the post-trial sentence be considered the baseline in practice. To continue the analogy, if most transactions are conducted at less than the recommended retail price, because the item is perpetually “discounted,” is the consumer really getting a “discount,” or has the baseline price been inflated to give the appearance of a “discount”? Lynch has argued:

In a system where ninety percent or more of cases end in a negotiated disposition, it is unclear why the “discounted” punishment imposed in that ninety percent of cases should not rather be considered the norm.

(Lynch 2003, p. 1401)

While this issue has plagued some regulatory bodies that aim to ensure consumers are protected, no definitive resolution has been found regarding when the “discounted price” becomes the baseline. Interestingly, the possibility that the baseline sentence may change has been considered in England and Wales. In one instance, sentencing guidelines were locally modified to assume a timely guilty plea, with the sentence in effect increasing where there was a late guilty plea or no guilty plea (Henham 2000, p. 439). Using guidelines in this way was normatively troubling as it came closer than before to accepting formally that the sentence differential is in reality a penalty for going to trial. However, while this potentially exposed the sentence differential, and the justice system, to criticism it did have advantages.

One notable advantage is that this approach could be considered more straightforward and practical. The usual alternative is a guideline or norm based on a first-time offender who is convicted following a trial. However, in most jurisdictions this is far from typical. For example, it has been noted that in England and Wales only 12 percent of offenders plead not guilty, “and only 10 percent appeared for sentencing without any prior convictions” (Ashworth and Roberts 2013, p. 7). Consequently, the percentage of those sentenced that meet both these conditions is “obviously much smaller than 10 percent” (Ashworth and Roberts 2013, p. 7). Thus, a criticism can be made of guidelines and norms that do not match the typical reality of most cases.

However, despite this, viewing the sentence differential as a discount is not unfounded (certainly it is by far the most common perspective in law), but it is highly questionable. Perhaps in the past when trials were shorter; for example, in England it was once normal for a single judge to hear between 12 and 20 cases a day, and guilty pleas were uncommon (Alschuler 1979, p. 9), it therefore may have been less controversial to consider the sentence differential a discount. However, given that sentences following a guilty plea now account for over 90 percent of all sentences in adversarial systems, this argument has become more tenuous. Yet, while this challenges the notion that the sentence differential is a discount, ultimately it may be impossible to settle on an objective answer to the question. Discounts and penalties, like notions of “gain” and “loss,” “are malleable concepts” (Bibas 2004, p. 2512). This means there may always be room for different views. Consequently, what may matter more, and perhaps be the question that should be asked, is what the defendant believes the sentence differential to be. While there is a growing interest in “user perspectives” of the justice system generally, so far there is only limited knowledge about defendant perspectives—a situation which we argue later needs urgently to be addressed.

One criticism of sentence differentials is that they may have a disproportionally negative effect on some minority groups, thereby making the practice indirectly discriminatory and undermining the principle of legal equality. However, identifying indirect discriminatory practices is a challenging task as “a wide variety of practices that are facially neutral … have racially disparate effects” (Tonry 2012, p. 87). For example, drug laws penalizing the use of crack cocaine one hundred times more severely than powdered cocaine seem prima facia racially neutral. However, if one group is more likely to use the heavily penalized drug, then these laws can have a disparate impact (Tonry 2011, p. 1).

One possible way the sentence differential may indirectly discriminate against minorities is if, for example, “black defendants are in a worse bargaining position than white ones, and that this differential bargaining power makes black defendants more likely to make worse bargains than similarly situated white defendants” (Savitsky 2012, p. 135). However, it also possible that inequality could result if minority groups are less likely to plead guilty. Indeed, Henham noted:

Research suggests that the fact ethnic minority offenders were more likely to contest the charges against them inadvertently subjected them to a form of indirect discrimination since the system encouraged sentence discounts for guilty pleas. (2001, p. 4)

One factor that may influence whether a defendant pleads guilty or contests the charges is their level of trust in the justice system and in legal actors. This is problematic as there is evidence that certain ethnic minority groups have less faith in the criminal justice system than majority groups (e.g., Shute, Hood, and Seemungal 2005). Such mistrust may be demonstrated by the higher proportion of black defendants in England who plead not guilty and choose to contest their case before a jury of their peers rather than by a judge sitting alone (Hood 1992, p. 196; Thomas 2010, p. 21). Furthermore, Michael Tonry has noted that “black defendants less often plead guilty and when they do, they do it later” (2012, p. 75).

These apparent differences in intergroup pleading behaviors are extremely significant in justice systems where pleading decisions may dramatically affect sentences. For example, if a “guilty plea discount” is one-third, then by not pleading guilty a defendant’s sentence will be 50 percent higher than a comparable defendant who pleads guilty. Indeed, Hood’s study suggested that pleading not guilty accounted for 13 percent of the total difference “in the black male general population and their proportion among those serving sentences” (1992, p. 203).³ Thus, lower levels of trust may mean that guilty plea discounts indirectly contribute to racial disparities in criminal justice systems.

Why some minorities have lower levels of trust is a complex question largely beyond the scope of this essay. However, one factor may be that some groups are more likely to feel victimized by agents of the criminal justice system. For example, minority groups may be targeted in various ways, such as through “racial profiling” (Tator and Henry 2006; Welch 2007). The result of this is that “put into a single sentence: young black men [and others] who believe themselves unfairly treated by the police understandably become angry and uncooperative, and are punished more severely as a result” (Tonry 2012, p. 74).

However, ultimately, understanding and empirically testing the potentially discriminatory effects of the sentence differential is not easy. There are many variables to consider (e.g., intersectionality between race, class, and gender⁴) and information is limited: particularly regarding defendants’ subjective perceptions and how these influence pleading decisions. Though, for now it seems safe to suggest that, while ostensibly neutral, the sentence differential may produce “harsher punishments” for black and ethnic minority defendants (Tonry 2012, p. 87). Unfortunately, while there has been recognition of the problem of indirect discrimination, no solution is forthcoming. Some jurisdictions have attempted to limit judicial discretion in sentencing in the hope that this will reduce the potential for racial disparity; for example, this was supposed to be one advantage of the Federal Sentencing Guidelines in the United States (Spohn 2008). However, such attempts to limit discretion have not been entirely successful. Firstly, if the content of guidelines is itself indirectly discriminatory, then little can be achieved. Secondly, while sentencing guidelines may limit discretion to some extent, this discretion may be displaced rather than eliminated (Baldwin and Hawkins 1984, p. 582). Indeed, one potential consequence of reduced judicial discretion can be increased prosecutorial discretion. This may be problematic as “prosecutorial discretion may be more erratic and harder to contain than judicial discretion” (Provine 1998, p. 831).

Consequently, without removing the sentence differential altogether, there is no clear way to prevent its potentially disparate impacts. However, it may be argued that the disparate effects of sentence differentials are not the problem but the symptom of a problem. For example, might the fundamental issue be related to why certain groups may mistrust the justice system, rather than how this mistrust manifests to their disadvantage? If this is correct, and the disparate impact of the sentence differential is the symptom, then even if sentence differentials were abolished the underlying systemic causes of the disparate impact would remain. However, regardless, this is yet another reason for more research into defendants’ perceptions about the sentence differential and wider criminal process within the context of their lived experience.

So far this essay has discussed a number of criticisms relating to the sentence differential: it undermines the presumption of innocence, it may have a racially disparate impact, and so forth. Given these issues, one might wonder why the sentence differential is tolerated or sometimes encouraged. Three main rationales are advanced to justify the sentence differential.

The first rationale is the remorse rationale. This supposes that guilty pleas demonstrate remorse and that this remorse warrants the reduction. The second rationale is the victim rationale. This supposes that reductions are justified on the basis that the guilty pleas spare victims from the ordeal of a trial. Finally, this essay considers the efficiency rationale. This supposes that reductions are justified because they induce guilty pleas, or at least earlier guilty pleas, thereby saving resources, such as court time and money.

Using remorse to justify sentence discounting relies on the presupposition that a contrite offender is less deserving of punishment or that the contrite offender is more worthy of mercy.⁵ It also relies on the assumption that the guilty plea is a sign of remorse. However, there are problems with this rationale in both practice and theory. The first problem lies in identifying genuine remorse. The second problem is explaining why, all things being equal, remorse justifies a lesser sentence. These are dealt with in turn.

Perhaps the biggest problem with the remorse rationale is that identifying genuine remorse is difficult (e.g., Bandes 2016; Leverick 2004, pp. 370–372). It might be assumed that the guilty plea itself is evidence of remorse. Ironically, however, the perception of a sentence differential means that this argument is weak. As long as defendants perceive that they can benefit from a guilty plea, it cannot be known whether a guilty plea evidences remorse, a tactical decision to try to benefit from a discount, or both. Additionally, even if the sentence differential were not a factor, there are many other reasons why a defendant might plead guilty (e.g., to avoid the stress and uncertainty of a trial, to be liberated from pretrial detention, etc.). This difficulty in distinguishing remorse from tactical maneuvering may be a reason not to rely on it is as a basis for granting guilty plea discounts. There is a risk that some defendants tendering guilty pleas for tactical reasons will be thought of as remorseful and benefit from a discount, while some pleading guilty because of genuine remorse will be thought of as disingenuous and not benefit. If this were to happen, it would be far from fair.

Aside from the difficulty of identifying remorse, it can be asked why remorse should justify a reduction in sentence. Hannah Maslen and Julian V. Roberts argue that “desert theory provides the primary theoretical basis for sentencing guidance” (2013, p. 125). If this is correct, the remorse rationale is questionable as it does not affect culpability or harm: remorse occurs after the fact of the offense (Leverick 2004, p. 370).⁶ Thus, considering remorse does not affect what the defendant did nor how blameworthy they were,⁷ it is hard to find a principled reason that explains why such a defendant should benefit from a reduction in sentence. Indeed, as suggested earlier, the sentence differential may undermine principled sentencing.

However, despite the normative problems of rewarding defendants for remorse create, there appears some public support for the idea that remorse is important and that it can provide an acceptable reason for a reduced sentence (Maslen and Roberts 2013, p. 124). Remorse can also be an important factor for policymakers as well. For example, in Scandinavia, reductions for admissions of guilt have been “rationalized in terms of contrition or acceptance of responsibility” (Tonry and Lappi-Seppälä 2011, pp. 16–17); meanwhile in the United States, discounts for “acceptance of responsibility” are also premised on a remorse rationale. Thus, a sentence differential (especially a large differential) because an offender is remorseful is hard to logically justify in terms of penal principle, remorse does have some enduring appeal.

Another rationale for sentence differentials asserts that this practice benefits victims of crime. In essence, the rationale is based on the assumption that a victim will be best served by a guilty plea because this means that he or she will not be subjected to the trial process. Indeed, it is true that a trial can be arduous for a victim in that, inter alia, it can require the victim to be cross-examined and have their integrity questioned—potentially a difficult experience some victims prefer to avoid (Dawes et al. 2011, para. 3.2).

However, while sparing victims further ordeal may appear to be kind, there are reasons to question to this rationale. One reason to object is that in many cases where a reduction is given, there is no victim and the only witnesses are police officers (page 2010, para. 7.1.2). In these cases, the victim rationale does not apply. Secondly, it is wrong to stereotype victims. Just as defendants’ circumstances vary, so do those of victims. These variations are significant as there are victims who are not well served by sentence discounting (Darbyshire 2000, p. 905). For example, some victims may prefer a contested trial if this means they will learn more about the offense and the offender—which can help them obtain closure. Additionally, some victims may resent being “spared” for the simple reason that the cost of this “kindness” is that the defendant who wronged them receives a lesser sentence; although this risk must be weighed against the increased odds that an offender will not be convicted without a guilty plea.

Thus, while “sparing” victims is a noble endeavour, these good intentions do not always have positive results. While the victim rationale seems plausible, it does not apply to many cases where a discount is given: as there is either no victim or the victim is not one who would benefit from being “spared.” Additionally, there is something troubling about the criminal process when it becomes so burdensome on victims that they need to be protected from it, potentially at the expense of penal principles. Indeed, Leverick has criticized the victim rationale for focusing attention on avoiding flawed criminal processes rather than improving them (Leverick 2004, p. 374).

A related drawback with the practice of not hearing from the victim at trial goes well beyond victims’ interests. It relates to the function of the criminal trial as a public communicative forum playing out the nature and limits of social mores. In this sense, a trial is not solely for the disposal of cases concerning the interests of its immediate participants, but it also serves as a forum for wider public discussion and debate. Through public displays of emotion, ritual, and drama, the trial may demonstrate the polity’s commitment to justice. It plays an essential cathartic role for victims and the public, which is increasingly encouraged to identify with “the victim” (Sparks 2011). In so doing, in daily practice and especially high-profile cases, the trial enables a moment of collective moral expiation that in turn helps constitute a sense of community. In this way, the phenomenon of the vanishing trial not only silences victims and defendants, but it also denies the public from hearing their stories and for the law to show publicly that it has listened to them before coming to a decision. The decreased incidence of trials entails the loss of public displays of emotion, drama, and ritual, in turn perpetuating a feeling that criminal justice has become a sterile automated process devoid of moral drama and meaning, an effect with tends to undermines public confidence in the legitimacy of sentencing and the wider criminal justice system (Tait 2002).

The efficiency rationale is the prevailing basis on which the sentence differential is justified. In essence, the argument typically put forward is that while trials may be the ideal way to safeguard the presumption of innocence, resource constraints dictate that this ideal cannot be achieved in most cases. As a consequence, it is argued that it is necessary to incentivise guilty pleas (or at least earlier guilty pleas) with sentence discounts. Thus, the efficiency rationale is based on a claim of pragmatic necessity: inferring that without guilty pleas the system would collapse under the increased workload. Prima facia this argument is persuasive. If the justice system were to fail, the consequences would be significant. Additionally, it is important to note that this argument, properly put, is not amoral. Since the justice system runs on public money, there is a moral duty to ensure it is spent wisely. Thus, the efficiency rationale has a moral (utilitarian) foundation, based on an interpretation of “the greater good.”

However, is this claim that plea-dependent sentence differentials save money factually accurate? While removing the sentence differential may result in fewer (and later) guilty pleas, it should again be remembered there are several other reasons why defendants plead guilty (Schulhofer 1984, p. 1040; Feeley 1979). This means it should not be assumed that all, or even most, guilty pleas are in fact the result of the sentence differential.

Moreover, a number of commentators have suggested that the caseload necessity thesis is not as certain as may be assumed. For example, Weigend notes that it

has little evidence to support it—the time and location of system changes from trial to non-trial adjudication are not related to significant increases in case input. This is especially true for countries where the introduction of bargained case dispositions did not result from overburdened courts’ search for an outlet but was the product of comprehensive legislative reform. (2006, p. 213)

Indeed, from both international and historical perspectives the empirical link between the rise of the sentence differential (and plea bargaining more generally) and level of workload is at best tenuous (e.g., Eisenstein and Jacob 1977; Feeley 1979, 1982; Heumann 1975, 1978; Mather 1979; Vogel 2007).⁸ By comparing low and high court volumes and across time, it has been found that a remarkably similar proportions of cases result in a guilty plea. McCoy sums up the point neatly:

The finding that a high percentage of cases conclude with guilty pleas even when there is very little caseload pressure undermines the conventional wisdom that explains plea bargaining in terms of efficiency. (1983: 59)

Ironically, the expectation, encouraged by the high profile of the sentence differential, that cases will invariably settle by way of guilty pleas or be dropped can itself lead to wasteful “churn” and delay in court hearings. For example, prosecutors may tend to be more improvident in how they libel charges than they would if they expected a trial, in the partial expectation that those charges will later become useful bargaining chips (Caldwell 2011, p. 65). Defense lawyers may respond to this by delaying settlement until trial, which may be “the best time to put on the screws” and get a better deal from a pressured prosecutor (Tata 2007, p. 512).

Indeed, in part because of the environment propagated by sentence discounting, prosecutors and defense lawyers tend to avoid a thorough examination of the evidence at the earliest opportunity, instead expecting a settlement later in the process or a withdrawal of the case (Tata and Stephen 2007; Bradshaw et al. 2012, para 6.14). This, of course, is self-perpetuating. All this leads to a lack of preparedness and an “adjournments culture” (Kemp 2008), which from the perspective of individual practitioners is rational and prudent, but from a systemwide perspective is inefficient. Thus, the sentence differential may work to normalize an inefficient culture and contribute to “phenomena aimed at delaying the progress of cases” (McInnes 2004, chap. 28).⁹

Additionally, there can be a tendency to assume that processes encouraging guilty pleas don’t cost anything. However, while such costs may be hard to quantify they do exist, and it is incorrect to assume that they are “trivial” (Schulhofer 1984, p. 1040). For example, guilty pleas may be seen as saving resources, but in reality there are costs, for example lawyers have to spend time persuading defendants to plead guilty. Furthermore, even seemingly minor delays related to guilty pleas have consequences (these delays may be the result of “cracked trials,”¹⁰ or delays resulting from cases unexpectedly going to trial). Schulhofer noted that in his sample various delays meant that the average case involved 35 minutes of waiting, and the “average guilty plea proceeding in fact required 55 minutes of courtroom time for the conviction stage alone” (1984, pp. 1056–1057). While these figures do not take account of the entirety of the potential culture of inefficiency that results from practices designed to promote guilty pleas, they do suggest that the costs might be more significant than generally assumed. Moreover, it is worth noting that even if the removal of the perception of a sentence differential did increase the workload of the justice system, it might adapt rather than grind to a halt. Some have suggested that a simplified trial procedure could be used, while others have argued that a total ban on guilty plea discounts may be unsustainable but a partial ban could work (Schulhofer 1984; Alschuler 1983; Gazal-Ayal 2005). Thus, while the necessity argument cannot be dispelled completely, equally, it should not automatically be believed to be true.

Finally, the widely cited concept of “efficiency” bears scrutiny. Efficiency is, of course, a laudable goal with which no one can reasonably disagree. However, implicit in the debate about the sentencing differential is profound disagreement about what counts as “efficient.” Measuring “efficiency” in the justice system is not a value-free exercise. It implies normative claims about what is “just,” what is “necessary,” and what is “wasteful” based on different and competing perspectives about the “correct” role of prosecution, defense, judge, and, indeed, the justice of the criminal process as a whole (Tata 2007). It is easy to imagine that “efficiency” is about doing things more productively. The prevailing image is the factory assembly line. Increased production of the desired output with no increase in input (e.g., labor, outlays, etc.) equals greater productivity. When the output is shoes or widgets, it is fairly clear and easy to measure. Yet, what should be the desired output of criminal justice? Is it the sheer number of cases processed? Surely, the output of a justice system must be justice. Therefore, an increase in case disposals that largely result in injustice is, by definition, self-defeating and therefore inefficient. In other words, when even the cheapest justice system produces injustice it fails to be efficient. As an analogy, you can drive quickly, but it cannot be efficient unless you are actually heading to your destination. In the same way, injustice can only ever be a mark of inefficiency: the system has failed to produce what it should. In other words, logically, it cannot be enough to justify the sentence differential on the grounds of cost alone. This is not to deny that the sentence differential may dispose of cases in a way that also dispenses justice, but the link between the two cannot be assumed: it is tenuous and contingent.

The concept of a sentence reduction resulting from a guilty plea (or its functional equivalent) is well-known. Even inquisitorial jurisdictions understand the concept of negotiated justice, and many systematically reward “confessions” through practices somewhat akin to plea bargaining (Rauxloh 2010; Langer 2004). However, a concept is not the same as the empirical reality of a practice. Consequently, this section focuses on the empirical questions regarding the sentence differential. In particular, this section aims to highlight that the reality of the sentence differential may not be as it is commonly assumed and note that there can be debate over whether there is a sentence differential. This is done by highlighting some of the difficulties in defining and measuring the sentence differential and some research that may challenge certain assumptions.

At first glance this empirical question may seem strange given that it is often suggested (in case law, statues, or guidelines depending on the jurisdiction) that guilty pleas result in a lower sentence, and that defendants may believe and be told they received a “discount.” However, just because the law suggests there should be a discount, and because defendants believe there is a discount, does not necessarily mean that this is the case. Nor does it assuage concerns that sentences may be inflated prior to the application of a discount in order to negate the effect of any apparent reduction (Chalmers et al. 2007, para 6.26). Thus, the empirical evidence needs to be scrutinized.

However, empirically verifying whether the sentence differential materializes is not a simple task. The first issue is conceptual. It is crucial that empirical studies are clear about exactly what is (and is not) being examined. For example, in US federal cases, Andrew Kim notes that some studies exclude the “acceptance of responsibility” discount for a guilty plea while others do not (2015, p. 1200). Given that this “acceptance of responsibility” feature can significantly reduce a sentence (Kim suggests by between 25 percent to 35 percent), its inclusion or exclusion should be noted. Likewise, other differences between prima facia similar studies can result in each measuring something differently under the same or similar headings. Abrams sets out what he means by the term “trial tax,” though this definition is not universally employed by all researchers. (This does not, of course, make Abrams or any other researcher wrong, but what it does underline is that to a considerable extent what and how researchers study is also driven by their normative values and beliefs about justice.) The second issue is the ability to identify and quantify any sentence differential. This requires a means of comparing otherwise equal cohorts of cases where the only variation is plea. To achieve this comparison demands a way of conceiving and representing the sentencing decision-making process (Tata 1997; 2007). How to understand and measure the exercise of discretionary decision-making is widely debated. For instance, should one take a behaviorist perspective in which certain stimuli (factors, variables) are thought to illicit predictable responses in the decision-maker? If so, even before one comes to measure the impact of factors, one first has to identify what does and does not count as a factor. On the other hand, should one take an interpretive approach to understand naturalistically how decision-makers interpret the cases before them (Hawkins 1992; Tata 2007)? If so, is it ever possible to quantify definitively what impact a guilty plea has on sentence? Arguably, both approaches are in fact justifiable, and judges themselves operate and flip between behaviourist and interpretive idioms. Thus, it would be sensible to research the sentence differential using both approaches.

One possible way to explore the sentence differential is to simply ask sentencers what effect a guilty plea has had. For example, some jurisdictions require sentencers to state openly in court what, if any, guilty plea discount has been applied. This information can be used for research such as that by Wren and Bartels (2015). While the particulars of the methodology are beyond the scope of this essay, it suffices to note that the research analyzed “300 decisions handed down by the [Australian Capital Territory] Supreme Court” (Wren and Bartels 2015, p. 372). Using these decisions allowed Wren and Bartels to compare the stated prediscount sentence with the actual sentence imposed to shed some light on how guilty plea discounts were operationalized in these cases (Wren and Bartels 2015, p. 372). However, while this method provided empirical evidence of a sentence discount, it depends on the prediscount sentence being correctly reported. This is potentially problematic for a number of reasons. For example, can sentencers always isolate and quantify the effect of a guilty plea accurately? Alternatively, might the stated prediscount sentence have been altered (even subconsciously) to negate, reduce, or increase any apparent sentence differential?

It might be thought that the issue can be easily determined through the use of guidelines that specify starting points and the required sentence differential. Yet, even relatively juridified and precise guidelines (in jurisdictions were these operate) may not be precise enough to allow empirical verification of guilty plea discounts. For example, even the relatively stringent pre-Booker US Federal Guidelines (United States v. Booker, 543 U.S. 220 [2005]) have only had a qualified success in curtailing judicial discretion in sentencing as they still leave room for “judicial resistance to policies deemed legal, but unjust” (Provine 1998, p. 825). Moreover, curtailing judicial discretion may displace discretion to prosecutors who now have “greater bargaining power concerning charge, plea, and sentence” (Hodgson and Roberts 2010, p.81). Thus, even in what appear to be highly regulated sentencing systems it is still wise to be cautious about making assumptions regarding the sentence differential. Indeed, Tonry has noted that while Washington State has come closest, “no jurisdiction has as yet devised an adequate system for controlling plea bargaining under a sentencing guidelines system” (Tonry 1996, p. 67).

Another way to explore the sentence differential is to compare guilty plea cases and not guilty plea cases using statistical techniques such as logistic regression, and so forth. The details of such methods are beyond the scope of this essay, but the results of some studies are worth noting. Many of these studies focus on the United States and, perhaps unsurprisingly, several support the idea that the sentence differential is substantial (e.g., King et al. 2005; Kim 2015; McCoy 2005, pp. 88–91). Yet, interestingly, not all research supports the notion that there is a large sentence differential.

David Abrams’s work is one such piece. Abrams explores whether the “trial penalty” exists using data from Cook County, Illinois. The study compares cases where there was a guilty plea with comparable cases where there was no guilty plea. The study is notable for its suggestion that the “trial penalty” may not exist and that defendants may actually fare better by pleading not guilty and going to trial (Abrams 2011, 2013). This suggestion runs contrary to what many intuitively believe, and the studies consequently garnered some attention and criticism (Alschuler 2013, p. 687; Kim 2015). Whether these criticisms should be accepted is a difficult question and beyond the scope of this essay; the aim here is merely to highlight that assumptions regarding the sentence differential may be open to challenge.

However, it should be pointed out that Abrams’s study is a poignant example of why it is crucial to be clear what a particular study is measuring. As suggested earlier, the sentence differential and trial penalty can be understood to mean different things. In Abrams’s study he takes the “trial penalty” to include the odds of no conviction. As Kim eloquently puts it: “‘Abrams’ Trial Penalty” = (trial sentence) × (odds of conviction) —plea sentence” (Kim 2015, p. 1217). Thus, Abrams is aiming to measure overall case outcomes (including whether there is a custodial sentence), and not just sentence outcomes.

While definitions are open to debate and not all agree with Abrams’s definition of the trial penalty (Abrams argues it is correct as in any trial there is a “substantial likelihood that a person charged may not be convicted,” 2013, p. 779), the important thing is to be clear regarding the definitions used in a particular study. However, these caveats aside, the apparent lack of an Abrams’s sentence differential could be considered troubling. It suggests that widely held perceptions that defendants will benefit from more lenient treatment by pleading guilty may (all else being equal) be exaggerated; though, of course, this definition of benefit does not account for the burdens an accused may face if he or she goes to trial.

Outside the United States another notable study worth mention is that by Goriely et al. (2001; Tata et al. 2004). This was a large-scale study of Scottish summary (nonjury) cases that focused on overall case outcomes (including the chances of conviction and sentencing outcomes). The study sought to compare the performance in otherwise similar cases of two types of defense lawyers working for legally aided clients.¹¹ By gathering data directly from the courts, the study compared the impact and timing of guilty pleas with cases where the accused pled not guilty on case outcomes. The study found that (with the exception of sex offense cases), generally speaking, the widely perceived reductions for guilty pleas did not materialize. It also suggested that there might be some benefit to pleading not guilty.

Much popular and policy discourses focus on the problem of “late” guilty pleas and the waste of time and resources where defendants plead guilty shortly before trial. It is often said that some defendants simply put off the inevitable “evil day” by continuing to deny their guilt until the trial (e.g., Pleasence and Quirk 2001, para 4.10.7). Yet, aside from whether a defendant may believe himself or herself to be guilty of the charges, there can be sound tactical reasons for maintaining a not guilty plea for as long as possible. For example, Goriely et al. found that the chances of no conviction rose markedly when prosecutors really had to examine the strength of their case:

The longer a case proceeds the more the likely the [prosecution] is to abandon it. While only 2 percent of cases were abandoned at the first hearing, 47 percent of cases resulted in a not guilty outcome after trial evidence was led. Other things being equal, an accused’s best means of avoiding a conviction is to maintain a not guilty plea at least until the morning of the trial and preferably beyond. The chance that the prosecution case will fold is small but real, and contrasts with the certainty of conviction if one pleads guilty.

(Goriely et al. 2001, pp. 97–98) ¹²

Consequently, it may be that not pleading guilty might be more advantageous than is sometimes suggested by the widely held idea of a large sentence differential. This issue could be particularly marked if it means that some defendants are pleading guilty with the false belief that there is a sentence differential. However, it is vital to note that statistical odds are not the same as the perceptions of defendants or indeed their experiences of the criminal justice system. The extent to which one is prepared to take the risk of pleading not guilty until trial and beyond depends, all else being equal, for example, on personal attitude toward risk (in the context of personal circumstances) and trust in the fairness of the system. While some defendants are prepared to take the risk of going to trial, others are keen to get a case over with, especially if they have been held in pretrial detention (Tata and Stephen 2007). Research has repeatedly suggested that while the lead up to the trial is not explicitly punitive, it tends to be experienced as such by many defendants (e.g., Feeley 1979; Wong 2012):

It is the cost of being caught up in the criminal justice system itself that is often most bothersome to defendants accused of petty offences, and it is this cost which shapes their subsequent course of action once they are entrapped by the system … In essence, the process itself is the punishment. The time, effort, money and opportunities lost as a direct result of being caught up in the system can quickly come to outweigh the penalty that issues from adjudication and sentence.

(Feeley 1979, pp. 30–31)

The stress and uncertainty of the pretrial process, not to mention physical deprivations of pretrial detention and bail controls, mean that even if the prosecution’s case is thought to be weak, for many defendants a guilty plea is a prudent (even if disingenuous) decision. Consequently, “there is a real concern that defendants plead guilty to crimes that they may not have committed or should not be found to be legally at fault” (Wong 2013, p. 271).

So far this essay has covered some of the fundamental issues relating to the plea-dependent sentence differential. However, this section briefly highlights some further questions and issues that follow from the fundamentals. The most obvious question to ask regards the empirical nature and extent of the sentence differential.

Here we confront an immediate problem: the deficient quality of data on which most sentencing research has to rely. To be able to quantify and measure sentence differentials depends on the ability to analyze sentencing data in a meaningful way. To isolate the impact of plea on sentence, research has to control for all other relevant sentencing information. The problem is that very often that information is of dubious quality.

For understandable reasons of convenience, resources, and access, most sentencing research (by official bodies and academic research) relies on official data sets. However, in many countries official data tend to be recorded for a variety of official purposes other than sentencing and so, from the perspective of sentencing, tends to be limited.

Critical to the measurement of sentencing differentials is the ability to control for information other than the plea. So, for example, how should research conceive of and measure offense seriousness? Most official data tend to have limited information about the seriousness of convictions. To take just one example, official data tend to focus mainly on recording what is deemed (often somewhat arbitrarily by a nonjudicial agency) to be the principal conviction, rendering other serious convictions as subconvictions, about which little may be known if they are even recorded. The result of this is that the seriousness of multi- and single-conviction cases are insufficiently distinguished, thus undermining the ability to make valid comparisons between cases (Tata 1997). There are two obvious solutions to this issue, though neither is easy. The first is that the quality of official data collected about sentencing needs to be improved markedly. The second is that where researchers are unable to gain access to data that allows for genuinely meaningful analyses about case seriousness and so forth, it will be necessary to collect their own data: this of course is time-consuming and challenging, but the quality of sentencing information tends to be far more satisfactory.

A second area that requires much closer research is defendants’ views and perspectives about the sentence differential. Most of the research conducted examines the views and practices of defense lawyers, prosecutors, and, to a lesser extent, judges. Yet, we know remarkably little about the perspectives of defendants. Much of what we think we know derives from the views and assumptions of practitioners. We need to know more from defendants and, indeed, the extent to which their interpretations and perspectives are shared by practitioners. How is plea decision-making approached and interpreted by defendants? How does it connect with their life contexts and previous encounters with the justice system? Do defendants consider the sentence differential to be a discount or a penalty? How do they understand the meaning of guilt in the particular context of their case? What impact do their perceptions about their treatment (including how to plead) play into wider views of the justice system and authority? How might a sense of processual fairness assist (or exacerbate) the desistance journey away from offending? For example, it was suggested that a guilty plea does not necessarily indicate remorse. However, even if this is true, it may be that the defendant nevertheless comes to internalize some aspect of the guilty plea. In the case of white-collar offenders, it has been suggested “acceptance or rejection of guilt has an impact upon the process of desistance at various stages” (Hunter 2015, p. 182).

This essay has explored the issues and questions surrounding the sentence differential that is perceived to exist in most Western jurisdictions. First, the essay discussed the argument that the sentence differential violates the ideal of a presumption of innocence; sometimes called a “trial tax” or “trial penalty” by opponents. This concern was explored in two sections.

Section II explored the philosophical nature of what constitutes a reward and what constitutes a discount. While there are valid arguments on both sides, and the matter is open to interpretation, the section highlighted that since most defendants plead guilty the “discounted” sentence is actually the one most commonly paid. Accordingly, it might make more sense to consider the postdiscount sentence as the baseline. However, while this has advantages, it was also recognized that it is normatively problematic as it would mean that sentences following a trial are increased. This, rather ungainly situation, raises the question of whether the sentence differential deters defendants from going to trial and thus compromising the presumption of innocence.

Section III explored the criticism that the sentence differential may have a disparate impact on certain groups, such as minorities. It was argued that, by nature, a sentence differential that favors guilty pleas (or disadvantages not guilty pleas) means that those pleading not guilty will fare worse. This is problematic since some groups appear less likely to plead guilty. This section concluded that it was likely that the sentence differential contributes to disparate sentencing outcomes.

Section IV of the essay explored why, considering the criticism and controversy, some justice systems allow the sentence differential to continue and others actually seek to encourage and bolster it. To this end three main rationales were discussed: the remorse rationale, the victim rationale, and the efficiency rationale. All three were found to be limited, but each has its appeal. Of the three, the efficiency rationale is generally considered dominant in legal discourse. However, this paper argues that for the sentence differential to be efficient, it must produce justice, otherwise it cannot, no matter how many trials it avoids, be considered efficient. The essay also questioned whether the sentence differential, and other practices related to securing guilty pleas, actually save resources, or whether there is in fact adaptive behavior and a consequential culture of delay.

Penultimately, section V explored the quantitative question of whether there is an empirically verifiable sentence differential. The essay outlined some of the difficulties that make empirical work on sentencing difficult. In particular, it noted that there are tricky conceptual issues to be wary of, especially as seemingly similar research may measure different things. However, while there is research supporting the common assumptions regarding the sentence differential, the essay also highlighted research that provides reason to question these assumptions. The essay also noted that while the lack of sentence differential might have advantages, if it were absent, it might mean defendants are pleading guilty on an incorrect basis.

Section VI focused on what further questions should be asked about the sentence differential and where future research might focus. It suggested that more needs to be learned about the empirical reality of the sentence differential. The section also argued that greater effort needs to be made to research the perspective of defendants, rather than simply relying on the accounts of practitioners.