The Yale Law Journal

May 2015

Federal Sentencing Error as Loss of Chance

In July 2010, a federal district court sentenced DeAngelo Whiteside to seventeen years and six months in prison for a drug offense.1 Under Fourth Circuit precedent, Mr. Whiteside’s two prior state drug convictions triggered application of the Federal Sentencing Guidelines’ “career offender” enhancement.2 On the facts of Mr. Whiteside’s case, the Guidelines recommended between twenty-one and twenty-eight years in prison.3 The district court arrived at its ultimate sentence after granting the government’s motion for a shorter sentence due to Mr. Whiteside’s cooperation.4

If Mr. Whiteside had been sentenced just over a year later, he would not have been a “career offender.”5 In 2011, the Fourth Circuit determined that the Circuit had misinterpreted which state convictions qualify as “prior felony convictions” that trigger the career offender enhancement.6 Under the proper interpretation, Mr. Whiteside’s prior offenses would not have warranted a heightened recommended sentence.7 Instead, the Guidelines would have recommended a maximum prison term of roughly fourteen years and six months.8 Assuming the same downward departure, Mr. Whiteside’s sentence, if determined today, would be nine years and four months—a difference of about eight years of his life.9

Federal courts are currently locked in a debate over what to do with sentences like Mr. Whiteside’s, in which the sentencing court misapplied10 the Federal Sentencing Guidelines’ career offender enhancement.11 The core issue in this debate is whether misapplication of the Guidelines may be challenged post-conviction on collateral review.12 In these cases, the sentencing court’s application of the legal standard for career offender status has been invalidated, typically because the circuit’s interpretation of a “prior felony conviction” has changed.13 The sentencing court’s use of that legal standard is, in retrospect, an erroneous application of law.14 The question is whether such misapplications of law are cognizable in a later challenge under 28 U.S.C. § 2255.15

The test for cognizability in these cases is whether sentencing error constitutes a “complete miscarriage of justice.”16 If it does, then sentences like Mr. Whiteside’scan be challenged on collateral review; if it does not, these sentences stand. Of the four circuits that have applied this test, the Seventh, Eighth, and Eleventh have held that career offender misapplication is not a fundamental miscarriage of justice.17 The First Circuit recently found a career offender claim cognizable on its facts but declined to consider whether sentencing errors arising from a change in legal interpretation give rise to a § 2255 challenge.18 In Whiteside, after a Fourth Circuit panel held that the error amounted to a fundamental miscarriage of justice, an en banc court reversed on the grounds that Mr. Whiteside’s appeal was untimely.19 All of these decisions, except the First Circuit’s, hinged on a single vote.20 Two were en banc.21

This Comment argues that courts have taken the wrong approach to cognizability. Circuit court opinions, and scholarly analysis of these opinions, have framed the argument over misapplication of the career offender enhancement as a conflict between individual fairness—the righting of a wrong by the legal system to an erroneously sentenced individual—and finality—the criminal justice system’s interest in leaving final sentences undisturbed.22 This Comment contends that disagreement over the cognizability of these claims is actually about the nature of the harm in sentencing error. What federal courts are asking, in effect, is whether the lost probability of a lower sentence is itself a cognizable injury. Despite the prevalence of language about “probability” and “risk” in the career offender opinions,23 courts rarely articulate the sentencing debate in these terms. This Comment focuses on the latent probability analysis in sentencing opinions. It argues that a new approach to cognizability—one characterized in terms of probability—would better address the harm at stake in sentencing.

The Comment proceeds in two Parts. Part I draws on an analogy to tort law to argue that sentencing debates are, at their core, about loss of chance. This Part highlights the role that probability plays in recent sentencing opinions. It argues that, as an empirical matter, loss of chance is an accurate way to describe sentencing error given the “anchoring effect” of the Guidelines on sentencing practices.

Part II makes the structural case for conceptualizing Guidelines sentencing error as a problem of probability. This Part argues that failure to recognize the probability dispute has obscured important debates about the continued vitality of the Guidelines system. After United States v. Booker, the Sentencing Guidelines are advisory in principle and influential in practice.24 Part II argues that treating Guidelines error as loss of chance—a loss that can constitute a fundamental miscarriage of justice in the career offender context—is necessary in order to enforce a Guidelines regime that is neither too rigid nor wholly indeterminate. The Comment concludes that a loss of chance framework can help address core concerns in federal sentencing law.

I. the probability debate

In appellate court opinions on career offender misapplication, judges have framed the debate over sentencing error in terms of fairness and finality.25 Circuit courts that have held career offender misapplication not cognizable have emphasized the need for finality in order to minimize systemic burdens on the justice system.26 In contrast, courts that have held sentencing error cognizable have stressed fairness to individual defendants.27 The Fourth Circuit, for example, characterized its holding that career offender error is a fundamental miscarriage of justice as a determination that finality should not “outweigh the plain injustice” of precluding post-conviction review.28 In holding career offender error non-cognizable, the Seventh Circuit majority faulted the dissent for failing to recognize “the difficulty of balancing ‘fairness’ (meaning what exactly?) against finality.”29 The Eleventh and Eighth Circuit opinions30 and academic discussion31 feature similar language.

This line of discourse has obscured the significance of the advisory Guidelines sentencing regime to these cases. In 2005, United States v. Booker invalidated the provisions of the Federal Sentencing Act making the Sentencing Guidelines mandatory.32 Now, though federal courts still determine the recommended Guidelines sentence, that determination is not binding. The shift to an advisory regime leaves the Guidelines’ juridical status unclear. The Supreme Court has called the advisory Guidelines the “lodestone”33 and “the starting point and initial benchmark”34 of federal courts’ sentencing.35 Yet Booker describes the Guidelines as “merely advisory,” and the significance of an “advisory” system is hotly contested.36 Ultimately, disagreement over career offender resentencing is about the meaning of the post-Booker Guidelines: is the probability of a higher sentence due to Guidelines error a harm of its own?

Circuit court opinions on sentencing error do not treat the probabilistic question inherent in these cases in a systematic way; they do not even use the phrase “loss of chance.”37 But the valuative and the empirical questions running through the career offender debate are the essential questions of probabilistic analysis: whether Guidelines error represents a lost opportunity for a better outcome and the value of that loss to the person harmed. Courts disagree about whether missing out on the chance of a lower sentence can ever be a fundamental miscarriage of justice.

A. Chance as Value

When cast in these terms, the sentencing misapplication argument raises issues analogous to those implicated by the loss of chance approach in tort law. In medical malpractice cases, many jurisdictions use the loss of chance approach.38 The idea underlying this doctrine is that the opportunity for a better outcome itself has value, such that deprivation of that chance may be a legally actionable harm.39 Sentencing error debates can be understood along similar lines. Judges who argue that career offender misapplication may be a fundamental miscarriage of justice see the chance of a lower sentence as a harm in itself.40 Those who reject cognizability do not view the probability of a different sentence as a harm requiring review because, even with a different Guideline recommendation, the judge may re-sentence the offender to the same term.41

In the medical malpractice context, loss of chance typically involves a two-step, proportional recovery analysis.42 In the first step, the injury is conceptualized as the loss of an opportunity for a better outcome.43 This characterization enables the plaintiff to recover when she cannot demonstrate preponderance-of-the-evidence causation of the ultimate injury but can show that it is more likely than not that the harm caused a diminished likelihood of a positive outcome.44 Imagine, for example, that a person who died of cancer initially had a forty-percent chance of recovery, but that her physician’s negligence decreased her chances to twenty-five percent. This person’s estate could not prove that the physician’s negligence caused her death.45 But a loss of chance framework would cast the fifteen-percentage-point diminution in the possibility of recovery as a legally cognizable harm.46 The loss of chance doctrine thus allows plaintiffs to recover where they otherwise might not.

Judges and scholars also use probabilistic reasoning in non-medical contexts. The Seventh Circuit applies a loss of chance approach to damages in employment discrimination cases.47 When a fire department violated § 1983 and Title VII by maintaining racial quotas in promotions, the Seventh Circuit explained that the harm at issue was the lost chance of promotion for white applicants.48 The Seventh Circuit required that the jury award damages equal to the product of the total lost benefits and the percentage chance of promotion in order to compensate for the lost opportunity.49 Standing doctrine is another area in which probabilistic reasoning has begun to take hold. As Justice Breyer has emphasized, “courts have often found probabilistic injuries sufficient to support standing.”50 Commentators have suggested that the loss of chance doctrine best explains the existence of “actual or imminent injury” for standing in cases finding probabilistic harm, such as cases in which applicants challenge university programs’ admissions policies.51

The role of probabilistic reasoning has not yet been recognized in the sentencing context.52 This is a mistake given the centrality of probability in sentencing error cases. In misapplication cases, a major—if only implicit—point of disagreement is whether to conceptualize the harm in sentencing error as the lost probability of a different sentence.53 Opinions holding career offender errors cognizable borrow the valuative terms of loss of chance doctrine. For example, the Fourth Circuit has argued that the “chance to be sentenced according to the factors that everyone agrees should apply” is significant in itself.54 These opinions have tended to acknowledge that a new sentence could be the same as the first, since both would be within the permissible statutory range, but have emphasized the likelihood of a new sentence under a different Guidelines determination.55

Other courts argue that, post-Booker, a sentence within the statutory bounds can never be a fundamental miscarriage of justice because that sentence is lawful and could be imposed again.56 The fact that a judge could order the same sentence, even under the new interpretation, means that the sentence must be just, no matter what the odds are that a judge would order a different sentence.57 These opinions do not view the lost chance of a lower sentence as an inherent harm. In arguing over the relative value of the probability of a lower sentence, these sets of opinions are actually, although they do not explicitly recognize it, arguing for and against conceiving of the harm from Guidelines sentencing error as loss of chance.

B. Anchoring Effects

The second step in loss of chance analysis starts from an empirical question: what is the value of the lost chance? In the medical malpractice context, the most common approach to this question is the proportional recovery rule, which values the lost chance as a compensable injury.58 Under this approach, the plaintiff’s decreased probability of a positive outcome is multiplied by the damages in the worse outcome in order to arrive at a final damages determination.59 This step incorporates a factual determination about the medical professional’s negligence into loss of chance analysis. If the decreased chance for a better outcome is de minimis, then there is no possibility of recovery.60

Debates over sentencing involve a similar empirical question about the extent to which the Guidelines influence judges. In tort law, the loss of chance question is how much a physician’s negligence lowered the probability of a different outcome. The parallel question for sentencing error is how much the erroneous Guidelines range affected the original sentence.61 If the Guidelines error lowered the chance of a different sentence, that decreased chance, together with the longer sentence, may make the harm a fundamental miscarriage of justice.62

There is reason to believe that the Guidelines do affect outcomes. In a heuristic pattern first identified by psychologists of judgment and decision-making in the 1970s, an initial quantitative starting point, or a suggested “anchoring” value, affects a final judgment by establishing a presumptive baseline.63 The final determination or estimate will be closer to the anchoring value. Altering the initial anchoring value thus changes the final estimate. Social science research suggests that the Guidelines may in fact have such an “anchoring effect.”64 The Supreme Court has identified such anchoring as the purpose of the Guidelines.65

Courts construing sentencing error as a fundamental miscarriage of justice tend to emphasize the Guidelines’ impact on judges’ decision-making.66 Using empirical evidence67 and sentencing judges’ accounts of how they arrived at their decisions,68 judges stress that the Guidelines are the basis from which sentencing begins. In its since-overturned panel opinion holding the career offender enhancement cognizable, for example, the Eleventh Circuit highlighted Sentencing Commission data indicating that eighty percent of sentences fall within the Guidelines range or result from government motions for a downward departure.69 The Eleventh Circuit panel also cited the sentencing judge’s statement that the career offender enhancement had “essentially doubled” the offender’s sentence.70

The converse view, expressed in opinions failing to find a fundamental miscarriage of justice, is that the Guidelines are “less serious” after Booker.71 This view takes at face value the idea that judges will apply the steps of Guidelines doctrinal analysis mandated by the Supreme Court. Opinions based on this belief emphasize that the sentencing process requires that the Guidelines “not even be presumed reasonable”72 and that the judge “make an independent determination.”73 The Seventh Circuit, for instance, concluded that, “calculating the guidelines range correctly . . . is less important now . . . .74 These opinions dispute the empirical significance of the Guidelines’ anchoring effect, arguing that in the career offender context the offender’s prior crimes independently explain why he or she received a higher sentence.75

Empirical evidence indicates that the latter approach is inaccurate.76 Studies have documented the anchoring effect’s impact on judges,77 including in criminal sentencing, even when judges determine the starting point for a criminal sentence by rolling dice and know that their baseline is completely random.78 In the aggregate, sentencing data suggest that the anchoring effect is borne out: 51.2% of all fiscal year 2013 sentences were within Guidelines range, and 27.9% were within the range of government-sponsored downward departures.79 In other words, only 20.9% of all fiscal year 2013 sentences fell outside of Guidelines- or government-sponsorship-based determinations.80 For career offenders in fiscal year 2012, roughly 30% of sentences fell within Guidelines range, while 41.1% of sentences were government-sponsored below-range sentences.81

The anchoring effect suggests that the Guidelines have a strong influence on offenders’ ultimate sentences, making it likely, although not certain, that an individual would have had a different sentence if the initial benchmark were different. Because the Guidelines still affect judges in practice, the defendant’s loss of chance in cases of sentencing error is real. In the career offender context, the likelihood that misapplied Guidelines affected the outcome—and the years at stake in that error—weigh in favor of cognizability.82

II. the advisory guidelines system’s continued legitimacy

Characterizing sentencing error as loss of chance captures the empirical reality that Guidelines errorsmake a harsher sentence more probable. Structural concerns also support a loss of chance framework. The post-Booker Guidelines run two risks. First, the fact that the regime is discretionary in principle but anchoring in fact may create a legitimacy deficit. Second, such a regime threatens to upset the delicate balance that the Guidelines have struck between indeterminacy and rigidity in criminal sentencing. Recognizing Guidelines error as injury due to loss of chance—and significant error in the career offender context as a fundamental miscarriage of justice—is necessary in order to avert these problems and reinforce the advisory Guidelines regime.

The loss of chance framework can help to avoid what procedural justice theory terms a “legitimacy deficit.”83 Drawing on social psychology, procedural justice scholars have identified the perception that a legal authority’s decision-making procedures are fair as a reason that people view the criminal justice system as an authoritative set of rules that they ought to obey.84 According to this line of reasoning, procedural transparency and consistency promote the rule of law. As Sarah French Russell points out, a systemic failure to “correct clear injustices that are easy to fix” undermines the current United States sentencing regime’s legitimacy.85

The legitimacy problem is particularly acute in the context of Guidelines misapplication. Tom Tyler has identified both the “quality of decision making” and the “understandability of actions” as “antecedents of procedural justice” that heighten people’s beliefs in the legitimacy of the legal system and actually reinforce its legitimacy.86 Perceptions of both of these characteristics are at stake in the cognizability of sentencing error. Intuitively, people often feel that the lost chance of a better outcome is itself an injury.87 The federal sentencing system’s failure to recognize this harm as significant may challenge perceptions of the quality of sentencing, particularly in light of the effects of anchoring bias.88 The perception that the resentencing system is arbitrary, or that it is functioning in a less discretionary manner than Booker suggests, risks undermining the legitimacy of post-Booker sentencing.89

The loss of chance framework also reinforces the Guidelines’ significance in the eyes of judges. For the past half-century, American sentencing has struggled with problems of both indeterminacy and rigidity.90 Sentencing reforms have vacillated between efforts to make the system more flexible, on one hand, and more predictable on the other.91 The paradox of the advisory Guidelines system is that it is both flexible and rigid: the Guidelines draw their power simultaneously from judges’ awareness of their advisory nature and from their anchoring effects on judicial decision-making. The current advisory Guidelines system strikes a balance between the harshness of mandatory Guidelines92 and the pre-1984 indeterminate federal sentencing regime, which created disparities in sentences that implicated issues of bias against minorities.93

This careful balance rests on judges’ continued recognition that the Guidelines are meaningful. If judges begin to perceive the “merely advisory” Guidelines as having little persuasive power, then the sentencing system may revert to pre-Guidelines indeterminacy.94 However, if judges fail to appreciate the Guidelines’ actual power, then judges may become less sensitive to miscalculation and less willing to confront the inequities that are systemically built into the Guidelines.95 Extensive scholarship on the anchoring effect suggests that the greatest threat to the post-Booker system is judges’ underestimation of the Guidelines’ effect on their own sentencing practices.96 When judges believe the Guidelines have no real force, they may fail to appreciate the subtle but real effect the Guidelines have in creating a baseline for their sentencing determinations. The loss of chance framework corrects for this error by signaling to judges that the Guidelines continue to have practical significance and by requiring judges to think carefully about the effect of Guidelines error in particular cases.

Ultimately, the failure to recognize that significant Guidelines error can be a fundamental miscarriage of justice—because it represents a loss of chance—risks devaluing the Guidelines in the eyes of judges.97 In order for the Guidelines to continue to function as a compromise between indeterminacy and rigidity, courts must act in accordance with the principle that the Guidelines matter. Framing the harm at stake in sentencing as a loss of chance of a lower sentence encourages such an approach. The loss of chance analogy, borrowed from tort law and familiar from other probabilistic reasoning in law, can help to maintain the integrity of the current Guidelines regime.


Framing sentencing error as loss of chance highlights an institutional value: the continued vitality of the advisory Guidelines regime. Underlying sentencing miscalculation cases is a larger post-Booker debate about the significance and structure of non-mandatory Guidelines.98 While scholars continue to debate whether the post-Booker system is an improvement, the advisory structure is here to stay. Booker unraveled the mandatory sentencing regime,99 and a return to an indeterminate sentencing system seems not only undesirable but also unlikely.100 The pressing question, then, is how best to cognize errors within the regime we have.

Academic commentary praising the advisory Guidelines system highlights its flexibility101 and its potential to make judges more cognizant of their choices.102 Because the Guidelines gain their force from their anchoring effect on judges,103 sentencing doctrines should recognize the Guidelines’ impact on the probability of a certain sentence. Guidelines errors of serious magnitude, like the career offender enhancement, should be understood as fundamental miscarriages of justice so that judges and other actors in the criminal justice system continue to place serious weight on the Guidelines.104 In a post-Booker world, the least desirable outcome would be a sentencing regime in which sentencing guidelines are considered insignificant105 but continue to have a major effect on sentencing.106 Such a system would combine the worst aspects of sentencing: the potential for indeterminacy and a lack of transparent reason.

Characterizing sentencing error as loss of chance averts this outcome. This framework explicitly recognizes the reality of the Guidelines’ impact on sentencing. It also acknowledges that when, for instance, nine years of a person’s life are at stake, the likely effect of an error in Guidelines calculation is a fundamental miscarriage of justice. In order to maintain the continued relevance of the Guidelines, courts should recognize that the baseline norm affects probabilities in a case like DeAngelo Whiteside’s. For Mr. Whiteside, and for other defendants sentenced in error, the loss of chance may itself be a significant harm.