How Categorical Rules Work

How Categorical Rules Work

The two excessiveness tests say when a punishment is too much, and they leave open a question of method: how does the Court decide that a whole sentencing practice, such as executing every member of a class of offenders, is forbidden in every case? The answer comes from Graham v. Florida, where the Court held that life imprisonment without parole for a juvenile who did not commit homicide is categorically cruel and unusual.¹ Along the way the opinion gathered the earlier proportionality decisions into a single framework.² A later section teaches Graham's facts and holding in full. Here it supplies the map, because the death-penalty decisions that come next were reasoned through the same steps, and reading them against one framework keeps each from being learned as a separate puzzle.

The framework begins with a sorting step. Graham explained that the cases implementing the proportionality precept fall into two classifications. In the first, the Court considers all the circumstances of the particular case to decide whether the sentence is unconstitutionally excessive, and Solem v. Helm is the example the Court gave.³ The second classification uses categorical rules to define the Eighth Amendment standard, as in Kennedy v. Louisiana, and such rules are appropriate when a sentencing practice is inherently disproportionate for a certain class of offenders or a certain class of crimes.⁴ The two classifications exist side by side because they answer different problems. A sentence whose fairness depends on the details, such as a repeat offender's record and the circumstances of the latest crime, can only be judged by looking at those details. A practice that is wrong for everyone in a group, whatever the details, can be decided once for the whole group, and a categorical rule does exactly that. Because one decision then reaches every future case in the class, the Court wants support for it from outside the Justices' own views, which is why the method that follows starts with evidence and only afterward turns to judgment.

Categorical rules come in two kinds, depending on what defines the class. In the first, the Court looks at the nature of the offense and concludes that the sentence is disproportionate for a particular class of crimes, as in Coker v. Georgia. In the second, the Court looks at the nature of the offender and concludes that the sentence is disproportionate for that class of offenders, as in Roper v. Simmons (the death penalty for juveniles) and Atkins v. Virginia (the death penalty for intellectually disabled offenders).⁵ These are not the only decisions of the kind. Where objective indicia of community values have shown a consensus that the death penalty is disproportionate for a certain class of cases, the Court has set substantive limits on its use: no death sentence for the rape of an adult woman in Coker, none for a felony-murder defendant absent a showing of a sufficiently culpable mental state in Enmund v. Florida, and none for a prisoner who is insane in Ford v. Wainwright.⁶ The sections on the offense-based and offender-based bars take up the main examples, and the same steps govern the rest.

The steps pair an outside check with an inside one. Objective indicia keep the decision tied to what legislatures and sentencers have actually done, so the Justices' private views do not carry the result alone. Yet that evidence can be incomplete or distorted, and the stakes, especially when the sanction is death, are too high to rest on a headcount. For that reason the Court follows consensus with its own judgment about culpability and severity.⁸ The consensus step also looks past the statute books to how a practice is used. In Graham, thirty-seven jurisdictions permitted life without parole for a juvenile nonhomicide offender, yet only 109 such offenders were serving the sentence nationwide. The Court found that number, standing alone, insufficient to show a consensus, but a closer comparison of the jurisdictions that permit the sentence with those that prohibit it showed a consensus developing, and the sentence's rarity suggested that it is not needed to serve legitimate penological goals.⁹

On an exam stem that recites how many states authorize a punishment, the temptation is to treat the count as the answer. Students lose this issue when they treat a head count of statutes as the end of the analysis. Consensus is the first step of the Court's inquiry, and the Court has also looked to the standards of the medical and psychological communities and to its own sense of moral proportionality, so a complete answer carries the facts through culpability, severity, and penological purpose as well.¹⁰ The four sections below take the steps one at a time, starting with the choice between case-by-case and categorical review, and the capital-punishment sections then run the full sequence on the offense-based and offender-based bars.

Case-by-Case Versus Categorical

Every proportionality challenge starts with a sorting decision, because the classification controls which method of review the Court uses. The Eighth Amendment guarantees individuals the right not to be subjected to excessive sanctions, and the Court's cases implementing that guarantee fall into two general classifications.¹² In the first, the Court considers all of the circumstances of the particular case to decide whether the sentence is unconstitutionally excessive, and Solem v. Helm is the example the Court itself gives.¹³ Solem held unconstitutional a sentence of life without parole for the defendant's seventh nonviolent felony, passing a worthless check, and that contrast is all this chapter needs from it.¹⁴ The point to hold is the method: the sentence is measured against one defendant's record and one crime, so the facts of that case supply the whole analysis. Challengers who proceed this way often struggle to establish a lack of proportionality, and Harmelin v. Michigan, which involved life without parole for possessing a large quantity of cocaine, is the leading example.¹⁵

The second classification uses categorical rules to define Eighth Amendment standards, and the Court reserves them for sentencing practices that are inherently disproportionate as applied to certain classes of offenders or certain classes of crimes.¹⁶ Two reasons explain why a class-wide challenge calls for a different method. The first is that the opening comparison used in individual cases, which sets the gravity of one offense against the severity of the sentence, does not advance the analysis when the target is a type of sentence applied to a whole class of offenders who committed a range of crimes, because there is no single offense to weigh.¹⁷ The second is that deciding the question one defendant at a time would ask every sentencer to repeat the same hard judgment. In Graham v. Florida the State and the United States proposed exactly that, asking sentencers to consider a juvenile's age and other mitigating factors before imposing life without parole. The Court declined, because it is difficult to tell the juvenile whose crime reflects transient immaturity from the rare juvenile whose crime reflects irreparable corruption, so an error-prone call would recur in every case.¹⁸ A categorical rule removes that call and gives sentencers and legislatures a clear line.¹⁹ Case-by-case review stays available where no class-wide disproportion can be shown and the fairness of the sentence depends on the particular facts.

A claim about a whole class looks like this in Roper v. Simmons. The objection there was to executing everyone who commits a capital crime while under 18, so the Court answered with a categorical rule instead of weighing each defendant's circumstances.²⁰ That example also shows the second fork inside the categorical classification, because the Court recognizes two kinds of categorical rule, distinguished by what defines the class. An offense-based rule looks at the nature of the crime and concludes that the sentence is disproportionate for a particular class of crimes. Coker v. Georgia, which barred death for the rape of an adult woman, and Kennedy v. Louisiana, which barred death for the rape of a child where the crime did not result, and was not intended to result, in the victim's death, are the examples.²¹ The class is defined by the crime, so the bar protects anyone convicted of that crime, whoever the offender is. An offender-based rule looks at the nature of the offender and concludes that the sentence is disproportionate for a class of people whose characteristics reduce their culpability, as in Roper (juveniles) and Atkins v. Virginia (intellectually disabled offenders).²² Here the class is defined by the person, and the bar applies whatever the crime.

The Court itself sorts the earlier cases along this line. Roper and Atkins turned on the offender's diminished personal responsibility for the crime, while Coker and Enmund v. Florida turned on the disproportion between death and the crime itself.²³ Enmund, for instance, overturned the death sentence of a defendant who aided a robbery during which a murder occurred but who did not kill, attempt to kill, or intend that a killing take place.²⁴ Graham v. Florida belongs on the offender side, and it is the case that carried the method beyond capital punishment. The earlier categorical cases involved the death penalty, whereas Graham presented a categorical challenge to a term-of-years sentence, life without parole for a juvenile who did not commit homicide, and the Court resolved it with the categorical approach.²⁵ The class in Graham has two features, youth and the absence of homicide, but the reasoning rests on the lesser culpability of juveniles that Roper recognized.²⁶ The boundary between the two classifications is not perfectly tidy for prison terms. Nothing obvious makes juveniles who commit nonhomicide offenses a better comparison class than nonviolent offenders generally, so it is possible, though unsettled, that sentences like those in Harmelin and Ewing could face the more demanding comparative standard Graham used, and it is equally possible that the Court will keep treating finite terms of years as too hard to compare across states.²⁷

On an exam stem, classification comes first and it follows the claim. Students misclassify when they sort by the type of penalty instead of by what the challenger asks the court to decide. Life without parole appears on both sides of the fork: Solem judged it case by case for a seventh nonviolent felony, while Graham barred it categorically for juveniles who commit nonhomicide crimes.²⁸ Once the claim is categorical, ask whether the class is defined by what was done or by who did it, because that choice points to offense-based or offender-based precedent. The next three sections take the steps of the categorical method one at a time, and the capital-punishment sections after them apply those steps to the offense-based cases first and the offender-based cases second.

Objective Indicia of Consensus

The first step of the categorical analysis sends the Court to evidence outside its own chambers. Before the Justices say anything about culpability or severity, they ask what the country's legislatures and sentencers have actually done with the practice under challenge, and whether those choices add up to a national consensus against it. The Court describes this inquiry as a search for objective indicia of society's standards, as expressed in legislative enactments and state practice.³⁰ The inquiry rests on the idea that the Eighth Amendment draws its meaning from the evolving standards of decency that mark the progress of a maturing society, and the best evidence of those standards is the legislation the country's legislatures have enacted.³¹ Jury sentencing decisions rank next, because they are a significant and reliable index of contemporary values.³²

The reason to start with outside evidence is that a judge asked what decency requires is tempted to answer with a personal moral preference. Legislative votes and jury verdicts check that temptation, since they are choices made by bodies answerable to the community and recorded in a form anyone can count. The Court has said that Eighth Amendment judgments should not be, or appear to be, merely the subjective views of individual Justices, and that they should be informed by objective factors to the maximum possible extent, including history, legislative attitudes, and the response of juries.³³ Counting legislatures and juries therefore keeps the Clause tied to what the public has chosen rather than to what nine people believe.

Consensus is a fact about the country at a particular moment, and the Court's treatment of intellectual disability shows that fact changing. In Penry v. Lynaugh the Court held that the Constitution did not categorically bar executing intellectually disabled offenders, because only two States had enacted bans and that was insufficient evidence of a national consensus.³⁴ By the time of Atkins v. Virginia, sixteen more States had enacted such bans.³⁵ The Court gave weight to the fact that the large number of prohibiting States was matched by the complete absence of States passing legislation to restore the power to execute, and to the fact that the legislatures that took up the question had voted overwhelmingly for the prohibition.³⁶ Even among States that still allowed these executions, the practice was uncommon.³⁷ The constitutional text stayed the same between the two decisions while the legislative record moved, and that movement is what produced a different answer to the same question.

The same pattern drove Roper v. Simmons, which overruled Stanford v. Kentucky, the decision that had allowed the execution of sixteen- and seventeen-year-old offenders.³⁸ After Stanford, the number of States prohibiting juvenile executions grew, and the practice became rare: since 1989 only six States had executed anyone for a crime committed as a juvenile.³⁹ Those facts show why the direction of change matters as much as the raw count. A modest number of prohibiting States, moving steadily the same way with no reversal, can show a consensus that a larger but static number would not. Justice Breyer's dissent in Glossip v. Gross, which is not the Court's holding, used the same measure for capital punishment generally, counting the States that still carry out executions and stressing the consistency of the direction of change.⁴⁰

Jury behavior counts alongside legislation, and Coker v. Georgia shows both kinds of evidence pointing the same way for an offense rather than an offender. In 1925, eighteen States, the District of Columbia, and the federal government authorized death for the rape of an adult woman, but by 1977 Georgia was the only jurisdiction that did.⁴¹ Georgia's own juries rarely used the power, since in at least nine of ten cases they declined to impose a death sentence.⁴² The legislative retreat and the jury reluctance reinforced each other, and the Court treated the legislative rejection as strong confirmation of its own judgment that death is a disproportionate penalty for that crime.⁴³

Statutes can mislead in the opposite direction from Atkins, which is why actual practice counts too. In Graham v. Florida, thirty-seven States and the District of Columbia permitted life without parole for a juvenile who committed a nonhomicide offense, six jurisdictions barred the sentence for every juvenile, and seven allowed it only for homicide, and the State and its amici argued from that statutory count that no consensus existed.⁴⁴ The Court looked behind the statute books to how the sentence was used. Only 109 juvenile offenders nationwide were serving it for nonhomicide crimes, and many of the jurisdictions that permitted it had no one serving it at all.⁴⁵ A law can remain on the books through inertia, because no legislator has a reason to repeal a penalty that no prosecutor seeks, so a permissive statute may reveal little about whether communities accept the sentence in practice. The Court did not treat the 109 as proof of consensus by themselves. It compared the jurisdictions that permit the sentence with those that prohibit it, found a consensus developing, and observed that the sentence's rarity suggested it is not needed to serve legitimate penological goals.⁴⁶

On an exam stem that recites statutory counts, the sound move is to read every column the facts supply: the statutes, the frequency of actual use, and the direction of change. Students lose this issue when they treat law on the books as the whole record, because a large number of authorizing statutes can coexist with a rarely used sentence, and a small number of prohibiting statutes can sit inside a clear trend. The Atkins pair supplies the matched contrast. Two prohibiting States against a growing list of bans and no reinstatements is the same question at two points in time, answered no and then yes.

Even a strong consensus is only the first half of the analysis. In Coker the Court said that the recent legislative and jury evidence did not wholly determine the controversy, because the Constitution contemplates that in the end the Court's own judgment will be brought to bear on the acceptability of the punishment.⁴⁷ A second reason follows from how the evidence is produced. A legislature that reads a Supreme Court decision as closing off a punishment has little reason to enact it, so a bare statute book can reflect the earlier decision as much as independent community values. That weakness is why the next step turns from counting to the Court's own judgment about the culpability of the class and the severity of the punishment.

Culpability and Severity

The count of legislatures and juries does not finish the categorical analysis, because the Court treats that evidence as one input and reserves a second step for itself. In Kennedy v. Louisiana, after surveying the legislative landscape, the Court said that the inquiry does not end with consensus. Whether a punishment is disproportionate also depends on the standards elaborated by controlling precedents and on the Court's own understanding of the Amendment's text, history, meaning, and purpose.⁴⁹ Its holding that death cannot be imposed for the rape of a child, where the crime did not result and was not intended to result in death, rested on both consensus and the Court's own independent judgment.⁵⁰ That is the shape of the whole categorical method. Step one gathers objective evidence, and step two, the subject here, is the Court's own assessment of how culpable the class is and how severe the punishment is.

Two reasons explain why the Court will not hand the question entirely to the count. The first concerns the interest the Clause protects. Evolving standards of decency must embrace respect for the dignity of the person, and the punishment of criminals must conform to that rule.⁵¹ Dignity is an interest a legislative majority can undervalue, because the people most affected by a harsh penalty are rarely the people who vote on it, and a rule that made the vote count conclusive would let the majority define the limit on its own power. The second reason is that death cannot be undone. The Court has said that the Eighth Amendment demands more than that a challenged punishment be acceptable to contemporary society, and that the Constitution contemplates the Court's own judgment on whether death is acceptable.⁵² Justice O'Connor's dissent in Roper, which disagreed with the result but accepted the method, described a "constitutional obligation" to judge for itself whether death is excessive for a particular offense or class of offenders.⁵³ A dissent in McCleskey v. Kemp, which is not the Court's holding, adds the reliability point: because death differs so much from any prison term, the level of rationality required in capital sentencing must be uniquely high.⁵⁴

What the Court does with that independent judgment is weigh the culpability of the offenders, in light of their crimes and characteristics, against the severity of the punishment.⁵⁵ The two sides must be weighed together because a punishment is excessive only in relation to what the offender deserves. The same sentence can therefore be proportionate for one offender and disproportionate for another whose capacity for blame is diminished. Capital punishment adds a ceiling on the culpability side: it must be limited to a narrow category of the most serious crimes, committed by offenders whose extreme culpability makes them the most deserving of execution.⁵⁶ Culpability can fall short of that ceiling for two different reasons. The offender may bear diminished personal responsibility, as in Roper and Atkins, or the crime itself may fall short, as in Coker and Enmund, where death was disproportionate to a crime that did not result, or was not intended to result, in the victim's death.⁵⁷ The line is graded rather than all-or-nothing, since in Tison v. Arizona defendants who did not themselves kill, but whose involvement in the events leading to the murders was active, recklessly indifferent, and substantial, kept their death sentences.⁵⁸ In Kennedy the state court found no characteristic of the petitioner that tended to mitigate his culpability, yet the Court still held that death was barred for the rape of a child where the victim did not die and death was not intended.⁵⁹

The juvenile cases show how culpability reasoning works on the offender side. Roper and Graham v. Florida rest on three gaps between juveniles and adults: a lack of maturity and an underdeveloped sense of responsibility, greater vulnerability to negative influences and outside pressures together with limited control over their own environment, and a character that is less well formed, so that their actions are less likely to be evidence of irretrievable depravity.⁶⁰ Each gap bears on blame for its own reason. Retribution punishes a choice, and the less capacity an offender had to weigh risk, resist pressure, or escape a crime-producing setting, the less of the choice belongs to the offender. The third gap works differently, because it undercuts the inference that a sentence of permanent exclusion depends on, namely that the crime reveals who the person will always be. Graham drew from these differences that juveniles are more capable of change than adults and so are less deserving of the most severe punishments.⁶¹

Severity is measured by what the sentence actually takes from the person, and that varies with the person even when the label stays the same. Life without parole is an especially harsh punishment for a juvenile, who will on average serve more years and a greater percentage of life in prison than an adult, so a 16-year-old and a 75-year-old sentenced to life without parole receive the same punishment in name only.⁶² The Court added that life without parole is the second most severe penalty the law permits and that, for a juvenile, it resembles a death sentence because it guarantees the juvenile will die in prison.⁶³ Putting the two sides together, Graham found reduced culpability and heightened severity. That mismatch conflicted with the principle that punishment be graduated and proportioned to the offense, and the sentence imposed on Graham, who did not kill or intend to kill, was unconstitutional.⁶⁴

At the opposite pole, an adult convicted of a deliberate killing, with no characteristic that reduces blameworthiness, presents culpability at its highest. The Court has said that the death penalty is not invariably unconstitutional, citing Gregg v. Georgia, which is why it has confined capital punishment to the narrowest category of cases rather than abolished it.⁶⁵

On an exam stem, the sequence is to find the consensus evidence first and then ask what the Court's own judgment adds. Students lose this step when they stop once the legislative count favors one side, because a count that favors the challenger still leaves the weighing to be done, and a count that is less clear does not remove the Court's judgment from the analysis either. On the culpability side, name the specific trait of the offender or feature of the crime that lowers blame. On the severity side, name what the sentence actually takes from this offender. The balance is still incomplete without one more question, whether any traditional purpose of punishment can justify the sanction for this class, and that question is the last piece of the Court's own judgment and the subject of the next section.

Penological Justifications

The last part of the Court's own judgment asks whether any recognized purpose of punishment can explain why the State needs this sanction for this class of offenders. The question follows from the weighing of culpability and severity. Once the Court has found a class less blameworthy and a sentence unusually heavy, the remaining issue is whether the sentence gives the State anything a lesser penalty would not. A penalty that serves no legitimate purpose inflicts suffering for its own sake, and the precept that punishment be graduated and proportioned to the offense leaves no room for that. Kennedy v. Louisiana identifies three principal rationales for punishment: rehabilitation, deterrence, and retribution.⁶⁷ Graham v. Florida tests four by adding incapacitation, and it serves as the worked model for this step.⁶⁸

In Graham the class was juveniles sentenced to life without parole for a nonhomicide crime, and the Court found that none of the four purposes justified the sentence.⁶⁹ Retribution comes first because it connects most directly to the culpability findings. It answers to blame, and the case for retribution is weaker with a minor than with an adult, so a sentence defended as deserved cannot rest on blame that the juvenile has less of.⁷⁰ Deterrence works only if the offender weighs a possible penalty before acting. Juveniles are less likely to take possible punishment into consideration when making decisions, so the threat of the harshest available sentence adds little to what an ordinary penalty already deters.⁷¹

Incapacitation and rehabilitation are the two purposes a State can most plausibly claim, and Graham's treatment of them is the most instructive part of the opinion. Incapacitation is an important goal because recidivism threatens public safety, and it can justify life without parole in other settings.⁷² To justify the sentence for juvenile nonhomicide offenders, however, the State would have to demonstrate affirmatively that all of them are fit to be treated as permanent prisoners.⁷³ That showing is out of reach because a life-without-parole sentence is a prediction made at the outset, while a juvenile's character is not yet fixed and most juveniles who engage in crime mature out of it.⁷⁴ Graham himself deserved to be separated from society for some time, given an escalating pattern of criminal conduct, but a present risk does not establish a risk for life, and later prison misbehavior would not have rescued the sentence because the judgment was made at the beginning.⁷⁵ Rehabilitation fails for a more direct reason. A sentence that bars reentry forecloses the very thing rehabilitation aims at, since it makes an irrevocable judgment about the person's place in society, and prisoners serving such sentences are often denied vocational training and other rehabilitative services.⁷⁶

The Court's conclusion followed from the combination: penological theory could not justify life without parole for juvenile nonhomicide offenders, and that finding joined the limited culpability of the class and the severity of the sentence to make the practice cruel and unusual.⁷⁷ The holding has a limit that matters for exam answers. A State need not guarantee eventual freedom to such an offender, but it must give some meaningful opportunity to obtain release based on demonstrated maturity and rehabilitation, and the Amendment leaves open that some of these offenders will remain behind bars for life.⁷⁸ What the Clause forbids is the judgment, made at the outset, that the offender will never be fit to reenter society.⁷⁹

The earlier cases run the same test on other classes and find the same weakness. In Roper v. Simmons, once the Court recognized the diminished culpability of juveniles, it followed that retribution and deterrence, the two rationales offered for the death penalty, apply to juveniles with lesser force than to adults.⁸⁰ The plurality in Thompson v. Oklahoma had reasoned the same way about offenders under sixteen: lesser culpability made execution poor retribution, and such offenders rarely engage in the cost-benefit analysis that gives any weight to the possibility of execution.⁸¹ In Atkins v. Virginia the Court found that the retributive value of executing offenders with intellectual disability diminishes, because their lesser culpability makes them less able to control their conduct and therefore less deserving of the most severe punishment.⁸² In each case the purpose is tested against the class, and the class's reduced blame is what drains the purpose of its force.

Kennedy adds a caution about why this step carries special weight when the sentence is death. Of the three principal rationales, retribution is the one that most often can contradict the law's own ends, and when the law punishes by death it risks a sudden descent into brutality, which is why capital punishment must be confined to a narrow category of the most serious crimes and the most extreme culpability.⁸³ Retribution looks backward and asks only what the offender deserves, so it supplies no internal check against escalation. In capital cases the Court's insistence on a narrow category supplies the check from outside.

A dissent in Gregg v. Georgia pressed the point further, and it is not the Court's holding. That dissent argued that death is unnecessary to deterrence or to any legitimate notion of retribution, and that sustaining death because the wrongdoer deserves it denies the wrongdoer's dignity.⁸⁴ The Court has not adopted that position. It has said that the death penalty is not invariably unconstitutional and has chosen instead to confine the instances in which it can be imposed.⁸⁵ The dissent is useful for the shape of its argument, but an exam answer should label it as a dissent and should not present it as the rule.

The adult side of that pair rests on the comparison drawn in Roper, where recognizing the diminished culpability of juveniles led the Court to conclude that retribution and deterrence apply to them with lesser force than to adults.⁸⁶ That comparison is why the same crime can fail the penological test for a seventeen-year-old and survive it for an adult. It also explains why the contrast with Graham's class is so sharp. Graham's sentence lost every purpose it was tested against, while the adult's sentence keeps the purposes that matter most.

The most common error is testing each purpose against the individual offender instead of the class. Graham could be dangerous, and incapacitation still failed, because the question was whether the State could show that all juvenile nonhomicide offenders are permanently dangerous, and a judgment made at sentencing could not show that.⁸⁷ A second error is to name a purpose and stop. A purpose counts only if the class's characteristics leave it with real force, so the answer must say why retribution or deterrence weakens for this class, or why incapacitation or rehabilitation remains available. With the penological question answered, the three-part method is complete, and the next sections apply it, beginning with the offense-based bars on capital punishment.