Predictive Sentencing

Normative and Empirical Perspectives

Jan W de Keijser, Julian V Roberts, Jesper Ryberg, Jan W de Keijser, Julian V Roberts, Jesper Ryberg

• Publisher: Hart Publishing
• Year: 2019
• ISBN:
  9781509921423
• Edition: 1
• Topic: Law
• Subtopic: Sentencing
• Index: Law

Predictive Sentencing addresses the role of risk assessment in contemporary sentencing practices. Predictive sentencing has become so deeply ingrained in Western criminal justice decision-making that despite early ethical discussions about selective incapacitation, it currently attracts little critique. Nor has it been subjected to a thorough normative and empirical scrutiny. This is problematic since much current policy and practice concerning risk predictions is inconsistent with mainstream theories of punishment. Moreover, predictive sentencing exacerbates discrimination and disparity in sentencing. Although structured risk assessments may have replaced 'gut feelings', and have now been systematically implemented in Western justice systems, the fundamental issues and questions that surround the use of risk assessment instruments at sentencing remain unresolved. This volume critically evaluates these issues and will be of great interest to scholars of criminal justice and criminology.

1

Introduction

Normative and Empirical Perspectives on Predictive Sentencing

JAN W DE KEIJSER, JULIAN V ROBERTS AND JESPER RYBERG

At sentencing, courts look backwards to punish the offender for the crime, and forwards to prevent further offending. The primary objectives of sentencing therefore fall into one of two categories: retributive and preventive. Retributive sentencing focuses on the seriousness of the offence and the offender’s culpability. Establishing the degree of harm and blameworthiness is challenging, but a court is at least dealing with a crime which has occurred. Preventive sentencing is a very different matter. Here the court is required to establish the offender’s risk of further offending and then craft a sentence to address this level of risk. The punishment is imposed for what offenders may do, not for what they have already done. Rather than addressing the past, the court must predict the future. Retributive sentencing creates ethical challenges for sentencers; preventive sentencing throws up even more controversies. This volume explores the normative, ethical and empirical challenges arising from predictive, risk-based sentencing.

With the introduction of the concept of selective incapacitation in the 1970s (see Greenberg 1975; Greenwood and Abrahamse 1982), predictive sentencing immediately attracted much academic attention and debate. While the phrase may often be associated with pseudo-science and crystal balls, predictive sentencing has become a vast enterprise, both in terms of the scientific development of risk assessment instruments and its application to criminal justice. All Western jurisdictions currently deploy a multitude of risk assessment instruments to guide aspects of criminal justice decision making. Consequently, it is used for a myriad of decisions, including bail, sentencing and release from imprisonment.

Risk assessment instruments attempt to predict future offending by combining a variety of static and dynamic criminogenic factors (see Mills, Kroner and Morgan 2011; Otto and Douglas 2010). High-risk offenders may be considered eligible for longer prison terms, while low-risk offenders may be considered more eligible for diversionary measures. Risk assessments are also used to identify criminogenic ‘needs’ that subsequent behavioural interventions may target (according to the ‘risk, needs, responsivity’ model; see Andrews and Bonta 2010). Structured risk assessment instruments are also promoted because they eliminate or minimise the subjective element in criminal justice decision making; that is, they replace more intuitive decision making (see Latessa and Lovins 2014).

I.Predictive Sentencing: A Widely Used But Controversial Practice

Predictive sentencing can be defined as using risk of future offending to influence the nature and quantum of punishment imposed. Although widely used, the practice remains inherently problematic. Concerns are often expressed about the accuracy of risk assessment instruments. Such instruments ascribe a particular risk level to individual offenders by applying statistical group-based models that combine a variety of risk factors. Risk scales may contain a small or a large number of such factors. Some of those factors are believed to reduce the risk of recidivism, while others are supposed to increase the risk of re-offending. These models inevitably produce errors, most importantly in terms of false positive predictions of dangerousness. Predictive sentencing also raises many fundamental legal and normative questions about its justification and application (see also Tonry 1987).

Predictive sentencing has become so deeply ingrained in Western criminal justice decision making that despite the early discussions about the ethics of selective incapacitation, it currently attracts little critique from practitioners. Nor has the practice been subjected to a comprehensive normative and empirical scrutiny. This is problematic since, as Tonry (2014: 170) noted, ‘much current policy and practice concerning risk predictions is flatly inconsistent with mainstream normative theories of punishment’. But the problems of predictive sentencing do not end there. Which characteristics are considered to be risk factors and why are they – or have they become – risk factors? These questions raise concerns about the discriminatory effects of using risk assessments at sentencing. Predictive sentencing appears to exacerbate discrimination and disparity in sentencing (cf Monahan and Skeem 2016). As such, Starr (2014) has described it as the scientific rationalisation of discrimination (see also van Eijk 2017). Although structured risk assessments may have replaced sentencers’ ‘gut feelings’ and have now been systematically implemented in Western justice systems, many fundamental issues remain unresolved.

II.Justifying Predictive Sentencing

A.Retributivism

The justifications for predictive sentencing may be sought in the expected future effects of risk-based sentencing or they may be sought in desert. A full retributive justification dictates a deserved punitive response to wrongdoing that is proportional to the harm done and the culpability of the offender. So is there a desert-based justification for predictive sentencing? At first glance, the answer would seem to be no. After all, desert is retrospectively focused on the issues of blameworthiness and harm. Moreover, predictive sentences would quickly be trumped by concerns for proportionality and equality at sentencing (von Hirsch and Kazemian 2009). Predictive sentencing therefore offends desert-based principles. Can someone be censured or blamed for their risk of criminal behaviour in the future, even if re-offending is almost a certainty? Put differently, does future risk affect the offender’s current penal desert? Moreover, does variation in assessed risk ceteris paribus justify variation in the severity of sentences? Despite such direct and obvious conflicts with desert, conceptualising risk or dangerousness as a reprehensible act or state of mind in itself draws risk assessment into the domain of retributivism.

Another role for retributivism with respect to predictive sentencing may lie in providing a limiting principle (grounded in desert) for constraining preventive interventions based on risk (see, eg, Morris and Miller 1985). However, as a limiting principle, the role for retributivism falls short of providing a general justifying aim (cf Hart 1968) for predictive punishment. Within retributivism, then, there are different forms of predictive sentencing. The views range from those who reject risk-based sentencing even for the most dangerous criminals (cf Petersen 2014) to others who accept risk within a retributive framework, albeit to a limited extent (eg, Duff 1998; Husak 2011; Morse 1996; Morris and Miller 1985). Obviously, for desert theory, these are difficult issues, but justifying predictive sentencing from a consequentialist point of view is also challenging.

B.Utilitarianism

Utilitarians justify interventions by reference to their net benefit for society. In order to be justified from this perspective, the benefits of predictive sentencing must outweigh the costs. However, it is unclear whether the benefits from taking risk into account do in fact outweigh the (added) harm arising from predictive sentencing. While an instrumentalist approach to predictive sentencing would appear obvious, it does grapple with the predictive validity of the risk assessment instruments as well as with the actual benefits that (predictive) sentencing produces. Moreover, if left unconstrained, an instrumentalist approach to predictive sentencing elicits the classic ethical concerns about justifying draconian or inhumane penal measures as well as justifying punishing those who have yet to commit a criminal offence. But there is more. The individual preventive effects of criminal sentences are not as great as one may expect. Thus, any justification of predictive sentencing is linked to the more general question of whether punishment can be justified on utilitarian grounds. Looking at the available empirical evidence, it is not obvious that that is the case.

For example, the literature shows that incapacitation is a very costly penal strategy, while the preventive effects, mainly as a result of overestimating the residual criminal careers, are modest (cf Auerhahn 2002; Blokland and Nieuwbeerta 2007; see also Stolzenberg and D’Alessio (1997) regarding the effectiveness of ‘three strikes’ sentencing laws). Similarly, individual deterrent sanctions (especially imprisonment) do not seem to achieve their intended effects (cf Nagin, Cullen and Jonson 2009; Wermink et al 2018), perhaps not even to the extent that we would expect for a sufficient instrumentalist justification. And finally, considering rehabilitation, the research has shown us that in order to have a chance to achieve rehabilitation, every intervention needs to be tailored to individual risk, criminogenic needs and offender responsivity to the treatment (Andrews and Bonta 2010; Smith, Gendreau and Swartz 2009). But even while such interventions generally produce lower rates of recidivism than custodial sentences, the question remains as to whether these are still sufficient to satisfy the instrumental justification for such interventions.

III.Prediction Issues: Validity and Relevance

Any justification of predictive sentencing must include a discussion of the predictive validity of the risk assessment instruments. How well do specific instruments predict who will re-offend and who will not (eg, Childs et al 2013; Farrington and Tarling 1985; Vincent et al 2008)? Of course, the higher the predictive validity of risk assessment instruments, the better an instrumental justification for predictive sentencing can stand its ground. Incapacitating offenders (most notably through imprisonment), in part or even largely based on such wrongful predictions, inflicts penal suffering and incurs criminal justice costs that cannot be justified. Nevertheless, at the aggregate level, it may still be rational (and cost-effective) to accept a certain proportion of false positives in order to preserve risk assessment in the sentencing equation. The challenge then lies in identifying the ‘tipping point’: what level of false positive predictions is considered acceptable in order to obtain a degree of collective crime prevention benefits? Morris and Miller (1985) argued that using predictions of dangerousness requires a policy judgement on how to balance the predicted risk and harm with the intrusions on individual liberties that result from intervening because of that prediction. It is disconcerting that neither the end-users nor criminal justice policy makers seem aware of or concerned about the issue. On the other hand, from a principled point of view, it has been pointed out that this can only be a conditional challenge to predictive sentencing (von Hirsch and Kazemian 2009). After all, with the continuous scientific development of these instruments and more validation studies which combine static and dynamic risk factors, the diagnostic value of risk assessments has increased over the past few decades. As such, the problem of false positives may diminish as an objection to predictive sentencing (cf Lippke 2008). What if we were able to predict dangerousness with great accuracy? Could – or should – ‘pure preventive detention’ as a form of pre-emptive criminal justice strike be permitted (see Morse 2004; and also Morris and Miller 1985)? But there is more to this. Again, from a principled point of view, if dangerousness itself is to be considered a punishable condition or ‘act’, then the whole issue of false positive predictions becomes trivial. In an individual case, the label ‘high risk’ as the outcome of a risk assessment instrument is unaffected by this person’s actual future behaviour.¹

IV.The Current Volume

If predictive sentencing conflicts with mainstream rationales for punishment, can it be justified? Moreover, as the practice has been so widely embraced by criminal justice practitioners, how strong is the societal support base for predictive sentencing? What do we currently know about the predictive validity of risk assessment tools and what developments or improvements can we expect from them in the future? Can such developments and improvements alleviate some of the more fundamental concerns that are being addressed? How are risk assessments integrated into sentencing practices in different jurisdictions and how are the immediate concerns related to fairness, human dignity, discrimination and validity dealt with in these practices? Answering these interrelated questions requires the collaboration of diverse disciplines. The contributors to this volume not only explore these questions from their own disciplinary backgrounds, but they also engage with contributions from other disciplines.

This volume aims to add to the existing literature on risk assessment at sentencing by keeping a focus on predictive sentencing and by bringing together and integrating empirical, legal and moral perspectives. The variation in disciplinary backgrounds of the contributors is most directly reflected in this volume by the rough division of chapters into two parts; the first part of the volume contains contributions with a normative and legal outlook on the issues at hand. Some contributors squarely focus on defining a consistent principled justification for the practice, while other contributors take a closer look at considerations external to mainstream moral legal rationales for punishment. This is done by some by reflecting on possibilities and constraints that are derived from concerns for human rights, fairness and dignity. Others compare predictive sentencing to other legal, social and economic practices in order to examine justifications for using risk assessments at sentencing. These include prior record enhancements, indeterminate sentences, the medical practice of quarantine, and exploration of the concept of non-punitive incapacitation.

The second part of the volume contains chapters with a predominantly empirical orientation. The contributors critically discuss the developments in predictive validity and application of risk assessment instruments, discuss the discriminatory effects of risk assessment instruments for offenders and for society, address societal views on predictive sentencing and explain the integration of risk assessments in specific sentencing practices.

The mix of moral, legal and empirical perspectives will invite the reader to weigh both moral theory and practice against one another. The empirical and normative contributions to this volume will first be preceded by an overview of predictive sentencing in Chapter 2, describing the historical development of risk assessment instruments and practices of predictive sentencing within and across various jurisdictions. As such, the overview chapter will set the stage for what follows.

This volume does not explore two issues relevant to predictive sentencing. The first concerns protective factors that diminish the risk of committing future crimes. While most contributions to the current volume focus on positive risk factors (ie, factors that are positively correlated with risk of future offending), does that mean that we are addressing only half the story? We think not. After all, criminal justice interventions aimed at creating or strengthening protective factors carry with them the same fundamental problems as those related to the use of positive risk factors for sentencing purposes. Not only do we need to address the predictive validity and effectiveness of (rehabilitative) efforts aimed at strengthening protective factors, but we also need to address the moral justification for such interventions. The punitive effect of such interventions may only be an unintended side-effect and, as a result, not in need of justification. Taking the punitive bite out of the equation in that manner is a rhetorical trick; something described by Hart as a ‘definitional stop’ argument (Hart 1968: 5). Second, the question of the relative effectiveness of criminal sanctions is not discussed. Of course, the issues of crime preventive effects, but also of harmful consequences of sanctions (see, for example, Welsh and Rocque 2014) are closely connected to the instrumental justification of predictive sentencing. As the effectiveness debate is relevant for any consequentialist justification of sentences per se, we have decided to keep the focus of the current volume squarely on risk assessment and its role in predictive sentencing.

References

Andrews, DA and Bonta, J (2010) ‘Rehabilitating Criminal Justice Policy and Practice’ 16 Psychology, Public Policy, and Law 39.

Auerhahn, K (2002) ‘Selective Incapacitation, Three Strikes, and the Problem of Aging Prison Populations: Using Simulation Modelling to see the Future’ 1 Criminology & Public Policy 353.

Blokland, AAJ and Nieuwbeerta, P (2007) ‘Selectively Incapacitating Frequent Offenders: Costs and Benefits of Various Penal Scenarios’ 23 Journal of Quantitative Criminology 327.

Childs, KC, Ryals, J, Frick, PJ, Lawing, K, Phillippi, SW, and Deprato, DK (2013) ‘Examining the Validity of the Structured Assessment of Violence Risk in Youth (SAVRY) for Predicting Probation Outcomes Among Adjudicated Juven...