THE CANADIAN SOCIOLOGICAL ASSOCIATION
LA SOCIÉTÉ CANADIENNE DE SOCIOLOGIE
info@csaa.ca
Publisher of / Éditrice de
The Canadian Review of Sociology
Revue canadienne de sociologie
BOOK REVIEW - RECENSION DE LIVRE
BERNARD E. HARCOURT, Against Prediction:
Profiling, Policing, and Punishment in an Actuarial Age. Chicago: University of
Chicago Press, 2007, vii + 336 p. + index.
Reviewed by Paul F. McKenna, Dalhousie University
Bernard Harcourt is a law professor at the University of Chicago and director of that school’s Center for Studies in Criminal Justice. He is the author of previous works examining the “broken windows” theory of policing developed by George Kelling and James Q. Wilson, as well as youth gun crime and public policy. The present publication explores Harcourt’s deepening concern over the dominance of actuarial science within the realm of criminal justice. Of particular interest is the impact of mathematically-based prediction upon forms of criminal profiling, most notably its racial strain.
Professor Harcourt divides his attention across three dimensions. First, there is an examination of the rise of actuarial methods as a pervasive paradigm. Secondly, he presents a critique of those methods through a careful analysis of their impacts on parole, penal policy and policing. Thirdly, the author posits a general theory of punishing and policing that relies upon randomization. Throughout, Harcourt challenges conventional contemporary thinking and practice which privileges actuarial science.
The first section addresses the gathering confidence of sociologists in the application of statistical methods for parole decision-making. Harcourt cites the work of University of Chicago sociologist Ernest Burgess in the 1920s as a landmark in the advance of actuarial methods. Burgess’s reliance on prediction and statistical method is a prelude to a significant transformation in parole practices in the United States. Harcourt draws a direct line from Burgess to his student, Ferris Laune, who became the first incumbent of the Sociologist and Actuary post with the Illinois State Penitentiary in 1933. It appears that sociologists are deeply implicated in the process of abandoning clinical approaches in favour of mathematical models of prediction for carceral purposes. For Harcourt, this transition is linked to a movement away from an earlier concentration on moral culpability, clinical diagnosis and treatment interventions to what has been termed the “new penology” which displays an indifference to individual cases and “small numbers” within the criminal justice system. Prognostication begins to supplant other approaches for dealing with offending populations. Harcourt’s thesis incorporates the view that this has had a deleterious effect on our conception of justice.
Increasingly, actuarial methods have made their way into the calculus of policing, punishment, and parole decisions. Harcourt suggests that econometric models actually serve to advance an efficiency argument for accepting racial profiling practices. For example, “good” policing may require profiling based upon categories, generalizations and stereotypes. The use of risk assessments, algorithms and criminal profiles allows law enforcement agencies to pinpoint with a high degree of accuracy the allocation of limited resources.
However, Harcourt joins several other legal scholars in expressing discomfort with the application of actuarial methods in the arena of criminal justice. Significant effort is placed on explaining the elements, and exposing the limits, of these econometric and actuarial approaches. Rational choice theory informs much of the work done on criminal profiling. This theory builds upon the notion that individuals are elastic with regard to policing activities, i.e., their offending will vary according to the likelihood of being detected; the cost of crime. Therefore, when police are making resource allocation decisions, they should be applauded for actively profiling based upon group offending rates. And yet, there is a palpable problem with this logic when considering how different groups actually respond to changes in policing. A group’s “relative elasticity” will be informed by many considerations. Harcourt asserts that there is no good reason to assume that high-offending groups will respond to policing in the same manner as low-offending groups. His assertion is based upon socio-economic, and other, factors. Furthermore, when it becomes apparent that state agencies are profiling any given “group” (e.g., airline hijackers, drug couriers, repeat offenders, and tax evaders) there is a strong likelihood that offending rates within the non-profiled population will increase, thereby nullifying the supposed claims of enforcement efficiency. Harcourt spends some time explaining that mathematical equations may be applied to an assessment of police racial profiling. For many readers, this section will be inaccessible. However, the author offers a simpler treatment in a following portion of the text.
Central to Harcourt’s critique is the problem of what he terms the “ratchet” effect. The ratchet effect occurs when a particular group has been identified as characteristic for some purpose. For example, when drug courier profiles were formulated in the 1970s by Drug Enforcement Agency (DEA) officers, it was claimed that more efficient (i.e., successful) searches could be carried out on the basis of such profiles. And yet, when this particular profiling tool was evaluated by a U.S. National Institute of Justice study in 1982, there was a 34% “success” rate based upon one hundred and forty-six passengers encountered. Clearly, if this profile template were applied as a matter of public policy, there would be exponential growth in the number of individuals searched according to that profile resulting in a self-fulfilling prophecy, but not necessarily improved drug enforcement. Harcourt explains the ratchet effect as a logical consequence of “statistical discrimination” with a potential for dire social implications. Again, if the police increasingly sample from within the high-offending population they will increasingly target that precise population and correspondingly reduce attention to the low-offending population. Over time, the ratchet effect will produce what can only be viewed as an unjust system of policing. The devastating impact of this effect is not difficult to comprehend when one considers the “minority” carceral populations in the United States and Canada. Also, the ratchet effect may have negative consequences for police-community relations. When aggressive police enforcement initiatives (for example, the Toronto Anti-Violence Intervention Strategy (TAVIS) mounted by the Toronto Police Service) relentlessly pursue a profiled high-offending population, it is not surprising when that targeted community becomes less willing to report crime, or to testify against and convict individuals in their midst.
Harcourt proceeds to question the related notion of incapacitation as a guiding principle within criminal justice. Studies conducted in Philadelphia in 1945 by Marvin Wolfgang, Robert Figlio and Thorsten Sellin involving 9,945 boys revealed that 6% of these youths were committing 50% of the crime. Absent any internal limiting principle, Harcourt ponders whether supporters of a policy of incapacitation might recommend the incarceration of the entire male population between the ages of 16 and 24 as conducive to public order. And, while building toward a defence of randomization as a method for ethical justice, Harcourt is not oblivious to the intellectual value of statistical insight. For example, he recounts the work done by Gould and Mastrofski (2004). These researchers led a highly qualified team observing the search practices of a top-ranked American police department. Their results indicated that 30% of the 115 searches violated the Fourth Amendment rights of suspects. Additionally troubling was a finding that fully 84% of searches involved African Americans.
Profiling has been used in probation and parole supervision programs. Indeed, Harcourt offers the reader a detailed overview of the relevant research on parole prediction models from 1923 to 1978, including summaries of the various actuarial methods applied. He notes that the application of risk assessment tools to aspects of criminal justice is growing. For example, the Level of Services Inventory – Revised (LSI-R) was developed in Canada in the 1970s and is widely applied across the U.S. The Discriminant Index Function (DIF) tool is used by the U.S. Internal Revenue Service to identify audit targets. Risk assessment instruments are widely applied for making sentencing decisions. Of course, a risk management mentality places a premium on risk avoidance. Avoidance in a criminal justice context logically leads to a desire to optimize incapacitation. It is, therefore, not surprising that sentencing and parole decisions rely heavily on predictive methods that submerge the individual beneath layers of actuarial formulae.
What is immediately fascinating to a reviewer grounded in Canadian policing is the serious attention paid to racial profiling. This fascination goes well beyond Professor Harcourt’s academic interest in this topic as an illustration of the invidious effects of actuarial practices on police methods. The operating reality in the United States is such that more than four hundred American law enforcement agencies have implemented data collection policies and practices that seek to track potential disparities in routine highway patrol searches. In spite of methodological complications and conundrums which accompany these undertakings, American police leaders are, apparently, seized with the notion that this is an appropriate response to concerns about disparities in the rate at which minority drivers are stopped and searched. Indeed, Northeastern University is the home of the Racial Profiling Data Resource Center which provides academic and practitioner research on this pivotal public policy issue. The U.S. reality is remarkable from a Canadian perspective. Indeed, the Kingston (Ontario) Police remain the only police service in this country to have even experimented with maintaining statistics on the racial background of individuals stopped under routine circumstances by its officers. The “problem” of racial profiling in policing can only be properly addressed from an operational or policy perspective when Canadian criminal justice leaders accept the fundamental need for collecting relevant data. Last year, a study was released by the U.S. National Council on Crime and Delinquency reporting that blacks in Wisconsin were imprisoned at nearly 20 times the rate of young whites in 2002. Focused Canadian research in these areas is sadly lacking in spite of sporadic efforts to redress this important gap.
What makes Harcourt’s publication so critically important for policy-makers, police, and criminal justice practitioners relates to the slipstream effect that a prejudice for actuarial approaches has for advancing racial profiling as a theory in practice. The author is genuinely troubled by the penetration of actuarial methods within the justice arena. He sees what amounts to an overthrowing of clinical practice by scientific approaches that emphasize efficiency, predictability and statistical rigour. Harcourt laments our tendency to become “slaves of probability.” The University of Chicago appears to be the locus and centre of gravity for a movement to supplant clinical expertise with statistical efficiency. Harcourt cites the fortuitous nature of actuarial instruments and their lack of any normative connection with criminal law and seeks to apply randomization as a necessary counterpoint for a “risk society” in an actuarial age.
His final chapter defends the use of randomization as the exclusive manner by which the criminal justice system can achieve a carceral population that truly reflects the actual offending population. Of course, several commentators have ridiculed the concept of random sampling when applied to the highly emotive issues of crime and disorder. However, Harcourt immunizes himself against the derisive dismissal of “junk science” which has often accompanied efforts at racial profiling data collection. Rather, he asserts that randomization may be used in precise ways to achieve ethical results. Random sampling is preferable to actuarial instruments of prediction because it preserves a closer alignment with the “real” offending population.
This publication offers valuable insight and argument for academics and policy-makers working at various corners of the criminal justice system. It is particularly instructive for those engaged in policing and public safety. The law and order stance that places a premium on targeted policing and tough punishment might be moderated when assessing the social disorganization that Harcourt alludes to in this work. The complete absence of any serious offender re-entry programs in Canada is yet another symbol that we have failed to make progress in emulating the efforts mounted in the U.S. to avoid being abject “slaves of prediction.” The hypothesis that any form of criminal profiling may fail to address the fundamental law enforcement goal of preventing crime is worthy of careful consideration. Indeed, Harcourt argues that actuarial methods may, in certain circumstances, actually produce an increase in crime. This work offers forceful insights into ways to escape the dominance of actuarial methods that are exogenous to criminal justice and return to more balanced practices of policing and punishing.
Paul F. McKenna, Dalhousie University.
TOP OF PAGE / HAUT DE LA PAGE
GENERAL INFORMATION
ON BOOK REVIEW / INFORMATIONS GÉNÉRALES SUR LES RECENSIONS
BOOK RECEIVED FOR REVIEW - LIVRES REÇUS
LIST OF BOOK REVIEWS / LISTE DES RECENSIONS
MAIN MENU / MENU PRINCIPAL
(2008-05-31)