Jeffrey A. Butts and Vincent Schiraldi
Recidivism is not a robust measure of effectiveness for community corrections agencies. When used as the sole measure of effectiveness, recidivism misleads policymakers and the public, encourages inappropriate comparisons of dissimilar populations, and focuses policy on negative rather than positive outcomes. Policymakers who focus on recidivism as evidence of justice effectiveness are confusing a complex, bureaucratic indicator of system decision-making with a simple measure of individual behavior and rehabilitation. Recidivism is at least in part a gauge of police activity and enforcement emphasis and, because of differential policing practices in minority communities, using recidivism as a key measurement may disadvantage communities of color. Relying on recidivism defines the mission of community corrections in law enforcement terms, relieving agencies of their responsibility for other outcomes such as employment, education, and housing.
In the following discussion, we describe the logical and practical problems that arise when recidivism is used as the principal outcome measure for community corrections agencies.
We recognize that recidivism will always be a feature of justice policy and practice. Recidivism offers a simple and familiar outcome measure for judging the effectiveness of justice interventions. Pointing out the logical flaws of recidivism will not diminish its salience for audiences disinclined to question its utility. Our purpose in this discussion is not to end the use of recidivism as a justice system measure but to illustrate its limits and to encourage the development and use of more suitable measures — namely, positive outcomes related to the complex process of criminal desistance.
Recidivism has long been a central concept in the assessment of justice policies and the evaluation of justice-related programs. Policymakers rely on recidivism to gauge the success of crime reduction efforts. Researchers test the value of crime control strategies by comparing their association with recidivism. The public and the media are accustomed to hearing about recidivism in popular discussions about crime prevention — i.e., “What’s the recidivism rate?” Asking about recidivism seems obvious and natural, as if one were asking, “Does this work?”
Recidivism may be defined as the “tendency to relapse into a previous condition or mode of behavior” (Merriam-Webster, 2017). Researchers have measured it in varied ways. Some studies define recidivism as any new arrest of an individual following justice intervention. Others measure it with new prosecutions or subsequent convictions. There are many approaches, some more complicated than others. Researchers may draw on several decades of research to fashion recidivism measures (Maltz, 1984; Schmidt and Witte, 1980, 1988; Waldo and Griswold, 1979). A casual reader of criminal justice research would not be faulted for assuming that recidivism — along with the general incidence of crime — is a foundational metric for public safety.
We argue that this is unwise. Recidivism is not a comprehensive measure of success for criminal justice in general or for community corrections specifically. When used to judge the effects of justice interventions on behavior, the concept of recidivism may even be harmful, as it often reinforces the racial and class biases underlying much of the justice system. We encourage justice systems to rely on more flexible and more responsive outcome measures. Community corrections agencies should encourage policymakers to rely on outcomes related to criminal desistance and the social integration of people on probation or parole. Measures focused on social development and community wellbeing are more useful for evaluating the effects of justice interventions, and they are less likely to distort policy discussions.
The Meaning of Recidivism
When researchers address the appropriate uses of recidivism, they concentrate on official data from the criminal justice system, thereby obscuring the social-structural, racial, and economic biases embedded in the justice process and built into the very notion of recidivism (King and Elderbroom, 2014; Zara and Farrington, 2016).
In his foundational discussion of recidivism, Maltz (1984:1) described recidivism as resulting from a “concatenation of failures”:
failure of the individual to live up to society’s expectations — or failure of society to provide for the individual; a consequent failure of the individual to stay out of trouble; failure of the individual, as an offender, to escape arrest and conviction; failure of the individual as an inmate of a correctional institution to take advantage of correctional programs — or failure of the institution to provide programs that rehabilitate; and additional failures by the individual in continuing in a criminal career after release.
Researchers have estimated recidivism using a variety of techniques, from the simplest binary measures (either someone did or did not recidivate), to relative trajectories (the pace of offending following intervention), and even “failure rate” or “hazard rate” models that account for the rate of increase and the overall frequency of recidivism events over time (Stollmack and Harris, 1974). The seriousness of subsequent offending is included n some definitions of recidivism as well (Maltz, 1984). A study that considers recidivism to mean any instance of subsequent offending or any legal action after a previous offense will report a higher prevalence than will a study that counts only subsequent offenses above a certain level of severity.
Traditionally, justice data are collected at the individual level and then combined to calculate what many practitioners call the “recidivism rate” of a group or population. The federal Bureau of Justice Statistics (BJS) publishes recidivism rates for cohorts of former prisoners and community supervision populations. Among the 2005 cohort, for example, data were collected for nearly 43,000 people placed on federal community supervision. The BJS analysis found that 18 percent had been arrested at least once within a year of being placed on community supervision. By the fifth year, more than two in five had been arrested at least once (Markman et al., 2016).
The prevalence of recidivism is naturally higher for populations with lengthier criminal careers and deeper penetrations into the justice system. It would be foolish to compare two recidivism figures without accounting for the population base. Whereas a one-year recidivism rate among first-time probationers may be 15 percent, the same figure for state prison inmates would usually exceed 50 percent. Obviously, this does not mean that probation is three times more effective than prison at curbing recidivism; the different numbers reflect the different populations.
Just as it would be inappropriate to compare recidivism for people coming out of prison with those supervised on probation, it could be deceptive to compare recidivism among the clients of different community programs. If the clients of one program differed from the other on any variables possibly related to recidivism (e.g., age, prior record, most serious offense ever, extent of drug use, schooling, employment history, social class, and race), it would be unfair to assess the relative effectiveness of both programs using a simple, common recidivism measure. The BJS data in figure 1, for example, show that post-release recidivism is related to age and prior record, respectively.
The notion that different populations exhibit different recidivism probabilities creates unanticipated consequences for service providers. The most successful probation clients, for example, should be eligible for early discharge from supervision, but they are often the most tractable clients to supervise. They generally pay their fees, come to appointments on time, and comply with probation requirements. These same characteristics make them less likely to experience subsequent justice contacts. Their presence among a probation population would lower an agency’s overall recidivism numbers (and improve its collection of fees). When community corrections programs are held strictly to account for client recidivism, it would be to their advantage to devise ways to hang on to their best performers by extending the length of supervision for low-risk clients.
Similarly, it would be wrong to compare recidivism in two states or cities without examining possible differences in the populations being compared or in the handling of offenders by each justice system. What if the first state rarely incarcerated offenders with a single prior conviction while the second state did so routinely? Post-release recidivism in the second state would undoubtedly be lower, but not because the correctional system in that state was more effective.
The base arrest rate for violent and property crimes in Memphis, Tennessee, for example, is more than three times higher than the arrest rate in New York City (Greene and Schiraldi, 2016). Comparing recidivism outcomes for probation agencies in Memphis and New York City using arrest data could be misleading. Comparing virtually any group of states or cities with simple, aggregate recidivism figures is inherently misleading and should constitute statistical malpractice. Yet, policymakers are routinely encouraged to do this by advocacy groups and professional justice organizations that publish reports and maps comparing aggregate recidivism figures at the state level.
Even analyzing recidivism in one jurisdiction over time may be problematic. Law enforcement, prosecutorial, and judicial practices change, and those changes affect recidivism. If law enforcement or the courts gradually became more lenient or more punitive in a city or state, the difference would be reflected in the recidivism of community corrections clients. In New York City, for example, misdemeanor drug arrests plummeted 50 percent between 2011 and 2015 following litigation, media attention, and public complaints about the police department’s “stop and frisk” policies (Greene and Schiraldi, 2016). A researcher could be making a serious error if he or she used drug arrest rates to evaluate the effectiveness of a New York program launched in 2012.
Using recidivism to assess the effectiveness of justice programs also presumes that justice interventions are designed primarily and explicitly to prevent crime. Certainly, justice systems intervene to prevent crime, but they pursue other objectives as well. Some policies in the justice system, in fact, are designed purely for retributive or punitive purposes without regard for their behavioral effects. In other words, the justice system punishes some people simply because they “deserve it” and not because punishment is expected to change behavior. Other programs in the justice system are designed to employ or educate people under community supervision. If the programs achieve those goals, they should be considered successful. Why should the effectiveness of all justice policy be judged according to individual recidivism outcomes when the goals of justice programs are more varied?
Recidivism as an Organizational Product
There is no perfect way to measure recidivism, but it is generally defined as a person’s return to crime following some form of intervention. This simple definition reveals several underlying problems: What must happen before someone can be said to have “returned to crime?” Who decides what constitutes “crime” in the first place? How can we know someone has “returned” to it? Some crime concepts evolve over time. Using cannabis for recreational purposes is now legal in many states. Other states have downgraded some crimes from felonies to misdemeanors. A comparison of recidivism in states with differing laws — or in single states before and after a change in laws — would have to adjust for such differences.
Moreover, it is never possible to detect all instances of recidivism. State and local governments in the U.S. do not have (and hopefully will never have) perfect data about all crimes. We do not observe the behavior of every individual in every community at every minute of the day. Thus, we can never know whether a person has committed a crime until the criminal act has been observed or reported.
Before the justice system may record an act of recidivism and attribute it to a specific person, several events must occur. First, a law violation must take place (except, of course, in cases involving false arrest, erroneous conviction, or noncriminal, technical violations of community supervision). Second, the violation must come to the attention of justice authorities in some way, either through citizen reports or direct observation. Third, an individual suspected of committing the violation must be identified, apprehended, and — for some recidivism measures — convicted and sentenced.
All data used to measure crime pass through the filter of justice bureaucracies — from the law enforcement agencies that receive citizen reports of crime and investigate those suspected of committing offenses, to the courts that examine the evidence in a prosecution before determining guilt and imposing punishment, to the probation departments, prisons, and other agencies that administer sanctions and services. Other than victimization surveys that generate population-level estimates rather than individual outcomes, all data used to measure crime are work products from the organizations that respond to crime. Some bureaucratic process involving human decision-making is required before the inherently unmeasurable act of crime may be defined as an occurrence of recidivism.
When policymakers use recidivism outcomes to judge the effectiveness of crime-reduction strategies, they fail to account for the bureaucratic contribution to recidivism. How many people are involved in the sequence of decisions required to qualify an outcome as recidivism, and what are their beliefs about the justice system and the individuals caught up in it? The answers vary from place to place, from time to time, and from case to case, depending on individual characteristics and social context.
Traffic infractions offer a clear example. The number of traffic citations written by a police department may be a useful measure of enforcement actions, but communities would never use the number of citations as a metric for judging actual improvements in traffic safety. Traffic citations are an imperfect reflection of the prevalence of moving violations by the driving public. The volume of traffic stops is obviously influenced by the rate of infractions among driving public, but it is also shaped by the distribution of police resources devoted to detecting infractions. If a police department increased the number of patrol cars along a roadway, or if it mounted new cameras at specific locations, the number of tickets from those locations would undoubtedly increase. The likelihood of any individual being stopped and ticketed would grow as well, even if the behavior of drivers had not changed. Similarly, if a police department moved more patrol cars to the west side of town, drivers who frequented the east side of town would have a lower probability of being ticketed, even if they committed violations just as often as west-side drivers. It might appear logical to judge the impact of a newly mandated traffic safety course by counting infractions (i.e., recidivism) among drivers following their completion of the course. However, an accurate test of effectiveness would first have to account for the base probability of infractions among drivers and the extent to which their chances of being ticketed were influenced by area of residence, daily schedule, type of car, and other easily observable characteristics that might increase their chances of being stopped, including race, class, age, gender, and perhaps the genre of music blasting from their car stereos.
An individual’s risk of contact with the justice system depends in part on the allocation of justice resources across communities and the varying social contexts in which resources are deployed. The chance of receiving a punitive response from the justice process after an initial contact depends in part on the availability of alternatives. Affluent communities enjoy more discretion to divert individuals from official processing because they have more worthwhile alternatives for police and courts to rely on in making diversion decisions. Alternatives are less available in neighborhoods of concentrated disadvantage, which are more likely to be communities of color and where officials in the justice system may be more likely to act with unconscious racial bias (Goff et al., 2016).
Individuals’ personal resources and attitudes may also affect how they are handled in the justice system. Suspects who are disrespectful and contemptuous of legal authority and those who abide by the “code of the street” are more likely to find themselves arrested and treated harshly by the justice system (Mears et al., 2016). Suspects willing to appear submissive and polite to authority figures, on the other hand, are more likely to be warned than arrested, more likely to be offered services rather than sanctions, and more likely to be treated in the community rather than incarcerated. Recidivism is not a sanitized measure of individual behavior; in part, it is a measure of how individuals are perceived when they come into contact with legal authorities.
The Sampling Effect of Official Data
Reported crimes are a sample of actual crime but not necessarily a representative one. Some illegal acts are observed directly by law enforcement, but most crimes must be reported by victims or witnesses. Depending on the nature of an offense, the location of an offense, and the inclination of residents to trust the police and report crimes, there may be large differences between the volume of criminal behavior and the number of criminal acts that come to the attention of legal authorities. The sampling effect is clear when official justice data are compared with self-reported data in those few instances where self-reported data exist.
Estimates of self-reported delinquency from the annual Monitoring the Future (MTF) study suggest that half of all teenagers have done something in the past year that could have resulted in an arrest (Miech et al., 2016). In a recent MTF report, 28 percent of 10th graders (typically 15-year-olds) admitted they had used an illegal drug at least once in the previous 12 months. The resident population of 15-year-olds in the United States is approximately 4 million. Thus, there could be more than 1 million drug arrests of 15-year-olds each year if 28 percent of those 4 million youths used illegal drugs and all their drug use was reported or observed and all drug laws were applied consistently and rigorously. According to data from the FBI’s Uniform Crime Reporting (UCR) Program, however, police nationwide make just 20,000 drug arrests involving 15-year-olds in a year (Federal Bureau of Investigation, 2015), which means the justice system handles only 2 percent of all possible drug crimes involving 15-year-olds. How many individual and socio-structural factors are involved in a sequence of actions leading to the arrest of just 2 percent of all possible candidates for arrest?
The rates of drug use versus drug arrests also differ dramatically by race. African American and white youth report similar rates of illegal drug use, but UCR data show that African American juveniles are 40 percent more likely than white youth to be arrested for drug offenses. There are many theories to explain the existence of such a difference but, for the purposes of this discussion, the most important point is that a stark discrepancy exists (Welty et al., 2016). With such an incongruity between reported drug use and drug arrests, using rearrest as a straightforward measure of individual behavior could inflate or deflate actual recidivism.
The situation is similar for all offenses to varying degrees. The National Survey on Drug Use and Health recently estimated that 4 percent of all 15-year-olds carried a handgun at least once in the previous year (Center for Behavioral Health Statistics and Quality, 2015). Thus, as many as 160,000 15-year-olds could be arrested for gun charges each year. Yet, the FBI (2015) reports only 4,000 annual arrests of 15-year-olds for all types of weapon offenses. Again, this suggests that police handle just 2 to 3 percent of all 15-year-olds who could be arrested for weapons.
Once an offense has been committed, the odds of justice intervention are low, and they vary depending on the offense, the resources available to process the offense, the personal characteristics of the suspect, and the neighborhood context of all the individuals involved in the offense. Serious offenses, of course, are more likely to be reported to law enforcement, but even reported crimes are not always “cleared” by the arrest of a suspected perpetrator. According to FBI data (2015), just under half of all violent crimes and less than 20 percent of serious property crimes will ever result in an arrest. What more would we know about recidivism if all crimes were reported and all offenders were identified?
The probability of an individual crime becoming an official instance of recidivism depends on many factors, including the social-cultural and political-economic environment, the resources available to sustain the sequence of organizational actions ending in recidivism, and the perceptions, beliefs, and biases of the human beings responsible for operating the justice system. The illegal behaviors of individuals are usually a necessary impetus for some interaction with the justice system (arrest, prosecution, and sentencing), but recidivism is inherently a measure of person-bureaucracy interactions. It is not simply an indicator of individual failure. Thus, it would be inappropriate to place the onus for recidivism entirely and exclusively on the individual (Kubrin and Stewart, 2006; Mears et al., 2008). Recidivism is a useful indicator of justice operations and the interactions of justice systems and individuals, but it is not a pure measure of community safety or individual rehabilitation.
So, What’s the Alternative?
Fortunately, there are alternatives to recidivism for assessing the effectiveness of community corrections. The first step is to reorient the goal of intervention to supporting desistance rather than preventing recidivism (Kazemian, 2015). In a desistance framework, crime reduction is viewed as a complicated change process in which individuals learn to be law abiding over time. Recidivism is a binary frame: People either succeed or they fail. Desistance allows for degrees of success even if there are occasional setbacks. One misdemeanor committed by a former armed robber with multiple prior offenses would be an instance of recidivism, but it might also be an indicator of progress toward eventual desistance.
The difference is more than rhetorical. Focusing on desistance instead of recidivism leads justice systems to reorient their operations and their measurement of success. A desistance framework encourages just ice agencies to promote and monitor positive outcomes. The British government recently published a comprehensive review of research literature about desistance (Ministry of Justice, 2014). The report asked the question, “What helps individuals desist from crime?” The research literature identified nine critical factors:
1. Getting older and maturing
2. Family and relationships
5. Hope and motivation
6. Having something to give to others
7. Having a place within a social group
8. Not having a criminal identity
9. Being “believed in”
Some of these factors would be difficult or expensive to measure, but a justice system that tracked them consistently would inevitably pursue a different intervention regime for justice-involved individuals. Sobriety and employment are already a target of community corrections agencies, but an agency focused on desistance would view such issues from an asset-based perspective rather than a deficit-based perspective. Asking probation officers to focus on “family and relationships” and “having a place within a social group” would revive the social work heritage of community corrections (as opposed to its modern law enforcement orientation) and create meaningful points of intervention. Measuring access to these desistance-promoting factors would help to redefine the role of community corrections. Rather than focusing their time on monitoring compliance and imposing punishments, probation workers would naturally concentrate on supporting positive changes and achieving success.
The developmental approach provides another compelling alternative to the recidivism regime — at least for adolescents and young adults. When healthy, pro-social development is viewed as a natural antidote to the normative tendency of youth to take risks and engage in illegal behavior, the justice system instinctively focuses on promoting desistance rather than suppressing recidivism (National Research Council, 2013). In the District of Columbia, for example, the Department of Youth Rehabilitation Services (DYRS) rejected a purely deficit-based approach to intervention and embraced the positive youth justice (PYJ) model (Butts, Bazemore, and Meroe, 2010).
The PYJ model encourages youth justice systems to focus on protective factors, strengths, and pro-social skills. The goal of intervention for youth is to facilitate their successful transition to adulthood, not only to reduce law enforcement contact. The PYJ model (see figure 2) includes 12 key components, depicted as a 2 x 6 matrix. Each cell in the matrix represents the interaction of two core assets needed by all youth (learning/doing and attaching/belonging) with six separate life domains (work, education, relationships, community, health, and creativity). To assess the effectiveness of youth justice, the model suggests that state and local governments measure activities and outcomes within each of the 12 combinations of assets and practice domains.
The Council of Juvenile Correctional Administrators (CJCA) has embraced the developmental approach to youth justice interventions (Harris, Lockwood, and Mengers, 2009). As the preeminent trade association for juvenile justice department leaders, CJCA collaborated with the federal Office of Juvenile Justice and Delinquency Prevention to reconsider the meaning and uses of recidivism. Inspired by the positive youth justice literature and the strengths-based approach of the District of Columbia’s DYRS (a member agency), CJCA created a Positive Youth Outcomes committee to embed the key elements of education and work, social connectedness, and health and well-being in measures of youth justice effectiveness. CJCA hopes that its members begin to track positive outcomes as systematically as they have tracked youth recidivism.
Even organizations that have traditionally encouraged policymakers to focus on recidivism have begun to acknowledge the need for a more diverse set of outcome measures (Seigle, Walsh, and Weber, 2014).
The criminal justice sector is increasingly aware that recidivism is insufficient for measuring the effectiveness of community corrections interventions on individuals or for assessing community well-being. As an outcome indicator, recidivism is subject to at least three significant limitations.
First, recidivism is not generated solely by the behavior of justice-involved individuals. It is at least partly a reflection of the supervision efforts of probation or parole agencies, as well as the intensity and consistency of policing. This is particularly true in the treatment of technical violations that may result in a probation/parole revocation and possibly incarceration. In community corrections, recidivism may be an indicator of the scale and intensity of surveillance and of the ability of people to keep up with payments ordered as financial punishment (Harris, 2016).
Second, the incidence of recidivism is influenced by situational factors and a myriad of forces reflecting the vulnerability of those under criminal justice supervision. People under correctional supervision in many communities are highly “arrest-able.” Complying with the conditions of supervision can be very difficult, especially in poor, high-crime neighborhoods. Such neighborhoods are policed intensively, and everyday activities such as traveling in cars, being in the proximity of a crime scene, or being subject to a consent search may result in arrest. Given that the scale and intensity of surveillance are likely to be especially high in minority neighborhoods and neighborhoods of concentrated disadvantage, focusing on recidivism will tend to promote racially disparate justice system impact. Simple comparisons of recidivism may fail to account for differences in community and social context and how they interact with individual factors (Sharkey and Faber, 2014).
Third, recidivism is a crude indicator of a complex and varied process. Community corrections workers engage with clients on issues such as employment, education, housing, the need for allied services such as substance abuse and mental health supports, and efforts to reconnect with family members and positive peers. Community corrections workers have substantial influence over, and can facilitate, client access to various types of services, supports, and opportunities. As such, each domain that may contribute to desistance and client success should be tracked using a variety of strategies. Measures of client progress cannot be assessed with a simple, binary metric like recidivism.
Fourth, recidivism is a noisy signal of new criminal conduct, but it does not lead to any new information that could be used to shape an effective response to that conduct. Outcome measures should not only identify problems but also should promote better solutions. Relying exclusively on comparative recidivism data to determine the need for additional justice sanctions comes close to satisfying that popular definition of insanity: repeating the same action while expecting different results.
We recommend several changes in practice and policy to reduce the justice system’s reliance on recidivism as a measure of public safety outcomes and, instead, position it as one measure among a range of measures.
1. Insist That Recidivism Comparisons Involve Appropriately Matched Groups
Policy makers and the public must learn that comparing recidivism results between dissimilar groups and populations is misleading and harmful. It will take time to shift public discourse about crime prevention measures, but researchers and practitioners must collaborate to ensure that all conversations about recidivism include essential, clarifying questions. Namely, how might this recidivism figure be different if the data were about individuals from another community, at a different time, and in a different social, political, cultural, and economic context? Essentially, the response to all claims about recidivism outcomes should be, “compared to what?” or “compared to whom?” Whenever the answer is “the state average” or “the national average,” or indeed any other “average,” the recidivism claim should be dismissed out of hand. Asking the questions can have several useful outcomes for policymakers looking to improve public safety. It could reduce the perverse incentives for programs and community corrections agencies to “hang on to” low-risk good performers, wasting resources and unnecessarily depriving people of their full liberty who no longer need supervision. It could also encourage programs to tackle more challenging clients with greater needs (and higher risk) if those programs know that they will be fairly evaluated for assisting people with thornier problems.
2. Use Other Measures to Assess the Effectiveness of Justice
The justice system should monitor and assess how people are reintegrated into a community following system contact. Rather than asking only, “What’s the recidivism rate?” policymakers and the public should learn to ask, “What’s the graduation rate, what’s the employment rate, and how many are now living independently?” If justice systems are to be truly correctional, whether in the community or not, policymakers should begin to hold them to more rigorous standards, including asking systems to measure what they promise to produce and not merely what they try to avoid. This would aid policymakers in ensuring that they are getting the most from limited resources.
3. Increase the Policy Salience of Desistance
Justice policy should reduce the importance of recidivism and focus on desistance. If community corrections programs were designed to facilitate desistance rather than simply combating recidivism, they would naturally focus their efforts on maximizing skills, strengths, and positive assets. This would have important implications for policy and practice. It would also require more ambitious data collection and analysis. Tracking desistance outcomes would have to involve more than law enforcement data. It would require ongoing contact with people on probation or parole after intervention, using repeated interviews or surveys that investigate attitudes, perceptions, and beliefs. Even a sample-based approach to collecting this information would be costly, but the investment could help to change the public conversation about crime, justice, and public safety. Focusing on a suite of outcome measures could help community corrections agencies to diversify their intervention approaches as well. Measuring positive outcomes inspires staff to pay more attention to connecting clients with services, supports, and opportunities that facilitate desistance, employing the familiar adage that “what you measure is what you get.”
Despite promising research on the potential for desistance-focused approaches to improve outcomes, community corrections agencies continue to rely on recidivism as the primary measure of their effectiveness. Policymakers and practitioners may grow impatient with researchers who incessantly criticize existing outcome measures and warn of measurement error and sample bias. Facing endless discussions of what appears to be methodological trivia, weary policymakers find comfort in the simple and easily interpretable measure of recidivism. Researchers and practitioners should work together to make policymakers less comfortable with their reliance on recidivism. Once the criminal justice field accepts the fact that simple comparisons of recidivism generate inaccurate, harmful, and often discriminatory conclusions, it may finally begin to make real progress in replacing recidivism with more flexible and positive outcome measures.
Butts, J.A., G. Bazemore, and A.S. Meroe (2010). Positive Youth Justice: Framing Justice Interventions Using the Concepts of Positive Youth Development. Washington, DC: Coalition for Juvenile Justice.
Center for Behavioral Health Statistics and Quality (2015). Behavioral Health Trends in the United States: Results from the 2014 National Survey on Drug Use and Health. Washington, DC: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration. HHS Publication No. SMA 15-4927, NSDUH Series H-50.
Federal Bureau of Investigation (2015). Crime in the United States 2014. Washington, DC: U.S. Department of Justice.
Goff, P.A., T. Lloyd, A. Geller, S. Raphael, and J. Glaser (2016). The Science of Justice: Race, Arrests, and Police Use of Force. Los Angeles, CA: Center for Policing Equity.
Greene, J.A., and V. Schiraldi (2016). “Better by Half: The New York City Story of Winning Large-Scale Decarceration while Increasing Public Safety.” Federal Sentencing Reporter 29(1):22-38.
Harris, A. (2016). A Pound of Flesh: Monetary Sanctions as Punishment for the Poor. New York, NY: Russell Sage Foundation.
Harris, P.W., B. Lockwood, and L. Mengers (2009). CJCA White Paper: Defining and Measuring Recidivism. Braintree, MA: Council of Juvenile Correctional Administrators.
Kazemian, L. (2015). Straight Lives: The Balance between Human Dignity, Public Safety, and Desistance from Crime. The City University of New York, John Jay College of Criminal Justice, Research & Evaluation Center.
King, R., and B. Elderbroom (2014). Improving Recidivism as a Performance Measure. Washington, DC: Urban Institute.
Kubrin, C.E., and E.A. Stewart (2006). “Predicting Who Reoffends: The Neglected Role of Neighborhood Context in Recidivism Studies.” Criminology 44:165-197.
Maltz, M.D. (1984). Recidivism. Orlando, FL: Academic Press.
Markman, J.A., M.R. Durose, R.R. Rantala, and A.D. Tiedt (2016). Recidivism of Offenders Placed on Federal Community Supervision in 2005: Patterns from 2005 to 2010. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics. NCJ 249743.
Mears, D.P., E.A. Stewart, P.Y. Warren, and R.L. Simons (2016). “Culture and Formal Social Control: The Effect of the Code of the Street on Police and Court Decision-making.” Justice Quarterly 34(2):217-247.
Mears, D.P., X. Wang, C. Hay, and W.D. Bales (2008). “Social Ecology and Recidivism: Implications for Prisoner Reentry.” Criminology 46:301-340.
Merriam-Webster Dictionary (2017). Definition of “recidivism.”
Miech, R.A., L.D. Johnston, P.M. O’Malley, J.G. Bachman, and J.E. Schulenberg (2016). Monitoring the Future: National Survey Results on Drug Use, 1975-2015: Volume I, Secondary School Students. Ann Arbor, MI: University of Michigan, Institute for Social Research.
Ministry of Justice (2014). Transforming Rehabilitation: A Summary of Evidence on Reducing Reoffending (2nd edition). London, England: Ministry of Justice Analytical Series.
National Research Council (2013). Reforming Juvenile Justice: A Developmental Approach. Committee on Assessing Juvenile Justice Reform. R.J. Bonnie, R.L. Johnson, B.M. Chemers, and J.A. Schuck (Eds.). Committee on Law and Justice, Division of Behavioral and Social Sciences and Education. Washington, DC: National Academies Press.
Schmidt, P., and A.D. Witte (1980). “Evaluating Correctional Programs: Models of Criminal Recidivism and an Illustration of Their Use.” Evaluation Review 4:585-600.
Schmidt, P., and A.D. Witte (1988). Predicting Recidivism Using Survival Models. New York, NY: Springer Verlag.
Seigle, E., N. Walsh, and J. Weber (2014). Core Principles for Reducing Recidivism and Improving Other Outcomes for Youth in the Juvenile Justice System. New York, NY: Council of State Governments Justice Center.
Sharkey, P., and J.W. Faber (2014). “Where, When, Why, and for Whom Do Residential Contexts Matter? Moving Away From the Dichotomous Understanding of Neighborhood Effects.” Annual Review of Sociology 40:559-579.
Stollmack, S., and C.M. Harris (1974). “Failure Rate Analysis Applied to Recidivism Data.” Operations Research 22(6):1192-1205.
Waldo, G., and D. Griswold (1979). “Issues in the Measurement of Recidivism.” In L. Sechrest, S.O. White, and E.D. Brown (Eds.), The Rehabilitation of Criminal Offenders: Problems and Prospects (pp. 225-250). Washington, DC: National Academy of Sciences.
Welty, L.J., A.J. Harrison, K.M. Abram, N.D. Olson, D.A. Aaby, K.P. McCoy, J.J. Washburn, and L.A. Teplin (2016). “Health Disparities in Drug- and Alcohol-Use Disorders: A 12-Year Longitudinal Study of Youths After Detention.” American Journal of Public Health 106(5):872-880.
Zara, G., and D.P. Farrington (2016). Criminal Recidivism: Explanation, Prediction and Prevention. New York, NY: Routledge.
Jeffrey A. Butts, Ph.D., is Director of the Research & Evaluation Center at John Jay College of Criminal Justice at The City University of New York.
Vincent Schiraldi, M.S.W., is a Senior Research Scientist at Columbia University School of Social Work and Co-Director of the Columbia University Justice Lab.
The authors are grateful for the support and assistance provided by members and staff of the Executive Session on Community Corrections, especially Ana Yáñez-Correa, Bruce Western, Kendra Bradner, and Rachel Corey.
This paper was prepared with support from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice, under contract number 2012-R2-CX-0048. The opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors and do not necessarily represent those of the Department of Justice.
Butts, Jeffrey A. and Vincent Schiraldi (2018). Recidivism Reconsidered: Preserving the Community Justice Mission of Community Corrections. Cambridge, MA: Program in Criminal Justice Policy and Management, Kennedy School, Harvard University.