Prisoner Reentry, Employment, Signaling, and the Better Identification of Desisters

Introduction to the Special Issue

Prisoner reentry remains one of the most critical social problems confronting America, one that brings with it many social and economic costs. The mass incarceration that has occurred in recent decades (Clear, 2007; Gottschalk, 2010; Pratt, 2009) has created a situation in which more than 1.6 million individuals currently are in state and federal prisons (West, 2010) and approximately 730,000 inmates are released annually (West, Sabol, and Greenman, 2010: 4). Estimates suggest that more than two thirds will be rearrested within 3 years of release (Langan and Levin, 2002). That, however, only skims the surface of the potential problems associated with reentry. Recidivism estimates based on official records data invariably understate the true amount of offending committed by ex-prisoners. In addition, recidivism constitutes but one dimension of relevance. Others include unemployment, homelessness, drug abuse, physical illness, and mental disorders (Mears and Barnes, 2010; Petersilia, 2003). Regardless of one's political leanings, the fact that ex-prisoners face challenges upon release that contribute to these adverse outcomes creates substantial cause for concern for the well-being of these individuals and, even more so, for the communities to which they return (Clear, 2007; Lattimore, Steffey, and Visher, 2010; Travis, 2005).

Over the past decade, policy makers and researchers have responded to the problem through a plethora of new policies and programs and through an increasingly large body of research on prisoner reentry. There is cause for being optimistic. Today, more is known about the characteristics of ex-prisoners, their experiences after release, and the types of programs and policies that may improve the likelihood of successful reentry (see, e.g., Cullen and Gendreau, 2000; Lattimore et al., 2010; Mears, 2010; Travis, 2005; see, generally, Wilson and Petersilia, 2011, and the recent special 2011 issue of the Prison Journal). Even so, many basic questions remain. There is, for example, the overarching question of whether incarceration, or lengthier incarceration, reduces recidivism or crime rates (Blumstein and Wallman, 2006; Cullen, Jonson, and Nagin, 2011; Nagin, Cullen, and Jonson, 2009). There also are important questions about the “black box” of prison experiences and their salience for the transition to society (Lattimore et al., 2010; Mears, 2008). What, for example, happens to inmates in prison and upon release, and how do these experiences affect reentry outcomes?

Successful efforts to promote large-scale improvements in prisoner reentry no doubt require more information about effective programs and policies. However, such efforts depend heavily on accurate information about the factors that give rise to the types of successful reentry experiences that benefit ex-prisoners and, in turn, society. It is here, the identification of what happens in prison and after and what it can tell us about the prospects for successful reentry (Visher and Travis, 2003), where there exists considerable room for advances. As but one example, despite the potentially central role of inmate social ties to families, friends, and others in their home communities, there are few rigorous assessments of inmate visitation during and after incarceration (Bales and Mears, 2008; Berg and Huebner, 2011; Hochstetler, DeLisi, and Pratt, 2010).

Prior scholarship points to a great deal of heterogeneity in the inmate experience and how it may influence inmate behavior (Adams, 1992; Bottoms, 1999; Maruna, 2001; Mooney and Daffern, 2011). In addition, desistance scholarship (see, e.g., Maruna, 2001; Piquero, Farrington, and Blumstein, 2003), many nuanced accounts of offending (e.g., Laub and Sampson, 2003), and research on criminogenic risks, inmate needs, and treatment responsivity (Andrews, Bonta, and Wormith, 2011) all underscore the notion that a diverse array of factors may contribute to the recidivism of any particular individual. To make matters even more complicated, there is also the potential for family and community contexts of various types to influence recidivism and other reentry outcomes (Kubrin and Stewart, 2006).

Against this backdrop, there is a clear and compelling need for developing a better understanding of who is likely to have a successful reentry experience, who is not, and why. Risk and needs prediction efforts have improved greatly in recent years, but there is much room for more progress, both in predicting recidivism and other outcomes. To be clear, better predictions can only take us so far. But without them, programs and policies will be less able to target scarce resources in as cost-efficient a manner as they otherwise might.

The Importance of the Special Issue Topic

The lead policy article by Bushway and Apel (2012, this issue) constitutes a new and interesting way to think about predicting future offending among released prisoners. Of course, many approaches arguably pass the test of being “interesting.” Their article, however, provides a potentially helpful way to advance efforts aimed at understanding reentry and, more important for the journal, Criminology & Public Policy, to inform discussions about specific ways that reentry can be improved.

Their central argument is that individuals, while in prison or upon release, issue “signals” that let us know that they are unlikely to recidivate. They focus on ex-prisoner employment program participation as a useful signal. But, as noted in the subsequent discussion, there are others that may work as well. No doubt, there is good reason to perhaps wax skeptical about some parts of the Bushway and Apel (2012) argument. At the same time, their article raises questions that should be relevant to scholars and policy makers alike. In this introduction, we detail some of the reasons for viewing the article as an important point of departure for reentry discussions and policy.

First, it draws attention to the need to improve predictive accuracy when attempting to identify individuals who are most at risk of misconduct while incarcerated and of recidivism while in society at large. Considered in this light, the significance of signaling is not just for reentry efforts; it may be useful to officials who wish to run safer and more orderly prison systems.

Second, it highlights that risk prediction can help us to allocate resources in ways that may be more effective. Armed with information about individuals most at risk, we can, for example, focus more attention on those who would benefit from some type of intervention. At the same time, Bushway and Apel (2012) underscore that even under the best prediction scenario there will be substantial error, including many false positives (e.g., individuals identified as high risk who are not) and many false negatives (e.g., individuals identified as low risk who in fact are high risk). This point is, as Gottfredson and Moriarty (2006) and Rhodes (2011) have emphasized, central to discussions of how we identify individuals for interventions. Yet it frequently seems to be omitted from research and policy discussions.

Third, the article raises the possibility of being able to have and use real-time information about recidivism risk, including changes in this risk. Individuals may participate in and complete employment programming, perhaps sending a signal that some change has occurred that makes them less likely to offend. Suppose, though, that they change again and stop working: that may signal a greater likelihood of recidivism. Also, in prisons, individuals may participate in different programs and do so to varying degrees. Having such information may signal their current likelihood of misconduct. Signaling, too, can be used to monitor changes in the likelihood of such misconduct. These possibilities are especially important given that collecting information about program participation on a regular per-inmate, by month, basis need not cost a great deal. By contrast, conducting frequent risk and needs assessments on inmates would be costly.

Fourth, the article by Bushway and Apel (2012) makes a compelling argument for using information about postrelease employment program participation to assess recidivism risk. This idea alone is intriguing, regardless of the other benefits of the article. Of course, the idea of funding employment programs solely to create a mechanism by which ex-prisoners can issue signals about potential desistance may seem to defy common sense. We want employment programming to do more than that. As the essayists in this issue and prior scholarship suggests, there seem to be potential recidivism-reducing benefits of employment programming. Even so, as Bushway and Apel note, the record based on extant studies creates pause for concern. Regardless, the relevance of their idea does not hinge on their focus on employment programming. There may be many other signals that may be “stronger.” Perhaps, for example, how inmates behave in prison, or whether and to what extent they participate in various voluntary (e.g., educational or vocational) programs, may predict better the likelihood of recidivism. Inmates who choose to refrain from misconduct or to choose to participate in nonmandatory programs, especially those that require considerable effort (a central prerequisite for a good signal), may be sending a message that they have changed. In turn, that change may signal that they will continue to engage in prosocial behavior upon release from prison (Mooney and Daffern, 2011).

The inmate visitation example discussed can illustrate the situation. Few inmates are visited, and those who are typically must have acted in ways that require substantial effort (e.g., refraining from the many opportunities to engage in misconduct). Accordingly, it may be that visitation serves as a rough proxy for an inmate's decision to act in ways that may reflect a diminished likelihood of offending. Only empirical research ultimately will be able to adjudicate the matter (see, e.g., Mears, Cochran, Siennick, and Bales, in press). The point is that the idea of signals provides a strategic method with which to identify the need to improve risk prediction efforts and, concomitantly, to show how to do so (e.g., exploring a range of potential signals).

Fifth, Bushway and Apel's (2012) article indirectly draws attention to the need for the study of factors that, at the moment of release, reflect the likelihood of offending and, in a related vein, the study of factors that, at any given time postrelease, reflect the likelihood of offending. For example, inmate visitation or participation in programming may provide strong signals of the likelihood of offending in the days, weeks, and months after release. But they would be “dated” quickly if inmates underwent changes that contributed to a reduced likelihood of offending. Signals such as participation in postrelease programming may be better because they provide “real-time” measures of potential changes in individuals. In short, there is, as the study's argument implies, a need for systematic investigation of in-prison signals and postrelease signals of the likelihood of offending. Whether, as the authors contend, these signals differ fundamentally from various types of risk factors, is in our view less important than the emphasis—from either a signaling or a risk factors perspective—on better identification of individuals who are most likely to offend at any given period in time.

It seems to us that there are many cost-effective opportunities to advance the understanding of how in-prison and postrelease experiences may provide information about the risk of offending. Consider, for example, what happens at some amusement parks. You take a ride, and during it, you answer a series of questions about, say, your personal preferences and wishes for the future. (Admittedly, this experience is unlikely at some older amusement parks.) At the end of the ride, you receive a printout that customizes a future that you might experience based on your answers. Thousands of people take this ride each day, and yet this real-time information can be collected, analyzed, and a prediction provided almost instantaneously.

It seems reasonable to expect that administrators of jails, prisons, and community supervision be allocated resources to compile information from inmates and those who are under supervision about standard risk and needs information and, more broadly, a range of questions that speak to in-jail, in-prison, postrelease behaviors, experiences, and program participation and performance. All of this information might provide insight into the likelihood of recidivism. It also could be used in a real-time manner to adjust for recent behaviors, experiences, and programming that might be predictive. Put differently, it seems imminently possible to create a real-time system of updated prediction that includes signaling data, information now collected and used in validated risk and needs instruments, and information reflecting real-time assessments of an individual's behavior, experiences, and program participation and performance while in prison and after release. The latter information would be especially important for capturing up-to-the-minute assessments of the likelihood of offending. These types of information could be compiled using information that inmates, prison officers, parole officers, and treatment providers provide, perhaps averaging the different sources of information to adjust for potentially distorted perceptions about a particular individual.

The point is that the notion of signaling underscores the need to improve our ability to predict recidivism and other prison and reentry outcomes and that improvements can be made on many fronts. Bushway and Apel's (2012) article is directly relevant, of course, to discussions about employment programs. However, as we have argued here, the broader relevance is the call to arms to improve the prediction of recidivism and other reentry outcomes and to make predictions reflect, to the extent possible, the real-time likelihood of these outcomes among ex-prisoners. That in turn requires collecting data on an ongoing basis during the postrelease period that can allow for detecting potential changes in the likelihood of offending. From this perspective, then, and notwithstanding Bushway and Apel's use of the term “desistance,” the focus turns to real-time monitoring of an individual's risk of offending, whether toward or away from such behavior and regardless of past behavior.

Any such monitoring raises with it attendant ethical concerns. For example, the information might be used to extend the formal social control of the state without in any clear way reducing the recidivism of the individual. Here, again, more accurate information about the full range of factors that contribute to an individual's offending—including the extent to which prisons are operated in ways conducive to rehabilitation; the extent to which community supervision is undertaken in a way that would promote deterrence and, at the same time, linkages to available services; and the extent to which the communities to which individuals return are equipped to facilitate successful reentry—would be helpful. In addition, we would want information on the types of approaches that would most cost-effectively improve outcomes among those who are at risk. Whether such approaches consist of more incarceration or of any of a variety of interventions is another question (see, generally, Cullen et al., 2011; Travis, 2005).

The Essays in this Special Issue

The essays in this special issue examine the signaling argument from different perspectives. Collectively, they provide a coherent and comprehensive assessment of the argument and of its implications for research and policy.

In his policy essay, Alex Piquero (2012, this issue) aims to situate signaling within the broader context of literature on reentry and desistance. Piquero argues that ex-prisoners should be taught to transition gradually back into employment and mainstream society by placing an emphasis on prevention strategies that teach these individuals to respect and obey the law. He and Tim Brennan (2012, this issue) note that it remains unclear the extent to which a signaling approach differs from a risk and need factor approach. Piquero and Brennan also emphasize that there is considerable heterogeneity in employment programming, that there are potentially differential effects of various types of programming, and as Edward Latessa (2012, this issue) observes, that there also may be differential effects depending on the degree to which employment programming is used for the “right” types of ex-prisoners.

Brennan (2012) further underscores the importance of targeting inmates for employment programs. According to Brennan, not all inmates will benefit from employment programs in the same way. He argues strongly that substantial questions exist about the extent to which current research on employment programming is sufficient for drawing conclusions about its ineffectiveness. A central policy issue raised by Brennan and echoed by Latessa (2012) is that we need better ways of identifying target populations to improve the assignment of cases to appropriate programs. More broadly, we need to be able to identify better what types of employment programming work for what kind of ex-prisoners and under what circumstances.

In his essay, Shadd Maruna (2012, this issue) compares the signaling argument to symbolic interactionism and labeling theory and argues that signaling can place rehabilitation in the hands of the inmate by focusing on factors that they can control. He views this partial empowerment of ex-prisoners as potentially helpful in achieving reductions in offending. Maruna highlights the problems inmates’ face in understanding and accepting the traditional risk assessment process. Many feel as though they are being assessed on factors outside of their control and find it difficult to show people they have changed. According to Maruna, signaling may help ex-prisoners make an invisible quality, such as a commitment to “going straight,” visible. At the same time, Maruna draws attention the conceptual challenges in defining and discussing desistance.

Edward Latessa (2012) acknowledges that although many empirical studies find that work and employment programs do not reduce recidivism, the programs should not be ignored and can be potentially important catalysts for desistance. He argues that work can create feelings of self-worth and dignity within an individual; these changes in turn may reduce recidivism. Latessa believes that employment success is more likely if the individuals’ attitude about work changes and they learn the skills necessary to be successful in the given job at the same time. He emphasizes that employment programming, whether it serves as a signal or not, stands as a type of effort that may or may not be effective, and that the central issue, which flows from a large body of research, is “that well-implemented correctional programs that target the right offenders, target criminogenic needs, and teach offenders new skills and behaviors can have an appreciable effect on recidivism.”

Essayist Dan Bloom (2012, this issue), like the other essayists, finds Bushway and Apel's (2012) argument compelling in many respects. At the same time, he raises questions that need addressing. What kind of information about employment programming participation or performance, for example, would be most helpful to employers as a signal? Among his many observations, Bloom notes that the transitional job experience may benefit the individual through a process similar to signaling. Transitional jobs are temporary positions that allow the individual to learn to work while on the job. Employers may be more willing to hire people who have demonstrated they can handle a job. Bloom acknowledges that more research is needed, but he believes that signaling may be especially important in an era when an increasing number of employers routinely conduct background checks on job applicants.

As Commissioner of the New York City Department of Correction, Dora Schriro (2012, this issue) offers a unique perspective on the policy implications of signaling programs. In an effort to reduce recidivism and the need for increased prison capacity, states have begun to offer some form of relief from the collateral consequences of a criminal conviction, many in the form of an employment certificate. Schriro agrees that the certificate can be used as a signal to prospective employers by validating the inmates’ efforts toward rehabilitation. The certificate does not obligate the employer to hire the ex-prisoner; it just must be viewed as evidence of rehabilitation. Schriro also highlights the importance of agencies using signaling programs in jails as well as in prisons. She, like Latessa (2012) and Brennan (2012), observes that employment programming may be effective for some inmates. She goes on to emphasize that it is not advisable to focus employment training solely on prisons because more individuals are sent to local jails than to state prisons and many are released directly back into the community. To overlook rehabilitation in jails, Schriro argues, is to miss an opportunity to protect the public. Schriro also draws our attention to the potential importance of employment programming for other outcomes, such as housing, that can contribute to and, on their own, constitute successful reentry. Not least, she points to the importance of a multifaceted approach to improving reentry outcomes and, as part of such an approach, the central role that signaling-based efforts may have.

Conclusion

For the indefinite future, communities across America will remain in need of programs and policies that can promote public safety and, most importantly, that can do so in a cost-efficient manner. Criminology can point to many advances to help achieve that goal. The identification of principles of effective intervention (Cullen and Gendreau, 2000) alone, for example, provides broad-based guidance on how correctional systems can promote successful reentry. In addition, the many reentry initiatives undertaken during recent presidential administrations, including those of presidents Clinton, George W. Bush, and Obama, hold much promise (Lattimore et al., 2010). Even so, there remains a strong and compelling need to understand better the factors, both in prison and after release, that can facilitate successful reentry.

Prediction efforts using signaling approaches may well be one approach that contributes to this broader goal. There is also the intriguing notion of empowering individuals while they are in prison and after they return to society to provide demonstrable evidence that they will refrain from offending. In this regard, the proposed certificates of rehabilitation stand out as an approach worth serious consideration. Perhaps such an approach would foster better prediction and, concomitantly, a better ability to target interventions toward the “harder” cases. As the essayists in this issue point out, such certificates are not unproblematic, and so the use of them would need to be undertaken with considerable care. Policy makers may want to consider federal funding to support pilot testing the idea, along with support for rigorous research that assesses the design of such certificates as well as their implementation, impact, and cost-efficiency.

Daniel P. Mears, Ph.D., is the Mark C. Stafford Professor of Criminology at Florida State University's College of Criminology and Criminal Justice. He conducts basic and applied research on a range of crime and justice topics. His work has appeared in Criminology, the Journal of Research in Crime and Delinquency, and other crime and policy journals as well as in a recent book, American Criminal Justice Policy (Cambridge University Press, 2010).

Julie Mestre, M.S., is a doctoral student in the College of Criminology and Criminal Justice at Florida State University. Her research interests include criminal justice policy evaluation and public opinion.