This paper explores recidivism — the relapse into criminal behavior following conviction and release — from multiple analytical angles. It begins by clarifying conceptual and operational definitions of recidivism, including the challenges posed by incomplete criminal justice data across enforcement, prosecutorial, court, and correctional agencies. The paper then reviews Langan and Levin's (2002) four measures of recidivism using a large-scale Bureau of Justice Statistics dataset, with particular attention to gender differences and timing of reoffense. It further examines the broader goals of correctional sanctions and concludes by outlining the Principles of Effective Intervention — risk, need, treatment, and fidelity — as evidence-based guidelines for reducing recidivism rates.
In the context of criminal justice, recidivism represents a relapse of criminal activity by a person after being convicted of an offense, punished, and ostensibly corrected (Maltz, 2001). Recidivism emerges from a series of failures: society's failure to meet the expectations of the individual, or the individual's failure to meet the expectations of society; the person's resultant failure to stay out of trouble; the individual's failure to avoid capture and conviction following the commission of an offense; the individual's failure as a correctional-institute inmate to make the most of correctional interventions — or the institution's failure to provide adequate rehabilitation programs; and the person's further failure to refrain from criminal activities following release (Maltz, 2001).
Recidivism, from a correctional perspective, has a fairly explicit meaning. The term derives from the Latin recidere, meaning "falling back" (Maltz, 2001). Recidivists are individuals who are not rehabilitated following release from confinement and instead revert to their former behavior by committing further offenses. While this conceptual definition seems straightforward, a more operational one — which allows for measurement — is considerably more complex (Maltz, 2001).
The data on which recidivism measurement is based is seldom comprehensive, and even when complete information can be obtained, data analysis methods are inconsistent. Substantial variation exists in recidivism measurement techniques (National Advisory Commission, 1973).
Consider recidivism measurement when comprehensive data is available. Complete information means that all offenses an individual has committed — whether or not he or she was arrested for all of them — are known to evaluators. This improbable condition also poses a challenge, as the term "crime" covers a wide range of behaviors (Maltz, 2001).
A child-molesting individual, for instance, may be detained, found guilty, and sentenced to a correctional intervention designed exclusively for treating such offenders. After release, he may genuinely stop molesting children. However, if he turns to other forms of crime — such as forgery or armed robbery — should he be considered a recidivist? From one perspective, he may be labeled a recidivist because he engaged in criminal behavior after release. From another, this new crime is entirely different; while the original offense has been curbed, he has resorted to a different type of harmful behavior. The question of whether an individual who moves from one type of crime to a completely different one should be classified as a recidivist remains unresolved (Maltz, 2001).
One could label an individual a recidivist only if he or she repeats the same type of offense for which he or she was originally convicted (Maltz, 2001). This definition, however, implies that offenders specialize in a single crime — a claim that contradicts available and emerging evidence (Petersilia, 1980; Chaiken & Chaiken, 1982; Goldstein, 1982; Miller & Dinitz, 1982). Moreover, creating a separate recidivism category for each individual offense type would correspondingly reduce the available information within each category (Maltz, 2001). The four broad categories of crime — personal, white-collar, property, and public-order — may be considered; however, no clear lines divide these categories, as many crimes (such as arson or robbery) may overlap, supersede, or magnify one another (Maltz, 2001).
In order to understand the full impact of "return to crime" and the time period in which the probability is highest, it is necessary to categorize re-imprisoned convicts, identify the most probable reasons and timing of such recurrence, and examine the influence of gender on the phenomenon.
The usual assumption is that comprehensive information is available regarding an individual's involvement with the criminal justice system; in practice, some information is inevitably lacking. Criminal justice dealings are reported by several different agencies; however, most states do not maintain a single, unified criminal-data repository. This can partly be attributed to the involvement of numerous governmental levels and jurisdictions within a state's criminal justice system (Maltz, 2001).
Enforcement agencies in a state operate at every governmental level: local police at the municipal level, sheriffs at the county level, and investigation bureaus and highway patrols at the state level. Centralized criminal and detention data reporting began more than 50 years ago, with ongoing expansions and improvements (Maltz, 1977). Today, it encompasses virtually every enforcement agency.
Prosecutors' offices and criminal courts are essentially county-level agencies, though with some exceptions. They do not have a single data repository comparable to the National Crime Information Center (NCIC) felony arrests program. Statistical Analysis Centers (SACs) are responsible for compiling all criminal justice data; however, their collection of trial and prosecution data has not advanced at the same pace as data collected from enforcement agencies (Maltz, 2001). Nevertheless, many larger prosecutorial offices have received Department of Justice support to set up PROMIS (Prosecutor's Management Information System) (National Institute of Justice, 1977) or similar IT-based systems for storing and retrieving information, which should allow them to provide more complete records than other jurisdictions (Maltz, 2001).
Correctional information can be obtained from county jails as well as state-level prisons and halfway houses. However, only a few states currently incorporate jail data into statewide correctional statistics (Maltz, 2001). States have been funded by the Department of Justice to develop Offender-Based State Correctional Information Systems (OBSCIS) for compiling statewide correctional information across all levels. At present, however, only some states are able to track individuals sentenced to jail on a routine basis with any assurance of comprehensiveness (Maltz, 2001).
Even when complete statewide information is available for analysis, it remains incomplete in cases where offenders are rearrested, tried, and/or convicted in another state. Recidivism rates can be geographically skewed if only data from the state conducting the study is used — a limitation that is always present when the recidivism marker is "return to imprisonment" (Maltz, 2001).
Arrests need not be the only indicators of recidivism. One way to reduce Type I errors — those arising from incorrect arrests — is to examine subsequent criminal justice proceedings following the arrest. If recidivism is defined as arrest followed by affirmative prosecution, then records must be reviewed to determine whether charges or other prosecutions have occurred. Analysts must examine records by reviewing every file available in the prosecutor's office. Except in a few automated offices, this typically means studying all case folders to clarify the defendant's name, the nature of the original charge, whether charges were dropped or pursued, and the current disposition of the cases. This information is often handwritten by staff attorneys onto the outside of the folder — sometimes in writing that is difficult to read. The task is further complicated by the fact that case folders may be kept in briefcases, stacked on desks, stored in attorneys' cars or homes, or in other locations convenient for attorneys but burdensome for researchers. Although this specific check on arrest quality would be valuable, obtaining comprehensive and accurate information cannot be guaranteed (Maltz, 2001).
Another disposition related to parole must be considered when defining recidivism. If an individual absconds — that is, ceases reporting to a parole officer and cannot be located — should he or she be labeled a recidivist? Some states, rather irrationally, treat absconders as default successes. When the justice system takes no action against an absconder because the person has not been found, the individual is recorded as having succeeded, since no records exist of arrest, parole violation, or any other negative action (Maltz, 2001).
A majority of states rely solely on data from within their own state to evaluate recidivism (Maltz, 2001). This practice is driven not by any theory of offender mobility but by the difficulty and uncertainty of obtaining data from other states, as well as a degree of caution about using information gathered under different criteria and standards (Maltz, 2001).
Data from the Federal Bureau of Investigation (FBI) may be used in special evaluations during follow-up periods to capture all arrests; however, this is the exception rather than the norm. Even when FBI data is used, it can be difficult to determine whether an arrest led to a conviction. This was demonstrated by a 1977 study by Kitchener and colleagues, which investigated federal prison releases despite having access to the FBI's NCIC data; a subsequent study by Stone-Meierhoefer and Hoffman (1980) found that disposition data were frequently missing, with gaps filled through telephone follow-ups with arresting authorities (Maltz, 2001).
Langan and Levin (2002) analyzed data gathered from the Bureau of Justice Statistics on prisoners released from state prison in 1994. Their research focused on four measures of recidivism: re-arrest, reconviction, and resentencing (with and without a new sentence) (Deschenes, Owen, & Crow, 2007). A sample of 272,111 prisoners — both male and female — was used. Analysis of this sample showed that, within three years of release (Deschenes, Owen, & Crow, 2007):
67.5% were rearrested; 46.9% were reconvicted; 25.4% were resentenced; and 51.8% were returned to prison due to a new crime or technical violation.
These combined statistics showed that the crimes with the highest re-arrest rates were (Deschenes, Owen, & Crow, 2007): robbery (70.2%); burglary (74%); larceny (74.6%); motor vehicle theft (78.8%); possession or sale of stolen property (77.4%); and weapons charges (70.2%).
The lowest re-arrest rates were among prisoners originally imprisoned for (Deschenes, Owen, & Crow, 2007): homicide (40.7%); rape (46%); other sexual assaults (41.1%); and driving under the influence (51.5%).
Another important factor Langan and Levin analyzed was the number of prisoners who returned to crime after release. The data show that, over an overall duration of three years, approximately 44% of all female prisoners released in 1994 were reconvicted (Deschenes, Owen, & Crow, 2007). Among all female offenders rearrested in the three-year follow-up, 23.3% were rearrested within the first six months and 34.5% within the first year. This rate increased by roughly 15% in each of the second and third years, indicating that the highest-risk period for re-arrest is the first year following release. Research on this sample further showed that the percentage of females who recidivated was substantially lower than that of males across all four measures: re-arrest, reconviction, resentencing, and return to prison (Deschenes, Owen, & Crow, 2007).
The inferences drawn from these findings offer insight into the likelihood of relapse into crime and the time periods of greatest susceptibility. Such an understanding can inform intervention design, improve anticipatory capabilities, and increase the chances of preventing recidivism.
The nature of a correctional intervention, to a considerable extent, also shapes the definition of recidivism employed (Maltz, 2001). Revisiting the child-molester example, the offender would be labeled a recidivist if the intervention under evaluation was a study of different bail caseload outcomes, but not if the intervention was specifically designed to modify child molesters' behavior. Therefore, when defining recidivism, both program type and crime type must be taken into account (Maltz, 2001).
"Risk, need, treatment, and fidelity principles explained"
"Limitations and value of recidivism as an evaluative tool"
Always verify citation format against your institution’s current style guide requirements.