Chapter One

Study Definitions and Methods

This chapter sets forth a detailed description of the definition of recidivism and the methodology used by the Legislative Program Review and Investigations Committee to measure the rate of recidivism. The cohort groups and sample of cases selected for in-depth analysis are also discussed.

What is recidivism?

Over the past six years, shifting priorities in the state's crime policy have increasingly stressed protection of public safety as a primary goal, imposing stricter mandates and policies on criminal justice agencies. The serious and increasingly severe consequences imposed through the state's criminal sentencing laws are meant to provide a deterrent to crime. Incarceration protects the public from offenders for a period of time. Eventually, however, nearly all inmates return to the community -- most within three years. Community supervision becomes an important public safety measure providing oversight of offenders released from prison and those never sent to prison. Finally, the state funds and provides a network of community- and prison-based treatment, education, and rehabilitative programs for pre-trial and convicted offenders, which can offer alternatives to future criminal activity.

In its broadest sense, recidivism can be defined as a public safety failure rate. Recidivism, more specifically and for the purposes of this study, is new criminal activity by a person after a criminal conviction that resulted in either imprisonment or other sanction (i.e., probation, diversionary sentence, or fine).

How was the rate of recidivism measured?

How recidivism is defined has a substantial impact on the identified rate of recidivism, for which there is no universally accepted method of measurement. After a literature review, the program review committee used multiple measurements in its analysis rather than relying on a single method.

The three measurements tracked to identify the overall rate of recidivism are:

As discussed below, each measure has strengths and weaknesses, but combined the three are more comprehensive and accurate means to measure the rate of recidivism in Connecticut. The basis for the program review committee's analysis was rearrest, reconviction, and sentencing data for convicted felons discharged from prison or sentenced to probation during 1997. The committee tracked criminal activity from the date of the offender's last discharge or sentencing in 1997 through December 31, 2000. This is known in criminal justice literature as the release threshold -- the period of time the offender is in the community and "at risk" of repeat criminal activity.

The criminal activity of convicted felons was tracked for three years because there is agreement among researchers and criminal justice administrators this is a sufficient follow-up period to identify a majority of the offenders who would eventually be rearrested for a new crime. The three-year period of 1998, 1999, and 2000 also covers the most recent full years of arrest and court data available. Since repeat criminal activity is tracked from a specific date in 1997 through December 31, 2000, the release threshold for some offenders is more than three years. For example, inmates discharged from prison in January 1997 were tracked beginning from that date, which would include the additional 11 months of 1997 plus the three years from January 1, 1998, through December 31, 2000.

Rearrest data. The rearrest rate was examined because an arrest is the initial response of the state against a person suspected of committing a crime, and it begins the criminal justice process. Arrests are an accepted measure of criminal activity and are used in other research and reporting requirements (e.g., the federal Uniform Crime Reports).

Leading criminal justice researchers have also concluded that arrests are a valid measure of recidivism even though some arrests do not result in convictions. The overall arrest rate in the cohort group tends to be higher than the reconviction rate, in part, because the court's lag time in disposing of cases meant the dispositions of some cases were not included in the study's time frame. The reduced number of convictions also can be attributed to the wide-spread practice of plea bargaining, the diversion of cases out of the criminal courts to pre-trial and alternative sentencing programs, revocations of probation or parole rather than prosecution for the new crime, reluctance of witnesses to cooperate, and due process issues rather than the innocence of the person arrested.

It should be noted an offender might be rearrested more than once during the three-year period. The program review committee analysis included a review of up to 15 rearrests per offender. Also, an offender may be charged with more than one offense per arrest. The analysis included an examination of up to three charges per arrest incident. The dispositions and sanctions imposed for each arrest were also reviewed.

Reconviction data. Reconviction data indicate a new arrest did occur and that, in fact, the offender was found guilty of the charge against him or her. It is measured by the court disposition -- or verdict -- for each criminal case. As previously stated, an offender may be charged with more than one crime per case. The reconviction rate was measured based on a guilty verdict for at least one of the three charges per arrest under analysis.

Reconviction data do not always indicate the seriousness of the offense, typically because of the practice of plea bargaining. However, the program review committee analysis examined any differences in the crime for which the offender was arrested and the crime for which he or she was convicted.

Sentencing data. Sentencing data are the narrowest measure, indicating a new offense occurred and the court imposed a sanction against the offender. The analysis includes a review of any sanction imposed by the court, the types of sentences (i.e., prison, probation, diversionary program, fine) and the length of the sentences.

What type of offenders were tracked?

Given the magnitude of the offender population, it would have been extremely difficult to include all cases in the analysis. The committee staff selected a representative sample of offenders. The initial selection criteria for sampling the offender population included persons who, in 1997, were:

A review of the criminal justice process showed offenders are at risk either at the end of their sentence as they transition from prison, or at the beginning of their sentence if they are sentenced to probation or an alternative sanction in lieu of prison. It was necessary then to select two samples of offenders: (1) inmates discharging from prison; and (2) probationers sentenced to community supervision.3

Offenders convicted of felony crimes were included for analysis, while persons convicted of misdemeanor crimes were not. The analysis was limited to felonies for two reasons. First, felonies are generally the more serious crimes, and policy decisions often are based on the felony -- especially violent -- offender population and their sentencing or program needs. Second, misdemeanors were excluded as a way to make the sample size manageable. There are many more misdemeanor arrests than felony arrests and, based on the selection criteria, almost 15,000 felony offenders were selected for analysis.

Only adult offenders were included for analysis because historically and philosophically the juvenile justice system is independent of the adult criminal system. The objectives of the juvenile and adult criminal systems are different. Also, juvenile court records are confidential and, under certain circumstances, erased when the child becomes an adult. It was determined by the program review committee that a study of recidivism among juvenile delinquents should be conducted separately from that of adult felons.4

Inmate cohort group. In general, inmates are more serious offenders who are being discharged after completing:

For the purposes of the sample, inmates who discharged from a DOC facility and were taken into custody by another state, federal, or military agency on an outstanding warrant or detainer were not included. (Since the inmate remained in a custodial situation, his or her opportunity to recidivate in Connecticut was low.) As a result, a total of 4,006 inmates convicted of a felony who discharged from prison in 1997 were selected for the study.

Probationer cohort group. In 1997, 10,402 adults were convicted of a felony and sentenced to probation or another type of sanction that did not impose a prison term. In general, probationers are the less serious offenders who were sentenced to:

Tracking the rates of recidivism among both the inmate and probationer cohort groups provides the most accurate measure of repeat criminal activity among convicted felons. However, the rates of the two groups should not be compared because differences are likely due to the nature and severity of the offenders' criminal history rather than group characteristics. Simply by the sentence imposed, the court has established a difference among the two groups. The court has found probationers are not a risk to public safety and can remain in the community whereas the inmates were found to pose some threat. Therefore, inmates have served time in prison, while probationers have not, although some may have had a prior incarceration as a result of another criminal conviction. Inmates are nearing or at the end of their sentence, while probationers are beginning their sentence. Furthermore, inmates are more likely to have a more extensive and serious criminal history than probationers. All of these factors have implications for whether offenders will reoffend.

What data were used to track recidivism rates?

Since no single, statewide database currently exists for tracking recidivism or offenders, data for this study were extracted and compiled from several sources within the state's criminal justice system. Four data elements -- offender's name, data of birth, DOC inmate number or court docket number, and State Police Bureau of Identification (SPBI) number -- were used to create an unique identifier for each offender to enable data from different agencies to be linked. The program review committee staff then established two offender-based databases -- one each for inmates and probationers -- that included all repeat criminal activity committed over the three-year period under analysis.

Department of Correction. DOC maintains an inmate-based database that captures historical information on prison admissions and discharges. The department also maintains automated information on sentence lengths, primary offense, demographics, and classification scores.7 From its database, DOC identified the 4,006 felony inmates who discharged in 1997 and provided information on each inmate.

An inmate may be discharged from a DOC prison more than once in a single year. For the purposes of this study, if an inmate had multiple discharges, the program review committee selected the last discharge in 1997 to avoid double counting inmates.

Judicial branch. The judicial branch operates two separate databases that were accessed for this study. The first contains information on court operations and tracks each docket -- or criminal case -- against a defendant, including data on the offense, disposition, and sentence. The second is a case management database for adult probation. It tracks those offenders sentenced to probation including their demographics, sentence, release conditions imposed by the court, program and treatment evaluation status, and violations of probation that may change the custodial status of the offender. The judicial branch identified 10,402 adult, felony offenders sentenced to probation or another sanction in 1997. It provided court and probation supervision data on each.

A defendant may have more than one docket pending during a single year. It is accepted court practice to combine dockets into a single case against a defendant, which would result in only one disposition and sentence. For the purposes of this study, the last docket disposed of in 1997 was selected.

State Police. The Division of State Police, within the Department of Public Safety, is the repository for all criminal history information on offenders. It maintains basic arrest, conviction, and sentencing information for all felony and misdemeanor arrests made by local and state police. This information is used to provide a chronological record ("rap sheet") of a person's criminal activity.8

The state police provided the recidivism component data for the two cohort groups from the date of either their discharge from prison or their sentence to probation through December 31, 2000. This information on rearrests, reconvictions, and sentencing is the primary data used to determine the rate of recidivism.

It should be noted offenders in both the inmate and probation samples may have been rearrested in another state or by a federal law enforcement agency during the period under review. The program review committee did not have access to that information and, therefore, only rearrests occurring in Connecticut are included in the database.

Board of Parole. The Board of Parole provided information on the release and supervision of those inmates who discharged from prison in 1997 after being granted parole. There were 458 (out of 4,006) felony inmates paroled in 1997.

Program and treatment participation. Although the purpose of the study was not to evaluate the effectiveness of prison and community-based treatment and service programs for criminal offenders, a random sample of each cohort group was selected to examine any differences in the recidivism rate among those offenders who participated in programs and those who did not. It should be noted, however, that this analysis will not conclusively determine whether programs resulted in less recidivism than would have occurred in their absence.

Program participation data are not maintained in the automated databases by the criminal justice agencies. These data are recorded in inmates' prison program files and DOC, probation, and parole community supervision files. To make the case file review manageable and obtain the most accurate information, the sample of offenders was limited. A 10 percent random sample of each cohort group was selected. The DOC, probation, and parole case files were then reviewed by program review committee staff to obtain the necessary data.


Program participation in prison and in the community was reviewed for the 423 inmates randomly selected. The inmates' participation in prison-based programs was tracked for the three years preceding their 1997 discharge from prison. For those offenders sentenced to less than three years, the study tracked their program participation from admission date to discharge date in 1997. Their participation in community-based programs was tracked during the period they were under supervision in the community by either DOC or the parole board.

The prison and community-based programs available to inmates include residential and non-residential options: academic education; addiction and mental health services; employment; religious; disciplinary, sex offender, and gang programs; domestic violence and family services; transitional services; vocational education; and other programs.

There were 1,211 probationers randomly selected. Their participation in community-based programs and services as ordered by the court as part of their sentence was reviewed. The type of programs provided to probationers is similar to those provided to the inmate sample.

Technical violation data. A technical violation of probation or parole is misbehavior by an offender under supervision that is not by itself a criminal offense. It is generally a violation of a release condition imposed by either the court or parole board that is intended to guide the offender's behavior while in the community. Technical violations do not usually result in a new arrest. For example, failing to report to a probation or parole officer for a scheduled office visit, missing a curfew, lack of employment or attendance at school, or testing positive for drug use are all common technical violations. Probation and parole officers have a range of graduated sanctions available to address technical violations from verbal reprimand and increased reporting requirements to referrals to treatment or service programs and reincarceration in prison.

It should be noted, however, that a series of relatively minor technical violations can indicate an offender's unwillingness to abide by the release conditions and may result in a more punitive response by the court or parole board. The court may convict the offender of a violation of probation (VOP) and require he or she serve any suspended portion of their sentence or impose a new criminal sentence. The parole board may revoke parole and require the offender to serve the unexpired portion of their sentence in prison.

Typically, research on the rate of recidivism does not track -- or measure -- technical violations of those offenders on probation or parole. This study did not include technical violations in the overall rate of recidivism, but a supplemental analysis of the rate of technical violations among probationers and parolees as a description of misbehavior in the community that does not rise to the level of an arrest is provided. The analysis of the rate and type of technical violations committed by the inmate and probationer samples is presented in Appendix A.

2 A felony is a criminal offense that may result in a sentence of more than one year in prison. A misdemeanor is a criminal offense that may result in a sentence of one year or less in prison.

3 Only about 25 percent of convicted offenders are sent to prison; three-quarters are supervised in the community under probation or a diversionary or alternative sentencing program.

4 In accordance with Public Act 00-172, the Connecticut Economic Policy Council (CPEC) is conducting a cost-benefit evaluation of programs serving juvenile offenders, including an analysis of recidivism among delinquents. The most recent interim report issued by CPEC is dated June 12, 2001.

5 Inmates sentenced to more than two years are eligible for parole release after serving 50 percent of their sentence, except for those convicted of a "serious, violent" offense who must serve at least 85 percent to be eligible.

6 Inmates sentenced to two years or less are eligible for several early release programs administered by the DOC after serving at least 50 percent of their sentence or if they are within 18 months of their discharge from prison.

7 Each convicted offender sentenced to prison is rated -- or classified -- for security and program purposes based on evaluations of his or her criminal and correctional history to determine the propensity for violence, mental health, health, and substance abuse status, educational attainment, gang involvement, and sex offender history. The overall classification score allows DOC to determine which facility and security level to assign the inmate and further allows for movement between facilities and security levels for inmate management and transition to the community.

8 The state police maintain an offender-based database that tracks individuals using a unique identification number called the State Police Bureau of Identification (SPBI) number. The SPBI number is assigned to a person upon their first arrest and is used to track all subsequent criminal activity. It is supported by -- or based on -- the offender's own unique fingerprints.


Return to Year 2001 Studies

Return to Table of Contents