During the 1996 legislative session, there was considerable discussion about the cost-effectiveness of Minnesota's prison system. Among the 50 states, Minnesota has one of the lowest rates of imprisonment, but its cost per inmate is among the nation's highest. At a time when Minnesota was contemplating construction of additional prison beds, legislators wondered whether spending more per inmate has resulted in better outcomes, such as reduced rates of offender recidivism or lower crime rates. They also wondered whether community-based alternatives to prison--another growing part of the state budget--adequately protect public safety. In light of these concerns, our study addressed the following questions:
To what extent are convicted Minnesota felons subsequently arrested, convicted, and imprisoned, and how do these recidivism rates compare with those found in other states?
How is recidivism related to offenders' criminal history, conviction offense, personal characteristics, program participation, and other factors? What types of new offenses do convicted felons commit?
We tracked 1,879 offenders released from prison in 1992 and 6,791 offenders sentenced to probation in 1992. For each offender, we examined recidivism for exactly three years--from a prisoner's date of release or from a probationer's date of sentencing. Previous research has indicated that it is possible to identify the majority of eventual recidivists by tracking rearrests over a three-year period. To identify instances of recidivism, we relied primarily on arrest and conviction information from the Minnesota Bureau of Criminal Apprehension (BCA). Our analysis examined arrests and convictions for felonies and gross misdemeanors, but it did not examine less serious offenses (that is, simple misdemeanors). In addition to BCA data, we used Minnesota Department of Corrections data on imprisonments, Federal Bureau of Investigation data on offenses in other states, and sentencing data from the Minnesota Sentencing Guidelines Commission.
Adult Felons in Minnesota
Historically, Minnesota has had fewer serious crimes per capita than the nation as a whole. Figure 1 shows that Minnesota's rates of adult arrest for serious violent and property crimes are relatively low, although Minnesota's adult arrest rates for violent crime rose faster than the nation's rates during the past decade. About 9,400 people were convicted of felony offenses in Minnesota during 1995, an increase of more than 3,000 felons since 1986.
Since 1980, Minnesota's sentencing guidelines have indicated the sentences that are presumed to be appropriate for convicted felons, based on (1) the severity of the felons' conviction offenses, and (2) the offenders' prior criminal records. The courts may depart from the guidelines under "substantial and compelling circumstances," but they have followed the guidelines nearly 90 percent of the time. <FN- Minnesota Sentencing Guidelines Commission, Minnesota Sentencing Guidelines and Commentary (St. Paul, August 1, 1995), 20.> The guidelines reflect a "just deserts" sentencing policy that links the severity of offenders' punishments to the severity of the crimes they have committed.
In part, the guidelines are intended to reserve prison space for the most serious offenders. Minnesota's courts have consistently sentenced about 20 percent of convicted felons to prison, with the remainder placed on probation in the community. The length of Minnesota prison sentences has increased in recent years, largely reflecting legislative actions. The median prison sentence in 1994 was 30 months. While many states have parole boards that decide when to release offenders from prison, nearly all imprisoned offenders in Minnesota are required by law to serve at least two-thirds of their sentence in prison before their "supervised release" to the community. <FN-Offenders who violate prison rules may be required to serve more than two-thirds of their sentence in prison. Also, the Commissioner of Corrections has discretion about when to release offenders imprisoned for crimes committed before May 1980, but there are relatively few of these offenders.>
Minnesota's eight state prisons currently incarcerate more than 5,000 male and female offenders. Minnesota imprisons fewer people per state resident than all states except North Dakota, but its daily cost per prisoner ($83 in fiscal year 1995) is among the nation's highest. Minnesota's higher cost per inmate largely reflects the state's relatively high prison staffing and salary levels. It may also reflect Minnesota prisons' wide array of work, education, and treatment programs, although these programs account for only 13 percent of prison expenditures. We found that about 70 percent of offenders in Minnesota prisons in October 1996 were participating in such programs.
Felons sentenced to probation and offenders on supervised release after serving time in prison are supervised by county staff in 31 counties and by Department of Corrections staff in the remaining 56. As of December 1995, there were 26,114 adult felons under community supervision in Minnesota. Most felons placed on probation serve jail time after sentencing, typically for less than two months.
Overall Recidivism Rates
"Recidivism" can be defined as an individual's return to crime following a criminal conviction. Our study--like nearly all recidivism studies--relied on official records of criminal activity and, therefore, only measured offenses that resulted in arrests, convictions, or imprisonments. There is no universally-accepted method of measuring recidivism, so our study used a variety of measures. To determine a reasonable estimate of the overall level of criminal behavior among convicted felons, most studies have examined the percentage of offenders who were rearrested during a particular follow-up period. Many studies have also examined the percentage of offenders reconvicted. Reconviction rates are somewhat lower than rearrest rates because (1) not all arrested persons are prosecuted and convicted, and (2) the convictions of some arrested offenders occur after the follow-up period. The percentage of offenders who are subsequently imprisoned will be still lower because some convicted recidivists are sentenced to probation rather than prison.
Figure 2 shows the percentages of Minnesota probationers and released prisoners who were rearrested, reconvicted, or imprisoned in Minnesota for new felonies or gross misdemeanors. We found that: In the three years following release from prison in 1992, 59 percent of offenders were rearrested in Minnesota (45 percent for felonies) and 45 percent were reconvicted (34 percent for felonies).
Another 5 percent of the prisoners were arrested in other states (but not in Minnesota) in the three years following their release. Thus, a total of 64 percent of Minnesota's prisoners were rearrested in the United States within three years of their release. In addition, a total of 40 percent of prisoners returned to prison in Minnesota within three years of release--28 percent for new offenses, and another 12 percent solely for technical violations of their release conditions, not for new offenses.
Minnesota's levels of recidivism were within the broad range of recidivism rates found in studies conducted elsewhere. These studies have usually found three-year rates of rearrest for released prisoners ranging from 50 to 70 percent, with reconviction rates usually ranging from 35 to 55 percent and cumulative reimprisonment rates (for new offenses and technical violations) ranging from 25 to 45 percent.
Figure 2 also shows that:
Nationally, recidivism studies have usually shown lower rates of recidivism for probationers than for released prisoners, as we found in Minnesota. This largely reflects the fact that the average probationer has a shorter history of criminal behavior than the average released prisoner, and the length of offenders' prior criminal records is an important indicator of their likelihood to reoffend. Figure 3 uses a measure of offenders' prior records--the sentencing guidelines' "criminal history score"--to show that probationers and prisoners with similar prior records of criminal activity had similar rates of rearrest in our follow-up period.
Recidivism Rates for Various Types of Felons
Under Minnesota law, a "felony" is any crime for which a prison sentence of more than one year may be imposed. Felonies include serious crimes against persons and property, as well as some drug crimes and a variety of other offenses. We examined whether the recidivism rates of offenders were related to the types of crimes for which they were sentenced.
Figure 4 shows that probationers and released prisoners who were originally convicted of property offenses were more likely than other offenders to be rearrested during the follow-up period. For instance, 66 percent of property offenders released from prison were arrested for a new felony or gross misdemeanor within three years, compared with 45 percent of violent offenders.
In addition, Figure 4 shows that offenders sent to prison for violent, property, and drug crimes were about equally likely to be arrested for a violent felony after their release from prison. In contrast, offenders placed on probation for a violent felony were more likely than other probationers to be subsequently arrested for a violent felony.
We also examined whether recidivism was related to the specific crimes for which convicted felons were imprisoned or sentenced to probation. As shown in Table 1, we found that:
The felony probationers most likely to be rearrested were car thieves (57 percent) and robbers (55 percent). The probationers least likely to be rearrested were homicide offenders (21 percent) and sex offenders (25 percent).
Table 1 also shows that many offenders were rearrested for crimes other than their original offense, suggesting that felons often do not "specialize" in one offense type.
In 1992, about 8 percent of convicted felons were placed on probation by the courts in cases where the sentencing guidelines recommended prison sentences. We found that:
In contrast, violent and drug offenders who were placed on probation in cases where the sentencing guidelines called for prison had recidivism rates less than or similar to those of other felony probationers sentenced for violent and drug offenses in 1992. We also examined the relationship between offenders' personal characteristics and their likelihood of committing repeat offenses. We found that recidivism rates were: (1) lower among older offenders, (2) somewhat higher among men than women, and (3) highest among black offenders and lowest among white offenders.
In addition, probationers and released prisoners in Hennepin and Ramsey counties had higher recidivism rates than offenders in most other counties. Released prisoners in the 29 Community Corrections Act (CCA) counties other than Hennepin and Ramsey collectively had higher recidivism rates than offenders in the 56 non-CCA counties.<FN- CCA counties receive state block grants to implement community corrections services. Plans for these services are developed with assistance from local advisory boards and approved by the Commissioner of Corrections.> For probationers, the collective recidivism rates in CCA counties other than Hennepin and Ramsey were about the same as the rates in non-CCA counties.
Recidivism and Participation in Prison Programs
Minnesota law requires the Commissioner of Corrections to "have wide and successful administrative experience in correctional programs embodying rehabilitative concepts" and to accept persons committed by the courts "for care, custody, and rehabilitation." <FN- Minn. Stat. §241.01, Subd. 1, 3a.> Minnesota prisons and communities have developed many correctional programs that are intended to change criminal behaviors and thoughts, provide education and job skills, and address chemical dependency problems. In addition, these programs are intended to reduce prison discipline problems by reducing inmates' idle time.
Careful studies in other states have shown that some programs have reduced recidivism, although not always by large amounts. Unfortunately, there is no clear consensus regarding which programs are most effective with various categories of offenders, and programs that have worked in certain circumstances have not always worked in others.
We examined the recidivism of offenders who participated in selected programs during or after their Minnesota prison stays. Our study did not isolate the impact of programs from other factors, and we do not know how the program participants would have behaved in the absence of the programs. <FN- The best studies evaluate program results by randomly assigning offenders to treatment programs or "control groups" that do not participate in the program. Such studies help to ensure that treated offenders have characteristics similar to those of offenders in the control group. This approach was not possible in our study, which tracked offenders who had already been released from prison. We examined whether recidivism patterns reflected factors such as criminal history and types of conviction offenses, but it is possible that factors for which we could not control also helped explain recidivism differences between treated and untreated offenders.> We found that:
We found that participants in a chemical dependency program at one prison (Stillwater) and education programs at two prisons (St. Cloud and Shakopee) had recidivism rates similar to those of other released inmates. We also found that inmates who were released to halfway houses because they were considered "public risks" had slightly higher recidivism rates than other inmates released in 1992.
About 27 percent of the sex offenders released in 1992 completed a treatment program in prison. Inmates with no felony sex offense convictions prior to their imprisonment offense who completed sex offender treatment had a lower rearrest rate for sex offenses (3 percent) than first-time sex offenders who never entered treatment (9 percent). Among inmates who entered prison with at least one prior sex offense conviction, treated and untreated inmates had about the same rates of rearrest for sex offenses (25 percent for treated offenders, 27 percent for untreated offenders).
Adequacy of Criminal History Data
To conduct this study, we relied considerably on arrest and conviction information in the Bureau of Criminal Apprehension's (BCA) official criminal history database. Unfortunately, law enforcement agencies often fail to send BCA fingerprint cards that are required by state law, and information submitted to BCA by the courts and county attorneys is not always consistent with information submitted by law enforcement agencies.
Mainly as a result of these problems, BCA estimates that 37 percent of the records it has received from courts and county attorneys--including records of numerous convictions--cannot be matched to arrest data. BCA excludes these records from its official criminal history database, although they are maintained in a "suspense file" that is largely inaccessible to criminal justice agencies.
We incorporated information from BCA's suspense file into our recidivism analysis. This increased our estimates of three-year recidivism rates by modest amounts--a 4 percentage point increase in the prisoner reconviction rate and a 3 percentage point increase in the prisoner rearrest rate. However,
These agencies regularly use the criminal history database to perform investigations and background checks, prepare recommendations for offender sentencing, make pre-trial release and bail decisions, and conduct research. For these purposes, it is important to have information on all instances of known criminal behavior. We reviewed BCA records for a large group of released prisoners and probationers and found that about half had at least one record (often a conviction) that was not recorded in BCA's official criminal history database. <FN We examined suspense file records for nearly 6,600 of the offenders whom we tracked in our recidivism analysis. About 54 percent had records in the suspense file, indicating that their full criminal histories were not reflected in BCA's criminal history database. Based on our review of 1992-95 suspense file records, we concluded that about half of such records were for convictions.>
Generally, policy makers and corrections officials hope that recidivism rates will be low, perhaps reflecting successful efforts to rehabilitate offenders and deter crime through the threat of sanctions. Of course, it may be a difficult task for the criminal justice system to change the well-established criminal behaviors of certain offenders, including many of the released prisoners and felony probationers we studied. Nevertheless, policy makers may be disappointed by Minnesota's recidivism rates, especially in light of Minnesota's relatively high prison expenditures per inmate. Although we found that Minnesota's recidivism rates are within the broad range of rates reported in other studies, it is not reassuring that nearly two-thirds of released prisoners and nearly half of felony probationers were rearrested for felonies or gross misdemeanors within three years.
Policy makers could imprison more criminals to keep them from reoffending, but it is possible that this would only postpone recidivism rather than reduce it. Such a solution would be extremely expensive if applied to large numbers of offenders.<FN- As noted earlier, only about 20 percent of the nearly 10,000 felons convicted annually in Minnesota are sentenced to prison. The daily operating cost per inmate for Minnesota prisons averaged $83 in fiscal year 1995, and the 1996 Legislature authorized the sale of $89 million in bonds to build an 800-bed prison.> Alternatively, the state can--and does--use imprisonment more selectively, but with greater risks to public safety and well-being. We offer no recommendations for changes in correctional programming or sentencing policy, but we do recommend that corrections officials regularly monitor recidivism. Specifically,
We think that the department should supplement reimprisonment measures with more comprehensive measures, including rearrest and reconviction rates. In addition, the department should change its method of counting offenders reimprisoned for new offenses to avoid undercounting the actual reimprisonment rate.<FN- When calculating the percentage of released prisoners returning to prison for new offenses, the department has not counted offenders who first returned to prison for technical violations and later were imprisoned for new offenses. For a three-year follow-up period, the department's method would understate the actual recidivism rate of 1992 releasees by 6 percentage points (22 vs. 28 percent).>
There are many potential state and local users of recidivism information, and it would be useful for these users to help design future measures of recidivism (and other outcomes). We recommend that:
The Department of Corrections should establish central, permanent records that indicate (1) the programs in which individual prisoners have participated (including dates of participation and whether the program was completed), and (2) whether inmates have been designated by the department as "public risks."
We think that Minnesota's official criminal history database needs more complete and accurate information on Minnesota convictions and other case "dispositions." <FN- Besides convictions, "dispositions" include records of dismissed cases, acquittals, and other case outcomes.> BCA has taken some steps already, training local officials in proper reporting of criminal history information and seeking federal funding for continued implementation of electronic fingerprinting technology. We recommend that:
We hope that local law enforcement and court officials will provide BCA with more complete, accurate information in the future, but in the meantime we think there should be a way for users of the official criminal history database to identify instances of serious criminal behavior that are not yet recorded in this database. We recommend that:
BCA or the Legislature may wish to provide criminal justice agencies with access to the suspense file, but restrict or prohibit access to others. Since the identities of many convicted offenders in the suspense file have not been positively established through fingerprints, the database should provide users with appropriate cautions about the suspense file information.
The Program Evaluation Division was directed to conduct this study by the Legislative Audit Commission on May, 1996. For a copy of the full report, entitled " Recidivism of Adult Felons", (97-01), 93 pp., published on January 13, 1997, please call 651/296-4708, e-mail Legislative.Auditor@state.mn.us, or write to Office of the Legislative Auditor, 658 Cedar St., St. Paul, MN 55155.
Staff who worked on this project were Joel Alter (project manager), Christina Tsuei, and David Chein.