By Tara Wheeler, Julia Bergeron-Smith, George Shaler, Lisa Sampson

Sex offender recidivism data can be difficult to comprehend, especially when conducting research across states. For example, sex offenders can be categorized in numerous ways: by the type of crime committed (e.g., rape, assault, exhibitionism), by offense severity, by victim age (adult or child). Likewise, recidivism definitions (e.g., rearrest, reconviction, or reincarceration) and timeframes (e.g., one-year, two-year, or three-years following release) can vary across local, state, and federal agencies. Thus, there is no single measure systemically used across jurisdictions. Recognizing the public’s concerns about sex offender recidivism, the Maine and Massachusetts Statistical Analysis Centers (SACs), proposed and received funding for a retrospective sex offender recidivism study through the Bureau of Justice Statistics, Department of Justice. This collaboration marks the first time either state has undertaken this type of study. This project studied the recidivism rates of Maine sex offenders who were released from prison between 2005 and 2019 and Massachusetts sex offenders released between 2009 and 2018. In total, the Massachusetts cohort was composed of 1,210 offenders and the Maine study of 905 offenders. Criminal history records were obtained for each offender from their respective states to determine whether the offender recidivated (i.e., committed post-release offenses that resulted in a conviction). To be included in the recidivism portion of this study, there must have been at least a five-year window between the time of release and the time the criminal history data was pulled. Overall, 880 offenders were included in the Massachusetts recidivism analysis and 661 in the Maine recidivism analysis. A primary interest area was to determine whether recidivism rates differed by offense severity (i.e., rape or non-rape) and victim age (i.e., child or adult), however, these two groupings are not entirely separate from one another— while rape is considered more severe than other types of sex offenses, crimes against children are also deemed to be more egregious than crimes against adults. Therefore, an offender typology—based on the original sex offenses associated with the commitment and subsequent release—was created that incorporates both victim type and offense severity. The first, and most severe, category is child rapist, which includes all offenders who committed a child rape offense, followed by the rapist category, which includes all other offenders who committed a rape offense. Next is child predator and is used for those who committed a sex offense against a child but did not commit a rape offense. Last, is the other category and captures those who did not fall within one of the prior three categories. To gain a better understanding of recidivism among the sex offender population, survival analysis (specifically, the Cox Proportional Hazards regression method) was conducted to determine which offender characteristics, if any, influenced recidivism rates. Using this approach, researchers were able to control for other known attributes. The attributes tested in this study were: • offender type • severity of sex offense(s) (Maine only) • number of sex offense(s) • commitment length (Maine only) • release type (supervision or discharge) • security level of facility offender was released from (Massachusetts only) • release age Key Findings • Age at earliest sex offense associated with commitment varied by offender type for both Maine and Massachusetts. Interestingly, child predators had a mean age that was statistically higher than that of rapists and child rapists. o In Massachusetts, the average offense age of child predators was 36.6 years old, which was significantly higher than the offense age of rapists (31.1 years old) and child rapists (33.9 years old). o For Maine, child predators were, on average, 35.0 years old at the time of their earliest sex offense associated with commitment, compared to 30.4 years old for child rapists and 31.1 years old for rapists. • Maine’s five-year recidivism rate (43%) was much higher than the Massachusetts five year recidivism rate (25%). While there are numerous potential scenarios that could explain the differences in recidivism rates, it is important to note that each state has its own criminal laws and procedures that, in turn, impacts the underlying study population and their recidivism rates. For instance, an offense that resulted in commitment to state prison in Maine might result in a county jail commitment for Massachusetts. Therefore, caution should be taken when making comparisons between the two states. • Offender type, severity of sex offense, and release age were found to be associated with Maine recidivism rates. When coupled with release age, release type was also found to have an influence on recidivism rates. Holding all other attributes constant: o Individuals whose most severe sex offense was a misdemeanor are expected to recidivate at a rate 45% higher than those with a felony level offense. o ‘Other’ type offenders (those whose offenses fell short of rape and did not target children) have an expected recidivism rate 45% higher than offenders who committed sex offenses against children (child predators and child rapists) o Age at release, both by itself and as an interaction with release type, is associated with recidivism rates. For every 10-year increase in age at release, the predicted recidivism rate decreased by 21%. However, for offenders who were released to supervision, the hazard rate decreases even further, with supervised offenders having a 38% decrease in risk for every 10-year increase of age at release. • Of the attributes tested, offender type, release type, release level, and release age were found to be associated with Massachusetts recidivism rates. Holding all other attributes constant: o Offenders released from a maximum-security level facility are predicted to recidivate at a rate 2.4 times higher than that of offenders released from a medium security level facility or lower. o ‘Other’ type sex offenders are predicted to recidivate at a rate 115% higher than those who committed child-based sex offenses (child predators and child rapists). Meanwhile, non-child rape offenders are expected to recidivate at a rate 58% higher than child offenders. o Discharged offenders released without supervision have an expected recidivism rate 61% higher than offenders who were supervised following release. o For every 10-year increase in age, the recidivism rate is expected to decrease by 37%.    

 Portland:  Maine Statistical Analysis Center;   Massachusetts Statistical Analysis Center, 2023. 50p.