Expanded view of recidivism in Alaska
Recidivism is a problem, both nationally and in Alaska, with many who are released from prison returning to the criminal justice system convicted of new crimes. As part of its Alaska Results First (RF) analysis, the Alaska Justice Information Center (AJiC) looked at recidivism rates for individuals convicted of crimes who were released from an Alaska Department of Corrections (DOC) facility in 2007. By following these offenders for eight years, AJiC is expanding our understanding of recidivism patterns for a large group of offenders, beyond any prior study.
While AJiC’s analysis is consistent with older two and three-year studies of recidivism conducted by the Alaska Judicial Council (Carns et al., 2007; Carns et al., 2011), additional years of study surface questions about recidivism patterns related to offense type and changes that occur beyond three years.
Differences among offense-based cohorts
Within the framework of the RF analysis, recidivism was defined as a new criminal conviction, measured by the time of the arrest that resulted in the conviction. Only the conviction for the first re-offense was counted when calculating recidivism. (Offenders were tracked a year and a half beyond the 8-year period to address lag time between arrest and conviction.)
To illustrate differences in recidivism related to crime type, we focused on recidivism for four groups of offenders. These RF cohorts were defined based on criteria for domestic violence (DV), sex offender, and Driving Under the Influence (DUI) therapeutic court programs.
Cumulative recidivism curves, like those in Figure 1, show the percentage of offenders who have recidivated for the first time by a given year. For example, among DV offenders, 41 percent recidivated in the first year after release. By the second year, 54 percent had recidivated and by the third year, 62 percent. By the eighth year, approximately 75 percent of offenders in this cohort had recidivated.
In general, recidivism curves rise sharply in the first year, and then begin to flatten. Although all RF cohorts followed this trend, there were differences among cohorts. Some cohorts rose more sharply, some flattened more quickly, demonstrating the differences in cumulative recidivism among the cohorts.
For instance, although the percentage of first time recidivists is highest in the first year among all cohorts, the rate for the DV cohort is 20 points higher than other cohorts.
Overall, the DV cohort had the highest rate of recidivism, and the sex offender cohort had the lowest rate, during each year of the follow-up period. DUI cohorts had recidivism rates in between these two. The greatest difference occurred in the third year, when 62 percent of offenders in the DV cohort and 35 percent of those in the sex offender cohort had recidivated.
Looking at the pattern beyond the three-year mark (the vertical line in Figure 1) we see that the gradual flattening of the recidivism curve does not continue smoothly among all cohorts. Curves for the DUI felon and sex offender cohorts begin to steepen slightly during the last two years, rather than continue to flatten.
In the following, we look at the three general offense types: sex offenders, domestic violence, and DUI offenders (misdemeanor and felony). We also look at cumulative recidivism rates of felons versus misdemeanants.
Least likely to recidivate: Sex offenders
After one year, 20 percent of sex offenders had recidivated, similar to the rate for the DUI-related cohorts. By year two, sex offenders had the lowest rate of recidivism of all cohorts. Over half remained clear of a new conviction for seven years after release. In all other cohorts, more than half of offenders recidivated by the fifth year or earlier.
Overall, these results are consistent with prior reports that sex offenders are less likely to recidivate than other offenders (Carns et al., 2007; Carns et al., 2011; Durose et al., 2014). Nonetheless, the steepening of the curve in the last two years surfaces questions about what might be accounting for a rise in recidivism when offenders are tracked for a longer period of time and how this trajectory might look if tracked even longer.
Consistent with prior research, when sex offenders recidivated, they were most often convicted of a misdemeanor (Myrstol, Rivera, & Parker, 2016). The RF analysis found 70 percent convicted of a misdemeanor and less than 10 percent convicted of another felony sex offense.
Most likely to recidivate: DV
Domestic violence is defined by Alaska Statute 18.66.990. A DV offense is determined by the relationship between the offender and the victim, and may involve a variety of offenses, including murder, assault, burglary, criminal trespass, arson, terroristic threatening, harassment, and violating a protective order. The Department of Public Safety (DPS) maintains a DV-conviction flag in offenders’ criminal history, but the information is not available in DOC records.
AJiC used DPS records from 2014 to identify the distribution of offenses associated with a DV-conviction. Misdemeanor assault (65.4%), violation of a DV protective order (7.1%), and assault 3 (5.3%) accounted for three quarters of convictions in the DV distribution. To develop the DV cohort, AJiC randomly selected offenders released from DOC in 2007 to match this distribution of convictions (Valle, 2017: 79, 80).
The DV cohort had the highest recidivism rate of all the RF cohorts. Within one year of their release, 41 percent had recidivated, twice the percentage seen for the other offense-specific cohorts. Although the cumulative recidivism rate remained higher for this cohort throughout the eight-year follow-up, the curve flattened markedly in the second year. With the exception of the higher recidivism rate in the first year, the recidivism curve was most similar to that of the DUI misdemeanor cohort.
Members of the DV cohort, like sex offenders, were most likely to reoffend by committing a misdemeanor. However, 60 percent of DV offenders who recidivated committed another offense associated with a DPS DV-conviction flag. A third of these offenses were assaults, most often assault in the fourth degree, a misdemeanor (Valle, 2017: 36).
Offenders convicted of DUI offenses were more likely to be reconvicted than sex offenders, and less likely to be reconvicted than DV offenders.
Recidivism patterns for DUI misdemeanant and DUI felon cohorts were very similar for the first four years after release, but then diverged. At eight years, felons had a 10-point higher rate of recidivism than misdemeanants (66% versus 56%).
For misdemeanant DUI offenders, the cumulative recidivism curve flattens beginning in year five. In each of the next four years, only about two percent were added to the ranks of recidivists.
In year five, the curve for felony DUI diverges from the misdemeanant curve, becoming steeper, and showing a higher cumulative recidivism rate. In the eighth year, the felony DUI curve turns upward. Here, as in the sex offender cohort, the longer time line surfaces questions about what might be influencing an upswing in recidivism when we look farther out. What would we see if we were to extend our analysis beyond year eight? Would it continue to go up, level off, or go down?
Felons versus misdemeanants
The upturn in the felony DUI but not the misdemeanor DUI curves, and upturn in the sex offender but not the mostly (85%) misdemeanor-based DV cohort, raises questions about what we might find if we looked at patterns of recidivism among general groups of felons versus misdemeanants over an eight-year period. Information collected by AJiC researchers while doing the RF analysis made it possible to do this analysis.
In the RF analysis, reconviction data were established for all convicted offenders released from DOC institutional custody in 2007, but recidivism rates were only computed for the cohorts used to model RF programs. To compute cumulative recidivism for all felons, we identified offenders whose jail time was associated with at least one felony conviction (N = 2,360). For misdemeanants, we included those with no felony convictions and at least one misdemeanor conviction (N = 8,659). Results are shown in Figure 2.
Offenders are at greatest risk for a return to crime during their first year post-release. Close to a third of offenders were arrested and later convicted of crimes committed during this period. This is more than twice the rate of first time recidivism seen in any other year. This result is important for policy because it suggests that the immediate reentry period is critical. Programs that facilitate the transition back to the community may have a particularly positive impact on recidivism.
At first, cumulative recidivism is slightly higher for misdemeanants than for felons, consistent with prior results (Carns et al., 2011). However, beginning in the third year, the curve for misdemeanants flattens more quickly than that for felons. The difference between the cohorts becomes smaller, until year five, when the two lines overlap for a couple of years. By year seven, the recidivism rates are slightly higher for felons than for misdemeanants.
For misdemeanants, the recidivism curve continues to flatten through the eight-year follow-up period. In contrast, the percentage of felons who recidivated for the first time increases in the eighth year over the prior year.
The general felony and misdemeanant analysis mirrors the findings of our RF cohort analysis, surfacing questions about why we are seeing a slight uptick in recidivism when we look farther out.
Recent AJiC research to support Alaska RF is providing a more nuanced look at recidivism among Alaska’s criminal offenders. In general, the RF findings corroborate reports of recidivism patterns one to three years after release (Carns et al. 2007; Carns et al., 2011). In particular, these results confirm the critical importance of supporting the early transition to the community.
What sets the RF study apart, however, is that it explores conviction patterns for general offenders beyond three years. This allows us to extend patterns and surface questions about longer term recidivism. Recidivism curves continue to flatten beyond the three-year mark of previous Alaska studies. However, this trend begins to change by year six.
In year six, we begin to see a difference between felony offense-based cohorts and misdemeanant cohorts. The felony cohorts’ recidivism curves rise more steeply relative to previous years while the misdemeanant curves continue to flatten. When we look at all felons and misdemeanants in the eighth year, the felons, who had been less likely than misdemeanants to recidivate initially, are slightly more likely to do so. Taken together, these findings hint at a possible emergence of higher long-term recidivism for felons versus misdemeanants. Further exploration of long-term patterns is important to understand factors that might explain and mitigate an increase in risk of recidivism after many crime-free years.
Araceli Valle is a research professional with the Alaska Justice Information Center (AJiC).
Carns, Teresa White; Cohn, Larry; & Martin, Stephanie. (2011). Criminal Recidivism in Alaska, 2008 and 2009. Anchorage, AK: Alaska Judicial Council.
Carns, Teresa White; McKelvie, Susan; Cohn, Larry; & Martin, Stephanie. (2007). Criminal Recidivism in Alaska. Anchorage, AK: Alaska Judicial Council.
Durose, Matthew R.; Cooper, Alexia D.; & Snyder, Howard N. (2014). “Recidivism Patterns of Prisoners Released in 30 States in 2005: Patterns from 2005 to 2010.” BJS Special Report. Bureau of Justice Statistics. NCJ 244205.
Myrstol, Brad A.; Rivera, Marny; & Parker, Khristy L. (2016). Alaska Sex Offender Recidivism and Case Processing Study: Final Report. Anchorage, AK: Alaska Justice Statistical Analysis Center, University of Alaska Anchorage. (JC 1408.02).
Valle, Araceli. (2017). Alaska Results First Initiative: Adult Criminal Justice Program Benefit Cost Analysis. Anchorage, AK: Alaska Justice Information Center, University of Alaska Anchorage.