This study was a successful collaboration between the Florida Correctional Privatization Commission, the Florida Department of Corrections and the Florida State University School of Criminology and Criminal Justice to examine whether public and private prisons produce different recidivism outcomes. Major advancements were made to the research methodologies used in previous studies of this important issue, including how exposure to public versus private prisons was measured, larger numbers of cases studied, and the increased breadth of control variables used in the analysis.
This study examined the effectiveness of private prisons, relative to public prisons, in terms of affecting recidivism rates as measured by a new offense or re-imprisonment after prison release. Six different levels and types of exposure to public and private prisons were quantified; three specific offender types were examined based on the different types of inmates private prisons house: adult males, adult females, and youthful offender males; and two measures of recidivism were examined (re-offense and re-imprisonment). Therefore, in total, there were thirty-six separate comparisons of recidivism rates between public and private prisons, each controlling for factors known to affect these rates after prison release (age at release, gender, prior recidivisms, etc.).
For adult males, there were no differences in recidivism rates found between public and private prisons, a result consistent with the previous study conducted of Florida inmates by Farabee and Knight (2002). Also consistent with Farabee and Knight, among youthful offender male inmates, no differences in recidivism rates were found between the public and private prisons.
For adult females, in only one of twelve measures of public and private prison exposure (the purest group – B1) and recidivism (re-offense) were private prisons found to significantly reduce the likelihood of recidivism relative to public prisons. In this analysis, the private prison exposure definition (A2) closest to that used by Farabee and Knight did not demonstrate an effect for re-offending or re-imprisonment. This could be because Farabee and Knight analyzed the latest of multiple prison releases from the same commitment rather than the first of multiple releases, which this analysis employed. Or it could result from Farabee and Knight not controlling for disciplinary reports, which are known to influence recidivism rates.
This study’s purer definition (B1) of private prison exposure did yield an effect, the result found by Farabee and Knight that females released from private prisons re-offend at lower rates is partially confirmed. In fact, they found that females released from public prisons were 25% more likely to re-offend than private prison females. The size of this effect is quite similar to this study’s finding that private prison females are 18% less likely to re-offend than public prison females. However, this study does not confirm Farabee and Knight’s finding of lower reimprisonment rates for private prison females. This difference is probably caused by their use of latest release instead of first release on a prison commitment, because counting recidivism beginning with the most recent of multiple prison releases ignores earlier returns to prison.
This study’s finding of a lower re-offense probability from private prison exposure should be taken as provisional for two reasons. First, the measurement’s statistical significance depends on forcing the exposure measure into a regression model; the treatment effect did not enter the statistical model based on its independent correlation with re-offense rates. Second, although the re-offense effect for females does appear when using the purest measure of exposure to private prisons, the number of females examined is also the smallest number of cases for a private prison group (640 cases) and substantially smaller than the other female private prison exposure groups (ranging from 1,133 to 1,866 cases). A future analysis with more cases that fit the purest exposure definition would be needed to determine whether this effect is an artifact of the relatively small number of cases.
In summary, in only one of thirty-six comparisons was there evidence that private prisons were more effective than public prisons in terms of reducing recidivism. This indicates that, at this time, public and private prisons are essentially the same in terms of their relative effectiveness in preventing inmates from being re-imprisoned for adult males, adult females, and youthful offender males after release from Florida’s prisons. Likewise, no effect appears on reoffense rates for adult males and youthful offender males attributable to private prison exposure. For adult females, only one of six measures of private prison exposure is associated with a lower re-offense probability for private prison inmates and this finding should be considered provisional until a larger number of cases are available to analyze.