Results: Adult Females

Table 8 presents the base recidivism rates for adult female inmates using the recidivism re-offense measure generated from the SAS© life-test procedure with follow-up periods of six months to sixty months. These data show that, in the short term through an eighteen month follow-up period, the re-offense rates of public and private female inmates are virtually identical within five of the six outcome measures. Only the B1 group shows public female inmates exhibiting re-offense rate differences greater than 2% with public females recidivating slightly more than private female inmates (21.7% versus 19.4%). In the longer follow-up periods of thirty and thirty-six months, again only the B1 group indicates any difference greater than 2% with public female inmates having higher re-offense rates (34.0%) than private female releases (31.7%). However, none of these re-offense rate differences between the public and private adult female inmates are statistically significant at the p<.05 level.

Re-imprisonment rates for adult female releases are displayed in Table 9, prior to controlling for known relevant predictors, for the public and private releases within each of the six definitional groups. For the short term re-imprisonment rates up to eighteen months after prison release, none of the differences in the public and private female releases are meaningful with all groups revealing a less than a 2% difference. For the longer term rates up to thirty-six months, only within the B1 group was there a difference of more than 2%. Within this group, public female inmates were re-imprisoned at a 2.8% higher rate than private females. However, none of these re-imprisonment rate differences between public and private inmates are statistically significant at the p<.05 level.

Table 10 presents the distributions of the private and public adult female inmates on the variables used in the final survival models to answer the question whether recidivism rates differ across the two groups. In terms of the demographic characteristics, adult females from public prisons were somewhat younger at release than those from private prison; however, the difference was not statistically significant (mean age: public = 33.4; private = 35.6). The racial and ethnicity distributions across the public and private groups were virtually the same. In contrast, private female inmates were somewhat more likely to have more prior recidivism events than public female inmates and these differences were statistically significant (average: private = 0.9; public = 0.7, p<.01) but are very similar in terms of the percentage that were close custody inmates (private = 29.5%; public = 28.9%). Public female inmates were much more likely to spend longer lengths of time in prison (average months: public = 20.4; private = 16.4) and have more disciplinary problems while in prison (average disciplinary reports: public = 2.8; private = 0.7). The median TABE score for public was slightly lower than for private female inmates (6.6 versus 7.3) and more likely to be released to some form of post-prison supervision (26.5% versus 22.0%). In terms of their most serious crime type, public female inmates are more likely to have committed a homicide, sex/lewd offense, or a robbery than private female inmates. However, the two groups are very similar in terms of their total number of convictions for property, drug, and weapons offenses.

The differences described on many of the control variables for the public and private adult female inmates indicate the need to control for these factors when evaluating whether the recidivism rates differ between the two groups. Public female inmates are younger at release, and have more disciplinary infractions as well as lower TABE scores. These characteristics are all associated with a higher likelihood of recidivism. In contrast, private inmates have more prior recidivism rates and shorter prison stays, and are less likely to have post-prison supervision, which are also predictive of higher recidivism rates.

Even though the base recidivism rate comparison between the public and private prison experience for adult females revealed no statistically significant differences, the differences in these populations indicates the need to use multivariate survival models to control for the effects of these variables on recidivism rates. Table 11 displays these results and shows that, for the re-offense recidivism measure, only the B1 group reveals a statistically significant difference with the private female group experiencing lower recidivism rates than the public female group. However, the private prison effect in the B1 group is marginally significant (p = .043), and when performing a stepwise survival model which requires an entry p value of .05 or lower, the treatment effect does not enter the model. These contradictory results may be due to the proximity of the p-value to the stepwise significance level threshold and some slight interaction between the treatment indicator variable and the other control variables. None of the re-imprisonment rate differences between public and private adult female inmates within each of the six methods of quantifying the type of prison experience were statistically significant.