Table of Contents | Previous | Next |
Results for Groups Defined by Time Since Release
CEO’s program model is designed to provide immediate employment and income to ex-prisoners in the critical weeks following release. A large majority of CEO’s overall client population comes to the program either immediately upon release or shortly thereafter.
As discussed earlier, however, a large proportion of the study sample members had been out of prison for a long time before coming to CEO. Once this became clear to the research team, there was great interest in looking at results separately for the 377 sample members who came to CEO within three months after release from state prison.1 This “recently released” group is closer to the population for whom the program was designed, and results for this group may provide better evidence about CEO’s impacts for its overall service population. More broadly, the “recently released” group better fits the usual definition of a “reentry” population, and the results for that group may help to test the widely held assumption that reentry programs are more effective if they begin working with ex-prisoners immediately after release.2
Tables 7, 8, and 9 show results for the same measures that were described above for the full sample, but focusing only on sample members who came to CEO within three months after release — that is, the “recently released” group.3 There is more uncertainty about these results because of the relatively small sample size, but the general pattern of results appears to be more positive than for the full sample, particularly with respect to recidivism outcomes. Results for the rest of the sample — those who came to CEO more than three months after release — are shown in Appendix Tables A.3-A.5, and they are less favorable.
Table 7 shows that the general pattern of employment impacts for the recently released group is similar to the results for the full sample — impacts are very large in the early quarters and not statistically significant by the end of the year — but the impact appears to decline more slowly for the recently released group.
| Outcome (%) | Program Group | Control Group | Difference (Impact) | Standard Error | |
|---|---|---|---|---|---|
| Neighborhood Work Project (NWP) employment | Ever employed in NWP Q1-Q4a | 73.0 | 0.5 | 72.4 *** | 3.832 |
| NWP employment Qtr 1 (quarter of random assignment) | 66.3 | 0.5 | 65.7 *** | 4.116 | |
| NWP employment Qtr 2 | 47.2 | 0.5 | 46.7 *** | 4.165 | |
| NWP employment Qtr 3 | 15.1 | 0.0 | 15.1 *** | 3.010 | |
| NWP employment Qtr 4 | 5.1 | 0.0 | 5.1 *** | 1.687 | |
| All UI-covered employment | Ever employed in Q1-Q4 | 85.4 | 53.3 | 32.1 *** | 4.700 |
| Employed Qtr 1 (quarter of random assignment) | 69.9 | 26.0 | 43.9 *** | 4.900 | |
| Employed Qtr 2 | 65.5 | 30.9 | 34.6 *** | 5.200 | |
| Employed Qtr 3 | 47.8 | 35.0 | 12.8 ** | 5.200 | |
| Employed Qtr 4 | 38.9 | 33.7 | 5.2 | 5.000 | |
| Employed all 4 quarters Q1-Q4 | 25.2 | 10.3 | 14.9 *** | 3.900 | |
| Sample size (total = 377) | 222 | 155 | |||
| SOURCES: MDRC calculations from CEO's Network for Information Gathering Evaluation and Learning (NIGEL) database and unemployment insurance wage records from New York State. NOTES: Includes sample members randomly assigned between January 2004 and October 2005. A total of two people were excluded from the UI-covered employment outcomes because of missing social security numbers. Results in this table are weighted by week of random assignment and adjusted for pre-random assignment characteristics. Significance levels are indicated as: *** = 1 percent; ** = 5 percent; * = 10 percent. The significance level indicates the probability that one would be making an error in concluding that there is a difference between research groups for the variable in question. The standard error is an estimate of the sampling variation of the impact and is used to assess its level of statistical significance, as shown by the p-value. a One control group member was employed in NWP during the follow-up period. |
Table 8 presents recidivism measures for the recently released group. It shows that the decrease in felony convictions for the full sample, discussed earlier, is concentrated in the recently released group, where there is a statistically significant impact of 4.5 percentage points (though few people in either group had felony convictions). In addition, there is a statistically significant 8 percentage-point decrease in the percent of sample members whose parole status was ever classified as revoked or absconded. These two effects translate into a sizable, 10 percentage-point decrease in overall reincarceration, as well as a 5 percentage-point decrease in the percent incarcerated for a new crime (both effects are statistically significant). There is also a decrease in the total number of months incarcerated in state prison during Year 1. Thus, at the end of Year 1, program group members in the recently released group were substantially more likely than their control group counterparts to remain under parole supervision and substantially less likely to be incarcerated.
Table 9, which combines employment and incarceration measures by quarter, shows, as expected, that CEO decreased the number of recently released people who were “not employed and incarcerated” at the end of Year 1. This table also makes clear that the impact on reincarceration emerged in the latter half of Year 1.
As a work program, CEO can only affect recidivism indirectly through its effects on employment. However, it is notable that a similar pattern of employment impacts for the recently released subgroup and for those who came to CEO later may be leading to much different patterns of impacts on recidivism. This pattern may reflect the fact that the recently released group is at higher risk of reincarceration. For example, 20 percent of control group members in the recently released group were incarcerated in state prison within one year after random assignment, compared to 12 percent of control group members who came to CEO more than 3 months after release (see Tables 8 and A.4). Similarly, the percent incarcerated for a new crime was 5 percent for control group members in the recently released group and only 2 percent for control group members who enrolled more than three months after release. This pattern is not surprising. People who are not in the recently released group had already stayed out of prison for a lengthy period before random assignment, so one would not necessarily expect them to have a high rate of incarceration in the subsequent year. It is possible that, while both groups of participants needed assistance finding jobs, CEO’s employment help was more crucial in avoiding reincarceration for the group at higher risk of returning to prison. Additional analysis of this pattern of results will be included in later reports.
The results for the recently released group are intriguing — few programs have generated substantial impacts on recidivism outcomes in rigorous studies — but not yet definitive. It will be important to see whether the impacts on felony convictions and incarceration hold up over a longer follow-up period (although one could argue that even a short-term delay in reincarceration is a positive result). In addition, data on incarceration in New York City jails will provide a more complete picture of incarceration.
| Program Group | Control Group | Difference (Impact) | Standard Error | |||
|---|---|---|---|---|---|---|
| Arrests and convictions | Arrested (%) | 21.8 | 27.0 | -5.2 | 4.600 | |
| Average number of arrests and first arrest (%)a | 0.3 | 0.4 | 0.0 | 0.073 | ||
| Number of months between random assignment | 3 months or less | 7.8 | 6.2 | 1.6 | 2.900 | |
| 4-6 months | 3.3 | 8.3 | -5.0 ** | 2.400 | ||
| 7-9 months | 5.9 | 6.2 | -0.3 | 2.600 | ||
| 10-12 months | 4.8 | 6.2 | -1.4 | 2.500 | ||
| Convicted of a crime (%)b | 14.1 | 20.3 | -6.2 | 4.000 | ||
| Convicted of a felony (%) | 1.7 | 6.2 | -4.5 ** | 2.000 | ||
| Convicted of a misdemeanor (%) | 9.6 | 11.6 | -2.0 | 3.300 | ||
| Other convictions (%) | 2.7 | 2.9 | -0.3 | 1.900 | ||
| Parole outcome | Absconded/revoked from parole (%)c | 18.8 | 27.0 | -8.2 * | 4.400 | |
| Incarceration in state prison | Incarcerated in state prison (%)d | 9.6 | 19.7 | -10.1 *** | 3.700 | |
| Incarcerated for a new crime (%)e | 0.5 | 5.1 | -4.7 *** | 1.600 | ||
| Incarcerated for a parole technical violation (%)f | 7.4 | 12.0 | -4.7 | 3.200 | ||
| Incarcerated for all other/unknown reasons (%)g | 1.8 | 2.5 | -0.7 | 1.500 | ||
| Average number of months incarcerated in prison | 0.4 | 0.8 | -0.4 ** | 0.183 | ||
| Status at the end of Year 1 | Incarcerated in state prison (%)h | 6.4 | 15.4 | -8.9 *** | 3.300 | |
| Not incarcerated and on parole (%)i | 86.7 | 78.4 | 8.3 ** | 4.000 | ||
| Not incarcerated and not on parole (%)j | 6.8 | 6.2 | 0.6 | 2.600 | ||
| Sample size (total = 377) | 222 | 155 | ||||
| SOURCE: MDRC calculations using data from the New York State (NYS) Division of Criminal Justice Services (DCJS). NOTES: Includes sample members randomly assigned between January 2004 and October 2005. Results in this table are weighted by week of random assignment and adjusted for pre-random assignment characteristics. Significance levels are indicated as: *** = 1 percent; ** = 5 percent; * = 10 percent. The significance level indicates the probability that one would incorrectly conclude that a difference exists between research groups for the corresponding variable. The standard error is an estimate of the sampling variation of the impact and is used to assess its level of statistical significance, as shown by the p-value. a Each arrest date is only counted as a single event. If there are multiple crimes or charges on the same date, only one event per date is recorded in the analysis. b A total of 21 convictions (less than 1%) were found to be associated with an arrest that occurred prior to random assignment. c Measure includes those that absconded and those with a discharge type of "Revoked," according to data from the NYS Division of Parole. d All prison incarcerations (regardless of parole status) according to data from the NYS Department of Corrections. Due to multiple incarcerations, subcategories are not mutually exclusive and may sum to a total greater than the "Incarcerated" measure. e Incarcerations for those committing new crimes, regardless of their parole status. f Incarcerations for those committing technical violations while on parole. g All other incarcerations, including those for technical violations while not on parole and for other unknown/unspecified reasons. h Incarceration status based on month 12 after random assignment. i "Not incarcerated/on parole" includes those not incarcerated and without any parole discharge data in the 12 months after random assignment. j "Not incarcerated/not on parole" includes those not incarcerated in Month 12 who also have a parole discharge in the 12 months after random assignment. |
| Outcome | Program Group | Control Group | Difference (Impact) | Standard Error | |
|---|---|---|---|---|---|
| Employed and not incarcerated (%) | Quarter 1 (quarter of random assignment) | 48.3 | 22.2 | 26.2 *** | 5.100 |
| Quarter 2 | 65.5 | 30.9 | 34.6 *** | 5.200 | |
| Quarter 3 | 47.7 | 34.3 | 13.5 ** | 5.200 | |
| Quarter 4 | 38.5 | 33.0 | 5.5 | 5.000 | |
| Employed and incarcerated (%) | Quarter 1 (quarter of random assignment) | 21.6 | 3.8 | 17.8 *** | 3.700 |
| Quarter 2 | 0.0 | 0.0 | 0.0 | 0.000 | |
| Quarter 3 | 0.0 | 0.7 | -0.7 | 0.600 | |
| Quarter 4 | 0.4 | 0.7 | -0.3 | 0.700 | |
| Not employed and not incarcerated (%) | Quarter 1 (quarter of random assignment) | 12.6 | 40.1 | -27.5 *** | 4.500 |
| Quarter 2 | 32.0 | 64.0 | -32.0 *** | 5.200 | |
| Quarter 3 | 46.3 | 54.0 | -7.7 | 5.400 | |
| Quarter 4 | 52.8 | 49.0 | 3.8 | 5.400 | |
| Not employed and incarcerated (%) | Quarter 1 (quarter of random assignment) | 17.5 | 33.9 | -16.4 *** | 4.600 |
| Quarter 2 | 2.5 | 5.1 | -2.6 | 2.100 | |
| Quarter 3 | 5.9 | 11.0 | -5.1 * | 3.000 | |
| Quarter 4 | 8.3 | 17.3 | -9.0 ** | 3.500 | |
| Sample size (total = 375) | 220 | 155 | |||
| SOURCE: MDRC calculations using data from New York State (NYS) unemployment insurance wage records and NYS Division of Criminal Justice Services (DCJS). NOTES: Includes sample members randomly assigned between January 2004 and October 2005. Two people were excluded from the sample due to missing social security numbers. Results in this table are weighted by week of random assignment and adjusted for pre-random assignment characteristics. Significance levels are indicated as: *** = 1 percent; ** = 5 percent; * = 10 percent. The significance level indicates the probability that one would incorrectly conclude that a difference exists between research groups for the corresponding variable. The standard error is an estimate of the sampling variation of the impact and is used to assess its level of statistical significance, as shown by the p-value. |
1 The “recently released” sample of 377 is 39 percent of the full sample of 977. However, 49 sample members are missing the latest prison release date. Thus, as shown in Table 2, the “recently released group” accounts for 41 percent of those for whom the latest release is known. (back)
2 Appendix Table A.1 shows that the baseline characteristics of the recently released group are generally similar to those for sample members who came to CEO longer after release. However, there are some expected differences. For example, the proportion with UI-covered employment in the year prior to random assignment is 34 percent for those who came to CEO more than three months after release and only 12 percent for the recently released group. In addition, the recently released group is more likely to be living with friends or relatives and less likely to be in transitional housing. Appendix Table A.2 shows that the pattern of program participation is very similar for the two groups. (back)
3 About 43 percent of the sample members in the recently released group came to CEO within one month after release from state prison (not shown in tables). (back)
| Table of Contents | Previous | Next |