Study 1
Dynia and Sung (2000) compared the recidivism rates of Drug Treatment Alternative to Prison (DTAP) program participants to similar people who did not participate in the program. The study sample included 487 people who were arrested in police undercover busy-and-bust operations for drug sales from December 1990 to December 1991. All of them were initially eligible to participate in DTAP. In total, 272 people opted to participate in DTAP treatment. Of those, 184 successfully completed treatment and 88 left treatment before completion. The remaining 215 from the study sample did not participate in DTAP, either because they declined to participate or because they were rejected during the screening process.
The study sample was divided into three groups: completers (n = 184), failures (n = 88), and nonparticipants (n = 215). Completers were 77 percent male, and 56 percent Hispanic, 34 percent Black, and 9 percent white, with an average age of 31 years. Failures were 76 percent male, and 63 percent Hispanic, 28 percent Black, and 9 percent white, with an average age of 28 years. Finally, nonparticipants were 80 percent male, and 57 percent Hispanic, 37 percent Black, and 6 percent white, with an average age of 30 years. There were no significant differences between the groups in terms of gender, race, and criminal history. The age between DTAP completers and failures was statistically significant; participants who completed treatment were slightly older than those who dropped out or were expelled.
Recidivism data were collected from New York State criminal history records. Official arrest charges were classified by penal law severity (felony versus misdemeanor) and offense type (drug versus nondrug). The analysis looked at the following time periods for program completers: 3 years before the DTAP arrest, time during treatment, and 3 years after treatment. For DTAP failures, the analysis looked at 3 years before the DTAP arrest, time during treatment, at-large time (time between treatment drop out and apprehension by the DTAP warrant enforcement team), and 3 years following release from prison. For nonparticipants, the analysis looked at 3 years before the DTAP arrest, pretrial, time during sentence, and 3 years after prison release or sentence competition.
To assess public safety risk, the analysis compared rearrest prevalence rates for DTAP completers and failures during treatment and at-large periods, respectively, to the rearrest rate for nonparticipants during the pretrial and sentence periods. To assess treatment effectiveness, rearrest rates were compared for the 3-year pre- and post-periods between the three groups. Two different analyses were performed: chi-square and survival analysis.
Study 2
Belenko, Foltz, Lang, and Sung (2004) used a longitudinal quasi-experimental design to analyze the recidivism impact of DTAP on program participants. The study included a prospective treatment sample of 150 people diverted to DTAP (90 of whom completed DTAP and 60 of whom dropped out or failed) and a comparison sample of 130 people sentenced to State prison. The study also used retrospective samples of 64 DTAP graduates and 68 DTAP dropouts who were randomly selected from all DTAP participants who entered the program before the current evaluation began (May 1995) but who graduated or dropped out during the same time that the prospective sample was in treatment. It was necessary to include this retrospective sample because DTAP residential treatment can last up to 24 months, and comparison group members are usually sentenced to several years in prison. Therefore, the retrospective sample ensured a sufficient number of participants were available to track recidivism in the community for a multiyear follow-up period.
Participation in the study was voluntary. The prospective treatment group members were repeatedly convicted of felony offenses and arrested in Brooklyn, N.Y., in 1995 and 1996 but diverted to DTAP. The comparison group members were repeat felons sentenced to prison from other boroughs of New York City. The comparison group was sentenced to prison terms that were equal to sentences that DTAP participants would have received if they had not been diverted to treatment (about 18 to 36 months). The study participants were 88 percent male, and 59 percent Hispanic and 38 percent African American. Average age of the participants was 33 years. There were no significant differences between the groups, except that comparison group members had a greater number of prior misdemeanor convictions.
At intake, a number of interview instruments were administered to prospective study participants, including items related to drug use, prior treatment history, criminal activity, employment and earnings, physical and psychological health, HIV risk behaviors, and social stability. Standardized assessment instruments included the Addiction Severity Index (ASI), the Michigan Alcohol Screening Test (MAST), the Risk Behavior Assessment, and the Texas Christian University’s Self Rating Form.
Data were collected from the offender tracking database maintained by the New York City Criminal Justice Agency on all arrests as of July 31, 2002. The follow-up period ranged from 0 to 103 months, because DTAP program participants completed treatment or members of the comparison group were released from prison for their original target arrest between January 6, 1994, and April 14, 2003.
Recidivism was measured as new arrest after admission (to DTAP treatment or prison) and new convictions for a criminal offense following release from prison or treatment. The analysis distinguished between charge severity (felony or misdemeanor) and charge type (drug and nondrug). The recidivism data also captured new prison and jail sentences that occurred after release from DTAP treatment or prison. The total months that study participants were incarcerated for these new sentences were found by summing the minimum sentence length received for prison and jail sentences.
DTAP dropouts and comparison group members had less opportunity to reoffend because they were incarcerated for part of the follow-up period, whereas DTAP participants who completed treatment had a longer time in the community and thus had more opportunity to recidivate. These factors could confound any observed treatment effect. Therefore, the analysis required an adjusted annualized rearrest rate to address the differential time in the community across groups.
The analysis used logistic regression to compare the various recidivism measures, and analysis of variance was used to compare group arrest rates. Survival analyses were conducted on the time to first rearrest, while regression models were used to predict the adjusted annual rearrest rate. Wald Chi-Square tests were conducted to examine differences in cumulative prevalence for the analysis of rearrest by time in the community.