Program Profile: Philadelphia (Pa.) Low-Intensity Community Supervision Experiment

This was an experiment conducted to examine the effects of lowering the intensity of community supervision with low-risk people in an urban community. The program is rated Promising. There were no statistically significant effects between groups on rearrest, time to rearrest, reincarceration, frequency of offending, or seriousness of offending. Overall, there was no evidence that reducing the intensity of supervision had any effect on the subsequent criminal behavior of low-risk people.

A Promising rating implies that implementing the program may result in the intended outcome(s).

This program's rating is based on evidence that includes at least one high-quality randomized controlled trial.

A Promising rating implies that implementing the program may result in the intended outcome(s).

This program's rating is based on evidence that includes at least one high-quality randomized controlled trial.

Program Goals

The Philadelphia Low-Intensity Community Supervision Experiment was conducted to examine the effects of lowering the intensity of community supervision with low-risk people in an urban community. The purpose of the experiment was to test an alternative to the regional supervision model used by the Philadelphia (Pa.) Adult Probation and Parole Department (APPD), for which caseloads per officer had increased steadily for decades. The goal was to test whether no- or low-risk individuals could be supervised in large caseloads without increasing recidivism and risk to the public.

Earlier research on supervision intensity concentrated on the intensification of supervision programs (ISPs) for serious, higher-risk individuals on probation and parole. Many of the studies examining the effects of ISP for low-risk individuals found that recidivism rates actually increased in some cases (Barnes et al. 2010). There is little research currently available that has looked at the effects of reducing the intensity of supervision for less serious individuals. This was the first randomized trial of low-intensity probation or parole conducted at any level of risk classification (low risk or high risk).

Target Population/Target Sites

The aim of the experiment was to test the effects of reducing the intensity of community supervision for individuals who would be at low risk of committing serious offenses such as murder, attempted murder, aggravated assault, robbery, or sex crimes. Low risk was defined as “a forecast of no charges for serious crimes within 2 years of the probation or parole case start date” (Barnes et al. 2010, 168).

Low-risk individuals were identified using a modified version of a random forests model, which was originally designed to forecast homicide or attempted homicide (Berk et al. 2009). The basic method involved using information on each individual (prior criminal record and other baseline data) to forecast risk at the beginning of supervision, based on the recent 2-year outcomes of individuals with similar characteristics under the supervision of the APPD.

The experiment was conducted in the western and northeastern regions of Philadelphia.

Program Theory

The experiment was informed by three perspectives in criminological theory: specific deterrence, defiance theory, and deviant peer contagion and deviancy training. Specific deterrence predicts that crime will increase if supervision is decreased, whereas defiance and deviant contagion predict that crime will go down if supervision is reduced. The experiment was guided by these theories but did not aim to test their strength.

Program Components

Individuals in the experimental treatment group received low-risk, low-intensity supervision. They were placed into a fairly sizeable caseload with other low-risk individuals: each probation officer received and maintained a caseload of about 400 individuals. With such large caseloads, probation officers were not able to invest a substantial amount of time on each case.

Low-risk individuals were intended to receive a considerably reduced level of supervision compared with the standard model of regional supervision. The low-risk supervision protocols included:

• Office reporting. Participants were to have a scheduled office visit with the probation officer once every 6 months. These visits concentrated on a review of the participant’s residence, employment, payments of fines, and compliance with other conditions.
• Telephone reporting. Participants were to have a scheduled telephone report once every 6 months, occurring about midway between office visits. These contacts concentrated on confirming details described in the office visits. They were not restricted from initiating additional telephone contact.
• Drug testing. Drug tests were administered only if required by court order. Probation officers were instructed to order a drug evaluation after no more than three positive urine tests. They could be referred for drug treatment if they requested it.
• Missed contacts. Arrest warrants were issued if there was no contact with the person for 6 months. If the individual surrendered voluntarily, the warrant could be removed with no criminal sanction.

Low-risk individuals were informed on their first visit that they were in a low-risk caseload and subject to the reduced reporting requirements described above. They were also informed that they would be transferred back to standard supervision if they were rearrested for a new crime or if an arrest warrant was issued because there was no contact for 6 months.

Key Personnel

The experiment was conducted with the cooperation of the APPD of the First Judicial District of Pennsylvania in Philadelphia. Low-intensity supervision was delivered by two probation officers, one for the western region and one for the northeastern region.

Study

The Philadelphia Low-Intensity Community Supervision Experiment conducted by Barnes and colleagues (2010) used a randomized trial to test the effects of lowering the intensity of community supervision with low-risk individuals. The field experiment took place from 2007 to 2008 in the western and northeastern regions of Philadelphia, Pa.

Data on a total of 12,233 probation cases involving 7,830 different individuals under active supervision was extracted from the case management database run by the Adult Probation and Parole Department (APPD). The data was linked to court history data from the Court of Common Pleas Case Management System. Eligible persons were identified with the random forests model (described in the Program Description). Cases were eligible for random assignment only when more than 50 percent of the votes forecasted an absence of serious offenses.

This narrowed the data down to 2,859 cases that were screened for exclusionary criteria before random assignment. People were ineligible if they were scheduled to end supervision within 60 days of the start of the experiment, they were ordered by court to a specialized unit, they were supervised in an existing low-risk caseload, or they were in potential direct violation of their sentence owing to an arrest occurring after the start of their supervision. This left a total sample size of 1,559 participants who were randomly assigned to either low-intensity community supervision (experimental treatment group) or standard regional supervision (control group).

In the western APPD region, 400 individuals were assigned to the low-risk caseload and 401 to the control group. In the northeastern APPD region, 400 individuals were assigned to the low-risk caseload and 358 to the control group. The experimental treatment group was 66.5 percent male, 48.0 percent African American, and 42.8 percent white, with an average age of 40.8 years. The control group was 67.6 percent male, 48.4 percent African American, 38.9 percent white, and with an average age of 40.6 years. There were no significant differences between groups on demographic variables.

The experiment was a single-blind study, because the probation officers of the people in the low-intensity supervision group were informed of their involvement in the experiment, whereas probation officers of the control group were not informed of their inclusion. Once the experiment began, the number of contacts with probation officers experienced by the experimental group was reduced by about 45 percent, while the control group had almost the exact same amount of contact as they had encountered during the previous year.

Recidivism was measured as any charges that occurred after the start of the experiment. Recidivism data was collected from court records and was limited to offenses that were dealt with by the Philadelphia courts (offenses that took place outside the city limits were excluded from the analysis). The data included information on time from release to first rearrest, reincarceration, frequency of offending, and seriousness of offending. Serious offenses were defined as murder, attempted murder, aggravated assault, or sex crimes. The period of outcome measurement was October 1, 2007, through Sept. 30, 2008.

The study used intent-to-treat analysis, which means all participants were retained in the treatment group even if they were not permitted to participate in the low-intensity supervision treatment. Approximately 17.8 percent of the experimental participants were retained in the analysis of the treatment group, even though they did not receive low-intensity supervision.

These sources were used in the development of the program profile:

These sources were used in the development of the program profile:

Ahlman, Lindsay C., and Ellen M. Kurtz. 2009. The APPD Randomized Controlled Trial in Low-Risk Supervision: The Effects of Low Risk Supervision on Rearrest. Philadelphia, Pa.: Adult Probation and Parole Department.

Barnes, Geoffrey C., Jordan M. Hyatt, Lindsay C. Ahlman, and Daniel T.L. Kent. 2012. “The Effects of Low-Intensity Supervision for Lower-Risk Probationers: Updated Results From a Randomized Controlled Trial.” Journal of Crime and Justice 35(2):200–220.

Berk, Richard A., Lawrence W. Sherman, Geoffrey C. Barnes, Lindsay C. Ahlman, and Ellen M. Kurtz. 2009. “Forecasting Murder Within a Population of Probationers and Parolees: A High-Stakes Application of Statistical Learning.” Journal of the Royal Statistical Society, Series A 172(Part 1):191–211.