Bureau of Prisons researchers strive to evaluate what works.

During the 1980s, the number of drug offenders incarcerated in the Federal Bureau of Prisons increased dramatically--from 5,100 inmates (25 percent of the sentenced population) in 1981 to 30,498 inmates (57 percent) in 1991.

Based on this shift in the inmate population, the BOP has begun actively re-examining its drug treatment efforts and has increased its capacity to provide new and expanded drug treatment programs throughout its facilities. This has been done through a six-part drug treatment strategy that involves screening, education, residential (unit-based) treatment, non-residential treatment, transitional treatment and evaluation. This article discusses the evaluation component of this strategy, focusing on research design and data collection efforts.

In 1990, the BOP entered into an agreement with the National Institute on Drug Abuse to evaluate its drug treatment strategy. To collect information on the impact of the BOP's treatment programs, the agencies developed a multi-site longitudinal evaluation, which has since become a critical element of the Bureau's overall treatment initiative.

The evaluation assesses the effectiveness of the Bureau's program--which incorporates both residential intervention and transitional treatment after release--in reducing recidivism and substance abuse relapse. In addition, it attempts to determine which individuals are likely to require intensive treatment and who might benefit from less intensive treatment.

Research Design

The research design proposes to evaluate the BOP's drug treatment programs by comparing inmates participating in various types of drug treatment programs with non-participants. Treatment program subjects are divided into three subgroups:

* those in the 12-month, 1,000-hour residential (unit-based) treatment program;

* those in the nine-month, 500-hour (unit-based) residential treatment program; and

* those in the non-residential (general population) treatment programs.

In the original research design, which has since undergone two significant revisions, residential treatment volunteers were to be randomly assigned to one of the three subgroups only if the number of volunteers exceeded the number of available treatment slots. In the event that too few inmates volunteered for the programs, the research design also incorporated features to address external and internal validity issues. The collection of comprehensive information on all research subjects ensured statistical control of pre-treatment differences between the comparison groups.

As the BOP's unit-based drug treatment programs evolved, however, it became increasingly clear that the number of volunteers was going to be low, largely because the BOP rapidly set up programs at additional facilities to provide easier access to treatment for more inmates. Therefore, the study design needed to be modified to better address these enrollment issues.

The original research design was thus replaced by a "quasi-experimental" method that targeted subjects coming from a modified set of comparison groups. The principal study group comprised the inmates who volunteered and started the residential treatment programs, including those who graduated, voluntarily withdrew or were expelled. Sociodemographic characteristics of these group members were used to select comparison subjects from the non-residential treatment and non-treatment groups. These matching variables included gender, age, race, drug-use history, and criminal history characteristics such as severity of instant offense, history of violence, sentence length, and security classification.

A third revision to the study design occurred when it became apparent that there was an insufficient number of subjects immediately available for community follow-up. This was directly related to the enrollment issue addressed by the first revision--namely, that inmates with substance abuse histories were not volunteering for treatment in the numbers that had been expected. Because the 1984 Sentencing Reform Act abolished parole in the federal criminal justice system, there was no clear "carrot and stick" approach to encourage treatment participation. While the BOP has always given priority to inmates closer to their release, treatment programs were filling up with inmates with longer sentence lengths. The BOP felt that denying treatment to inmates with long sentences was unethical when treatment slots were readily available. Thus, the research plan was again changed by expanding the number of research sites to increase the pool of offenders who might be released to the community, increasing the number of available subjects to help assess the outcomes of relapse and recidivism. Currently, residential treatment participants are identified as research subjects at 19 of the Bureau's 31 treatment facilities and a larger pool of subjects have been added to the study. Data Collection

The pre-treatment data collection battery includes measures of the inmates' motivation and cognitive characteristics as well as psychological/psychiatric impairment information and sociodemographic/background data. Background data comes from both the client and from official records. This information includes prior criminal history, drug-use history, institution disciplinary records, urine test results, and program and work involvement.

Data collected on the treatment process includes detailed descriptions of the treatment programs and assessment of whether the program has been implemented as it was intended. Variables observed include staff, focus of treatment, program acceptance by staff and therapist empathy.

Research staff monitor treatment delivery by placing an evaluation specialist at designated research sites. These sites are facilities with residential programs of either nine or 12 months in duration. The researchers prepare detailed program descriptions and track inmates throughout their residential treatment experiences. They administer assessment instruments and attitudinal questionnaires to inmates upon program admission, during treatment and at the end of the treatment phase.

A clear advantage of having on-site specialists is the ability to merge data collected from the subject with information made available by the treatment staff. These data include performance on assignments, cooperation in program objectives, rule infractions and attendance.

Post-release Data

After release from the BOP's correctional facilities, research subjects may participate in further treatment in the community corrections environment. Community supervision and treatment staff provide data on client adjustment and program involvement, and additional information is taken from official BOP records. Staff collect data on program elements within the community corrections centers as well as urine test results, details of employment, educational experiences, and new criminal activity or rule infractions.

The U.S. probation/parole offices play an important role in assessing program outcomes after research subjects are released from custody. Evaluation staff

make three contacts with these authorities per subject. The contact intervals are six months after release, 18 months after release, and at the end of supervision or upon revocation.

The purpose of these contacts are to provide information on community outcomes such as arrest rates, employment rates, relapse to drug use through urine testing, type and intensity of continued substance abuse treatment, and residential and family environment information.

The Bureau of Prisons has made it a priority to provide appropriate treatment to offenders based on their individual needs. Data collected during this evaluation initiative can help guide managers as to what works so they can develop and implement programs that are truly effective.

Anita Arcidiacono is drug evaluation administrator with the Office of Research and Evaluation of the Federal Bureau of Prisons.