The purpose of this study was to provide a comprehensive assessment of the usefulness and effectiveness of prediction and classification of offenders under community supervision. The researchers sought to test the accuracy of the most commonly used third generation prediction and classification instrument (i.e., Wisconsin Risk and Need Assessment) and explore the creation of a new fourth generation prediction model. Specifically, goals of this study were to (1) test the predictive validity of the variables and determine the extent to which index predictors can be created by combining variables, (2) assess how well the current risk assessment instrument as applied in Texas correctly sorts offenders into differential supervision levels, (3) determine the predictive accuracy of models constructed using the Wisconsin Risk and Need variables, (4) develop an improved prediction instrument, and (5) develop objective measures of criminogenic need factors that produce more valid and reliable predictors.
A felony cohort data collection instrument was developed by the Texas Department of Criminal Justice (TDCJ) and the Criminal Justice Policy Council (CJPC) to test the validity of the Wisconsin Risk and Need Instrument in use in Texas, as well as to develop "better" predictor variables for a variety of dependent variables (e.g., rearrest, probation failure, drug use, or absconding). Using the felony cohort data instrument, which included all of the Wisconsin Risk and Need variables as well as a combination of various static and dynamic predictor variables, the Texas Department of Criminal Justice Community Justice Assistance Division (TDCJ-CJAD) collected detailed statewide information on 3,405 felony offenders placed on probation in Texas during October 1993. Specifically, the form was completed by a probation officer on all felony probation intakes made in Texas during October 1993 at the time the initial case classification risk/needs assessment was conducted, which was within 45 days of intake. Additionally, follow-up forms were developed and administered to track the offenders' progress at one year, two years, and three years.
To ensure data validity and reliability, TDCJ-CJAD and CJPC conducted a total of ten training sessions across the state attended by 519 representatives of Community Supervision and Corrections Departments (CSCDs). Site validity checks were also conducted during October and the first part of November 1993. Completed felony cohort forms were checked for accuracy against the intake and supervision information located in the offender's file. The CSCD staff assisted in pulling and reviewing the files. The CSCDs not involved in site validity checks sent their first batch of completed forms to TDCJ-CJAD and CJPC for an initial review. The designated Evaluation Coordinator for each CSCD was instructed to collect all Felony Cohort data forms for the department, ensure that forms were completed for each original felony supervision intake, and mail the completed forms to TDCJ-CJAD. Lastly, forms delivered to TDCJ-CJAD were manually checked for reporting errors.
The initial sample consisted of 4,929 offenders convicted of at least one felony offense that resulted in placement on community supervision during October 1993 in Texas. Of these 4,929 offenders, valid felony cohort forms were completed for 4,245 individuals. Thus, the felony cohort sample consists of 4,235 offenders from 116 Community Supervision and Corrections Departments (CSCDs).
Some offenders in the sample were under indirect supervision at the time the questionnaire was completed. Indirect supervision occurs when an offender transfers to another county or state, absconds, or is serving time in a jail, prison, or other secure residential facility. Since these offenders were not available to be interviewed, only demographic, offense, and sentence information were completed. Those offenders under indirect supervision were not included in this study.
Additionally, follow-up information was not available over the three-year follow-up period for 288 offenders. Therefore, the 288 offenders were filtered from the dataset and were not included in this study. This made the final sample 3,405.
Offender risk and need scores were computed using the current Texas Wisconsin Risk and Need weighting scheme.
Mode of Data Collection:
Data were obtained from face-to-face interviews by a probation officer with the offender and from official records including such documents as the pre- or post-sentence investigation reports, conditions of probation documents, Department of Public Safety (DPS) rap sheets, and other probation intake documents.
Description of Variables:
Variables include probationer information, current offense, criminal history, social history, substance abuse, probation sanctions, case classification risk items, and case classification need items. Additional variables include felony cohort one-year follow-up data form questions, felony cohort second-year follow-up data form questions, and felony cohort third-year follow-up data form questions.
The number of forms completed by each Community Supervision and Corrections Department (CSCD) was checked against the number of "Felony Original Probation Placements" reported to the Texas Department of Criminal Justice Community Justice Assistance Division (TDCJ-CJAD) on the required Monthly Community Supervision and Corrections Report (MCSCR, section II.A1 1.). During October 1993, 4,929 community supervision intakes were reported by CSCDs on the MCSCR. There were 4,245 valid felony cohort forms completed. This amounts to an 86 percent response rate. No systematic reason for missing forms was found during the error resolution phase.
Presence of Common Scales:
The Wisconsin Risk and Needs Assessment was used in this study. Additionally, the principal investigator constructed an education scale, an employment scale, a substance abuse scale, a current offense scale, and a criminal history scale.
Extent of Processing: ICPSR data undergo a confidentiality review and are altered when necessary to limit the risk of
disclosure. ICPSR also routinely creates ready-to-go data files along with setups in the major
statistical software formats as well as standard codebooks to accompany the data. In addition to
these procedures, ICPSR performed the following processing steps for this data collection:
Standardized missing values.
Checked for undocumented or out-of-range codes.