Classifying Inmates for Strategic Programming in the New York Department of Corrections, 1997-1998 (ICPSR 3205)

Time Period(s) View help for Time Period(s)

Date of Collection View help for Date of Collection

Data Collection Notes View help for Data Collection Notes

Hide

Study Purpose View help for Study Purpose

The main goal of this study was to construct and test a statistically based system of classifying inmates for one or more types of Substance Abuse Intervention Division (SAID) sponsored treatment programs within the New York Department of Corrections (DOC) system. In the New York DOC, detainees generally were screened and identified for SAID treatment on the 10th or 11th day after admission to jail. DOC planned to create two sets of treatment services (each requiring about 45 days to administer) for detainees that qualified for treatment: one for those who were likely to go to state prison and another for those who were likely to be released back into the community after serving a jail sentence or serving time for pre-trial detention. DOC's primary objective was to be able to identify two groups on the day of SAID screening: detainees who would be released to the community but would stay approximately 45 days or more in the system, and detainees who would go to state prison but remain at least 45 days in the system. Building statistical models to identify these two groups was the central objective of the study.

Study Design View help for Study Design

DOC provided data items from its IIS database on each of the cases. While original plans called for researchers to gather additional publicly available data on each case for purposes of analyses, DOC administrators indicated their interest in conducting one set of analyses that used DOC data only. They were concerned about the department's ability to quickly routinize the process of obtaining and integrating the added data so that it could be employed in classifying inmates for the new SAID program within 11 days of admission to custody. Researchers agreed to collect two sets of data: one that employed only DOC data and another that included the supplementary data from the Division of Criminal Justice Services (DCJS) and the New York City Criminal Justice Agency (CJA). Matching and merging the DCJS and CJA data with the original DOC sample reduced the size of the sample. To obtain the DCJS data, researchers provided DCJS with state identification numbers and arrest dates for the original cases. DCJS returned perfect matches on a subset of these cases. Some of the initial matches were dropped from the sample because cases were sealed (because the charge was dismissed or was disposed as a juvenile case) and could not be released to the researchers under state statute. Since the information returned by DCJS included multiple events when they existed for a given arrest date, the dataset was reduced to information associated with the top charge for each arrest date. This ensured that the DCJS data would be comparable to the DOC data, also based on the top charge.

Sample View help for Sample

The analytic sample used for the prediction models was a sample of recent jail inmates identified as eligible for admission to SAID. Selected from the DOC Inmate Information System (IIS) database, the initial sample included people who entered DOC custody as detainees (the sample did not include offenders entering with jail sentences) and who were released between July 1, 1997, and June 30, 1998. This group was then screened for SAID eligibility. To be eligible for SAID, inmates could not have any convictions for any weapons-related infractions, convictions for any violence-related infractions within five years, arson-related convictions or charges, or convictions or current charges for sex-related offenses. Further screening criteria were specified through discussions with DOC to ensure that the dataset included only detainees who would be eligible for the new, planned SAID program. This meant ensuring that detainees in the sample had been in custody at least 11 days (and had not received a jail sentence by that time) and were not technical parole violators. Also excluded from the sample were small subgroups of inmates identified as cases who would be treated differently by the court (or appeared to have been based on their actual time in custody) when compared to the general population of SAID-eligible inmates. These cases included inmates who had absconded or escaped from DOC custody, or left custody as a result of death, suicide, a warrant being lifted, transfer to other jurisdictions of the criminal justice system, or left custody for an unknown reason.

Universe View help for Universe

Detainees of the New York Department of Corrections.

Unit(s) of Observation View help for Unit(s) of Observation

Data Source View help for Data Source

Data Type(s) View help for Data Type(s)

Description of Variables View help for Description of Variables

The DOC data include gender, ethnicity, age, self-reported drug use, marital status, and citizenship. Information on the current case includes bail amount, type and severity of the charge, whether the charge was remanded, borough of arraignment, and whether there was a warrant existing at the time of arrest. Prior criminal record information includes prior convictions, history of violence, and disciplinary history. SAID eligibility information includes total length of stay in DOC, whether the inmate was prison-bound or not, and DOC classification score. Part 2, DOC and Supplementary Data, includes all the DOC items stated in Part 1 as well as other data collected from DCJS and CJA. Additional data items include flags indicating whether the charge was for an offense involving drugs, a firearm, another weapon, or a child victim, a DWI or DWAI arrest, whether the person was eligible for probation, prior arrests, prior jail, prison, and probation sentences, the length of time a defendant had been residing at current address, and DCJS's release recommendation.

Response Rates View help for Response Rates

Not applicable.

Presence of Common Scales View help for Presence of Common Scales

None.

Hide

Original Release Date View help for Original Release Date

2003-06-13

Version History View help for Version History

Hide