For Internal Use Only Adapting BJS s web-based recidivism analysis tool for state-level queries Matt Durose Statistician Bureau of Justice Statistics National Corrections Reporting Program Data Providers Meeting Aurora, Colorado April 2014
Outline of Presentation Methods to convert state criminal history records into national recidivism databases Overview of the new BJS prisoner recidivism study Demonstration of the BJS web-based recidivism analysis tool Discussion on adapting the BJS recidivism analysis tool for state queries
BJS Criminal History Record Research Program In 2008, BJS entered into data sharing agreement with the FBI to directly access criminal history records through the Interstate Identification Index via Nlets for data analysis purposes. The FBI s III is an automated pointer system that allows authorized agencies to determine whether any state repository has criminal history records on an individual. Nlets is a computer-based system that links together state, local and federal criminal justice agencies for the purpose of information exchange.
Rap sheet parsing programs BJS partnered with Nlets to develop- A secure, automated system to retrieve large sets of criminal records on study cohorts. States respond automatically to III record requests over the Nlets network. Software programs parse key data elements from rap sheets in every state. Collates the multi-state records into a uniform structure and produces a relational database. Parsing process leverages the standardized XML format used by some states to transmit rap sheets.
Rap sheet fields in the uniform record layout Subject segment BJS case number Name State identification number (SID) FBI identification number Date of birth Gender Race Social security number Court/Sentencing segment Court disposition date Court name and ORI Statute number NCIC code Offense description Type of court disposition Sentences (prison, fine, etc.) Arrest segment Date of offense Date of arrest Arresting agency name and ORI Statute number NCIC code Charge description Charge severity Arrest disposition Arrest disposition date Supervision segment Supervision date (admission, release, etc.) Agency name and ORI Supervision description
Recidivism of Prisoners Released in 2005 In 2011, BJS used this new data collection process to conduct a recidivism study of state inmates released in 2005. Study included 30 states that were able to provide 2005 release records and FBI or SID numbers. BJS sampled ~70,000 released inmates to represent the ~405,000 released in the 30 states in 2005. Sample included ~2,300 subjects from each state.
Recidivism of Prisoners Released in 2005 Alaska Michigan Oklahoma Arkansas Minnesota Oregon California Missouri Pennsylvania Colorado Nebraska South Carolina Florida Nevada South Dakota Georgia New Jersey Texas Hawaii New York Utah Iowa North Carolina Virginia Louisiana North Dakota Washington Maryland Ohio West Virginia
Recidivism of Prisoners Released in 2005 BJS study excluded Inmates whose sentence was less than 1 year Transfers of custody Releases due to death Releases on bond Releases to seek an appeal Escapes from prison or AWOL.
Recidivism of Prisoners Released in 2005 BJS study excluded about 1,600 sampled prisoners who died during the 5-year recidivism period. Post-release information on deceased subjects obtained from Fingerprint-verified notices in the FBI s Interstate Identification Index (III). Social Security Administration s public Death Master File (DMF).
External Validity Checks on Criminal Records Demographic data in NCRP prisoner records used to verify correct criminal history records received on sampled prisoners. 98.9% match on date of birth. 99.9% match on gender. 99.9% match on race.
Software to convert state rap sheet data into research databases with common codes BJS and NORC to developed software to convert the criminal history data on the released prisoners into research databases with nationally standardized codes Primary component of conversion software includes: State-to-state crosswalk tables to convert each state-specific RAP sheet data into a uniform coding structure. Thousands of state statutes and offense descriptions mapped into about 100 BJS offense codes. Thousands of state court disposition descriptions mapped into about 25 BJS court disposition codes.
Software to convert state rap sheet data into research databases with common codes Benefits of conversion software - Recoding enables BJS to analyze data across states. Software requires less hand-coding each time rap sheets are processed for a research study. New coding rules are added to the existing crosswalk libraries.
Software to convert state rap sheet data into research databases with common codes Criminal history flat file Standardized criminal history variables on each subject from earliest to most recent arrest Criminal history summary file Tallies an individual s arrests, convictions, and incarcerations in criminal career Includes in-state and out-of-state activity prior to and following a target date (i.e., date of prison release) Files can be analyzed with basic statistical software such as SPSS or SAS
New BJS Recidivism Analysis Tool BJS will launch a new data analysis tool for the 2005 prisoner recidivism study on its website Users can conduct their own analyses on former prisoners by specific demographic, criminal history, and prison sentence attributes Graphs to show the monthly and annual recidivism rates following release by different offender characteristics The tool will also provide standard errors/confidence intervals associated with the recidivism estimates
Recidivism of Prisoners Released in 2005 Several Post-Prison Recidivism Measures Arrest Adjudication (dismissed, acquitted, convicted) Conviction Incarceration sentence (prison or jail) Prison sentence Return to prison (technical violation or new court commitment) Return to prison based on both NCRP admission records and criminal history data.
Contact information Matt Durose matt.durose@usdoj.gov Bureau of Justice Statistics www.bjs.gov (202) 307-0765