Lowering Recidivism With Performance Measurements/EBP and Predictive Technology Presented by: Charlie Granville CEO, Capita Technologies Chris Baird Former Director of Research, National Council on Crime and Delinquency
Charlie Granville Chris Baird CEO, Capita Technologies 15 years experience designing and building criminal justice systems that enable and support Evidence-based Practices Expert on predictive analytics Chris Baird Former Director of Research, National Council on Crime and Delinquency Created the National Institute of Corrections Model Probation and Parole program
Why EBP & Predictive Methodologies A well-designed, automated system provides an opportunity to significantly advance the Corrections field.
The Impact of Evidence-Based Practices An organization with the right knowledge, experience and Evidence-Based Practices can provide services that have a positive impact on changing human behavior resulting in lower violations and recidivism rates.
Missing From Many Systems Method(s) to determine Best Practices Tracking to measure use Ability to share and utilize Evidence
Recidivism Rates Are Through the Roof Source: Bureau of Justice Statistics: Recidivism of Prisoners Released in 30 States in 2005: Patterns from 2005 to 2010 http://www.bjs.gov/index.cfm?ty=pbdetail&iid=4986
Reducing Recidivism by Only 10% Can Save Millions – Even Billions $635 Million First Year Savings in Corrections Alone - State of Recidivism 2011, Pew Research Does not include additional savings from: Direct Costs – Police, Sherriff, Court Costs Indirect Costs – Lost Productivity, Victim Costs, other Social Costs Total savings reaches $10 Billion over first 5 years
States Have Shown Recidivism Can Be Reduced -9.1% -9.4% -5.8% -8.9% -7.1% -19.3% -17.9% -10.0% Council of State Governments Justice Center, Reducing Recidivism: States Deliver Results (New York: Council of State Governments Justice Center, 2014)
What Did They Do? Tailored approaches to individual needs Provided continuity of care from incarceration to community Expanded use of risk assessments Data collection and performance measurement Fidelity to evidence-based practices Council of State Governments Justice Center, Reducing Recidivism: States Deliver Results (New York: Council of State Governments Justice Center, 2014)
Today, We Aren’t Tracking What Works “For most correctional programs, the only evidence available is evidence that they exist.” - Dennis Wagner, NCCD Director of Research Does the program exist? ? Did the program positively impact offenders? ? Was an individual case plan successful? ? Did different officers work better with different offenders?
Evidence Based Practices Make The Difference 2. Analyze & Adjust 3. Put Into Practice 1. Collect Evidence “We think one of the most important parts of [our state’s reforms] is data collection and evidence-based practices, essentially making sure we’re spending money where results are predictable and the best results will be achieved.” -Georgia Governor Nathan Deal
Data is King: Collect All The Evidence Risk and Need Assessments Case Plans Offender Demographics Location of Home Classifications Family History Education Level Institutional Behavior Employment Contacts Offense Information Sanctions and Violations Criminal History Program Costs Programs and Interventions
Key Factors in Data Collection Quantity Quality Structured Data
Bottom Line “If you can't measure it, you can't improve it.” - Peter Drucker
Step 2: Analyze and Adjust Analyzing and Adjusting Requires The Collected Evidence Functional Domain Expertise Computerized System to Manage Data, and Analyze Processes and Store Best Practices
Step 2: Identify Group Characteristics Gender = Offender Data EBP Group Library Age= x-y Ethnicity = Risk Scores = Social Attitude Family Mental/Medical Education/skills Dependencies Influencers Aggression Repeat offender Crime Type = Violent, nonviolent Class type Modeling Analysis Other… Industry recidivism knowledge Performance Group 1 Positive Behavior
Step 2: Analyze and Adjust What to Analyze Contact Compliance Provider Performance: Completion %, Recidivism Rate, Duration Officer Performance Rewards, Sanctions, Violations Recidivism Rate
Step 2: Analyze and Adjust What to Adjust Contact Frequency Sanctions and Violations Programs and Providers Caseloads Measurements
Step 3: Put Into Practice Retrain Update computerized systems to reflect new knowledge Reallocate resources Modify EBP documentation
The Next Step in the Puzzle: Predictive Analytics
What is “Predictive Analytics”? “Predictive analytics uses algorithms to find patterns in data that might predict similar outcomes in the future.” - Forrester Research Enabled by Big Data (Requires Lots of Evidence) Dynamic (Changes Based on Historical Evidence) Requires Technology
What Agencies Will Be Able to Do Uncover Patterns That Were Previously Hidden Examples: Identify when someone is likely to re-offend Improve program selections for individuals Identify if a program is working for specific offender Improve officer assignment Improve supervision policies and procedures Improve use of funds
Predictive Analytics Requirements A Strong Model Known Outcome
Building Robust Models Know historical data Know historical outcomes associated with that data Build a model Use it to predict future events based on new data
Dynamic Decision Making Dynamically create and provide guidance and creation for your most important planning and supervision decisions.
Preparing Your Agency Collect Structured Data and Outcomes Identify Outcomes You Want to Predict Benchmark Where You Are Today
Closing Thoughts Statistics are only as good as their source and collection process Large data samples are critical for continuous measurements to dynamically improve Best Practices Utilizing Best Practices is the missing piece of the puzzle With the right approach to information and automation, it can now be done routinely
Questions