pretrial-summit-risk-assessment-tool-evaluation

Indiana Office of Court Services Pretrial Pilot Study ‐ October 4, 2019Indiana Pretrial Pilot Project • In 2014, the Indiana Supreme Court authorized the development of a pretrial releas

Trang 1

Indiana Office of Court Services Pretrial Pilot Study ‐ October 4, 2019

Indiana Office of Court Services

Pretrial Pilot Study

Brad Ray, Ph.D.

Director of the Center for Behavioral Health and Justice

Wayne State University

Spencer G. Lawson, M.S.

Michigan State University

Wayne State University Center for Behavioral Health and Justice

Eric Grommon, Ph.D.

Associate Professor O’Neill School of Public and Environmental Affairs Indian University – Purdue University Indianapolis

Indiana Pretrial Pilot Project

The Center for Behavioral Health and Justice is a community‐engaged research group who works directly with agencies to address social problems through data‐driven decision making in the development, implementation, and evaluation of policies and initiatives

Spencer G. Lawson, M.S.

School of Criminal Justice Michigan State University

Evan Lowder, Ph.D.

Assistant Professor Criminology, Law and Society George Mason University

Trang 2

Indiana Pretrial Pilot Project

• In 2014, the Indiana Supreme Court authorized the development of a pretrial release pilot project in collaboration with the Evidence Based Decision Making Initiative

• Criminal Rule 26: “In determining whether an arrestee presents a substantial risk of flight or danger to self or other persons or to the public, the court should utilize the results of an evidence‐based risk assessment…”

• Indiana Criminal Rule 26 was the foundation for the 11 pretrial pilot project counties

• Evaluation efforts started in 2015 and was initially aimed at examining the barriers and facilitators towards implementation of the Indiana Risk Assessment System – Pretrial Assessment Tool (IRAS‐PAT) and has more recently focused on validation

Indiana Pretrial Risk Assessment Tool

• Administered to determine

risk of pretrial failure

• Self‐report and verification

• Low, Moderate, or High

Trang 3

EPIS Framework: IRAS‐PAT

A developmental evaluation, aimed at course corrections toward the utilization and validation of a structured risk assessment tool

Exploration Phase

• Every state has enacted legal reforms governing pretrial release with the goal of exploring alternatives to money bail

• Unstructured professional judgments have repeatedly been shown to be less accurate than evidence‐based risk assessment approaches

• Structured risk assessment tools are informed by more than 65 years of rigorous research into factors associated with public safety risks

Trang 4

Exploration Phase

Trang 5

Preparation Phase

Pretrial Pilot Counties

• Allen

• Bartholomew

• Grant

• Hamilton

• Hendricks

• Jefferson

• Monroe

• Porter

• St. Joseph

• Starke

• Tipton

Facilitators

• Pretrial recommendations helpful at initial hearings

• Established local teams committed to the pretrial risk assessment process

Barriers

• Lack of integration across local data systems

• Concerns about the IRAS‐PAT instrument

Trang 6

Preliminary integration and analysis of linked INCite and Odyssey data: 2014 ‐ 2016

Trang 7

• The IRAS‐PAT was not being used in the decision‐making process

• Counties were using the IRAS‐PAT at varying points in the criminal justice system, with different populations, and different release schedules:

• Allen County had a target population of non‐violent F5/F6 arrestees and used an additional risk tool for pre‐screening

• Bartholomew County had a target population which largely consisted of those arrestees with warrants issues or charges filed

Evaluation plan shifted to validation of the IRAS‐PAT

Trang 8

Implementation Phase

• Allen  Mar. 2016

• Bartholomew  Sept. 2016

• Grant  Aug. 2017

• Hamilton  June 2016

• Hendricks  Jan. 2016

• Jefferson  Oct. 2016

• Monroe  Oct. 2016

• Porter  Mar. 2017

• St. Joseph  Oct. 2016

• Starke  Jan. 2016

• Tipton  Oct. 2016

Validation requires approximately one‐ year of IRAS‐PAT implementation and an additional year for follow‐up

IRAS‐PAT Validation

• Validation is the process of demonstrating the quality of a measure, the scores

obtained with the measure, or the interpretation of those scores

• Example: mental illness

• Kessler (6) screening

instrument

• Distress from serious mental

illness in past 30 days

Trang 9

Kessler (6) Example

18%

27%

0%

10%

20%

30%

2017 2019

County

A

22%

25%

0%

10%

20%

30%

2017 2019

County

B

21% 22% 0%

10%

20%

30%

2017 2019

County

D

18%

16%

0%

10%

20%

30%

2017 2019

County

E

21% 22%

0%

10%

20%

30%

2017 2019

County

J

19%

32%

0%

10%

20%

30%

40%

2017 2019

County

I

22%

19%

0%

10%

20%

30%

2017 2019

County

H

16%

12%

0%

10%

20%

30%

2017 2019

County

G

Data Source: K6 Collection 2017 & 2019

IRAS‐PAT Validation

• The IRAS is a suite of assessments: Community Supervision and Screening Tool (CSST), Community and Screening Tool (CST), Prison Intake Tool (PIT), and Reentry Tool (RT)

The IRAS‐PAT measures the risk for failure‐to‐appear and reoffend while on pre‐trial supervision

Pretrial risk assessment tools are designed to inform not replace the exercise of judicial decision‐making and discretion

No matter how well validated in other jurisdictions, assessments should be subjected to local evaluation

Trang 10

• Record linkage between three systems

• We cannot validate the IRAS‐PAT without local jail data

• We need to know when a person entered the jail, left the jail, and a broad understanding of the arrest charges

• From a research standpoint nothing has proven more challenging than the collection of local jail data

Unique Individual Event INCite

Jail Odyssey

IRAS‐PAT Validation: Monroe County

• Data Sources

• Pretrial records

• Jail bookings

• Odyssey court records

• Matching

• Offense date (O)

• Arrest date (J)

Sample Flow Chart

Trang 11

Distribution of Risk Classifications

Predictive Validity: Any Failure‐to‐Appear

Trang 12

Predictive Validity: Any Arrest

Predictive Validity: Any Misconduct

Trang 13

Predictive Validity

Trang 14

• Preliminary results suggest moderate to strong predictive accuracy of IRAS‐PAT

assessment in Monroe County, Indiana

• IRAS‐PAT assessments predict arrest with more accuracy than FTA

• High proportions of defendants at moderate (33%) and high (52%) risk levels engage

in any pretrial misconduct; should be closer to 18% and 29% respectively based on instrument design

Continue working with counties to obtain local booking data to

provide a local validation of the IRAS‐PAT Need to assure we have adequate sample size for validation Data needed to be cleaned and linked prior to validation

Trang 15

• Allen  Mar. 2016

• Bartholomew  Sept. 2016

• Grant  Aug. 2017

• Hamilton  June 2016

• Hendricks  Jan. 2016

• Jefferson  Oct. 2016

• Monroe  Oct. 2016

• Porter  Mar. 2017

• St. Joseph  Oct. 2016

• Starke  Jan. 2016

• Tipton  Oct. 2016

Validation requires approximately one‐ year of IRAS‐PAT implementation and an additional year for follow‐up

Trang 16

Sustainment Phase

Quality Assurance and Future Research

• Determine validation within county specific practices

• Investigate pretrial supervision on court appearance and public safety outcomes

• Inform supervision practices for optimal outcomes by risk‐level

• Examine whether the implementation of pretrial risk assessments improves the fairness of pretrial release decisions for racial‐ethnic minorities (funded by the National Institute of Justice)

Questions & Discussion

Michigan Mental Health Diversion Council

Định dạng
Số trang	16
Dung lượng	0,99 MB