kirk r. williams, ph.d. professor of sociology and co-director

40
Findings from a Dual Generic and Specific Risk Assessment Process for Domestic Violence Perpetrators in Connecticut Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director Presley Center for Crime and Justice Studies University of California, Riverside Riverside, CA 92521 (951) 827-4604 [email protected]

Upload: camdyn

Post on 16-Jan-2016

23 views

Category:

Documents


0 download

DESCRIPTION

Findings from a Dual Generic and Specific Risk Assessment Process for Domestic Violence Perpetrators in Connecticut. Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director Presley Center for Crime and Justice Studies University of California, Riverside Riverside, CA 92521 - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Findings from a Dual Generic and Specific Risk Assessment Process for Domestic Violence

Perpetrators in Connecticut

Kirk R. Williams, Ph.D.

Professor of Sociology and Co-Director

Presley Center for Crime and Justice Studies

University of California, Riverside

Riverside, CA 92521

(951) 827-4604

[email protected]

Page 2: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

A Brief History of the CT Domestic Violence Risk Assessment Project

Page 3: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

• Intake interviews and risk assessments are conducted by family relations counselors with all domestic violence perpetrators after arrest

• Results are used used to make two recommendations at arraignment, held approximately 24 hours after arrest

Page 4: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Placement

• Referral to family services for pre-trial supervision

• Referral to State’s Attorney for prosecution, nolle (no prosecution but case stays open for 13 months. If violations occur, the case can be reopened, with prosecution of old and new offenses), continued monitoring to ensure compliance with court orders, or dismissal

Page 5: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Protective Orders:

• No protective order

• Partial or limited protective order

• Residential stay away protective order

• Full no contact protective order

Page 6: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

• Family Services wanted an instrument to assess the risk of repeat offending that could be administered quickly, given the short time constraints and conditions under which assessments are done

Page 7: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

• The original Domestic Violence Screening Instrument (DVSI) met their criteria

• It was developed and validated in Colorado (Law and Society Review, 2004, Vol. 28: 437-455)

Page 8: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

It was adopted in May, 2002 as the risk assessment instrument to be used in CT

Page 9: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

After analyzing repeated pilot samples and incorporating input from Family Relations Counselors (FRCs) during 2003, the instrument was revised (now the DVSI-R) and implemented statewide in 2004

Page 10: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

The DVSI-R includes 11 items, seven of which address the behavioral history of offenders, and the other four items pertaining to substance abuse, objects used as weapons, or children present during prior or current violent incidents, and employment status

Page 11: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

It also includes summary ratings (low, medium, high) of imminent risk of future violence to the victim or some other person known to the victim

Page 12: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Scoring categories range from zero (no evidence) to two or three depending on the item

Assessors score each item after reviewing the available sources (police report, defendant interview, registry, record checks, victim input if available)

Page 13: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Items are automatically (computerized system) summed to derive total risk scores, ranging from zero to 28.

Page 14: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Determining the Predictive Validity of the DVSI-R

• An initial study was done with 14,970 assessments between September 1, 2004 and May 2, 2005

• The results showed that higher scores on the DVSI-R were significantly related with a greater likelihood of recidivism during this eight month period (see Public Health Reports, 2006, Vol. 121: 400-408.)

Page 15: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Conducted two 18-Month Recidivism Studies

Sample 1 included assessments during October-November, 2005 (N = 3,796)

Sample 2 included assessments during February-March, 2007 (N = 3,569)

Page 16: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

DVSI-R Total Risk Scores Were Associated With (AUC in parentheses):

• New family violence offenses (Sample 1 .680; sample 2 .690)

• Violations of restraining and protective orders

(Sample 1 = .733; sample 2 = .751)

• Multiple arrests during the follow-up period

(Sample 1 = .738; sample 2 = .738)

• Multiple victims and victimizations

(Sample 1 = .746; sample 2 = .738)

Page 17: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

In short, the higher the DVSI-R risk score, the greater the probability of family violence recidivism, the potential seriousness of the offending, and non-compliance with court orders

Page 18: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Finding the Limitation of Generic Risk Assessment for Domestic Violence

Offenders

Both samples included sub-sets of cases that had previously been on probation and assessed with a generic risk assessment instrument:

The Level of Service Inventory Revised (LSI-R)

Page 19: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

• The LSI-R is a 54 item risk-needs assessment instrument administered to all convicted offenders placed on probation in Connecticut for any criminal offense

• The 54 items are distributed across ten domains

Page 20: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

• Criminal history (10 items)• Education/Employment (10 items)• Financial problems (2 items)• Family/Marital (4 items)• Accommodation - living arrangements (3

items)• Leisure/Recreation (2 items)• Nature of companions (5 items)• Substance abuse problems (9 items)• Emotional problems (5 items)• Attitude/Orientation (4 items)

Page 21: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

• Items are scored, and scores are automatically (computerized system) summed for each domain for a “needs” sub-score and across domains for a total “risk” score

• Total risk score is used to determine the level of supervision, and domain sub-scores are used to link needs to services for case management

Page 22: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

The empirical relation between between the LSI-R and the DVSI-R

These two sets of total risk scores are not highly correlated

• Sample 1 r = .172 (N = 920)

• Sample 2 r = .150 (N = 808)

Page 23: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

920 of the 3,796 cases in the 2005 statewide DVSI-R validation study (sample 1) had previously received the LSI-R

56.7% (N = 152) of those scoring in the lowest quartile of the LSI-R (N = 268) scored in the top half of the DVSI-R distribution

Page 24: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

808 of the 3,569 cases in the 2007 statewide DVSI-R validation study (sample 2) had previously received the LSI-R

70.6% (N = 149) of those scoring in the lowest quartile of the LSI-R (N = 211) scored in the top half of the DVSI-R distribution

Page 25: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

The cases scoring low on the LSI-R but high on the DVSI-R “fall through the cracks” in terms of calculated levels of supervision (administrative monitoring or low)

They are “false negatives” in the assessment of risk when the LSI-R is use alone, posing public safety (in the home) and criminal justice liability problems

Page 26: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Connecticut initiated a pilot study to determine if including the DVSI-R in the post-conviction risk and needs assessment process would identify high risk DV cases and elevate them to a higher level of supervision

Page 27: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Collected data for a “test” sample of 220 cases between August 20, 2007 and February 20, 2008 in two pilot sites

Page 28: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Collected a comparison sample of 260 cases six-months prior to the “test” period

Page 29: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Key Findings

25.4

35.0

010

2030

40

comparison group test group

% o

f Cas

es

Upg

rad

ed t

o A

ny

Hig

her

Lev

el

Calculated to Assigned Levels of Supervision by Comparison and Test Sample

Page 30: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

020

4060

80

0 .5 1 1.5 2 0 .5 1 1.5 2

comparison group test group%

of C

ase

s U

pgra

ded

One

or

Tw

o L

eve

ls

upgraded_levelsCalculated to Assigned Levels of Supervision by Comparison and Test Samples

Page 31: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

23.4

9.2

05

1015

2025

no upgrade upgraded%

Re-

Arr

est

ed

One

or

Mo

re T

imes

Domestic Violence Re-Arrests by Any Upgrades

Page 32: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Conclusions

Including the DVSI-R in the risk assessment process was associated with significant upgrading from calculated to assigned levels levels of supervision

Page 33: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Overriding DV cases from lower to higher levels of supervision was associated with a lower likelihood of domestic violence recidivism six months after assessment

Page 34: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Connecticut launched a statewide dual assessment process last summer for all convicted domestic violence offenders placed on probation

Page 35: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Preliminary evidence from 1,267 recent cases in 2009 suggest upgrading levels of supervision is occurring:

– 46.9% of the 207 cases with low calculated levels of supervision were assigned medium levels

– 43.5% of those 207 cases were assigned high levels of supervision, and

– 46.5% of 340 cases with medium calculated levels of supervision were assigned high levels of supervision

Page 36: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

The dual assessment process will be followed through next year to determine:

– whether this pattern of increasing levels of supervision is continuing, and

– whether it is associated with domestic violence recidivism reduction

Page 37: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Thank You!

Page 38: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director

Setting Sun Over the Pacific

Page 39: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director
Page 40: Kirk R. Williams, Ph.D. Professor of Sociology and Co-Director