not protectively marked integrated offender management: a multiagency desistance programme c/supt...
TRANSCRIPT
NOT PROTECTIVELY MARKED
Integrated Offender Management:
A Multiagency Desistance Programme
C/Supt Andy Williams MSt Barak Ariel PhD
NOT PROTECTIVELY MARKED
• Evidence Based Practice – necessity• Theory• History & Evaluation• Bristol IOM• Does it work?
To Cover
EBP – A Necessity?(timing has never been better/worse!)
NOT PROTECTIVELY MARKED
Theory
• Life Course Desistance (Glueck & Glueck, 1930; Sampson & Laub, 1993; West & Farrington, 1977)
• Defiance, Deterrence and Irrelevance (Sherman, 1993)
• (Offender Desistance Policing (Sherman and Neyroud, 2012)
NOT PROTECTIVELY MARKED
Research
• Desistance Studies (Liverpool & Sheffield)– Re-integration– Procedural Fairness– Compliance
• Historical research of Intensive Supervision
NOT PROTECTIVELY MARKED
NOT PROTECTIVELY MARKED
NOT PROTECTIVELY MARKED
Research of IOM
• Home Office policy launch 2009 p.9 “No specific impact evaluation of IOM”
• London Diamond Project (Dawson et al, 2011)
• Process Evaluation of 5 Pilots (Senior et al, 2011)
Limited quantitative data / Limited evidence
NOT PROTECTIVELY MARKED
What is Bristol IOM?
NOT PROTECTIVELY MARKED
Carrot and Stick + No Choice
NOT PROTECTIVELY MARKED
IOM
• Started in 2009
• Co-located multiagency – Police, Prisons, Probation & CJIT
• Selection / Supervision / Surveillance
• No Choice (expect defiance?)
• Fast-track Pathways treatment
NOT PROTECTIVELY MARKED
Pathways – Criminogenic Need
• Accommodation• Alcohol• Attitudes, Thinking
and Behaviour• Children, Family of
Offenders
• Drugs• Employment, Training
and Education• Finance, Budgets and
Debt• Mental and Physical
Health
NOT PROTECTIVELY MARKED
RESEARCH – DOES IT WORK?
NOT PROTECTIVELY MARKED
Research Questions
1. Are post-IOM arrest reductions associated with IOM?
2. Any reductions in seriousness of arrest?
3. Any differences between those with and without pathways?
NOT PROTECTIVELY MARKED
Methodology• Data: cohort of 155 Bristol’s 2009-2010 most prolific offenders• Coded pathways – Inter-rater reliability test (Cronbach
alpha .97)• All offenders entered programme after being assessed as
serious/prolific• 111 offenders under police supervision + pathway treatments • 39 offenders under police supervision + no pathway• Poisson Generalized Linear Model
NOT PROTECTIVELY MARKED
Findings
NOT PROTECTIVELY MARKED
Baseline Data
n 155Treatment year 2010Age 31.7 (8.49)Ethnicity (% none-White) 15.5%Gender (% male) 92.9%Mean number of arrests 6-months pre IOM 1.86 (2.5)***
Mean number of days in Prison pre IOM 42.9 (93.5)73% with 0 days
OASYS Risk Level pre-treatment (scale 1-3)2.64 (0.5) (sys mis 59%)
NOT PROTECTIVELY MARKED
Statistically significant differences between treated and untreated groups (t-tests or chi-square tests, depending on data distribution) - * p < .05; ** p < .01; *** p < .001
Age Curve
NOT PROTECTIVELY MARKED
Contact / Pathway Treatment Group
NOT PROTECTIVELY MARKED
Number of Contacts During IOM (treatment group) 44.6 (SD=46.5)
No Pathway No Contacts 28%Single Pathway (within treatment group) 12%Multiple Pathways (within treatment group) 88%Pathways:
Drugs 37%Accommodation 27%Employment Training & Education 25%Attitudes Thinking & Behaviour 12%Mental & Physical Health 12%Alcohol 8%Finance Benefit & Debt 4%Children & Family of Offenders 3%
Findings – Arrest Rates
NOT PROTECTIVELY MARKED
Seriousness of Arrest
NOT PROTECTIVELY MARKED
NOT PROTECTIVELY MARKED
Bristol IOM 2010 Programme
Poisson Generalised Linear Model Parameter Estimates
Parameter B S.E.
95% Wald Confidence Interval
Lower UpperWald Chi-
SquareTreatment .877*** .2713 .345 1.409 10.448Ethnicity -.029 .2187 -.458 .400 .018Age -.012 .0113 -.034 .010 1.114Gender -.542 .4612 -1.446 .362 1.382Arrest (pre-IOM) .187*** .0390 .111 .264 23.029Z Days on IOM -.272* .1313 -.529 -.014 4.286Pathways:
Accommodation .768** .3143 .152 1.384 5.969Drugs -.488* .2889 -1.054 .078 2.855Mental & Physical Health -1.161** .5024 -2.145 -.176 5.338Finance Benefit & Debt -.245 .5206 -1.265 .776 .221Alcohol -.214 .3716 -.942 .515 .331Attitudes Thinking & Behaviour
-.011 .4004 -.796 .774 .001
Employment & Education .343 .3054 -.256 .941 1.259Children & Family of Offenders
.831 .5444 -.236 1.898 2.331
Bristol IOM 2010 Programme
Estimated Marginal Means (Post-Treatment Arrests)^
MeanStd. Error
95% Wald Confidence Interval
Lower Upper
IOM Untreated Group 1.28 .263 .76 1.79
IOM Treated Group .53 .079 .38 .69
^ Covariates appearing in the model are fixed at the following values: Age=31.99; percent minority =27%; percent female = 6%; Pre-IOM arrest mean=1.79; Days on IOM (Z-Score) =.0149
Limitations & Caution
• No causal inference (RCT next step?)
• Regression to the mean
• Selection bias
• 6 month period
• Prison
NOT PROTECTIVELY MARKED
Key Findings
• IOM participants have more than twice the likelihood to desist from crime
• The effect of IOM appears stronger than any other predictor, including offender’s age, gender and criminal record
• IOM Treatment - 78% before-after reduction in arrests
• No IOM treatment before-after 197% increase in arrests
• IOM treatment 67% reduction in seriousness of arrest
• No IOM treatment 15% increase in seriousness of arrest
• Most ‘effective’ pathway = Accommodation
NOT PROTECTIVELY MARKED
Next Steps?• Role of Police in achieving accelerated
Desistance?• Offender Desistance Policing (Sword of
Damocles, Sherman and Neyroud, 2012) – great opportunity
• Police / Offender relationship – ‘rich potential’?
• RCT for IOM?
NOT PROTECTIVELY MARKED
NOT PROTECTIVELY MARKED
C/Supt Andy Williams MSt
07768 327764