Você está na página 1de 46

Institute of Criminology University of Cambridge

Institute of Psychology University of Erlangen-Nuremberg

What Works in Reducing Reoffending: A Global Perspective


Friedrich Lsel

First European Seminar of the STARR Project, 27 April 2010, Cambridge, UK

Offender Rehabilitation: Recent History (1)


Strong rehabilitation optimism in the 1960s and early 1970s Nothing works (Martinson, 1974) Lack of methodologically sound research Support for non-empirical arguments (e.g. getting tough on crime, high costs, forced personality change, informalisation of criminal justice, just deserts punishment) Revitalisation of rehabilitation in the 1990s Better theoretical foundation More longitudinal research on desistance

Recent History (2)


More controlled evaluations Systematic research reviews (meta-analyses) More international exchange What works instead of nothing works Programme accreditation & quality management (e.g., Canada, England & Wales, Netherlands, Scotland, Sweden) Large-scale implementation of offending behaviour programmes

England & Wales: National Offender Management Service (NOMS): Pathways to Reducing Re-offending (2006)

Attitudes, Thinking & Behaviour Programmes


Social learning theories Reinforcement & modeling Social information processing & problem solving Social integration & bonding Action theory Theries of moral attitudes & beliefs Neuropsychological theories

Evidence comes by replication


More than 700 more or less well controlled evaluations of correctional treatment worldwide Majority from North America For nearly all programmes: positive vs. zero or sometimes even negative effects Many factors are relevant: Methodology, samples, context, culture, staff motivation, implementation a.s.f. No sound conclusions from one single study Systematic reviews/meta-analyses of good quality primary studies

Study Quality: Maryland Scale


(Sherman et al., 1997, modified)

1. No control/comparison group (CG) 2. Nonequivalent CG (e.g., refusers) 3. CG judged to be equivalent (e.g., comparison of offender characteristics) 4. Systematic strategy applied to achieve equivalence (e.g., matching) 5. Randomized assignment (not impaired by attrition etc.)

Mean Effect Sizes (d) in Meta-Analyses on General Offender Treatment


Tong & Farrington (2007) MacKenzie (2006) Lipsey & Landenberger (2005) Lipton et al. (2002) Lipton et al (2002) Dowden & Andrews (2000) Redondo et al. (1999) Pearson et al. (1998) Lipsey & Wilson (1998) Lipsey (1992) Andrews et al. (1990) Gottschalk et al. (1987) Lsel et al. (1987)

In total > 500 studies

0,00

0,05

0,10

0,15

0,20

0,25

0,30

Mean Effect Sizes of Offender Rehabilitation (1)


General and violent offenders Typical range: d = 0.10 - 0.30 Example: d = 0.20 50% recidivism in CG vs. 40% in TG 10 percentage points or 18% reduction Small but highly significant

Mean Effect Sizes (d) in Meta-Analyses on Juvenile Offender Treatment


Latimer et al. (2003) Cleland et al. (1997) Redondo et al. (1999) Petrosino (1997) Lipsey & Wilson (1998) Lipsey (1992) Andrews et al. (1990) Whitehead & Lab (1989) Gottschalk et al. (1987) Garrett (1985)

0,00

0,10

0,20

0,30

0,40

Mean Effect Sizes (d) in Meta-Analyses on Sexual Offender Treatment


d

0,6 0,5 0,4 0,3 0,2 0,1 0


) ) ) ) ) ) ) ) ) ) ) 95 99 99 00 00 02 89 06 06 05 04 19 19 19 20 20 20 19 20 20 20 20 l( .( r( l( .( .( .( .( ll ( r( .( l l l l l l l e e a e ta ne H ta ta ta ta ta s nd ck e e e e e L u ie bo xa m er er os by on ar zz le r h k li s A C A gh al Sc Fu an & Po lla W H & a el G el tz s ei R L

Meta-Analysis of High-quality Studies on Sex Offender Treatment


Schmucker & Lsel (2010): Update of Lsel & Schmucker (2005) Only high-quality studies Level 3 of Maryland Scale Treatment & Control Group equivalent 26 eligible studies, 28 independent comparisons between TG and CG N = 8,347 offenders (3,919 treated) Mainly Cognitive-behavioural programmes (n = 20)

Recidivism Rates
(Length of follow up: M = 5.7 years)
% 50,0
40,0 30,0 Treatment group Control group

26% reduction
32,4 24,0

27% reduction
20,0
12,4 9,0

10,0 0,0

sexual

all offences

Type of reoffending

Some Outcome Moderators


Larger effects in smaller samples high risk offenders group programmes with some individual sessions better descriptive validity

Mean Effect Sizes of Offender Rehabilitation (2)


In total ca. 10% - 30% reduction in reoffending Statistically and practically significant Benefit-cost ratios: 1 7 payoff for each invested Lifetime costs for 1 persistent offender ca. 1 M Effect sizes in medicine: Radiation + chemotherapy of brain tumors: ca. 10% Aspirin therapy of cardiovascular events: ca. 13% Passive smoking and lung cancer: ca. 22% Drug treatment of depression: ca. 56%

Variation in Outcome Between Offender Rehabilitation Programmes


Previous figures: overall means Differences between various programmes & even between studies on one and the same programme No single magic bullet or gold standard programme Programme content is only one reason for outcome differences Many other factors play a role They explain more variance than programme content What works question often too simple

What works: Relatively consistent positive effects


Basic education (e.g. literacy, maths) Vocational & employability programmes (if useful in life) Cognitive-behavioural programmes (CBT; e.g., Reasoning & Rehabilitation, Cogn. Restructuring, Moral Reconation Therapy, Anger Management) Most important elements in CBT: Anger management & interpersonal problem solving Structured therapeutic communities (TCs), milieu therapy & Social-therapeutic prisons (Germany)

What works: Relatively consistent positive effects


Multisystemic Therapy (MST) and other multimodal family-oriented programmes for young offenders Restorative Justice (e.g., offender-victim mediation, restitution; not for all groups) Drug courts sychopharmacological/substitution drug treatment

Promising: Not yet well-replicated and/or rather small effects


Intensive supervision (probation, parole) Electronic monitoring (problem of IS & EM: revocation, violation of orders) Counseling (juveniles & adults) Mentoring programmes Challenge programmes Social case work Pharmacological treatment for aggressiveimpulsive offenders (SRIs) and sexual offenders (hormonal treatment)

What doesnt work: Mean zero or even negative effetcs


Pure sanction (e.g. custodial vs. non-custodial sentences, longer prison sentences, regular sanction vs. diversion) Purely deterrent measures (e.g. Scared Straight, shock incarceration for juveniles) Strict discipline (e.g. Boot Camps, when without treatment elements) Purely psychodynamic and unstructured therapeutic approaches Inappropriate programmes may even harm

The Good Lives Model (Ward, 2002): Nine Goods


Life (including healthy living and functioning) Knowledge Excellence in work and play Excellence in agency (autonomy, self-directedness) Inner peace (freedom from emotional turmoil, stress) Friendship and community (intimate partner, family relations) Spirituality (finding meaning and purpose in life) Happiness Creativity

Evaluation of the Good Lives Model


Increasingly popular in practice Strength-oriented (as is MST) Relations to humanistic psychotherapy, positive psychology & resilience research Some indirect support from desistance research But: Not yet replicated empirical evidence in evaluations Conclusion: Perhaps promising

Principles of Appropriate Offender Treatment


Risk principle (Adequate intensity) Need principle (Address criminogenic needs) Responsivity principle (Adequate modes of learning and teaching) Up to 60% reduction of recidivism when all three R-N-R principles realized Smaller effect when 2 or 1 principle present; slightly negative effect when no R-N-R principle addressed Problems of R-N-R approach: Some circularity re. responsivity; risk not individualized Andrews et al. (1990)

Transfer of What Works Principles into Policy & Practice


Best example: England & Wales Ambitious political aims in the 1990s and 2000s: 5% & 10% reduction of reoffending Systematic programme development, accreditation & implementation Intensive investment of Home Office/Ministry of Justice Coordinated action: NOMS, HMPS, NPD, CSAP More than 40 accreditated programmes Quality management (assessment, staff training & supervision, monitoring & audit, outcome evaluation)

Criteria of the Correctional Services Accreditation Panel (CSAP) in E & W


1. Clear model of change (evidence based) 2. Selection of offenders (e.g., adequate assessment, risk level) 3. Targeting a range of dynamic risk factors 4. Effective methods (i.e., cognitivebehavioural) 5. Skills orientation (concrete behaviour)

CSAP Accreditation Criteria


(Continued) 6. Sequencing, intensity and duration
7. Engagement and motivation 8. Continuity of programme and services (aftercare; community services) 9. Maintaining integrity (context; selection, training, and supervision of personnel) 10. Ongoing evaluation
Based on principles of effective intervention: Andrews (1995), Lsel (1995)

Accredited Programmes in E & W (1)



Fast & widespread programme implementation More than 30,000 offenders p.a. on programmes Some problems of integrity & continuity Sometimes violation of eligibility critera Non-starters and dropout problems in the community Not much well-controlled evaluation; often weak designs (e.g. non-equivalent CG, short follow-up) Only small correlations between intermediate factors (e.g. psychometric) and recidivism

Accredited Programmes in E & W (2)


Some encouraging results (Harper & Chitti,2005): Well-implemented cognitive-behavioural programmes for medium to high risk offenders Aggression replacement training in probation Sex offender treatment programmes Drug treatment programmes Motoring offence programmes Programmes including family relations in treatment and aftercare However, methodologically weaker data suggest some success of the treatment policy

Harper & Chitti (2005)

Adult Re-offending Rates in E & W


Percentage of offenders committing at least one offence within one year of discharge from prison or commencement of probation supervision
50 45 40 35 30 25 20 2000 2002 2003 2004 2005 2006
Note. No valid data available for 2001

2000-06: - 11%

MoJ Evidence Digest (2009)

Adult Re-offending Rates in E & W


Number of offences per 100 offenders within one year of discharge from prison or commencement of a court order under probation supervision
230 210 190 170 150 130 110 90 70 50 2000 2002 2003 2004 2005 2006
Note. No valid data available for 2001

2000-06: - 23%

MoJ Evidence Digest (2009)

Re-Offending Rates of Prisoners by Sentence Length (HMPS 2007)


80 70 60 50 40 30 20 < 1 year 1&<2 years 2&<4 years 4 years 2000 2002 2003 2004

Diff. actual/ predicted

- .02%

- 6.3%

- 10.1%

- 13.4%

Total: - 3.1%

Re-offending rates for offending behaviour programmes


Predicted rates 75 70 65 60 % 55 50 45 40 35 30 Non-starters
Diff = 7.3%

Actual rates

Starters / Dropouts
Diff = 4.8%

Completers
Diff = 26.4%

Interim Accredited Programmes Software: 2004 Community Cohort (Hollis, RDS NOMS, 2007) Reduction in Re-Offending Using Predicted Data
Total (n = 25,255) Drink impaired drivers (1,148) General offend. behav. (12,924) Domestic violence (1,148) Sex offender progr. (741) Substance misuse (5,081) Anger management (1,148) Women's programme (62) 0 5 10 15 20 25 30 35 40

* * * * * *

Conclusions for international practice


In total positive results of correctional programmes However, most what works research from North America Experience in Europe: Local adaptation and evaluation necessary Very few controlled studies in non-Englishspeaking countries Problems of transfer from/to other countries Remember: Many factors determine outcome

Factors that have an impact on outome in meta-analyses: Examples (1)


Programme content/type (++) Content of the control condition (+) Quality of evaluation design (?) Length of follow up (-) Hard/official outcome measures (-) Small vs. large samples (++) High vs. low risk offenders (++) Juvenile vs. adult offenders (+)

Factors that have an impact beyond programme content: Examples (2)


Many dropouts, counted as treated (--) Voluntary vs. mandatory treatment (?) Integrity of programme implementation (+) Staff training & supervision (+) Staff-prisoner relations/social climate (+) Demonstration project vs. routine practice (+) Researchers involved in programme development/delivery (+) Community vs. custody (+)

Problems of transnational transfer


Different cultural traditions Different justice systems Different age ranges for juveniles Different minority problems Different organizational structures of prison & probation Different staff roles

Problems of transnational transfer


Different experiences in evaluation Differences in financial resources Different weighting of various penal aims: compensation of guilt, protection of the public, general deterrence, rehabilitation, restitution Differences in crime policy Differences in incarceration rates

Incarceration Rates

Incarceration Rates in Europe


Per 100,000 population in 200-08 ( Walmsley, 2009)
Russia Belarus Georgia Ukraine Latvia Estonia Lithuania Poland Czech Republic Spain England & Wales Scotland Hungary Slovakia Bulgaria Romania Serbia Greece 629 468 415 323 288 259 234 221 182 160 153 152 149 148 134 124 122 109

Incarceration Rates in Europe


Per 100,000 population in 200-08 ( Walmsley, 2009)
Netherlands France Austria Croatia Italy Belgium Germany Northern Ireland Switzerland Sweden Norway Slovenia Denmark Finland Iceland Liechtenstein 100 96 95 93 92 93 89 88 76 74 69 65 63 64 44 22

: England & Wales

Incarceration rates
Since the 1990s increase in many countries Not clearly related to crime rates Political decisions Recently decrease in several countries Incarceration very costly Lack of resources for effective treatment and prevention? (e.g. California) Reduction by effective offender treatment & less short-term imprisonment The chance of older populations

Conclusions and Perspectives (1)


Global findings on rehabilitation programmes positive Continuity & stepwise improvement instead of pendulum movements in policy Cultural adaptation & differentiation Quality management, process & outcome evaluation Centers of excellence, clearing houses Knowledge transfer and management of innovation

Conclusions & Perspectives (2)


Integrating correctional programmes with the whole range of services (no silo approach) Linkage to routine case work Relations to natural processes of desistance Relations to developmental prevention Crime not only a criminal justice issue (e.g. public health, education, welfare & economy)

Contact: starr@crim.cam.ac.uk fal23@cam.ac.uk

Você também pode gostar