Saving Children from a Life of Crime

David P. Farrington

Cambridge University

July 20, 2017

Carleton University

Criminal justice prevention (deterrence, incapacitation, rehabilitation)

Community prevention (targeting community risk factors and social conditions such as collective efficacy, cohesiveness or disorganization, or using the community as the context)

Situational prevention (reducing opportunities in the physical environment)

Developmental/risk-focused prevention (targeting early risk and protective factors)

Crime Reduction/Prevention Strategies

Identify key risk factors for offending and implement prevention techniques designed to counteract them

Identify key protective factors and implement techniques designed to enhance them

Public health method. For example: Key risk factors for coronary heart disease include

smoking, a fatty diet, lack of exercise

Therefore, encourage people to stop smoking, eat more healthily, take more exercise

Risk-Focused Prevention

Saving Children Froma Life of Crime

Book by David P. Farrington and Brandon C. Welsh (Oxford University Press, 2007)

Later article: Farrington, D. P. and Welsh, B. C. (2014) Saving

children from a life of crime: The benefits greatly outweigh the costs! International Annals of Criminology, 52, 67-92.

Key conclusions:

Crime can be reduced by intervening early in life to tackle key risk factors

Key individual, family, peer, school, and community risk factors are reviewed

Effective individual, family, peer, school, and community interventions are reviewed

There is a need to establish a national strategy or national agency for early prevention in all countries

Saving Children

1. Risk factors

2. Effective programs

3. Benefit: cost ratios of interventions

4. Risk-focussed prevention strategy

5. Conclusions

Outline of Lecture

1. Risk Factors

A factor that predicts a high probability of offending (e.g. poor parental supervision)

Since the definition depends on prediction, longitudinal data are needed to study risk factors

Focus on changeable risk factors

Establish risk factors in key longitudinal studies and systematic reviews of findings in longitudinal studies

How replicable are risk factors over time and place? comparison of risk factors in London and Pittsburgh by Farrington & Loeber (1999)

What is a Risk Factor?

Important Risk Factors for Delinquency

Individual: impulsiveness/hyperactivity/risk-taking, low school attainment, low empathy

Family: poor supervision, harsh/erratic discipline, cold/rejecting attitude, low parental involvement, child abuse/neglect, broken families, criminal parents, young parents

Peer: delinquent siblings, delinquent friends

School: high delinquency rate school

Socio-economic: low income, poor housing

Community: high crime neighbourhood

See Farrington, D. P. (2015) The developmental evidence base: Psychosocial research. In Crighton, D. A. and Towl, G.J. (Eds.) Forensic Psychology (2nd ed.). Chichester: Wiley (pp. 161-181).

Similarity of risk factors in two generations

No. Risk factorG2 offending

(OR)G3 offending

(OR)

1 Convicted father at 32 4.38 2.86

2 Convicted mother at 32 3.21 3.23

3 Authoritarian father 1.41 0.91

4 Authoritarian mother 1.39 0.91

5 Young father 1.12 1.01

6 Young mother 1.41 1.30

7 Nervous/Depressed father 1.42 0.82

8 Nervous/Depressed mother 1.58 1.63

9 Uninvolved father 1.44 0.99

10 Harsh discipline/Physical punishment 1.83 2.01

No. Risk factorG2 offending

(OR)G3 offending

(OR)

11 Poor supervision 2.64 2.78

12 Parental conflict 2.24 1.05

13 Disrupted family/Separated from child 2.56 1.88

14 Low family income/Low take home pay 2.37 3.20

15 Large family size 2.48 2.23

16 Poor housing 2.12 3.23

17 Low social class 1.43 2.10

18 Low attainment/Early school leaving 2.77 4.83

19 High daring/Risk taking under 12 3.26 3.33

20 High troublesomeness/Suspended from school 4.17 4.61

Risk factors are not necessarily causes

It would be most efficient to target causes in intervention research

X causes offending if changes in X are followed by changes in offending

Establish causes either in experiments or by studying changes within individuals in longitudinal studies

Farrington et al (2002): in Pittsburgh Youth Study, changes within individuals in parental supervision, but not in peer delinquency, were followed by changes within individuals in delinquency

Need more research on within-individual change

What are the Causes of Offending?

2. Effective Programs

Home visiting (Olds)

Pre-school (Schweinhart)

Parent training (Sanders)

Skills training (Augimeri)

School-based (Salmivalli)

Home/community programs with older children (Alexander, Chamberlain, Ross)

Multi-systemic therapy (MST) (Borduin)

Focus on results of some key experiments, especially those with long-term follow-ups

See Farrington, D. P. (2015) The developmental evidence base: Prevention. In Crighton, D. A. and Towl, G.J. (Eds.) Forensic Psychology (2nd ed.). Chichester: Wiley (pp. 141-159).

Effective Programs

Nurse Family Partnership program

400 mothers randomly assigned to:

home visits from nurses during pregnancy

home visits in pregnancy and infancy

control: no home visits

Visits every two weeks: nurses give advice about child-rearing, nutrition, infant development

Find (15 year follow-up): experimental children had half as many arrests. Biggest effect and benefits > costs for lower class unmarried mothers

Eckenrode (2010): 25% of treated vs 37% of controls arrested; bigger effects with girls

Bilukha et al. (2005) systematic review

David Olds (Elmira, NY)

About 120 children age 3 randomly assigned to pre-

school or control groups

Experimental children get daily pre-school program plus

weekly home visits

The pre-school program was designed to increase

thinking and reasoning ability and school achievement

Find: By age 27, many benefits; experimental children

have half as many arrests as controls. Benefits per child

= $88,000, costs per child = $12,000, hence 7:1 ratio

Schweinhart et al. (2005): Age 40 follow-up: benefit:

cost ratio 17:1

Larry Schweinhart: Perry Pre-School Program (Ypsilanti, MI)

Triple-P Positive Parenting Program: can be used for primary prevention (media-based) or for high risk children or clinic samples

305 high-risk children randomly assigned to experimental or control conditions

Experimental parents receive training in 17 child management strategies, with modelling, role-playing, feedback and homework

Find: experimental children’s antisocial behaviour improved

Thomas & Zimmer-Gembeck (2007), Nowak & Heinrichs (2008) systematic reviews

Matt Sanders (Brisbane)

Leena Augimeri (Toronto)

Target boys aged 6-11 referred by police

Based on skills training, cognitive problem solving, self-control, anger management: aim to control impulsiveness

SNAP: Stop now and plan. Snap fingers

Stop: calm down, take deep breaths, count to 10

Now and: use coping statements, think what to say to remain calm: this is hard but I can do it!

Plan: effective solutions to interpersonal problems

Teach children to identify triggers: what makes them angry or upset

Leena Augimeri (Continued)

12 week program; group of about 7 children meet for 90 mins with facilitator (6-7, 8-9, 10-11)

Social skills, self-control, group discussion, modelling, behavioural rehearsal, role playing, relaxation training, home practice exercises

Topics include: joining in, dealing with anger, avoiding trouble, dealing with peer pressure, apologising, stopping stealing

Evaluation by Koegl et al. (2008): program is effective

Independent evaluations by Lipman (2008) d=.41; Burke & Loeber (2015) d = .40

Concrete materials for students, teachers and parents; utilization of Internet and Virtual Learning Environments/Computer games with an anti-bullying content, embedded in lessons.

Web-based questionnaire for students

Web-based discussion forum for teachers & teacher training

Increased playground supervision; distinctive vests for teachers during recess time; reorganization of school space

Peer-support group for victims of bullying

Information for parents

Karna et al. (2011) randomize 78 schools to experimental or control: program is effective with children ages 10-12

Ttofi & Farrington (2011) systematic review of bullying prevention

Christina Salmivalli (Finland): KiVabullying prevention program

Functional Family Therapy: Aim to change family contingencies to increase positive and decrease negative behavior

Aim to modify family communication patterns to be clearer and more reciprocal, considering alternative solutions to problems: work with entire family

86 delinquents randomly assigned to FFT or control conditions

Find: 26% of experimental delinquents reoffended, versus 55% of controls

Review by Sexton & Alexander (2000)

James Alexander (Utah)

Treatment Foster Care (TFC): foster parents use behaviour management methods to provide boys with a structured daily living environment, with close supervision and clear rules and limits

79 chronic male delinquents randomly assigned to TFC or group homes (in which group work, confronting negative behaviour, individual therapy)

Find: TFC boys have lower official and self-reported delinquency in a one year follow-up

MacDonald & Turner (2007) systematic review

Patti Chamberlain (Oregon)

Robert Ross (Ontario)

Reasoning and Rehabilitation program:

Key elements: self-control (teach offenders to stop and think), social skills (negotiating, how to respond to criticism, apologising), thinking skills (how to analyse interpersonal problems), creative thinking (consider prosocial options), critical reasoning, social perspective taking (see the other’s viewpoint), values enhancement (care about victims, empathy), emotional control (anger management), helper therapy (become prosocial trainers for other offenders)

Tong & Farrington (2008) systematic review: R&R is effective, more so in community than in institutions

Multi-Systemic Therapy (MST): family intervention to promote the parent’s ability to monitor and discipline the adolescent, peer intervention to promote prosocial friends, school intervention to enhance competence; work with family/peer/school, youth may not be present (unlike FFT)

176 serious delinquents (mean age 14) randomly assigned to 6m MST or individual therapy

Sawyer & Borduin (2011): the MST group had fewer felony arrests (33% vs 55%) and fewer years incarcerated (5.3 vs 7.9) up to age 37

Curtis et al. (2004), Littell et al. (2005) reviews

Charles Borduin (Columbia, MO)

Campbell Collaboration Crime & Justice Group: freely

available reviews at:

www.campbellcollaboration.org

What doesn’t work? Deterrent strategies: Scared

Straight (Petrosino), Boot Camps (Wilson), Official

processing (Petrosino)

What does work? Parent training (Piquero), child skills

training (Losel), mentoring (Tolan), self-control

programs (Piquero), bullying prevention (Ttofi),

cognitive-behavioural interventions (Lipsey)

Systematic reviews

Farrington, D.P. et al. (2017) Systematic reviews of the effectiveness of developmental prevention programs in reducing delinquency, aggression and bullying. Aggression and Violent Behavior, 33, 91-106.

50 systematic reviews of individually-based, family-based, and school-based interventions

All types of programs were effective

Median odds ratio = 1.46, corresponding approximately to a one-quarter reduction in prevalence

Family-based programs were most effective, followed by individually-based programs, with school-based programs somewhat less effective

Review of Systematic Reviews of Developmental Prevention

3. Benefit: Cost Ratios of Interventions

Very convincing argument to policy-makers: for every £1 spent on the program, £5 are saved

Calculate the costs of the program: capital versus recurring, average versus marginal

Calculate the benefits of the program, especially in terms of crimes prevented

Take account of inflation (by discounting) if the benefits are in the future; £1 in 10 years’ time is not worth the same as £1 today

Work out benefit: cost ratio: very useful measure of effectiveness

The metric matters! E.g. 8% decrease in reconviction = 7 to 1 benefit-to-cost ratio for restorative justice, according to the Washington State Institute for Public Policy

Cost-Benefit Analysis

Cost-benefit analysis of SNAP program of Augimeri

Program costs $4,641 per boy on average (2012 $)

Effect size d between .2 and .4

Corresponds to 18% to 33% decrease in offending

Saves 1.25 to 2.29 convictions per boy (age 12-20) based on convictions of 376 SNAP boys

Saves $9,493 to $17,404 per boy (discounted)

Benefit: cost ratio 2.1 to 3.8 for convictions

Scaling up to self-reports: benefit: cost ratio 17 to 32

Farrington & Koegl (2015)

Which is better? (in terms of reducing crimes and monetary benefits exceeding monetary costs)

More imprisonment?

More rehabilitative programs in prison/probation?

More police?

More community penalties?

More community programs?

More situational crime prevention?

More developmental crime prevention?

Choosing Crime Prevention Strategies

Advantage: all costs and benefits calculated comparably for Washington State (April 2012)

Olds: benefit $22,781, cost $9,600 (per child) B:C 2.4 to 1

Preschool: benefit $22,457, cost $7,523 B:C 3.0 to 1

FFT: benefit $33,967, cost $3,261 B:C 10.4 to 1

TFC: benefit $39,197, cost $7,922 B:C 4.9 to 1

MST: benefit $32,121, cost $7,370 B:C 4.4 to 1

Conclude: invest in prevention programs in childhood and adolescence, to reduce the prison population and save money

Steve Aos (Washington State)

4. Risk-Focussed Prevention Strategy

Since there are multiple risk factors, there should be

multiple-component interventions targeted on child,

family, peers, schools and communities

Generally, these are more effective than single

component interventions

But hard to identify active ingredients and decide

which elements of a package are more effective

How learn from experience and improve multiple-

component interventions?

Important multiple-component intervention:

Communities That Care (CTC)

Multiple-Component interventions

Key community leaders meet and agree to implement CTC

Set up Community Board to take charge of CTC on behalf of the community

Audit of problems and risk and protective factors using surveys (school, community) and records (police, school, social, census)

Assess existing resources, choose programs from a menu of strategies that have been proved to be effective in high-quality evaluations

Implement programs, evaluate effectiveness

Communities that Care

Prenatal/postnatal home visiting programs

Preschool intellectual enrichment programs

Parent training

Child skills training

Teacher training/curriculum development

Anti-bullying progams

Media campaigns

Situational prevention

Policing strategies

The Menu of Strategies

24 communities: 12 matched pairs

One community in each pair randomly assigned to CTC, one control

Student surveys from grades 5-10 (ages 10-15)

Over 4000 E & C students followed up to grade 10

Find decreases in:

Alcohol use 11%

Cigarette use 15%

Delinquency 11%

Marijuana use 11%

David Hawkins (2012) Evaluation

Kuklinski (2012) CBA of CTC

Based only on reductions in smoking and delinquency

Benefits were projected over the lifetime of participants (up to age 74)

CTC reduced the onset of smoking by grade 8 by 38%, and onset of delinquency by grade 8 by 21%

Smoking benefits estimated to be $812 per youth

savings from reduced crime were estimated to be $4,438 per youth (2004$)

total savings came to $5,250 per youth

Av. cost of program per youth was $513 to $991

therefore, benefit:cost ratio = 5.30 to 10.23 (to 1)

Kuklinski (2015) CBA of CTC

Based on reductions in delinquency, alcohol use, smoking cigarettes

Benefits projected up to age 65

Sustained abstinence up to grade 12:

Delinquency 41.7% vs 33.0%, alcohol use 32.2% vs 23.3%, smoking 49.9% vs 42.8%

Benefits per youth (2011$): Delinquency $4,477, alcohol use $287, smoking $45

Average cost of program per youth over 5 years was $556

Discounting over time, benefit:cost ratio = 8.22 to 1

5. Conclusions

Offenders differ significantly from non-offenders in many respects, including impulsiveness, empathy, low intelligence and low school achievement, poor parental supervision, child physical abuse, punitive or erratic parental discipline, cold parental attitude, parental conflict, disrupted families, antisocial parents, large family size, low family income, antisocial peers, high delinquency-rate schools, and high crime neighbourhoods.

More longitudinal studies are needed in different countries, with frequent measurement of risk factors and offending, to study within-individual changes in risk factors and in offending.

Research Implications

High quality evaluation research shows that many programs are effective in reducing delinquency, and that in many cases the financial benefits of these programs outweigh their financial costs.

More experimental and quasi-experimental evaluations are needed.

The best programs include general parent education in home visiting programs, parent management training, pre-school intellectual enrichment programs, child skills training, anti-bullying programs, mentoring, FFT, MTFC and MST.

Communities That Care is a useful overarching program.

Implications for Interventions

A national prevention agency could provide technical assistance, skills and knowledge to local agencies in implementing prevention programs, could provide funding for such programs, and could ensure continuity, co-ordination and monitoring of local programs.

It could provide training in prevention science for people in local agencies, and could maintain high standards for evaluation research. It could also act as a centre for the discussion of how policy initiatives of different government agencies influence crime and associated social problems.

It could set a national and local agenda for research and practice in the prevention of crime, drug and alcohol abuse, mental health problems and associated social problems.

It could also maintain a computerized register of evaluation research and advise governments about effective and cost-effective crime prevention programs.

National Prevention Agency