AGGRESSION REPLACEMENT TRAINING (ART): AN AUSTRALIAN YOUTH JUSTICE

EVALUATION

Matthew R. Currie

Psychological Sciences and Statistics

Faculty of Life and Social Sciences

Swinburne University of Technology

Hawthorn, Victoria, Australia

A thesis submitted in partial fulfillment for the degree of

Professional Doctorate in Clinical Psychology

September, 2010

ii

Abstract

Aggression Replacement Training (ART) is a multimodal, cognitive-behavioural intervention for

aggressive youth. The results of several outcome studies from the US, UK and Europe indicate that

ART is effective in reducing aggressive behaviours and thoughts and recidivism among offenders.

This study evaluated the effectiveness of ART for aggressive juvenile offenders (aged 17 to 20 years)

at an Australian youth justice custody centre. The results of an initial pilot study (N= 5) indicated

overall support for the clinical utility of ART. Thus, with some methodological and programmatic

amendments a larger (N= 20) study was undertaken. A repeated measures design was used with

pre-treatment, post-treatment, 6-month, and 24-month follow-up assessments. Results showed

significant reductions in self-reported aggressive behaviours and thoughts, cognitive distortions, and

impulsivity and some improvement in social problem solving skills. The effect sizes were moderate to

large and the treatment effects were maintained at two year follow-up. Contrary to expectations,

custody worker ratings of participant’s aggressive/antisocial behaviours did not show significant

reduction at the six month follow-up. Additionally, these ratings were not clinically significant at any

time point. This measure had not previously been used in a correctional setting, however, the

workers may have been unreliable respondents due to their lack of knowledge of the ART

participants. Despite efforts, a waitlist comparison group could not be obtained. Results from this

study provide support, in terms of youth self-report measures, for the application of ART with

moderate to high risk aggressive offenders in an Australian youth justice custodial setting. Further

investigation with a control comparison and a reliable non-self-report informant is required to confirm

the findings. The results also support the cultural and age-related applicability of the learning

procedures inherent in ART for this group of moderate to high risk young offenders.

iii

DECLARATION

I declare that this thesis is my own account of my research and does not contain work that has been

previously submitted for a degree at any institution or for publication, without due acknodelgement.

Matthew R. Currie

iv

ACKNOWLEDGEMENTS

My eternal gratitude to my supervisor Katie Wood who saw to it that I actually completed this thesis.

There were a few occasions where I seriously doubted that I would. In addition to her role as my

coordinating research supervisor, Katie has provided invaluable support as a professional practice

supervisor. Working with mostly involuntary, violent young offenders has been, and continues to be,

highly rewarding and at times really, really challenging. Katie has been a wonderful mentor in helping

me to reflect on my personal journey in the work I do, as well as build on my strengths to do it better.

I would also like to thank my colleagues Cath Powell and Charlene Pereira, both, particularly Cath,

helped facilitate all the ART programs that were part of this thesis. Most importantly, they also helped

me to navigate the complexities of the youth justice system. I also wish to thank Michael Crewdson,

Felicity Dunn, and Tania Nahum all of whom, a various stages of my professional development have

acted as clinical supervisors. Their contribution to my personal and professional development has

been highly rewarding. I would like to thank each of the young men who participated in this research

and freely gave of their time at each of the follow-up assessments. In most cases they did this long

after exiting the criminal justice system, which stands as testimony to the effectiveness of ART.

Finally, big respect to Arnold Goldstein and his colleagues for developing ART and making sure it is

readily available in the service of youth.

v

CONTENTS

Abstract .............................................................................................................................................ii

Table of Figures................................................................................................................................ix

List of Tables .................................................................................................................................... x

Chapter 1 Introduction and Overview of Thesis ............................................................................... 1

The Disruptive Behaviour Disorders: Definitions and Classifications....................................... 5

Attention-Deficit/Hyperactivity Disorder (ADHD)............................................................. 6

Oppositional Defiant Disorder (ODD) and Conduct Disorder (CD) ................................. 8

Comorbidity and disruptive behaviour disorders.................................................................... 11

DBD’s and Aggression: Developmental pathways................................................................. 11

DBD’s and Offenders............................................................................................................. 15

Summary ............................................................................................................................... 16

Chapter 2 Changing Offending Behaviour: What works? .............................................................. 18

The Risk-Needs-Responsivity Model: The core principles of “what works” ........................... 18

Effective Treatment with Offenders ....................................................................................... 22

Summary of “What Works” .................................................................................................... 24

Chapter 3 Aggressive Behaviour: The theoretical context ............................................................. 25

Behaviourism: Operant learning ............................................................................................ 25

Social learning theory ............................................................................................................ 26

Cognitive-Behavioural Therapy (CBT) and It’s Underlying Theory ........................................ 27

Summary ............................................................................................................................... 32

Chapter 4 Aggression Replacement Training: A multimodal, cognitive-behavioural intervention .. 33

Skillstreaming ........................................................................................................................ 33

Skillstreaming: Content and procedure......................................................................... 34

vi

Defining Social Competence, Social Skills and Social Problem Solving................................ 36

Social Skills Training and Aggressive Youth.......................................................................... 39

Social Problem-Solving and Aggressive Youth...................................................................... 43

Summary ............................................................................................................................... 46

Anger Control Training .......................................................................................................... 47

Anger control training: Content and procedure............................................................. 48

Anger, Self-regulation and Aggression .................................................................................. 50

Impulsivity and Aggression.................................................................................................... 55

Evaluations of ACT................................................................................................................ 60

Summary ............................................................................................................................... 62

Moral Reasoning Training ..................................................................................................... 63

Moral Reasoning Training: Content and procedure………………………………………..64

Moral Reasoning: The theories of Piaget and Kohlberg ........................................................ 66

Moral Reasoning and Social Perspective Taking .................................................................. 72

Moral Reasoning and Offending Behaviour ........................................................................... 75

Self-Serving Cognitive Distortions and Antisocial Youth........................................................ 79

Summary ............................................................................................................................... 83

Chapter 5 Previous ART Evaluations and Rationale for the Present Study................................... 86

Previous Evaluations of ART................................................................................................. 86

Rationale for the Present Study............................................................................................. 90

Chapter 6 Pilot Study..................................................................................................................... 92

Method .................................................................................................................................. 92

Participants................................................................................................................... 92

Measures...................................................................................................................... 92

Procedure..................................................................................................................... 95

vii

Results .................................................................................................................................. 97

Discussion ........................................................................................................................... 100

Methodological and Programmatic Issues.................................................................. 103

Limitations and Summary of Findings ........................................................................ 106

Chapter 7 Main Study .................................................................................................................. 109

Aims and Hypotheses.......................................................................................................... 109

Method ................................................................................................................................ 110

Participants................................................................................................................. 110

Recruitment ................................................................................................................ 111

Measures.................................................................................................................... 112

Procedure................................................................................................................... 115

Results ................................................................................................................................ 117

Preparation of Data .................................................................................................... 117

Dealing with Missing Data .......................................................................................... 118

Statistical Analysis...................................................................................................... 119

Scale Reliability and Intercorrelations......................................................................... 120

Aggression: Youth self-report ...................................................................................... 124

Cognitive Distortions and Behavioural Referents: Youth self-report............................ 130

Social Problem Solving: Youth self-report ................................................................... 141

Impulsivity: Youth self-report ....................................................................................... 144

Adult Behavior Checklist: Worker-report ..................................................................... 145

Discussion ........................................................................................................................... 147

Aggressive Behaviours and Thoughts ........................................................................ 148

Cognitive Distortions and Moral Reasoning ................................................................ 155

Social Problem Solving ............................................................................................... 157

viii

Impulsivity ................................................................................................................... 162

Maintenance of Treatment Gains ................................................................................ 165

Group Dynamics: Process versus content .................................................................. 168

Clinical and Policy and Planning Implications ............................................................. 173

Methodological Considerations: Strengths and limitations .......................................... 176

Directions for Future Research ................................................................................... 179

Summary and Concluding Comments......................................................................... 181

References ................................................................................................................................... 185

Appendix A: ART Group Materials................................................................................................ 222

Appendix B: Ethics………………………………………………………………………………………...230

Appendix C: Measures……………………………………………………………………………………243

Appendix D: Single case study ..................................................................................................... 268

Appendix E: Statistical output ....................................................................................................... 272

ix

Table of Figures

Figure 1. Research plan of thesis. .................................................................................................... 4

Figure 2. A continuum of methods in behavioural and cognitive therapies. .................................... 29

Figure 3. The anger control chain. .................................................................................................. 49

Figure 4. Determinants and concequences of anger ...................................................................... 52

Figure 5. Agression Questionnaire Total Scale score T1 to T4. ................................................... 126

Figure 6. Physical Aggression T1 to T4. ....................................................................................... 127

Figure 7. Hostility from T1 to T4. .................................................................................................. 128

Figure 8. Indirect Aggression from T1 to T4. ................................................................................ 129

Figure 9. HIT total from T1 to T4................................................................................................... 132

Figure 10. Self-Centered Thinking from T1 to T4.......................................................................... 133

Figure 11 . Blaming Others from T1 to T4. ................................................................................... 134

Figure 12. Minimizing/Mislabeling from T1 to T4. ......................................................................... 135

Figure 13 . Assuming the Worst from T1 to T4. ............................................................................ 136

Figure 14. .Lying from T1 to T4. ................................................................................................... 137

Figure 15 . Stealing from T1 to T4. .............................................................................................. 138

Figure 16 . Oppositional Defiant from T1 to T4. ............................................................................ 139

Figure 17 . HIT Physical Aggression T1 to T4. ............................................................................. 140

Figure 18 .Social Problem Solving Total Scale from T1 to T4....................................................... 143

Figure 19 . Impulsiveness from T1 to T4. ..................................................................................... 144

x

List of Tables

Table 1 Skillstreaming skill- "keeping out of fights" ......................................................................... 35

Table 2 Kohlberg's stages of moral reasoning ............................................................................... 69

Table 3 Selman's stages of perspective taking.............................................................................. 73

Table 4 Gibbs' stages of sociomoral reasoning .............................................................................. 76

Table 5 Pre- and post- treatment raw scores on the AQ, SSRS, and HIT Questionnaires ............ 97

Table 6 Pilot study pre- and post- treatment descriptives and Wilcoxon T-values for all outcome measures ................................................................................................................................. 99

Table 7 Methodological changes in the main study: Summary and rationale ............................... 108

Table 8 ART group completers, non-completers, waitlist controls, and refusals for main study .. 112

Table 9 Reliability coefficients for all dependent measures at T1, T2, T3 and T4......................... 121

Table 10 Inter-correlations for all dependent measures in the main study at T1……………………123

Table 11 Descriptive statistics and main effect ANOVA results from T1 to T4 on the AQ ........... 125

Table 12 Descriptive statistics and main effect ANOVA results from T1 to T4 on the HIT ........... 131

Table 13 Descriptive statistics and main effect ANOVA results from T1 to T4 on the SPSI ........ 142

Table 14 T-score descriptive statistics and main effect ANOVAS from T1 to T3 for the ABCL.... 146

Table E1 Inter-correlations for all dependent measures in the main study at T2…………………...317

Table E3 Descriptives and main effect ANOVA results from T1 to T4 on the AQ (LOCF) ........... 318

Table E3 Descriptives and main effect ANOVA results from T1 to T4 on the HIT (LOCF)........... 319

Table E4 Descriptives and main effect ANOVA results from T1 to T4 on the SPSI (LOCF) ....... 320

Table E5 Descriptives and main effect ANOVA results from T1 to T4 for Impulsiveness (LOCF).321

Table E6 T-score descriptivs and main effect ANOVAS from T1 to T3 for the ABCL (LOCF) ..... 322

xi

1

Chapter 1

Introduction and Overview of Thesis

The World Health Organisation (WHO) has reported that violence is one of the leading global

public health concerns today (Krug, Dahlberg, Mercy, Zwi, & Lazano, 2002). Aggressive behaviour

and violent offending represents one of the most prevalent, stable, and costly problems in society

(Guerra & Slaby, 1990; Kazdin, 1987). The individual and financial burden of violent offending

includes increased public expenditure associated with policing, probable recurring imprisonment of

offenders and victim costs (Polaschek & Dixon, 2001; Serin, Gobeil, & Preston, 2009). Recent

Australian statistics indicate that offences defined as “acts intended to cause injury” account for the

highest proportion (18%) of the total adult (median age of 33 years) prison population (Australian

Bureau of Statistics [ABS], 2008; p.21). According to the ABS, the proportion of violent offenders

among younger prisoners (18-25 years) is even greater, accounting for 25% of the total prison

population. Moreover, ABS crime statistics indicate an average yearly increase in assaults by 5% for

each year between 1995 and 2007 (Australian Institute of Criminology [AIC], 2009). Although most

correctional jurisdictions in Australia offer some form of anger management program for violent

offenders, there are few published evaluations of such interventions (Howells & Day, 2002). Indeed,

Serin et al. (2009) noted that despite the considerable financial, social and psychological costs

associated with aggressive offending there is relative lack of accumulated knowledge in the

international literature regarding the effectiveness of programs targeting violent offending.

According to Goldstein and Glick (1987) interventions aimed at remediating antisocial and

aggressive behaviours in young people need to address the cognitive, behavioural and affective

components of aggression to show even moderate treatment gains. Multimodal, cognitive-

2

behavioural interventions have been shown to be most effective in treating the complex set of

behaviours associated with conduct related disorders once they have become an entrenched

behavioural repertoire among antisocial youth (Fonagy & Kurtz, 2002; Kazdin, 1997; Lochman et al.,

2000). Aggression Replacement Training (ART; Goldstein, Glick, & Gibbs, 1998) represents an

exemplar of this type of intervention. Several researchers (e.g., Barnoski, 2004; Fonagy & Kurtz,

2002; Palmer, 2007; Polaschek, 2006) have indicated that ART is one of the most efficacious

interventions in working with moderate to high risk violent young offenders. This thesis reports the

findings of a two-year longitudinal evaluation of ART in an Australian, custodial youth justice setting

with males aged 17 to 20 years. The results of two studies (i.e., a small pilot study and a larger multi-

informant main study) are reported and discussed.

Chapter 1 begins with an examination of the Disruptive Behaviour Disorders (Attention

Deficit/Hyperactivity Disorder, Oppositional Defiant Disorder, and Conduct Disorder) and their

relationship to youth aggression and offending behaviour. In Chapter 2 the “what works” research

relevant to offending behaviour (Andrews, Bonta, & Hoge, 1990) is examined, emphasising the risk-

need-responsivity model of offender treatment and meta-analytic research that shows cognitive-

behavioral interventions as the most effective in treating offender populations. Chapter 3 discusses

the current theoretical perspectives that inform ART, including operant and social learning theories

which represent the theoretical basis of CBT interventions with aggressive youth and offender

populations. Chapter 4 provides a detailed examination of the cognitive-behavioural, multimodal

components of ART (i.e., Skillstreaming, Anger Control Training, and Moral Reasoning Training) in

terms of their respective practice content and procedures, theory, and empirical evidence. In Chapter

5 the published evaluations of ART to date are reviewed which leads into the rationale informing the

present study.

3

Chapter 6 reports the pre- to post- treatment results of the initial pilot study of ART in an

Australian youth justice custodial setting. Firstly, the aims, hypotheses and methodology of the pilot

study are described. Then the results of the pilot study will be reported and briefly discussed, with an

emphasis on the key learnings and how they translated into programmatic and methodological

refinements for the Main study. Chapter 7 begins with a detailed description of the methodology

applied in the Main study followed by details of data preparation and analysis procedures.

Subsequently the results for the Main study are reported specific to the targets of change assessed

(i.e., youth self-report -aggression, -cognitive distortions, -social problem solving and -impulsiveness,

and custody worker-reports of youth aggression and antisocial behaviour). Chapter 8, the final

chapter, discusses the results of the main study. Again this chapter is structured according to the

specific targets of change. The maintenance of treatment gains at two-year follow-up is then

examined. Group process variables, such as the therapeutic alliance and group cohesion, is then

discussed as they relate to the effectiveness of ART and indeed any intervention with offenders. The

final sections of the Discussion review the clinical and policy and planning implications of this thesis,

followed by the methodological strengths and limitations. This thesis will conclude with

recommendations for future research and final comments summarising how this study has

contributed to offender treatment literature broadly and the ART literature specifically. Figure 1

provides an overview of the research plan and changes implemented from Studies One (pilot study)

to Two (main study).

4

Figure 1. Research plan of thesis.

Note. NS = non-significant; AQ = Aggression Questionnaire (Buss & Warren, 2000); HIT = How I Think Questionnaire (Barriga, Gibbs, Potter, & Liau, 2001); SSRS = Social Skills Rating System (Gresham & Elliot, 1990); SPSI = Social Problem Solving Inventory- Revised (D'Zurilla, Nezu, & Maydeu-Olivares, 2002); I7 = Impulsiveness Scale (Eysenck, Pearson, Easting, & Allsopp, 1985); ABCL = Adult Behavior Checklist (Achenbach & Rescorla, 2003).

5

The Disruptive Behaviour Disorders: Definitions and Classifications

The two most commonly used systems of classification for psychiatric disorders- the

International Classification of Diseases- Tenth Edition (ICD-10; WHO, 1992) and the Diagnostic and

Statistical Manual of Mental Disorders Fourth Edition- Text Revised (DSM-IV-TR; American

Psychiatric Association [APA], 2000) describe a set of diagnostic criteria characterised by

externalising behaviours (e.g., lying, stealing, oppositionality, and aggression) throughout childhood

and adolescence. These disorders are commonly referred to as the disruptive behaviour disorders

(DBD’s) which comprise Attention Deficit Hyperactivity Disorder (ADHD) (or Hyperkinetic Disorders

{HKD] in the ICD-10) Oppositional Defiant Disorder (ODD) and Conduct Disorder (CD) (Cowling et

al., 2005). In Australia, the DSM is the most widely used system of classification for clinical disorders.

Thus, for the purpose of the present review, discussion of the diagnostic categories relevant to

aggressive and antisocial behaviour will focus primarily upon the criteria described in the DSM-IV-TR

(APA, 2000).

Lochman, Whidby and Fitzgerald (2000) defined aggression as “a set of primarily

interpersonal actions that consist of verbal or physical behaviours that are destructive or injurious to

others or to objects” (p.31). Throughout the course of “normal” psychosocial development, evidence

for aggressive behaviour can be seen in the social interactions of almost all children and adolescents

to varying degrees (Hawley, 1999; Tremblay, 2000). In cases where aggression becomes a more

entrenched pattern of social interaction, however, a disruptive behaviour disorder (DBD) diagnosis

might be indicated (Lochman et al.). The externalising behaviours associated with the DBD’s violate

the rights of others and common social rules and standards (APA, 2000). Epidemiological studies

indicate that conduct related disorders are increasing in clinical and non-clinical populations with

6

DBD’s being the most common reason for referral to child and adolescent psychiatric services

(Kazdin, 2002; Phelps & McClintock, 1994; Webster-Stratton & Dahl, 1995). In Australia, a national

survey of 4,500 young people aged between 4-17 years reported a prevalence rate of 14.2% for

DBD’s (Sawyer et al., 2000).

Kazdin (1987) reported that an early onset of a DBD can be related to continuing aggressive

and anti-social behaviours into adolescence and adulthood (e.g., school drop out, alcoholism, drug

abuse, juvenile delinquency, adult crime, marital/interpersonal disruption, and poor physical and

mental health outcomes). Given the reported stability of aggressive and antisocial behaviour across

time and setting, and that the complexity of behaviours vary qualitatively and quantitatively, the task

of identifying effective interventions has proven to be highly problematic (Conduct Problems

Prevention Research Group, 1999a; 1999b; Kazdin, 1987; 1997). In this section, the DBD’s are

reviewed in terms of their diagnostic characteristics, prevalence, comorbidity and relevance to

offender populations. It is important to note that the participants in the present study were not

assessed for DBD’s. Rather, the aim of this section is to provide a context for the nosology of violent

offending behaviour.

Attention-Deficit/Hyperactivity Disorder (ADHD). ADHD is defined as a pervasive pattern

of inattentive behaviour and/or hyperactivity-impulsivity that is more frequent and severe than is

typically observed in individuals of the same developmental level (APA, 2000). Children and

adolescents with inattentive behaviour often make careless mistakes with schoolwork, find it hard to

persist with tasks and are easily distracted. Those with problems in the area of

hyperactivity/impulsivity often fidget and talk excessively, interrupt others, and are constantly “on the

go” (APA, 2000). The DSM-IV-TR (APA, 2000) distinguishes between three subtypes of ADHD based

on the predominant symptom pattern for the past six months. Children and adolescents with

7

symptoms of both inattentiveness and hyperactivity-impulsivity are diagnosed with ADHD-Combined

Type; those with primarily inattentive symptoms are diagnosed with ADHD- Predominantly Inattentive

Type; and those with primarily hyperactivity-impulsivity symptoms are diagnosed with ADHD-

Predominantly Hyperactive-Impulsivity Type (APA, 2000). Additionally, the DSM-IV-TR requires an

age of symptom onset before seven years and the presence of symptoms across two or more

settings (e.g., school and home). In contrast, the ICD-10 hyperkinetic disorder does not distinguish

between subtypes, rather a single disorder is described relative to the presence of both inattentive

and hyperactive-impulsivity symptoms which is consistent with the DSM-IV-TR ADHD-Combined

Type. The ICD-10, unlike the DSM-IV-TR, does not allow for comorbid diagnoses, with the exception

of the presence of conduct related problems warranting a diagnosis of hyperkinetic conduct disorder.

These differences in diagnostic classification have implications for the reported prevalence rates of

HKD and ADHD. ADHD has consistently been shown to have a significantly higher prevalence rate

than HKD (Lee et al., 2008).

The prevalence rate for ADHD among school-aged children has been estimated to

approximate 3-7% (APA, 2000). Polanczyk et al. (2007) reported a global prevalence rate of 5%.

Given that epidemiological studies represent approximations of clinical diagnosis, the accuracy of

prevalence rates is dependent on having reliable diagnostic tools (Rowland, Lesesne, & Abramowitz,

2002). In the case of ADHD, diagnosis is primarily based on teacher and parent ratings of child

behaviour, hence the rigors of clinical assessment are often lacking in the assessment of this

condition (Rowland et al.). Moreover, Rowland et al. argued that the presence of comorbid conditions

(e.g., ODD, CD, anxiety and learning disorders) might confound accurate diagnosis, which is likely to

contribute to inflated prevalence rates. Population estimates of ADHD vary greatly with upper

estimates of up to 17.8% (Goldman, et al., 1998). In Australia, epidemiological data indicate a

8

prevalence rate for ADHD of 15% for boys and 7% for girls between the ages 4-17 years. The

disorder was more prevalent among primary school aged boys (19%) than adolescent males (10%)

(Sawyer et al., 2000). However, this survey was not based on the full diagnostic criteria, which might

have artificially inflated the prevalence rates. The gender differences reported by Sawyer et al.

approximate other epidemiology research in which the diagnosis of ADHD among males is estimated

to be greater than females by ratios ranging from 2:1 to 9:1 (APA, 2001). Several explanations have

been proposed to help explain these gender differences, including biased sampling (i.e., differences

in gender ratios tend to decrease among community as opposed to clinical samples) and possibly

biased diagnostic criteria and tools which favour detection in males (Barkley, 1995; Goldman et al.,

1998; Waschbusch & King, 2006).

The early symptoms of inattention, hyperactivity, and impulsivity are often detected in the first

years of school (Goldstein & Rider, 2005). The effects of ADHD cause significant psychosocial

impairment across multiple domains including academic failure, poor interpersonal relations, poor

self-esteem, and self-regulatory deficits (Barkley, Fischer, Edelbrock, & Smallish, 1990). High rates of

psychosocial impairment across the lifespan have been reported for those diagnosed with the

disorder. For example, 32 to 40% are estimated to drop out of school, 50-70% reported having few

friends, 70-80% reported underachievement in the work place, 40-50% engaged in antisocial

behaviour, and 18-25% have been shown to develop personality disorders in adulthood (International

Consensus Statement on ADHD, 2002).

Oppositional Defiant Disorder (ODD) and Conduct Disorder (CD). The DSM-IV-TR (APA,

2000) defines ODD as a persistent pattern of angry, antagonistic, hostile, vindictive, and irritable

behaviour toward authority figures and/or peers for at least six months. The problem behaviours

occur more often than is typical of similarly matched peers in terms of age and developmental level.

9

The onset of oppositional behaviours usually occurs in the home and later transfers to other settings,

such as school, usually before eight years of age. In contrast, CD is characterised by “a repetitive

and persistent pattern of behaviour in which the basic rights of others or major age-appropriate

societal norms or rules are violated” (APA, 2000; p.93). These behaviours fall into four categories: (1)

actual or threatened physical aggression toward people or animals; (2) deliberate destruction or

damage to property; (3) acts of wilful deception and theft; and (4) serious violation of societal rules.

These diagnostic categories are typically manifested through antisocial behaviours such as recurrent

instances of bullying, frequent physical fights, deliberate destruction of other people’s property,

breaking into houses or cars, staying out late at night despite parental prohibitions, running away

from home, and/or frequent truancy from school. To meet DSM-IV criteria for CD, children and

adolescents must exhibit three or more of these behaviours during the past 12 months, with at least

one behaviour present in the last six months (APA, 2000).

Unlike ADHD, a diagnosis of ODD or CD does not require that the behaviours are present in

more than one setting. Yet, as with ADHD, a multi-informant (i.e., teacher, parent, self-report and

clinical assessment) approach to diagnosis is usually employed and it is not uncommon for

disagreement between informants, which can complicate diagnosis and influence prevalence rates

(Bird, Gould, & Staghezza, 1992; Werry, 1997). Prevalence rates for ODD have been estimated to be

between 2 to 15% (Nock, Kazdin, Hiripi, & Kessler, 2007). According to Goldstein and Rider (2005)

these variations in estimates are likely to be attributable to small sample sizes and unreliable

diagnostic methods. Australian Epidemiological data indicate a prevalence rate for CD of 3% among

4-17 year olds (Sawyer et al., 2000). The prevalence rate was higher for males (4.4%) than females

(1.6%) which is consistent with other reports of gender differences for CD (APA, 2000). In a review of

the existing literature Kazdin (1987) estimated a US population prevalence rate of CD ranging from 4

10

to 10%. Among juvenile offenders, however, the prevalence of CD is far greater. For example, Fazel,

Doll, and Langstrom (2008) reported a prevalence rate of 52.8% for CD for both males and females

in juvenile detention. For males, this represented 5 to 10 times greater risk for the development of CD

compared to community samples. CD has consistently been shown to be the most prevalent

psychiatric disorder among young offender populations (Abrantes, Hoffmann, & Anton, 2005). Yet,

even after excluding CD, Teplin, Abram, McClelland, Dulcan, and Mericle (2002) reported that almost

60% of males in juvenile detention met criteria for one or more psychiatric disorders.

There is debate among researchers as to whether ODD merely represents a milder, earlier

developmental manifestation of CD (Maughan, Rowe, Messer, Goodman, & Meltzer, 2004; Nock et

al., 2007). The diagnostic criteria for ODD in both the DSM-IV-TR (APA, 2000) and ICD-10 (WHO,

1992) allude to this purported developmental progression from ODD to CD in precluding a diagnosis

of ODD in the presence of CD. Certainly, there is much research showing that the early onset of

behavioural problems in childhood, consistent with a diagnosis of ODD, represents a key risk factor in

the development of CD (Broidy, et al. 2003; Dodge & Pettit, 2003; Loeber, 1991).

Adolescents who are most likely to be chronically antisocial are those who first evidenced

symptoms of ODD in the preschool years. Thus, the primary developmental pathway for

serious conduct disorders in adolescents and adulthood appears to be set in the preschool

period. ODD is a sensitive predictor of subsequent CD, in that nearly all youths have shown

previous ODD (Webster-Stratton & Dahl, 1995; p. 336).

While some youth with early onset ODD go on to develop CD, many do not (Loeber, Burke, Lahey,

Winters, & Zera, 2000; Nock et al., 2007). Nevertheless, comorbidity between the two conditions is

more the norm than the exception (Loeber, 1991; Webster-Stratton & Dahl, 1995).

11

Comorbidity and disruptive behaviour disorders. Kazdin (1997) suggested that comorbidity

among the DBD’s is very common among cases referred for treatment. He reported that as many as

70% of the cases treated in clinical settings meet DSM criteria for at least two disorders. Sawyer et

al. (2000) reported that 23% of all children and adolescents in their sample who met criteria for one

disorder (ADHD, CD or Depressive Disorder) also reported symptoms consistent with diagnosis of a

second. Overall, males had a higher rate of comorbidity (27%) than females (15%). Less than 1% of

children and adolescents met criteria for all three disorders. The literature abounds with reports of

comorbid relationships between ADHD, ODD, CD, learning difficulties, mood disorders, anxiety

disorders, communication disorders, and Tourette's Disorder (APA, 2000; Biederman, Newcorn, &

Sprich, 1991). In an eight year follow-up study, Barkley et al. (1990) found that 80% of the children

with ADHD in their study were still hyperactive as adolescents and that 60% of them had developed

ODD or CD. Numerous studies (e.g., Barkley et al., 1990; Biederman, Petty, et al., 2008; Bird et al.,

1994) have found that young people with comorbid ADHD, ODD or CD, as opposed to one or the

other, are more likely to have greater cognitive impairment, increased family dysfunction, poorer

psychosocial outcomes (e.g., peer rejection, educational and employment failure, substance abuse,

ongoing antisocial behaviour) and earlier onset of conduct related problems.

DBD’s and Aggression: Developmental pathways

While a diagnosis of one DBD does not in, and of itself, indicate an increased likelihood of

persistent antisocial or aggressive behaviours, the characteristics of these externalising disorders

(i.e., a persistent pattern of impulsive, hyperactive, defiant and aggressive behaviours across various

settings) have consistently been shown to predict adverse psychosocial outcomes throughout

12

childhood, adolescence and into adulthood (Goldstein & Rider, 2005; Kazdin, 1987). Hence, current

causal conceptualisations of the development and persistence of serious conduct-related problems

aim to integrate multiple risk factors in a cumulative explanatory model (Dodge & Pettit, 2003; Frick,

2004). According to Dandreaux and Frick (2009) the large number of common risk factors associated

with CD can be grouped as either dispositional (e.g., difficult child temperament, impulsivity, deficits

in social information processing) or environmental (e.g., parental psychopathology, family conflict,

association with a deviant peer group). Similarly, Nietzel, Hasemann and Lynam (1999) proposed an

etiological sequence comprising multiple risk factors in the development of violent offending

behaviour. They suggested four main contributing pathways: (1) Distal antecedents, including

biological (e.g., neurochemical abnormalities) psychological (e.g., impulsivity) and environmental

(e.g., family functioning) factors; (2) Early indicators- for CD such as poor parenting and early

aggression; (3) Developmental processes- school failure, cognitive style (e.g., hostile attributions)

and substance abuse; and (4) Maintenance variables- antisocial peer associations, opportunism, and

socioeconomic deprivation. In assessing individual level of risk for severity and stability of aggressive

and antisocial behaviours, a cumulative approach has been proposed that takes account of multiple

risk factors from pre-birth through childhood and adolescent development (Frick, 2004; Nietzel,

Hasemann, & Lynam, 1999).

A robust predictor of increasing antisocial behaviour in terms of frequency and type (e.g.,

violent offending) has been the subtyping of CD according to age of onset- childhood or adolescent

onset type (Moffit et al., 2008). Additionally, several researchers (e.g., Broidy, et al. 2003; Dodge &

Pettit, 2003; Frick, 2004; Webster-Stratton & Dahl, 1995) have identified early physical aggression as

a precursor to the development of CD. Bor, Najman, O’Callaghan, Williams, and Anstey (2001)

conducted a longitudinal study of 3,792 Australian mothers and their children at age five with follow-

13

up at age 14 years. The strongest predictor of adolescent delinquency was aggressive behaviour at

age five. Bor et al. concluded that aggression in the early years is a reliable and powerful predictor of

continuing adolescent aggressive behaviour. The stability of aggressive and antisocial behaviour

across time suggests that the prognosis is likely to be poor for some children. Indeed, Webster-

Stratton and Dahl (1995) suggested that almost 50% of severely conduct disordered children

continue to be antisocial into adulthood. Gelhorn, Sakai, Price and Crowley (2007) reported an even

higher rate of antisocial behaviour persistence. Seventy-five percent of their large (N=43,093)

retrospective, American, stratified sample who met criteria for CD also met criteria for adult Antisocial

Personality Disorder (APD). Moreover, persistent antisocial behaviour was most strongly predicted by

victim and violence orientated offences, including “Stealing with Confrontation”, “Cruelty to People”,

“[use of] Weapons”, and “Lies” (p.536).

The findings of Gelhorn et al. (2007) and Bor et al. (2001) fit within the developmental

framework of CD, which distinguishes between a childhood-onset pathway and an adolescent-onset

pathway (e.g., Fergusson & Horwood, 2002; Frick, 2004; Moffitt, Capsi, Harrington, & Milne, 2002;

Odgers et al., 2008). Drawing on data from the Dunedin Longitudinal Study, Moffitt and her

colleagues (1993; 2001; 2002) provided a detailed examination of a relatively uncommon group of

life-course-persistent (LCP) individuals compared to the more prevalent adolescence-limited (AL)

group of antisocial offenders. LCP is characterised by early childhood cognitive deficits, difficult

temperament, and hyperactivity, which are exacerbated by environmental risks including poor

parenting, economic impoverishment and family discord. In later childhood and throughout

adolescence, environmental risk factors broaden to include problematic social relationships such as

peer isolation and conflict with teachers, and a later tendency toward antisocial peer associations and

poor relations with employers and partners. By contrast, AL antisocial behaviour is thought to co-

14

occur with the onset of puberty and is consistent with the normative task of individuation and

separation from parents and the greater influence of peers. The early environmental and

neurobiological risk factors associated with LCP are not typically seen in the AL group. However, in

cases where AL youth become involved in criminal activities, they are less likely to be convicted of a

violent offence or to spend time in prisons compared to youth in the LCP group (Odgers et al., 2008).

Research has consistently shown that childhood onset of conduct related problems, compared to

adolescent onset, is predictive of poorer outcomes throughout adolescence and into adulthood, such

as increased impulsivity, criminality and aggression (Frick, Cornell, Barry, Bodin, & Dane, 2003;

Moffitt, 1993).

In addition to age of onset, developmental research has implicated a difficult childhood

temperament in the development of antisocial behaviour (Frick & Morris, 2004). Specifically,

emotional regulation, particularly regulation of negative emotions combined with a lack of fearful

behavioural inhibitions, and callous-unemotional (CU) traits (i.e., a lack of empathy and guilt) have

consistently been shown to be associated with the development and maintenance of severe conduct

problems (Frick, 2004; Frick & Morris, 2004). This combination of emotional dysregulation and CU

traits are more commonly associated with the early onset LCP group of conduct disordered youth

(Moffitt, 1993). Moreover, consistent with the construct of adult psychopathy, the presence of CU

traits in youth has been shown to predict heightened levels of aggressive offending, increased

severity and number of conduct problems, more police contact, and a greater likelihood of family

histories of APD compared to antisocial youth without CU traits (Frick et al., 2003; Frick & Loney,

1999).

Frick and his colleagues (2005) investigated the predictive utility of CU traits relative to

severity and stability of conduct problems among a community sample of school children in Grades 3,

15

4, 6, and 7 over a four year period. Parent and teacher ratings on the Callous-Unemotional subscale

of the Antisocial Process Screening Device (APSD; Frick & Hare, 2001) were used to classify

participants as high or low on CU traits at each of the three follow-up assessments, each one year

apart. Parents and teachers also completed measures related to DSM-IV (APA, 2000) criteria for

ODD, CD and ADHD. Children completed a self-report measure of delinquency, which enabled the

researchers to determine the number and type of offences committed over the four year period. Four

groups (Control; CU-only; conduct problems [CP]- only; and CU+CP) of approximate equal size (n =

25) were compared on repeated measures over the four year period. As predicted, children with

conduct problems and CU traits showed a more severe and chronic pattern of antisocial behaviour

compared to the control, or CU-only or CP-only groups. Specifically, the CU+CP group showed

significantly higher levels of conduct problems, self-reported delinquency, and parent-reported police

contacts. Despite the limitation of a small sample size, Frick and colleagues demonstrated that the

presence of CU traits can predict conduct problem severity among a subgroup of at risk youth, in

terms of heightened aggressive and antisocial behaviour.

DBD’s and Offenders

Comorbid psychiatric illness among youth in juvenile detention has been widely reported

(Abram, Teplin, McClelland & Dulcan, 2003; Atkins et al., 1999; Fazel et al., 2008; Teplin, Abram,

McClelland, Dulcan & Mericle, 2002). Fazel et al. (2008) conducted a meta-analysis of 25 surveys

conducted in various locations around the world, including Australia, in which estimates of psychiatric

disorders for male and female youth in juvenile detention (mean age 15.6 years, range 10-19 years).

They found that the incidence of ADHD among incarcerated youth was two to four times greater

compared to the general adolescent population and adolescents in juvenile detention were

16

approximately 10 times more likely to suffer a psychiatric illness than their non-offender peers. Eme

(2008) reported similar high levels of ADHD among juvenile offenders compared to age-related

peers. In a small Australian study, Bickel and Campbell (2002) reported that 98% of the juvenile

detainees (aged 12 – 18 years, 43 males and 7 females) in their study met criteria for CD, 46% met

criteria for ADHD, 46% were suffering a mood disorder, and 32% suffering an anxiety disorder.

Comobidity with CD was high, 52% were found to have three or more mental health problems in

addition to CD. These results were based on the Adolescent Psychopathology Scale (Reynolds,

1998) a semi-structured psychometric tool designed to assess the potential presence of DSM-IV

diagnostic categories. The absence of a full diagnostic clinical interview in this study may have

inflated reported prevalence rates.

Summary

In this chapter, the DBD’s were reviewed in order to place adolescent and early-adult

criminality, in particular violent offending, within the context of current clinical nosology. As noted, a

diagnosis of one of the DBD’s does not necessarily suggest a heightened risk of criminal behaviour.

However, numerous studies have reported a high rate of comorbidity between the DBD’s (Barkley et

al., 1990; Kazdin, 1997; Sawyer et al., 2000). There is evidence that early behavioural problems in

childhood represent a key risk factor in the development of CD. There is also evidence to support the

subtyping of CD according to age-of-onset. Childhood onset, as opposed to later adolescent onset,

has been linked to a life-course persistent (LCP) group of antisocial offenders who are unlikely to just

grow out of aggressive and antisocial behaviours (Frick, 2004; Moffitt et al., 2002; Odgers et al.,

2008). Additionally, LCP offenders have been shown to be at greater risk of long-term negative

psychosocial outcomes including increased impulsivity, criminality, and aggression and early-adult

17

development of APD. The presence of CU traits in combination with conduct related problems has

also been identified as a major risk factor in the developmental trajectory of increasing aggressive

and antisocial behaviours (Frick et al., 2005). CD is the most common psychiatric disorder among

young offenders, although high rates of comorbidity with other psychiatric disorders have been

reported (Bickel & Campbell, 2002; Fazel et al., 2008). Of particular relevance to the present study is

the finding that early physical aggression is a reliable predictor for the development of CD and

ongoing antisocial aggressive behaviours (Broidy, et al. 2003; Dodge & Pettit, 2003; Frick, 2004;

Webster-Stratton & Dahl, 1995).

The next chapter provides an overview of the key principles that constitute the “what works’

literature relevant to offenders. The empirical evidence with respect to the effectiveness of specific

modes of intervention, particularly cognitive-behavioural therapies, aimed at reducing criminal

offending is then reviewed.

18

Chapter 2

Changing Offending Behaviour: What works?

Since the mid-1970’s, there has been a pervasive belief that lasting behaviour change

among criminal offenders is at best unlikely, if not impossible. This pessimistic view was primarily

informed by Martinson’s (1974) influential publication- “What works? Questions and answers about

prison reform”. Here, Martinson summarised the results of several hundred studies according to

negative versus positive outcomes. Hollin, Browne and Palmer (2004) criticised this approach

arguing that it failed to account for differences in the types of intervention employed, setting of

intervention, and the outcome measures used. They argue that more recent advances in statistical

procedures, such as meta-analysis, provide a more accurate picture in aggregating the results of

multiple empirical outcome studies. Indeed, according to Hollin and his colleagues, meta-analytic

procedures have significantly altered the pessimistic view that “nothing works” to a more empirically

validated position that behaviour change and transfer of gain can be achieved through certain

offender treatment programs. In this chapter, an overview of the core principles of “what works” in

treating adult and juvenile offenders, that is the criminogenic risk-need-responsivity model (Andrews,

Bonta, & Hoge, 1990) is reviewed. Then the empirical literature relevant to the general CBT approach

with offender populations is examined.

The Risk-Needs-Responsivity Model: The core principles of “what works”

Andrews and his colleagues (Andrews, Bonta, & Hoge, 1990; Andrews et al., 1990; Bonta &

Andrews, 1996) were the first to examine the specific variables associated with offender rehabilitation

and reductions in recidivism (Hornsveld, Nijman, Hollin, & Kraaimaat, 2008). Andrews et al. (1990)

19

described five core principles that constitute effective practice in treating offenders with the aim of

reducing rates of re-offending: (1) Risk- level of risk of reoffending should be determined in order to

match intensity of service provision (i.e., higher risk = more intensive). Risk factors are categorised as

either static- that is unchangeable (e.g., family history, age of first offence) or dynamic- that is

potentially changeable (e.g., criminal associates and cognitions, impulsivity, education, employment,

social skill and problem solving deficits); (2) Need- criminogenic needs include antisocial attitudes

and beliefs, antisocial peer associations, substance abuse, unstable accommodation and social

supports, lack of employment and education, hostility and anger and poor social problem solving

skills. In contrast, non-criminogenic needs include variables such as self-esteem, psychological

distress, group cohesion, and community participation (Ogloff & Davis, 2004). With respect to

reducing recidivism, the criminogenic needs should be the primary targets of change; (3)

Responsivity- interventions should be matched to meet risk and need principles in addition to

developmental needs, motivation, cognitive capacity, and learning styles; (4) Program Integrity-

interventions should have built in processes of monitoring treatment fidelity. That is, practice should

be firmly placed within a body of evidence-based theory and program design and delivered within

clearly defined parameters (i.e., a practice manual); and (5) Professional Discretion- notwithstanding

the above mentioned principles, it is a fundamental tenet of best-practice that professional/clinical

judgement should provide for scope in adapting to situational and individual client need (Andrews &

Bonta, 2003; Andrews & Dowden, 2007).

Andrews and Dowden (2006) conducted a meta-analysis that included 374 primary studies to

test the empirical validity of the risk principle (i.e., the extent to which assessment of level of risk

enhances correctional treatment outcomes). The inclusion criteria required the primary studies to

20

have reported a follow-up assessment period, a treated group of offenders compared to a non-

treated or non-equivalent treatment group, a measure of recidivism, and a data-set that could provide

for the calculation of an effect size based on recidivist data. Additionally, the primary studies were

coded for the presence of appropriate treatment factors (i.e., adherence to the core principles of need

and responsivity). Overall, moderate support for the risk principle (eta squared = .17, p>.00) was

found. This finding was the same irrespective of treatment setting (i.e., community or institutional).

However, there were age effects showing that the risk principle did not reach statistical significance

with older offenders (20-years or older) although the treatment effects were significant for the

principles of need and general responsivity irrespective of age. Andrews and Dowden suggested that

this finding might be partly explained by poor reporting practices of risk for the older participants and

a method of coding risk that was based upon procedures outlined in earlier research with younger

offenders. In other words, the classification methods for determining high versus lower risk offenders

might have been an inaccurate test of the risk principle with older offenders. Notably, the risk

principle only enhanced treatment effectiveness in studies where the principles of need and general

responsivity were adhered to. General responsivity refers to the use of theoretically informed

principles of offender treatment such as social learning theory (see Chapter 3) and cognitive–

behavioural treatment approaches. This finding suggested the importance of the interaction effects of

the risk-need-responsivity model as a holistic approach to offender treatment, which is consistent with

the findings of an earlier meta-analysis conducted by Andrews et al. (1990).

Ward and Stewart (2003) criticised the risk-need-responsivity model for its over-emphasis on

criminogenic needs as targets of change, to the exclusion of more psychological and interpersonal

orientated needs that serve to improve offenders quality of life. According to these authors, the

21

dynamic targets of change (i.e., criminogenic needs) in the risk-need-responsivity model represent an

approach to offender rehabilitation that is driven by an overarching correctional policy that is

“primarily concerned with risk reduction” (p.125). Ward and Stewart argued that reductions in

recidivism in offender treatment have become the benchmark of correctional program design and

effectiveness to the exclusion of other more subjective human, and at least equally relevant, clinical

targets of change. They proposed a broader conceptualisation of need that encompasses a holistic

model of human well-being, including the “psychological needs for relatedness, autonomy, and

competence” (p.134). Some support for Ward and Stewart’s contention is indicated by the use of

recidivism as the predominate measure of outcome in assessing offender treatment programs.

Several researchers (Losel, 2001; Serin et al., 2009; Tate, Reppucci, & Mulvey, 1995) have indicated

that the sole use of recidivism, while an important measure of outcome, limits our understanding of

program effectiveness and fails to account for other important clinical outcomes. In essence, Ward

and Stewart argue for a strengths-based model of intervention in offender treatment that builds upon

universal psychological and humanistic needs, thereby reducing the inclination toward antisocial and

criminal behaviour. In reply, Bonta and Andrews (2003) suggested that Ward and Stewart had over-

simplified the risk-need-responsivity model by suggesting that the model does not take account of

offender strengths or broader universal needs. The Level of Service Inventory- Revised (LSI-R)

developed by Andrews and Bonta (1995) is a widely used measure of offender risk and needs, which

according to Ogloff and Davis (2004) addresses offender strengths in the absence of a risk (e.g.,

antisocial peer associations) domain. Moreover, Ogloff and Davis suggested that implicit to the

responsivity principle is the notion of working with individual offenders with an understanding of their

individual risks and needs, including strengths.

22

Effective Treatment with Offenders

Several meta-analytic studies (e.g., Andrews et al., 1990; Izzo & Ross, 1990; Landenberger

& Lipsey, 2005; Lipsey, Chapman, & Landenberger, 2001; Lipsey & Wilson, 1998) have reported

favourable results in treating offender populations, specifically reduced rates of recidivism, where

structured CBT programs have been employed. Izzo and Ross (1990) conducted a meta-analysis of

46 studies of behaviour change programs with juvenile offenders ranging in age from 11 to 18 years

with a mean age of 13.2 years. The majority of studies analysed involved male offenders or mixed

gender groups of participants. Studies were initially classified according to their theoretical

conceptualisation of antisocial behaviour (i.e., social learning, behaviourist, cognitive-behavioural,

systems theory, reality therapy, and other). Consistent with findings from the vast majority of

psychotherapy outcome studies (see Hubble, Duncan & Miller, 1999 for review) Izzo and Ross found

no significant difference between theoretical models in terms of reducing recidivism. However, their

component analysis showed that programs that included a cognitive component (i.e, social skills

training, social problem solving, modelling, role-playing, and cognitive-behaviour modification) were

more than twice as effective as programs that did not. Further, regression analysis showed that two

variables significantly accounted for the variance in effect size: (1) the presence of a cognitive

component; and (2) a community treatment setting rather than an institutional setting. These two

factors have consistently been shown to produce greater effect sizes in treating both adult and

juvenile offenders (Andrews et al., 1990; Goldstein et al., 1998; Wilson, Bouffard, & MacKenzie,

2005). However, this is not to suggest that programs delivered in institutional settings cannot produce

significant change, despite limitations such as limited opportunity for rehearsal of newly acquired

prosocial skills and the omnipresent influence of an antisocial peer group.

23

Lipsey, Chapman and Landenberger (2001) reviewed 14 studies in which cognitive-

behavioural group treatments were employed with juvenile (aged 15-18 years) or adult (aged 20-30

years) offender populations. In nine of the studies reviewed, the intervention took place in a custodial

setting, whereas the remaining five were administered in a community setting while participants were

either on a parole or probation order. The results showed an overall reduction in recidivism of about

one third for CBT treatment participants compared to the treatment-as-usual (i.e., standard parole,

probation or prison case-management) control group. Specifically, those who received the CBT

treatment reoffended at a rate of 26 percent, compared to 38 percent for the control group.

Importantly, Lipsey and colleagues noted that the largest reductions in recidivism were associated

with ‘”demonstration programs’” implemented by researchers and administered by personnel with

high level training in CBT and an emphasis upon treatment integrity. The smaller effect sizes were

associated with programs conducted by lesser trained staff and in institutional settings.

To explore these and other variations in treatment effects, Landenberger and Lipsey (2005)

extended their 2001 research by examining potential moderating effects of CBT interventions with 58

juvenile and adult offenders. Three moderator variables were identified that explained a significant

proportion of the variance in recidivism rates: (a) the level (i.e., low, moderate or high) of re-offence

risk of participants, with higher levels of risk being associated with larger effect sizes; (b)

maintenance of treatment integrity; and (c) the inclusion or exclusion of specific treatment

components. Inclusion of anger control/self-regulation instruction and interpersonal problem solving

skill development in the treatment were associated with larger effect sizes. In contrast, smaller effect

sizes were found to be associated with the inclusion of victim impact and behaviour modification

treatment components. Additionally, results of the meta-analysis were consistent with previous

findings in showing the general CBT approach to produce a 50% greater reduction in recidivism over

24

and above that of the control group. Once the effects of the moderator variables were controlled for,

there were no differences in the effectiveness between specific, “brand name” CBT interventions

compared to a general CBT intervention. The authors concluded that the general CBT approach

appeared to have a positive effect upon adult and juvenile offender recidivism rates.

Summary of “What Works”

The risk-need-responsivity model of offender assessment and treatment is recognised as

having provided a clear and empirically supported framework for effective intervention with offenders

(Ogloff & Davis, 2004). Further refinement of this model has been recommended (Andrews &

Dowden, 2007; Bonta & Andrews, 2003) including increased recognition of the role of broader

universal and psychological needs of offenders (Ward & Stewart, 2003). With regard to specific

modes of intervention in working with adult and juvenile offenders, there is now a substantial body of

research demonstrating that the CBT approach is effective in reducing reoffending (Lipsey, 2009).

Importantly, the meta-analytic research indicates several key treatment components that contribute to

greater reductions in recidivism, including high-level training of therapists, maintenance of treatment

integrity and the inclusion of anger control/self-regulation training and interpersonal problem solving

skill development. Notwithstanding the importance of reductions in recidivism, several researchers

have indicated the need for offender treatment evaluation studies to include other relevant measures

of clinical importance in order to improve the depth of understanding relative to treatment

effectiveness (Losel, 2001; Serin et al., 2009; Tate et al., 1995).

In the next chapter, the relationship between operant and social learning theories are

discussed in relation to cognitive-behavioural conceptualisations of aggression, thus providing an

overview of the theoretical underpinnings of ART.

25

Chapter 3

Aggressive Behaviour: The theoretical context

This chapter examines the theoretical context of aggressive behaviour relative to cognitive-

behavioural conceptualisations, which underpin contemporary approaches to behaviour change

aimed at treating aggression. Firstly, a brief overview of Operant (Skinner, 1938, 1953) and Social

(Bandura, 1977) learning theories will be presented. These theoretical perspectives are described to

illustrate the lineage of cognitive-behavioural conceptualisation of aggression. Operant and social

learning theories have served to inform cognitive-behavioural theories of aggression which

represents the theoretical foundation of ART (Hollin, 2004).

Behaviourism: Operant learning

The behaviourist theory of operant learning (Skinner, 1938, 1953) proposed that behaviour is

contingent upon environmental consequences. Skinner (1938, p.138) formulated this theory in terms

of the three-term contingency model of behaviour- an Antecedent precedes a Behaviour, which in

turn produces a Consequence (i.e., ABC). Operant conditioning or learning was further elaborated by

Skinner with reference to two behavioural consequences that either increase (i.e., reinforcement) or

decrease (i.e., punishment) the frequency of a given behaviour. Two reinforcement contingencies

were proposed: (1) Positive reinforcement increases behaviour when the consequence of that

behaviour is experienced as rewarding (e.g., an aggressive youth threatens another youth with

violence if he does not hand over money. The money is promptly handed over and the aggressive

behaviour is positively reinforced); and (2) Negative reinforcement increases the likely repetition of a

behaviour through the removal of an aversive or unwanted consequence (e.g., an adolescent

26

becomes verbally abusive toward a parent when told she is grounded. The parent gives in and allows

the child to go out. The aggressive behaviour has been negatively reinforced by the removal of the

unwanted consequence (i.e., the grounding). In contrast, Skinner described two forms of punishment

that decrease the likelihood of a given behaviour: (1) Positive punishment occurs when a behaviour is

followed by an unwanted consequence (e.g., a youth commits a violent offence and is given a

custodial sentence by the courts); and (2) Negative punishment represents the removal of something

desirable following a behaviour (e.g., loss of privileges such as a parent grounding an adolescent). A

major criticism of operant learning theory has been its failure to provide detail of the role of internal,

cognitive processes and how they contribute to the learning and repetition of behaviour (Palmer,

2003). Bandura’s (1977) theory of social learning attempted to address the socio-cognitive aspects of

learning.

Social learning theory

Social learning theory (Bandura, 1977) was developed in response to the perceived

limitations of the behaviourist conceptualisation of learning theories, in which the acquisition of

behaviour was strictly viewed in terms of environmental consequences (Palmer, 2003). Bandura

attempted to explain behaviour by building on operant learning theories by including the internal,

cognitive process of observational learning. In essence, Bandura proposed that learning could occur

through observing the actions of another (i.e., modelling). The likelihood of the modelled behaviour

being repeated is dependent upon subsequent environmental consequences (i.e., reinforcement or

punishment). According to social learning theory, two main principles of motivation also play an

important role in maintaining behaviour: Vicarious learning operates through observing the

consequences of behaviour for others (optimally an esteemed other); and Self-reinforcement, which

27

refers to intrinsic motivation for behaviour implicit in self-validating thoughts and feelings (e.g., feeling

pride after completing a task and self-talk like “I did well”). Vicarious learning and self-reinforcement

serve to increase or decrease the repetition of a given behaviour (Bandura, 1977, 1986; Palmer,

2003).

Several researchers (Akers, 1999; Bandura, 1973; Nietzel, 1979) have applied social

learning principles to antisocial behaviour, arguing that salient models for aggressive and offending

behaviour include: family members; antisocial associates; and cultural influences (e.g., television,

film, videogames). In defining the interaction between the individual and environment, Goldstein

(1994) proposed a three-tier model of analysis for the geographical ecology of aggression: the

macrolevel provided an analysis of aggression at the international, national and regional level; the

mesolevel examined violence at the level of communities; and the microlevel analyses examined

violence as it occurs in the home. Goldstein et al. (1998) clearly view the development of aggressive

behaviour in terms of social learning theory, they wrote “Aggression is primarily learned behaviour,

learned by observation, imitation, direct experience and rehearsal” (p.3). While social learning theory

endeavored to take greater account of internal cognitive processes than earlier behaviourist learning

principles (Skinner, 1938; 1953) critics of the model argued that it did not go far enough in explaining

behavioural acquisition with respect to cognition (Palmer, 2003; Sestir & Bartholow, 2007). The next

section will examine cognitive-behavioural therapy (CBT) and its underlying theory which has

attempted to address this issue.

Cognitive-Behavioural Therapy (CBT) and It’s Underlying Theory

CBT draws heavily from the learning theories of behaviourism (Skinner, 1938, 1953) and the

increasing interest in the role of cognition pre-empted in social learning theory (Bandurra, 1977). In

28

particular, CBT with aggressive youth emphasises the importance of understanding the thoughts and

perceptions that lead to aggressive behaviour when a young person is faced with a perceived threat

or aggravation, actual or illusory (Lochman et al., 2000). According to Goldstein and Glick (1987)

interventions for antisocial and aggressive behaviour in young people need to address the cognitive,

behavioural and affective components of aggression to show even moderate treatment gains. CBT

interventions aim to do this through training in self-instruction and self-regulation techniques, social

problem-solving, affect labelling, social perspective taking (or moral reasoning), relaxation training,

and anger management skills. Figure 2 provides a useful representation of where various CBT

interventions sit within a continuum from purely cognitive to purely behaviourally oriented methods of

intervention.

29

Figure 2. A continuum of methods in behavioural and cognitive therapies.

*Source: McGuire (2001; p.216)

The continuum of cognitive versus behavioural methods shown in Figure 2 is a diagrammatic

representation of the cognitive-behavioural theory and methods that underlie ART. Typically CBT

based treatments employ an integrated multimodal approach in addressing the complexities of child,

adolescent and adult aggression (Fonagy & Kurtz, 2002; Kazdin, 1997; Lochman et al., 2000). The

theoretical basis of the CBT approach to understanding aggressive youth is based upon Novaco’s

(1979, 1994) conceptualisation of anger in adults, in which the relationship between precipitating

events and anger arousal is considered to be indirect. Rather, it is the individual’s cognitive

expectations and appraisals that serve to mediate an angry or aggressive response to a given

30

stimulus. Similarly, cognitive-behavioural considerations of child and adolescent development,

including maladaptive and antisocial behaviour, are framed in terms of the interaction between

cognitive, behavioural and interpersonal processes, as articulated in social information-processing

(SIP; Crick & Dodge, 1994; Dodge, Laird, Lochman, & Zelli, 2002) models. SIP models provide a

theoretical framework that specifies the mental steps involved in social-cognition. The model

proposes that in responding to social stimuli, an individual initially encodes cues, then makes mental

representations of those cues, which are associated with emotional states and goals, and

subsequently generates possible behavioural responses, evaluates potential responses, and finally

makes a decision and enacts the selected behaviour. SIP theorists propose that through social

learning processes including situational specific reinforcement, modelling, and efficacy expectancies,

children develop stable patterns of social-cognitive processing, personality-like traits that serve as

templates that inform social responses (Dodge et al., 2002).

Crick and Dodge (1994) proposed a non-linear, six-stage cognitive process involved in

children’s competent performance in social situations. In step 1, children encode aspects of the

external (i.e., an event) and internal (i.e., self-talk) cues. It is hypothesised that not all cues are

attended to but that social cues are selectively attended to. Step 2 involves interpretation of the

social cues, which may call upon mental representations stored in long-term memory. Other

interpretative processes include causal attributions, attribution of intent, and evaluation of goal

attainment. Self-evaluation occurs with respect to self-efficacy and outcome expectancies based

upon similar past experiences. All of these processes are influenced by social schemata, scripts and

social knowledge stored in long-term memory, which in turn can be altered or revised based upon

new experiences. Following encoding and interpretation of cues, Crick and Dodge proposed that in

Step 3, children enter into a process of goal clarification by way of selecting a desired outcome (e.g.,

31

keeping out of trouble or getting even). At Step 4, the child will extract from memory possible

responses to immediate social cues. Faced with a novel situation, the child may generate new

behaviour alternatives. Step 5 involves the evaluation of possible responses based on outcome

expectancies, self-efficacy beliefs, and evaluation of responses in terms of appropriateness. In Step

6, the selected response is behaviourally enacted. The entire mental process will begin again given

the external response of other players in the situation, which will feedback into the model at different

steps. The SIP model of behavioural acquisition imbues the individual with agency in actively

interpreting social information through cognitive processes (Sestir & Bartholow, 2007).

Dodge et al. (2002) reported strong evidence for the construct validity of the SIP model.

Using a sample comprising 387 children, from kindergarten to third grade, they analysed responses

to hypothetical vignettes that described social interactions involving either peer provocation or peer

group entry dilemmas. Children’s open ended responses designed to assess a hostile attribution

bias (i.e., a tendency to attribute hostile intent on the part of others) were rated in terms of intent-

“benign”, “hostile” or “don’t know”. A measure of social problem solving was coded for type of

response: aggressive (i.e., physical, verbal or threat); competent (i.e., socially appropriate); authority-

punish (i.e., appeals to have the provocateur punished); authority-intervene; passive-inept (i.e., a

non-assertive response); irrelevant other (i.e., a nonsense response); and unable to provide further

response. Children’s goal orientation- instrumental or pro-social was rated and emotional

understanding (personal and other) was also assessed. Several measures of parent and teacher

ratings of aggressive behaviour were also administered. Confirmatory factor analysis revealed

support for a partial mediation model of teacher rated aggression consistent with the SIP model.

Children’s emotional understanding (self and other) and specific social information processing

patterns (i.e., hostile attributions, hostile response generation, positive aggression evaluation, and

32

instrumental goal orientation) predicted aggressive behaviour in the classroom. Dodge and his

colleagues concluded that the support they found for a multi-dimensional information processing

model (i.e., SIP) as opposed to a uni-dimensional perceptive (e.g., social intelligence), has

implications with respect to treatment. Specifically interventions aimed at remediating aggressive

behaviour should include components that address the multiple cognitive processing deficits and

distortions (Dodge et al., 2002; Dodge & Pettit, 2003).

Summary

Social information processing theory, unlike its behavioural and social learning predecessors,

was specifically developed in order to address issues of prevention and/or intervention among

aggressive children and youth (Sestir & Bartholow, 2007). Cognitive-behavioural theory has drawn

extensively on these perspectives and ART represents a multimodal, cognitive-behavioural

intervention aimed at altering the behaviours of persistently aggressive youth (Hollin, 2004; Palmer,

2003). In the next chapter the content and procedures and theory and evidence pertaining to each of

the three cognitive-behavioral components of ART is elaborated.

33

Chapter 4

Aggression Replacement Training: A multimodal, cognitive-behavioural intervention

ART (Goldstein et al., 1998) was developed within a cognitive-behavioural theoretical

framework, comprising three CBT components that are delivered in three weekly group training

sessions (one meeting per week for each of the three components) over a 10-week period. The

sessions typically last for about one hour and are co-facilitated. (1) Skillstreaming (the behavioural

component) is designed to teach the young person effective social skills aimed at displacing

aggressive behaviour; (2) Anger Control Training (the emotional component) aims to reduce the

frequency of anger arousal and to teach techniques of self-control when anger is aroused; and (3)

Moral Reasoning Training (the cognitive/values component) is designed to “facilitate progress along

the natural stage-sequential trajectory of moral-cognitive development so that youths will make more

mature decisions in social situations” (Goldstein et al., p.107).

Skillstreaming

The origins of Skillstreaming as a psychoeducational intervention date back some 30-years. At

that time, Goldstein (1973) referred to the intervention as Structured Learning Therapy- a prescriptive

behavioural psychotherapy designed for working with adult psychiatric patients prior to

deinstitutionalisation. Skillstreaming, conceptualised within cognitive-behavioural theory, represents a

social skills training (SST) curriculum aimed at increasing the repertoire of prosocial behaviours of

aggressive children and youth through a psychological skills training approach (Goldstein et al.,

1998). Learning to resolve problematic or conflictual social situations through the development of

social and problem-solving skills represents a cornerstone of CBT interventions with aggressive

34

children and youth (Fonagy & Kurtz, 2002). In this section, the theory, evidence and practice of SST

and social problem solving will be examined with specific reference to the research pertaining to

aggressive youth and young offenders.

Skillstreaming: Content and procedure. Skillstreaming is a manualised cognitive-

behavioural treatment consisting of clearly defined, systematic procedures designed to ensure

treatment integrity (Goldstein et al., 1998). The Skillstreaming curriculum comprises nine sequential

steps: (1) Define the skill- initially this is undertaken by group participants as a brainstorming session

(e.g., “What does ‘keeping out of fights’ mean for the group”?), the group responses are then

reframed by the trainers into a succinct definition of the given skill. Additionally the meaning and

reasoning behind each step is discussed and defined as a thinking (i.e., cognitive) or action (i.e.,

behavioural) or both (e.g., listening) step; (2) Model the skill- trainers model successful use of the skill

for trainees; (3) Establish trainee skill need- here the trainers ask group members to identify when,

where and with whom they could use the skill (e.g., “When was the last time you could have used the

skill… “Keeping out of fights”?); (4) Select a role-player- the trainer selects a group member to role-

play their scenario for the group; (5) Set up the role-play- the main actor selects a co-actor to role-

play their scenario with them, the role-plays are determined by group members responses at Step 3;

(6) Conduct the role-play; (7) Provide performance feedback in a prescribed order (i.e., co-actor,

group members, co-trainer, main-trainer, and main-actor); (8) Select next role-player and repeat

steps 5-7; and (9) Assign skill homework- Skillstreaming practice sheet (see Appendix A). Table 1

shows the social skill- “keeping out of fights”, trainer notes, suggested content for trainer models, and

comments used in teaching the skill.

35

Table 1

Skillstreaming skill- "keeping out of fights"

KEEPING OUT OF FIGHTS Definition: Avoiding unnecessary conflicts by controlling angry feelings (conflicts can be physical and/or verbal). STEPS TRAINER NOTES 1. Stop and think about what triggered you to want to fight. (Think)

Stay calm. What event or action are you becoming angry about?

2. Decide what you want to happen in your future. (Think)

What will be the long-range outcome? Do you want to be treated with respect? What will increase your personal power?

3. Think about other ways to handle the situation besides fighting. (Think)

You might negotiate, stand up for your rights, ask for help, use humour, or pacify the person.

4. Decide on the best way to handle the situation and do it. (Think & Act)

Which of your options from step four is least likely to escalate the situation?

SUGGESTED CONTENT FOR MODELLING School or neighbourhood: Main actor tells classmate that he/she wants to talk out their differences instead of being pressured to fight. Main actor asks a neighbour to speak with their kids about throwing rocks at him/her rather than beating them up. Home: Main actor resolves potential fight with older sibling by asking parent to intervene. Main actor resolves an argument with a parent by agreeing to complete a task as requested, thus pacifying the parent. Peer group: Main actor goes for help when he/she sees peers fighting on school steps. Main actor decides to ignore the insulting remarks made by a peer who feels it is necessary to fight to “prove yourself.” Comments: Defining what a fight entails is generally useful when doing this skill. Many trainees see fights as merely physical, which is not the case. Prior to teaching or reviewing this skill, it is often useful to review or teach Using Self-Control. These are both excellent skills to use as reinforcement with the Self-Control Chain.

36

Ten core social skills are delivered over the 10-week Skillstreaming curriculum form the

standard ART curriculum (see Appendix A). While the skills and skill steps change on a weekly basis,

the procedures do not. It is important to note that the overarching goal of Skillstreaming is to prepare

participants for real-life pro-social social interaction. As such, the content of each session should

parallel interpersonal relationships that are relevant to youth. The models provided by trainers

therefore entail scenarios that depict social competence in youth-relevant relationships including

peers, family members, teachers and other significant agents of change in the young peoples lives,

this maximises the potential for transfer and maintenance of treatment gains (Goldstein, 2004).

Defining Social Competence, Social Skills and Social Problem Solving

According to Bierman and Welsh (2000) social competence represents a combination of

behavioural, cognitive, and affective skills that when well integrated enable an individual to apply

those combined skills adaptively in multiple social situations. Bierman and Welsh also make

reference to the notion that social competence is a subjective measure of an individual’s ability to

perform social tasks successfully within a given social milieu. The performance of a given social skill

and the capacity to socially problem solve are likely to be part of that assessment. Similarly, Cook et

al. (2008) argued that social competence represents the ability to successfully negotiate a wide array

of interpersonal interactions in a variety of social settings. They argued that social skills are learned

behaviours that are associated with the subjective measure of social competence.

One of the earliest definitions of the social skills construct was provided by Argyle and Kendon

(1967). They proposed three interrelated components that constitute socially skilled performance: (1)

social perception- refers to the ability to perceive and correctly interpret verbal and nonverbal social

cues. In the previously described social information processing (SIP; Crick & Dodge, 1994) model,

37

social perception represents the combination of Step 1- encoding and Step 2- interpretation; (2)

social cognition- following perception and interpretation of social cues, the individual makes a

decision regarding a suitable response. Here an amalgamation of SIP Steps 3- goal clarification, and

Step 4- Response access or construction drawn from memory schemas, and Step 5- response

decisions are performed; and (3) social performance- refers to directly observable social behaviour,

which is represented at Step 6- behavioural enactment of the SIP model.

In a comprehensive review, Gresham, Cook, Crews, and Kern (2004) attempted to resolve

disparities in defining the social skill construct. They examined and synthesised the SST literature

published from 1980 to 2004, in part to assess the construct validity of SST interventions. Based on

their findings, they concluded that social skills involved three main areas: (a) social interaction; (b)

prosocial behaviour; and (c) social-cognitive skills, all three of which form the basis of a SST

intervention. Furthermore, these domains are consistent with the construct validity reported for

behaviour rating scales such as the Social Skills Rating System (Gresham & Elliot, 1990) and the

School Social Behaviour Scales (Merrel, 2002) which are widely used measures in assessing child

and adolescent social skills.

Social problem solving was defined by D’Zurilla and Nezu (2001) as “the self-directed

cognitive-behavioural process by which a person attempts to identify or discover effective or adaptive

solutions for specific problems encountered in everyday living” (p.112). The ability to problem solve

within a social context, like the effective use of social skills, plays a key role in the construct of social

competence (McGuire, 2001). D’Zurilla and Goldfried (1971) conceptualised problem solving in terms

of a five-stage, sequential cognitive process: (1) problem orientation; (2) defining and formulating the

problem; (3) generating alternative solutions; (4) decision making; and (5) implementation of a

solution and verification of effectiveness. The incorporation of the cognitive steps delineated in

38

D’Zurilla and Goldfried’s model is paralleled in the SIP model later proposed by Crick and Dodge

(1994) as a way of understanding social decision making processes (McGuire, 2001).

Social problem solving training approaches, like social skills training, specify a series of

sequential steps in teaching social problem solving skills (McMurran, Egan, & Duggan, 2005). Those

steps broadly follow the problem solving model developed by D’Zurilla and Goldfried (1971). For

example, Spivack and Shure’s (1982) interpersonal cognitive problem-solving (ICPS) training teaches

six key ICPS skills to aggressive children and youth: (a) generation of alternative solutions; (b)

consideration of consequences of behaviour to self and other; (c) development of means-ends (i.e.,

sequential) thinking; (d) promotion of social-causal thinking (i.e., the impact of behaviour upon

others); (e) sensitivity to problems (i.e., knowledge of situational specific problems that can occur);

and (f) dynamic orientation (i.e., understanding that behavioural motivations may not always be

clear). While social problem solving training is not an explicit component of ART (Goldstein, Glick, &

Gibbs, 1998) the concepts and indeed some of the procedures are intrinsic to all components of the

program.

Social skills training, as the behavioural enactment of social competence, is more closely

aligned to a behavioural theoretical approach (see Figure 2, p.30). Whereas, social problem solving

training has greater association with cognitively oriented modes of behaviour change. While these

therapeutic methods are presented as distinct modalities of intervention, it is important to note that all

are variants of CBT and that in practical terms there is considerable overlap in the processes and

procedures of each of these modes of change (Gundersen & Svartdal, 2006; Longmore & Worrell,

2007). Moreover, as already indicated, both social skills and social problem-solving represent

interdependent components of the social competence construct with measurable improvement in one

suggesting improvement in the other (Cunliffe, 1992; McGuire, 2001).

39

Social Skills Training and Aggressive Youth

Social skill deficits and poor social competence are evident in the aetiology and maintenance

of numerous psychological problems of childhood and adolescence, including: depression (Reed,

1994; Reinecke & Ginsburg, 2008; Segrin, 2000); social anxiety (Beidel, Turner, Young, & Paulson,

2005); social phobia (Spence, 2000); Autism Spectrum Disorders (Bellini & Peters, 2008; White,

Keonig, & Scahill, 2007); ADHD (Fenstermacher, 2006; Pfiffner & McBurnett, 1997); and conduct

problems (Bullis, Walker, & Sprague, 2001; van Manen, Prins, & Emmelkamp, 2004). Similarly,

social problem-solving deficits among aggressive and antisocial youth have also been reported

(Cunliffe, 1992; Goldstein, 1999; Lochman et al., 2000). Compared to their socially competent peers,

antisocial youth have deficits in social competence that are a significant risk factor in the

maintenance of aggressive behaviours and poor psychosocial outcomes into adulthood (Moffitt et al.,

2002; Patterson, Reid, & Dishion, 1992). This is not to suggest, however, that all young offenders

consistently show deficits in social competence, or that the construct can be used to differentiate

offender from non-offender populations (McGuire, 2001; McMurran, 2005).

Gresham (1997) distinguished between social skill acquisition deficits (i.e., the child does not

possess the skill) and social skill performance deficits (i.e., the skill is known but poorly enacted).

SST programs work from the premise that aggressive children and youth show evidence of both

(Goldstein, 1999). Spence (2003) suggested that performance deficits among aggressive youth

might in part be attributable to behavioural contingencies that serve to reinforce antisocial/aggressive

behaviour. She wrote:

40

“For example, the conduct disordered adolescent may find it more effective, and may receive

more positive reinforcement from the deviant peer group, if they engage in physical violence

rather than appropriate conflict resolution skills” (p.86).

Hence SST interventions such as Skillstreaming generally employ principles of operant learning

(Skinner,1953) for contingency management (e.g., reward systems) and social learning (Bandura,

1977) in which three specific behaviour change strategies are used: (a) an exemplar models the

desired behaviour (modelling); (b) frequent opportunity for guided practice and rehearsal of the

desired behaviour (role-play) preferably across various settings (in-vivo); and (c) the modelling and

role-playing need to provide experiences of success in various settings (reinforcement) to ensure

transfer and maintenance of skill acquisition. Cook and his colleagues (2008) summarised four

common objectives of SST programs, including Skillstreaming: (1) promotion of prosocial skill

acquisition; (2) improving social competence through accurate skill performance; (3) reducing or

replacing antisocial behaviours; and (4) ensuring transfer and maintenance of social skills beyond the

treatment setting.

Support for the effectiveness of SST as stand alone intervention in treating various emotional

and behavioural disorders of childhood and adolescence is mixed (Cook et al., 2008). In particular,

the effectiveness of SST with secondary school aged youth has been questioned on the basis of the

widely held belief that once antisocial behaviour has become entrenched in adolescence, the

effectiveness of behaviour change programs is at best limited (Cook et al.). Contrary to this view,

Gresham, Cook, Crews, and Kern (2004) reported that 64% of students with emotional and/or

behavioural disorders (EBD) improved on measures of social skills at post-treatment follow-up,

compared to 35% in a no-treatment control condition. This meta-analysis of six meta-analyses

included 338 studies with more than 25,000 children and youth between the ages of 3-18 years.

41

In contrast, Quinn, Kavale, Mathur, Rutherford and Forness (1999) conducted a meta-

analysis of 35 studies published between 1981 and 1994. Studies were selected for inclusion on the

basis of three criteria. First, participants in each of the studies had to meet eligibility for special

education under the US Federal Government classification of emotionally disturbed. Second, the

focus of the research had to be an investigation of the effectiveness of an SST program. Third, the

data within each study had to allow for the calculation of an effect size. The average age of

participants was 11.53 years. The analysis of the pooled data revealed a small mean effect size of

.199. According to Gresham and his colleagues (2004) of the 35 studies included in the Quinn et al.

meta-analysis only two met the first of their inclusion criteria. Additionally, the Quinn et al. analysis

did not require the primary studies to have a control group, or specific research design (i.e.,

experimental or quasi-experimental). A further criticism was the inclusion of studies that employed

personality measures and an overemphasis on academic achievement as the main outcome

measure of improvement in social skills. Gresham et al. argued that these variables are not targets of

change within the SST construct, and thus the Quinn et al. meta-analysis is “tainted theoretically by

construct irrelevant variance” (Gresham et al., p. 39).

Drawing on the meta-analysis of Gresham et al. (2004) Cook et al. (2008) conducted a

“mega-analysis” (p.134) to examine the effectiveness of SST for secondary school aged students

either suffering or at risk of EBD (N=5000). The results were similar to that of the earlier study by

Gresham et al. Approximately two thirds of secondary students with EBD, compared to one third of

controls, showed evidence of improved social skills following SST. However, the inclusion of only

experimental or quasi-experimental research designs in this “mega-analysis” precluded an

assessment of SST effectiveness under “real world” conditions. Cook and his colleagues suggested

42

that future researchers should turn their attention to the evaluation of SST in school and community

based settings to determine effectiveness beyond the controlled experimental condition.

van Manen, Prins and Emmelkamp (2004) conducted an effectiveness study in which a

social-cognitive intervention program (SCIP) based on Crick and Dodge’s (1994) SIP model (n= 42)

and a stand alone SST intervention (n= 40), and a waitlist control group (n= 15) were compared on

several behavioural, affective, and cognitive outcome measures. Participants included 97 aggressive

boys, between 9- to 13- years, attending outpatient mental health clinics throughout the Netherlands.

All participants met DSM-IV (APA, 1994) criteria for CD or ODD, or Disruptive Behaviour Disorder-

Not Otherwise Specified. The SCIP included Spivack and Shure’s (1982) ICPS training and a social-

cognitive component designed to increase self-reflection and perspective taking abilities and reduce

self-centred cognitive distortions (drawing on the work of Selman, 1971, 1980) and an affective self-

control training component. Both the SST and social-cognitive treatment conditions were delivered

over 11-weeks in a group treatment format. Various parent, teacher and child self-report measures of

aggressive behaviour, self-control, social-cognitive skills, and social skills were administered at pre-

and post- treatment and one -year follow-up. Results showed a significant reduction in aggressive

behaviour and significant improvement across all other measures for both treatment conditions,

compared to no change for the waitlist control group. Additionally, there was a significant difference

between the SCIP and SST treatment conditions with the SCIP condition showing improvement on

more of the outcome measures at one-year follow-up. At one-year follow-up effect sizes of 0.56 for

the SST condition and 0.76 for the SCIP condition were reported. The greater effect size for the SCIP

treatment is not surprising given that it is in line with current treatment recommendations for conduct

disordered children and youth (Conduct Problems Prevention Research Group, 1999a, 1999b). The

SCIP represents a multi-modal treatment package similar to ART. Importantly, and in contrast to the

43

findings of Quinn et al. (1999) SST also showed a significant, moderate strength effect when

considered as a stand alone treatment. It was concluded that interventions with aggressive children

and youth that address both cognitive deficiencies and cognitive distortions such as the SCIP (and

indeed ART) show greater improvement after treatment than interventions that only address cognitive

deficiencies as with stand alone SST programs (van Manen et al., 2004).

In a review of the SST theory and research literature, Bullis, Walker and Sprague (2001)

identified several areas of weakness in the implementation and evaluation of SST programs. In

particular, they noted the relative lack of longitudinal research that might provide empirical support for

the maintenance and transfer of post treatment gains. In the absence of such data, it is not possible

to determine the extent to which SST programs provide a long-term preventative effect in reducing

aggressive and antisocial behaviour and the associated negative life outcomes (Cook et al., 2008).

Moreover, Cook etal. indicated that SST interventions needed to be provided within a

“comprehensive intervention approach” (p.82) for high risk antisocial youth. Again this

recommendation is consistent with the cognitive-behavioural multi-modal approach, which the

research findings clearly support over that of a single method approach (Conduct Problems

Prevention Research Group, 1999a, 1999b; Kazdin, 1987, 1997).

Social Problem-Solving and Aggressive Youth

As previously stated, ART does not include a specific social problem-solving component.

However, the cognitive-behavioural training procedures that are common to social problem-solving

treatment models, such as that previously described in Spivack and Shure’s (1982) interpersonal

cognitive problem-solving (ICPS) program, are consistent with, explicit and/or implicit, aspects of

program content across all three components of ART (i.e., Skillstreaming, Anger Control Training,

44

and Moral Reasoning Training). The social problem-solving evaluation research is, therefore, not

reviewed here. However, as this study was interested in evaluating potential changes in social

problem solving abilities following treatment, a social problem solving measure was included. Hence,

this section will briefly review the evidence pertaining to social problem-solving deficits among

aggressive and offender youth populations.

Numerous studies have identified various social-cognitive skill deficits and distortions

among aggressive children and youth (Dodge & Coie, 1987; Dodge & Pettit, 2003; Hollin, 2003;

Järvinen, 2001; Lansford et al., 2006). Specifically, compared to their non-aggressive age-related

peers, aggressive youth have been shown to: misattribute hostile intentions to others and fail to take

into account non-hostile cues, usually referred to as a hostile attribution bias (Crick & Dodge, 1994;

Dodge & Coie, 1987; Järvinen, 2001); select instrumental goals- such as winning a game, rather than

interpersonal goals- like the development or maintenance of friendship (Lansford et al., 2006; Slaby &

Guerra, 1988); generate fewer alternative solutions to social problems (Palmer & Hollin, 1999) and

rely on aggressive responses and a tendency to evaluate the outcomes of those responses in

favourable terms (Dodge, Greenberg, & Malone, 2008).

Slaby and Guerra (1988) examined the cognitive mediators of aggression among 144

participants (equal number of males and females) ranging in age from 15-18 years, with a mean age

of 17-years. They compared three groups of adolescents: (a) antisocial/violent juvenile offenders; (b)

high-aggressive secondary school students; and (c) low-aggressive secondary school students.

Three categories of cognitive content were assessed: (a) beliefs legitimising aggression; (b) beliefs

about the expected outcome for the aggressor; and (c) beliefs about the expected outcome for the

victim. It was predicted that, compared to the low-aggressive students, the aggressive offenders and

high-aggressive students would be more likely to hold beliefs that aggression increases self-esteem

45

and avoids a negative image among peers and that victims deserve what they get and don’t really

suffer anyway.

With respect to social-cognitive processes, Slaby and Guerra (1988) made hypotheses

relating to six information-processing components delineated in Crick and Dodge’s (1994) model.

Specifically, it was hypothesised that compared to low-aggressors, the aggressive offenders and

high-aggressors would: (a) seek less information; (b) attribute hostility to others and respond with

aggression; (c) generate fewer social solutions; (d) anticipate fewer consequences for an aggressive

response; and (e) select fewer “best” and “second best” solutions (i.e., “a solution judged by others to

be nonviolent and goal-directed”; Slaby & Guerra; p.581), which would less likely to be rated as

effective. The results indicated support for all of the hypotheses. A significant positive relationship

was also found between higher levels of aggression and cognitive content that endorsed aggressive

behaviour, which was inversely related to social problem solving abilities.

Järvinen and Pakaslahti (1999) reported similar findings in a study comparing the social

problem solving strategies employed by aggressive versus sociable children and youth in a seven-

year longitudinal, repeated measures study. A total of 47 aggressive children, 26 girls and 21 boys,

were assessed at baseline (10-11 years of age) via peer nominations for level of aggressiveness

(e.g., “Who starts fighting in a conflict situation in the school yard”?) versus non-aggressiveness (e.g.,

“Who never fights”?). The same participants were followed-up seven years later (17-18 years of age)

and again peer nominations were used to determine levels of aggression and a vignette based

measure of social problem solving was administered at both assessment points. Results revealed

overall stability of aggressive behaviour, although 11 participants moved from the non-aggressive to

the aggressive group and five female participants originally deemed aggressive moved into the non-

aggressive group. Moreover, stability of aggressive behaviour for both male and female participants

46

was predicted by aggressive childhood problem solving strategies and a lack of non-aggressive

solutions. A strength of the Järvinen and Pakaslahti study was the length of interval between initial

and follow-up assessments, which provided reliable, long-term evidence in support of the reported

association between social problem solving deficits and aggressive behaviour. However, a high

attrition rate from the original cohort (N=120) due to problems tracking original participants, made it

difficult to determine if the lack of males in the childhood aggressive group who subsequently might

have moved into the late-adolescent non-aggressive group is attributable to a real-world phenomena

or a biased sample. Järvinen and Pakaslahti concluded that their results were consistent with the

findings of other research (Crick & Dodge, 1994; Dodge & Coie, 1987; Slaby & Guerra, 1988) in

which teaching problem solving strategies to aggressive children and youth can mitigate aggressive

behaviour over time.

Summary

The research discussed in the previous sections clearly shows an association between

cognitive deficiencies (e.g., social skills and social problem solving) and cognitive distortions (e.g.,

hostile attribution bias and cognitive content advocating aggression) which have been implicated in

the onset and maintenance of aggressive behaviour (Gibbs, 2003; Goldstein et al., 1998; Lansford et

al., 2006; van Manen et al., 2004). As already noted, ART attempts to address both deficiencies and

distortions through the cognitive, behavioural and affective components of the program. In particular,

the aim of Skillstreaming is to provide frequent opportunity, through role-plays, for youth to practice

pro-social skill acquisition and non-aggressive social problem solving strategies. Cook et al. (2008)

indicated that the evidence in support of SST interventions has been mixed, particularly with respect

to youth in the mid- to late- adolescent age range. Indeed none of studies reported in this section

47

included participants above the age of 18-years. Given the age range of participants in the present

studies (i.e., 17 to 19.5 years for both pilot and main studies) there is an opportunity to address this

gap in the research literature relative to the effects of SST interventions with aggressive young

adults. In the next section, the theory and evidence pertaining to Anger Control Training, the affective

component of ART, is reviewed.

Anger Control Training

According to Goldstein et al. (1998) the intention of Anger Control Training (ACT) is twofold:

(a) to reduce the frequency of anger arousal and (b) to teach techniques of self-control when anger is

aroused. ACT, the affective component of ART, was originally developed by Feindler and Ecton

(1986) and later adapted by Goldstein and colleagues to fit within the multi-component intervention of

ART. Prior to Novaco’s cognitive-behavioural conceptualisation of anger arousal and subsequent

advances in approaches to treatment, the predominant approach in treating aggressive youth relied

on the provision of contingencies within a behaviour modification framework (Feindler & Baker,

2004). The use of contingencies within the controlled treatment setting usually saw initial reduction in

aggressive behaviour, however these gains frequently failed to transfer when the youth returned to

their natural environment. A natural environment which according to Goldstein et al. provided

frequent reinforcement of the notion that “might make’s right”. This is the earliest, most immature

stage of sociomoral development in which physical dominance is the basis for moral decision making

(this concept is further elaborated in the next section- Moral Reasoning). This coupled with a

behaviourist approach that failed to take full account of the cognitive and sociomoral components of

adolescent aggression was largely responsible for the apparent failure of transfer. In their approach

to treating adolescent anger problems, Feindler and Ecton drew upon Novaco’s (1975) adaption of

48

stress inoculation training (SIT) which Meichenbaum (1977) developed for the treatment of anxiety

disorders and impulsive children. In this section, the theory underpinning cognitive-behavioural

approaches to anger management interventions is outlined. The research concerned with evaluating

these interventions is reviewed, with specific reference to evaluations of ACT as a stand alone

intervention.

Anger control training: Content and procedure. Anger control techniques provide

instruction in self-regulatory skills that can serve to inhibit emotional and physiological arousal, in

anger provoking situations. This is achieved through a didactic, psycho-educational approach that

trains young people to become increasingly aware of the interaction between the cognitive,

physiological and behavioural components of their anger. ACT is typically presented in weekly,

group sessions. A minimum 10-week program is optimal, although programmes can be adapted to

individual treatment and extended time periods with chronically aggressive youth (Feindler &

Guttman, 1994).

Interpersonal problem-solving skills and verbalisation of cognitive self-statements are first

modelled for participants. Then participants perform role-plays of real-life anger provoking situations

in which anger reduction techniques, such as deep breathing, counting backwards and use of

pleasant imagery, are used. Participants learn to identify physiological cues (e.g., clenched fists,

shallow/rapid breathing) that represent the early indicators of anger arousal, as well as self-

statements that are likely to increase or maintain arousal (e.g., “He meant that”. “I’ll show him”). ACT

homework tasks include variations of Feindler and Guttman’s (1994) “Hassle Log”. The Hassle Log

is a cognitive-behavioural self-monitoring task specifically designed for monitoring adolescent anger.

It consists of a “what”’, “when”’, “where” and “‘with whom” format for monitoring anger, in addition to a

49

self-evaluation of behavioural responses and an anger rating scale. The content of the hassle log is

used in the ACT session as the material for the role-plays.

Although the content differs, the procedure in ACT is similar to that of Skillstreaming.

Goldstein et al. (1998) outlined nine sequential steps involved in ACT comprising the Anger Control

Chain. As shown in Figure 3, a new strategy is introduced into the Anger Control Chain each week,

so that the role-play scenarios build upon previous sessions.

Figure 3. The anger control chain.

As shown in Figure 3, the Anger Control Chain begins with teaching youth to identify two

types of Triggers- (a) external triggers- the anger arousing event; and (b) internal triggers- the

negative self-statements that increase and perpetuate anger arousal (e.g., “Who does he think he is?

I’ll show him”); Cues/Body Signs- physiological cues of anger; Anger Reducers- strategies aimed at

reducing anger arousal through the use of cognitive distracters, relaxation techniques, and buying

time to think through a non-aggressive response (i.e., deep breathing, backward counting, and

pleasant imagery); Reminders- short positive self-statements that are self-instructional, increase self-

50

control, and are designed to counter the negative self-statements at the stage of internal triggers

(e.g., “I’m in control”, “its not worth it”); Thinking Ahead- designed to teach consequential thinking in

broad terms including short-term, long-term, internal (i.e., “how will I feel”?), external (i.e.,

consequences imposed by an external agent), and social (e.g., loss of friends) consequences; Angry

Behaviour Cycle- up until this point the focus of ACT has been upon how participants should respond

to perceived provocation by others (i.e., external triggers). In week six, the Angry Behaviour Cycle

explores what participants do to initiate anger in others and how one person’s angry response can

often trigger a similar response in another and keep angry feelings going back and forth without

resolution; Social Skills- here Skillstreaming merges with ACT, in that participants select an

appropriate social skill to incorporate into the role-plays of the anger control chain; Self-evaluation-

teaches participants to realistically self-reflect on: (a) how well they have handled a conflict; (b) self-

reward for handling it well; and (c) critical self-evaluation of what could have done better (i.e., self-

coaching). The final two weeks of ACT are dedicated to the role-play rehearsal of the full sequence of

the Anger Control Chain.

Anger, Self-regulation and Aggression

Kassinove and Tafrate (2006) argued that confusion in distinguishing between the emotional

state of anger and related cognitive and behavioural constructs such as aggression, hostility and

violence has led to confusion among clinicians and researchers regarding the goals of treatment for

anger-related disorders. This, in turn, has contributed to ineffective treatment models that have

tended toward a hegemonic approach in which a specific aspect of the construct has been

emphasised (e.g., behavioural) over other important aspects (e.g., physiological or cognitive arousal).

Current theories of anger arousal are conceptualised in terms of a multidimensional construct

51

comprising affective, behavioural, cognitive and physiological aspects that affect both experience and

expression (Spielberger, Reheiser, & Sydeman, 1995). The definition of anger provided by Novaco

(1998) held significance for the interpersonal aspect of the emotional state- “a negatively toned

emotion subjectively experienced as an aroused state of antagonism toward someone or something

perceived to be the source of an aversive event” (p.13). Navaco’s (1975) model of anger as a

frequent, although not necessary, precursor to interpersonal violence, has been the predominant

influence in contemporary cognitive-behavioural conceptualisations of both functional and

dysfunctional manifestations of this common emotional state (Hollin & Bloxsom, 2007). Figure 4

provides a schematic representation of Novaco’s interactional model of anger in terms of its

determinants and consequences.

52

Figure 4. Determinants and concequences of anger.

*adapted from Hollin and Bloxsom (2007)

According to Novaco (1975, 1997) the subjective experience of anger represents a complex

interplay between environmental activation (i.e., perceived threat or verbal or physical provocation)

physiological arousal (e.g., increased heart rate, increased body temperature, muscular tension)

cognitive structures and processes (e.g., attributions, affect labelling, information processing,

schemas) subjective affective experience (i.e., intensity and frequency) and behavioural enactment

(e.g., physical and emotional withdrawal, physical or verbal aggression). Novaco’s multi-component

model indicates the potential for bi-directional change between the behavioural, cognitive and

physiological aspects of anger arousal, so that change in one aspect of the model can produce

53

reciprocal change in another. This is consistent with the aforementioned social information

processing formulations of aggressive behaviour that serve to inform contemporary approaches to

treatment (Hollin & Bloxsom, 2007).

In addition to the influence of Novaco’s (1975) model of anger, ACT has drawn extensively

on the work of early developmental theorists (Luria, 1961; Vygotsky, 1962) in which the role of self-

talk has been implicated in self-regulation and behavioural disinhibition. Self-talk, also known as

private speech, inner dialogue, self-statements, and inner speech, refers to the experience of talking

to oneself, initially through external/overt verbalisations and later internal/covert speech (Brinthaupt,

Hein, & Kramer, 2009). Both Luria and Vygotsky provided evidence that in early cognitive

development, children are reliant upon the verbal commands of adult caregivers for regulation of

behaviour. Subsequently, the child’s overt self-verbalisations increasingly provide opportunity for self-

regulation, usurping the need for external, usually parental, controls. Eventually the child’s use of

overt self-verbalisations gives way to covert, inner speech representing a critical developmental step

toward self-directed and self-motivated regulation of behaviour. Several studies (see Harris, 1990 for

review) have provided support for Vygotsky’s early developmental theories. A series of investigations

conducted by Luria showed some support for the effectiveness of interventions aimed at remediating

the consequences of failures in the development of internalised private speech among children

showing signs of hyperactivity and poor impulse control. However, Luria’s research conclusions have

been questioned on the basis of small samples and overreliance upon observational design, which

was poorly described and lacking in the detail required for replication (Harris, 1990).

The intervention procedures employed by Luria were later adapted by Meichenbaum and

Goodman (1971) in an investigation of the effects of cognitive self-instructional training as a means of

developing increased self-control among behaviourally disordered second grade children (N=15). The

54

training included a sequence of procedures similar to those employed in ACT: (a) the trainer models

task performance and self-instructs aloud while the child observes; (b) the child performs the same

task while the trainer instructs the child aloud; (c) the child performs the task again while self-

instructing aloud; (d) the child again performs the task while whispering self-instructional statements;

(e) and finally the child performs the task with covert, inner self-instruction (without whispering). The

results of their study revealed a significant improvement for the treatment group on measures of

cognitive performance and impulsivity compared to an attentional control group (i.e., cognitive

modelling without self-instructional training) and a no-treatment control group. The improvements

were maintained at one month follow-up. In a further study, Meichenbaum and Goodman reported

observed differences in the content of overt self-statements by children assessed as impulsive

compared to children assessed as reflective. According to their observations, the impulsive children

tended to employ immature and self-stimulatory self-statements, whereas the private speech of the

reflective children was observed to be significantly more self-guiding and instrumental in content with

greater use of inaudible verbalisations (i.e., private speech). With reference to the development of

internalised private speech and its relationship to anger control, Novaco wrote: “A basic premise is

that anger is fomented, maintained, and influenced by the self-statements that are made in

provocation situations” (p.17). Similarly, Bandura (2007) noted the significance of self-talk in the

acquisition of socio-cognitive skills among children, suggesting that “models verbalise aloud their

reasoning strategies as they engage in problem-solving activities. The thoughts guiding their

decisions and actions are thus made observable and acquirable” (p.57). Given that impulsivity is a

criminogenic risk factor (Andrews et al., 1990; Andrews & Dowden, 2007) the relationship between

impoverished private speech and impulsivity reported by Meichenbaum and Goodman has important

implications for interventions with offender populations.

55

Impulsivity and Aggression

The role of poor impulse control (i.e., impulsivity) has been well established as an important

risk factor in the development and maintenance of antisocial and aggressive behaviour (Barratt,

1994; Eysenck & Eysenck, 1977; Eysenck & McGurk, 1980; Moffitt, 1993). Moreover, the longitudinal

research conducted by Moffitt and her colleagues (1993; 2001; 2002) has implicated deficits in

impulse control in the development of the previously defined life course-persistent (LCP) offending.

As noted, impulsivity has also been identified as a dynamic risk factor (i.e., changeable) among

offender populations (Andrews, 1996). According to Andrews and his associates (Andrews, Bonta &

Hoge, 1990; Andrews et al., 1990) the dynamic risk factors should be the target of any treatment

where the overarching goal is to reduce recidivism (i.e, Need Principle). In defining impulsivity,

Eysenck and Eysenck (1977) described both pro-social and antisocial aspects of the construct. They

found that impulsiveness was related to the higher order personality factors of Psychoticism (i.e,

social non-conformity, characteristically associated with asocial pathological tendencies) and to a

lesser extent Extroversion (i.e., outgoing sociability, with a tendency to seek out novel situations).

Eysenck et al. (1985) later defined socially acceptable impulsivity as Venturesomeness, not unlike

Zuckerman’s (1979) conceptualisation of sensation-seeking, and socially unacceptable impulsivity

was labeled Impulsiveness. Similarly, Dickman (1990) reported a two-factor structure for

impulsiveness which he defined as either functional (i.e., Venturesomeness) or dysfunctional.

Functional impulsiveness was more closely aligned with extroversion, while dysfunctional

impulsiveness was found to have higher correlations with psychoticism.

Another line of research supporting the relationship between impulsivity and aggression has

been the typological study of reactive versus proactive aggression. Numerous studies (Crick &

56

Dodge, 1991; Dodge & Coie, 1987; Marcus & Kramer, 2001; Raine et al., 2006) have found support

for the broad distinction made between reactive and proactive aggressive types, each with differing

etiological and developmental trajectories. Reactive aggression has been defined as a “hot blooded”,

defensive or retaliatory response to a perceived threat and is associated with autonomic arousal and

emotional reactivity. In contrast, proactive aggression has been defined in terms of goal directed or

instrumental aggression with the intent to gain power over others in the interests of material gain

and/or dominance (Dodge et al., 1990; Dodge, 1991). There is some research to suggest that

physiological underarousal is associated with proactive aggression, whereas overarousal has been

associated with reactive aggression (Raine et al.; Scarpa & Raine, 1997). Although impulsivity has

been found to be significantly associated with both reactive and proactive aggressive subtypes, the

relationship has generally been found to be stronger with the reactive type (Felthous & Barratt, 2003;

Raine et al.).

Several studies (Carroll et al., 2006; Luengo, Carrillo-de-la-Peña, Otero, & Romero, 1994;

Smith & Waterman, 2006) have investigated differences on measures of impulsivity among offender

compared to non-offender populations. Luengo et al. found significant differences in impulsivity

between non-institutionalised delinquent and non-delinquent adolescents in a community sample.

They assessed a large (N=1,226) mixed gender sample of school aged adolescents- 12 to 18 years

(M=14.28 years; SD= 1.91) at two time points, with one year between assessments. Participants

were categorised as non-delinquent, minor-delinquent or major-delinquent depending on an offence

history. According to scores on the Esyenck Impulsivity Scale for children (Eysenck, Easting, &

Pearson, 1984) both the delinquent groups reported higher impulsivity at both time points and higher

impulsivity was associated with future risk of antisocial behaviour. The strongest correlations between

57

impulsivity and types of antisocial behaviour were associated with rule breaking, vandalism, and

aggression.

An Australian study conducted by Carroll et al. (2006) investigated differences in impulsivity

among 129 (114 males, 15 females) adolescents (M age = 15.52 years). Three equal sized (n = 43)

groups of early-onset, late-onset and non-offenders were compared on an abbreviated version of the

Impulsiveness Questionnaire (Eysenck & Eysenck, 1977) self-reported offence history, and several

tests of cognitive function. Results showed that all seven of the offence history variables (i.e., school

misdemeanors, stealing, soft- and hard- drug use; vehicle-related offenses, abuse of property, and

physical aggression) were positively associated with the two offender groups. Consistent with

Moffitt’s (1993, 2001, 2002) conceptualisation of LCP offenders, Carol et al. found that the early-

onset offender group reported significantly higher incidences of offending behaviours than the late-

onset group, except for soft-drug use and abuse of property. The offender groups also reported

significantly higher levels of delinquency than the non-offender group. Impulsivity also differed

between the offender and non-offender groups. Although no significant difference on impulsiveness

was found between the late and early onset offenders, both showed greater impulsivity than their

non-offending age-related peers.

Smith and Waterman (2006) examined self-reported aggressive behaviour and impulsivity

among a group of incarcerated violent and non-violent male (n=115) and female (n=133) offenders,

compared to male (n=114) and female (n=122) non-offenders in the UK. The age range of female

participants was 15-49 years and for males 18-49 years. The violent male and female offenders did

not differ significantly with respect to self-reported aggression and impulsivity scores. Additionally,

Smith and Waterman found a significant increasing linear trend in aggression and impulsivity from

non-offenders, to non-violent offenders, to violent offenders. As predicted, violent offenders reported

58

significantly higher levels of impulsivity than the other groups. Smith and Waterman concluded their

results showed further support for the contention that increased impulsivity can reliably distinguish

offender from non-offender populations and violent compared to non-violent offenders, although not

violent males compared to violent females.

Many of the impulsivity treatment outcome studies have been clinical trials of the effects of

psycho-stimulant medication on impulsivity related to childhood ADHD. Similarly, therapeutic

outcomes with respect to psychological interventions aimed at reducing impulsivity have tended to

report findings relative to school-aged children (Baer & Nietzel, 1991; Kendall & Finch, 1978;

Meichenbaum & Goodman, 1971) with or without a diagnosis of ADHD. Baer and Nietzel conducted

a meta-analysis comprising 36 primary outcome studies in which CBT treatments of childhood (M

age = 9.62 years) impulsivity was investigated. They reported that CBT interventions for impulsivity

produced an average post-treatment improvement of between one third to three quarters of a

standard deviation compared to non-treated or placebo-treated controls. However, impulsivity scores

for the treated children were close to the normative group average both before and after treatment.

Baer and Nietzel concluded that this finding may have been due to confounding the definitions of

impulsivity with externalizing behaviours in the primary studies, arguing that the two constructs may

or may not be correlated. In addition, they questioned the validity of the main measure of impulsivity

(i.e., Matching Familiar Figures Test; Kagan, 1966 as cited in Baer & Nietzel) and the reliability of the

normative data provided. Overall, the findings of this meta-analysis were inconclusive with respect to

determining the effectiveness of CBT interventions with impulsive children. Moreover the

generalisability of these results to aggressive offenders was limited.

In a study of young-adult university students, McMurran, Blair and Egan (2002) investigated

the relationship between aggression, impulsiveness, social problem solving, and alcohol use among

59

70 non-offender, British males (M age = 27 years). Contrary to expectation alcohol use did not

correlate with any of the other measures in the study. However, impulsivity was found to be

negatively related to social problem solving, and social problem solving was negatively related to

aggression. There was no direct relationship between impulsivity and aggression, rather the

relationship was found to be mediated by social problem solving. That is, heightened impulsivity

predicted social problem solving deficits, which in turn predicted increased aggression in this student

sample. McMurran et al. suggested that impulsivity in early childhood may act as a learning obstacle

with respect to the acquisition of social problem solving abilities. Later, social problem solving deficits

may contribute to an increased propensity to employ aggression in an attempt to resolve perceived

social difficulties. With respect to reducing aggression, McMurran et al. concluded that treatments

that teach social problem solving skills to aggressive youth are likely to prove beneficial. Equally,

interventions aimed at reducing impulsivity in early childhood, such as self-instructional training

(Meichenbaum & Goodman, 1971) might prevent social problem solving deficits thereby reducing the

likelihood of the development of later aggression.

One study that did examine impulsivity as a treatment outcome with aggressive offenders

was the first evaluation of ART conducted by Goldstein and Glick (1987). They reported a significant

reduction for worker-reported impulsivity in the treatment group (n=24) compared to a “brief

instruction control group” (n=24) and a no-treatment control (n=12). The ART-group, compared to

both control groups, was also found to have significantly fewer incidences of acting out behaviours,

as indicated by institutional incidence reports. These results were replicated for the 36 control group

participants who later completed the ART program. Despite the well established link between

impulsivity, delinquency and aggression, the empirical investigation of impulsivity as a treatment

outcome among offenders has been relatively lacking. Some evidence in support of the CBT

60

treatment procedures included in ART (e.g., self-instructional-, social skills-, and anger control-

training) has been provided in studies showing reduced impulsivity with children (Baer & Nietzel,

1991; Kendall & Finch, 1978; Meichenbaum & Goodman, 1971). However, there is a need for further

investigation of the effectiveness of these methods among older, aggressive offenders and the

purported relationship with increased impulsivity.

Evaluations of ACT

There are few published evaluations of ACT as a stand alone intervention. The two studies

reviewed here provide some evidence for effectiveness in treating aggressive adolescents.

Additionally, these studies provide some insight into the difficulties faced by program

facilitators/researchers in working in real-world environments. Issues such as maintenance of

treatment integrity and facilitator training and competency are important considerations in program

outcomes studies and can have significant implications in regard to outcomes. This was certainly the

case in the evaluation of ACT conducted by Nugent, Champlin and Wiinimaki (1997).

Nugent et al. (1997) evaluated ACT with 102 males aged between 12 and 18 years (M age =

14.7 years) living in a residential group home for adolescent males with chronic behavioural

problems. Of the total sample, nine living in one group home received ACT, while four in another

group home did not, and the remaining 89 participants formed a no-treatment comparison group

drawn from a previous study conducted by the researchers. Variations in length of custody meant

that participants in the treatment group received varying lengths of ACT. After controlling for the

covariate- length of time in custody, the results showed that the length of time in ACT was

significantly related to a decrease in externalising behaviours as rated on the Child Behavior

Checklist (CBCL; Achenbach, 1991) and case note reports of acting out behaviours. The behavioural

61

improvement for ACT participants was significantly greater compared to the adolescents in the non-

treatment group home and the matched control group of typical delinquent adolescents in custody.

The decrease in externalising behaviours for ACT participants paralleled the commencement of the

intervention in the group home. At the cessation of the intervention there was a significant increase in

acting-out behaviours.

According to Nugent et al. (1997) the ACT program finished at a time when one of the group

leaders became angry and verbally aggressive toward a participant in the presence of other

residents, which resulted in the termination of the employee and the ACT program. They noted that

the potential effect of adverse modelling by the group leader may have influenced the failure of

maintenance of treatment gains. Other threats to the validity of the study included the small sample

size in the ACT treatment group. In addition, only two participants were assessed at treatment-end

for incidence of acting-out behaviours, necessitating a single-case study design for the post-ACT

results. Additionally, it would seem that five ACT participants did not receive the full 10-week

treatment due to an open group format such that each time a new group member entered the group,

the program started again from week one. The use of a convenience sample within a residential

facility necessitated an open group format. Given the structured and sequential format of ART, this

represented a significant departure from the program structure described by Goldstein et al. (1998)

and a threat to treatment integrity. As previously discussed there is a necessary balance required

between the “what works” principles of program integrity and professional discretion. Inevitably the

demands of a given treatment setting require some flexibility in program structure and available

staffing in terms of facilitator training and experience. However, as already discussed, the meta-

analytic studies have shown that greater effect sizes have been associated with programs that

62

maintained treatment integrity and facilitators with high level training in CBT (Lipsey et al., 2001;

Landenberger & Lipsey, 2005).

A more robust evaluation of ACT was conducted by Sukhodolsky et al. (2009) in which 26

young people between 11 and 16 years of age (24-boys, 2-girls, mean age 12.7years) with Tourette’s

Syndrome and comorbid disruptive behaviour problems were randomly and equally assigned to ACT

or treatment-as-usual (TAU). Several clinician, parent and child report measures of anger and

externalising behaviours were administered to all participants at baseline and treatment-end (i.e., 10-

weeks later). At base-line, all participants showed high levels of non-compliance and just over half

showed explosive anger outbursts. At post-treatment follow-up, ACT participants showed significantly

greater reduction on these measures than the TAU, with improvement evident on both parent and

blinded clinician ratings. A large effect size (1.8) was reported for ACT on parent ratings of disruptive

behaviour. Similarly, the blind clinician ratings showed a 69% behavioural improvement for ACT

participants compared to 15% for the TAU control condition. Sukhodolsky et al. suggested caution in

generalising their findings given the small and homogenous sample in relation to ethnicity and socio-

economic status. Moreover only three participants met criteria for CD and none were outside the

normal range of scores on the CBCL (Achenbach, 1991) subscale delinquent behaviour, suggesting

that their results may not generalise to adolescents with more serious forms of conduct related

problems. A further limitation of Sukhodolsky et al’s. study was the lack of longitudinal follow-up

making it difficult to know the extent to which treatment gains were maintained over time and setting.

Summary

ACT has been developed within a sound developmental (Luria, 1961; Vygotsky, 1962) and

cognitive-behavioural (Meichenbaum & Goodman, 1971; Novaco, 1975, 1994) framework. As

63

previously discussed, there is a substantial body of research that has evaluated the broad CBT

approach in treating offenders (Andrews, et al., 1990; Izzo & Ross, 1990; Landenberger & Lipsey,

2005; Lipsey et al., 2001; Lipsey & Wilson, 1998). However, despite the often reported association

between impulsivity and offending behaviour (Andrews, 1996; Carroll et al., 2006; Luengo et al.,

1994; Smith & Waterman, 2006) there is a paucity of research in which impulsivity has been

investigated as a treatment outcome among offender populations. Thus, it is unclear if the treatment

procedures employed in ACT can effect changes in impulsivity. There are notably few outcome

studies in which ACT as a stand alone intervention has been specifically investigated. Moreover, the

aforementioned evaluations of ACT are limited in generalisability by a combination of several

methodological limitations including small sample sizes, simple pre- and post- treatment research

design without longitudinal follow-up, and in the case of Nugent et al. (1997) apparent threats to

treatment integrity. Notwithstanding, Hollin (2004) argued that anger management programs such as

ACT should be viewed as a component, as opposed to a stand alone intervention, in working toward

reducing aggressive behaviours and cognitions among violent offenders. The next section reviews

the theory and evidence for the third, and final, component of ART- Moral Reasoning Training.

Moral Reasoning Training

Explanations of antisocial and aggressive behaviour, like any human behaviour, are framed

within the context of a complex interplay between internal and external and individual and social

phenomena (Palmer, 2003). The research relevant to effective behaviour change with violent

offenders provides clear support for treatment programs that employ multiple modes of change (i.e.,

behavioural, affective, and cognitive) in targeting the complex set of variables associated with

aggressive behaviour (Andrews, 1995; Hollin, 2004; Hollin & Palmer, 2006). Potter, Gibbs and

64

Goldstein (2001) argued that effective treatment programs for antisocial youth must incorporate a

moral learning component. Proponents of this view contend that the frequently reported failure of

offender treatment programs to generalise beyond the treatment setting can in part be attributed to

the omission of program content aimed at increasing social perspective taking and social problem

solving abilities and reducing self-centred cognitive distortions that serve to maintain antisocial

behaviour (Gibbs, 2003; Goldstein, et al., 1998). In the previous sections of this chapter the

behavioural (Skillstreaming) and affective (Anger Control Training) components of ART were

explained in terms of their applied procedure and theoretical underpinnings. In this section the

cognitive/values component of ART (Moral Reasoning Training) will be examined.

Moral reasoning training: Content and procedure. Moral Reasoning Training aims to

“facilitate progress along the natural stage-sequential trajectory of moral-cognitive development so

that youths will make more mature decisions in social situations” (Goldstein et al., 1998; p. 107). Prior

to the moral reasoning session (i.e., at the end of Skillstreaming or Anger Control Training) group

participants read a short story or problem situation in which a protagonist is faced with a moral

dilemma (see Appendix A for example) followed by a series of closed-questions to which group

members provide responses. The problem situations are portrayed in various settings (e.g., home,

school, work, custodial facility) and are designed to facilitate a peer group discussion of moral values

such as honesty, individual and community responsibility, and prosocial decision making within peer

and family relationships and broader community and social interactions. The problem situations

provide role-taking opportunities including social perspective-taking while also stimulating discussion

on the role of cognitive distortions in morally immature decision making and subsequent adverse

consequences at the individual and social levels (Gibbs, 2004).

65

Goldstein and his colleagues (1998) described four phases of moral reasoning facilitation:

Phase 1: Introducing the problem situation- here the aim is to ensure that all group members clearly

understand the essence of the problem situation (e.g., “Who can tell the group what the main themes

in this problem are?” and “Why is that a problem?”). Also the relevance of the problem situation to the

everyday lives of group members needs to be made explicit (e.g., “Do problems like this really

happen?” or “Who has been in a situation like this?”); Phase 2: Cultivating mature morality- the aim of

this phase is to establish moral maturity as the group norm by initially attending to morally mature

reasoning among group members. This is achieved by listing prosocial decisions on a whiteboard

and relabelling the “should” as strong (e.g., “It takes courage to stand up for, and do, the right thing”);

Phase 3: Remediating moral developmental delay- having established moral maturity as the group

norm in the previous phase, the task of the facilitator turns to directly addressing the problem of moral

developmental delay. Group members who have responded in ways indicative of moral immaturity

are asked to explain their reasoning to the group. Morally mature group members are encouraged to

respectfully challenge antisocial decision making by indentifying implied thinking errors and creating

role-taking opportunities (e.g., “What happens to families/friends/communities in which everybody

only thinks about there own needs or wants?”); and finally Phase 4: Consolidating mature morality- at

this stage of the moral reasoning session, the morally immature group members, having heard the

mature group decisions, are invited to change their responses in order to come to a morally mature,

prosocial group consensus or at least a morally mature majority (e.g., “Does the group agree that not

stealing the car is the majority position?”). Finally the group facilitator reviews the morally mature

decision making of the group emphasising the strength and maturity in prosocial behaviour (e.g., “I’m

impressed with the groups ability to make strong decisions based upon sound and mature reasons).

Moral reasoning training is designed to remediate socio-moral developmental delay through

66

establishing a positive peer culture as the norm. It is not intended as opportunity for the group

facilitators to impose their own moral, religious or political beliefs upon young people. This is

achieved through the use of a Socratic dialogue, circular questioning and respectful challenging of

antisocial thinking errors (Gibbs, 2004; Goldstein et al., 1998).

Moral Reasoning: The theories of Piaget and Kohlberg

The earliest psychological study of moral reasoning from a cognitive-development

perspective can be traced to the work of Piaget (1932) who was interested in the cognitive processes

and structures of children’s intelligence, not just the content (Palmer, 2003). According to Piaget

(1952) children employ distinct reasoning structures that inform their thinking at four stages of

cognitive development: the sensorimotor stage (birth– 18-months)- the child relies on the physical

senses (e.g., taste, touch) to make sense of the world and his/her understanding of objects and

persons is limited to the physical attributes, the child is totally egocentric; the preoperational stage

(18-months– six years)- symbolic thought processes are evident through the emerging ability to

classify objects according to similarities and differences and hold mental representations of objects in

mind without needing to see or act upon the object. Perspective taking skills are not yet fully

developed, the child believes that his/her perspective is the same for others; the concrete operational

stage (six years– early adolescence) - here the child begins to develop the capacity for abstract

thinking and inductive reasoning. Beginning to demonstrate the capacity to generalise from a specific

circumstance or concept to broader rules of understanding, however the capacity for abstraction is

limited to concrete examples (Palmer, 2003). At this stage of cognitive development the pervasive

egocentricism characteristic of the earlier two stages shifts toward a developing sense of self as

being distinct from other (Gibbs, 2003); and the formal operational stage (early adolescence

67

onwards) - at this stage an individual develops abstract reasoning skills that allows for complex

problem solving in terms of possible consequences of a given action. The attainment of formal

operational thinking sees the individual develop the ability to plan into the future and consider multiple

perspectives, leaving behind the egocentric biases of earlier cognitive stages (Gibbs, 2003; Palmer,

2003).

Piaget (1932) proposed that children’s moral development is a corollary of their cognitive

development that can be seen in two stages of moral reciprocity: reciprocity as a fact- a “you scratch

my back, I’ll scratch yours” mentality (p.323); and reciprocity as an ideal- a more mature moral

perspective in which the early-adolescent begins to take account of psychological and circumstantial

motives for behaviour and moves toward a moral reasoning perspective more akin to “do unto others

as you would have them do unto you” (p.272). The major advance of Piaget’s conceptualisation of

moral development was his proposition that children actively construct their capacity for moral

reasoning, that they are not just passive recipients of social and parental moral values. According to

Piaget, it is through social interaction particularly age-related peers that moral development

progresses beyond egocentricism toward an increasing capacity to deduce the perspective of others

(Gibbs, 2003; Palmer, 2003).

Kohlberg (1969) expanded upon Piaget’s model of moral development by initially attempting

to replicate his findings and then expand them beyond adolescence into adulthood. Kohlberg, like

Piaget (1932, 1952) was interested in the development of moral reasoning in terms of the cognitive

structures and processes that lead to higher stages of development. Kohlberg (1969) proposed that

progression from one stage of moral development to the next was sequential and invariant. As

outlined in Table 2, Kohlberg’s (1969) proposed a six-stage, trichotomous system for understanding

moral development, each stage represents increasing complexity in terms of reasoning and

68

abstraction abilities. The stages in Kohlberg’s (1969) model are defined with respect to the structures

of social interaction at each level. These structures operate within the individual as a model or

schema for making moral decisions (Palmer, 2007). An important extension of Piaget’s work by

Kohlberg was the incorporation of social perspective-taking, at the individual and collective levels.

According to Kohlberg consideration of the processes and structures of moral development

necessitated examination of the qualitative differences in social perspective-taking at various stages

of the model (Palmer, 2007).

69

Table 2

Kohlberg's stages of moral reasoning

Note. adapted from Palmer (2003)

At the preconventional stage, rules and social norms are viewed as external to the self and

the individual has no perceived agency in the development of a moral code. Moral decision making is

based upon the individual’s relationship to authority figures such as parents, teachers and police and

the avoidance of punishment. With respect to criminal behaviour, offending is justified in terms of

“might makes right”, an entirely egocentric position in which the decision to commit a crime is based

upon perceived differences in power (i.e., Stage 1). At stage 2, the complete egocentricism of Stage

Level I: Preconventional reasoning

Stage 1 Moral reasoning is based upon avoiding punishment and obeying perceived authority figures

Stage 2 Moral reasoning is egocentric, with the person’s own needs being of greatest importance.

Reasoning is based on the perceived balance of rewards and punishment

Level II: Conventional Reasoning

Stage 3 Moral reasoning is determined by other peoples needs, with personal relationships assuming

importance

Stage 4 Moral reasoning is based on maintaining society’s rules and laws in order to maintain social

cohesion

Level III: Postconventional reasoning

Stage 5 Moral reasoning is underpinned by an understanding that society’s laws are a contract

between the individual and society. However under certain circumstances these laws can be

broken

Stage 6 Moral reasoning is determined by self-chosen ethical principles that are consistent over time

and situations, and these may over-rule society’s laws if they come into conflict with each

other

70

1 has progressed toward a greater capacity to take account of the motives and perspectives of

others. However, social perspective taking is limited to a pragmatic exchange- “you scratch my back,

I’ll scratch yours”. Offending behaviour is justified in terms of weighing up risk versus gain (e.g., “Will I

get caught?”) and pro-social decision making is primarily based upon deals- “if I do this, what’s in it

for me”. The conventional level of reasoning see’s the individual move toward greater comprehension

of the need for social norms and rules in order to maintain interpersonal and community relationships.

At Stage 3 reasoning, there is recognition of the need for mutuality in relationships based upon

building reciprocal trust and care, a “treat others as you would hope they would treat you”. Offending

behaviour at this stage can be justified with respect to maintaining a relationship. Gibbs (2004)

suggested that the risk at this stage of moral development can be an overemphasis on the opinions

of others to the extent that the individual becomes a “moral marshmallow” (p.54) whereby peer

influence leads to antisocial behaviour. While at Stage 3 there is evidence of mutuality in personal

relationships, Stage 4 represents these same principles applied for the betterment of the broader

community. At Stage 4, the individual begins to ask “How am I contributing to society?” and moral

decision making is characterised by a sense of fairness and equity for the good of the social whole.

Criminal behaviour is justified with respect to the greater interest of society, such as when a workers

union takes strike action in support of the rights of workers as a collective (Gibbs, 2004; Palmer,

2003).

The post conventional level (Stages 5 and 6) represents a transition from the internalisation

of societal expectations and rules (i.e., Conventional Reasoning) to a greater capacity for abstraction

and differentiation between the self and other (Colby, Kohlberg, Gibbs, & Lieberman, 1983). Here the

individual formulates a set of personal values, that are consistent with universal morals, however the

self-formulated values take precedence over societies. Thus the principles of the social contract

71

implied in Stages 3 and 4 are upheld in Stages 5 and 6 only where there is consistency between the

universal principles of justice, equality and dignity for all human beings. At the post conventional

level, offending behaviour can be justified where these basic principles of human rights are in conflict

with the law such as when a peaceful protester is charged with a civil order offense for participating in

a human rights protest (Palmer, 2003; 2005). It is noteworthy that in the results reported for

Kohlberg’s (1963) original child and adolescent sample, only 20% of the oldest age cohort (16-years)

showed evidence of Stage 5 moral reasoning and even fewer (5%) Stage 6. These higher stages of

moral development, particularly Stage 6, have been found to be so rare, outside of academic

philosophers, that many researchers have questioned their actual existence (Gibbs, 2003). Indeed in

later investigations of the model Colby et al. removed Stage 6 from the study design.

Colby et al. (1983) reported the findings of a 20-year, cross-sectional, longitudinal study of

moral development in which several of the assumptions underlying Kohlberg’s theory were tested.

Specifically, the invariant order of the stages was examined, as well as the relationship between

moral development and age, socio-economic status, and education. The sample comprised 58 boys

aged 10-, 13-, and 16- years of age at time-1, with five follow-up interviews conducted at 3-4 year

intervals. Participants were asked to provide justifications for their responses to nine different

hypothetical moral dilemmas. Their responses were then blind scored by researchers in terms of

Kohlberg’s moral stages. As predicted, participants progressed through the developmental stages in

the hypothesised, sequential order. Consistent with the theoretical tenets of a cognitive-

developmental model, the moral development of participants was also found to be significantly and

positively associated with age, socio-economic status, IQ and level of education. Four percent of

participants showed evidence of reversal to an earlier stage. This finding stands in contrast to earlier

results in which approximately 20% of the sample showed reversals (Kohlberg & Kramer, 1969). The

72

substantial reduction in the frequency of reversals was largely due to what Colby et al. (1983)

referred to as refinement in the scoring systems that provided for re-scoring of the participants

responses. Gibbs (2003) suggested that more recent advances in the measurement of moral

reasoning, where the potential for biases in scoring have been reduced, indicate that reversals are

common. According to Gibbs, people are likely to reason across various stages depending upon

situational variables, this represents a major departure from Kohlberg’s (1969) original invariance

proposition.

Moral Reasoning and Social Perspective Taking

Kohlberg (1969) like Piaget (1932) viewed moral development to be closely linked to

cognitive development and that higher stages of moral thought required the capacity to take the role

of another or social perspective taking (Palmer, 2007). Social perspective taking has been defined in

terms of the increasing capacity of children to “recognise, articulate, and coordinate the internal

states of others” (Marsh, Serafica, & Barenboim, 1980, p.140). There is general support for the

purported developmental sequence of social perspective taking that see’s a child develop beyond

early stages of egocentrism toward an increasing capacity to take account of differing thoughts and

feelings between self and other, although they frequently confuse the two, by about the age of six-

years (i.e., simple perspective taking). By the age of eight- or nine- years, the preadolescent is

capable of placing her/himself in the shoes of the other as distinct from the self and there is also

recognition that others can do the same. Sometimes referred to as third-party perspective taking, by

approximately 10-years of age a child is able to take into account multiple perspectives (i.e.,

simultaneous perspective taking) (Marsh et al., 1980; Selman, 1980). Table 3 shows Selman’s (1980)

73

stages of social perspective taking which like Kohlberg’s stages of moral development consists of

three levels, each comprising two stages.

Table 3

Selman's stages of perspective taking

Note. Adapted from Palmer (2007)

The first two stages of Selman’s model (preconventional reasoning) are characterised by an

egocentric bias. At Stage 1, the child is incapable of imagining the perspective of the other, their

Level I: Preconventional reasoning – concrete individual perspective- simple perspective taking

Stage 1 The individual has a self-centred view of the world, with little recognition of other people’s

perspective

Stage 2 Early recognition that others have their own view of the world, however the individuals

personal view takes precedence

Level II: Conventional Reasoning – member of society perspective- sequential perspective taking

Stage 3 The individual takes account of the perspectives of those in close association (e.g., family,

friends)

Stage 4 The individual has developed the capacity to distinguish between a societal perspective and

the perspective of the individuals within that society

Level III: Postconventional reasoning – prior to society perspective- simultaneous perspective taking

Stage 5

The individual has a perspective that acknowledges the rights of others regardless of a social

contract (i.e., prior to society). Differing perspectives are able to be taken into account and

combined in a rational way, although there is recognition that this may not always be possible

Stage 6 The individual takes a perspective that is derived from holding a consistent set of moral

principles

74

worldview is entirely self-centred. Stage 2 sees the early development of social perspective taking in

which the child can begin to appreciate that others have their own feelings and thoughts, however

there is a tendency toward minimisation of the other-perspective in favour of self- needs and wants.

By Stage 3, the all consuming egocentric biases give way partially to a greater concern for the

perspectives of others, although this decentration is limited to concern for the perspectives of

significant others such as family and friends. Stage 4 represents the development of a broader

understanding of social perspectives that encompasses the capacity to take account of differences

between individualistic and collective (social) perspectives. At the post-conventional level (Stages 5

and 6) the development of complex systems of perspective taking, whereby a consistent set of moral

values form the basis of perspective taking, is consolidated. Kohlberg (1976) suggested that Piaget’s

cognitive-developmental stages were integral to Selman’s (1971) stages of social perspective taking,

which were in turn a necessary, although not sufficient, precursor to his moral stages.

A study conducted by Walker (1980) provided some support for this proposition. He

hypothesised that the attainment of stage 3- conventional moral reasoning would require the

beginnings of Piaget’s (1932) cognitive-developmental stage- Formal Operational and Selman’s

(1971) Stage 3 of social perspective taking. Participants included 146 fourth- (M age = 9.9 years)

through seventh- (M age =13.0 years) grade children. Girls (n= 80) and boys (n= 66) were

approximately evenly distributed throughout the four grades. Several measures were administered to

assess Piaget’s cognitive developmental level, as well as a measure of social perspective taking and

Kohlberg’s moral judgement interview at pre- and post- test, and a six-week follow-up assessment. A

treatment and control group was compared on all measures. The treatment-group received a single,

45-minute session designed to stimulate moral development up to stage 3 through the use of moral

dilemma role-plays and discussion of potential stage 3 problem solutions. Findings supported the

75

hypothesis, and Kohlberg’s (1976) original supposition, that both cognitive and perspective-taking

development were prerequisites for moral development. Additionally, results showed that the moral

education program based on Kohlberg’s approach produced stable and generalised treatment gains

at six week follow-up. The treatment group showed an average increase of approximately one-half

stage in moral development compared to the no-treatment control group. Walker concluded that a

moral education curriculum that incorporated techniques aimed at increasing problem-solving and

perspective-taking skills of participants (as does ART) can affect commensurate improvements in

moral reasoning abilities.

Moral Reasoning and Offending Behaviour

Gibbs (2003) revised and expanded upon the work of Kohlberg by placing greater emphasis on

the role of deficits in social perspective taking and empathy and the use of self-serving cognitive

distortions in the development and maintenance of antisocial behaviour among youth. Gibbs, Potter,

Barriga, and Liau (1996) used the “three-D’s” to refer to the difficulties seen in antisocial youth: (a)

developmental delay in moral judgement; (b) self-serving cognitive distortions; (c) and social skill

deficiencies. According to Gibbs (2003) prosocial behaviour can in part be attributed to mature moral

perception and subsequent decision making relative to actual behaviour. Conversely, antisocial and

aggressive behaviour can be partly understood with respect to developmentally delayed or immature

morality.

Table 4 provides a summary of Gibbs’ theory of sociomoral reasoning, a reformulation of

Kohlberg’s (1969) model of moral development applied to antisocial and chronically aggressive

youth. Gibbs (2003) distinguished between immature versus mature sociomoral reasoning. He

employed the term sociomoral reasoning, as opposed to moral reasoning, in order to avoid pejorative

76

and simplistic connotations of right and wrong. The first two stages presented in Table 4 represent

immature reasoning that is characterised by superficial and egocentric moral judgements. Gibbs

argued that the developmentally delayed youth makes moral decisions based upon observable

features of the situation including physical strength and physical consequences (stage 1) or tit for tat,

pragmatic and reciprocal exchanges (stage 2).

Table 4

Gibbs' stages of sociomoral reasoning

Immature Sociomoral Reasoning

Stage 1: Unilateral and physicalistic- “might makes right”

Pervasive egocentric bias is evident. Sociomoral reasoning is based upon unilateral authority that

emphasis physical strength. Consequences are understood in absolute and inflexible terms, “all or nothing

thinking”

Stage 2: Exchanging and instrumental- “you scratch my back…”

The total egocentric bias of stage 1 gives way to sociomoral reasoning that takes account of social

interactions, however morality is still external to the self and superficial. Moral decision making is

understood in purely pragmatic exchanges- “what’s in for me?” or “eye for an eye”

Mature Sociomoral Reasoning

Stage 3: Mature and prosocial- “do unto other…”

Sociomoral reasoning moves beyond the individualistic and pragmatic concerns of stage 2 toward a

deeper consideration of interpersonal relationships (i.e., social empathy). A greater understanding and

willingness to adhere to prosocial norms in order to maintain social order and intrapersonal approval

Stage 4: Systemic and standard- “am I contributing?”

Here sociomoral reasoning incorporates mature social perspective taking in which complex social systems

are able to be considered. The development of conscience and social responsibility informs a sense of

fundamental human rights and social justice.

Note. Adapted from Palmer (2003, 2007)

77

A brief case example of immature reasoning is provided by Gibbs when he asked Joey a 15-

year old youth attending a specialist school:

“Why is it so important to obey the law?” I asked Joey. “Because, [pause], like in a store, you

may think no one sees you, but they could have cameras!” he replied. His other explanations

were generally similar: Keeping promises to other is important because if you don’t, they

might find out and get even; helping others is important in case you need a favour from them

later (Gibbs, 2003, p. 136).

The egocentric and overly pragmatic moral justifications made by Joey are consistent with the moral

reasoning of a child rather than those of mid-adolescence, by which time normal moral development

would predict egocentric decentration and an increasing capacity to take the perspective of others

(Gibbs, 2003; Palmer, 2007). There is a substantial body of research (Blasi, 1980; Nelson, Smith, &

Dodd, 1990; Palmer, 2003, 2005; Stams et al., 2006) supporting the relationship between

developmental delay in moral reasoning and offending behaviour. Specific to violent offending, the

ability to perspective take and undertake moral decision making has been shown to be arrested

among chronically aggressive youth (Gibbs, 2003; Goldstein, et al., 1998; Lochman, et al., 2000;

Palmer, 2005).

In an early narrative review of the literature, Blasi (1980) examined the relationship between

moral reasoning and several theoretically associated cognitive and behavioural outcome measures.

Blasi reviewed 15 studies in which delinquent and non-delinquent participants were compared, 12

studies comparing moral reasoning and moral action, 17 studies where the relationship between

moral reasoning and degree of honesty was assessed, and 11 studies where the relationship

between moral reasoning and resistance to social conformity was examined. Overall, the research

78

showed a significant relationship between moral thinking and moral behaviour, which was particularly

so in studies where delinquent and non-delinquent youth were compared. Of the 15 delinquent

versus non-delinquent studies compared, 10 showed the moral reasoning stage of the delinquent

youth to be significantly lower then their non-delinquent age-matched peers. Support was also found

for the hypothesis that higher stages of moral reasoning were associated with greater levels of

honesty and altruism. However, contrary to expectation, higher levels (i.e., postconventional) of moral

reasoning were not strongly associated with social non-conformity. Blasi identified discrepancies in

scoring procedures of moral reasoning measures and poor operationalisation of moral behaviour as

limiting the generalisability of the studies included in his review. Also, although extensive in terms of

the number of available studies reviewed and comprehensive within the narrative approach, Blasi’s

investigation did not provide quantitative analysis. Meta-analytic procedures conducted since have

provided substantial statistical support (i.e., aggregated effect sizes) for Blasi’ conclusions regarding

antisocial youth. For example, Nelson et al.’s (1990) meta-analysis included 15 published and

unpublished studies, which compared moral immaturity of delinquent and non-delinquent youth. The

total combined sample comprised 673 male and female participants, with an average age of 15.34-

years (age range 11-17 years). Results showed that the moral reasoning of the juvenile delinquents

was significantly lower (immature) than non-delinquent age-related peers. Nelson et al. were unable

to distinguish type of offence (violent, nonviolent) because the studies did not consistently provide

this information.

In an attempt to address several of the limitations identified in previous studies, Stams et al.

(2006) conducted a large scale (N = 4,814) meta-analysis comprising 50 studies in which groups of

delinquent (n = 2,316) and non-delinquent (n = 2,498) youth (age range 10-20 years) were compared

on moral reasoning and numerous moderator variables (e.g., type of offence, gender, diagnosed

79

psychopathic disorder, IQ, incarcerated versus non-incarcerated, length of incarcerated). Stams and

colleagues partially replicated the study conducted by Nelson et al. (1990) and also examined the

potential impact of the moderator variables on the relationship between moral immaturity and juvenile

delinquency. Consistent with the findings of Nelson et al., it was found that delinquent youth scored

significantly lower on moral reasoning compared to non- delinquent youth. Additionally, several of the

moderator variables were found to significantly influence moral reasoning scores, including: gender

(males were lower on moral reasoning than females); later adolescence was associated with larger

effect sizes than early or middle adolescence; delinquents with psychopathic personality traits

showed significantly lower moral judgement scores than non-psychopathic delinquents and even

greater moral immaturity when compared to non-delinquents. An institutionalisation effect was also

reported- incarcerated delinquents showed greater moral immaturity than non-incarcerated

delinquents and the longer the period of incarceration (18-months or more) the greater the

discrepancy in moral reasoning ability. Given the large sample size, Stams et al. concluded that the

statistical support provided for the relationship between delayed moral judgment and delinquency

should be considered robust. However, the results did not imply causation (i.e., low moral reasoning

causes offending behaviour) as all the studies included in the analysis were cross-sectional. Notably,

the relationship between stages of moral reasoning and moral behaviour has been reported to vary

depending on social context and purpose (Hains, 1984; Krebs & Denton, 2005).

Self-Serving Cognitive Distortions and Antisocial Youth

Delays in moral development and social perspective taking are not the only schemas

associated with antisocial and aggressive behaviour (Gibbs, 2003). Cognitive distortions have been

80

defined as “inaccurate or rationalising attitudes, thoughts, or beliefs concerning one’s own or others

social behaviour” (Liua, Barriga, & Gibbs, 1998, p. 337). The use of cognitive distortions has been

widely described in relation to various internalising problems such as depression and anxiety (Beck,

1996, 1967; Ellis, 1962) and externalising problems such as the criminal personality (Yochelson &

Samenow, 1976) and mechanisms for prosocial disengagement (Bandura, 1973) and avoidance of

cognitive dissonance (Gibbs, 2003) and cognitive processes that lead to the misinterpretation of

social information processing cues leading to aggressive acts (Dodge & Coie, 1987; Dodge,

Greenberg & Malone, 2008).

According to Gibbs (2003) and Goldstein et al. (1998) the developmental delays in moral

reasoning among antisocial youth are also associated with two cognitive distortions or thinking errors:

Primary Cognitive Distortion- this thinking error is characterised by a tendency among egocentric

young people to attribute far greater importance to one’s own views, thoughts feelings, needs and

wants at the expense of others legitimate views; and Secondary Cognitive Distortions which are

represented by three self-centred thinking errors that serve to reinforce the primary distortion: (1)

Blaming Others- the misattribution of responsibility for one’s own antisocial behaviours to sources

external to the self, including blaming the victim (e.g., “he asked for it”), blaming intoxication or a bad

mood, or blaming ones own history of victimisation; (2) Minimising/ Mislabelling- attempting to make

one’s antisocial behaviour seem less then, or different to what it really is, such as arguing that it did

not cause any real harm or even distorting antisocial actions as being admirable (e.g., “it was just a

joke”, “everybody lies its no big deal”) or labelling victims in derogatory, dehumanising terms; (3)

Assuming the Worst – is frequently expressed through attributing hostility to others as a

rationalisation for ones own aggressive behaviour or alluding to the misnomer that behaviour change

(self or others) is impossible (e.g., “I’ve always been this way”). The primary and secondary thinking

81

errors are thought to maintain antisocial and aggressive behaviour through rationalisations that serve

to prevent cognitive dissonance (i.e., feelings of guilt, shame, remorse) (Barriga, Gibbs, Potter, &

Liau, 2001). There is a substantial body of research in which the relationship between self-serving

cognitive distortions and antisocial behaviour among youth has been investigated (Palmer, 2007).

Barriga, Landau, Stinson, Liau and Gibbs (2000) conducted a study designed to investigate

the discriminative power of types of cognitive distortions (i.e., self-serving versus self-debasing)

relative to type of behavioural problems (i.e., internalising versus externalising) among female and

male youth. A group of incarcerated delinquents (M age= 15.8 years) was compared to a group of

non-delinquent High School students (M age = 16.4 years). Both groups were approximately matched

for socio-economic and ethnic similarity. Several cognitive and adolescent problem, self-report

measures were administered and misconduct reports for the delinquent youth were also examined

two months after initial assessment. The results showed that both self-serving and self-debasing

cognitive distortions were more prevalent among delinquent youth compared to non- delinquent

youth. The self-serving cognitive distortions were more often associated with externalising disorders

and self-debasing distortions were more associated with internalising disorders. Support for the

specificity of type of cognitive distortion by type of disorder was also found even after controlling for

comorbidity. Barriga and his collegues recommended that youth who miminise/mislabel their

antisocial behaviour require interventions aimed at increasing perspective taking skills (i.e., victim

empathy). While internalising youth who distort their behavioural consequences in terms of self-

blame and self-debasing distortions (e.g., catastrophising) are more likely to benefit from cognitive

restructuring methods that objectively examine the evidence and challenging irrational beliefs.

Liua, et al. (1998) investigated the relationship between self-serving cognitive distortions and

overt versus covert behavioural referents among a group of 52 male juvenile delinquents compared

82

to 51 High School, non-delinquents aged between 14- and 18- years. The How I Think (HIT)

Questionnaire (Barriga, et al., 2001 see Measures section for description) makes the broad

distinction between overt cognitive distortions involving direct confrontational antisocial behaviour

such as physical aggression (e.g., “people need to be roughed up once in while”) versus covert

cognitive distortions involving non-confrontational antisocial behaviours such as lying and stealing

(e.g., “If someone is careless enough to lose a wallet they deserve to have it stolen”). Similarly,

several researchers (Loeber, Lahey, & Thomas, 1991; Loeber & Schmaling, 1985; Patterson, 1982)

have found support for distinct dimensions of antisocial behaviour comprising overt (e.g., fighting,

arguing) versus covert (stealing, lying, fire setting) antisocial acts. Liua et al. found strong support for

their main hypothesis that overt versus covert cognitive distortions would be directly associated with

their respective behavioural (overt vs covert) referents. Additionally, there was a strong, positive

relationship between self-serving cognitive distortions and antisocial behaviour for both the

delinquent and non-delinquent groups. Both self-serving cognitive distortions and antisocial

behaviour were higher among the delinquent compared to the non-delinquent group. Path-analysis

revealed that overt cognitive distortions predicted overt behavioural referents (i.e., physical

aggression). By contrast, covert cognitive distortions predicted covert antisocial behaviour (e.g.,

stealing). It was concluded that the inclusion of cognitive restructuring techniques aimed at reducing

cognitive distortions among antisocial youth can serve to reduce both overt and covert antisocial

behaviour.

Barriga, Hawkins and Camelia (2008) extended the earlier research of Liua et al. (1998) by

partial replication and examining the cross-cultural validity of the earlier findings among a group of

239 males aged 10-19 years (mean age = 14.22 years) from schools on the Caribbean Island of

Curacao. The results were consistent with the findings of Liua et al. Self-serving cognitive distortions

83

were significantly associated with externalising behaviours and self-debasing distortions were

associated with internalising behaviours. Further analysis of participants self-reported aggressive

behaviours revealed specific associations between endorsement of cognitive distortion types and

type of aggression (i.e., physical or verbal aggression). Physically aggressive behaviour was

significantly related to cognitive distortions theoretically underpinning such behaviour, whereas verbal

aggression was associated cognitive distortions specifically related to oppositional-defiance. Contrary

to the argument that there is a generic criminal mind set (see Yochelson & Samenow, 1976) Barriga

et al. suggested that there is a tendency on the part of antisocial male youth to employ rationalising

or guilt neutralising cognitive distortions relative to specific behaviours. That is, there appears to be

group differences depending on type of offending behaviour.

Summary

In the previous section, the theories of moral development proposed by Piaget (1932) and

Kohlberg (1969) were reviewed in terms their influence on the development of the procedures and

content inherent in the Moral Reasoning component of ART. Several researchers (Blasi, 1980;

Nelson et al., 1990; Palmer, 2003, 2005; Stams et al., 2006) have reported evidence of sociomoral

developmental delay as a correlate of offending behaviour. Gibbs (2003) has emphasised the role of

deficits in Selman’s (1971) stages of social perspective taking and the use of self-serving cognitive

distortions among antisocial and aggressive youth. Persistently aggressive youth tend to lack

perspective taking and moral decision making abilities (Gibbs, 2003; Goldstein et al., 1998; Lochman

et al., 2000; Palmer, 2005). Consistent with developmental delay in moral reasoning, Goldstein et al.

(1998) described four cognitive distortions (Self-centred, Blaming Others, Assuming the Worst, and

Minimising/Mislabeling) that antisocial and aggressive youth frequently employ as way of rationalising

their harmful behaviour and avoiding cognitive dissonance. The procedures employed in Moral

84

Reasoning Training are designed to challenge these self-serving cognitive distortions and provide

frequent opportunity for social perspective taking. In the next section, previous evaluations of ART

will be reviewed.

85

86

Chapter 5

Previous ART Evaluations and Rationale for the Present Study

Previous Evaluations of ART

There are numerous published descriptions of ART and adaptations of the program in

various settings and among diverse populations (Goldstein, 2004). These include: school settings

(Amendola & Oliver, 2003; Gundersen & Svartdal, 2006; Roth & Striepling-Goldstein, 2003; Salmon,

2003) children and youth with Autism Spectrum Disorders (Moynahan, 2003) youth in residential care

(Barnoski, 2004; Nugent, Bruley, & Allem, 1999) young offenders (Goldstein & Glick, 1994;

Holmqvist, Hill, & Lang, 2009) adult offenders (Hatcher et al., 2008; McGuire & Clark, 2004) and

violent forensic psychiatric patients (Hornsveld, 2005; Hornsveld, Nijman, Hollin, & Kraaimaat, 2008).

Although some of these studies have included evaluation data, there are relatively few published

outcome studies of ART (Hornsveld et al.). Notwithstanding, several researchers (Fonagy & Kurtz,

2002; Goldstein, 2004; Palmer, 2007; Polaschek, 2006) have indicated that ART has demonstrated

itself to be one of the most efficacious interventions in working with aggressive youth. Improvement

in pre- to post-treatment follow-up measures of anger control, decreased frequency of acting-out

behaviours, increases in prosocial behaviours and significant reduction in reconviction rates have all

been reported (Goldstein, 2004).

Goldstein and Glick (1994) reported the findings of one of the earliest effectiveness studies

of ART in the treatment of a community based sample of post-release delinquents. Three groups

(i.e., “ART for youths and parents/family members”, “ART for youths only”, and “no-ART control

group”) were compared on a measure of “total skill change”. While there were no significant

87

differences between the ART groups, there were differences between these groups and the controls.

Compared to the no-ART control group, the two treatment groups showed an overall increase in

interpersonal skills and a decrease in self-reported anger in response to mild anger-provoking

situations. The ART for youths and parents/family members group were the least likely to be re-

arrested (15%) followed by the ART for youth only group (30%) while the no-ART control group were

most likely to be re-arrested (43%).

In order to assess transfer of treatment effects beyond the institutional setting, Goldstein and

Glick (1994) evaluated the carry-over of ART treatment gains post release back into the community.

Of the 54 youths released from a juvenile detention centre in New York, 17 had received ART while

37 had not. Parole Officers completed a global assessment measure on all 54 participants. On four of

the six psychosocial domains assessed (i.e., home & family, peer, legal, and overall, but not school

or work) ART recipients rated better than the non-ART group. However no statistical significance data

were reported.

Studies conducted with juvenile offenders have reported decreased recidivism rates for ART

program participants compared to no-ART or waitlist control groups. Barnoski (2004) conducted a

two-year longitudinal evaluation of ART by assigning 1,500 juvenile offenders to either an ART group

or a wait-list control group. Results from the 18-month follow-up showed a marked reduction in

recidivism (24%) for the ART participants compared to the control group. More recently, McGuire and

Clark (2004) adapted the ART youth program for adult inmates in the United Kingdom. ART is an

accredited program within the UK justice system. Reconviction rates for ART participants at a one

year follow-up showed a clear reduction (i.e., 20.4% compared to 34.5% for non-ART participants).

Hatcher and her colleagues (2008) extended the original study conducted by McGuire and

Clark (2004) by examining reconviction rates for ART completers compared to non-completers and a

88

no ART control group. Participants included 106 adult male, mostly violent, offenders placed on a

community-based order. Using a national offender database, matched pairs of research participants

were extracted on the basis of offence type, age, criminogenic risk scores, and number of previous

convictions. Participants were aged between 18 to 53 years, with a mean age of 27.42 years. Exactly

half were allocated to ART while the other half were allocated to the control group. At the 10-month

follow-up period, 47 of 106 research participants (44.34%) had been reconvicted, of those 38.3%

were reconvicted of violent offences while the remaining comprised various property offences. Just

over half (50.9%) of the no-ART control group were reconvicted, compared to 39.2% of the treatment

group. At 10-month follow-up, there was a 13.3% decrease in reconviction for the treatment group

compared to the control comparison. Comparison of those participants who completed ART

compared to matched non-completers revealed a 7.8% increase in reconviction rates for the non-

completer group. Hatcher et al. noted that the reconviction rates were favourable in comparison to

meta-analytic studies in which an average reduction of 10% in reconviction rates for offender

treatment groups compared to controls was reported. Hatcher et al. noted their small sample size

compared to other similar studies. It was further noted that as a violence offence-specific program,

ART cannot attract the numbers of participants that broad-based offending programs (e.g., cognitive

skills or problem-solving) do.

Gundersen and Svartdal (2006) reported the outcomes of a 24-session intervention based on

ART, delivered in school settings by teachers. Participants were 16 girls (mean age = 14.1 years) and

49 boys (mean age = 12.6 years). From pre- to post- treatment, there was significant improvement in

social skills for parent and teacher report but not for youth self-report. Parent and teacher ratings of

problem behaviours showed a significant reduction for the ART treatment group compared to the no-

ART control group. For the youth self-report of problem behaviours and cognitive distortions

89

associated with aggression, the treatment and control groups showed significant improvement. The

authors suggested that this result could partly be explained by a possible dilution effect between

treatment conditions as some participants were drawn from the same classrooms resulting in a

positive peer influence between members of the two conditions.

Aggression control therapy, described by Hornsveld (2004; 2005) represents an adaptation

of ART for working with violent forensic inpatients or outpatients. The program involves 15, 90-

minute, weekly sessions, with eight participants per group and is designed for adult forensic patients

with a diagnosis of either APD and/or a comorbid psychotic disorder (in remission) or for young adult

forensic patients with a diagnosis of CD or APD. The program consisted of all ART components, with

a self-regulation module added. Hornsveld (2005) reported the outcomes of two groups of male

forensic patients- 109 inpatients with a mean age of 32.5-years and 44 younger outpatients (mean

age = 23.4 years) all with a history of serious violent offending. Participants completed a battery of

self-report measures designed to assess personality traits, hostility, aggressive behaviour, social

skills, and social anxiety at commencement of the treatment, treatment-end, and 15-week post-

treatment follow-up. Results showed a significant reduction in self-reported hostility and aggressive

behaviour at treatment-end with those gains maintained at post-treatment follow-up. There was no

significant difference on scores for social anxiety or social skills. In explaining these non-significant

findings, Hornsveld noted that compared to Dutch norms, the program participants reported above

average scores on aggressive behaviour and below average social anxiety and greater social

competence at intake assessment. This apparent inflated sense of social competence and lack of

social anxiety might be also partly explained by elevated narcissism that some researchers have

identified as typical of the Cluster B personality disorders including APD (Hare, Hart, & Harpur, 1991)

and violent offending (Baumeister, Smart, & Boden, 1996; Bushman & Baumeister, 2002).

90

Rationale for the Present Study

The ART evaluation literature to date has shown considerable promise in terms of

generalisability across time and setting in working with aggressive youth (Hollin, 2003). However,

none of the previous evaluations of ART have included a separate analysis of the different

components of the intervention, making it difficult to determine the relative impact of each component

(Palmer, 2007). Furthermore, a limitation of the ART evaluation studies among offender groups has

been the overreliance upon recidivism as the sole or main measure of outcome. As previously

discussed, several researchers have voiced criticism of offender outcome studies for this reason

(Serin et al., 2009; Tate et al., 1995). It is argued that while recidivism represents an important

treatment outcome among offender populations, other more psychologically orientated variables

provide important information in regard to clinically relevant outcomes. The present study aimed to

address these potential limitations by including psychological measures that are designed to assess

the cognitive, behavioural and affective targets of change in ART (i.e., aggressive behaviours and

thoughts, aggression-related cognitive distortions, social problem solving, and impulsivity).

Notably, the majority of ART evaluation research has consisted of participants in mid-

adolescence or older offenders. To date, no known study has investigated the effectiveness of ART

among young adults (e.g., 18-20 years). The present study aimed to assess the effectiveness of ART

in treating young-adult, Australian, male offenders. Further, to date, there has been no published

evaluation ART in Australia. Indeed, there is a relative absence of published research in which

treatment programs for aggressive youth, particularly violent juvenile offender programs, have been

empirically investigated in Australia (Boni, 1999; Howells & Day, 2002). Ogloff (2002) commented

that “…there is a very real need for Australian-based research in offender rehabilitation. Very real

91

questions exist concerning the applicability of international offender rehabilitation research to

Australian society” (p.246). As such, another important aim of the studies comprising this thesis was

to determine the applicability of an American developed program in terms of its cultural relevance to

Australian youth.

Several researchers have noted the unique difficulties that both researchers and clinicians

can encounter in implementing and evaluating programs in correctional settings, including

institutional resistance to outsiders (Fleck, Thompson, & Narroway, 2001; Hollin, 2001b; Tate et al.,

1995; Wakai, Shelton, Trestman, & Kesten, 2009). These authors have noted the importance of

piloting a program in order to ascertain and find solutions to potential obstacles before entering into a

major research project. Hence, an initial pilot study (see Chapter 5, next) aimed to identify potential

research, institutional and programmatic obstacles before undertaking the main study. Furthermore,

the pilot study provided an opportunity to determine the appropriateness of the measures employed

in a youth justice custodial setting. Although the selected measures have well established

psychometric properties in community settings, there is relatively little, if anything known about their

reliability and validity in a custodial setting.

92

Chapter 6

Pilot Study

An extended version of this pilot study has been published previously (Currie, Wood,

Williams, & Bates, 2009). As previously, discussed the main aims of the pilot study were: (a) to

determine the appropriateness of the measures employed in a youth justice custodial setting; and (b)

investigate the overall effectiveness of ART in simple pre- and post- treatment research design

before undertaking a larger outcome evaluation. To this end, three specific hypotheses were made,

from pre-treatment to treatment-end it was predicted that participants would-

1. report an overall reduction in aggressive thoughts and behaviours;

2. report fewer self-serving, aggression-related cognitive distortions; and

3. report an increase in pro-social skills.

Method

Participants

Six young men (17-18 years) initially agreed to participate in the pilot study. All had

committed violence-related offences and were serving a custodial sentence at a Victorian state

Juvenile Justice Centre. In week three of the 10-week program, one participant self-selected out of

the group and the research. The remaining five participants completed the 10-week ART program.

Measures

Aggressive behaviours and thoughts were assessed using the Aggression Questionnaire

(AQ; Buss & Warren, 2000) which is designed to assess aggressive tendencies among children and

93

adults. The instrument is written at a third-grade reading level, and comprises 34-items providing

scores on five subscales: Physical Aggression (e.g., “I may hit someone if he or she provokes me”);

Verbal Aggression (e.g., “My friends say that I argue a lot”); Anger (e.g., “At times I get very angry for

no good reason”); Hostility (e.g., “At times I fell like I’ve gotten a raw deal out of life”); and Indirect

Aggression (e.g., “I sometimes spread gossip about people I don’t like”). Respondents rate each

item on a 5-point scale from “Not at all like me” to “Completely like me”. Higher scores indicate

higher levels of aggression. Scores can be classified in terms of severity based on their percentile

rank according to standardized norms. Buss and Warren provided seven classifications- “low” (2nd–

14th percentile), “low average” (15th–27th percentile), “average” (28th–71st percentile), “high

average” (72nd–81st percentile), “high” (82nd–97th percentile), and “very high” (98th percentile and

above). The AQ was standardized in a large (N = 2,038) community based sample and reported

reliability (Cronbach’s alpha) is moderate to high for the subscales, with alphas ranging from .71 to

.88. Internal consistency for the total scale score was very high (alpha = .94).

The Social Skills Rating System (SSRS; Gresham & Elliot, 1990) was used to measure

social skills. In the present study, the self-report, Secondary Student Form was used, comprising 34

items that are rated on a 3-point scale (“Never”, “Sometimes” and “Very Often”) across four

subscales: Cooperation (e.g., “I use my free time in a good way”) Assertion (e.g., “I make friends

easily”) Empathy (e.g., “I say nice things to others when they have done well”) and Self-control (e.g.,

“I ignore other children when they tease me or call me names”). Gresham and Elliot reported

reliability coefficients ranging from adequate to high (i.e., alpha coefficients from .67 to .77 for the four

subscales, and .83 for the total scale score).

Cognitive distortions were assessed using the How I Think Questionnaire (HIT: Barriga, et

al., 2001). This is a 54-item self-report measure designed to assess four categories of self-serving

94

cognitive distortions (i.e., Thinking Errors): Self-Centered (“When I get mad, I don’t care who gets

hurt”) Blaming Others (“If I made a mistake its because I got mixed up with the wrong crowd”)

Minimizing/Mislabelling (Everybody lies, its no big deal”) and Assuming the Worst (“Its no use trying

to stay out of fights”). These four cognitive distortions represent inaccurate or biased ways of

perceiving and interpreting environmental stimulus, which have been shown to play a central role in

antisocial behaviour and criminogenic thinking (Gibbs, 2003). Items comprising first-person

statements are rated on a 6-point scale ranging from “Agree Strongly” to “Disagree Strongly”. The

HIT can be administered in groups or with individuals. It is typically completed in 5 to 15 minutes and

requires only a fourth-grade reading level. Percentile rankings based on the normative sample

provide for three qualitative severity descriptors- non-clinical (<50th – 72nd percentile), borderline-

clinical (74th – 82nd percentile), and clinical (84th – 100th percentile). Moderate to high internal

consistency was reported, with Cronbach’s alphas ranging from .63 to .92 for the cognitive distortion

subscales and the behavioural referent subscales (i.e., Lying, Stealing, Oppositional Defiant, and

Physical Aggression). Estimates of internal consistency were very high for the total scale score

(alpha = .95).

Treatment integrity was assessed on an ongoing basis via video footage using the

Washington State Aggression Replacement Training Quality Assurance (QA) Form (Hayes,

2001). This tool assesses adherence to the ART treatment model across multiple process (e.g., “Was

a positive climate established through welcoming students?”) and content (e.g., “Did most youth

complete the [homework]?”) domains, relative to each of the program components (i.e.,

Skillstreaming, Anger Control Training, and Moral Reasoning Training). A categorical (yes/no)

checklist format is used, with different weightings for each of the items assessed. For each of the

three program components, treatment adherence scores range from 0-100 for the main-trainer and 0-

95

9 for the co-trainer, higher scores indicate greater treatment adherence. The role of the main trainer

is emphasised in the QA assessment, which is reflected in the allocation of scores for each trainer.

Detailed qualitative corrective feedback is also provided. All scores are combined to provide an

overall competency ranking (i.e., 0 = “Not Competent”, 1 = “Borderline Competent”, 2 = “Competent”,

and 3 = “Highly Competent”) for each of the three ART program components (see Appendix C for

example of a completed QA Form).

Procedure

Prior to commencing this pilot study, approval was sought and granted from relevant Human

Research Ethics committees (i.e., Department of Community Services Victoria and Swinburne

University of Technology see Appendix B). Initial referral to ART was made by the health workers

(i.e., psychologists or social workers) who undertook a case-management role of the young people

for the duration of their custodial sentence. The health workers determined level of criminogenic risk

(moderate to high) for inclusion in the ART program. Risk was assessed using the Victorian

Offending Needs Indicator for Youth (VONIY; Department of Human Services Victoria- Youth Justice,

2004). The VONIY provides scores across several domains of ciminogenic risk and need in addition

to a score (low-moderate- high-intensive) relative to level of intervention required. To date, there has

been no published psychometric evaluation of this assessment tool. The Principal Researcher then

met with each young person individually to explain ART and the expectations and requirements for

participation in the program. The plain language statement and consent form (see Appendix B) was

read to potential participants and explanation offered for any terms or content requiring clarification.

Voluntary participation was emphasized and it was explained that participation in ART was not

96

conditional upon participation in the research. After a “cooling off” period of two days, all of the initial

six referrals agreed to participate in the program and the research.

Screening for history of substance use and previous psychiatric diagnoses was conducted

during the initial intake assessment. Exclusion criteria included current or recent (previous 6-months)

history of psychotic symptoms. None of the participants reported a history of psychosis. The self-

report measures were administered one-on-one, which also provided an opportunity to assess the

reading capability of each young person before entry into the program. All participants appeared to

have at least adequate reading, writing and comprehension skills required for participation in ART.

The program commenced in the week following the intake assessments.

The program was facilitated by the present author (provisional psychologist) and a female

colleague (Master of Social Work). Only the Principal Researcher had received accredited training in

ART, so he main-trained for the duration of the 10-week program, while the co-trainer assisted. Both

facilitators had previous experience in cognitive-behavioural group therapy programs. It is important

to note that in the interests of maintaining treatment integrity and providing a forum for clinical

supervision relative to ART both facilitators participated in regular supervision. For all groups, the

participants gave informed consent for the ART sessions to be videotaped in weeks three and eight.

These videotapes were then sent to an experienced ART Master Trainer in Washington State, USA.

After viewing the sessions written feedback was provided using the ART QA Form (see Appendix C).

Additionally, the co-facilitators attended fortnightly, in person, supervision with a clinical supervisor.

97

Results

Statistical analysis was conducted using SPSS version-14. Several researchers (e.g., Cohen,

1994; Kazdin, 1999; Kendall, Marrs-Garcia, Nath, & Sheldrick, 1999) have been critical of the

arbitrary cut off of p<.05 in determining statistical significance, particularly in the case of

psychotherapy outcome studies. Hence, as this pilot study was exploratory we set a liberal p-value of

.10 in determining statistical significance.

Table 5 shows case by case, pre- to post- treatment raw scores and indication of clinical

versus non-clinical scores for all participants on the three total scale score outcome measures.

Table 5

Pre- and post- treatment raw scores on the AQ, SSRS, and HIT Questionnaires (N=5)

Case

AQ

Pre Post

SSRS

Pre Post

HIT

Pre Post

A 105* 108* 50.0 50.0 3.64* 4.33*

B 96 89 42.0 48.0 3.23* 3.54*

C 113* 77 47.0 49.0 3.38* 3.11*

D 129* 99 61.0 63.0 4.31* 3.79*

E 90 80 66.0 75.0 2.41 1.92

Note. AQ= Aggression Questionnaire; SSRS= Social Skills Rating System; HIT= How I Think

Questionnaire; * = Score within clinical range.

98

As shown in Table 5, participants raw scores for the total scale score on the AQ showed a downward

trend from pre- to post- treatment in all but one case, A, who reported a slight increase in overall

aggression at post- treatment. An improvement in social skills was evident at post-treatment, with

scores on the SSRS slightly higher at post-treatment, except for case A, who reported no change.

For all cases, except “E”, participants remained within the clinical range for cognitive distortions at

post-treatment on the HIT. The HIT scores for cases A and B showed an increase in cognitive

distortions at post-treatment, while the remaining three cases reported a decrease at post-treatment.

Given the small sample size (N=5) a one-tailed, Wilcoxon signed-ranks test was employed to

test for significant differences between pre-and post- treatment scores on the outcome measures.

Table 6 shows means, standard deviations, and Wilcoxon signed-ranks T-values for the AQ, SSRS,

and HIT. As predicted, a comparison of mean aggression scores showed a significant overall

reduction from pre- to post- treatment (p = .06). The AQ subscales of Anger (p = .06) and Hostility (p

= .03) also showed a significant reduction at post-treatment assessment. As predicted, results

revealed a significant increase in scores on the Self-Control subscale (p = .03) and the total scale

score (p = .06) on the SSRS from pre- to post- treatment. Contrary to expectation, there was no

significant difference between pre- and post- treatment scores on the total scale score for the HIT (p

= .50) or any of its subscales.

99

Table 6

Pilot study pre- and post- treatment descriptives and Wilcoxon T-values for all outcome measures

(N=5)

Pre-Treat Post-Treat

Variable M (SD) M SD

Wilcoxon

T value

AQ Total 106.60 (15.27) 90.60 (12.97) 1.0*

- Physical 31.00 (4.85) 26.60 (6.80) 1.0

- Anger 22.20 (2..05) 19.60 (2..07) 0.0*

- Hostility 22.00 (8.15) 16.00 (3.94) 0.0*

- Indirect 17.00 (5.29) 14.80 (3.90) 2..0

- Verbal 14.40 (2.30) 13.60 (1.34) 1.0

SSRS Total 53.20 9.98 57.00 11.77 0.0*

- Cooperation 13.20 1.48 13.20 4.66 2..0

- Assertion 15.80 1.92 16.40 1.82 1.0

- Self-Control 9.20 4.32 11.80 4.15 0.0*

- Empathy 15.00 3.54 15.60 3.13 2..0

HIT Total 3.39 .69 3.33 .91 2..0

- Assuming the Worst 3.16 .79 3.22 .85 2..0

- Blaming Others 3.46 .87 3.46 1.03 2..0

- Minimizing/Mislabeling 3.69 .75 3.56 .90 2..0

- Self-Centered 3.27 .62 3.13 .92 1.0

Note. *p<.10; AQ= Aggression Questionnaire; SSRS= Social Skills Rating System; HIT= How I

Think Questionnaire

In order to illustrate qualitative change, a single case study- The case of “C” is presented in

Appendix D.

100

Discussion

The results supported the predicted reduction in overall aggression and increased pro-social

skills from pre- to post- treatment. However, contrary to expectation there was no change in cognitive

distortions associated with antisocial and aggressive behaviour.

With respect to the predicted change in aggressive thoughts and behaviours, there was a

significant reduction in scores on the Anger (i.e., irritability, easily frustrated, affective lability) and

Hostility (i.e., feelings of bitterness, paranoia, social isolation) subscales of the AQ, which are thought

to represent an “internalising approach to anger” (Buss & Warren, 2000, p.15). According to Buss

and Warren, high scorers on the Anger subscale are most likely to respond to CBT interventions

aimed at identifying common external triggers, internal physiological cues, and techniques in arousal

reduction such as cognitive distraction and relaxation techniques. These modes of change are

consistent with the Anger Control Training component of ART. High scorers on the Hostility scale are

thought to be responsive to treatments aimed at challenging the consistent and erroneous hostile

attributions made toward others (Buss & Warren, 2000). As previously discussed, improving social

problem solving skills represents an implicit aspect of the social skills training component of ART (i.e.,

Skillstreaming). By teaching the substitution of social skills in place of aggression and the breaking

down of those skills into sequential skill steps, ART participants learn to become aware of the

perspectives of others, increase their capacity to interpret social cues, and rehearse different social

solutions to potentially anger arousing situations. Overall, the reduction in Anger and Hostility scores

suggested a change in cognitive appraisals of anger provoking situations and increased capacity to

employ self-control techniques when anger is aroused.

Results also revealed a significant increase in Self-Control on the SSRS (Gesham & Elliot,

101

1990). According to Gresham and Elliot, this subscale provides a measure of non-aggressive

behavioural responses employed when faced with conflict situations, such as being teased, and in

situations requiring prosocial skills such as taking turns or compromising. This finding is consistent

with an increase in participants repertoire of pro-social skills taught in Skillstreaming and the self-

regulatory techniques acquired through ACT. While these positive effects are promising, the lack of

change in participants cognitive distortions from pre- to post- treatment needs to be considered in the

context of the theoretical underpinnings of CBT, and specifically the rationale for including the moral

reasoning component in ART.

Several researchers (e.g., Hayes, Strosahl, Bunting, Twohig, & Wilson, 2004; Longmore &

Worrell, 2007; Maruna & Mann, 2006) have questioned the value of cognitive interventions in

ameliorating symptoms associated with psychopathology. Hayes and his colleagues argued that

symptom reduction in CBT is often evident “before procedures thought to be central to its success

have been implemented” (p.15). They further argued that behaviour change is not necessarily

dependent upon cognitive change. In a review of CBT component analysis studies, Longmore and

Worrell found no significant difference in effectiveness between the cognitive and behavioral

elements of CBT in all but one of the 13 studies reviewed. Consistent with Hayes (2004) it was

concluded that cognitive interventions appear to add nothing over and above that of the behavioural

components of CBT. However, as Longmore and Worrell pointed out, it may be the combined multi-

modal packaging of CBT that makes it effective. In other words, an intervention employing any CBT

component may have implications for change in other components. However, the component

analysis research reviewed by Longmore and Worrell was limited to studies of depression and

anxiety. There are differences in the cognitions associated with these disorders and those thought to

be associated with the conduct related disorders and indeed differences in treatment approach.

102

Moreover, as noted earlier, the results from several meta-analytic studies (Andrews et al., 1990; Izzo

& Ross, 1990; Landenberger & Lipsey, 2005; Wilson et al., 2005) have found that inclusion of a

cognitive component has consistently been shown to produce greater effect sizes in treating both

adult and juvenile offenders.

As described previously, the moral reasoning component of ART employs these techniques

in challenging cognitive distortions specifically relevant to aggressive youth. Palmer (2007) noted that

while ART outcome research indicates overall effectiveness, the exact contribution of each module

cannot be determined until component analysis is undertaken. Specifically, she questioned if the

moral reasoning component provides any added value over and above that of typical anger

management programs that do not include this cognitive/values component. Gundersen and Svartdal

(2006) attempted to analyse the relative effectiveness of the different components of ART, and found

a significant improvement in prosocial thinking (i.e., a reduction in cognitive distortions as assessed

by the HIT (Barriga et al., 2001) for both the treatment and control conditions. They explained the

finding in terms of a potential dilution effect between the two conditions as both groups of children

continued to interact in the school/classroom setting. Notably, Gundersen and Svartdal only delivered

four to five sessions of moral reasoning as part of the overall ART treatment and still found a positive

effect. Consistent with Longmore and Worrell (2007) they suggested that training youth in any one of

the components (Skillstreaming- behavioural; Anger Control Training- affective; and Moral

Reasoning- cognitive/values) appears to generalise across others.

Some methodological and sample differences between the present study and Gundersen

and Svartdal (2006) also offer some explanation for the different outcomes on the HIT (Barriga et al.,

2001). The Gundersen and Svartdal study was conducted in a school setting, with male (n = 49) and

female (n = 16) participants, with a combined average age of 13.4 years. In contrast, this pilot study

103

was conducted in a custodial setting, with five males between the ages of 17-18 years, who had

committed serious violent offenses. The differences in the degree and persistence of behavioural

disorder between the two samples are also evident in their respective pre to post mean scores on the

HIT. The treatment group in the Gundersen and Svartdal study was just within the clinical range of

scores at pre-treatment (M=3.03) while in the present study participants were well within the clinical

range (M=3.39). Given their age, history of aggressive behaviour and early incarceration it is

reasonable to hypothesise that the ART participants would fit within the childhood-onset or LCP

developmental pathway for antisocial behaviour (Frick, 2004; Moffitt et al., 2002). Certainly, they

would seem to be further along in the developmental trajectory of the LCP subgroup than the

participants described in the Gundersen and Svartdal study and more entrenched in the cognitive

distortions associated with criminogenic thinking. As already noted, the HIT (Barriga, et al., 2001) is a

specific measure of cognitive distortions associated with antisocial and criminal behaviour. Hence the

present results might reflect a lack of cognitive change specific to these distortions (i.e., Self-centred,

Blaming Others, Assuming the Worst, and Minimising/Mislabelling) rather than cognition broadly. The

SSRS (Gresham & Elliot, 1990) and the AQ (Buss & Warren, 2000) are not only measures of

behavioural activation, both include items designed to assess cognition, and pre- to post- change

was found in the expected direction on these measures.

Methodological and Programmatic Issues

With respect to the exploratory aim of the pilot study to evaluate the cultural relevance of

ART in an Australian cultural context, the Moral Reasoning problem situations were written in a

language specific to American youth. Feedback given by participants at the conclusion of the pilot

program indicated the need to re-write the problems in a vernacular more familiar to Australian youth.

104

Therefore, expressions such as “Midtown Bar and Grill” became “local pub”, “AWOL” became

“escaped”, “going steady” became “going out with”, “nark” became “dob”, “play ball” became “play

footy” etc. Other changes to language were also made to some of the Skillstreaming and ACT

participant handouts. Apart from these minor adaptations, the feedback from participants in this pilot

study suggested that the ART program content was culturally relevant and applicable to aggressive

Australian youth.

At the time of commencing the pilot study, the Victorian Juvenile Justice Review (Day,

Howells, & Rickwood, 2003) had made recommendations regarding the implementation of evidence-

based programs, including ART, across both community and custodial youth justice services in the

State of Victoria. Several researchers have noted that the implementation of a new program and its

evaluation in a correctional setting can be fraught with difficulty (e.g., Fleck et al., 2001; Hollin, 2001b;

Tate et al., 1995; Wakai et al., 2009). Thus, another aim of the pilot study was to provide an

opportunity for the early identification of potential institutional and programmatic obstacles in

preparation for the main study. Apart from program/research participants, the stakeholders in a

custodial setting are numerous and their investment, or lack thereof, can have major implications to

the success of a program (Fleck et al., 2001; Wakai et al., 2009). Several initial hurdles to

implementation had to be overcome throughout the pilot period, including logistical problems such as

timetabling the three ART sessions per week within existing therapeutic, educational, and vocational

programming, and finding a suitable and available group-work space. Clarification of the research

design, particularly the inclusion/exclusion criteria, with the Health Services team who were

responsible for the overall case management of each youth and hence the initial referrals to ART was

also required. By far the most challenging and ongoing problems were related to communication

between the Centre staff and the program/research facilitators. Although the facilitators ensured

105

regular communication with the Health Services team, it became clear that on a daily basis the unit

staff, who undertook an individual key-worker role with the youth, had the most frequent contact with

them. As such, by week five, a written outline of the weekly program content was distributed on a

weekly basis and regular verbal progress reports were also given to the unit staff. These procedures

are consistent with recommendations made by Goldstein et al. (1998) relative to ART implementation

and enhancement of transfer of treatment gains. Goldstein et al. referred to these potential agents of

change as “Transfer Coaches”.

As discussed previously, Andrews et al. (1990) described Program Integrity (i.e., the in built

processes for monitoring treatment fidelity) as one of the five core principles in effective offender

treatment. Several meta-analyses (Andrews & Dowden, 2006; Landenberger & Lipsey, 2005; Lipsey

et al., 2001) provided strong evidence for the importance of the integrity principle. Similarly, in the

largest evaluation of ART to date, Barnoski (2004) found that treatment integrity and the related

concept of facilitator competency significantly influenced ART outcomes. Where an ART trainer was

ranked as “not competent” on the Washington State ART QA measure (see Measures section above)

recidivism data at the 18-month follow up was found to be slightly higher, although not significantly

different to that of the no-treatment control group. In contrast, “competent” or “highly competent”

delivery of ART was associated with a 24% reduction in recidivism over and above that of the control

group this difference was significant. Treatment integrity for the pilot study was assessed in sessions

3 and 8 by sending the video recordings of these sessions to the Washington State QA Specialist.

The facilitators were ranked as “highly competent” on the QA measure (see Appendix C).

106

Limitations and Summary of Findings

A limitation of the pilot study was the sole use of self-report outcome measures. A number of

researchers (e.g., Kazdin & Weisz, 1998; Moffitt et al., 2008; Posavac & Carey, 2003) have indicated

the need for a multi-informant approach in evaluation research as important criteria for identifying

effective treatments. Moreover, subsequent to the pilot study, it was concluded that the

aforementioned key-worker role undertaken by the unit staff provided an opportunity to assess

potential behavioural and cognitive change of ART participants through a suitable non-self-report

measure. The addition of such a measure in the main study offered a broader, multi-informant

approach to evaluating the effectiveness of ART. Several other methodological changes were

implemented at the conclusion of the pilot study. Table 7 provides a summary and rationale for those

changes relevant to the key learning’s from the pilot study. Each of these changes are elaborated

upon in the next chapter which describes the research design and results of the main study, which

followed by a discussion of the combined pilot and main study findings.

The present results provide some support for the efficacy of ART in reducing self-reported

aggression and increasing social skills among Australian youth justice clients serving a custodial

sentence for violence or violence related offences. However, the findings also revealed a lack of

change specific to cognitive distortions associated with aggressive and antisocial behaviour. Of

particular relevance to the findings of this study and the subsequent main study of ART was the

conclusions made by several researchers (Beck, 1996; Longmore & Worrell, 2007; Maruna & Mann,

2006) that the overall effectiveness of CBT interventions might lie in the interaction effects of the

components or as is the case with ART, a multimodal approach. Moreover, the distinction between

cognitive, behavioural and affective components in CBT might represent more of an arbitrary, theory

107

driven construct rather than the actual practice of the model in an applied setting. In conclusion, the

findings of this small, pilot study warrant further investigation with a larger sample, over an extended

period so that generalisability and maintenance of treatment gains can be assessed beyond the

treatment setting and over time.

108

Table 7

Methodological changes in the main study: Summary and rationale

Change Rationale

Inclusion of waitlist control group Allow for statistical comparison between treatment group

and waitlist control at T2

Upper age range increased to 22-years of

age

As recommended by Health Services staff at the

custody centre increase in upper age limit in line with

centre’s upper age limit and increased potential subject

pool

Removal of SSRS Outside revised upper age range

Inclusion of ABCL Provide for multi-informants and reliable/valid multi-

factorial behavioural outcome measure

Inclusion of SPSI-R:S Outcome measure relevant to reported social problem

solving deficits identified as a dynamic risk factor among

offenders

Inclusion of Impulsivity measure Outcome measure relevant to reported increased

impulsivity identified as a dynamic risk factor among

offenders

Change of language employed in some

moral reasoning problems

Replacement of American-centric language to culturally

relevant Australian youth vernacular

Note. T2 = Time-2 assessment; SSRS = Social Skills Rating System; ABCL = Adult Behaviour Checklist; SPSI-R:S = Social Problem Solving Inventory- Revised: Short Form

109

Chapter 7

Main Study

Aims and Hypotheses

This chapter presents the method and results for the main study, which included the key

learning’s from the pilot study (see Table 7). In addition to extending the research done in the pilot

study, there were several other aims specific to the main study. As previously discussed, most of the

evaluations of ART to date have evaluated its effectiveness with adolescent populations (i.e., up to

18 years of age). A few published studies (Hatcher et al., 2008; Hornsveld, 2005; McGuire & Clarke,

2004) have reported outcomes for older, adult offenders. To date, there are no known published

outcome studies of ART among the age cohort that sits between adolescence and adulthood (i.e.,

early adulthood 18-24 years). Thus, one the aims of the main study was to investigate the

effectiveness of ART in an Australian custodial youth justice (i.e., 18-22 years) setting with a group of

young-adult male offenders. Consistent with Palmer’s (2007) recommendation regarding the need to

assess the relative contribution of each of the respective components of ART (i.e., cognitive,

behavioural, and affective) the measures employed in the main study were selected on the basis of

their theoretical relationship to each of the programs targets of change. Palmer argued that compared

to the anger control and social skills training procedures in ART, that there was relatively less

evidence for improvements in moral decision making and perspective-taking associated with the

moral reasoning component. Thus, the intention here was to provide evidence of change relative to

the specific targets of change implied within the three components of ART, particularly moral

reasoning.

110

In line with reommnedations made by Serin et al. (2009) and Tate et al. (1995) another aim

of the main study was to examine the clinical significance of potential post-treatment changes not just

criminogenic risk reduction. To this end, four specific hypotheses and two research questions were

posed. From pre-treatment (intake) to post-treatment (24-month follow-up) it was predicted that ART

participants would demonstrate:

1. A significant reduction in the frequency of self-reported and case-worker reported

aggressive behaviours and thoughts.

2. Fewer self-reported cognitive distortions.

3. A significant increase in self-reported social problem solving abilities.

4. A significant reduction in impulsivity

Additionally two broad research questions were posed:

I. Does participation in ART reduce aggressive symptoms, compared to a waitlist control

group?

II. If there is evidence of post-treatment gain/s are they maintained over time (i.e., from

treatment-end to 24-month follow-up)?

Method

Participants

Conistent with the pilot, participants in the main study (N=20) were serving at least a three

month custodial sentence at a youth custody centre located in Victoria, Australia for violent or

violence-related offences. The age range of participants was 18.7 to 20.5 years of age (M=19.6;

111

SD=.60). On average participants had completed 9.4 years of education. Most (74%) came from

intact two-parent families. With respect to family criminal history, 50% indicated that their father had a

previous criminal conviction and 5% stated that their mother had a previous conviction. Ten percent

of participants stated that their father or mother suffered from either an Axis I or II psychiatric

condition, while a quarter (25%) of participants indicated that they did not know. Nine young people

(45%) stated that they had previously been diagnosed with an Axis I disorder, including depression,

anxiety, and/or one of the DBD’s.

Recruitment

Recruitment into the main study was according to the same procedures described in the

pilot. All participants referred to ART met eligibility for the program in terms of clinically judged

cognitive ability, type of offence (i.e., violence-related) level of ciminogenic risk (moderate to high)

length of sentence (i.e., at least three months) and no self-reported history of psychotic symptoms in

the last six months. Table 8 shows completers, non-completers, waitlist, and refusals for each of the

ART groups conducted over the two year intervention period.

112

Table 8

ART group completers, non-completers, waitlist controls, and refusals for main study

Commence Dates Completers Non-completers Waitlist Refusal

Group 1 (Mar. 2005) 6 0 0 0

Group 2 (Oct. 2005) 3 3 0 0

Group 3 (Jan. 2006) 4 0 0 1

Group 4 (Mar. 2006) 4 2 1 1

Group 5 (July 2006) 3 3 1 0

TOTAL 20 8 2 2

Thirty-two young men were referred to ART by the Health Services team. Of those, 20 completed the

treatment and their data were analysed for the main study. The inclusion of a waitlist control group

was intended to provide an examination of the potential contribution of extra-therapeutic factors (i.e.,

changes on outcome measures irrespective of treatment) at treatment end for the treatment group

compared to the waitlist group (i.e., T2). However, institutional factors beyond experimental control

did not allow for a waitlist control group sizable enough for statistical comparison. These factors

included the length of custody before parole, participation in other co-occurring vocational (e.g.,

forklift driving) and educational (e.g., literacy) programs, and other criminongenic need (e.g., drug

and alcohol) and non-criminogenic need (e.g., peer education) programs.

Measures

The Aggression Questionnaire (Buss & Warren, 2000) and the How I Think Questionnaire

(Barrriga et al., 2001) were retained. The Social Problem Solving Inventory- Revised: Short Form

(SPSI-R:S; (D'Zurilla, Nezu, & Maydeu-Olivares, 2002) was added to the battery of youth self-report

113

measures in order to assess potential changes in problem solving abilities and the Impulsiveness

subscale of the I7 questionnaire (Eysenck, Pearson, Easting, & Allsopp, 1985) was included to assess

impulsivity. As previously discussed, several researchers (Kazdin & Weisz, 1998; Moffitt, et al., 2008;

Posavac & Carey, 2003) have indicated the need for a multi-informant approach in outcome research

as important criteria for identifying effective treatments. Thus, the Adult Behaviour Checklist (ABCL;

Achenbach & Rescorla, 2003) was added in the main study to provide for other informants (i.e.,

custodial key-workers).

The Social Problem Solving Inventory- Revised: Short Form (SPSI-R:S; (D'Zurilla et al.,

2002) is a 25-item, self-report instrument designed to assess the ability to resolve problems faced in

everyday life for ages 13 and older. Respondents rate their responses on a 4-point Likert scale from

0 (“Not at all true of me”) to 4 (“Extremely true of me”). The SPSI-R:S comprises two adaptive

problem solving dimensions- Positive Problem Orientation (PPO; e.g., “Whenever I have a problem I

believe that it can be solved”) and Rational Problem Solving (RPS; e.g., “When I am trying to solve a

problem, I think of as many options as possible until I cannot come up with any more ideas”) and

three dysfunctional dimensions- Negative Problem Orientation (NPO; e.g., “I feel threatened and

afraid when I have an important problem to solve”) Impulsivity/Careless Style (ICS; e.g., “When I am

trying to solve a problem I go with the first good idea that comes to mind”) and Avoidance Style (AS;

e.g., “When a problem occurs in my life I put off trying to solve it for as long as possible”). The

dysfunctional dimensions were reverse scored for calculation of the total scale score, as described by

D’Zurila et al. High total scale scores indicate greater overall social problem solving abilities. The

SPSI-R:S provides seven qualitative severity descriptors based on standard scores, which have a

mean of 100 and standard deviation of 15. For standard scores of 55 and below a respondent is

categorised as “extremely below the norm group average”; 56-70 “very much below the norm group

114

average”; 71-85 “below norm group average”; 86-114 “norm group average”; 115-129 “above norm

group average”; 130-144 “very much above norm group average”; and 145 and above “extremely

above the norm group average”. D'Zurilla et al. indicated that a standard score of one or more

standard deviations above or below the norm group average should be considered a substantial

departure from the average. The normative sample of young adults (age 17-39 years) consisted of

950 American college students. D'Zurilla et al. reported adequate reliability estimates for the SPSI-

R:S subscales with alpha’s ranging from .74 for ICS up to .83 for NPO and AS, and a high of .89 for

the total scale score.

Impulsivity was assessed using the I7 Questionnaire (Eysenck et al., 1985) which is a self-

report, 54-item, forced choice format (yes/no) questionnaire that includes three scales:

Impulsiveness, Venturesomeness, and Empathy. The present study employed the 19-item, uni-

dimensional I7Impulsiveness scale (e.g., “Do you often do things on the spur of the moment?”)

designed to assess poor impulse control. Higher scores on the I7 indicate greater impulsivity. The

scale was validated with a mixed gender sample, ranging in age from 16-89 years. Eysenck et al.

reported good internal consistency for males (alpha = .84) on the I7.

The Adult Behavior Checklist (ABCL; Achenbach & Rescorla, 2003) represents a revision

and increase in aged norms of the earlier Achenbach (1997 cited in Achenbach & Rescorla) Young

Adult Behavior Checklist (YABCL) which was normed for ages 18 to 30. The ABCL is completed by

“people who know the adult well, such as spouses, partners, friends, roommates, therapists, parents,

adult children, and other relatives” (p.6). In this study, the 118 problem behaviour items of the ABCL,

which are rated on a three point rating scale (“0=Not True”, “1=Somewhat or Sometimes True”,

“2=Very True or Often True”) and constitute the total scale score. The ABCL comprises eight

empirically-based syndrome subscales designed to assess co-occurring patterns of psychopathology:

115

Anxious/Depressed (14-items; e.g., “feels worthless, nervous tense”) Withdrawn (nine items; e.g.,

“trouble making friends”) Somatic Complaints (nine items; e.g., “tired without good reason”) Thought

Problems (12-items; e.g., “strange behaviour, strange ideas”) Attention Problems (17-items; e.g.,

“cannot concentrate”) Aggressive Behaviour (16-items; e.g., “attacks/threatens people”) Rule

Breaking Behaviour (13-items; e.g., “trouble with the law”) and Intrusive (six items; e.g., “brags a lot;

shows off”). The ABCL provides scores for two global second order factors: Internalizing- assesses

problems within the self, and Externalizing- items associated with conflicts with other people and

social mores. The ABCL raw scores are converted to T scores (M= 50; SD= 10). A T score of 65 to

69 is considered to be in the “borderline clinical range” and T scores of 70 and above are considered

to be within the “clinical” range. According to Achenbach and Rescorla, both the borderline and

clinical range of scores significantly differentiate between people referred to mental health services

for behavioural, emotional, and social problems and non-referred demographically matched

individuals. The normative sample for ABCL was based on a large (N= 1,660) American national

probability sample. Mean alpha coefficients of .85 for the Empirically Based Scales were reported

(Achenbach & Rescorla). Internal consistency was very high for the Internalizing (alpha = .92) and

Externalizing (alpha = .93) factors, and the Total Problems scale (alpha = .97). Relevant to the

purposes of the present study only the Aggressive and Rule-Breaking syndrome scales were

analysed. The Internalizing and Externalising and total scale scores were also included in the present

analysis.

Procedure

The ART co-trainers were the same in the main study as the pilot study. However,

subsequent to the “highly competent” QA assessment in the pilot study (see Appendix C) the roles of

main- and co- trainer were alternated between the co-facilitators from week to week for the five

116

groups comprising the main study. Several procedures aimed at ensuring treatment integrity were

employed in the main study. These included the use of self-assessment session checklists for each

of the three ART components (see Appendix C). Additionally, the co-trainers attended fortnightly

supervision with a clinical psychologist experienced in adolescent group therapy. Also, the present

author maintained regular email contact with the Washington State ART Master Trainer in which

issues of treatment integrity and group process were canvassed.

The research procedure for the main study was the same as for the pilot with the exception

of additional post-treatment assessment points at six month post-treatment follow-up (T3) and two

year post-treatment follow-up (T4). The key-worker report measure (i.e., ABCL; Achenbach &

Rescorla, 2003) was administered at T1, T2 and T3 but not T4. The ABCL was not administered at

T4 because at this point participants would have either exited from the justice system altogether or if

convicted of another offence would have most likely entered the adult justice system. Even for those

who might have re-entered the youth justice system, it was likely that they would not have been

allocated the same key-worker, making comparison of the measure at T4 problematic. This

procedure was consistent with a repeated- measures, multi-informant research design (Posavac &

Carey, 2003).

117

Results

Preparation of Data

All data screening and analysis was performed using SPSS version 17. Prior to analysis, all

variables were screened for data entry errors, outliers and violations of the assumptions for

multivariate analysis across all data collection points (i.e., T1 to T4). Potential outliers were assessed

using histograms, box plots, and descriptive statistics. The extent to which extreme values influenced

the distribution was determined by comparing the mean values with the 5% trimmed mean (Pallant,

2007). Where a substantial difference was identified, the outlier was replaced with the next closest

mean score at that time point which was not an outlier (Tabachnick & Fidell, 2001). This procedure

was implemented for one outlier on each of the SPSI-R subscales “Positive Problem Orientation”

(PPO) and “Negative Problem Orientation” (NPO) both at T2. These scores were causing significant

positive skew in the distribution (i.e., particularly low scores).

To assess significant departures from normality, skewness, kurtosis and standard error (SE)

scores were calculated. Tabachnick and Fidell (1996) outlined procedures for determining statistical

significance of non-normality in small to moderate sample sizes based upon the calculation of a z–

score for the skewness and kurtosis values. The critical z–score of 2.33 for one-tailed tests (p =.01)

was used as the cut-off for significant departure from normality for all dependent variables at each of

the data collection points (T1 to T4). There were no significant departures from normality except on

several of the subscales for the worker-report ABCL. Transformations were not performed on the

ABCL as it was scored using computer scoring software and all scores were entered as T–scores.

Moreover, potential problems with the validity of this scale with respect to its use in a correctional

118

setting were identified and will be discussed later in the General Discussion (see section- Social

Skills and Social Problem Solving).

Dealing with Missing Data

Missing data, or missing cases as in the present study, is common in most studies,

particularly in research involving longitudinal follow-up (Pallant, 2007; Streiner, 2002). In the present

study, there were three missing cases at the six month follow-up (T3) and six missing cases at the

two year follow-up (T4) for the youth self-report data. None of the missing cases were the same at T3

as T4. For the worker-report ABCL, five cases were missing at T3. Although there are several

suggested methods, each with various strengths and weaknesses, there is no agreed gold standard

regarding the best approach to use with missing data. All have important implications relative to the

validity of the study (Streiner, 2002; Tabachnick & Fidell, 2001).

Streiner (2002) reviewed several of the most common approaches to dealing with missing

cases, including multiple regression imputation methods to predict missing values. However, this

procedure requires a large sample size and thus was not feasible in the present study. The two

methods, reviewed by Streiner, that were possible in the present study were replacement of missing

values with the group mean at that data collection point or the “last observation carried forward”

(LOCF) for specific missing cases. Replacing missing cases with the group mean increases the risk

of a Type I error (i.e., concluding a significant difference, when in fact there is not one). In contrast,

the LOCF approach is more conservative with respect to the risk of a Type I error, but increases the

risk of failing to detect significant improvement at post-treatment or the degree of improvement (i.e.,

Type II error). In line with Streiner’s recommendation, the present study initially ran the statistical

analysis for both the group mean replacement and LOCF methods to assess for differences. The

119

results of this comparison revealed no substantial differences in the results for each the respective

methods (see Appendix D for SPSS output for mean replacement and LOCF methods). Therefore, as

SPSS provides a function for this procedure, the present study employed the group mean

replacement method for missing cases in all statistical analysis.

Statistical Analysis

A series of repeated measures analysis of variance (ANOVA) were conducted to assess

main effect differences across each of the four data collection time points (i.e., the independent

variable) for all dependent measures. The univariate main effect was reported where the sphericity

assumption was met. Where the assumption was violated, the multivariate main effect (Wilks Lamda)

was reported. For significant main effects, a linear trend effect (i.e., consistent decrease or increase

on a given variable) was calculated using SPSS polynomial function to test for significant linear

change from T1 to T4. In cases where the main effect was significant but the linear effect was not, a

polynomial quadratic trend effect (i.e., an initial decrease/increase is later followed by an

increase/decrease on a given variable) between T1 and T4 was calculated. In cases where the main

effect was significant but neither the linear nor quadratic effects were, a planned contrast using

SPSS- “difference” function, was performed (Rosenthal, 1985). To guard against Type I familywise

error, a conservative alpha (p < .01) for all significance testing was employed (Tabachnick & Fidell,

2001).

Effect sizes (partial eta squared) were calculated for all significant main and interaction

effects. Consistent with Cohen’s (1988) interpretive guidelines, the strength of an effect size was

reported as small (.01), medium (.06), or large (.14). Additionally, several researchers (Kazdin, 1999;

Kendall, Marrs-Garcia, Nath & Sheldrick, 1999; Wise, 2004) have recommended that treatment

120

effectiveness studies report changes based on normative data for the outcome measures used.

Thus, the group means for participants in the present study were compared to the available

normative data (i.e., percentiles or standard scores or T scores) for each of the dependent measures,

thereby providing a measure of clinical significance.

Scale Reliability and Intercorrelations

Table 9 shows the reliability coefficients (Cronbach’s alpha) for all dependent variables used in

the present study at each data collection point. Reliability estimates for the ABCL (Achenbach &

Rescorla, 2003) were not calculated as the scores reported for this scale were T-scores based on the

normative data provided by Achenbach and Rescorla.

121

Table 9

Reliability coefficients for all dependent measures at pre-treatment (T1), treatment-end (T2), 6-month

follow-up (T3), and 24-month follow-up (T4)

Cronbach’s alpha

Pre

(n = 20)

Post

(n = 20)

6-month

(n = 17)

24-month

(n = 14)

N of Items

AQ Total .94 .93 .92 .95 34

- Physical .92 .94 .86 .94 8

- Anger .76 .82 .69 .88 7

- Hostility .82 .77 .65 .76 8

- Indirect .77 .66 .62 .74 7

- Verbal .62 .87 .61 .75 5

HIT Total .93 .95 .96 .95 54

- Self-Centred .80 .85 .88 .74 9

- Blaming Others .75 .84 .88 .81 10

- Minimizing/Mislabeling .80 .89 .93 .85 9

- Assuming the Worst .72 .83 .89 .81 11

- Lying .77 .85 .86 .72 8

- Stealing .91 .91 .96 .89 11

- Oppositional Defiant .70 .83 .88 .74 10

- Physical Aggression .78 .92 .91 .85 10

SPSI- R:S Total .87 .80 .68 .89 25

- Positive Problem Orientation .83 .86 .65 .92 5

- Negative Problem Orientation .82 .62 .41 .79 5

- Rational Problem Style .77 .85 .83 .95 5

- Impulsive/Carelessness Style .60 .54 .86 .79 5

- Avoidance Style .88 .93 .79 .86 5

I7 Impulsiveness Scale .80 .82 .74 .85 19

Note. AQ = Aggression Questionnaire; HIT = How I Think Questionnaire; SPSI- R:S = Social Problem

Solving Inventory-Revised: Short Form

122

As indicated in Table 9, most measures showed moderate to very high internal consistency at each

of the data collection points. There were some instances where alpha coefficients fell below the

recommended cut-off of .70 for scale reliability for research purposes (DeVellis, 2003). However,

these instances of low reliability need to be considered in light of the small sample size in the present

study. It is not uncommon for low Cronbach values to be found in such cases (DeVellis, 2003;

Pallant, 2007).

Table 10 shows the intercorrelations between all dependent measures at T1. A similar table has

been included for T2 assessments (see Appendix E). Due to sample attrition at T3 and T4, no

correlation matrices were included for these data collection points. The direction of the correlations

(Pearson’s r) within scales was as expected relative to the construct validity of each measure. For the

AQ scale the strength of significant intercorrelations ranged from moderate (r=.57) for anger and

verbal aggression and strong (r=.89) for the AQ total scale and physical and indirect aggression

respectively. The strength of association between the HIT and it subscales ranged between moderate

(r=.39) for physical aggression and lying to strong (r=.91) for the HIT total and minimizing/mislabeling

and oppositional defiant and self-centered. As expected the dysfunctional problem solving

approaches- NPO, ICS, and AS were all showed moderate to high, negative associations with the

SPSI-R:S total scale. While the positive problem solving approaches- PPO and RPS showed

moderate positive relationships to the total scale.

123

Table 10

Inter-correlations for all dependent measures in the main study at T1 (N=20)

Note. All 1-tailed correlations; *p<.05, **p<.01 AQ = Aggression Questionnaire Total; PHY = Physical Aggression, ANG = Anger, HOS = Hostility, IND = Indirect Aggression, VER = Verbal Aggression; HIT = How I Think Questionnaire Total: SC = Self-Centred, BO = Blaming Others, MM = Minimising/Mislabelling, AW = assuming the Worst, L = Lying, S = Stealing, OD = Oppositional Defiant, PA = Physical Aggression; SPSI = Social Problem Solving Inventory-Revised: Short Form Total: PPO = Positive Problem Orientation, NPO = Negative Problem Orientation, RPS = Rational Problem Solving, ICS = Impulsive/Careless Style, AS = Avoidance Style; I7 = Eysenck Impulsiveness Scale.

AQ PHY ANG HOS IND VER HIT SC BO MM AW L S OD PA SPSI PPO NPO RPS ICS AS I7

AQ -

PHY .89** -

ANG .77** .82** -

HOS .71** .39*. .34 -

IND .89** .73** .60** .65** -

VER .74** .63** .57** .32 .63** -

HIT .63** .49* .33 .71** .56** .28 -

SC .62** .53** .29 .62** .50* .33 .90** -

BO .39* .14 .16 .61** .35 .18 .86** .73** -

MM .67** .61** .40 .62** .64** .33 .91** .76** .66** -

AW .59** .46* .31 .69** .51* .17 .94** .79** .76** .86** -

L .46* .25 .28 .74** .52** .10 .70** .51* .60** .68** .70** -

S .40* .23 .13 .53** .40* .19 .90** .78** .90** .80** .81** .57** -

OD .61* .49* .27 .68** .43* .26 .89** .91** .73** .76** .85** .45* .73** -

PA .70** .72** .48* .47* .56** .39* .84** .81** .60** .81** .82** .39* .62** .83** -

SPSI -.21 .07 -.01 -.68** -.13 .06 -.45* -.31 -.51* -.32 -.48* -.50* -.40* -.45* -.16 -

PPO .18 .21 -.08 .01 .31 .20 -.06 .10 -.20 .01 -.16 -.08 -.17 .10 .07 .54** -

NPO .34 .02 .18 .71** .22 .16 .33 .27 .43 .15 .35 .39* .25 .41* .12 -.84** -.25 -

RPS .10 .00 .01 .13 .05 .19 .05 .18 .12 -.09 -.06 .11 -.01 .02 .05 .62** .51* .20 -

ICS .46* .19 .03 .65** .51* .29 .54** .52** .56** .43* .42* .45* .42 .54** .41* -.54** .24 .57** .30 -

AS .15 -.11 -.18 .71** .11 -.15 .44* .38* .43* .32 .43* .59** .36 .43* .14 -.81** -.09 -.75** .26 .60** -

I7 .36 .24 .15 .30 .44* .26 .41* .34 .44* .41* .28 .05 .40* .37 .48* -.30 .08 -.30 .08 .56** .25 -

124

With regard to the intercorrelations between measures, Table 10 shows that for the HIT and AQ Total

Score and subscales most were significantly and positively related. As cognitive distortions increased

so too did aggression. The exception was for the AQ subscales Anger and Verbal Aggression, which

were not significantly related to any of the HIT subscales except Physical Aggression. Surprisingly,

there were few significant associations between the AQ and the SPSI-R:S. Overall, the only AQ

subscale that was significantly related to social problem solving was Hostility. As hostility increased

general social problem abilities (i.e., SPSI-R:S total) tended to decrease and dysfunctional problem

solving approaches (i.e., NPO, ICS and AS) tended to increase. Not surprisingly, the HIT and the

SPSI-R;S total scores showed mostly significant and inverse relationships, as cognitive distortions

increased social problem solving abilities tended to decrease. As expected, Impulsiveness showed a

significant, positive association with ICS on the SPSI-R:S. Except for Indirect Aggression,

Iimpulsiveness was not significantly related to the AQ. Overall, the strength of intercorrelations at T2

(see Appendix E) tended to increase slightly, which might have been due to possible practice effects.

Aggression: Youth self-report

Table 11 shows means, standard deviations, and main effect ANOVA results from T1 to T4

on the AQ total scale score and each of its subscales.

125

Table 11

Descriptive statistics and main effect ANOVA results from T1 to T4 on the AQ (N=20)

Pre (T1) Post (T2) 6-month (T3) 24-month (T4) Main Effect

M (SD) M (SD) M (SD) M (SD) F(3,57) p Partial η2

AQ Total 102.25 (26.00) 89.35 (24.51) 83.26 (20.73) 85.14 (22.01) 4.91 .004 .32

Physical 26.85 (8.86) 21.40 (9.00) 20.94 (6.56) 20.14 (7.49) 6.07 .001 .31

Anger 19.41 (5.04) 18.23 (5.79) 17.47 (4.64) 15.93 (5.58) 2.62 .06 Ns

Hostility 21.60 (7.69) 18.10 (5.76) 17.23 (4.70) 20.00 (5.16) 4.40 .007 .19

Indirect 16.55 (4.97) 14.45 (4.48) 15.00 (3.89) 14.43 (3.83) 4.82* .01 .46

Verbal 16.40 (3.80) 16.70 (4.84) 15.23 (3.36) 14.64 (3.41) 1.99 .13 Ns

Note. AQ = Aggression Questionnaire; Physical = Physical Aggression; Indirect = Indirect

Aggression; Verbal = Verbal Aggression; * = Multivariate Wilks’ Lambda (df = 3,17) reported

due to lack of sphericity; ns = non-significant

As shown in Table 11, there was a significant main effect for time on the AQ Total scale score,

Physical Aggression, Hostility, and Indirect Aggression. However, there was no significant change for

Anger and Verbal Aggression. The effect sizes reported for the significant main effects were large,

ranging from .19 for Hostility to .46 for Indirect Aggression (Cohen, 1988).

Figures 5 to 8 show the direction of change on the AQ from T1 to T4 for each of the

significant main effects. To test linear or quadratic change from T1 to T4, a series of polynomial trend

analyses were conducted.

126

Figure 5. Agression Questionnaire Total Scale score T1 to T4.

As shown in Figure 5, the linear trend for the AQ total score was significant (F (1,19)= 7.88, p= .01,

partial η2 = .29). There was a significant linear reduction in overall self-reported aggression from T1

to T2 (F (1,19)= 12.65, p= .002, partial η2 = .40) with a large effect size. There was no significant

change between T2 and T3 (F (1,19)= 5.25, p= .03) or between T3 and T4 (F (1,19)= 1.73, p= .20).

According to the clinical cut-offs for the AQ Total Score, the average score at T1 fell at the 76th

percentile (“High Average”) with all subsequent mean scores falling between the 50th and 60th

percentile (within the “normal” range) (Buss & Warren, 2000).

127

Figure 6. Physical Aggression T1 to T4.

The linear trend effect shown in Figure 6 for Physical Aggression was significant (F (1,19)= 12.10, p=

.003, partial η2 = .39). There was a significant reduction in self-reported physical aggression from T1

to T2 (F (1,19)= 16.43, p= .001, partial η2 = .46). The effect sizes were large. From T2 to T3 (F

(1,19)= 3.21, p= .09) and T3 to T4 F (1,19)= 3.98, p= .06) there was no significant change in self-

reported physical aggression. According to the percentile rankings and qualitative descriptors for the

AQ, the mean scores on this subscale at T1 fell at the 90th percentile (well within the “clinical” range).

All subsequent mean scores fell between the 50th and 55th percentile (“normal” range) (Buss &

Warren, 2000).

128

Figure 7. Hostility from T1 to T4.

While the main effect for Hostility was significant, the linear trend was not (F (1,19)= 1.46, p = .24).

However the quadratic effect was significant (F (1,19)= 9.11, p= .007, partial η2 = .32). Planned

contrasts revealed a significant reduction in hostility from T1 to T2 (F (1,19)= 8.70, p= .008, partial

η2 = .31). Again these effect sizes were large according to Cohen’s (1988) criteria. There was no

significant change between T2 and T3 (F (1,19)= 3.83, p= .07) or between T3 and T4 (F (1,19)= 1.10,

p= .31). The group means on Hostility were within the “normal” range of scores at all time points.

129

Figure 8. Indirect Aggression from T1 to T4.

Both the linear (F (1,19)= 2.33, p = .14) and quadratic (F (1,19)= 1.16, p = .30) trend effects for

Indirect Aggression were non-significant. As the main effect was significant, planned contrasts were

conducted. The tests showed a significant (F (1,19)= 11.04, p= .004, partial η2 = .37) decrease in

indirect aggression between T1 and T2. There was no significant change between T2 and T3 (F

(1,19)= .46, p= .51) or between T3 and T4 (F (1,19)= .77, p= .39). The group means for Indirect

Aggression remained within the “normal” range at all time points.

Summary for aggression results. The AQ Total and Physical Aggression scores showed a

significant linear reduction between T1 and T4. The linear trend for Hostility was not significant but

130

the quadratic effect was. For Indirect Aggression both the linear and quadratic effects were

significant. For each of these significant changes planned contrasts revealed that the reductions were

evident at post-treatment (T2). Given that there was no significant increase or decrease on these

measures at T3 or T4, the results suggest that the reductions in self-reported aggressive behaviours

and thoughts were maintained at two-year (T4) follow-up.There was no significant change on the AQ

subscales Anger and Verbal Aggression. All scores on the AQ were either clinical or borderline

clinical at T1 and T2 but had fallen within the “normal” range by T3 and remained there at T4.

Cognitive Distortions and Behavioural Referents: Youth self-report

Table 12 shows the descriptive statistics and main effect ANOVA results for the HIT Total

Scale, the four cognitive distortions (i.e., Self-centred, Blaming Others, Minimizing/Mislabeling,

Assuming the Worst) and the four behavioural referents (i., Lying, Stealing, Oppositional Defiant, and

Physical Aggression) from T1 to T4.

131

Table 12

Descriptive statistics and main effect ANOVA results from T1 to T4 on the HIT (N=20)

Pre (T1) Post (T2) 6-month (T3) 24-month (T4) Main Effect

M (SD) M (SD) M (SD) M (SD) F(3,57) p Partial η2

HIT Total 3.17 (.70) 2.78 (.76) 2.58 (.88) 2.13 (.50) 11.66 .000 .38

Self-Centered 3.18 (.83) 2.79 (.77) 2.67 (.87) 2.21 (.53) 8.54 .000 .31

Blaming Others 3.04 (.76) 2.73 (.80) 2.56 (.87) 2.02 (.55) 9.70 .000 .34

Minimizing/ Mislabel 3.22 (.89) 2.80 (.90) 2.50 (1.04) 2.02 (.59) 11.26 .000 .37

Assuming the Worst 3.21 (.68) 2.78 (.73) 2.55 (.85) 2.39 (.56) 6.65 .000 ..29

Lying 3.28 (.79) 2.82 (.84) 2.74 (.89) 2.46 (.56) 7.48 .000 .28

Stealing 2.76 (1.04) 2.50 (.84) 2.32 (.98) 1.68 (.48) 8.91 .000 .32

Oppositional Defiant 3.47 (.72) 3.04 (.79) 2.71 (.93) 2.31 (.60) 12.98 .000 .41

Physical Aggression 3.20 (.77) 2.78 (.89) 2.57 (.91) 1.97 (.59) 13.02 .000 .40

Note. HIT = How I Think Questionnaire.

The main effects for all HIT subscales and the total scale score were highly significant (p < .001). The

effect sizes were large ranging from .28 (Lying) to .41 (Oppositional Defiant and Physical

Aggression).

Figures 9 to 13 plot the direction of change for the group means between T1 and T4 on the

HIT Total Score, cognitive distortion and behavioural referent subscales. To test linear change from

132

one time point to the next, a series of polynomial trend analyses were conducted. Clinical significance

based on the normative data provided for the HIT was also reported.

Figure 9. HIT total from T1 to T4.

The linear trend effect displayed in Figure 9 was significant with a large effect size (F (1,19)= 11.65,

p= .000, partial η2 = .38). There was no significant change between T1 and T2 (F (1,19)= 4.52, p=

.05) and T2 and T3 (F (1,19)= 5.21, p= .03). However, there was a significant reduction on the HIT

total score between T3 and T4 (F (1,19)= 34.11, p= .000, partial η2 = .64). The group mean at T1

(88th percentile) was clinically significant while at T2 it had dropped to within the borderline clinical

133

range (74th percentile). At T3 (62nd percentile) and T4 (well below the 50th percentile) scores were

within the “normal” range (Barriga et al., 2001).

Figure 10. Self-Centered Thinking from T1 to T4.

The trend effect for Self-centered Thinking (see Figure 10) was significant (F (1,19)= 36.50, p= .000,

partial η2 = .66). While there was no significant difference between T1 to T2 (F (1,19)= 2.84, p= .11)

or between T2 to T3 (F (1,19)= 3.71, p= .07) there was a significant reduction in self-centred thinking

between T3 to T4 (F (1,19)= 26.87, p= .000, partial η2 = .57). The effect size was large. The mean

score for Self-Centered Thinking at T1 reached clinical significance (83rd percentile) while all post-

treatment scores fell within the “normal” range.

134

Figure 11 . Blaming Others from T1 to T4.

The linear trend effect shown in Figure 11 for Blaming Others was also significant (F (1,19)= 30.19,

p= .000, partial η2 = .61). Between T1 and T2 (F (1,19)= 2.14, p= .16) and between T2 to T3 (F

(1,19)= 3.41, p= .08) there was no significant change in participants tendency to blame others.

However, there was a significant reduction between T3 and T4 (F (1,19)= 29.14, p= .000, partial η2 =

.61). The effect sizes were large. At T1, participants scores were “borderline clinical” (81st percentile)

and thereafter fell within the “normal” range (Barriga et al., 2001).

135

Figure 12. Minimizing/Mislabeling from T1 to T4.

Figure 12 shows a highly significant linear trend effect for Minimizing/Mislabeling (F (1,19)= 51.92, p=

.000, partial η2 = .73) with a significant reduction between T3 and T4 (F (1,19)= 38.40, p= .000,

partial η2 = .67). The effect sizes for the linear trend and planned contrast were large. Planned

contrasts showed that there was no significant difference between T1 and T2 (F (1,19)= 4.13, p=

.06,) or T2 and T3 (F (1,19)= 5.20, p= .03). Normative data indicated that at T1 participants scores

fell within the “clinical” range (88th percentile). At post-treatment assessment (T2) mean scores were

136

“borderline” clinical (77th percentile) and subsequently dropped within the “normal” range. By T4

scores were below the 50th percentile (Barriga et al., 2001).

Figure 13 . Assuming the Worst from T1 to T4.

The linear trend effect for Assuming the Worst was significant (F (1,19)= 24.21, p= .001, partial η2 =

.56). Participants tendency to assume the worst showed a significant reduction between T3 and T4 (F

(1,19)= 12.86, p= .002, partial η2 = .40) but not between T1 and T2 (F (1,19)= 6.75, p= .02) or T2 and

T3 (F (1,19)= 5.55, p= .03). The group mean at T1 was clinically significant (89th percentile) while at

137

T2, it fell within the “borderline clinical” range (78th percentile). By T3 (66th percentile) and T4 (56th

percentile) the group average had fallen within the “normal” range.

Figure 14. .Lying from T1 to T4.

Figure 14 shows the linear trend effect for Lying, which was significant (F (1,19)= 23.99, p= .001,

partial η2 = .56) and the effect was large. There was a significant reduction in participants self-

reported lying between T1 and T2 (F (1,19)= 9.22, p= .007, partial η2 = .33) and between T3 and T4

(F (1,19)= 14.53, p= .001, partial η2 = .43) but not T2 to T3 (F (1,19)= 2.80, p= .11). The group mean

138

at T1 fell within the “borderline clinical” range (79th percentile). All subsequent scores fell within the

“normal” range (Barriga et al., 2001).

Figure 15 . Stealing from T1 to T4.

The declining linear trend for Stealing was also significant (F (1,19)= 29.00, p= .001, partial η2 = .60)

with a significant reduction between T3 and T4 (F (1,19)= 38.07, p= .000, partial η2 = .67) but not

between T1 and T2 (F (1,19)= 1.12, p= .30) or T2 and T3 (F (1,19)= 2.36, p= .14). At T1, scores fell

within the “clinical” range (88th percentile) while at T2 they were classified as “borderline clinical” (80th

percentile). By T3 and T4 scores had fallen within the “normal” range (72nd and below the 50th

percentiles, respectively).

139

Figure 16 . Oppositional Defiant from T1 to T4.

The linear trend for the Oppositional Defiant subscale shown in Figure 16 was significant and large (F

(1,19)= 54.94, p= .001, partial η2 = .74). Although the reduced scores between T1 and T2 were non-

significant (F (1,19)= 5.04, p= .04) there was a significant reduction between T2 and T3 (F (1,19)=

8.92, p= .008, partial η2 = .32) and between T3 and T4 (F (1,19)= 28.33, p= .000, partial η2 = .60).

Group means were well within the “clinical” range at T1 (89th percentile) but had fallen within the

“borderline clinical” range (77th percentile) at T2. By T3 and T4, the group mean fell within the

“normal”range (63rd and well below the 50h percentiles, respectively).

140

Figure 17 . HIT Physical Aggression T1 to T4.

The linear trend effect for the HIT subscale- Physical Aggression (Figure 17) was significant and

large (F (1,19)= 41.78, p= .001, partial η2 = .69). There was a significant reduction between T3 and

T4 (F (1,19)= 38.39, p= .000, partial η2 = .67) but not between T1 and T2 (F (1,19)= 4.90, p= .04) or

T2 and T3 (F (1,19)= 4.31, p= .05). The group mean at T1 was high and clinically significant (88th

percentile). By T2, the group mean was just within the “borderline clinical” range (73rd percentile). By

T3 and T4, scores were within the “normal” range (65th and less than the 50th percentiles,

respectively).

141

Summary of cognitive distortions and behavioural referents. The linear trend analysis

for the HIT Total Score and all of its subscales showed significant reductions between T1 and T4.

Group means for the HIT fell within the “clinical” or “borderline clinical” range at T1 but fell within the

“normal” range by the six month follow-up (T3). The reductions in cognitive distortions were

maintained at the two-year follow-up (T4).

Social Problem Solving: Youth self-report

Table 13 shows descriptive statistics and main effect ANOVA results from T1 to T4 on the

Social Problem Solving Inventory-Revised: Short Form (D’Zurilla et al., 2002).

142

Table 13

Descriptive statistics and main effect ANOVA results from T1 to T4 on the SPSI (N=20)

Pre (T1) Post (T2) 6-month (T3) 24-month (T4) Main Effect

M (SD) M (SD) M (SD) M (SD) F(3,57) P Partial

η2

SPSI Total 11.28 (2.80) 12.72 (2.94) 12.72 (2.51) 13.40 (2.35) 4.92 .004 .21

PPO 12.45 (4.62) 13.95 (4.44) 14.82 (2.92) 15.64 (3.62) 3.82 .02 ns

NPO 7.50 (4.74) 6.45 (3.49) 6.41 (3.04) 6.07 (3.19) 0.83 .48 ns

RPS 8.65 (3.77) 10.35 (4.77) 10.47 (3.87) 12.07 (4.90) 3.48 .02 ns

ICS 8.70 (3.37) 7.90 (3.18) 8.32 (4.47) 8.79 (3.62) 0.33 .80 ns

AS 8.50 (6.24) 6.35 (5.57) 6.94 (3.93) 5.86 (3.73) 2.03 .12 ns

Note. SPSI = Social Problem Solving Inventory; PPO = Positive Problem Orientation; NPO = Negative Problem

Orientation; RPS = Rational Problem Solving; ICS = Impulsive/Careless Style; AS = Avoidance Style; ns = non-

significant.

The main effect for the SPSI total scale score was significant. The increasing mean scores between

T1 and T4 was consistent with significant improvement in overall social problem solving. However,

the main effects for the SPSI subscales did not reach significance at the conservative p < .01. Figure

18 shows the trend effect for the SPSI total scale from T1 to T4.

143

Figure 18 .Social Problem Solving Total Scale from T1 to T4.

The linear trend effect shown in Figure 18 was significant and large (F (1,19)= 8.69, p= .008, partial

η2 = .31). There was significant improvement in overall social problem solving between T1 and T2 (F

(1,19)= 9.61, p= .006, partial η2 = .34) but not between T2 and T3 (F (1,19)= 3.65, p= .07) or T3 and

T4 (F (1,19)= 3.55, p= .08). The interpretive guidelines for the SPSI-R:S are based on standard

scores, which have a mean of 100 and a standard deviation of 15 (D’Zurilla et al., 2002). According to

these guidelines, the group means for the current participants fell within the “norm group average” at

all time points (i.e., a standard score between 86-114) for the Total Score and subscale scores.

Summary of social problem solving. The youth self-report scores for social problem

solving showed significant improvement from T1 to T4 on the SPSI total scale. However, there was

144

no significant change on the five SPSI subscales. All scores on the SPSI were within the normal

range at all data collection points.

Impulsivity: Youth self-report

The main effect for Impulsiveness was significant with a large effect size (F(3,57)= 4.91, p=

.004, partial η2 = .21). Figure 19 shows the mean scores from T1 to T4 for the youth self-report

scores on Impulsiveness.

Figure 19 . Impulsiveness from T1 to T4.

145

The linear trend effect displayed in Figure 19 was significant and large (F (1,19)= 14.58, p= .001,

partial η2 = .43). There was a significant reduction in participant’s self-reported impulsiveness

between T2 (M = 11.10, SD = 4.29) and T3 (M = 9.82, SD = 3.56) (F (1,19)= 8.34, p= .009, partial

η2 = .31). The reductions between T1 (M = 12.50, SD = 3.91) and T2 (F (1,19)= 1.89, p= .19) and T3

and T4 (M = 8.93, SD = 4.08) (F (1,19)= 5.68, p= .03) were non-significant. In the normative sample,

a mean score of 9.84 (SD= 4.13) was reported for males between 16 and 19 years of age (Eysenck

et al., 1985). In the present study, the group mean for impulsiveness at T1 fell at the 74th percentile,

above the normative average. By T3, the significantly reduced group mean (M= 9.82) was

comparable to the normative average for similar aged males (Eysenck et al).

Adult Behavior Checklist: Worker-report

Table 14 shows the T-score group means, standard deviations and ANOVA main effect

results for the ABCL from T1 to T3, as reported by custodial workers. None of the main effects for the

worker-report ABCL were significant at the conservative p < .01. Moreover, the T- score group means

and standard deviations show a relatively flat distribution between T1 and T3.

146

Table 14

T-score descriptive statistics and main effect ANOVAS from T1 to T3 for the ABCL (N=20)

Note. T-scores have a mean of 50 and a standard deviation of 10; * Multivariate test- Wilks’

Lambda was reported due to lack of sphericity- df = (2,18); ns = non-significant.

According to the clinical cut-off scores for the ABCL Total Problem, Internalising and

Externalizing scales, a T-score of 60-63 (approximately the 84th through to the 90th percentiles) is

classified as “borderline clinical”. T-scores greater or equal to 64 are considered to fall within the

“clinical” range (Achenbach & Rescorla, 2000). For the Syndrome scales (e.g., Aggression, Rule-

breaking Behaviour) a slightly higher cut-off is indicated with borderline T-scores ranging from 65 to

69, and T scores needing to be greater than 70 to be categorised as “clinical” . At no time in the

present study did any of the worker-reported ABCL scores reach borderline or clinical cut-offs.

In the next chapter, the results for the main study are discussed relative to the specific

targets of change in ART.

Pre (T1) Post (T2) 6-month (T3) Main Effect

M (SD) M (SD) M (SD) F(2,38) P Partial η2

ABCL Total 55.80 (8.14) 52.50 (9.50) 52.33 (7.35) 3.64* .05 Ns

Aggression 56.95 (6.83) 57.25 (8.01) 56.33 (7.59) .17* .86 Ns

Rule Breaking 64.25 (10.82) 61.75 (6.57) 60.93 (7.49) 1.43 .25 Ns

Internalizing 55.15 (10.12) 51.25 (8.68) 51.80 (9.22) 1.75 .19 Ns

Externalizing 58.85 (7.19) 57.38 (10.29) 56.00 (9.70) 2.08* .15 Ns

147

Discussion

The aims of the main study were firstly to evaluate the effectiveness of ART in Australia with

a group of young adult, violent male offenders. Consistent with Palmer’s (2007) recommendation, the

second aim was to provide further evidence for improved morel decision-making and social

perspective taking abilities thought to be associated with the moral reasoning training procedures in

ART. To this end, the measures used in the main study were selected on the basis of their theoretical

relationship to the specific cognitive, affective and behavioural targets of change implicit within each

of three components of ART. The third aim of the main study was to add to the existing “what works”

literature by examining treatment outcomes relative to behavioural and psychological change (i.e.,

aggressive thoughts and behaviours, self-serving cognitive distortions, social problem solving, and

impulsivity) not just criminogenic risk reduction as assessed by recidivism (Serin et al., 2009; Tate et

al., 1995).

Several hypotheses were made relative the aims of the main study. From pre-treatment (T1)

to 24-month (T4) follow-up, it was predicted that ART participants would show evidence of significant:

(1) reductions in self-reported and worker-reported aggressive behaviours and thoughts; (2)

reductions in self-reported aggressive and antisocial cognitive distortions; (3) improvement in self-

reported social problem solving abilities; and (4) reductions in self-reported impulsivity. In this

chapter, the results of each of the main study hypotheses are discussed relative to the targets of

change implicit in ART (i.e., Aggressive Behaviour and Thoughts, Cognitive Distortions and Moral

Reasoning, Social Problem Solving, and Impulsivity).

The present study also included two broad research questions: (1) Does participation in ART

reduce aggressive symptoms compared to a waitlist control group? and (2) If there is evidence of

148

post-treatment gain/s will they be maintained over time (i.e., from treatment-end [T2] to 24-month

follow-up [T4])? As previously mentioned, it was not possible to address the first research question

because of insufficant numbers in the waitlist group. This issue will be further discussed in the section

on Limitations. The evidence in support of the second research question will be discussed in the

section- Maintenance of Treatment Gains. A notable absence in the offender rehabilitation literature

is discussion of therapeutic process variables such as the therapeutic alliance or group cohesion, and

how these factors contribute to treatment effectiveness (Dowden & Andrews, 2004; Holmqvist, Hill, &

Lang, 2007; Ross, Polaschek, & Ward, 2007). The importance of attending to group processes in

ART was a key learning from this study, and will be discussed in the section- Group Process versus

Content. The final sections of this chapter will discuss the Limitations, Clinical Implications and

recommendations for Future Research, and provide a Summary of Findings and some Concluding

Comments with respect to the main study.

Aggressive Behaviours and Thoughts

As hypothesized, there were significant reductions in youth self-reported aggressive

behaviours and thoughts following participation in ART. The AQ Total score showed a significant

reduction at treatment-end (T2) with a large effect size (Cohen, 1988). At the six-month (T3) and two-

year (T4) follow-up, there was no further change (increase or decrease) on the AQ Total. This finding

suggests that there was an immediate post-treatment gain and that those gains were maintained at

two-year follow-up with no further increases or decreases. According to Buss and Warren (2000) the

AQ Total score provides a measure of aggression relative to both behavioural frequency (e.g., “I get

into fights more than most people”) and cognitive intensity (e.g., “At times I feel like a bomb ready to

explode”). The measurement of frequency and intensity is consistent with Novaco’s (1975, 1997)

149

multi-component conceptualisation of anger (see Figure 4, p.52). Moreover, Novaco’s multi-

component model has been operationalised by Feindler and her colleagues (Feindler, 2003; Feindler

& Guttman, 1994; Feindler & Ecton, 1986) in the Anger Control Training (ACT) component of ART.

The multi-component targets of change in ACT include: environmental activation (i.e., External

Triggers); physiological arousal (i.e., Cues/Body-Signs; Reducers); cognitive structures and

processes (i.e., verbalization of self-talk; Internal Triggers; Reminders; Thinking Ahead; Self-

Evaluation); subjective affective experience (i.e., Internal Triggers; Reminders); and behavioural

enactment (i.e., Social Skills). The significant change on the AQ Total score, the maintenance of

those changes at two year follow-up, and the large effect size (partial η2 = .32) indicated strong

support for the multi-component training procedures employed in ACT.

In addition, the AQ subscales- Physical Aggression, Hostility and Indirect Aggression all

showed significant reductions at treatment-end. As was the case with the AQ Total score, the change

on these measures was evident at T2 and there was no further change (increase or decrease) at

subsequent follow-ups. Again, these results indicate support for maintenance of the initial treatment

gains at two year follow-up and, according to Cohen’s (1988) criteria, the effects sizes were large.

High scorers on the Physical Aggression subscale typically lack impulse control and knowledge of

non-aggressive responses (i.e., pro-social skill deficits) when faced with perceived provocation or

frustration (Buss & Warren, 2000). In addition, high scorers on this subscale are likely to employ

cognitive distortions to justify their aggressive acts such as “they asked for it” (i.e., Blaming Others

and Mislabeling). The presence of poor impulse control and the use of cognitive rationalisations are

consistent with the dynamic risk factors which Andrews et al. (1990) argued should be the primary

targets of change in treatment aimed at ameliorating criminal behaviour. Poor impulse control, lack of

150

pro-social skills, and the use cognitive distortions to justify and maintain aggressive behaviour were

key characteristics of the participants in the present study prior to completing ART.

While the Physical Aggression subscale represents a tendency toward the externalisation of

aqgression, the Hostility and Indirect Aggression subscales reflect a propensity toward angry

rumination (i.e., internalisation of anger). High scores on Hostility are consistent with feelings of

paranoia (e.g., “I sometimes feel that people are laughing at me behind my back”) and the cognitive

distortion Assuming the Worst (e.g., “I can’t help losing my temper a lot”). Buss and Warren (2000)

noted that individuals scoring high on Hostility are likely to demonstrate hostile attributions toward

others. As previously discussed, several researchers (e.g., Barriga et al., 2001; Dodge et al., 2002;

Dodge & Pettit, 2003) have identified hostile attributions as being pervasive in the social-cognitive

appraisals of aggressive children and youth. The reduced Hostility scores for participants in the

present study might reflect an increased capacity to attend to non-hostile social cues and respond

accordingly through the use of one of the prosocial skills taught in Skillstreaming.

Indirect Aggression (e.g., “If I’m angry enough I may mess up someone else’s work”)

measures the tendency toward passive or non-confrontational aggression. High scores on this scale

are associated with oppositional and avoidant behaviours among children and antisocial personality

characteristics among adults, with concomitant chronic levels of frustration and angry cognitions

(Buss & Warren, 2000). Both Berkowitz (1990) and Novaco (2007) discussed the relationship

between chronic negative affect (e.g., frustration, annoyance, irritability) and the propensity to

experience anger-related feelings and thoughts, which in turn is associated with a “hair-pin” trigger for

aggressive behaviour. Further, Novaco noted the role of accumulated negative affect and angry

rumination in generating aggressive social interactions. Similarly, Gilbert and Daffern (in press)

151

described processes of over-learning and automatization in the development of aggression-related

cognitions and their association with schematic biases that interpret benign social situations as

threatening and requiring an aggressive response. This is consistent with anecdotal reports from

participants in the present study. They indicated a tendency to find themselves in anger provoking

situations that easily escalated into physical and/or verbal aggression. The reduced tendency toward

angry rumination (i.e., less Hostility and Indirect Aggression) for the ART participants, and the

maintenance of those gains at two year follow-up, may well be associated with a reduction in the

tendency to self-generate, albeit unconsciously, interpersonal conflict and subsequent aggression

relative to heightened levels of anger arousal.

In addition to testing for statistically significant change, this study was also interested in

examining “real-life” or clinically significant change. Several researchers (e.g., Kazdin, 1999; Kendall

et al., 1999; Wise, 2004) have described various procedures for determining clinical significance.

Consistent with Kazdin (1999) the present study determined clinical significance by comparing the

group means at each data collection point with the clinical cut-offs provided for each of the measures

used. Prior to participation in ART (T1) the group means on the AQ Total and the Physical

Aggression, Anger and Hostility subscales were clinically significant, while the group means for

Verbal and Indirect Aggression fell within the borderline clinical range. By post-treatment (T2) scores

on the AQ total, Physical Aggression, Hostility and Indirect Aggression fell within the “normal” range,

and by six month follow-up (T3) all scores were within “normal” range, where they remained at 24-

month follow-up (T4). These results indicated that the post-treatment reductions in participant’s self-

reported aggression were associated with clinically relevant symptom reduction. While these

improvements in aggressive related symptoms might transfer into very real improvements in the daily

social functioning of these youth, information from other informants known to the participants (e.g.,

152

family members, peers, intimate partners) would be needed to support this. Additionally, Kazdin

(1999) noted that post-treatment clinical change may or may not be associated with improved

functioning in daily life.

The youth self-report reductions in aggressive behaviours and thoughts were consistent with

previous evaluations of ACT as a stand alone intervention (Nugent et al., 1997; Sukhodolsky et al.,

2009) and as part of the multi-component ART program (Gunderson & Savardtal, 2006; Hornsveld,

2005). Although there were limitations to the Nugent et al. study (i.e., unequal sample sizes between

treatment and control groups and compromised treatment integrity) their results showed significant

behavioural improvement on the CBCL (Achenbach, 1991) for ACT participants compared to the

comparison groups. However, in contrast to the present study, Nugent et al. reported that these

treatment gains were not maintained beyond completion of ACT. They suggested that this might have

been due to the adverse effects of the program ending prematurely as a result of the aggressive

behavior of the group leader. Sukhodolsky et al. compared 26 young people (M age = 12.7years) with

comorbid disruptive behaviour disorders and Tourette’s Syndrome. They reported a significant

reduction in parent, clinician and child report measures of anger and externalizing behaviours

compared to the treatment as usual group. Given that only two participants in this study met criteria

for CD and none were outside the normal range of scores on the CBCL, Sukhodolsky et al. cautioned

against generalising their findings to adolescents with more serious forms of conduct related

problems. Given that particpants in the present study were moderate to high risk, young adult, violent

offenders, the results provide further evidence that the treatment procedures contained within the

ACT component of ART, are effective in treating more serious forms of conduct related problems.

However, several researchers (e.g., Beck, 1996; Longmore & Worrell, 2007; Maruna & Mann, 2006)

have noted that the overall effectiveness of CBT interventions appears to lie in the interaction effects

153

of the affective, cognitive and behavioural treatment components implicit in the CBT approach. This is

consistent with the theoretical underpinnings of ART in which a multimodal approach to altering

entrenched and difficult to change aggressive behaviour is recommended (Goldstein et al., 1998).

The significant reductions in youth-reported aggressive behaviours and thoughts also

support the findings of two previous evaluations of ART. Gunderson and Savardtal (2006) reported

significant reductions in child, parent and teacher ratings of problem behaviours at post-treatment

follow-up for ART participants in a school setting (M age = 13.4 years). There were substantial

differences in the age, treatment setting, and degree of aggressive cognitions and behaviours

between the Gunderson and Savardtal study and the present study. A more valid comparison of ART

outcomes relative to these factors was reported by Hornsveld (2005). He reported the post-treatment

findings for an adaptation of ART (i.e., Aggression Control Therapy) which, in addition to the three

standard components of ART, included a component designed to improve goal setting abilities and

positive self-reinforcement for goal attainment. Participants included 109 older (M age = 32.5 years)

forensic inpatients and 44 younger (M age = 23.4 years) outpatients, all with a history of violent

offending and a diagnosis of APD and/or CD. Similar to the findings of the present study, results

showed significant reductions in self-reported aggressive behaviours and thoughts at treatment-end

and these gains were maintained at follow-up assessments (i.e., 5-15 weeks after treatment-end).

Consistent with the principle of criminogenic risk (Andrews et al., 1990) Hornsveld indicated that the

intervention was most successful with participants who reported entrenched and higher levels of

anger.

Subsequent to the pilot study the worker-report ABCL (Achenbach & Rescorla, 2003) was

added to allow for a multi-informant research design and thereby increase the external validity of the

study (Kazdin & Weisz, 1998; Posavac & Carey, 2003). The ABCL was administered to custodial

154

workers from pre-treatment (T1) to six month follow-up (T3). Contrary to expectation, there was no

significant change at any time point on this measure. Also contrary to expectation, the group T-score

means for participants in the main study fell within the “normal” range at all assessment points. Given

that the ABCL is designed to assess both externalising and internalising symtomatology these results

were particularly surprising and go against the high prevalence rates reported for comorbid

psychiatric symtoms among offender populations (Abram et al., 2003; Atkins et al., 1999; Teplin et

al., 2002). For example, one large international meta-analysis reported that male and female youth in

juvenile detention were approximately 10 times more likely to suffer a psychiatric illness than their

non-offender peers (Fazel et al., 2008). It is possible that the reliability of the ABCL in this setting was

compromised by some worker’s relative lack of knowledge of participants.

Achenbach and Rescorla (2003) indicated that the ABCL can be completed by a person who

knows the individual well (e.g., partners, friends, roommates, therapists, parents, etc). Respondents

are required to rate each item based on the preceding six months. Given that participants were

serving a custodial sentence at the time of this study, access to informants from their personal lives

was impractical. Moreover, the present study was interested in the observations of the same

informant from pre-treatment to post-treatment to six month follow-up. Although the custodial

workers had an individual key worker role with detainees for the duration of their custody, there was

considerable variation between workers in the extent to which they engaged with the youth relative to

this role and more specifically their actual knowledge of the youth. Additionally, as noted by several

researchers, the selection of a suitable non-self-report measure was hindered by the lack of

psychometrically evaluated measures that have been assessed for use in correctional settings

(Shelton, Sampl, Kesten, Zhang, & Trestman, 2009; Wakai, Shelton, Trestman, Kesten, 2009). With

hindsight, and similar to the methodology employed by Nugent et al. (1997) a more reliable measure

155

of worker-reported changes in detainees aggressive and antisocial behaviour might have been

sourced through examination of worker-reported incident reports during the custody period or direct

semi-structured interviews with workers.

Cognitive Distortions and Moral Reasoning

The prediction that ART participants would report fewer cognitive distortions following

treatment was supported. The main and trend effects for the HIT total score, and the cogntive

distortions- Self-centered Thinking, Blaming Others, Minimising/Mislabeling, and Assuming the

Worst were all significant and the effect sizes were large (Cohen, 1988). Similarly, the HIT

behavioural referents- Lying, Stealing, Oppositional Defiant, and Physical Aggression showed

significant reductions from T1 to T4 with large effect sizes.

The reduction in cognitive distortions is generally consistent with findings reported by Nas,

Brugman and Koops (2005) who evaluated the effectiveness of the EQUIP program (Gibbs, Potter, &

Goldstein, 1995) with 108 males (12-21 years) in the Netherlands. The treatment and control groups

were recruited from high-security juvenile correctional facilities and all particpants had committed

serious violent offences (e.g., murder, violent robbery, rape). The EQUIP program represents an

extension of ART. Nas et al. reported significant reductions in cognitive distortions for the treatment

group but not the control group. At post-treatment the HIT Total score and the subscales- Self-

centred, Minimizing/Mislabeling, Blaming Others, Lying, and Stealing, showed significant reductions.

However, in contrast to the present study, Nas et al. did find significant reductions on Assuming the

Worst, Oppositional Defiant, or Physical Aggression. This inconsistency might be accounted for by

the offenders being higher risk than those in the current study. None of the participants in the present

study had committed rape or murder. Given that Nas et al. did not report longitudinal follow-up, it is

156

not known if the treatment gains were maintained. In the present study, the follow-up data indicated

that the reductions in cognitive distortions were maintained at two-year follow-up.

The present study did not include a direct measure of moral reasoning, instead, the HIT was

used as a measure of primary (i.e., self-centered) and secondary (i.e., blaming others,

minimising/mislabeling, and assuming the worst) cognitive distortions. As previously discussed, these

have been researched extensively with respect to their role in preventing cognitive dissonance and

thereby maintaining antisocial attitudes and behaviour (Barriga et al., 2000; Gibbs, 2003; Liua,

Barriga & Gibbs, 1998). In addition, there is a substantial body of research in which the moral

reasoning of offenders has been shown to be arrested at the early egocentric stages of moral

development such as Stage 1- “might makes right” and the pragmaticism of Stage 2- “you scratch my

back…” (Blasi, 1980; Nelson et al., 1990; Stams et al., 2006). Moreover, Selman’s (1971) stages of

social perspective taking were recognised by Kohlberg (1976) as a necessary precursor to the

development of more mature moral reasoning. Without having assessed moral reasoning

specifically, the present study cannot know the extent to which, if any, participation in ART has

progressed the socio-moral development of participants. However, according to Barriga et al. (2001)

and Gibbs (2003) the cognitive distortions assessed by the HIT represent a corollary of

developmental delays in sociomoral and social perspective taking that is characterised by a

pervasive egocentric bias. Thus, the reductions in cognitive distortions reported by the youth in this

study suggest evidence of what Gibbs (2003) referred to as movement from immature moral

superficiality toward more “social decentration or social construction” (p.32) or what Kohlberg (1969)

called higher levels of moral reasoning. Goldstein et al. (1998) defined the aims of moral reasoning

training in exactly these terms. Furthermore, there was evidence that the reductions in cognitive

distortions were clinically significant. Participants scores on the HIT at T1 placed them within the

157

subclinical range, and more often in the clinical range, whereas by T3 (6-month follow-up) all scores

had fallen to within the “normal” range of scores (Barriga et al., 2001).

Numerous researchers (e.g., Barriga et al., 2000; Day & Howells, 2008; Goldstein et al.,

1998; Palmer, 2007; Yochelson & Samenow, 1976) have indicated that a priority for offender

treatment programs should be the correction of cognitive distortions and the promotion of social

perspective-taking opportunities, particularly with serious violent offenders. However, as noted by

Palmer (2007) few studies have undertaken component analysis in order to determine the additive

value of moral reasoning training procedures to standard anger management interventions for violent

offenders. The substantial reductions in self-reported aggression, impulsivity and cognitive distortions

for partcipants in the present study provide evidence to support moral reasoning as part of a multi-

component CBT treatment approach. Further support is also evident for the argument that the

perspective taking opportunities provided through the Moral Reasoning component of ART should be

considered an essential component in offender rehabilitation (Potter et al., 2001).

Social Problem Solving

Subsequent to the pilot study, the SSRS (Gresham & Elliot, 1990) was replaced with the

SPSI-R:S (D'Zurilla et al., 2002) to account for the increased age range (i.e., 17-20 years) of

participants (the SSRS is not designed for use with individuals older than 18-years). Use of the SPSI-

R:S was also consistent with the broader theoretical conceptualisation and measurement of social

competence, which has been variously defined as comprising a combination of behavioural, cognitive

and affective skills (Bierman & Welsh, 2000; Cook et al., 2008). McGuire (2001) indicated that on a

continuum of cognitive to behavioural CBT interventions, social skills training is more behaviourally

oriented whereas social problem solving interventions are more closely aligned to cognitive theory

158

and methods of behaviour change (see Figure 2, p.29). However, both social skills and social

problem solving represent interdependent components of the social competence construct, with

improvement in one suggesting improvement in the other (Cunliffe, 1992; McGuire, 2001). Given the

multiple targets of cognitive and behavioural change in ART, the SPSI-R:S was considered to be a

suitable measure of the cognitive change aspects of social competence (i.e., social problem solving).

In addition, it was not possible to source an actuarial measure of social skills suitable for use in a

correctional setting with young adult offenders.

The results of the main study show partial support for the predicted increase in social

problem solving abilities among participants. There was a significant increase in general social

problem solving abilities (i.e., SPSI-R:S Total scale) at post-treatment (T2). However, none of the

other main effects for the SPSI-R:S subscales showed significant change at the conservative p-value

of .01. D'Zurilla et al. (2002) defined the Total scale as a global indicator of problem solving ability.

The higher scores at post-treatment for participants in this study indicated an increased capacity for

constructive problem solving. The improvement in global social problem solving and the maintenance

of those gains at six and 24-months follow-ups was consistent with Sukhodolsky et al.’s (2005) study

which showed improvements in parent reported problem behaviors, including aggression, and child-

reported anger and social problem solving at post-treatment and three month follow-up assessments.

This was for both treatment groups, which involved a modified version of ACT (Feindler & Ecton,

1986) in which SST components from ART or Social Problem Solving Training (SPST) components

were added.

As previously discussed, in order to reduce the potential for Type I, familywise error (i.e.,

false positive) a conservative p-value of .01 was set for significance testing in the main study.

However, the main effects for the adaptive social problem solving subscales PPO and RPS were

159

close to significance at the less conservative p = .02. In contrast, the dysfunctional problem solving

subscales NPO, ICS, and AS were not even close to reaching significance. These results suggest

that the significant improvement in global social problem solving was largely attributable to increased

adaptive social problem solving abilities, not reduced dysfunctional social problem solving. However,

these results need to be considered in light of the lack of clinical significance on the SPSI-R:S at any

time point, including at intake (T1). The interpretive guidelines provided by D'Zurilla et al. (2002)

indicate that the group means from T1 to T4 on all of the domains assessed by the SPSI-R:S were

within the “norm group average”. Thus, while there was statistically significant improvement in global

social problem solving, the group means for participants in the present study indicate that on average

their social problem solving abilities were comparable to “normal” age-related peers. These results

stand in contrast to previous research indicating that violent offending is associated with clinically

significant social problem solving deficits among juvenile (Cunliffe, 1992; Dodge et al., 1990;

Goldstein, 1999; McMurran, et al., 2005; Slaby & Guerra, 1988) and adult (McMurran, 2005;

McGuire, 2001) offenders. There are several possible theoretical and methodological explanations for

the lack of clinical significance.

Similar to the findings of the present study, Hornsveld (2005) reported that participants in his

evaluation of ART (adapted for forensic patients) showed slightly above average social competency

scores, based on Dutch norms. Hornsveld noted that this finding was inconsistent with behavioural

observations made by program facilitators and research evidence in which deficits in social

competency among violent offenders has often been reported. An association between aggressive

behaviour and unrealistic positive self-evaluations has been reported by several researchers (e.g.,

Baumeister et al., 1996; Bushman & Baumeister, 2002; Orobio de Castro, Brendgen, Boxtel, Vitaro,

& Schaepers, 2007). Contrary to the popular belief that low self-esteem underlies aggression,

160

Bushman and Baumeister found that violent offenders scored significantly higher on measures of

narcissism than non-offenders, and that there was no significant difference in self-esteem scores

between the two groups. They argued that narcissism, not self-esteem, is related to violent offending

and that such offenders are likely to self-evaluate more positively than evaluations relating to social

competence made by others. Similarly, overestimation of self-reported social competence, compared

to peer and teacher reports, has been reported relative to the social-information processes of

aggressive children (Dodge, 1980; Lochman & Dodge, 1998; Orobio de Castro et al., 2007).

Consistent with previous research that has demonstrated a link between violent offending

and increased narcissism, participants in this study might have rated themselves as more socially

competent than they actually are Overall, the behavioural observations of participants in the present

study indicated support for the contention that violent offenders are likely to evidence deficits in age

appropriate social competencies. Moreover, consistent with an over-estimation of self-reported social

competencies, participants frequently bemoaned the repetitive (i.e., over-learning) and simplified pro-

social skill steps employed in ART. These complaints were often voiced within aggressive/antisocial

language and behaviour and initial rehearsal of pro-social skills again served to reinforce the

perception of the facilitators that these youth lacked social competence despite their claims to the

contrary.

Another possible explanation for the lack of clinically significant scores on the SPSI-R:S

might have been that participants in the present study were deliberately “faking good”. As with any

self-report measure, D'Zurilla et al. (2002) discussed the potential for respondents to report falsely

inflated social problem solving abilities in the interests of social desirability. Given there was no

indication of socially desirable response sets with the other measures employed in this study, it

161

seems reasonable to assume that impression management was not systematically employed by

participants when they responded to the SPSI-R:S. However, Nas et al. (2005) reported a significant

positive relationship between social desirability and social skills in their evaluation of ART.

Participants in their study with higher scores on social desirability also tended to self-report better

social skills. Future research would benefit from the inclusion of a specific social desirability measure

as a way of determining the extent to which, if any, participants employ impression management in

their response sets. Also, a non-self-report measure of social problem solving completed by workers

in a semi-structured interview would provide a more objective measure of the domain.

It is also possible that the scores on the SPSI-R:S for participants in the present study

represented an accurate self-appraisal of social problem solving abilities and that on average, in this

small sample, their aggressive behaviour was not directly related to social problem solving deficits.

D'Zurilla and Maydue-Olivares (1995) distinguished between process versus outcome measures of

social problem solving. They indicated that process measures, such as the SPSI-R:S, are designed

to assess the cognitive processes in finding appropriate solutions in problematic situations. In

contrast, outcome measures assess the actual implementation of those solutions. D'Zurilla and

Maydue-Olivares indicated that the cognitive steps involved in social problem solving processes may

or may not be correlated with solution implementation depending upon situational factors, individual

expectancies, and affective responses to the perceived social problem. This aligns with social

information processing (SIP) models of aggressive behaviour that specify a series of complex mental

steps before an individual reaches the final stage of solution implementation (Dodge, 1994; Dodge,

1986; Dodge et al., 2008). Future evaluations of ART would benefit from the inclusion of an outcome

social problem solving measure in order to assess actual solution implementation. The Interpersonal

Problem-Solving Assessment Technique (IPSAT; Getter & Nowinski, 1981)) which is a semi-

162

structured paper-pencil measure, might be an appropriate measure of this sort, given the sample

characteristics of the present study. The IPSAT assess both social problem solving processes (i.e.,

generation of alternative solutions and decision making) and outcomes (i.e., respondents are

required to indicate which solution they would implement given a series of socially problematic

vignettes) (D'Zurilla & Maydue-Olivares, 1995).

Impulsivity

The linear reduction for impulsivity from pre-treatment to 24-month follow-up was significant

and the effect was large. The reduction was also clinically significant. At intake (T1) the mean of

impulsiveness for participants in the present study was above that of the normative average for

similar aged males reported by Eysenck et al. (1985). This finding is consistent with the results of

previous research in which impulsivity has been related to the development and maintenance of

aggressive behaviour (Barratt, 1994; Eysenck & Eysenck, 1977; Eysenck & McGurk, 1980; Moffitt,

1993; White et al., 1994) and as a dynamic risk factor (i.e., changeable) among offender populations

(Andrews, 1996). At six month follow-up (T3) the average level of impulsiveness for participants in

this study was comparable to the normative average, with a further, though non-significant, reduction

at the two year follow-up (T4).

The initially elevated scores on impulsivity for participants in this study are consistent with

other research showing heightened impulsivity among delinquent youth in general (Carroll et al.,

2006; Luengo et al., 1994) and violent offenders, in particular (Smith & Waterman, 2006). Luengo et

al. compared scores for three (non-delinquent, minor-delinquent, and major-delinquent) groups of

adolescents (12-18 years) on the children’s version of I7 Impulsiveness scale (Eysenck et al., 1985).

Results showed that both delinquent groups were higher on impulsivity than the non- delinquent

163

group. In addition, higher impulsivity was associated with increased risk of future antisocial

behaviour. Carroll et al. reported similar findings in comparing groups of early-onset offenders, late-

onset offenders, and non-offenders among Australian adolescents. Although there were no significant

differences between the offender groups, offenders were significantly higher on impulsivity than non-

offenders. In assessing impulsivity among violent versus non-violent adult offenders, Smith and

Waterman reported an increasing trend on aggression scores from non-offenders, to non-violent

offenders, to violent offenders. Moreover, the violent offenders were significantly more impulsive than

the other two groups. Smith and Waterman concluded that heightened impulsivity can reliable

distinguish offenders from non-offenders, and violent offenders from non-violent offenders. This

supported the argument that impulsivity, as a dynamic crimnogenic risk factor, is an appropriate and

important target of change among violent offenders (Andrews, 1996).

From a theoretical perspective, the present results indicate further support for the

effectiveness of the CBT training procedures (e.g., self-instructional training, think-out-loud, problem

solving, consequential thinking) that have been shown to reduce impulsiveness and aggression with

children and adolescents (Baer & Nietzel, 1991; Luria, 1961; Kendall & Finch, 1978; Meichenbaum &

Goodman, 1971). Moreover, the present results extend this earlier research by demonstrating that

these training strategies can be effective in reducing impulsivity and aggression among older violent,

young-adult offenders. According to Eysenck (2004) the I7 impulsiveness scale represents a measure

of dysfunctional impulsiveness associated with cognitive deficits such as lack of planning and

consequential thinking, and antisocial behavior, including aggression. Ramirez and Andreu (2006)

suggested that the lack of consequential thinking among impulsive children and youth is associated

with a tendency to select small, immediate rewards over larger, delayed rewards. Similarly, in

describing the inability of aggressive and antisocial youth to delay gratification from within a self-

164

centered worldview, Carducci (1980) described such youth “as fixated at a level of getting their own

throbbing needs met, regardless of the effects on others” (p.157). Here Carducci alluded to the

inability of impulsive youth to think beyond short-term, self-centred consequences. By implication, it is

possible that the three components of ART can, as Goldstein et al. (1998) suggested, replicate the

early developmental processes that lead to emotional self-regulation and mature socio-moral

reasoning. These replication procedures might serve as the basis for the remediation of aggression-

related cognitive and behavioural deficits, even among older offenders, as in the present study.

As previously noted, despite impulsivity having been identified as a significant risk factor for

non-violent and violent offending (Andrews, 1996; Carroll et al., 2006; Luengo et al., 1994; Waterman

& Smith, 2006) there are few published studies in which it has been investigated as a treatment

outcome among offender populations. There is limited research, therefore, with which to compare the

present findings. The reduction in self-reported impulsivity for the current youth was similar to the

findings reported by Goldstein and Glick (1987). In their pre/post evaluation of ART, they found that

worker-rated (not self-report as in the present study) impulsivity scores showed significant reduction

at post-treatment for ART participants (n=24) compared to a no-treatment (n=12) and a “brief

instruction” (n=24) control groups. These results were replicated with the 36 control participants who

later completed the ART program. The present study extends the findings of Goldstein and Glick by

providing follow-up assessments at six- and 24- month follow-up in which the reductions in impulsivity

were maintained and by demonstrating that ART can reduce impulsivity among an older, young-adult

group of violent offenders.

165

Maintenance of Treatment Gains

The present study posed the research question- are ART-treatment gain/s maintained over

time (i.e., from treatment-end [T2] to 24-month follow-up [T4])? Support for therapeutic maintenance

was indicated by the significant linear trend effects showing reductions in self-reported aggressive

behaviour and thoughts, cognitive distortions, impulsivity, and a significant increasing trend for

general social problem solving abilities. Furthermore, from pre-treatment to two-year follow-up all

significant effects were in the predicted direction; at no time was there a marked deterioration on

cognitive or behavioural outcomes. Consistent with these results, Hornsveld (2005) found that ART

participants reported significant reductions on hositility and aggressive behaviour at treatment-end

and that those gains were maintained at 15-week follow-up. More often, maintenance of treatment

gains among offender groups has been reported in terms of reconviction rates for ART participants

compared to untreated controls. For example, in the largest evaluation of ART to date (N=1,500)

Barnoski (2004) reported a significant reduction in recidivism for juvenile offenders who completed

ART (18.8%) compared to a no-treatment control group (24.8%) at 18-month follow-up. Similarly,

McGuire and Clarke (2004) reported a reconviction rate of 20.4% for ART participants, compared to

34.5% for controls at one year follow-up of adult offenders in the UK. At 10-month follow-up, Hatcher

et al. (2008) reported on overall decrease of 13.3% in reconviction rates for an ART treatment group

compared to the no-treatment control. A few researchers (Goldstein et al., 1998; Goldstein &

Martens, 2000; Hollin, 1999) have noted that the investigation of treatment maintenance, or

longitudinal follow-up, has been relatively lacking in the offender “what works” literature.

Several procedures have been described to enhance the transfer and maintenance of

therapeutic gains (Goldstein et al., 1998; Goldstein & Martens, 2000). With respect to transfer, there

166

are five core techniques that have been incorporated into the training procedures of ART: (1) General

Principles- the trainee is provided with rules, strategies, and organising principles that lead to

successful performance of the skill in real life situations and the ability to adapt these principles to

various settings and players; (2) Overlearning- here the trainee is given repeated opportunities to

practice a given skill set, initially after observing an exemplar, then in his own role-play’s and again

through observation or participation as a co-actor in the role-plays of other group members. Goldstein

et al. stated that “to maximize transfer through overlearning, the guiding rule is not “practice makes

perfect’”(implying that one simply practices until one gets it right and then moves on) but “practice of

perfect makes perfect’ (implying that numerous trials of correct responses are needed after the initial

success) (p.160); (3) Identical Elements- the closer the practice environment resembles the actual

environment the higher the probability the skill will be transferred; (4) Stimulus Variability- transfer is

enhanced by practicing the skill in a variety of situations; and (5) Mediated Reinforcement (or real life

reinforcement) - transfer of a newly acquired skill is enhanced when the skill is repeatedly reinforced

in real life. According to Goldstein et al. such reinforcement is mediated by the trainee, not by others.

This is achieved through teaching the trainee procedures including self-reinforcement, self-

instruction, self-correction, and self-rewarding. Each of these transfer enhancers are built into the

ART curriculum.

Goldstein and his colleagues (1998; 2000) have also emphasised the importance of

engaging significant others (e.g., family members, pro-social peers, and support workers) in the

treatment process. They referred to these agents of change as Transfer Coaches, who have the

primary role of providing support, encouragement and reinforcement outside the treatment setting.

Evidence suggests that the greater the number of external support systems recruited for these

purposes, the greater the chance of transfer and maintenance of pro-social gains (Goldstein &

167

Martens, 2000; McGuire, 2003). Recruitment of significant others in the treatment process has also

been found to support the use of relapse prevention strategies for several problem behaviours,

including substance abuse (Marlett & Gordon, 1985; Peters, 1993) sex offences (Launay, 2001;

Polaschek, 2003; Price, 1999) and violence (Bush, 1995). While Bush argued that the onus of

maintenance of treatment gains is the responsibility of the offender, staff should be trained in roles

similar to that of the transfer coach. In his research, the role of the treatment team, which included

parole and correctional officers, was to ensure that instruction in “self-risk management” strategies

were built into program content and maintained at post-release.

As already noted, a key learning from the pilot study was the importance of ensuring ongoing

communication, in the form of progress reports and program updates, between the custodial workers

and the ART facilitators. This exchange of information was instrumental in facilitating the workers role

as Transfer Coach, which undoubtedly contributed to skill transfer and maintenance. Goldstein et al.

(1998) argued that the frequently reported failure of behaviour change programs to show

maintenance of gains beyond the treatment setting can in part be explained by the ultimate return to

a family, peer group and/or community that is unfamiliar, unable and/or unwilling to engage the youth

in the newly acquired pro-social skills and attitudes. Thus, the well learned aggressive and antisocial

cognitions and behaviours are rapidly reinstated. The present study did not engage participants

significant others in the role of Transfer Coaches post-release. This represents a limitation in the

present study with respect to ensuring relapse prevention strategies were in place prior to release

back into the community. Despite this, there was evidence of treatment maintenance at two year

follow-up for these youth, which stands as testimony to the multimodal training procedures in ART. In

addition, the evidence for treatment maintenance provides further support for the criminogenic risk

168

principle which indicates that higher risk offending behaviour (such as violent offending) requires

more intense intervention, such as ART.

Group Dynamics: Process versus content

Interest in the therapeutic process has only recently started to receive attention in the CBT

literature and much less in the offender treatment literature (Marshall, 2009; Marshall & Burton,

2010). To date, the overarching goal of offender treatment programs has been the prevention or

reduction of recidivism (Marshall & Serran, 2004; Tate et al., 1995). The emphasis on criminogenic

risk and need has resulted in more attention been given to program content and technique rather

than to process variables, including the impact of the therapeutic relationship on treatment outcome

(Holmqvist et al., 2007; Marshall & Burton, 2009; Ross, Polaschek, & Ward, 2007). However, the

importance of the therapeutic alliance in predicting successful therapeutic outcome has been well

established in individual psychotherapy (Horvath & Luborsky, 1993; Horvath & Symonds, 1991;

Raue, Goldfried, & Barkham, 1997). Large meta-analyses have repeatedly shown that the quality of

the therapeutic relationship accounts for up to 30% of the variance in psychotherapy outcome studies

(Miller, Duncan, & Hubble, 1997).

Ross et al. (2007) defined the therapeutic alliance (TA) in terms of “a collaborative

relationship between therapist and client that can facilitate positive change for the client” (p.463).

They further argued that the operationalisation of the TA in the treatment of offender populations

does not differ significantly from other client groups. Bordin (1979) conceptualised the TA as a

working alliance between the therapist and client, comprising three core features that are negotiated

between therapist and client: (1) an agreed set of therapeutic goals; (2) agreement in undertaking

specific tasks aimed at attainment of those goals; and (3) a bond between therapist and client that is

169

characterised by warmth, empathy, support, and understanding. “The effectiveness of therapy is a

function in part, if not entirely, of the strength of the working alliance” (Bordin, 1979, p.253).

According to Andrews and his colleagues (Andrews & Bonta, 2003; Andrews et al., 1990; Bonta &

Andrews, 2003) non-criminogenic needs, including the therapeutic alliance, fit within the principle of

responsivity. This aspect of the “what works” principle speaks largely to the relational processes that

underpin effective intervention with offenders. The importance of relational/process variables in the

delivery of ART became increasingly evident throughout the duration of each program, and had

important implications for improving treatment effectiveness throughout the three year intervention

period comprising this study.

The concept of cohesiveness is also a core relational component in the effective outcome of

group interventions (Hornsey, Dwyer, & Oei, 2007; Marmarosh, Holz, & Scottenbauer, 2005; Yalom,

1995). According to Yalom (1995) the complexities of the relational aspects of therapy are increased

in the group setting by multiple intra- and inter- personal relationships and roles (i.e., therapist, co-

therapist, and several clients). This issue is further complicated in working with involuntary clients, as

was the case in the present study (Goldstein & Martens, 2000; Trotter, 2006). Yalom distinguished

between the process and content of group therapy in terms of the explicit material presented,

including the task at hand (i.e., content) compared to the meaning/s underlying that material relative

to the interpersonal relationships in the “here and now” of the group (i.e., process). In the early stages

of delivering ART, in particular the first pilot program, there was an emphasis upon program content

over process.

In Bordin’s (1979) conceptualisation of the working alliance, the first two core features- goals

and tasks are for the most part predetermined by the principal aim (i.e., aggression replacement) and

content of ART. However, Goldstein and his colleagues (1998) were clear that the prescriptive

170

programming within ART does not preclude ensuring participants are given an opportunity for

programmatic “buy in”. While a standard 10-week ART curriculum was employed in the present

study, amendments were made to the social skills taught and the moral reasoning dilemmas selected

for discussion within a given group depending upon the needs of the group. Goldstein et al. indicated

that such amendments have proven to be an effective method in increasing trainee motivation.

Additionally, establishing the need for a given skill set represents part of the “buy in” process. For

every new skill (cognitive, behavioural and affective) taught throughout the ART curriculum, the

therapist poses the question “why is it important to be able to… [e.g., identify triggers and body

signs]”. With experience in delivering ART, the co-therapists became increasingly aware that such

open ended questions all too often were met with a glassy-eyed silence from group members. Hence

the “why” question was reframed as “when was the last time being able to… [e.g., think ahead] would

have been important for you to know”. The nuance in the later wording of this question reflected our

assumption that the group members needed this particular cognitive, behavioural and/or affective skill

and also ensured that each participant identified a personally relevant scenario for the role-plays. The

opportunity to negotiate with the co-trainers regarding the content of the program provided in-vivo

practice opportunities for several targets of change within ART. These included pro-social modeling,

social skills training, social problem solving and exercising affective self-control within the group

relationship. The following case illustrates this point.

L. was serving a custodial sentence for a serious assault against another young man whom

L. claimed had bullied him at school (i.e., Blaming Others; MinimsingMislableing). L. had

consistently demonstrated resistance to participation in ART, frequently indicated that he had

no use for and was above the program content of ART. By approximately mid-way through

the 10-week curriculum, he attended group and described an altercation he had had with a

171

staff member regarding a request to have a protein drink made available to detainees at the

centre. The staff member had indicated that this would not be possible as it was against

Centre policy. In response L. became verbally abusive. After initial angry complaining by

group members, a discussion led the group, and L, to concede that his aggressive approach

was ineffective and perhaps an attempt at negotiation might prove more fruitful. [The

skillstreaming skill of negotiation is considered an “advanced’ social skill and is generally

reserved for the later stages of an ART program, or in some instances is beyond the grasp of

less socially skilled participants]. Hence it was agreed by the group that they would use this

session for the practice of the skill “negotiation”. After L. had practiced his scenario in a role-

play he indicated that he felt confident to attempt using the skill with Centre staff. At the next

session, L reported that staff had been responsive to his non-aggressive approach and that

they would put his request to the Centre manager for consideration. By the following week, L

reported to the group that his request had been approved, with some conditions.

This case example highlights the need for flexibility on the part of the therapist in delivering

interventions in a correctional setting. Andrews et al. (1990) referred to this in the “what works”

principles of Responsivity and Professional Discretion. The value of such flexibility for the client was

further evident in his subsequent behaviour and attitude within ART. While his resistance continued

to be evident at other times in the group, there was a notable shift in his willingness to participate and

contribute to group cohesiveness.

Andrews and Kiessling (1980) identified five core characteristics of correctional staff that

were thought to increase therapeutic effectiveness with offenders. They defined the first component

in terms of effective use of authority, characterised by a “firm but fair” approach in interacting with

172

offenders. Secondly, service providers should model and reinforce pro-social values, attitudes and

behaviours. The third aspect involves directly teaching problem solving skills to offenders, such as

consequential thinking and alternative solutions. The fourth characteristics involved demonstrating

effective use of community resources through brokerage, advocacy and referral. The fifth component,

described as the most important, was summarised as relationship factors comprising open, warm,

and enthusiastic communication by service providers. These relational qualities foster mutual respect

which directly and positively influences treatment outcomes (i.e., therapeutic alliance or cohesiveness

in the group setting). As shown in the study by Nugent et al. (1997) the failure of a service provider to

model these prosocial relational qualities can seriously compromise treatment integrity. In this study,

the aggressive behaviour toward clients by a group leader was likely to have contributed to the failure

of the program to show maintenance of treatment gains.

The few studies in which the relationship between TA and treatment outcomes has been

investigated specifically among young offenders have reported mixed findings. Chassin, Young and

Light (1980) found that practical support, such as task and goal orientated approaches, was indicated

by young offenders as more important to them than insight or interpersonal orientated therapies.

Similarly, Gold and Osgood (1992) found that the most effective treatment outcomes for juvenile

detainees were associated with a staff approach that was authoritative, like Andrews and Kiessling’s

(1980) “firm but fair” approach, rather than an approach that focused upon the emotional problems of

young offenders. Florsheim, Shotorbani, Guest-Warnick, Barratt and Hwang (2000) examined the

relationship between working alliance in the treatment of 121 delinquent boys (M age = 15.6 years)

placed in community-based residential programs. Measures of the working alliance between a key

staff member (as nominated by the individual youth) and each boy were administered along with the

CBCL (Achenbach, 1991) parent, teacher and youth self-report versions. Florsheim et al. reported

173

that a positive working alliance at three months into treatment predicted significant improvement on

the CBCL and recidivist measure at one year follow-up. They also reported a positive relationship

between treatment effectiveness and the working alliance.

The findings of Florsheim et al. (2000) are consistent with the view that the working alliance

(Bordin, 1979) or group cohesion (Yalom, 1995) is as much of a pragmatic concern in successful

treatment outcome as is the explicit goal/s or task/s therein. This was the certainly the experience of

the present researcher in the delivery of ART. The importance of attending to both process (i.e.,

group cohesion, working alliance) and content (i.e., the explicit manualised tasks) in the service of

treatment integrity was indicated in the ART QA assessment feedback provided by the Washington

State Master Trainer (see Appendix C). Moreover, the ongoing use of the ART self-assessment

session checklists by the co-trainers in this study not only ensured adherence to treatment integrity

but also provided a framework for reflective practice relative to the group processes. These practice

reflections provided the material for much of the discussion in the fortnightly supervision attended by

the ART co-trainers, which in turn ensured role and task clarity and treatment integrity.

Clinical and Policy and Planning Implications

As noted in the introduction, violent offenders account for 25% of the current young-adult

(18-25 years) prison population in Australia (ABS, 2009) and recent Australian crime data suggested

that assault-related crime is increasing at a rate of 5% for each year between 1995 and 2007 (AIC,

2009). These crime statistics highlight the need for effective, evidence-based interventions aimed at

violent offenders in Australia. Additionally, it is imperative that these interventions be evaluated

among Australian populations (Ogloff, 2002). The results of this study provide preliminary evidence

that the American developed ART program is effective in treating Australian violent young-adult

174

offenders in a custodial setting. This in turn suggests at least some reduction in the burden of

financial, psychological and social costs associated with aggressive and antisocial behaviours

(Werry, 1997).

Perhaps the most important clinical implication of the present study was that the delivery of

ART in an institutional setting, as opposed to a community setting, produced generally positive and

clinically significant changes in self-reported aggressive behaviours and thoughts, ciminogenic

thinking errors, level of impulsivity, and social problem solving abilities. Numerous researchers (e.g.,

Andrews et al., 1990; Izzo and Ross, 1990; Lipsey et al., 2001; Wilson et al., 2005) have reported

that offender treatment programs delivered in a community setting are likely to produce better

outcomes in terms of recidivism compared to those delivered in custodial settings. Although the

present study did not compare ART outcomes with respect to treatment setting (i.e, community

versus custodial) it is clear that ART can be highly effective in a custodial setting despite the limited

opportunity to practice the newly acquired prosocial skills and a ubiquitous antisocial peer influence.

The potential for peer-contamination when aggressive/antisocial youth are aggregated

together for extended periods in institutional settings has been well documented (Dodge, Lansford, &

Dishion, 2006; Greenwood, 2006; Lipsey, 2006; Osgood & Briddell, 2006). The results of the present

study have implications with respect to the timing of cognitive-behavioural interventions within these

settings. Most of the youth involved in this study were released shortly after completion of ART and

subsequently returned to their respective communities, families, peers and intimate relationships. The

previously discussed maintenance of treatment gains at two year follow-up, suggests that the youth

did apply at least some of the newly acquired pro-social skills, attitudes and behaviours in their real-

life environment at the two year follow-up (i.e., transfer and maintainence). This provides further

175

support for the cognitive-behavioural content and procedures in ART, even in the case of later

intervention, as in the present study, with violent young-adult offenders in an institutional setting.

Previous evaluations of ART have assessed outcomes relative to participants in mid-

adolescence (Goldstein & Glick, 1994; Gundersen & Svartdal, 2006; Barnoski, 2004) or older

offenders (Hatcher et al., 2008; McGuire & Clark, 2004). Given the age range of participants in the

present studies (17- 20.5 years) there is now evidence that ART and its modes of change (i.e., social

learning principles and procedures) are effective for early-adult male offenders. The positive

individual effects of participation in ART and the maintenance of those gains at two year follow-up

suggest potential improvements in at least some of the multiple individual, familial, and social costs of

aggressive and antisocial behavior described by Werry (1997). At two year follow-up one participant

commented that he had on several occasions used the ACT technique of “If… Then… Thinking

Ahead” (i.e., consequential thinking) to avoid escalating interpersonal conflict with his partner. In

terms of reoffending, of the 14 participants who could be contacted at two year follow-up, only two

self-reported that they had reoffended, and only one of those reported an assault-related offence.

Twelve of the 14 participants indicated that they were currently employed at two year follow-up.

As well as clinical implications, the results of this study have important implications relative to

social policy and planning. The three year ART intervention and evaluation period comprising this

study was funded by the Victorian Department of Human Services - Youth Justice Division.. The

funding followed from the recommendation made in the Victorian Juvenile Justice Rehabilitation

Review (Day et al., 2003) that ART, as an evidence-based program for violent offenders, be

implemented across the State’s youth justice system. At the time of writing these concluding

comments, ART is no longer being delivered in the State of Victoria. Hollin and Palmer (2006) noted

the difficulty for researchers in conducting longitudinal outcome studies in that the length of the

176

research can outlast a government’s term in office or sometimes rapidly changing policy directions.

They suggested that these circumstances can lead to “policy-led evidence” as apposed to “evidence-

led policy” (p.267). The two-year evaluation of ART in Washington State juvenile justice (Barnoski,

2004) was the culmination of a 10 year implementation and dissemination process. That process

included a policy and legislative commitment to evidence-based programs, including ART, and the

training of multiple ART facilitators at every juvenile justice county court (i.e., “evidence-led policy”). It

is difficult to determine the extent to which the results of this study can influence social policy and

planning in Australia. However, Hollin and Palmer indicated that research such as this can, and

should play an important role in contributing to evidence-led policy and that there is a responsibility

and need for clinician-researchers to take up that role. Simiarly, Goldstein (1998; 1999; 2004)

strongly advocated the role of the ART trainer as a “scientific practitioner”. A role that he argued

required a commitment to empirical evaluation of the program and dissemination of those findings to

other practitioners and researchers and policy makers.

Methodological Considerations: Strengths and limitations

There were a number of limitations to the present study that may limit the generalisability of

the results. The original research design included the recruitement of participants via a staged

process and depending upon length of sentence allocation to either a waitlist control group or an

immediate (i.e., most immediate start date for next ART program) treatment group. However, several

institutional and sentencing factors meant that the number of waitlist participants was too small to

form a meaningful comparison group. The length of stay at the custody centre was the most frequent

impediment. Some explanation of the Victorian juvenile justice system is necessary to elaborate on

this limitation.

177

In the State of Victoria, where this study was conducted, there exists a diversionary approach

to youth justice known as the “dual-track” system. This system allows magistrates to request a “Youth

Training Centre (YTC) suitability” report, completed by a youth justice worker, before sentencing.

Where a young person aged 18-21 years is deemed suitable, the court can order that a custodial

sentence be served in a youth justice centre (separate from offenders aged under 18-years) as an

alternative to imprisonment in an adult correctional facility. This ensures that 50% of all custodial

sentences for youth (aged 18 to 21 years) are served at a YTC rather than adult correctional facility.

The dual-track system is built on the principles of diversion and offender rehabilitation and there is an

expectation that the young person will participate in offence specific rehabilitation programs during

the period of YTC custody. This emphasis on rehabilitation also means that the length of stay in

custody at a YTC is usually approximately half of the actual sentence, the remaining period is served

on a community based order. Hence, a young person who is given a 12-month custodial sentence

will only usually serve six months in a YTC (Hanson, 2009). This meant that a young person who was

allocated to the ART waitlist control group was often due for release before completion of the

program, or he had commenced another program during the wait period that clashed with the

timetabling of ART.

Another limitation of the present study was the lack of recidivist data. This occurred despite

exhaustive efforts on the part of the researcher to gain access to the appropriate police records

database. Informed consent was obtained in the initial plain language and consent form for the

researcher to access the youth justice electronic records. However, given the age of participants in

this study, it became evident that if they reoffended after completion of ART that most likely they

would be dealt with through the adult courts and correctional system and these offences would not be

recorded in the youth justice data. Despite having obtained additional signed consent from the 14

178

participants that could be found for the final data collection at T4, and approval from the university

and police ethics committees, numerous administrative delays from within the department

responsible for releasing the requested data meant that the recidivist data was not obtained in time

for inclusion in this thesis. Similar difficulties of program implementation, administration and research

encountered in this study have been discussed by other researchers (Hollin, 1995, 2001; Tate et al.,

1995; Wakai et al., 2009).

A number of researchers (e.g., Cotter, Burke, Stouthhamer-Loeber, & Loeber, 2005; Hollin &

Palmer, 2006; Posavac & Carey, 2003) have discussed the problem of sample attrition in longitudinal

research designs. At intake all participants agreed to provide a contact address and telephone

number of a significant other who would be most likely to know the whereabouts of the participant at

follow-up assessment points. Despite several attempts and alternate methods (i.e., cross-checking

last known contact details through youth justice data records and online telephone listings) three

participants were non contactable at the six month follow-up, and six different participants were non

contactable at the two year follow-up. It was unclear whether this attrition represented a random or

systematic (e.g., more likely to have reoffended) sample effect. However, given that different

participants were missing at each follow-up, and other than one case at the two-year follow-up, no

participant refused further participation, it seems reasonable to assume that the attrition was not due

to a systematic sample bias. Irrespective, participant losses can bias research outcomes (Posavac &

Carey).

Cotter et al. (2005) noted that there is general agreement among researchers that a capture

rate of 80% is sufficient in longitudinal research with respect to maintaining the validity of a study. At

the two year follow-up, there was a 70% capture rate for the present study. Given the small sample

size, there were limited options for dealing with the missing cases. Thus, as suggested by Streiner

179

(2002) the statistical analyses were conducted for the group mean replacement and “last observation

carried forward” (LOCF) methods. Both methods produced similar results, so the mean replacement

method was used as SPSS provides a function for this method. This method was preferential to the

loss of statistical power associated with not including the cases at T3 and T4 (Streiner, 2002;

Tabachnick & Fidell, 2001).

The present study would have benefited from a diagnostic interview that provided a clinical

diagnosis relative to the DBD’s and APD and potential comorbidities. The inclusion of such an

assessment in future research could provide an index of problem severity enabling a determination of

ART effectiveness relative to clearly defined DSM-IV-TR (APA, 2000) diagnostic criteria. Such an

approach would be consistent with an assessment of the dynamic risk factors described by Andrews

et al. (1990) that in principle are changeable and have been shown to be predicative of recidivism.

Furthermore, this type of assessment could also provide information relative to the age of onset of a

specific diagnosis or conduct-related problems, and the presence of CU traits. As previously

discussed, early onset CD and CU traits have been shown to be associated with the life-course

persistent (LCP) group of antisocial offenders who are at greater risk of long-term negative

psychosocial outcomes including criminality, violent interpersonal relationships and development of

APD (Frick, 2004; Moffitt et al., 2002). To date, there has been no known published evaluation of

ART with respect to a formal assessment of problem severity and clinical presentation.

Directions for Future Research

There are several lines of research enquiry indicated by the strengths and limitations of this

study. Replication with a larger sample, including some form of comparison control group, is needed.

A particular criticism of offender outcome studies has been the over-reliance on recidivism as the

180

sole determinant of outcome or self-report measures to the exclusion of comparative data provided

by other respondents and clinical relevant measures (Hollin & Palmer, 2006; Serin et al., 2009; Tate

et al., 1995). The present study did attempt to address these concerns by including a worker-report

measure of aggressive behaviours and thoughts (i.e., ABCL; Achenbach & Rescorla, 2003) and

multiple self-report measures relevant to ART targets of change and criminongenic needs. However,

given that the questionable reliability of the worker reports, future evaluations of ART in correctional

settings should attempt to compare self- and other- report measures. As previously discussed,

recidivist data, worker-reported incident reports, and semi-structured interviews with case workers

would be recommended here.

The advantage of assessing TA or group cohesion in future ART evaluations has already

been discussed. It may be, consistent with the findings of psychotherapy outcome studies (see Miller

et al., 1997 for review) that the effectiveness ART is highly dependent upon these relational, process

factors. Certainly, several researchers with interest in the area of offender treatment (Andrews &

Dowden, 2004; Marshall & Serran, 2004; Polaschek, 2006; Trotter, 2006) have noted that further

examination of these factors is warranted.

A recent Australian report indicated a high level of overrepresentation of indigenous, relative

to non-indigenous, youth in detention across Australia. In 2006, the national average indicated that

indigenous young people were 21 times more likely to be in detention than non-indigenous age-

related peers (Taylor, 2007). The findings of one Australian state report (Harding et al., 1995 as cited

in Howells & Day, 2002) indicted that 45% of non-sexual violent assault charges were recorded

against indigenous adults. This finding is staggering when considered in light of population estimates

showing that indigenous Australians make up only 2.5% of the total population. Participants in the

present study did not include indigenous youth. However, given their over representation among

181

young offenders in Australia, it is important that the application of ART, with respect to the program

content and processes, be evaluated among this group (Howells & Day, 2002). In a qualitative review

of appropriate intervention strategies with violent aboriginal offenders, Mals, Howells, Day and Hall

(2000) noted that while there are areas of overlap with respect to ciminogenic risk and need between

indigenous and non-indigenous violent offenders, there are important differences relative to the

principle of responsivity. In particular, they suggested that differences in literacy, increased incidence

of alcohol fueled and family violence, and cultural norms- particularly with rural compared to urban

aborigines, are likely to require amendments to program implementation and processes. At the time

of writing, a 12-month pilot program of ART has been completed in the Australian State of

Queensland. Here the prevalence of indigenous youth (10-17 years) in detention is higher than non-

indigenous youth. Initial anecdotal reports indicate that the content of ART is applicable, with some

amendments to language and variations in cultural norms portrayed in the moral reasoning problems,

with young indigenous offenders in this State. However, a systematic research protocol is yet to be

undertaken.

Summary and Concluding Comments

The results of this study provided support for the predicted reductions in aggressive

behaviour and thoughts, antisocial and aggressive cognitive distortions and impulsivity. Some

support was also found for improved social problem solving abilities on the youth self-report

measures following ART participation. The changes on these measures were evident at treatment-

end (T2) and were maintained at the six month (T3) and two year (T4) follow-ups. Hollin and Palmer

(2006) noted that assessment of the relationship between statistical and clinical outcomes in offender

research is particularly lacking. Consistent with others (e.g., Kazdin, 1999; Kendall et al., 1999; Wise,

182

2004) they argued that the assessment of clinical significance is necessary to determine treatment

effects with respect to meaningful long-term outcomes. The present study has contributed to the

existing “what works” literature, and more specifically the ART outcome evaluation research, by

including clinically relevant measures of outcome and comparing the group means of participants in

the present study with the normative data available for each measure. These comparisons showed

that at each data collection point, the reductions in aggressive behaviours and thoughts, cognitive

distortions, and impulsivity were clinically significant. At the six month, post-treatment follow-up, at

the latest, all group means on these self-report measures had fallen to within the “normal” range.

The group means on social problem solving for participants in this study did not reach clinical

significance at any time. Similarly, the workers ratings of participants aggressive behaviours and

thoughts at pre-treatement, post-treatment, and six month follow-up remained within the “normal”

range at each data collection point. This finding in combination with the lack of statically significant

change on the ABCL (Achenbach & Rescorla, 2003) was discussed with respect to potentially

unreliable respondents in a correctional setting. One way of addressing this issue might be through

involving custody workers in the delivery of ART. This approach is used in Washinton State where

community and custodial workers (i.e., parole and probation officers and youth workers) are trained in

ART and participate in delivery of the program. Involving workers in the actual delivery of ART is

likely to increase their understanding of the program content and procedures, as well as increasing

programtic “buy in”, which in turn is likely to increase particpants skill transfer and maintenance of

treatment gains.

The results of this study also indicate support for the training procedures (i.e., modeling, role-

plays, “thinking out loud”, perspective taking opportunities, identification of cognitive distortions,

cognitive restructuring) employed in each of the three components of ART. In particular, Palmer

183

(2003) noted that until component analysis of ART was undertaken it was difficult to determine the

relative contribution of each of the modes of behavioural and cognitive change included in the

intervention, particularly what, if anything, the moral reasoning component adds to standard anger

management procedures. The clinically and statistical significant reductions on the measure of

cognitive distortions (i.e., HIT; Barigga et al., 2001) supported the added value of the moral reasoning

training component. This result also supported the argument that effective interventions with

chronically aggressive youth must include a component that addresses the underlying self-serving

cognitive distortions that serve to rationalise and maintain aggressive and antisocial behaviour (Day

& Howells, 2008; Barriga et al., 2000; Gibbs, 2003 Goldstein et al., 1998; Yochelson & Samenow,

1976). The present research has contributed to the clinical/forensic research by providing evidence of

the effectiveness of ART and its training procedures with a group of young –adult, moderate to high

risk offenders. Most of the previous evaluations of ART have investigated outcomes among

adolescent or adult participants. This was the first known study to provide evidence of the programs

effectiveness with young-adults (17-20 years).

While the overall effectiveness of anger management programs has been reasonably well

established, evidence supporting these interventions with violent offenders is relatively lacking (Day,

Howells, Mohr, Schall, & Gerace, 2008) Broadly in Australia, while “evidence-based practice” has

become the catch-cry of government and non-government community service agencies, including

youth justice, there are few published studies in which systematic evaluation of human service

programming has been undertaken (Ainsworth, 2001; Boni, 1999; Brann, Coleman, & Luk, 2001).

With respect to offender rehabilitation research in Australia, Ogloff (2002) noted that much of our

evidence-based practice is based upon programs that have been developed in America and that

“very real questions exist concerning the applicability of international offender rehabilitation research

184

to Australian society” (p.246). The present study has provided initial evidence to support the

effectiveness of the American developed ART program among violent, young-adult offeders in

Australia. Other than minor amendments to the language used in some components of the program

(mainly moral reasoning problems) the present research has demonstrated that ART is culturally

applicable to Australian youth.

The present results also indicated support for each of the multi-modal components (i.e.,

social skill training, anger control, and moral reasoning) that comprise ART. Palmer (2007) noted the

lack of research evidence pertaining to the effectiveness of moral reasoning training procedures

among offender populations. Through the selection of measures that assessed the specific cognitive,

affective and behavioural targets of change in ART, this study provided evidence that moral

reasoning training can reduce ciminogenic cognitive distortions, which suggested an increased

capacity for moral descion making and social perspective taking. Notably, much of the previous

research evidence for the treatment procedures included in ART has involved child or adolescent

samples. This study has provided evidence of clinical effectiveness with young adults, who present

with well practiced and chronic aggression-related behaviours and cognitions. Moreover, the reported

treatment gains were maintained at the two-year follow-up. These findings have added to the

international “what works” literature by providing further evidence in support of social learning theory

(Bandura, 1977) and its operationalisation in CBT interventions for moderate to high risk, young-

adult, violent offenders.

185

References

Abram, K. M., Teplin, L. A., McClelland, G. M., & Dulcan, M. K. (2003). Comorbid psychiatric

disorders in youth in juvenile detention. Arch Gen Psychiatry, 60, 1097-1108.

Abrantes, A. M., Hoffmann, N. G., & Anton, R. (2005). Prevalence of co-occurring disorders among

juveniles committed to detention centers. International Journal of Offender Therapy and

Comparative Criminology, 49, 179-193.

Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/ 4-18 and 1991 profile. Burlington,

VT: University of Vermont, Department of Psychiatry.

Achenbach, T. M., & Rescorla, L. A. (2003). Manual for the ASEBA adult forms and profiles.

Burlington, VT: University of Vermont, Research Center for Children, Youth & Families.

Ainsworth, F. (2001). After ideology: the effectiveness of residential programs for ‘at-risk’

adolescents. Children Australia, 26, 11-18.

Akers, R. L. (1999). Criminological theories: introduction and evaluation (2nd ed.). Chicago, IL:

Fitzroy Dearborn Publishers.

Amendola, A. M., & Oliver, R. W. (2003). LSCI and aggression replacement training: a mutli-modal

approach. Reclaiming Children and Youth, 12, 181-185.

American Psychiatric Association. (2000). The diagnostic and statistical manual of mental disorders-

fourth edition- text revised [DSM-IV-TR]. Arlington, VA: Author.

Andrews, D. A., & Bonta, J. (1995). The level of service inventory- revised (LSI-R). Toronto, ON:

Multi-health Systems.

Andrews, D. A., & Bonta, A. (2003). The Psychology of Criminal Conduct (3rd ed.). Cincinnati, OH:

Anderson Publishing.

186

Andrews, D. A., & Dowden, C. (2007). The risk-need-responsivity model of assessment and human

service in prevention and corrections: crime-prevention jurisprudence. Canadian Journal of

Criminology & Criminal Justice, 49, 439-464.

Andrews, D. A., & Dowden, C. C. (2006). Risk principle of case classification in correctional

treatment: a meta-analytic investigation. International Journal of Offender Therapy and

Comparative Criminology, 50, 88-100.

Andrews, D. A., & Kiessling, J. J. (1980). Program structure and effective correctional practices: a

summary of the CaVIC research. In R. R. Ross & P. Gendreau (Eds.), Effective correctional

treatment (pp. 441-463). Toronto: Butterworths.

Andrews, D. A., Bonta, J., & Hoge, R. D. (1990). Classification for effective rehabilitation:

rediscovering psychology. Criminal Justice and Behavior, 17, 19-52.

Andrews, D. A., Zinger, I., Hoge, R. D., Bonta, A., Gendreau, P., & Cullen, F. T. (1990). Does

correctional treatment work? a clinically relevant and psychologically informed meta-analysis.

Criminology, 28, 369-404.

Ang, R. P. (2003). Social problem-solving skills training: does it really work? Child Care in Practice, 9,

5- 13.

Argyle, M., & Kendon, A. (1967). The experimental analysis of social performance In L. Berkowitz

(Ed.), Advances in experimental social psychology (Vol. 3, pp. 55-98). New York: Academic

Press.

Atkins, D. L., Pumariega, A. J., Rogers, K., Montgomery, L., Nybro, C., Jeffers, G., et al. (1999).

Mental health and incarcerated youth I: prevalence and nature of psychopathology. Journal

of Child & Family Studies, 8, 193-204.

187

Australian Bureau of Statistics [ABS]. (2006). Population distribution, aboriginal and Torres Strait

islander Australians. Canberra, ACT: Author.

Australian Bureau of Statistics [ABS]. (2008). Prisoners in Australia. Canberra: Australia: Author.

Australian Institute of Criminology [AIC]. (2009). Australian crime: facts & figures 2008. Canberra,

Australia: Author.

Baer, R. A., & Nietzel, M. T. (1991). Cognitive and behavioral treatment of impulsivity in children: a

meta-analytic review of the outcome literature. Journal of Clinical Child Psychology, 20, 400-

412.

Bandura, A. (1973). Aggression: a social learning analysis. Englewood Cliffs, NJ: Prentice-Hall.

Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1986). Social foundations of thought and action: a social cognitive theory. Englewood

Cliffs, NJ: Prentice Hall.

Bandura, A. (2007). Albert Bandura. In G. Lindzey & W. M. Runyan (Eds.), A history of psychology in

autobiography: volume IX (pp. 43-76). Washington, DC: American Psychological Association

Bandura, A., Ross, D., & Ross, S. A. (1961). Transmission of aggression through imitation of

aggressive models. Journal of Abnormal and Social Psychology, 63, 575-582.

Barkley, R. A. (1995). Sex differences in ADHD. ADHD Report, 3, 1-4.

Barkley, R. A., Fischer, M., Edelbrock, C. S., & Smallish, L. (1990). The adolescent outcome of

hyperactive children diagnosed by research criteria I: an eight year follow-up study. Journal

of the American Academy of Child & Adolescent Psychiatry, 29, 546-557.

Barnoski, R. (2004). Outcome evaluation of Washington State’s research-based programs for

juvenile offenders. Washington State: Washington State Institute for Public Policy.

188

Barratt, E. S. (1994). Impulsiveness and aggression. In J. Monahan & H. J. Steadman (Eds.),

Violence and mental disorder: developments in risk assessment (pp. 61–79). Chicago, IL:

University of Chicago Press.

Barriga, A. Q., & Gibbs, J. C. (1996). Measuring cognitive distortion in antisocial youth: development

and preliminary validation of the "how I think" questionnaire. Aggressive Behavior, 22, 333-

343.

Barriga, A. Q., Gibbs, J. C., Potter, G. B., & Liau, A. K. (2001). How i think (HIT) questionnaire

manual. Champaign, IL: Research Press.

Barriga, A. Q., Hawkins, M. A., & Camelia, C. R. T. (2008). Specificity of cognitive distortions to

antisocial behaviours. Criminal Behaviour and Mental Health, 18, 104-116.

Barriga, A. Q., Landau, J. R., Stinson, B. L., Liau, A. K., & Gibbs, J. C. (2000). Cognitive distortions

and problem behaviors in adolescents. Criminal Justice and Behavior, 27, 35-56.

Bates, G. W., & Levery, B. J. (2003). Social problem solving and vulnerability to depression in a

clinical sample. Psychological Reports, 92, 1277-1283.

Beck, A.T. (1996). Beyond belief: a theory of modes, personality, and psychopathology. In P.M.

Salkovskis (Ed.), Frontiers of Cognitive Therapy (pp. 1-25). New York: Guilford Press.

Beck, R., & Fernandez, E. (1998). Cognitive-behavioral treatment of anger: a meta-analysis.

Cognitive Therapy and Research, 22, 63-74.

Beidel, D. C., Turner, S. M., Young, B., & Paulson, A. (2005). Social effectiveness therapy for

children: three-year follow-up. Journal of Consulting and Clinical Psychology, 73, 721-725.

Bellini, S., & Peters, J. K. (2008). Social skills training for youth with autism spectrum disorders. Child

and Adolescent Psychiatric Clinics of North America, 17, 857-873.

189

Berkowitz, L. (1990). On the formation and regulation of anger and aggression: a cogntive-

neoassociationistic analysis Amercian Psychologist, 45, 494-503.

Berkowitz, M. W., & Gibbs, J. C. (1985). The process of moral conflict resolution and moral

development. New Directions for Child and Adolescent Development, 29, 71-84.

Bickel, R., & Campbell, A. (2002). Mental health of adolescents in custody: the use of the ˜adolescent

psychopathology scale" in a Tasmanian context. Australian & New Zealand Journal of

Psychiatry, 36, 603-609.

Biederman, J., Newcorn, J., & Sprich, S. (1991). Comorbidity of attention deficit hyperactivity disorder

with conduct, depressive, anxiety, and other disorders. American Journal of Psychiatry, 148,

564-577.

Biederman, J., Petty, C. R., Dolan, C., Hughes, S., Mick, E., Monuteaux, M. C., et al. (2008). The

long-term longitudinal course of oppositional defiant disorder and conduct disorder in ADHD

boys: findings from a controlled 10-year prospective longitudinal follow-up study.

Psychological Medicine, 38, 1027-1036.

Bierman, K. L., & Welsh, J. A. (2000). Assessing social dysfunction: the contributions of laboratory

and performance-based measures. Journal of Clinical Child Psychology, 29, 526-539.

Bird, H. R., Gould, M. S., & Staghezza, B. (1992). Aggregating data from multiple informants in child

psychiatry epidemiological research. Journal of the American Academy of Child and

Adolescent Psychiatry, 31, 78-85.

Bird, H. R., Gould, M. S., & Staghezza-Jaramillo, B. M. (1994). The comorbidity of ADHD in a

community sample of children aged 6 through 16 years. Journal of Child and Family Studies,

3, 365-378.

190

Blasi, A. (1980). Bridging moral cognition and moral action: a critical review of the literature.

Psychological Bulletin, 88, 1-45.

Boni, N. (1999). Youth and serious crime: directions for Australasian researchers into the new

millennium, Children and Crime: Victims and Offenders. Melbourne, Aust: Australian Institute

of Criminology.

Bonta, J., & Andrews, D. A. (2003). A commentary on Ward and Stewart's model of human needs.

Psychology, Crime & Law, 9, 215 - 218.

Bor, W. (2004). Prevention and treatment of childhood and adolescent aggression and antisocial

behaviour: a selective review. Australian & New Zealand Journal of Psychiatry, 38, 373-380.

Bor, W., Najman, J.M., O’Callaghan, M., Williams, G.M., & Anstey, K. (2001). Aggression and the

development of delinquent behaviour in children. Australian Institute of Criminology: trends

and issues, 207. Canberra, ACT: Australian Institute of Criminology.

Bordin, E. S. (1979). The generalizability of the psychoanalytic concept of the working alliance.

Psychotherapy: Theory, Research & Practice, 16, 252-260.

Brad, J. B., Roy, F. B., Sander, T., Ehri, R., Sander, B., & Stephen, G. W. (2009). Looking again, and

harder for a link between low self-esteem and aggression. Journal of Personality, 77, 427-

446.

Brann, P., Coleman, G., & Luk, E. (2001). Routine outcome measurement in child and adolescent

mental health service: an evaluation of HoNOSCA. Australian and New Zealand Journal of

Psychiatry, 35, 370-376.

Brinthaupt, T. M., Hein, M. B., & Kramer, T. E. (2009). The self-talk scale: development, factor

analysis, and validation Journal of Personality Assessment, 9, 82-92.

191

Brodbeck, C., & Michelson, L. (1987). Problem-solving skills and attributional styles of agoraphobics.

Cognitive Therapy and Research, 11, 593-610.

Broidy, L. M., Nagin, D. S., Tremblay, R. E., Bates, J. E., Brame, B., Dodge, K. A., et al. (2003).

Developmental trajectories of childhood disruptive behaviors and adolescent delinquency: a

six-site, cross-national study. Developmental Psychology, 39, 222-245.

Bullis, M., Walker, H. M., & Sprague, J. R. (2001). A promise unfulfilled: social skills training with at-

risk and antisocial children and youth. Exceptionality, 9, 67-90.

Bush, J. (1995). Teaching self-risk management to violent offenders. In J. McGuire (Ed.), What

works: reducing re-offending: guidelines from research and practice (pp. 139-154).

Chichester: John Wiley & Sons.

Buss, A. H., & Warren, W. L. (2000). Aggression questionnaire: manual. Los Angeles, CA: Western

Psychological Services.

Caldwell, M. F., & Van Rybroek, G. J. (2005). Reducing violence in serious juvenile offenders using

intensive treatment. International Journal of Law and Psychiatry, 28, 622-636.

Carducci, D. J. (1980). Postive peer culture and assertiveness training: complementary modalities for

dealing with disturbed and disturbing adolescents in the class room. Behavioral Disorders, 5,

156-162.

Carroll, A., Hemingway, F., Bower, J., Ashman, A., Houghton, S., & Durkin, K. (2006). Impulsivity in

juvenile delinquency: differences among early-onset, late-onset, and non-offenders. Journal

of Youth & Adolescence, 35, 517-527.

Catalano, R. F., Loeber, R., & McKinney, K. C. (1999). School and community interventions to

prevent serious and violent offending. . Washington D.C.: Office of Juvenile Justice and

Delinquent Prevention.

192

Chambers, J. C., Eccleston, L., Day, A., Ward, T., & Howells, K. (2008). Treatment readiness in

violent offenders: The influence of cognitive factors on engagement in violence programs.

Aggression and Violent Behavior, 13, 276-284.

Chassin, L., Young, R. D., & Light, R. (1980). Evaluations of treatment techniques by delinquent and

disturbed adolescents. Journal of Clinical Child Psychology, 9, 220-232.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ:

Lawrence Erlbaum Associates.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112 , 155-159.

Cohen, J. (1994). The earth is round (p< .05). American Psychologist, 49, 997-1003.

Colby, A., Kohlberg, L., Gibbs, J. C., & Lieberman, M. (1983). A longitudinal study of moral judgment.

Monographs of the Society for Research in Child Development, 48, 1-96.

Conduct Problems Prevention Research Group. (1999a). Initial impact of the fast track prevention

trial for conduct problems 1: the high-risk sample. Journal of Consulting and Clinical

Psychology, 67, 631-647.

Conduct Problems Prevention Research Group. (1999b). Initial impact of the fast track prevention

trial for conduct problems II: classroom effects. Journal of Consulting and Clinical

Psychology, 67, 648-657.

Cook, C. R., Gresham, F. M., Kern, L., Barreras, R. n. B., Thornton, S., & Crews, S. D. (2008). Social

skills training for secondary students with emotional and/or behavioral disorders: a review

and analysis of the meta-analytic literature. Journal of Emotional & Behavioral Disorders, 16,

131-144.

Corder, G. W., & Foreman, D. I. (2009). Nonparametric statistics for non-statisticians: a step-by-step

approach. New Jersey: Wiley.

193

Cotter, R. B., Burke, J. D., Loeber, R., & Mutchka, J. (2005). Predictors of contact difficulty and

refusal in a longitudinal study. Criminal Behavior and Mental Health, 15, 126-137.

Cotter, R. B., Burke, J. D., Stouthamer-Loeber, M., & Loeber, R. (2005). Contacting participants for

follow-up: how much effort is required to retain participants in longitudinal studies?

Evaluation and Program Planning, 28, 15-21.

Cowling, V., Costin, J., Davidson-tuck, R., Esler, J., Chapman, A., & Niessen, J. (2005). Responding

to disruptive behaviour in schools: collaboration and capacity building for early intervention

Australian e-Journal for the Advancement of Mental Health, 4.

Crick, N. R., & Dodge, K. A. (1996). A review and reformulation of social information-processing

mechanisms in children's social adjustment. Psychological Bulletin, 115, 74-101.

Crick, N. R., & Dodge, K. A. (1996). Social information-processing mechanisms in reactive and

proactive aggression. Child Development, 67, 993-1002.

Cunliffe, T. (1992). Arresting youth crime: a review of social skills training with young offenders.

Adolescence, 27, 891-901.

Currie, M. R., Wood, C. E., Williams, B., & Bates, G. W. (2009). Aggression replacement training in

Australia: youth justice pilot study. Psychiatry, Psychology & Law, 16, 413-426.

Dandreaux, D. M., & Frick, P. J. (2009). Developmental pathways to conduct problems: a further test

of the childhood and adolescent-onset distinction. Journal of Abnormal Child Psychology, 37,

375-385.

Day, A., & Howells, K. (2002). Psychological treatments for rehabilitating offenders: evidence-based

practice comes of age. Australian Psychologist, 37, 39-47.

Day, A., Howells, K., & Rickwood, D. (2003). The Victorian juvenile justice rehabilitation review.

Melbourne, Aust: Department of Human Service, Victorian Government.

194

Day, A., Howells, K., Mohr, P., Schall, E., & Gerace, A. (2008). The development of CBT

programmes for anger: the role of interventions to promote perspective-taking skills.

Behavioural and Cognitive Psychotherapy, 36, 299-312.

de Castro, B., Brendgen, M., Van Boxtel, H., Vitaro, F., & Schaepers, L. (2007). “Accept me, or

else…”: disputed overestimation of social competence predicts increases in proactive

aggression. Journal of Abnormal Child Psychology, 35, 165-178.

Deffenbacher, J. L. (2003). Psychosocial Interventions: anger disorders. In E. F. Coccaro (Ed.),

Aggression: Psychiatric Assessment and Treatment (pp. 293-312). New York: Marcel Dekker

Inc.

DeRubeis, R.J., Tang, T.Z., & Beck, A.T. (2001). Cognitive therapy. In K.S. Dobson (Ed.), Handbook

of cognitive behavioural therapies (pp.349-392). New York: Guilford Press.

DeVellis, R. F. (2003). Scale development: theory and applications (2nd ed.). Newbury Park, CA:

Sage.

Dickman, S. J. (1990). Functional and dysfunctional impulsivity: personality and cognitive correlates.

Journal of Personality and Social Psychology, 58, 95-102.

Dodge, K. A. (1980). Social cognition and children's aggressive behavior. Child Development, 51,

162-170.

Dodge, K. A. (1986). A social information-processing model of social competence in children In M.

Perlmutter (Ed.), Minnesota Symposium on Child Psychology (Vol. 18, pp. 77-125). Hillsdale,

NJ: Erlbaum.

Dodge, K. A. (1991). Reactive and proactive aggression. In D. J. Pepler & K. H. Rubin (Eds.), The

development and treatment of childhood aggression (pp. 201-218). Hillsdale, NJ: Lawrence

Erlbaum Associates.

195

Dodge, K. A., Bates, J. E., & Pettit, G. S. (1990). Mechanisms in the cycle of violence. Science, 250,

1678-1683.

Dodge, K. A., & Coie, J. D. (1987). Social-information-processing factors in reactive and proactive

aggression in children's peer groups. Journal of Personality and Social Psychology, 53,

1146-1158.

Dodge, K. A., Greenberg, M. T., & Malone, P. S. (2008). Testing an idealized dynamic cascade

model of the development of serious violence in adolescence. Child Development, 79, 1907-

1927.

Dodge, K.A., Laird, R., Lochman, J.E., & Zelli, A. (2002). Multidimensional latent-construct analysis of

children's social information processing patterns: Correlations with aggressive behavior

problems. Psychological Assessment, 14, 60-73.

Dodge, K. A., Lansford, J. E., & Dishion, T. J. (2006). The problem of deviant peer influences in

intervention programs. In K. A. Dodge, T. J. Dishion & J. E. Lansford (Eds.), Deviant peer

influences in programs for youth: problems and solutions (pp. 3-13). New York, NY US:

Guilford Press.

Dodge, K. A., & Pettit, G. S. (2003). A biopsychosocial model of the development of chronic conduct

problems in adolescence. Developmental Psychology, 39, 349-371.

Dodge, K.A., Pettit, G.S., Bates, J.E., & Valente, E. (1995). Social information-processing patterns

partially mediate the effect of early physical abuse or later conduct problems. Journal of

Abnormal Psychology, 104, 632-643.

Dowden, C., Antonowicz, D., & Andrews, D. A. (2003). The effectiveness of relapse prevention with

offenders: a meta-analysis. International Journal of Offender Therapy and Comparative

Criminology, 47, 516-528.

196

D'Zurilla, T. J., & Goldfried, M. R. (1971). Problem-solving and behavior modification. Journal of

Abnormal Psychology, 78, 104-126.

D'Zurilla, T. J., & Maydeu-Olivares, A. (1995). Conceptual and methodological issues in social

problem-solving assessment. Behavior Therapy, 26, 409-432.

D'Zurilla, T. J., Nezu, A. M., & Maydeu-Olivares, A. (2002). Social problem solving inventory- revised

(SPSI-R). Toronto, Canada: Multi-Health Systems Inc.

Eme, R. (2008). Male life-course persistent antisocial behavior: a review of neuro-developmental

factors. Aggression and Violent Behavior, 14, 348-358.

Eme, R. F. (2008). Attention-Deficit/Hyperactivity Disorder and the juvenile justice system. Journal of

Forensic Psychology Practice, 8, 174 - 185.

Enticott, P. G., & Ogloff, J. R. P. (2006). Elucidation of impulsivity. Australian Psychologist, 41, 3-14.

Eysenck, S. B. G. (2004). How the impulsiveness and venturesomeness factors evolved after the

measurement of psychoticism. In R. M. Stelmack (Ed.), On the psychobiology of personality:

essays in honour of Marvin Zuckerman (pp. 107-112). Oxford, UK: Elsevier Inc.

Eysenck, S. B. G., & Eysenck, H. J. (1977). The place of impulsiveness in a dimensional system of

personality description. British Journal of Social and Clinical Psychology, 16, 57-68.

Eysenck, S. B. G., & McGurk, B. J. (1980). Impulsiveness and venturesomeness in a detention

centre population. Psychological Reports, 47, 1299-1306.

Eysenck, S. B. G., Easting, G., & Pearson, P. R. (1984). Age norms for impulsiveness, venturism,

and empathy in children. Personality and Individual Differences, 5, 315-321.

Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, J. F. (1985). Age norms for impulsiveness,

venturesomeness and empathy in adults. Personality and Individual Differences, 6, 613-619.

197

Fazel, S., Doll, H., & Langstrom, N. (2008). Mental disorders among adolescents in juvenile detention

and correctional facilities: a systematic review and metaregression analysis of 25 surveys.

Journal of the American Academy of Child & Adolescent Psychiatry, 47, 1010-1019.

Feindler, E. L. (2003). Anger control training. In Research Press (Ed.). Aggression replacment

training (ART): a comprehensive intervention for aggressive youth [video] Champaign, Ill:

Research Press.

Feindler, E. L., & Baker, K. (2004). Current issues in anger management interventions with youth. In

A. P. Goldstein, R. Nensén, B. Daleflod & M. Kalt (Eds.), New perspectives on Aggression

Replacement Training: practice, research, and application (pp. 31-50). Chichester, UK: John

Wiley & Sons Ltd.

Feindler, E. L., & Ecton, R. B. (1986). Adolescent anger control: cognitive-behavioural techniques.

Elmsford, NY: Pergamon Press.

Feindler, E. L., & Guttman, J. (1994). Cognitive-behavioural anger control training In C. W. LeCroy

(Ed.), Handbook of child and adolescent treatment manuals (pp. 170-199). New York:

Lexington Books.

Feindler, E. L., & Starr, K. E. (2003). From steaming mad to staying cool: a constructive approach to

anger control. Reclaiming Children and Youth, 12, 158-160.

Felthous, A. R., & Barratt, E. S. (2003). Impulsive aggression In E. F. Coccaro (Ed.), Aggression:

Psychiatric assessment and treatment (pp. 123-148). New York: Marcel Dekker.

Fenstermacher, K. (2006). Effectiveness of a computer-facilitated interactive social skills training

program for boys with attention deficit hyperactivity disorder. School Psychology Quarterly,

21, 197-224.

198

Fergusson, D. M., & Horwood, L. J. (2002). Male and female offending trajectories. Development and

Psychopathology, 14, 159-177.

Fleck, D., Thompson, C. L., & Narroway, L. (2001). Implementation of the problem solving skills

training programme in a medium secure unit. Criminal Behaviour & Mental Health, 11, 262-

273.

Florsheim, P., Shotorbani, S., Guest-Warnick, G., Barratt, T., & Hwang, W.C. (2000). Role of the

working alliance in the treatment of delinquent boys in community-based programs. Journal

of Clinical Child Psychology, 29, 94-94.

Fonagy, P., & Kurtz, Z. (2002). Disturbances of conduct. In P. Fonagy, M. Target, D. Cottrell, J.

Phillips, & Z. Kurtz. What works for whom? a critical review of treatments for children and

adolescents (pp.106-192). New York: Guilford Press.

Frick, P. J. (2004). Developmental pathways to conduct disorder: Implications for serving youth who

show severe aggressive and antisocial behaviour. Psychology in the Schools, 41, 823-834.

Frick, P. J., Cornell, A. H., Barry, C. T., Bodin, S. D., & Dane, H. E. (2003). Callous-unemotional traits

and conduct problems in the prediction of conduct problem severity, aggression, and self-

report of delinquency. Journal of Abnormal Child Psychology, 31, 457-470.

Frick, P., & Hare, R. (2001). Antisocial process screening device (APSD): technical manual. Toronto,

ON: Multi-Health Systems.

Frick, P. J., & Loney, B. R. (1999). Outcomes of children and adolescents with conduct disorder and

oppositional defiant disorder. In H. C. Quay & A. Hogan (Eds.), Handbook of disruptive

behaviour disorders (pp. 507-524). New York: Plenum.

Frick, P., & Morris, A. S. (2004). Temperament and developmental pathways to conduct problems

Journal of Clinical Child & Adolescent Psychology, 33, 54-68.

199

Frick, P. J., Stickle, T. R., Dandreaux, D. M., Farrell, J. M., & Kimonis, E. R. (2005). Callous–

unemotional traits in predicting the severity and stability of conduct problems and

delinquency. Journal of Abnormal Child Psychology, 33, 471-487.

Gannon, T. A., & Wood, J. L. (2010). Introduction: Special issue on group processes and aggression.

Aggression and Violent Behavior, 15, 85-86.

García-Forero, C., Gallardo-Pujol, D., Maydeu-Olivares, A., & Andrés-Pueyo, A. (2009).

Disentangling impulsiveness, aggressiveness and impulsive aggression: an empirical

approach using self-report measures. Psychiatry Research, 168, 40-49.

Gelhorn, H. L., Sakai, J. T., Price, R. K., & Crowley, T. J. (2007). DSM-IV conduct disorder criteria as

predictors of antisocial personality disorder. Comprehensive Psychiatry, 48, 529.

Gendreau, P. (1996). The principles of effective interventions with offenders. In A. Harland (Ed.),

Choosing correctional options that work (pp.117-130). Thousand Oaks, CA: Sage.

Getter, H., & Nowinski, J. K. (1981). A free response test of interpersonal effectiveness Journal of

Personality Assessment, 45, 301-308.

Gibbs, J. C. (2004). Moral reasoning training: the values component. In A. P. Goldstein, R. Nensén,

B. Daleflod & M. Kalt (Eds.), New perspectives on Aggression Replacement Training:

practice, research and application (pp. 51-72). Chichester, UK: John Wiley & Sons Ltd.

Gibbs, J.C. (2003). Moral development and reality: beyond the theories of Kohlberg and Hoffman.

Thousand Oaks, CA: Sage Publications

Gilbert, F., & Daffern, M. (in press). Integrating contemporary aggression theory with violent offender

treatment: How thoroughly do interventions target violent behavior? Aggression and Violent

Behavior, Corrected Proof.

200

Gold, M., & Osgood, D. W. (1992). Personality and peer influence in juvenile corrections.

Contributions in Criminology and Penology, 38, 163-178.

Goldberg, A. D. (1998). Arnold Goldstein: From counselor to psycho-educator. Journal of Counseling

& Development, 76, 351-357.

Goldman, L. S., Genel, M., Bezman, R. J., Slanetz, P. J., Karlan, M. S., Davis, R. M., et al. (1998).

Diagnosis and treatment of attention-deficit/hyperactivity disorder in children and

adolescents. Journal of the American Medical Association, 279, 1100-1107.

Goldstein, A. P. (1973). Structured learning therapy: toward a psychotherapy for the poor. New York:

Academic Press.

Goldstein, A. P. (1999). The prepare curriculum: teaching prosocial competencies (revised edn.).

Champaigne, IL: Research Press.

Goldstein, A. P., Sprafkin, R. P., & Gershaw, N. J. (1979). Structured learning therapy: training for

community living. Psychotherapy: Theory, Research & Practice, 16, 199-203.

Goldstein, A.P (2004). Evaluations of effectiveness. In A.P. Goldstein, R. Nensén, B. Daleflod, & M.

Kalt, (Eds). New perspectives on aggression replacement training: practice, research and

application. Chichester, UK: John Wiley & Sons Ltd.

Goldstein, A. P., & Glick, B. (1987). Aggression replacement training: a comprehensive intervention

for aggressive youth. Champaign, IL: Research Press.

Goldstein, A.P., & Glick, B. (1994). The prosocial gang: implementing aggression replacement

training. Thousand Oaks, CA: Sage.

Goldstein, A.P., Glick, B., & Gibbs, J.C. (1998). Aggression replacement training: a comprehensive

Intervention for aggressive youth (Revised Edition). Illinois, USA: Research Press.

201

Goldstein, S., & Rider, R. (2005). Resilience and the disruptive disorders of childhood. In R. Brooks &

S. Goldstein (Eds.), Handbook of resilience in childhood (pp. 203-222). New York: Kluwer

Academic/Plenum Publishers.

Greenwood, P. (2006). Promising solutions in juvenile justice. In K. A. Dodge, T. J. Dishion & J. E.

Lansford (Eds.), Deviant peer influences in programs for youth: problems and solutions. (pp.

278-295). New York: Guilford Press.

Gresham, F. M. (1997). Social competence and students with behavior disorders: where we've been,

where we are, and where we should go. Education and Treatment of Children, 20, 233-249.

Gresham, F. M. (2002). Best practices in social skills training. In A. Thomas & J. Grimes (Eds.), Best

Practice in School Psychology IV (pp. 1029-1040). Washington, DC: National Association of

School Psychologists

Gresham, F. M., & Elliott, S. N. (1990). Social skills rating system manual. Circle Pines, MN:

American Guidance Service.

Gresham, F. M., Cook, C. R., Crews, S. D., & Kern, L. (2004). Social skills training for children and

youth with emotional and behavioral disorders: validity considerations and future directions.

Behavioral Disorders, 30, 32-46.

Guerra, N. G., & Slaby, R. G. (1990). Cognitive mediators of aggression in adolescent offenders: II.

intervention. Developmental Psychology, 26, 269-277.

Gundersen, K., & Svartdal, F. (2006). Aggression replacement training in Norway: outcome

evaluation of 11 Norwegian student projects. Scandinavian Journal of Educational

Research, 50, 63-81.

202

Hains, A.A. (1984). Variables in social cognitive development: moral judgment, role-taking, cognitive

processes, and self-concept in delinquents and non-delinquents. Journal of Early

Adolescence, 4, 65-74.

Hanson, S. (2009). Youth Justice in Victoria: the benefits of Victoria’s youth justice system and the

challenges ahead. Melbourne, Aust.: YMCA World Alliance Youth Justice Campaign.

Hare, R. D., Hart, S. D., & Harpur, T. J. (1991). Psychopathy and the DSM—IV criteria for antisocial

personality disorder. Journal of Abnormal Psychology, 100, 391-398.

Harris, K. R. (1990). Developing self-regulated learners: the role of private speech and self-

instructions. Educational Psychologist, 25, 35 - 49.

Hatcher, R. M., Palmer, E. J., McGuire, J., Hounsome, J. C., Bilby, C. A. L., & Hollin, C. R. (2008).

Aggression replacement training with adult male offenders within community settings: a

reconviction analysis. Journal of Forensic Psychiatry & Psychology, 19, 517 - 532.

Hawley, P. H. (1999). The ontogenesis of social dominance: a strategy-based evolutionary

perspective. Developmental Review, 19, 97-132.

Hayes, C. (2001). The Washington State ART Qaulity Assurance form. Washington State Institute for

Public Policy [WSIPP}. WA: WSIPP

Hayes, S.C. (2004). Acceptance and commitment therapy and the new behaviour therapies. In S.C.

Hayes, V.M. Follette, & M.M. Linehan (Eds.), Mindfulness and acceptance: expanding the

cognitive behavioural tradition (pp.1-29). New York: Guilford Press.

Hayes, S.C., Strosahl, K.D., Bunting, K., Twohig, M., & Wilson, K.G. (2004). What is acceptance and

commitment therapy? In S.C. Hayes and K.D. Strosahl (Ed’s.). A practical guide to

acceptance and commitment therapy (3-29). New York: Springer.

203

Hemphill, S. A., Smith, R., Toumbourou, J. W., Herrenkohl, T. I., Catalano, R. F., McMorris, B. J., et

al. (2009). Modifiable determinants of youth violence in Australia and the United States: a

longitudinal study. Australian and New Zealand Journal of Criminology, 42, 289-309.

Hoge, R. D., & Andrews, D. A. (1996). Assessing the youthful offender: issues and techniques. New

York: Plenum Press.

Hollin, C. R. (1999). Treatment programs for offenders: meta-analysis, "what works," and beyond.

International Journal of Law and Psychiatry, 22, 361-372.

Hollin, C. R. (2001a). Editorial: Social problem solving and offenders. Criminal Behaviour and Mental

Health, 11, 204-209.

Hollin, C. R. (2001b). The role of the consultant in developing effective practice. In G. A. Bernfeld, D.

P. Farrington & A. W. Leschied (Eds.), Offender rehabilitation in practice: implementing and

evaluating effective programmes (pp. 269-287). Chichester: Wiley.

Hollin, C.R. (2003). Aggression replacement training: putting theory and research to work. Reclaiming

Children and Youth- the journal of strength-based interventions, 12, 132-135.

Hollin, C.R. (2004). Aggression replacement training: the cognitive-behavioural context. In A. P.

Goldstein, R. Nensén, B. Daleflod & M. Kalt (Eds.), New perspectives on Aggression

Replacement Training: practice, research and application (pp. 51-72). Chichester, UK: John

Wiley & Sons Ltd.

Hollin, C. R. (2010). Commentary: directions for group process work. Aggression and Violent

Behavior, 15, 150-151.

Hollin, C. R., & Bloxsom, C. A. J. (2007). Treatments for angry aggression In T. A. Gannon, T. Ward,

A. R. Beech & D. Fisher (Eds.), Aggressive offender’s cognition: theory, research, and

practice (pp. 215-229). Chichester, UK: John Wiley & Sons Ltd.

204

Hollin, C. R., Browne, D., & Palmer, E. J. (2004). Delinquency and adolescent offenders.

Camberwell, VIC: Acer Press.

Hollin, C. R., & Palmer, E. J. (2006). Offending behaviour programmes: controversies and resolutions

In C. R. Hollin & E. J. Palmer (Eds.), Offending behaviour programmes: development,

application, and controversies (pp. 247-278). Chichester, England: John Wiley & Sons Ltd.

Holmqvist, R., Hill, T., & Lang, A. (2007). Treatment alliance in residential treatment of criminal

adolescents. Child & Youth Care Forum, 36, 163-178.

Holmqvist, R., Hill, T., & Lang, A. (2009). Effects of aggression replacement training in young

offender institutions. International Journal of Offender Therapy and Comparative Criminology,

53, 74-92.

Hornsveld, R. H. J. (2004). Aggression control therapy for forensic psychiatric patients: Development

and preliminary results In A. P. Goldstein, R. Nensén, B. Daleflod & M. Kalt (Eds.), New

perspectives on aggression replacement training: practice, research, and application (pp.

189-196). Chichester, England: John Wiley & Sons Ltd.

Hornsveld, R. H. J. (2005). Evaluation of aggression control therapy for violent forensic psychiatric

patients. Psychology, Crime & Law, 11, 403-410.

Hornsveld, R. H. J., Nijman, H. L. I., Hollin, C. R., & Kraaimaat, F. W. (2008). Aggression control

therapy for violent forensic psychiatric patients. International Journal of Offender Therapy

and Comparative Criminology, 52, 222-233.

Horvath, A. O., & Luborsky, L. (1993). The role of the therapeutic alliance in psychotherapy. Journal

of Consulting and Clinical Psychology, 61, 561-573.

Horvath, A. O., & Symonds, B. D. (1991). Relation between working alliance and outcome in

psychotherapy: a meta-analysis. Journal of Counseling Psychology, 38, 139-149.

205

Howells, K., & Day, A. (2002). Grasping the nettle: treating and rehabilitating the violent offender.

Australian Psychologist, 37, 222-228.

Howells, K., Day, A., Bubner, S., Jauncey, S., Parker, A., Williamson, P. & Heseltine, K. (2001). An

evaluation of anger management programs with violent offenders in two Australian states.

Report to the Criminology Research Council.

Hubble, M. A., Duncan, B. L., & Miller, S. D. (Eds.). (1999). The heart & soul of change: what works

in therapy. Washington, DC: American Psychological Association.

International consensus statement on ADHD. (2002). Clinical Child and Family Psychology Review,

5, 89-111.

Izzo, R. L., & Ross, R. R. (1990). Meta-analysis of rehabilitation programs for juvenile delinquents: a

brief report. Criminal Justice and Behavior, 17, 134-142.

Jaffee, S., & Hyde, J. S. (2000). Gender Differences in Moral Orientation: a Meta-Analysis.

Psychological Bulletin, 126, 703.

Kassinove, H., & Tafrate, R. C. (2006). Anger-related disorders: basic issues, models, and diagnostic

considerations In E. L. Feindler (Ed.), Anger-related disorders: a practitioners guide to

comparative treatments (pp. 1-28). New York: Springer Publishing Company Inc.

Kazdin, A.E. (1987). Treatment of antisocial behaviour in children: Current status and future

directions. Psychological Bulletin, 102, 187-203.

Kazdin, A.E. (1997). Practitioner review: psychosocial treatments for conduct disorder in children.

Journal of Child Psychology and Psychiatry, 38, 161-178.

Kazdin, A.E. (1999). The meanings and measurement of clinical significance. Journal of Consulting

and Clinical Psychology, 67, 332-339.

206

Kazdin, A. E. (2002). Psychosocial treatments for conduct disorder in children and adolescents. In P.

E. Nathan & J. M. Gorman (Eds.), A guide to treatments that work (2nd ed., pp. 57-86). New

York: Oxford University Press.

Kazdin, A. E., & Weisz, J. R. (1998). Identifying and developing empirically supported child and

adolescent treatments. Journal of Consulting and Clinical Psychology, 66, 19-39.

Keltikangas-Järvinen, L. (2001). Aggressive behaviour and social problem-solving strategies: a

review of the findings of a seven-year follow-up from childhood to late adolescence. Criminal

Behaviour and Mental Health, 11, 236-250.

Keltikangas-Järvinen, L., & Pakaslahti, L. (1999). Development of social problem-solving strategies

and changes in aggressive behavior: a 7-year follow-up from childhood to late adolescence.

Aggressive Behavior, 25, 269-279.

Kendall, P. C., & Finch, A. J. (1978). A cognitive-behavioral treatment for impulsivity: a group

comparison study. Journal of Consulting and Clinical Psychology, 46, 110-118.

Kendall, P.C., Marrs-Garcia, A., Nath, S.R., & Sheldrick, R.C. (1999). Normative comparisons for the

evaluation of clinical significance. Journal of Consulting and Clinical Psychology, 67, 285-

299.

Kohlberg, L. (1969). Stage and sequence: The cogntive-developmental approach to socialization. In

D. A. Goslin (Ed.), Handbook of socialization theory and research (pp. 347-480). Chicago:

Rand McNally.

Kohlberg, L. (1976). Moral stages and moralization: The cognitive-developmental approach. In

T.Lickona (Ed.), Moral development and behavior (pp. 31-53). New York: Holt, Rinehart &

Winston.

207

Kohlberg, L. (1984). Essays on moral development: the psychology of moral development.

SanFrancisco, CA: Harper & Row.

Kohlberg, L., & Krammer, R. (1969). Continuities and discontinuities in childhood and adult moral

development. Human Development, 12, 93-120.

Krebs, D. L., & Denton, K. (2005). Toward a more pragmatic approach to morality: a critical

evaluation of Kohlberg's model. Psychological Review, 112, 629-649.

Krug, E. G., Dahlberg, L. L., Mercy, J. A., Zwi, A. B., & Lozano, R. (2002). World report on violence

and health Geneva: World Health Organisation

Landenberger, N.A., & Lipsey, M.W. (2005). The positive effects of cognitive-behavioural programs

for offenders: A meta-analysis of factors associated with effective treatment. Journal of

Experimental Criminology, 1, 451-476.

Lansford, J. E., Malone, P. S., Dodge, K. A., Crozier, J. C., Pettit, G. S., & Bates, J. E. (2006). A 12-

year prospective study of patterns of social information processing problems and

externalizing behaviors. Journal of Abnormal Child Psychology, 34, 715-724.

Launay, G. (2001). Relapse prevention with sex offenders: practice, theory and research. Criminal

Behaviour & Mental Health, 11, 38.

Lee, S. I., Schachar, R. J., Chen, S. X., Ornstein, T. J., Charach, A., Barr, C., et al. (2008). Predictive

validity of DSM-IV and ICD-10 criteria for ADHD and hyperkinetic disorder. Child Psychology

and Psychiatry, 49, 70-78.

Liau, A. K., Barriga, A. Q., & J.C, G. (1998). Relations between self-serving cognitive distortions and

overt vs. covert antisocial behavior in adolescents. Aggressive Behavior, 24, 335-346.

Lipsey, M. W. (2006). The effects of community-based group treatment for delinquency: a meta-

analytic search for cross-study generalizations. In K. A. Dodge, T. J. Dishion & J. E. Lansford

208

(Eds.), Deviant peer influences in programs for youth: Problems and solutions. (pp. 162-184).

New York: Guilford Press.

Lipsey, M. W. (2009). The primary factors that characterize effective interventions with juvenile

offenders: a meta-analytic overview. Victims & Offenders, 4, 124-147.

Lipsey, M.W., Chapman, G.L., & Landenberger, N.A. (2001). Cognitive-behavioural programs for

offenders. Annals of the American Academy of Political and Social Science, 578, 144-157.

Lipsey, M.W., & Wilson, D.B. (1998). Effective intervention for serious juvenile offenders: A synthesis

of research. In R. Loeber and D.P. Farrington (Eds). Serious and Violent Juvenile Offenders:

Risk factors and successful interventions (pp.313-344).. Thousand Oaks, CA: Sage.

Lochman, J.E., Whidby, J.M., & FitzGerald, D.P. (2000). Cognitive-behavioral assessment and

treatment with aggressive youth. In P.C. Kendall (Ed.). Child & adolescent therapy:

cognitive-behavioral procedures (pp.31-87). New York: Guilford Press.

Loeber, R. (1991). Diagnostic conundrum of oppositional defiant disorder and conduct disorder.

Journal of Abnormal Psychology, 100, 379-390.

Loeber, R. (1991). More enduring than changeable? Journal of the American Academy of Child and

Adolescent Psychiatry, 30, 303-397.

Loeber, R., Burke, J. D., Lahey, B. B., Winters, A., & Zera, M. (2000). Oppositional defiant and

conduct disorder: a review of the past 10 years, Part I. Journal of the American Academy of

Child & Adolescent Psychiatry, 39, 1468-1484.

Loeber, R., Burke, J., & Pardini, D. A. (2009). Perspectives on oppositional defiant disorder, conduct

disorder, and psychopathic features. Journal of Child Psychology and Psychiatry and Allied

Disciplines, 50, 133-142.

209

Longmore, R.J., & Worrell, M. (2007). Do we need to challenge thoughts in cognitive behaviour

therapy? Clinical Psychology Review, 27, 173-187.

Losel, F. (2001). Evaluating the effectiveness of correctional programs: bridging the gap between

research and practice In G. A. Bernfeld, D. P. Farrington & A. W. Leschied (Eds.), Offender

rehabilitation in practice (pp. 67-92). Chichester, UK: Wiley.

Luengo, M. A., Carrillo-de-la-Peña, M. T., Otero, J. M., & Romero, E. (1994). A short-term

longitudinal study of impulsivity and antisocial behavior. Journal of Personality and Social

Psychology, 66, 542-548.

Luria, A. R. (1961). The role of speech in the regulation of normal and abnormal behaviour. New

York: Liveright.

Maag, J. W. (2006). Social skills training for students with emotional and behavioral disorders: a

review of reviews. Behavioral Disorders, 32, 4-17.

Mals, P., Howells, K., Day, A., & Hall, G. (2000). Adapting violence rehabilitation programs for the

Australian aboriginal offender. Journal of Offender Rehabilitation, 30, 121-135.

Marsh, D. T., Serafica, F. C., & Barenboim, C. (1980). Effect of perspective-taking training on

interpersonal problem solving. Child Development, 51, 140-145.

Marshall, W. L., & Burton, D. L. (2010). The importance of group processes in offender treatment.

Aggression and Violent Behavior, 15, 141-149.

Marshall, W. L., & Serran, G. A. (2004). The role of the therapist in offender treatment. Psychology,

Crime & Law, 10, 309 - 320.

Martinson, R. (1974). What works? questions about prison reform. The Public Interest, 35, 22-54.

Maruna, S., & Mann, R.E., (2006). A fundamental attribution error? Rethinking cognitive distortions.

Legal and Criminological Psychology, 11, 155-177.

210

Maughan, B., Rowe, R., Messer, J., Goodman, R., & Meltzer, H. (2004). Conduct disorder and

oppositional defiant disorder in a national sample: developmental epidemiology. Journal of

Child Psychology and Psychiatry, 45, 609-621.

McGuire, J. (2001). What is problem solving? a review of theory, research and applications. Criminal

Behaviour and Mental Health, 11, 210-235.

McGuire, J. (2003). Maintaining change: converging legal and psychological initiatives in a

therapeutic jurisprudence framework. Western Criminology Review, 4, 108-123.

McGuire, J., & Clark, D. (2004). A national dissemination program. In A.P. Goldstein, R. Nensén, B.

Daleflod, & K. Mikael (Ed’s). New perspectives on aggression replacement training:

practice, research and application (pp.139-150). Chichester, UK: John Wiley & Sons Ltd.

McMurran, M. (2005). Social problem solving and offenders: reflections and directions. In M.

McMurran & J. McGuire (Eds.), Social problem solving and offending: evidence, evaluation

and evolution (pp. 297-308). Chichester: John Wiley & Sons Ltd.

McMurran, M., Egan, V., & Duggan, C. (2005). Stop & think! social problem-solving therapy with

personality disordered offenders. In M. McMurran & J. McGuire (Eds.), Social problem

solving and offending: evidence, evaluation and evolution (pp. 207-220). Chichester: John

Wiley & Sons Ltd.

Meichenbaum, D. H. (1977). Cognitive-behavior modification: an integrative approach. New York:

Plenum.

Meichenbaum, D. H., & Goodman, J. (1971). Training impulsive children to talk to themselves: a

means of developing self-control. Journal of Abnormal Psychology, 77, 115-126.

Merrel, K. (2002). School social behavior scales (2nd ed.). Eugene, OR: Assessment Intervention

Services.

211

Miller, S. D., Duncan, B. L., & Hubble, M. A. (1997). Escape from Babel: toward a unifying language

for psychotherapy practice. New York: Norton.

Moffitt, T. E. (1993). Life-course-persistent and adolescence-limited antisocial behavior: a

developmental taxonomy. Psychological Review, 100, 674-701.

Moffitt, T. E., Arseneault, L., Jaffee, S. R., Kim-Cohen, J., Koenen, K. C., Odgers, C. L., et al. (2008).

Research review: DSM-V conduct disorder: research needs for an evidence base. Journal of

Child Psychology and Psychiatry and Allied Disciplines, 49, 3-33.

Moffitt, T. E., & Caspi, A. (2001). Childhood predictors differentiate life-course persistent and

adolescence-limited antisocial pathways among males and females. Development and

Psychopathology, 13, 355-375.

Moffitt, T. E., Caspi, A., Harrington, H., & Milne, B. J. (2002). Males on the life-course-persistent and

adolescence-limited antisocial pathways: follow-up at age 26 years. Development and

Psychopathology, 14, 179-207.

Moynahan, L. (2003). Enhanced aggression replacement training with children and youth with autism

spectrum disorder. Reclaiming Chiidren and Youth, 12, 174-180.

Muñoz, L., Frick, P., Kimonis, E., & Aucoin, K. (2008). Types of aggression, responsiveness to

provocation, and callous-unemotional traits in detained adolescents. Journal of Abnormal

Child Psychology, 36, 15-28.

Nas, C. N., Brugman, D., & Koops, W. (2005). Effects of the EQUIP programme on the moral

judgement, cognitive distortions, and social skills of juvenile delinquents. Psychology, Crime

& Law, 11, 421- 434.

Nelson, W. J., & Birkimer, J. C. (1978). Role of self-instruction and self-reinforcement in the

modification of impulsivity. Journal of Consulting and Clinical Psychology, 46, 183-183.

212

Nelson, J. R., Smith, D. J., & Dodd, J. (1990). The moral reasoning of juvenile delinquents: a meta-

analysis. Journal of Abnormal Child Psychology, 18, 231-239.

Nezu, A. M. (1985). Difference in psychological distress between effective and ineffective problem-

solvers. Journal of Counseling Psychology, 32, 135-138.

Nietzel, M. T. (1979). Crime and its modifications: a social learning perspective. Oxford: Pergamon.

Nietzel, M. T., Hasemann, D. M., & Lynam, D. R. (1999). Behavioral perspectives on violent behavior.

In V. B. V. Hasselt & M. Hersen (Eds.), Handbook of psychological approaches with violent

offenders: contemporary strategies and issues. New York: Kluwer Acadamic/Plenum Press.

Nock, M. K., Kazdin, A. E., Hiripi, E., & Kessler, R. C. (2007). Lifetime prevalence, correlates, and

persistence of oppositional defiant disorder: results from the national comorbidity survey

replication. Journal of Child Psychology & Psychiatry, 48, 703-713.

Novaco, R. W. (1975). Anger control: the development and evaluation of an experimental treatment

Lexington, MA: Lexington Books.

Novaco, R. W. (1979). The cognitive regulation of anger and stress. In P. C. Kendell & S. C. Hollon

(Eds.), Cognitive-behavioral interventions: theory, research and procedures (pp. 241-285).

New York: Academic Press.

Novaco, R. W. (1997). Remediating anger and aggression with violent offenders. Legal and

Criminological Psychology, 2, 77-88.

Novaco, R.W. (2007). Anger dysregulation. In T. A. Cavell & K. T. Malcolm (Eds.), Anger, aggression

and interventions for interpersonal violence (pp. 3-54). Mahwah, NJ: Lawrence Erlbaum

Associates.

213

Novaco, R. W., & Welsh, W. N. (1989). Anger disturbances: cognitive mediation and clinical

precriptions. In K. Howells & C. R. Hollin (Eds.), Clinical approaches to violence Chichester,

UK: John Wiley & Sons Ltd.

Nugent, W. R., Bruley, C., & Allen, P. (1999). The effects of aggression replacement training on male

and female antisocial behavior in a runaway shelter. Research on Social Work Practice, 9,

466-482.

Nugent, W. R., Champlin, D., & Wiinimaki, L. (1997). The effects of anger control training on

adolescent antisocial behavior. Research on Social Work Practice, 7, 446-462.

Odgers, C. L., Moffitt, T. E., Broadbent, J. M., Dickson, N., Hancox, R. J., Harrington, H., et al.

(2008). Female and male antisocial trajectories: from childhood origins to adult outcomes.

Development and Psychopathology, 20, 673-716.

Ogloff, J. (2002). Offender rehabilitation: from "nothing works" to what next? Australian Psychologist,

37, 245-252.

Ogloff, J., & Davis, M. R. (2004). Advances in offender assessment and rehabilitation: contributions

of the risk-needs-responsivity approach. Psychology, Crime & Law, 10, 229-242.

Osgood, D. W., & Briddell, L. O. (2006). Peer effects in juvenile justice. In K. A. Dodge, T. J. Dishion

& J. E. Lansford (Eds.), Deviant peer influences in programs for youth: problems and

solutions. (pp. 141-161). New York: Guilford Press.

Palmer, E. J. (2003). Offending behaviour: moral reasoning, criminal conduct, and the rehabilitation

of offenders. Devon, UK: Willan Publishing.

Palmer, E. J. (2005). The relationship between moral reasoning and aggression, and the implications

for practice. Psychology, Crime and Law, 11, 353-361.

214

Palmer, E.J. (2007). Moral cognition and aggression. In T.A. Gannon, T. Ward, A.R. Beech, & D.

Fisher (Eds.), Aggressive offenders cognition: theory, research, and practice (pp.199-214).

West Sussex, England: John Wiley & Sons Ltd.

Palmer, E. J., & Hollin, C. R. (1999). Social competence and sociomoral reasoning in young

offenders. Applied Cognitive Psychology 13, 79-87.

Parker, J. S. (2009). Distinguishing Between Early and Late Onset Delinquents: Race, Income,

Verbal Intelligence and Impulsivity. North American Journal of Psychology, 11, 273-284.

Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992). A social interactional approach: antisocial boys

(Vol. 4). Eugene, OR: Castalia.

Pfiffner, L. J., & McBurnett, K. (1997). Social skills training with parent generalization: treatment

effects for children with attention deficit disorder. Journal of Consulting and Clinical

Psychology, 65, 749-757.

Phelps, L., & McClintock, K. (1994). Papa and peers: a biosocial approach to conduct disorder.

Journal of Psychopathology and Behavioral Assessment, 16, 53-67.

Piaget, J. (1932). The moral judgment of the child. London: Routledge & Kegan Paul.

Piaget, J. (1952). The origins of intelligence in children. New York: International University Press.

Polanczyk, G., de Lima, M. S., Horta, B. L., Biederman, J., & Rhode, L. A. (2007). The worldwide

prevalence of ADHD: a systematic review and metaregression analysis. American Journal of

Psychiatry, 164, 942-948.

Polaschek, D. L. L. (2003). Relapse prevention, offense process models, and the treatment of sexual

offenders. Professional Psychology: Research and Practice, 34, 361-367.

215

Polaschek, D. L. L. (2006). Violent offender programmes: concept, theory and practice In C. R. Hollin

& E. J. Palmer (Eds.), Offending behaviour programmes: development, application, and

controversies (pp. 113-154). Chichester, England: John Wiley & Sons Ltd.

Polaschek, D. L. L., & Dixon, B. G. (2001). The violence prevention project: The development and

evaluation of a treatment programme for violent offenders Psychology, Crime & Law, 7, 1-23.

Posavac, E. J., & Carey, R. G. (2003). Program evaluation: methods and case studies (6th ed.).

Upper Saddle River, NJ: Prentice Hall.

Potter, G. B., Gibbs, J. C., & Goldstein, A. P. (2001). The equip program implementation guide:

teaching youth to think and act responsibly through a peer-helping approach. Champaign, IL:

Research Press.

Price, D. M. (1999). Relapse prevention and risk reduction: results of client identification of high risk

situations. Sexual Addiction & Compulsivity, 6, 221-232.

Quinn, M. M., Kavale, K. A., Mathur, S. R., Rutherford, J. R. B., & Forness, S. R. (1999). A meta-

analysis of social skill interventions for students with emotional or behavioral disorders.

Journal of Emotional & Behavioral Disorders, 7, 54.

Rahill, S. A., & Teglasi, H. (2003). Processes and outcomes of story-based and skill-based social

competency programs for children with emotional disabilities. Journal of School Psychology,

46, 413-429.

Raine, A., Dodge, K., Loeber, R., Gatzke-Kopp, L., Lynam, D., Reynolds, C., et al. (2006). The

reactive-proactive aggression questionnaire: differential correlates of reactive and proactive

aggression in adolescent boys. Aggressive Behavior, 32, 159-171.

216

Ramírez, J. M., & Andreu, J. M. (2006). Aggression, and some related psychological constructs

(anger, hostility, and impulsivity): some comments from a research project. Neuroscience

and Biobehavioral Reviews, 30, 276-291.

Raue, P. J., Goldfried, M. R., & Barkham, M. (1997). The therapeutic alliance in psychodynamic-

interpersonal and cognitive-behavioral therapy. Journal of Consulting and Clinical

Psychology, 65, 582-587.

Reed, M. K. (1994). Social skills training to reduce depression in adolescents. Adolescence, 29, 293-

303.

Reinecke, M. A., & Ginsburg, G. S. (2008). Cognitive-behavioral treatment of depression during

childhood and adolescence. In J. R. Z. Abela, & B. L. Hankin (Eds.), Handbook of depression

in children and adolescents. (pp. 179-206). New York: Guilford Press.

Reynolds, W. M. (1998). The adolescent psychopathology scale (APS): administration and

interpretation manual. Odessa, FL: Psychological Assessment Resources.

Rosenthal, R. (1985). Contrast analysis: focused comparisons in the analysis of variance.

Cambridge, Eng: Cambridge University Press

Ross, E. C., Polaschek, D. L. L., & Ward, T. (2008). The therapeutic alliance: a theoretical revision for

offender rehabilitation. Aggression and Violent Behavior, 13, 462-480.

Roth, B. S., & Striepling-Goldstein, S. (2003). School-based aggression replacement training

Reclaiming Children and Youth, 12, 138-141.

Rowland, A. S., Lesesne, C. A., & Abramowitz, A. J. (2002). The epidemiology of attention-

deficit/hyperactivity disorder (ADHD): a public health view. Mental Retardation &

Developmental Disabilities Research Reviews, 8, 162-170.

217

Salmon, S. (2003). Teaching empathy: the PEACE curriculum. Reclaiming Children and Youth, 12,

167-173.

Sawyer, M. G., Arney, F. M., Baghurst, P. A., Clark, J. J., Graetz, B. W., & Kosky, R. J. (2000).

Mental health of young people in Australia: child and adolescent component of the national

survey of metal health and well-being. Canberra: Commonwealth Dept. of Health and Aged

Care.

Scarpa, A., & Raine, A. (1997). Phychophysiology of anger and violent behavior. Psychiatric Clinics

of North America, 20, 375-394.

Segrin, C. (2000). Social skills deficits associated with depression. Clinical Psychology Review, 20,

379-403.

Selman, R. L. (1971). The relation of role taking to the development of moral judgment in children.

Child Development, 42, 79-91.

Selman, R. L. (1980). The growth of interpersonal understanding: developmental and clinical

analysis. New York: Academic Press.

Serin, R. C., Gobeil, R., & Preston, D. L. (2009). Evaluation of the persistently violent offender

treatment program. International Journal of Offender Therapy and Comparative Criminology,

53, 57-73.

Serin, R. C., & Preston, D. L. (2001). Managing and treating violent offenders. In J. B. Ashford, B. D.

Sales & W. H. Reid (Eds.), Treating adult and juvenile offenders with special needs. (pp. 249-

271): American Psychological Association.

Sestir, M. A., & Bartholow, B. (2007). Theoretical explanations of aggression and violence In T. A.

Gannon, T. Ward, A. R. Beech & D. Fisher (Eds.), Aggressive offenders' cognition: theory,

reasearch and practice. West Sussex, EN: John Wiley & Sons.

218

Shelton, D., Sampl, S., Kesten, K. L., Zhang, W., & Trestman, R. (2009). Treatment of impulsive

aggression in correctional settings. Behavioral Sciences & the Law, 27, 787-800.

Skinner, B. F. (1938). The Behavior of Organisms. New York: Appleton-Century-Crofts.

Skinner, B. F. (1953). Science and Human Behavior. New York: Macmillan.

Slaby, R. G., & Guerra, N. G. (1988). Cognitive mediators of aggression in adolescent offenders: 1.

assessment. Developmental Psychology, 24, 580-588.

Smith, P., & Waterman, M. (2006). Self-reported aggression and impulsivity in forensic and non-

forensic populations: the role of gender and experience. Journal of Family Violence, 21, 425-

437.

Spence, S. H. (2000). The treatment of childhood social phobia: the effectiveness of a social skills

training-based, cognitive-behavioural intervention, with and without parental involvement.

Journal of Child Psychology & Psychiatry & Allied Disciplines, 41, 713-726.

Spence, S. H. (2003). Social skills training with children and young people: theory, evidence and

practice. Child and Adolescent Mental Health, 8, 84-96.

Spielberger, C. D., Reheiser, E. C., & Sydeman, S. J. (1995). Measuring the experience, expression,

and control of anger. Issues in Comprehensive Pediatric Nursing, 18, 207-232.

Spivack, G., & Shure, M. B. (1982). Interpersonal cognitive problem solving and clinical theory. In B.

Lahey & A. E. Kazdin (Eds.), Advances in Clinical Psychology (Vol. 5, pp. 323-372). New

York: Plenum.

Stams, G. J., Brugman, D., Dekovic, M., van Rosmalen, L., van der Laan, P., & Gibbs, J. C. (2006).

The moral judgment of juvenile delinquents: a meta-analysis. Journal of abnormal child

psychology, 34, 697-713.

219

Stouthamer-Loeber, M., & Loeber, R. (2002). Lost opportunities for intervention: undetected markers

for the development of serious juvenile delinquency. Criminal Behaviour & Mental Health, 12,

69-82.

Streiner, D. L. (2002). The case of missing data: methods of dealing with dropouts and other

research vagaries. Canadian Journal of Psychiatry, 47, 68-75.

Sukhodolsky, D. G., Golub, A., Stone, E. C., & Orban, L. (2005). Dismantling anger control training

for children: a randomized pilot study of social problem-solving versus social skills training

components. Behavior Therapy, 36, 15-23.

Sukhodolsky, D. G., Vitulano, L. A., Carroll, D. H., McGuire, J., Leckman, J. F., & Scahill, L. (2009).

Randomized trial of anger control training for adolescents with tourette's syndrome and

disruptive behavior. Journal of the American Academy of Child & Adolescent Psychiatry, 48,

413-421.

Sullivan, C. J., Veysey, B. M., & Dorangrichia, L. (2003). Examining the relationship between problem

history and violent offending in high-risk youth. Journal of Offender Rehabilitation, 38, 17 -

39. Viewed July 08, 2009,

Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Boston, Mass: Allyn &

Bacon.

Tate, D. C., Reppucci, N. D., & Mulvey, E. P. (1995). Violent juvenile delinquents: treatment

effectiveness and implications for future action. Amercian Psychologist, 50, 777-781.

Taylor, N. (2007). Juveniles in detention in Australia, 1981-2006. Canberra, Aust: Australian Institute

of Criminology

Teplin, L. A., Abram, K. M., McClelland, G. M., Dulcan, M. K., & Mericle, A. A. (2002). Psychiatric

disorders in youth in juvenile detention. Arch Gen Psychiatry, 59, 1133-1143.

220

Tremblay, R. E. (2000). The development of aggressive behaviour during childhood: what have we

learned in the past century? International Journal of Behavioral Development, 24, 129-141.

van Manen, T., Prins, P., & Emmelkamp, P. (2004). Reducing aggressive behavior in boys with a

social cognitive group treatment: results of a randomized, controlled trial. Journal of the

American Academy of Child & Adolescent Psychiatry, 43, 1478-1487.

Vorrath, H. H., & Brendtro, L. K. (1985). Postive peer culture (2nd ed.). Hawthorne, NY: Aldine.

Vygotsky, L. S. (1962). Thought and language. Cambridge, MA: M.I.T. Press.

Wakai, S., Shelton, D., Trestman, R. L., & Kesten, K. (2009). Conducting research in corrections:

challenges and solutions. Behavioral Sciences & The Law, 27, 743-752.

Walker, L. J. (1980). Cognitive and perspective-taking prerequisites for moral development. Child

Development, 51, 131-139.

Ward, T., & Stewart, C. (2003a). Criminogenic needs and human needs: a theoretical model.

Psychology, Crime & Law, 9, 125 - 143.

Ward, T., & Stewart, C. (2003b). The relationship between human needs and ciminogenic needs.

Psychology, Crime & Law, 9, 219 - 224.

Waschbusch, D. A., & King, S. (2006). Should sex-specific norms be used to assess attention-

deficit/hyperactivity disorder or oppositional defiant disorder? Journal of Consulting and

Clinical Psychology, 74, 179-185.

Webster-Stratton, C., & Dahl, R. W. (1995). Conduct disorder. In M. Hersen & R. T. Ammerman

(Eds.), Advanced Abnormal Child Psychology (pp. 333-352). Hillsdale, New Jersey:

Lawrence Erlbaum Associates.

Werry, J. S. (1997). Severe conduct disorder: some key issues. Canadian Journal of Psychiatry, 42,

577-583.

221

White, J. L., Moffitt, T. E., Caspi, A., Bartusch, D. J., Needles, D. J., & Stouthamer-Loeber, M. (1994).

Measuring impulsivity and examining its relationship to delinquency. Journal of Abnormal

Psychology, 103, 192-205.

White, S. W., Keonig, K., & Scahill, L. (2007). Social skills development in children with autism

spectrum disorders: a review of the intervention research. Journal of Autism &

Developmental Disorders, 37, 1858-1868.

Wilson, D. B., Bouffard, L. A., & MacKenzie, D. L. (2005). A quantitative review of structured, group-

oriented, cognitive-behavioral programs for offenders. Criminal Justice and Behaviour, 32,

172-204.

World Health Organisation [WHO]. (1992). The international classification of disease (10th ed.).

Geneva: Author.

Yalom, I. D. (1995). The theory and practice of group psychotherapy (4th ed.). New York: Basic

Books.

Yochelson, S., & Samenow, S. E. (1976). The Criminal Personality: a profile for change (Vol. 1). New

York: Jason Aronson.

Zuckerman, M. (1979). Sensation seeking: beyond the optimal level of arousal. Hillsdale, NJ:

Erlbaum.

222

Appendix A: ART Group Materials

A1. Standard ART Curriculum

A2. Skillstreaming Practice Sheet

A3. Hassle & Thought Trap Log

A4. Moral Reasoning Problem (e.g., Marks Problem)

A5. Leader Notes: Marks Problem

A6. Thought Traps (i.e., cognitive distortions)

223

Standard ART Curriculum

Week Skillstreaming Anger Control

Training Moral Reasoning

1 Making A Complaint ABC’s Of Anger Jim’s Problem

2 Responding to

Failure/Disappointment

Triggers & Body

Signs Tom’s Problem

3 Dealing with Someone Else’s Anger Reducers Mark’s Problem

4 Dealing with Group Pressure Reminders George’s Problem

5 Dealing with an Accusation

If… Then.. Thinking

Ahead

Andy’s Problem

6 Deciding What Caused a Problem

The Angry Behavior

Cycle

Sarah’s Problem

7 Getting Ready for a Difficult

Conversation

Social Skills Leigh’s Problem

8 Keeping Out of Fights Self-Evaluation Paul’s Problem

9 Understanding the Feelings of Others Rehearse Full-

Sequence Shaun’s Problem

10

Expressing Affection or Appreciation

Full-Sequence

Johnny’s Problem

224

225

226

Name: ___________________________________ Mark’s Problem Situation

Mark has been going out with a girl named Maria for about two months. It used to be a lot of fun to be with her, but lately it’s been sort of a drag. There are some other girls Mark would like to go out with. Now Mark sees Maria coming down the school hallway. What should Mark say or do?

1. Should Mark avoid the subject with Maria so Maria’s feelings aren’t hurt?

a. Should avoid subject b. Should bring it up c. Can’t decide

2. Should Mark make up an excuse, like being too busy to see Maria, as a way of breaking up?

a. Excuse b. No excuse c. Can’t decide

3. Should Mark simply start going out with other girls so that Maria will get the message?

a. Yes b. No c. Can’t decide

4. Let’s change the situation a bit. What if Mark and Maria have been living together for several years and have two small children? Then should Mark still break up with Maria?

a. Should break up b. No, shouldn’t break up c. Can’t decide

5. Let’s go back to the original situation. This is what happens: Mark does break up with Maria. He let her know how he feels and starts dating another girl. Maria feels hurt and jealous and thinks about getting even somehow. Should Maria get even?

a. Yes, should get even b. No, shouldn’t get even c. Can’t decide

6. What if the tables were turned, and Maria broke up with Mark. Should

Mark get even somehow?

a. Yes, should get even b. No, shouldn’t get even c. Can’t decide

7. How should Mark respond to Maria’s feelings?

227

Leader Notes: Mark's Problem Situation Mark's Problem Situation continues the theme of mature, caring relationships but focuses on the problem of ending a dating relationship that is going nowhere. The main value of this problem situation for moral judgment development arrives with discussion of the last questions, which concern vengeance. As with most of the problem situations, many group members choose positive responses. The majority position tends to be that Mark should discuss breaking up (Question 1) rather than making up an excuse (Question 2) or simply starting to date other girls (Question 3). Accordingly most of the open-ended suggestions (in response to Question 4) are positive: "Just tell her you'd like to date other girls"; "Be considerate and remember she's human, too"; "Explain how you feel, that you don't want to settle down'; "Listen to what she has to say about it'. Of the responses we have heard our favourite is "I think we should see other people. What do you think?" The group member who gave this response also indicated that he would first try to, "Work things out", before breaking up with Maria. As to the reasons for bringing the subject up (Question 1), one group member pointed out that Mark "should be man enough to tell her"; if he doesn't, another suggested, "Maria might lose a chance to get another boyfriend," and "would be hurt more in the long run", than by just being told. Speaking more pragmatically (against the idea of simply starting to date other girls), another group member suggested that then those girls could find out how Mark treated Maria and dump him for being a two-timer. Of course, not all of these responses are positive. On the open-ended question, one group member wrote, "Do things to make Maria drop him." Another wrote that he would say, "I'm dumping you, bag!" These group members may also advocate avoiding Maria or making up an excuse. After discussion, however, they are often willing to acquiesce to the majority position and thereby make a positive group discussion possible. An abrupt turnabout occurs on Question 5, in which Mark and Maria are live-in partners with two small children. Then the majority usually favours not breaking up, on the grounds that Mark has a responsibility to the children (e.g. "The kids should have both a dad and a mom."). If he left, "it would hurt the kids, because they would feel it was their fault." One group member suggested, "He loved her once. Why should one argument make him not love her again?" A pragmatic group member pointed out that he might have to pay child support if he leaves. The majority position continues to be positive on the vengeance questions, 6 and 7. The majority is against either Maria's getting even if Mark breaks up (Question 6) or Mark's getting even if the tables are turned (Question 7). Suggestions are that, "Mark should just tell himself that it's her loss"; "Let bygones be bygones." Mark, "wouldn't want her to get even with him [so he shouldn't do that to her; Stage 3] and if one of them retaliated, "there would just be more trouble." One group member suggested, somewhat ominously that Mark shouldn't get even because he "might do something really bad and wind up in here." Count on several group members advocating retaliation, however - especially by Mark against Maria. Reasons included, "Give her a taste of her own medicine" and suggested that Mark should get even because, "He'd be mad" and, as further justification, disclosed that he himself had gotten mad and

228

beaten up several girls who had left him for other guys. He remained silent when a peer asked, "Does that make it right?" and asked why he nonetheless thought it was wrong if Maria got even with Mark. Nor would he acquiesce to a group decision against getting even. At least he felt peer group opposition and perhaps this is the reason he was more accommodating to positive majority positions on subsequent occasions. It is sometimes helpful to ask the group exactly what is meant by "getting even." Responses range from, "Showing off [ to Maria] with a new girlfriend;' to, "Telling him [the new boyfriend] that she was a good lay for you," to "Slashing their tyres,' -- or faces! The majority will often brand these responses, once stated for group consideration, as immature or destructive. Nonetheless, many group members will comment that although Mark and/or Maria shouldn't get even, they probably would. If the group is still developing, the group leader may need to model re-labelling -- that is , comment on how much strength and courage it takes not to, "Give in to childish desires to get even." The degree of positive content may be surprising and should be encouraged. The group leader should comment on the great potential the group has shown for becoming a positive group. Using re-labelling, the group leader should emphasize that a strong group is one where members care about another's feelings. Bear in mind, however that the group members expressing more negative sentiments may be speaking more candidly; their words may be consistent with the actual behaviour of the majority. After all, consider how common "payback" or vengeance is in the daily life of the troubled adolescent! Similarity, in social skills exercises, initial absence of caring about another's feelings can be striking. Clearly the group challenge is to accomplish the translation of responsible words into responsible actions.

229

230

Appendix B: Ethics

B1. Participant Information and Consent Form (orginal 2004)

B2. Swinburne University Human Research Ethics Committee (HREC)- Approval (2004)

B3. Department of Human Services Victoria HREC- Approval (2005)

B4. Swinburne University HREC- Ammedment (2008)

B5. Partcipant Information and Consent form for the Release of Information from

Victorian Police Data (LEAP) Records

B6. Swinburne University HREC- Ammedment Approval (2009)

B7. Victoria Police HREC- Approval (2009)

231

Participant Information and Consent Form

Full Project Title: Aggression Replacement Training (ART): An evaluation study

Principal Researcher: Matthew Currie

Associate Researcher(s): Dr Glen Bates & Dr Katie Wood

1. Your Consent

You are invited to take part in this research project.

This Participant Information contains detailed information about the research project. Its purpose is to explain to you as openly and clearly as possible all the procedures involved in this project before you decide whether or not to take part in it.

Please read this Participant Information carefully. Feel free to ask questions about any information in the document. You may also wish to discuss the project with a relative or friend or your health worker. Feel free to do this.

Once you understand what the project is about and if you agree to take part in it, you will be asked to sign the Consent Form. By signing the Consent Form, you indicate that you understand the information and that you give your consent to participate in the research project.

You will be given a copy of the Participant Information and Consent Form to keep as a record.

2. Purpose and Background

The purpose of this project is to evaluate the effectiveness of Aggression Replacement Training (ART) with Australian youth.

Up to 30 young men from Malmsbury Youth Training Centre will participate in this project.

We know that some young people have problems managing their anger, which can damage personal relationships and cause trouble with the law. We are interested in seeing if we can help you find ways to deal with problems other than using aggression. To do this we want to run a program called Aggression Replacement Training or A.R.T. at Malmsbury. ART was originally developed in America and there have been good results for young people with it there. However, we need to know how and if it will work with Australian youth, that what this study is for.

You are invited to participate in this research project because your health and key workers at Malmsbury thought that this program could provide you with some useful skills. That is, you might learn some new skills to help you control your anger and get along better in your personal relationships.

The results of this research will be used to help the principal researcher. Matthew Currie, to obtain a degree.

232

3. Procedures

Participation in this project will involve

• Attendance at three ART group sessions per week for 10-weeks. Each session lasts about one hour, so that’s a total of about three hours per week. We will run ART at Malmsbury on three different days of the week and you will be in the group with about seven other guys.

• We will get about 16 guys to sign up for the program. The participants that are due for release earlier will go into the first group and those who have a longer stay at Malmsbury will go into the second group- which will start as soon as the first group finishes. So half the guys will start the program immediately and the other half will start 10-weeks later when the first group has finished. We do this so that we can see if there is any difference between a group of guys who have done ART compared to a group who have not, sometimes called a waitlist-control group.

• We will ask you some questions about your mental health, your drug use and your family background. We only do this part of the testing once- just before the program begins. This will take about 2-hours. We will ask everyone participating in the study to complete the same paper/pencil questionnaires just before the program begins, then again 10-weeks later (end of the program), and six months later, 12-months later and 2-years later. This is so that we can see what effect ART has over time. Each time you do the questionnaires it will take about 30 to 45-minutes.

• We will also ask your key-worker or probation officer (post-release) to complete some questionnaires about how they see you relate to others.

3.1 Consent for the Release of Information

We are seeking your consent for the release of information from The Department of Human Service- Juvenile Justice regarding your convictions just before the program begins, then six months, 12-months and 2-years after the program ends. DHS-JJ already has these records. Again we do this so that we can see if participation in ART can reduce or prevent re-offending.

3.2 Video Taping of ART Sessions

Because ART is new to Australia the ART-Trainer must go through an accreditation process. We have to videotape some of the ART sessions and send it to our teacher in the US, Chris Hayes (ART Master Trainer). It’s like a project we have to pass, Chris is the only person who will see the tapes, after he has watched them they are destroyed. The tapes will never be shown on TV, or sold by anyone to make money. If you sign the consent form, it means you are giving your permission for your ART program to be taped, and for the people listed above to see the tapes.

4. Possible Benefits

Possible benefits from ART might include an increase in your ability to control your anger, which in turn might help you to get along better with others and keep out of trouble, as well overall better self-esteem. We cannot guarantee that this will happen for you but there is evidence from the USA that some of these benefits have occurred for many of the people who have done the program.

233

5. Possible Risks

We do not believe that there are any serious risks associated with your participation in the program. However, the program will challenge you to think and do things differently and sometimes that might cause you some anxiety. If this happens for you we will provide lots of opportunities for you discuss this with the ART-Trainers. If that doesn’t help you could talk to your key-worker or health-worker at Malmsbury.

6. Alternatives to Participation

Alternative procedures/alternative treatments include individual meetings with your key-worker and health-worker at Malmsbury.

7. Privacy, Confidentiality and Disclosure of Information

We are serious about keeping your details confidential. You need to be aware that there are some situations where we would need to change that. You don’t have to tell us anything you don’t want to, but if you choose to tell us about illegal behaviour or a situation that puts yourself or others at risk of harm, then we are required to report that information to the appropriate authorities. This might mean contacting the police or your health-worker. We would let you know if this was going to happen. If you have any questions, or do not understand what is happening, make sure you say something. You can tell your key-worker, health-worker, or ask us.

8. New Information Arising During the Project

During the research project, new information about the risks and benefits of the project may become known to the researchers. If this occurs, you will be told about this new information. This new information may mean that you can no longer participate in this research. If this occurs, the person(s) supervising the research will stop your participation and discuss the reasons with you and your health-worker at Malmsbury.

9. Results of Project

The results of this study will not be available for some time because it spans a number of years. However, by about midway through 2007 some early results will be available and the final results will be available by about midway through 2008. The results will written up in a brief report which will be put up on the Menzies web site www.menzies.org.au. You can also contact Matthew Currie (Principal Researcher) at Menzies on 9784-9700 and he will discuss the results with you and post you a copy of any reports. The results will not have your name or any identifiable information in them, only group statistics will be reported.

10. Further Information or Any Problems

If you require further information or if you have any problems concerning this project (for example, any side effects), you can contact the principal researcher (Matthew Currie) or the associate researcher- Dr Katie Wood from Swinburne University. The researchers responsible for this project are:

Matthew Currie Dr Katie Wood Menzies Inc. Faculty of Life & Social Science

234

PO Box 55 OR PO Box 218 Frankston 3199 Hawthorne 3122 Ph: 9784-9700 Ph: 9214-5740

11. Other Issues

If you have any complaints about any aspect of the project, the way it is being conducted or any questions about your rights as a research participant, then you may contact

The Chair Human Research Ethics Committee Swinburne University of Technology PO Box 218, Hawthorn 3122 Tel: 9214 5223

You will need to tell The Chair the name of one or both of the researchers given in section 10 above.

12. Participation is Voluntary

Participation in any research project is voluntary. If you do not wish to take part you are not obliged to. If you decide to take part and later change your mind, you are free to withdraw from the project at any stage.

Your decision whether to take part or not to take part, or to take part and then withdraw, will not affect your routine treatment, your relationship with those treating you or your relationship with anyone at Malmsbury.

Before you make your decision, a member of the research team will be available to answer any questions you have about the research project. You can ask for any information you want. Sign the Consent Form only after you have had a chance to ask your questions and have received satisfactory answers.

13. Ethical Guidelines

This project will be carried out according to the National Statement on Ethical Conduct in Research Involving Humans (June 1999) produced by the National Health and Medical Research Council of Australia. This statement has been developed to protect the interests of people who agree to participate in human research studies.

The ethical aspects of this research project have been approved by the Human Research Ethics Committee of Swinburne University and the Department of Human Services Victoria.

14. Reimbursement for your costs

You will not be paid for your participation in this project.

235

Consent Form (JJ)

Full Project Title: Aggression Replacement Training (ART): An evaluation study

I have read and I understand the Participant Information.

I freely agree to participate in this project according to the conditions in the Participant Information, including the video-taping of some of the ART session and my permission for the principal researcher (Matthew Currie) to obtain information from DHS client records regarding my convictions.

I will be given a copy of the Participant Information and Consent Form to keep.

The researcher has agreed not to reveal my identity and personal details if information about this project is published or presented in any public form.

Participant’s Name (printed) ……………………………………………………

Signature Date

Name of Witness to Participant’s Signature (printed) ……………………………………………

Signature Date

Researcher’s Name (printed) Matthew Currie

Signature Date

Note: All parties signing the Consent Form must date their own signature.

236

237

238

August 9, 2008

Research Ethics Officer

Keith Wilkins

kwilkins@swin.edu.au

Room: 60Wm705a

Mail No: H68

Phone: 9214-5218

Dear Keith,

Re: Amendment- HREC No. 03/27 Aggression Replacement Training (ART): An evaluation

study

The above mentioned Doctoral research project is due to undertake final data collection (i.e., 2-year

follow-up) from participants (N=20). In the original Participant Information and Consent Form (dated

16/01/2006) signed consent was obtained for the following clause-

3.1 Consent for the Release of Information

We are seeking your consent for the release of information from The Department of Human Services-

Juvenile Justice regarding your convictions just before the program begins, then six months, 12-

months and 2-years after the program ends. DHS-JJ already has these records. Again we do this so

that we can see if participation in ART can reduce or prevent re-offending.

This de-identified data is needed in order to track recidivism, which is an important outcome measure

in determining clinical significance subsequent to treatment. However it has become apparent that

those individuals who may have re-offend subsequent to leaving Malmsbury Youth Justice Centre

would be likely to enter the adult, penal system and therefore would not be listed on the Juvenile

Justice data collection system. It is for this reason that the Principal Researcher has sought

permission from Victoria Police- Research & Project Governance to access the police LEAP system.

Victoria Police has given ‘in principle’ support for this de-identified data to be provided, however they

have indicated (see attached letter) that specific consent must be sought from research participants.

In order for this written consent to be obtained we are seeking an amendment to above mentioned

HREC application to mail out an amended consent form (see attachment) to be signed by

239

participants and returned in a stamped, addressed envelope to Dr Katie Wood (Research

Supervisor).

In addition we are seeking ethics approval to provide participants with a $50 gift certificate to be

provided upon receipt of the signed consent form and completion of the final set of questionnaires

relating to the 2-year follow-up (final) data collection. This is intended as an incentive to return the

signed consent form promptly and remuneration in recognition of participant’s ongoing willingness to

give of their time over the duration of this longitudinal evaluation study.

Your timely response to this request for amendment would be greatly appreciated.

Yours sincerely,

Dr Katie Wood Mr Matthew Currie

(Research Supervisor) (Principal Researcher)

240

Participant Information and Consent Form

Full Project Title: Aggression Replacement Training (ART): An evaluation study

Principal Researcher: Matthew Currie

Associate Researcher(s): Dr Katie Wood

Consent for the Release of Information from Victorian Police Data (LEAP) Records

The purpose of this additional consent form is to give Matthew Currie (Principal Researcher) your consent for the release of information from Victoria Police regarding your convictions 6-months and 24-months after your completion of the ART program. Matthew Currie agrees to de-identify the information by applying a code that only he can match to your name, so the information will be confidential. The purpose of this information is so that we can see if participation in ART can reduce or prevent re-offending. On return of this signed consent form in the pre-paid, self addressed envelope you will receive a gift voucher to the value of $50 in recognition of my time and effort in agreeing to participate in this study.

Participation is Voluntary

Participation in any research project is voluntary. If you do not wish to take part you are not obliged to. If you decide to take part and later change your mind, you are free to withdraw from the project at any stage.

Participant’s Name (print) :…………………………………………………

Signature :……………………………………………………Date:……………………………………………………

Researcher’s Name: Matthew Currie

Signature :……………………………………………………Date:………………………………………………

Note: All parties signing the Consent Form must date their own signature.

241

From: Keith Wilkins [KWilkins@groupwise.swin.edu.au] Sent: Monday, 5 October 2009 12:49 PM To: Matt Currie; Catherine Wood Cc: Glen Bates Subject: SUHREC Project 03/27 Ethics Clearance Confirmed for Modified/Extended Project (VPHREC Project 82/09) Re: FW: VPHREC 82/09 Approval To: Dr Catherine Wood/Mr Matthew Currie, FLSS Dear Katie and Matt SUHREC Project 03/27 Aggression Replacement Training (ART): An evaluation study Dr Catherine Wood, FLSS; Mr Matthew Currie Project Duration Extended to 31/12/2009 [Approved Modifications: June 2005, Aug/Sep 2008, Sep 2009 (Victoria Police Request VPHREC Project 82/09)] I refer to your email of 30 September 2009 attaching evidence of approval from the Victoria Police Human Research Ethics Committee to involve Victoria Police in the way outlined in prevous communication. In light of this information, which was put to the Chair of SUHREC for endorsement, I confirm on-going Swinburne ethics clearance in line with standard conditions (reprinted below). - All human research activity undertaken under Swinburne auspices must conform to Swinburne and external regulatory standards, including the National Statement on Ethical Conduct in Human Research and with respect to secure data use, retention and disposal. - The named Swinburne Chief Investigator/Supervisor remains responsible for any personnel appointed to or associated with the project being made aware of ethics clearance conditions, including research and consent procedures or instruments approved. Any change in chief investigator/supervisor requires timely notification and SUHREC endorsement. - The above project has been approved as submitted for ethical review by or on behalf of SUHREC. Amendments to approved procedures or instruments ordinarily require prior ethical appraisal/ clearance. SUHREC must be notified immediately or as soon as possible thereafter of (a) any serious or unexpected adverse effects on participants and any redress measures; (b) proposed changes in protocols; and (c) unforeseen events which might affect continued ethical acceptability of the project. - At a minimum, an annual report on the progress of the project is required as well as at the conclusion (or abandonment) of the project. - A duly authorised external or internal audit of the project may be undertaken at any time. Please contact me if you have any queries about on-going ethics clearance, citing the SUHREC project number. Copies of clearance emails should be retained as part of project record-keeping. As before, best wishes for the continuing project. Yours sincerely Keith Wilkins Secretary, SUHREC

242

Mr Matthew Currie Faculty of Life and Social Science Dept of Psychology Swinburne University of Technology PO Box 218 HAWTHORN VIC 3122 29 September 2009 Dear Mr Currie,

Strategic Research Unit Strategy and Policy

Level 3, Bldg C, 637 Flinders St.

Melbourne 3005 Victoria, Australia

Telephone 9247 6873 Facsimile 9247 6712

Ethics.committee@police.vic.gov.au www.police.vic.gov.au

Re: VPHREC 82/09: Aggression Replacement Training (ART): An evaluation study. Your response provided 26 September 2009 has been considered out of sessions. I am now pleased to advise you that your application has received formal approval. I draw your attention to the terms of the ‘Declaration by researcher(s)’ in your application, including the following requirements: • To provide a final report and a copy of any published material at the end of the research project,

and • To notify VPHREC in writing immediately if any change to the project is proposed and await

approval before proceeding with the proposed change. If you have any queries or require further clarification please contact Georgina Lee. Please note that you also require approval from the Research Coordinating Committee (RCC) before commencing your research. Yours sincerely, Dr. Georgina Lee

243

Appendix C: Measures

C1. Social Skills Rating System (SSRS): Secondary youth version

C2. Aggression Questionnaire (AQ)

C3. How I Think (HIT) Questionnaire

C4. Social Problem Solving Inventory- Revised: Short form (SPSI-R:S)

C5. I7 Impulsiveness scale

C6. Adult Behavior Checklist (ABCL)

C7 ART Trainer self-assessment checklists

C8 Washington State QA Assessment

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

Skillstreaming Session

ART Assessment

Location ____________ Date______ Trainer ______________ Co-trainer _______________

Session Number ___

Please place a check in the box � that best describes what happened in this session. YES ▼

NO ▼

1. Was a positive climate established through welcoming the students? ...............................................................................

2. Were issues since the last skill training session resolved?................................................................................................

3. Were group norms reviewed, emphasizing positive participation? .....................................................................................

4. Did most youth complete the Skillstreaming Homework Report?........................................................................................

5. Were the Skillstreaming Homework Reports used to review last week’s skill?................................................................

6. Were homework efforts appropriately and genuinely acknowledged?................................................................................

7. Were achievements rewarded?..............................................................................................................................................

8. Were the Skillstreaming Homework Reports collected or kept in the students’ folders?...................................................

9. Were visual aids used (skill cards distributed; skill title and skill steps displayed)?...........................................................

10. Was the new skill correctly introduced, defined, and briefly explained in understandable language? .............................

11. Was the new skill perfectly modelled by the trainer and co-trainer using a relevant adolescent situation?.....................

12. Did someone point to the skill steps during the modelling and role plays? ................................................................

13. Did each youth express how the skill could be personally useful? .....................................................................................

14. Did each youth correctly role-play the skill as the main actor?............................................................................................

15. Did each youth pick their own role play partner?..................................................................................................................

16. Did each youth provide performance feedback? ..................................................................................................................

17. Was the order of performance feedback correct (co-actor, group members, co-trainer, trainer, main actor)?................

18. Were new Skillstreaming Homework Reports given to each youth as homework and the top half filled out?.................

19. Was behaviour appropriately managed?...............................................................................................................................

20. Did the session pace keep the students interested and active?..........................................................................................

21. Did the students appear to understand the skill being taught in this session?................................................................

22. Does the primary Trainer interact with the youth in a positive manner?.............................................................................

23. Does the Co-trainer interact with the youth in a positive manner?......................................................................................

24. Does the Co-trainer aid the Trainer in delivering the curriculum? .......................................................................................

261

Anger Control Training Session

ART Assessment

Location____________ Date______ Trainer ____________ Co-Trainer _______________

Session Number ___

Please place a check in the box � that best describes what happened in this session. YES ▼

NO ▼

1. Was a positive climate established through welcoming the students? ................................................................

2. Were issues since the last Anger Control Training session resolved? ................................................................

3. Were group norms reviewed, emphasizing positive participation? ................................................................ 4. Were all ACT concepts covered to this point reviewed? (e.g. ABC’s, Personal Power,

etc)................................................................................................................................................

5. Did most youth complete the hassle log(s)? ................................................................................................

6. Were the hassle logs used to review the Anger Control Chain? ................................................................

7. Were efforts appropriately and genuinely acknowledged? ................................................................................................

8. Were achievements rewarded?................................................................................................................................

9. Were the hassle logs kept in the students’ folders? ................................................................................................

10. Was the new ACT concept correctly introduced, defined, and briefly explained in understandable language? ................................................................................................................................................................

11. Was the Anger Control Chain correctly reviewed?................................................................................................

12. Were visual aids used (poster of the Anger Control Chain or other visual aids)?................................

13. Was the Anger Control Chain perfectly modelled by the trainer and co-trainer using a relevant adolescent situation? ................................................................................................................................

14. Did someone point to the Anger Control Chain links during the modelling and role plays?................................

15. Did each youth express how the concept could be personally useful?................................................................

16. Did each youth correctly role-play the Anger Control Chain as the main actor?...............................................................

17. Did each youth pick their own role-play partner?................................................................................................

18. Did each youth provide performance feedback during the class?................................................................

19. Was the order of performance feedback correct (co-actor, group members, co-trainer, trainer, main actor)? ................................................................................................................................................................

20. Were new hassle logs given to each youth as homework? ................................................................................................

21. Was behaviour appropriately managed during the session?..............................................................................................

22. Did the session pace keep the students interested and active?................................................................

23. Did the students appear to understand the ACT lesson? ................................................................................................

24. Does the primary Trainer interact with the youth in a positive manner?................................................................

25. Does the Co-trainer interact with the youth in a positive manner?................................................................

26. Does the Co-trainer aid the Trainer in delivering the intervention? ................................................................

262

Moral Reasoning Session

ART Assessment

Location____________ Date______ Trainer ______________ Co-trainer ______________

Problem Situation _______________________________ Session Number ___

Please place a check in the box � that best describes what happened in this session. YES ▼

NO ▼

1. Was the Problem Situation Chart made before the group? ................................................................................................

2. Did the Trainer study the Problem Situation Chart for patterns of thinking and decide the order to discuss the questions?.............................................................................................................................................................

3. Was a positive climate established through welcoming the students? ................................................................

4. Were issues since the last moral reasoning session resolved? ...........................................................................................

5. Were Group Norms for the discussion session reviewed (page 111 A.R.T. Revised Ed.)? ................................

6. Were the four Thought Traps reviewed as established by the Washington State A.R.T. Quality Assurance process?................................................................................................................................................................

7. Was the Problem Situation read to the class? .......................................................................................................................

8. Was the real problem correctly defined? ................................................................................................................................

9. Was the real problem related to the lives of the group members?.......................................................................................

10. Was moral maturity established through: ..............................................................................................................................

a) Eliciting mature responses first? ................................................................................................................................

b) Reconstructing less mature responses, and.....................................................................................................................

c) Listing them on an easel pad, chalk board, or white board? ...........................................................................................

11. Were more mature group members used to challenge the less mature reasoning and Thought Traps expressed by some group members? ................................................................................................................................

12. Was the list of mature reasons used to challenge the less mature reasoning and Thought Traps expressed by some group members? ................................................................................................................................

13. Were role-taking opportunities created for the youth? (e.g. "What would the world be like if everyone behaved that way?" "What would you do if you were _____?") .........................................................................................

14. Was there an attempt made to make positive decisions and mature reasons unanimous or a group decision? ................................................................................................................................................................

15. Were comments by each group member acknowledged?................................................................................................

16. Were individuals within the group praised for positive decisions and mature reasons?................................

17. Was the whole group praised for positive decisions and mature reasons?................................................................

18. Were all group members involved in the discussion? ................................................................................................

19. Was behaviour appropriately managed during the session?...............................................................................................

20. Did the session pace keep the students interested and active?..........................................................................................

21. Did the Trainer remain objective during the session? ................................................................................................

22. Does the Trainer interact with the youth in a positive manner?...........................................................................................

23. Does the Co-trainer interact with the youth in a positive manner?......................................................................................

263

Anger Control Training Session Adherence

Washington State ART Assessment

Jurisdiction: Melbourne Aust. Date: July 2005 Instructor: Matt Currie Co-Instructor: Cath Powell

Week Number: 6

Please place a "1" in the box that best describes what happened in this session. Item

Weight YES=1 NO=0

1. Was a positive climate established through welcoming students? 2 1

2. Were any issues since the last anger control training session dealt with? 2 1

3. Were group norms reviewed, emphasizing positive participation? 2 1

4. Were all relevant ACT concepts covered to this point reviewed? 3 1

5. Did most youth complete the hassle log(s)? 11 1

6. Were the hassle logs used to review the anger control chain? 3 1

7. Were efforts honestly and genuinely acknowledged? 2 1

8. Were achievements rewarded? 2 1

9. Were the hassle logs kept in the students’ folders? 2 1

10. Was the new ACT concept correctly introduced, defined, and briefly explained in understandable language? 11 1

11. Was the Anger Control Chain correctly reviewed? 3 1

12. Were visual aids used (poster of the Anger Control Chain or other visual aids)? 2 1

13. Was the Anger Control Chain perfectly demonstrated by the trainer (as Main Actor) and co-trainer using a relevant adolescent situation? 11 1

14. Did someone point to the Anger Control Chain concepts during the demonstration and practice sessions? 2 1

15. Did each youth express how today’s information could be personally useful? 2 1

16. Did each youth correctly practice the Anger Control Chain as the main actor? 11 1

17. Did each youth pick their own practice session partner? 2 1

18. Did each youth provide performance feedback during the class? 2 1

19. Was the order of performance feedback correct (co-actor, group members, co-trainer, trainer, main actor)? 2 1

20.Were new hassle logs given to each youth as homework? 3 1

21. Was behavior appropriately managed during the session? 3 1

22. Did the session pace keep the students interested and active? 3 1

23. Did the students appear to understand the ACT lesson? 3 1

24. Does the primary instructor interact with the youth in a positive manner? 11 1

25. Does the co-instructor interact with the youth in a positive manner? 3 1

26. Does the co-instructor aid the trainer in delivering the intervention? 3 1

Instructor Score 100 Co-Instructor Score 9

264

Social Skills Training Session Adherence

Washington State ART Assessment

Jurisdiction: Melbourne Aust. Date: July 2005 Instructor: Matt Currie Co-Instructor: Cath Powell

Week Number: 6

Please place a "1" in the box that best describes what happened in this session. Item

Weight

YES=1 NO=0

1. Was a positive climate established through welcoming students? 3 1

2. Were any issues since the last social skills training session dealt with? 1 1

3. Were group norms reviewed, emphasizing positive participation? 2 1

4. Did most youth complete the Social Skills Homework Report? 11 1

5. Were the Social Skills Homework Reports used to review last week’s social skill? 3 1

6. Were homework efforts appropriately and genuinely acknowledged? 3 1

7. Were homework achievements rewarded? 3 1

8. Were the Social Skills Homework Reports collected or kept in the students’ folders? 3 1

9. Were visual aids used (skill cards distributed and social skill title and steps displayed)? 3 1

10. Was the new social skill correctly introduced, defined, and briefly explained in understandable language? 11 1

11. Was the new social skill perfectly demonstrated by the trainer (as Main Actor) and co-trainer using a relevant adolescent situation?

11

1

12. Did the someone point to the social skill steps during the demonstration and practice session? 2 1

13. Did each youth express how the social skill could be personally useful? 3 1

14. Did each youth correctly practice the social skill as the main actor? 11 1

15. Did each youth pick their own practice session partner? 1 1

16. Did each youth provide performance feedback? 3 1

17. Skill steps read to the class by the assigned youth during performance feedback? 1 1 18. Was the order of performance feedback correct (co-actor, group members, co-trainer, trainer, main actor)? 2 1 19. Were new Social Skills Homework Reports given to each youth as homework and the top section filled out? 3 1

20. Was behavior appropriately managed? 3 1

21. Did the session pace keep the students interested and active? 3 1

22. Did the students appear to understand the skill being taught in this session? 3 1

23. Does the primary instructor interact with the youth in a positive manner ? 11 1

24. Does the co-instructor interact with the youth in a positive manner? 3 1

25. Does the co-instructor aid the instructor in delivering the curriculum? 3 1

Instructor Score 100 Co-Instructor Score 9

265

Moral Reasoning Session Adherence

Washington State ART Assessment

Jurisdiction: Melbourne Aust. Date: July 2005 Instructor: Matt Currie Co-Instructor: Cath Powell

Week Number: 6

YES=1 NO=0

Please place a "1" in the box that best describes what happened in this session. Item

Weight ▼

1. Was the Problem Situation Chart made before the group? 3 1

2. Did the instructor study the Problem Situation Chart for patterns of thinking and decide in what order to discuss the questions? 11 1

3. Was a positive climate established through welcoming students? 3 1 4. Were issues since the last moral reasoning session dealt with?. 3 1

5. Were Group Norms for the discussion session reviewed (page 111 in the book)? 3 1

6. Were the four thinking errors reviewed? 3 1 7. Was the Problem Situation read to the class? 3 1 8. Was the real problem correctly defined and related to the lives of the group members? 11 1

9. Was moral maturity established through: a) Eliciting mature responses first? 5 1 b) Reconstructing less mature responses, and 5 1

c) Listing them on an easel pad, chalk board, or white board? 5 1 10. Was there an attempt made to make positive decision and mature reasons unanimous or a group decision? 2 1 11. Were comments by each group member acknowledged? 3 1

12. Were individuals within the group who evidenced more mature reasoning ecouraged to explain their reasoning? 3 1 13. Was the whole group praised for positive decisions and mature reasons? 3 1

14. Were all group members involved in the discussion? 11 1

15. Did the instructor encourage clear explanation of each person's answer? 3 1 16. Was behavior appropriately managed during the session? 3 1

17. Did the session pace keep the students interested and active? 3 1 18. Did the instructor remain objective during the session? 3 1

19. Does the primary instructor interact with the youth in a positive manner? 11 1

20. Does the co-instructor interact with the youth in a positive manner? 3 1 21. Does the co-instructor aid the trainer in delivering the intervention? 3 1 Instructor Score 100 Co-instructor Score 9 Comments

266

Washington State

Aggression Replacement Training

Trainer Competence Rating

Delivery Skills

Adherence Clarity: Understandable by

youth

Simplicity: Without unnecessary

complication or explanation

Objectivity: Presented as factual content, without moralizing or debate

Pacing: Not too fast, not too

slow

Engagement: Interesting and

involving

Total

Social Skills Training

3 3 3 3 3 3 18

Anger Control Training

3 3 3 3 3 3 18

Moral Reasoning 3 3 2 3 2 3 16

Total 9 9 8 9 8 9 52

Scoring:

0 = Not Competent

1 = Borderline Competent

2 = Competent

3 = Highly Competent

267

Matt Currie

Social Skills Training (week 6)

Comments: The Trainer and Co-trainer are a very effective team. They balance each other well. It is clear who is the main Trainer. The group norms were effectively used by the Trainers to manage group behavior in an assertive but not demeaning manner which avoided power struggles. Number 5 - The Trainer conducted a good review of the Skill Homework. At least one youth had used the skill (Dealing with Group Pressure) poorly. The Trainer could have analyzed the failure of the skill by having the youth look at each step and praising the youth for success up to the point that he gave into the group. One could phase it as a failure to complete the last step successfully. The Trainer did a particularly nice job of dealing with one youth who said the skill was "Stupid". In spite of the youth's initial resistance the Trainer was able to obtain a commitment from the youth to try the skill in the future. Number 10 - There are official definitions of each skill on page 25 of the Washington State A.R.T. Manual, Revision 16. There are language issues with our different dialects of English. The Trainer defined the new skill so the youth appeared to understand it. I encourage the Trainer to emphasize that this skill is about deciding on our role in the problem, since most aggressive youth use the thinking error of Blaming Others and see problems as primarily caused by others. Number 11 - The Trainer could read each step of the skill during the model if the youth do not appear to understand what the Trainer is doing. These youth appeared to understand what the trainer was doing.

Anger Control Training (week 6)

Comments: This was another well run group. There had evidently been a critical incident back at the residential unit just prior to group. The Trainer and Co-trainer did an excellent job of managing the extra energy this had created in the group. The review of the previous week’s concept of Thinking Ahead went very well. The Trainer did a good job of moving on when one youth said he experience no “Internal Consequence” of behavior. The Trainer did a great job of working with one youth around a Hassle that the youth had about changing rooms. The youth ended up endorsing Anger Control as something that really worked for him in that particular situation. The Angry Behavior Cycle was correctly introduced as what we do that others become angry about. The Trainer should sell this concept as our ability to take more power and control over our own lives by controlling the Triggers we through out. This continues the theme that Anger Control Training is about gaining more power and control over our own lives by controlling ourselves. It also avoids the youth saying they do not want the change any of their behaviors. The focus becomes taking more power and control not changing. T he Trainer and Co-trainer did a very good job of working with the youth on the things that they do that other become angry about. The Trainer and Co-trainer work well as a team when one youth (Hassan) was having trouble. The Trainer proceeded with the class while the Co-trainer worked individually with the youth.

Moral Reasoning (week 6)

Matt you have great rapport with this group. I think you need to just keep track of when you are losing the attention of the group, as in pace- for the most part there was a really great lively discussion happening, however just toward the end of the group they were getting ants in their pants. I have a couple of suggestions to help with running too long- 1) we don't review SS or ACT in MR session; and 2) you don't need to spend to much time on questions were it becomes clear the group is split or heading in an anitsocial response set. WARM-UP EXERCISE- this was brilliant, I really enjoyed how you laid the ground work for the young people around establishing their relationship values "what works" + " what get in the way". This laid the foundation for what proved to be some very mature responses from the young people throughout the session. THINKING ERRORS- again great use of thinking errors in session having young people do a "bah, bah" sound. I also liked your reframe of the revenge question in Marks Problem (i.e., do you want revenge to be part of your life"). Overall you facilitated really well.

268

Appendix D: Single case study

An Illustrative Single Case Study: The case of “C”

Background

In 2005, C was serving a six month custodial sentence at a Victorian Youth Justice Centre

for assault and assault related offences, reportedly while substance affected and in the company of

peers. At the time of participation in ART, C was 18 years old. He was the second oldest child in a

sibship of five. Prior to incarceration, he resided with his mother, three brothers and sister. The

family immigrated to Australia in 2002, fleeing civil war in their homeland. Client records indicated

that the family was well educated and of some standing in their community prior to the civil unrest.

Family relationships were reported to be close and loving. C’s father was assumed to be deceased

after 10 years of no contact. However, some time after the family arrived in Australia, news was

received that the father had survived and entered into a new marriage and begun a new family. The

client reported no ill-feeling toward his father, as his religion allowed for multiple wives. C conveyed

an eagerness to initiate contact with his father again, but in 2004 he received news that his father had

died. According to C, this news precipitated his violent and antisocial behaviour leading up to the

offence.

Participation in ART

C initially presented as angry and mistrustful within the group. In the first session, he

commented that “it’s a jungle out there” referring to life in and outside of the institution and the

apparent need to be hyper-vigilant for potential, real or imagined, threat. Within the first few sessions,

C disclosed information regarding his experiences while living through the Civil War, stating that he

had witnessed friends and relatives beaten and killed. The ART trainers had some concern regarding

269

C’s ability to follow the program material. Initially, he appeared to struggle with understanding the

program content and performing the sequential steps in the role-plays. The trainers discussed the

extent to which cultural differences (e.g., variations in social norms) might affect his ability to fully

benefit from ART.

By midway through the program (week 5), the trainers’ concerns regarding C’s ability to follow

the material appeared unfounded, the following note was written by the trainers at that time-

“[the client] is excited because he thinks he is getting the process”.

In week 8 we wrote-

“[the client] presented calm and focused. Challenged other group members appropriately.

Completed tasks well. Appears to be practicing ART steps outside of group”.

At that time, C was challenged to respond to a series of incidents involving another young

person making ongoing threats and racial taunts. During the break of the session in week 8, the ART

trainers witnessed such an incident. Later the client stated that he had drawn upon some of the skills

he learned in the program to control his anger. C’s Health Worker and other Centre staff noted that

he demonstrated a high level of self-control and mature problem solving skills by avoiding aggressive

confrontation and seeking staff support to manage what was a volatile situation. Information from C’s

case file notes in week nine of the 10-week program stated-

[The client’s] behaviour since his last Incident Report has improved dramatically. He is

learning that instead of dealing with issues through physical violence, mediation is the best

possible way to sort through issues. [The client] has requested mediation on a number of

occasions. Centre Management have acknowledged the improved behaviour and approach

that [the client] adopts when dealing with challenging situations. He has consistently

270

demonstrated his ability to remove himself from negative peers and obtains advice and

solutions to manage these situations.

Psychometric Testing: Pre to post scores

On the AQ total score C’s pre-treatment raw score was 113 placing him at the 85th percentile

and within the clinical range of scores. Whereas at post-treatment follow-up his self-reported level of

aggression was 77, at the 40th percentile and within the normal range of scores on the AQ. A similar

pattern of clinically significant reductions in aggressive behaviours and thoughts was evident for C

across all of the AQ subscales: Physical, pre-treatment score placed him, at the 92nd percentile,

compared to post-treatment score at the 60th percentile; Verbal, pre- 81st percentile, at post-treatment

his 55th percentile; Anger, pre-treatment 70th percentile, compared to a post-treatment 50th percentile;

Hostility, pre- 93rd percentile, compared to post-treatment 33rd percentile; and Indirect, pre- 23rd

percentile, while at post-treatment assessment his score placed him at 5th percentile. Apart from

Indirect aggression, which was within the normal range of scores at both pre- and post- treatment

assessment, all of C’s subscale scores on the AQ were within the clinical or subclinical range at pre-

treatment. At post-treatment follow-up C’s self-reported aggressive behaviours and thoughts were

within the normal range of scores compared to the normative sample provided by Buss and Warren

(2000). These results show clear evidence of clinically meaningful change for C subsequent to

participation in ART, which is consistent with the case note entries cited above.

C’s scores on the SSRS showed a slight increase in total social skills from pre-treatment with

a raw score of 47 (42nd percentile), compared to post-treatment with a raw score of 49 (50th

percentile). Both pre- and post- scores were within the normal range. Scores on the HIT total

remained within the clinical range at both pre-treatment (raw score= 3.38; 92nd percentile) and post-

271

treatment (raw score= 3.11; 86th percentile) assessment, indicating a slight, though non-significant,

reduction in cognitive distortions at treatment end.

C completed ART in mid- July 2005 and made a valuable contribution to the group. In

particular, he provided his peers with real-life modelling of ART in practice. These changes are partly

reflected in C’s pre- to post- scores on the outcome measures. Although scores on the HIT remained

stable and clinically significant, C demonstrated effective application of pro-social skills, emotional

self-regulation, and consequential thinking when he did not react to ongoing provocation from a peer

within the institutional setting.

272

Appendix E: Statistical output

E1. SPSS output for mean replacement method

Table E1: Inter-correlations for all dependent measures in the main study at T2

Table E2: Descriptives and main effect ANOVA results from T1 to T4 on the AQ (LOCF method)

Table E3: Descriptives and main effect ANOVA results from T1 to T4 on the HIT (LOCF method)

Table E4: Descriptives and main effect ANOVA results from T1 to T4 on the SPSI (LOCF method)

Table E5: Descriptives and main effect ANOVA results from T1 to T4 for Impulsiveness (LOCF)

Table E6: T-score means and main effect ANOVAS from T1 to T3 for the ABCL (LOCF method)

273

AQ Total (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

AQtotal 102.2500 25.99570 20

AQtotalT2 89.3500 24.51052 20

AQtotalT3 83.2600 20.72772 20

AQtotalT4 85.1420 22.01162 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .580 9.662 5 .086 .804 .929 .333

Tests of Within-Subjects Effects

Measure:MEASURE_1

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 4390.255 3 1463.418 4.911 .004 .205

Greenhouse-Geisser 4390.255 2.411 1820.995 4.911 .008 .205

Huynh-Feldt 4390.255 2.786 1575.794 4.911 .005 .205

factor1

Lower-bound 4390.255 1.000 4390.255 4.911 .039 .205

Sphericity Assumed 16986.903 57 298.016

Greenhouse-Geisser 16986.903 45.807 370.834

Huynh-Feldt 16986.903 52.935 320.900

Error(factor1)

Lower-bound 16986.903 19.000 894.048

274

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 3296.367 1 3296.367 7.880 .011 .293

Quadratic 1092.538 1 1092.538 3.691 .070 .163

factor1

Cubic 1.350 1 1.350 .008 .932 .000

Linear 7948.064 19 418.319

Quadratic 5623.802 19 295.990

Error(factor1)

Cubic 3415.038 19 179.739

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 3328.200 1 3328.200 12.648 .002 .400

Level 3 vs. Previous 3145.032 1 3145.032 5.253 .034 .217

factor1

Level 4 vs. Previous 839.290 1 839.290 1.732 .204 .084

Level 2 vs. Level 1 4999.800 19 263.147

Level 3 vs. Previous 11375.448 19 598.708

Error(factor1)

Level 4 vs. Previous 9204.495 19 484.447

275

AQ Physical Aggression (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

AQphy 26.8500 8.86314 20

AQphyT2 21.4000 8.99942 20

AQphyT3 20.9410 6.55720 20

AQphyT4 20.1420 7.48934 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .753 5.033 5 .412 .862 1.000 .333

Tests of Within-Subjects Effects

Measure:MEASURE_1

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 560.238 3 186.746 6.065 .001 .242

Greenhouse-Geisser 560.238 2.587 216.569 6.065 .002 .242

Huynh-Feldt 560.238 3.000 186.746 6.065 .001 .242

factor1

Lower-bound 560.238 1.000 560.238 6.065 .024 .242

Sphericity Assumed 1755.078 57 30.791

Greenhouse-Geisser 1755.078 49.151 35.708

Huynh-Feldt 1755.078 57.000 30.791

Error(factor1)

Lower-bound 1755.078 19.000 92.373

276

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 423.660 1 423.660 12.101 .003 .389

Quadratic 108.159 1 108.159 3.720 .069 .164

factor1

Cubic 28.420 1 28.420 1.005 .329 .050

Linear 665.202 19 35.011

Quadratic 552.413 19 29.074

Error(factor1)

Cubic 537.463 19 28.288

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 594.050 1 594.050 16.431 .001 .464

Level 3 vs. Previous 202.757 1 202.757 3.209 .089 .144

factor1

Level 4 vs. Previous 170.723 1 170.723 3.980 .061 .173

Level 2 vs. Level 1 686.950 19 36.155

Level 3 vs. Previous 1200.604 19 63.190

Error(factor1)

Level 4 vs. Previous 814.934 19 42.891

277

AQ Anger (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

AQanger 19.4115 5.03732 20

AQangerT2 18.2345 5.79046 20

AQangerT3 17.4705 4.63531 20

AQangerT4 15.9290 5.57687 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .567 10.060 5 .074 .707 .798 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 127.779 3 42.593 2.620 .059 .121

Greenhouse-Geisser 127.779 2.121 60.246 2.620 .082 .121

Huynh-Feldt 127.779 2.395 53.361 2.620 .074 .121

factor1

Lower-bound 127.779 1.000 127.779 2.620 .122 .121

Sphericity Assumed 926.760 57 16.259

Greenhouse-Geisser 926.760 40.298 22.998

Huynh-Feldt 926.760 45.498 20.369

Error(factor1)

Lower-bound 926.760 19.000 48.777

278

AQ Hostility (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

AQhos 21.6000 7.69415 20

AQhosT2 18.1000 5.75738 20

AQhosT3 17.2345 4.69635 20

AQhosT4 20.0000 5.16058 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .581 9.613 5 .087 .749 .854 .333

Tests of Within-Subjects Contrasts

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 32.098 1 32.098 1.464 .241 .072

Quadratic 196.282 1 196.282 9.114 .007 .324

factor1

Cubic .993 1 .993 .114 .739 .006

Linear 416.644 19 21.929

Quadratic 409.197 19 21.537

Error(factor1)

Cubic 164.894 19 8.679

279

Tests of Within-Subjects Contrasts

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 245.000 1 245.000 8.701 .008 .314

Level 3 vs. Previous 136.817 1 136.817 3.834 .065 .168

factor1

Level 4 vs. Previous 20.883 1 20.883 1.097 .308 .055

Level 2 vs. Level 1 535.000 19 28.158

Level 3 vs. Previous 678.072 19 35.688

Error(factor1)

Level 4 vs. Previous 361.584 19 19.031

280

AQ Indirect Aggression (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

AQind 16.5500 4.96806 20

AQindT2 14.4500 4.47772 20

AQindT3 15.0000 3.87977 20

AQindT4 14.4290 3.83494 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .373 17.482 5 .004 .705 .796 .333

Multivariate Testsb

Effect

Value F Hypothesis df Error df Sig.

Partial Eta

Squared

Pillai's Trace .460 4.821a 3.000 17.000 .013 .460

Wilks' Lambda .540 4.821a 3.000 17.000 .013 .460

Hotelling's Trace .851 4.821a 3.000 17.000 .013 .460

factor1

Roy's Largest Root .851 4.821a 3.000 17.000 .013 .460

281

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 88.200 1 88.200 11.040 .004 .368

Level 3 vs. Previous 5.000 1 5.000 .455 .508 .023

factor1

Level 4 vs. Previous 16.356 1 16.356 .766 .392 .039

Level 2 vs. Level 1 151.800 19 7.989

Level 3 vs. Previous 209.000 19 11.000

Error(factor1)

Level 4 vs. Previous 405.751 19 21.355

282

AQ Verbal Aggression (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

AQverbal 16.4000 3.80305 20

AQverbalT2 16.7000 4.83518 20

AQverbalT3 15.2345 3.36436 20

AQverbalT4 14.6420 3.41216 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .724 5.713 5 .336 .825 .958 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 56.365 3 18.788 1.985 .126 .095

Greenhouse-Geisser 56.365 2.475 22.778 1.985 .139 .095

Huynh-Feldt 56.365 2.874 19.613 1.985 .129 .095

factor1

Lower-bound 56.365 1.000 56.365 1.985 .175 .095

Sphericity Assumed 539.537 57 9.466

Greenhouse-Geisser 539.537 47.017 11.475

Huynh-Feldt 539.537 54.603 9.881

Error(factor1)

Lower-bound 539.537 19.000 28.397

283

HIT Total (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITtotal 3.1687 .70479 20

HITtotalT2 2.7806 .76124 20

HITtotalT3 2.5769 .88224 20

HITtotalT4 2.1317 .50313 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .809 3.751 5 .586 .878 1.000 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 11.184 3 3.728 11.659 .000 .380

Greenhouse-Geisser 11.184 2.633 4.247 11.659 .000 .380

Huynh-Feldt 11.184 3.000 3.728 11.659 .000 .380

factor1

Lower-bound 11.184 1.000 11.184 11.659 .003 .380

Sphericity Assumed 18.227 57 .320

Greenhouse-Geisser 18.227 50.033 .364

Huynh-Feldt 18.227 57.000 .320

Error(factor1)

Lower-bound 18.227 19.000 .959

284

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 10.986 1 10.986 49.153 .000 .721

Quadratic .016 1 .016 .039 .846 .002

factor1

Cubic .182 1 .182 .577 .457 .029

Linear 4.247 19 .224

Quadratic 7.998 19 .421

Error(factor1)

Cubic 5.981 19 .315

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 3.013 1 3.013 4.524 .047 .192

Level 3 vs. Previous 3.163 1 3.163 5.214 .034 .215

factor1

Level 4 vs. Previous 10.092 1 10.092 34.114 .000 .642

Level 2 vs. Level 1 12.653 19 .666

Level 3 vs. Previous 11.527 19 .607

Error(factor1)

Level 4 vs. Previous 5.621 19 .296

285

HIT Self-centred (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITsc 3.1778 .82945 20

HITscT2 2.7944 .76919 20

HITscT3 2.6727 .86629 20

HITscT4 2.2074 .53087 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .721 5.796 5 .327 .813 .942 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 9.597 3 3.199 8.544 .000 .310

Greenhouse-Geisser 9.597 2.440 3.934 8.544 .000 .310

Huynh-Feldt 9.597 2.825 3.397 8.544 .000 .310

factor1

Lower-bound 9.597 1.000 9.597 8.544 .009 .310

Sphericity Assumed 21.343 57 .374

Greenhouse-Geisser 21.343 46.351 .460

Huynh-Feldt 21.343 53.684 .398

Error(factor1)

Lower-bound 21.343 19.000 1.123

286

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 9.197 1 9.197 36.497 .000 .658

Quadratic .034 1 .034 .058 .812 .003

factor1

Cubic .366 1 .366 1.254 .277 .062

Linear 4.788 19 .252

Quadratic 11.007 19 .579

Error(factor1)

Cubic 5.548 19 .292

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 2.939 1 2.939 2.843 .108 .130

Level 3 vs. Previous 1.964 1 1.964 3.713 .069 .163

factor1

Level 4 vs. Previous 9.091 1 9.091 26.873 .000 .586

Level 2 vs. Level 1 19.641 19 1.034

Level 3 vs. Previous 10.052 19 .529

Error(factor1)

Level 4 vs. Previous 6.428 19 .338

287

HIT Blaming Others (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITbo 3.0350 .75622 20

HITboT2 2.7300 .80269 20

HITboT3 2.5640 .87054 20

HITboT4 2.0160 .55428 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .866 2.550 5 .769 .908 1.000 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 10.954 3 3.651 9.703 .000 .338

Greenhouse-Geisser 10.954 2.724 4.022 9.703 .000 .338

Huynh-Feldt 10.954 3.000 3.651 9.703 .000 .338

factor1

Lower-bound 10.954 1.000 10.954 9.703 .006 .338

Sphericity Assumed 21.450 57 .376

Greenhouse-Geisser 21.450 51.749 .415

Huynh-Feldt 21.450 57.000 .376

Error(factor1)

Lower-bound 21.450 19.000 1.129

288

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 10.388 1 10.388 30.193 .000 .614

Quadratic .295 1 .295 .621 .440 .032

factor1

Cubic .271 1 .271 .877 .361 .044

Linear 6.537 19 .344

Quadratic 9.031 19 .475

Error(factor1)

Cubic 5.883 19 .310

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 1.861 1 1.861 2.141 .160 .101

Level 3 vs. Previous 2.029 1 2.029 3.408 .081 .152

factor1

Level 4 vs. Previous 11.562 1 11.562 29.140 .000 .605

Level 2 vs. Level 1 16.509 19 .869

Level 3 vs. Previous 11.312 19 .595

Error(factor1)

Level 4 vs. Previous 7.539 19 .397

289

HIT Minimizing/Mislabeling (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITmm 3.2222 .88816 20

HITmmT2 2.8000 .89660 20

HITmmT3 2.4972 1.03864 20

HITmmT4 2.0171 .58668 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .660 7.354 5 .196 .797 .920 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 15.456 3 5.152 11.255 .000 .372

Greenhouse-Geisser 15.456 2.391 6.465 11.255 .000 .372

Huynh-Feldt 15.456 2.759 5.603 11.255 .000 .372

factor1

Lower-bound 15.456 1.000 15.456 11.255 .003 .372

Sphericity Assumed 26.092 57 .458

Greenhouse-Geisser 26.092 45.428 .574

Huynh-Feldt 26.092 52.415 .498

Error(factor1)

Lower-bound 26.092 19.000 1.373

290

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 15.352 1 15.352 51.915 .000 .732

Quadratic .017 1 .017 .030 .865 .002

factor1

Cubic .088 1 .088 .170 .684 .009

Linear 5.618 19 .296

Quadratic 10.654 19 .561

Error(factor1)

Cubic 9.819 19 .517

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 3.565 1 3.565 4.128 .056 .178

Level 3 vs. Previous 5.282 1 5.282 5.201 .034 .215

factor1

Level 4 vs. Previous 13.537 1 13.537 38.402 .000 .669

Level 2 vs. Level 1 16.410 19 .864

Level 3 vs. Previous 19.296 19 1.016

Error(factor1)

Level 4 vs. Previous 6.698 19 .353

291

HIT Assuming the Worst (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITaw 3.2091 .68419 20

HITawT2 2.7818 .72632 20

HITawT3 2.5461 .84718 20

HITawT4 2.3897 .55684 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .822 3.468 5 .629 .885 1.000 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 7.636 3 2.545 7.652 .000 .287

Greenhouse-Geisser 7.636 2.655 2.876 7.652 .000 .287

Huynh-Feldt 7.636 3.000 2.545 7.652 .000 .287

factor1

Lower-bound 7.636 1.000 7.636 7.652 .012 .287

Sphericity Assumed 18.961 57 .333

Greenhouse-Geisser 18.961 50.449 .376

Huynh-Feldt 18.961 57.000 .333

Error(factor1)

Lower-bound 18.961 19.000 .998

292

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 7.256 1 7.256 24.208 .000 .560

Quadratic .367 1 .367 1.010 .328 .050

factor1

Cubic .013 1 .013 .038 .848 .002

Linear 5.695 19 .300

Quadratic 6.904 19 .363

Error(factor1)

Cubic 6.362 19 .335

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 3.651 1 3.651 6.747 .018 .262

Level 3 vs. Previous 4.038 1 4.038 5.553 .029 .226

factor1

Level 4 vs. Previous 4.158 1 4.158 12.856 .002 .404

Level 2 vs. Level 1 10.283 19 .541

Level 3 vs. Previous 13.816 19 .727

Error(factor1)

Level 4 vs. Previous 6.145 19 .323

293

HIT Lying (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITlying 3.2813 .78836 20

HITlyingT2 2.8188 .84339 20

HITlyingT3 2.7423 .88711 20

HITlyingT4 2.4630 .55965 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .777 4.468 5 .485 .844 .984 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 6.922 3 2.307 7.483 .000 .283

Greenhouse-Geisser 6.922 2.531 2.735 7.483 .001 .283

Huynh-Feldt 6.922 2.952 2.345 7.483 .000 .283

factor1

Lower-bound 6.922 1.000 6.922 7.483 .013 .283

Sphericity Assumed 17.576 57 .308

Greenhouse-Geisser 17.576 48.087 .366

Huynh-Feldt 17.576 56.089 .313

Error(factor1)

Lower-bound 17.576 19.000 .925

294

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 6.407 1 6.407 23.990 .000 .558

Quadratic .168 1 .168 .454 .508 .023

factor1

Cubic .347 1 .347 1.203 .287 .060

Linear 5.075 19 .267

Quadratic 7.024 19 .370

Error(factor1)

Cubic 5.477 19 .288

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 4.278 1 4.278 9.220 .007 .327

Level 3 vs. Previous 1.894 1 1.894 2.801 .111 .128

factor1

Level 4 vs. Previous 4.693 1 4.693 14.533 .001 .433

Level 2 vs. Level 1 8.816 19 .464

Level 3 vs. Previous 12.850 19 .676

Error(factor1)

Level 4 vs. Previous 6.136 19 .323

295

HIT Stealing (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITsteal 2.7591 1.04333 20

HITstealT2 2.5045 .84124 20

HITstealT3 2.3162 .97784 20

HITstealT4 1.6813 .47991 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .709 6.091 5 .298 .835 .971 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 12.695 3 4.232 8.908 .000 .319

Greenhouse-Geisser 12.695 2.504 5.070 8.908 .000 .319

Huynh-Feldt 12.695 2.914 4.356 8.908 .000 .319

factor1

Lower-bound 12.695 1.000 12.695 8.908 .008 .319

Sphericity Assumed 27.077 57 .475

Greenhouse-Geisser 27.077 47.572 .569

Huynh-Feldt 27.077 55.373 .489

Error(factor1)

Lower-bound 27.077 19.000 1.425

296

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 11.709 1 11.709 29.000 .000 .604

Quadratic .723 1 .723 1.268 .274 .063

factor1

Cubic .263 1 .263 .583 .454 .030

Linear 7.671 19 .404

Quadratic 10.841 19 .571

Error(factor1)

Cubic 8.564 19 .451

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 1.296 1 1.296 1.118 .304 .056

Level 3 vs. Previous 1.993 1 1.993 2.356 .141 .110

factor1

Level 4 vs. Previous 14.292 1 14.292 38.071 .000 .667

Level 2 vs. Level 1 22.026 19 1.159

Level 3 vs. Previous 16.071 19 .846

Error(factor1)

Level 4 vs. Previous 7.133 19 .375

297

HIT Oppositional Defiant (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITod 3.4650 .72277 20

HITodT2 3.0350 .78558 20

HITodT3 2.7065 .92820 20

HITodT4 2.3080 .59865 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .888 2.102 5 .835 .928 1.000 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 14.471 3 4.824 12.981 .000 .406

Greenhouse-Geisser 14.471 2.784 5.197 12.981 .000 .406

Huynh-Feldt 14.471 3.000 4.824 12.981 .000 .406

factor1

Lower-bound 14.471 1.000 14.471 12.981 .002 .406

Sphericity Assumed 21.180 57 .372

Greenhouse-Geisser 21.180 52.901 .400

Huynh-Feldt 21.180 57.000 .372

Error(factor1)

Lower-bound 21.180 19.000 1.115

298

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 14.436 1 14.436 54.940 .000 .743

Quadratic .005 1 .005 .011 .918 .001

factor1

Cubic .029 1 .029 .075 .788 .004

Linear 4.992 19 .263

Quadratic 8.698 19 .458

Error(factor1)

Cubic 7.489 19 .394

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 3.698 1 3.698 5.040 .037 .210

Level 3 vs. Previous 5.908 1 5.908 8.923 .008 .320

factor1

Level 4 vs. Previous 11.577 1 11.577 28.333 .000 .599

Level 2 vs. Level 1 13.942 19 .734

Level 3 vs. Previous 12.579 19 .662

Error(factor1)

Level 4 vs. Previous 7.764 19 .409

299

HIT Physical Aggression (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

HITpa 3.2000 .76914 20

HITpaT2 2.7800 .89183 20

HITpaT3 2.5705 .91408 20

HITpaT4 1.9710 .58887 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .775 4.521 5 .478 .870 1.000 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 15.704 3 5.235 13.016 .000 .407

Greenhouse-Geisser 15.704 2.609 6.020 13.016 .000 .407

Huynh-Feldt 15.704 3.000 5.235 13.016 .000 .407

factor1

Lower-bound 15.704 1.000 15.704 13.016 .002 .407

Sphericity Assumed 22.925 57 .402

Greenhouse-Geisser 22.925 49.568 .463

Huynh-Feldt 22.925 57.000 .402

Error(factor1)

Lower-bound 22.925 19.000 1.207

300

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 15.183 1 15.183 41.780 .000 .687

Quadratic .161 1 .161 .314 .582 .016

factor1

Cubic .361 1 .361 1.090 .310 .054

Linear 6.905 19 .363

Quadratic 9.735 19 .512

Error(factor1)

Cubic 6.285 19 .331

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 3.528 1 3.528 4.903 .039 .205

Level 3 vs. Previous 3.520 1 3.520 4.307 .052 .185

factor1

Level 4 vs. Previous 15.459 1 15.459 38.388 .000 .669

Level 2 vs. Level 1 13.672 19 .720

Level 3 vs. Previous 15.526 19 .817

Error(factor1)

Level 4 vs. Previous 7.651 19 .403

301

SPSI-R:S Total (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

SPSItotal 11.2800 2.79729 20

T2SPSItotal 12.7200 2.94039 20

T3SPSItotal 12.7176 2.51401 20

T4SPSItotal 13.4000 2.34498 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .642 7.859 5 .165 .764 .874 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 47.814 3 15.938 4.915 .004 .206

Greenhouse-Geisser 47.814 2.291 20.868 4.915 .009 .206

Huynh-Feldt 47.814 2.623 18.230 4.915 .006 .206

factor1

Lower-bound 47.814 1.000 47.814 4.915 .039 .206

Sphericity Assumed 184.825 57 3.243

Greenhouse-Geisser 184.825 43.533 4.246

Huynh-Feldt 184.825 49.834 3.709

Error(factor1)

Lower-bound 184.825 19.000 9.728

302

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 40.420 1 40.420 8.691 .008 .314

Quadratic 2.870 1 2.870 .853 .367 .043

factor1

Cubic 4.524 1 4.524 2.643 .120 .122

Linear 88.369 19 4.651

Quadratic 63.932 19 3.365

Error(factor1)

Cubic 32.525 19 1.712

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 41.472 1 41.472 9.613 .006 .336

Level 3 vs. Previous 10.300 1 10.300 3.649 .071 .161

factor1

Level 4 vs. Previous 26.948 1 26.948 3.553 .075 .158

Level 2 vs. Level 1 81.968 19 4.314

Level 3 vs. Previous 53.632 19 2.823

Error(factor1)

Level 4 vs. Previous 144.116 19 7.585

303

SPSI-R:S Positive Problem Orientation (Mean replacement method))

Descriptive Statistics

Mean Std. Deviation N

PPO 2.4900 .92332 20

PPOT2 2.8400 .76667 20

SMEAN(sumPPOT3) 14.8235 2.92422 20

SMEAN(sumPPOT4) 15.6429 3.62168 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .553 10.484 5 .063 .735 .836 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 111.890 3 37.297 3.819 .015 .167

Greenhouse-Geisser 111.890 2.206 50.716 3.819 .026 .167

Huynh-Feldt 111.890 2.508 44.607 3.819 .021 .167

factor1

Lower-bound 111.890 1.000 111.890 3.819 .066 .167

Sphericity Assumed 556.670 57 9.766

Greenhouse-Geisser 556.670 41.918 13.280

Huynh-Feldt 556.670 47.658 11.680

Error(factor1)

Lower-bound 556.670 19.000 29.298

304

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 109.246 1 109.246 6.761 .018 .262

Quadratic 2.317 1 2.317 .302 .589 .016

factor1

Cubic .327 1 .327 .060 .809 .003

Linear 306.991 19 16.157

Quadratic 145.818 19 7.675

Error(factor1)

Cubic 103.860 19 5.466

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 45.000 1 45.000 5.897 .025 .237

Level 3 vs. Previous 52.717 1 52.717 3.714 .069 .164

factor1

Level 4 vs. Previous 72.328 1 72.328 3.386 .081 .151

Level 2 vs. Level 1 145.000 19 7.632

Level 3 vs. Previous 269.700 19 14.195

Error(factor1)

Level 4 vs. Previous 405.826 19 21.359

305

SPSI-R:S Negative Problem Orientation (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

sumNPO1 7.5000 4.74064 20

sumT2NPO1 6.4500 3.48644 20

SMEAN(sumT3NPO1) 6.4118 3.04456 20

SMEAN(sumT4NPO1) 6.0714 3.18656 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .863 2.611 5 .760 .913 1.000 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 22.941 3 7.647 .830 .483 .042

Greenhouse-Geisser 22.941 2.740 8.372 .830 .474 .042

Huynh-Feldt 22.941 3.000 7.647 .830 .483 .042

factor1

Lower-bound 22.941 1.000 22.941 .830 .374 .042

Sphericity Assumed 525.069 57 9.212

Greenhouse-Geisser 525.069 52.062 10.085

Huynh-Feldt 525.069 57.000 9.212

Error(factor1)

Lower-bound 525.069 19.000 27.635

306

SPSI-R:S Rational Problem Solving (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

sumRPS 8.6500 3.77352 20

sumRPST2 10.3500 4.77135 20

SMEAN(sumRPST3) 10.4706 3.86778 20

SMEAN(sumRPST4) 12.0714 4.90396 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .709 6.094 5 .298 .804 .929 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 117.256 3 39.085 3.498 .021 .155

Greenhouse-Geisser 117.256 2.411 48.644 3.498 .031 .155

Huynh-Feldt 117.256 2.786 42.095 3.498 .024 .155

factor1

Lower-bound 117.256 1.000 117.256 3.498 .077 .155

Sphericity Assumed 636.958 57 11.175

Greenhouse-Geisser 636.958 45.800 13.907

Huynh-Feldt 636.958 52.925 12.035

Error(factor1)

Lower-bound 636.958 19.000 33.524

307

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 107.846 1 107.846 8.015 .011 .297

Quadratic .049 1 .049 .004 .953 .000

factor1

Cubic 9.362 1 9.362 1.442 .245 .071

Linear 255.665 19 13.456

Quadratic 257.957 19 13.577

Error(factor1)

Cubic 123.336 19 6.491

308

SPSI-R:S Impulsive/Carelessness Style (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

sumICS1 8.7000 3.37327 20

sumT2ICS1 7.9000 3.17722 20

SMEAN(sumT3ICS1) 8.3529 4.47144 20

SMEAN(sumT4ICS1) 8.7857 3.61544 20

Mauchly's Test of Sphericityb

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .682 6.773 5 .239 .780 .897 .333

Tests of Within-Subjects Effects

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 9.724 3 3.241 .328 .805 .017

Greenhouse-Geisser 9.724 2.341 4.154 .328 .755 .017

Huynh-Feldt 9.724 2.690 3.615 .328 .784 .017

factor1

Lower-bound 9.724 1.000 9.724 .328 .574 .017

Sphericity Assumed 563.186 57 9.880

Greenhouse-Geisser 563.186 44.475 12.663

Huynh-Feldt 563.186 51.113 11.019

Error(factor1)

Lower-bound 563.186 19.000 29.641

309

SPSI-R:S Avoidance Style (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

sumAS1 8.5000 6.23656 20

sumT2AS1 6.3500 5.56564 20

SMEAN(sumT3AS1) 6.9412 3.92657 20

SMEAN(sumT4AS1) 5.8571 3.72555 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .750 5.095 5 .405 .837 .975 .333

Tests of Within-Subjects Effects

Measure:MEASURE_1

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 79.023 3 26.341 2.033 .119 .097

Greenhouse-Geisser 79.023 2.511 31.469 2.033 .131 .097

Huynh-Feldt 79.023 2.925 27.020 2.033 .121 .097

factor1

Lower-bound 79.023 1.000 79.023 2.033 .170 .097

Sphericity Assumed 738.450 57 12.955

Greenhouse-Geisser 738.450 47.712 15.477

Huynh-Feldt 738.450 55.567 13.289

Error(factor1)

Lower-bound 738.450 19.000 38.866

310

I7 Impulsiveness (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

EIStotal 12.5000 3.91354 20

EIStotalT2 11.1000 4.29075 20

EIStotalT3 9.8230 3.55758 20

EIStotalT4 8.9290 4.08417 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .863 2.606 5 .761 .920 1.000 .333

Tests of Within-Subjects Effects

Measure:MEASURE_1

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 145.108 3 48.369 4.909 .004 .205

Greenhouse-Geisser 145.108 2.760 52.577 4.909 .005 .205

Huynh-Feldt 145.108 3.000 48.369 4.909 .004 .205

factor1

Lower-bound 145.108 1.000 145.108 4.909 .039 .205

Sphericity Assumed 561.652 57 9.854

Greenhouse-Geisser 561.652 52.438 10.711

Huynh-Feldt 561.652 57.000 9.854

Error(factor1)

Lower-bound 561.652 19.000 29.561

311

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Linear 143.760 1 143.760 14.583 .001 .434

Quadratic 1.280 1 1.280 .121 .732 .006

factor1

Cubic .068 1 .068 .007 .932 .000

Linear 187.309 19 9.858

Quadratic 200.696 19 10.563

Error(factor1)

Cubic 173.647 19 9.139

Tests of Within-Subjects Contrasts

Measure:MEASURE_1

Source factor1 Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Level 2 vs. Level 1 39.200 1 39.200 1.887 .186 .090

Level 3 vs. Previous 78.171 1 78.171 8.334 .009 .305

factor1

Level 4 vs. Previous 97.859 1 97.859 5.681 .028 .230

Level 2 vs. Level 1 394.800 19 20.779

Level 3 vs. Previous 178.207 19 9.379

Error(factor1)

Level 4 vs. Previous 327.264 19 17.224

312

ABCL- Total (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

ABCL Full Scale

Worker

55.80 8.141 20

T2ABCLtot 52.50 9.498 20

T3ABCLtot 52.3325 7.34608 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .653 7.664 2 .022 .743 .791 .500

Multivariate Testsb

Effect

Value F Hypothesis df Error df Sig.

Partial Eta

Squared

Pillai's Trace .288 3.644a 2.000 18.000 .047 .288

Wilks' Lambda .712 3.644a 2.000 18.000 .047 .288

Hotelling's Trace .405 3.644a 2.000 18.000 .047 .288

factor1

Roy's Largest Root .405 3.644a 2.000 18.000 .047 .288

313

ABCL- Internalizing (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

ABCL Internalizing 55.15 10.122 20

T2ABCLint 51.25 8.675 20

T3ABCLint 51.8000 9.21783 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .997 .058 2 .971 .997 1.000 .500

Tests of Within-Subjects Effects

Measure:MEASURE_1

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 178.233 2 89.117 1.748 .188 .084

Greenhouse-Geisser 178.233 1.994 89.404 1.748 .188 .084

Huynh-Feldt 178.233 2.000 89.117 1.748 .188 .084

factor1

Lower-bound 178.233 1.000 178.233 1.748 .202 .084

Sphericity Assumed 1937.767 38 50.994

Greenhouse-Geisser 1937.767 37.878 51.158

Huynh-Feldt 1937.767 38.000 50.994

Error(factor1)

Lower-bound 1937.767 19.000 101.988

314

ABCL- Externalizing (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

ABCL Externalizing 58.85 7.191 20

T2ABCLext 57.38 10.285 20

T3ABCLext 56.0000 9.69536 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .721 5.891 2 .053 .782 .839 .500

Tests of Within-Subjects Effects

Measure:MEASURE_1

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 81.257 2 40.629 1.043 .362 .052

Greenhouse-Geisser 81.257 1.564 51.968 1.043 .349 .052

Huynh-Feldt 81.257 1.679 48.402 1.043 .353 .052

factor1

Lower-bound 81.257 1.000 81.257 1.043 .320 .052

Sphericity Assumed 1480.048 38 38.949

Greenhouse-Geisser 1480.048 29.708 49.819

Huynh-Feldt 1480.048 31.897 46.401

Error(factor1)

Lower-bound 1480.048 19.000 77.897

315

ABCL- Aggression (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

ABCL Aggression 56.95 6.832 20

T2ABCLagg 57.25 8.006 20

T3ABCLagg 56.3325 7.59270 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .664 7.382 2 .025 .748 .798 .500

Multivariate Testsb

Effect

Value F Hypothesis df Error df Sig.

Partial Eta

Squared

Pillai's Trace .017 .156a 2.000 18.000 .857 .017

Wilks' Lambda .983 .156a 2.000 18.000 .857 .017

Hotelling's Trace .017 .156a 2.000 18.000 .857 .017

factor1

Roy's Largest Root .017 .156a 2.000 18.000 .857 .017

316

ABCL- Rule Breaking Behaviour (Mean replacement method)

Descriptive Statistics

Mean Std. Deviation N

ABCL Rule Breaking 64.25 10.818 20

T2ABCLrule 61.75 6.568 20

T3ABCLrule 60.9325 7.48660 20

Mauchly's Test of Sphericityb

Measure:MEASURE_1

Epsilona Within Subjects Effect

Mauchly's W

Approx. Chi-

Square df Sig. Greenhouse-

Geisser Huynh-Feldt Lower-bound

dimension1 factor1 .951 .907 2 .635 .953 1.000 .500

Tests of Within-Subjects Effects

Measure:MEASURE_1

Source Type III Sum

of Squares df Mean Square F Sig.

Partial Eta

Squared

Sphericity Assumed 119.494 2 59.747 1.425 .253 .070

Greenhouse-Geisser 119.494 1.906 62.683 1.425 .254 .070

Huynh-Feldt 119.494 2.000 59.747 1.425 .253 .070

factor1

Lower-bound 119.494 1.000 119.494 1.425 .247 .070

Sphericity Assumed 1593.502 38 41.934

Greenhouse-Geisser 1593.502 36.220 43.995

Error(factor1)

Huynh-Feldt 1593.502 38.000 41.934

317

Table E1

Intercorrelations among all dependent measures at T2 (N=20)

AQ -phy -ang -hos -ind -ver HIT -sc -bo -mm -aw -l -s -od -pa SPSI -ppo -npo -rps -ics -as I7

AQ -

- phy .88** -

- ang .87** .84** -

- hos .64** .34. .43* -

- ind .82** .58** .52* .59** -

- ver .81** .61** .67** .31 .74** -

HIT .62** .65** .65** .49* .37 .26 -

- sc .50* .53** .53** .47** .29 .16 .95** -

- bo .54** .60** .62** .42* .30 .17 .96** .91** -

- mm .67** .70** .68** .50* .45* .31 .95** .85** .87** -

- aw .63** .65** .66** .44* .35 .34 .97** .90** .90** .91** -

- l .56** .48* .54** .70** .40* .18 .56** .80** .75** .87** .80** -

- s .48* .50* .54** .42* .34 .12 .90** .89** .93** .90** .89** .82** -

- od .56** .60** .63** .40* .25 .29 .93** .89** .92** .81** .93** .67** .82** -

- pa .63** .76** .66** .28 .33 .36 .90** .86** .85** .87** .87** .60** .76** .85** -

SPSI -.22 -.04 -.24 -.55** -.17 .09 -.38 -.33 -.39* -.31 -.40* -.45* -.49* -.38* -.07 -

- ppo .18 .14 -.07 -.06 .34 .43* -.30 -.26 -.32 -.23 -.33 -.26 -.44* -.29 -.09 .68** -

- npo .40* .09 .33 .72** .40* .23 .16 .12 .11 .19 .17 .40* .15 .11 -.03 -.68** -.18 -

- rps -.05 -.14 -.12 -.14 .02 .25 -.52** -.44* -.50* -.51* -.56* -.47* -.66** -.48* -.29 .70** .66** -.10 -

- ics .52* .32 .43* .44* .53** .46* .27 .25 .32 .21 .27 .23 .31 .34 .12 -.54** -.04 .33 -.25 -

- as .16 -.11 .05 .60** .25 -.06 .07 .10 .12 -.02 .08 .21 .13 .10 -.15 -.82** -.31 .80** -.26 .47* -

I7 .25 .18 .12 .36 .32 .02 .49* .43* .53** .45* .45* .39* .60** .46* .31 -.47* -.28 .10 -.48* .51* .28 -

Note. All 1-tailed correlations; *p<.05, **p<.01 AQ = Aggression Questionnaire; phy = Physical Aggression, ang = Anger, hos = Hostility, ind = Indirect Aggression, ver = Verbal Aggression; HIT = How I Think Questionnaire: sc = Self-Centred, bo = Blaming Others, mm = Minimising/Mislabelling, aw = assuming the Worst, l = Lying, S = Stealing, od = Oppositional Defiant, pa = Physical Aggression; SPSI = Social Problem Solving Inventory-Revised: Short Form: ppo = Positive Problem Orientation, npo = Negative Problem Orientation, rps = Rational Problem Solving, ics = Impulsive/Careless Style, as = Avoidance Style; I7 = Impulsiveness Scale

318

Table E2

Descriptives and main effect ANOVA results from T1 to T4 on the AQ (LOCF method)

Pre (T1) Post (T2) 6-month (T3) 24-month (T4) Main Effect

M (SD) M (SD) M (SD) M (SD) F(3,57) p Partial WW

AQ Total 102.25 (26.00) 89.35 (24.51) 88.45 (20.97) 86.30 (25.13) 4.71 .005 .20

Physical 26.85 (8.86) 21.40 (9.00) 21.05 (7.36) 20.14 (7.49) 6.16 .001 .25

Anger 19.41 (5.04) 18.23 (5.79) 18.05 (4.86) 16.90 (6.32) 2.12* .13 ns

Hostility 21.60 (7.69) 18.10 (5.76) 17.95 (5.40) 19.40 (6.22) 3.71 .01 .16

Indirect 16.55 (4.97) 14.45 (4.48) 15.35 (3.99) 14.60 (4.50) 5.49* .008 .49

Verbal 16.40 (3.80) 16.70 (4.84) 16.05 (4.05) 14.75 (4.25) 1.74* .20 ns

Note. AQ = Aggression Questionnaire; Physical = Physical Aggression; Indirect = Indirect Aggression; Verbal = Verbal Aggression; * = Multivariate

Wilks’ Lambda (df = 3,17) reported due to lack of sphericity; ns = non-significant

319

Table E3

Descriptives and main effect ANOVA results from T1 to T4 on the HIT (LOCF method)

Pre (T1) Post (T2) 6-month (T3) 24-month (T4) Main Effect

M (SD) M (SD) M (SD) M (SD) F(3,57) p Partial WW

HIT Total 3.17 (.70) 2.78 (.76) 2.48 (.93) 2.31 (.76) 8.07 .000 .30

Self-Centered 3.18 (.83) 2.79 (.77) 2.63 (.88) 2.42 (.83) 5.04 .004 .21

Blaming Others 3.04 (.76) 2.73 (.80) 2.54 (.93) 2.26 (.76) 5.56 .002 .23

Minimizing/ Mislabel 3.22 (.89) 2.80 (.90) 2.44 (1.07) 2.24 (.76) 8.40 .000 .31

Assuming the Worst 3.21 (.68) 2.78 (.73) 2.56 (.86) 2.49 (.76) 6.15 .001 .24

Lying 3.28 (.79) 2.82 (.84) 2.68 (.91) 2.61 (.81) 6.58 .001 .26

Stealing 2.76 (1.04) 2.50 (.84) 2.26 (.99) 1.97 (.85) 4.85 .005 .20

Oppositional Defiant 3.47 (.72) 3.04 (.79) 2.72 (.95) 2.50 (.80) 9.06 .000 .32

Physical Aggression 3.20 (.77) 2.78 (.89) 2.55 (1.00) 2.15 (.72) 9.41 .000 .33

320

Table E4

Descriptives and main effect ANOVA results from T1 to T4 on the SPSI (LOCF method)

Pre (T1) Post (T2) 6-month (T3) 24-month (T4) Main Effect

M (SD) M (SD) M (SD) M (SD) F(3,57) p Partial WW

SPSI Total 11.28 (2.80) 12.72 (2.94) 12.48 (2..60) 12.89 (2.87) 3.80 .01 .17

PPO 12.45 (4.62) 13.95 (4.44) 14.95 (3.12) 15.05 (4.12) 3.36 .03 ns

NPO 12.50 (4.74) 13.55 (3.49) 12.90 (3.96) 13.60 (3.84) .50 .68 ns

RPS 8.65 (3.77) 10.35 (4.77) 10.65 (3.90) 11.40 (5.28) 2.39 .07 ns

ICS 11.30 (3.37) 12.10 (3.18) 11.50 (4.55) 11.20 (4.25) .34 .79 ns

AS 11.50 (6.24) 13.65 (5.57) 12.40 (4.91) 13.20 (4.21) 1.57 .21 ns

Note. SPSI = Social Problem Solving Inventory; PPO = Positive Problem Orientation; NPO = Negative Problem Orientation; RPS = Rational

Problem Solving; ICS = Impulsive/Careless Style; AS = Avoidance Style; ns = non-significant

321

Table E5

Descriptives and main effect ANOVA results from T1 to T4 for Impulsiveness (LOCF method)

Pre (T1) Post (T2) 6-month (T3) 24-month (T4) Main Effect

M (SD) M (SD) M (SD) M (SD) F(3,57) p Partial WW

Impulsiveness 12.50 (3.91) 11.10 (4.29) 10.10 (3.68) 9.35 (4.88) 4.05 .01 .18

322

Table E6

T-score descriptivs and main effect ANOVAS from T1 to T3 for the ABCL (LOCF method)

Pre (T1) Post (T2) 6-month (T3) Main Effect

M (SD) M (SD) M (SD) F(2,38) p Partial WW

ABCL Total 55.80 (8.14) 52.50 (9.50) 53.15 (8.86) 1.66 .20 ns

Aggression 56.95 (6.83) 57.25 (8.01) 57.60 (8.80) .10 .91 ns

Rule Breaking 64.25 (10.82) 61.75 (6.57) 61.60 (8.64) 1.03 .39 ns

Internalizing 55.15 (10.82) 51.25 (8.68) 51.80 (10.07) 1.66 .20 ns

Externalizing 58.85 (7.19) 57.38 (10.29) 57.35 (11.56) .39 .40 ns

Note. T-scores have a mean of 50 and a standard deviation of 10