Preface

This report on the patterns and pathways to antisocial and criminal behaviours among Australian adolescents represents the first publication from the collaborative project between the Australian Institute of Family Studies and Crime Prevention Victoria.

The setting for the project is the longitudinal community study, the Australian Temperament Project (now in its twentieth year), which is itself a collaboration between researchers from the Institute, the Royal Children’s Hospital, and the University of Melbourne.

The study involves a representative sample of more than 2400 children and families living in urban and rural areas of Victoria. With its focus on children’s psychosocial development from infancy to adolescence, the study provides a rare and valuable opportunity to explore the development of teenage antisocial behaviour in an Australian context.

The origins of many problems in adolescence and adulthood can be traced back to early childhood. This report makes a substantial contribution to our understanding of how and why antisocial behaviours develop in childhood and adolescence, and identifies opportunities for assisting vulnerable youngsters to move onto more positive pathways. In doing so, it adds to the evidence base for policy and practice regarding Australian children and their families.

I commend Patterns and Precursors of Adolescent Antisocial Behaviour and am confident it will be of interest and value to the research community, to policy makers, and to parents, teachers and professionals who work with children and families. In particular, it is hoped that the report, in addressing current policy concerns, will facilitate government and community efforts to ensure the very best outcomes for all our children and their families.

David I. Stanton
Director
Australian Institute of Family Studies 

Executive summary

Adolescence is a crucial time for the emergence of antisocial and criminal behaviour which, for some, persists into adulthood; at considerable cost to individuals, families and the wider community. Much research has been devoted to the identification of risk factors associated with the occurrence of criminal and antisocial behaviour, with the aim of preventing such problems. However, much of the research has been cross-sectional or covered restricted age spans, conducted in other countries, employed disadvantaged samples, and focused on males. Its applicability to the Victorian and Australian context is uncertain.

Few Australian studies have examined the precursors of and pathways to antisocial behaviour from the earliest years of life. The present study, a collaborative project between Crime Prevention Victoria (Victorian Department of Justice), and the Australian Institute of Family Studies, uses data from the Australian Temperament Project to describe patterns and precursors of antisocial behaviour among a representative community sample of Victorian adolescents.

In this report a considerable amount of statistical data is presented, so that those who are interested can examine them in detail. However, at key points throughout the report summaries of the results are provided so that those who wish can bypass the statistical details.

Australian Temperament Project

The Australian Temperament Project (ATP) is a large scale, longitudinal study that has, to date, followed Victorian children from infancy to 17-18 years of age. The initial sample comprised 2443 infants (aged 4-8 months) and their parents, who were representative of the Victorian population at that time (1983). In total, twelve waves of data have been collected, via annual or biennial mail surveys. Using age-appropriate measures, data have been collected on aspects such as the child’s temperament, behavioural and emotional adjustment, academic progress, health, social skills, peer and family relationships, as well as family functioning, parenting practices and family socio- demographic background. Parents, teachers and the children themselves have acted as informants at various stages during the project. During the last three data collection waves in 1996, 1998 and 2000, when participants were aged 13-14, 15-16 and 17-18 years, adolescents answered questions regarding their engagement in antisocial acts.

Frequency of antisocial acts across the adolescent years

Antisocial behaviour was quite common among ATP participants over the period 13-14 to 17-18 years. One of the most common types of antisocial behaviour was property offences, with approximately 10-20 per cent of participants engaging in acts such as theft or vandalism. Cigarette and alcohol use were also common (39 per cent and 85 per cent respectively, at 17-18 years); however, fewer participants had used marijuana (increasing from 6 per cent at 13-14 years to 19 per cent at 17-18 years) and very few (less than 4 per cent) had used “hard drugs”. Authority conflict and violent antisocial acts were much less common, with the exceptions of skipping school (a high of 43 per cent at 17-18 years) and involvement in physical fights (a high of 34 per cent at 13-14 years). About one in ten participants had been in contact with the police for offending, but only a very small number had been charged (2-3 per cent), appeared in court (about 1 per cent), or been convicted of a crime (less than 1 per cent).

Frequency of antisocial acts among males and females

Cigarette use, alcohol use, and skipping school were the most common antisocial behaviours for both males and females. A higher proportion of males than females had engaged in violent and drug-related antisocial acts such as physical fighting (for example, 52 per cent of males at 13-14 years compared with 15 per cent females); been suspended/expelled from school (ranging from 6 to 9 per cent males compared with 2 to 4 per cent of females); committed property offences such as driving a car without permission (5-19 per cent males; 2-11 per cent females) and damaging property (19-32 per cent males; 8-11 per cent females); and been in contact with the criminal justice system (for example,19 per cent males and 6-8 per cent females had been in contact with the police for offending). Females, on the other hand, were more likely than males to have engaged in graffiti during early adolescence (11 per cent females compared with 7 per cent males at 13-14 years).

Patterns of antisocial behaviour over time

Different patterns of antisocial behaviour were identified among participants over 13-14, 15- 16 and 17-18 years of age, leading to the formation of three groups. These were: 844 “Low/non antisocial” (those who exhibited no or low levels of antisocial behaviour at all timepoints); 88 “Experimental” (those who exhibited high antisocial behaviour – three or more different antisocial acts in the past year – at only one timepoint during early-to-mid adolescence); and 131 “Persistent” (those who reported high antisocial behaviour – three or more different antisocial acts in the past year – at two or more timepoints, including the last data collection wave at 17-18 years). A further 103 were not included, as they did not fit the criteria for the three groups.

Predictors of antisocial behaviour across time

No significant differences were found between the two antisocial groups and the low/non antisocial group during infancy and early childhood. The first group differences emerged at the beginning of primary school (5-6 years). Clear and consistent differences between the persistent and low/non antisocial groups were observed from this time on. During mid childhood, the persistent antisocial group had higher levels of acting out, aggressive and hyperactive behaviour problems, and were more inclined to display volatility and to experience difficulties in maintaining attention than the low/non antisocial group. In late childhood, the persistent antisocial group continued to display problematic behaviour, and in addition were less cooperative, had poorer self-control, had poorer relationships with parents, and were more likely to have friends who engaged in antisocial behaviour.

The experimental and low/non antisocial groups did not differ significantly until early adolescence. During adolescence, the experimental group resembled the persistent group on many domains, although generally was less dysfunctional. The two antisocial groups were significantly more problematic than the low/non antisocial group on a wide range of domains, including school progress, attraction to risk taking, coping styles, parent-child relationships, and parenting style. Towards the end of adolescence, this pattern of differences appeared to change, with the experimental group becoming more similar to the low/non antisocial group.

Predictors of antisocial behaviour across domains of functioning

Group differences typically centred on temperamental characteristics such as negativity, volatility and low persistence, as well as aggressive, acting out and hyperactive behaviour problems, to the disadvantage of the antisocial groups. Powerful group differences were also observed in the domains of social competence, association with antisocial peers, school adjustment during adolescence, coping styles and involvement in risk-taking activities. Less powerful but significant group differences were also observed in family structural characteristics, parenting practices, and family relationships.

Gender differences in the predictors of antisocial behaviour

The effects of gender on the prediction of antisocial behaviour were investigated by (1) controlling for the effects of gender, and (2) conducting separate analyses for males and females. These analyses revealed a similar pattern of results to those described above, however, group differences during the early to mid primary school years were generally fewer, when sex-specific analyses were conducted.

Group differences over time

Key findings

1. Some degree of antisocial behaviour is “normal” in adolescence

Consistent with previous research, the findings of this study suggest that some degree of antisocial behaviour is common among adolescents. However, there are distinct patterns both in the timing, the frequency, and the nature of the antisocial behaviours, which need to be taken into consideration by prevention strategies.

2. Early interventions to divert children from pathways to persistent antisocial behaviour are most appropriate during the primary school years for the majority of young people

The current findings suggest that parents, teachers, clinicians and policy makers should focus on the early primary school years as a critical time for intervention in attempting to prevent the development of persistent antisocial behaviour. In this study, group differences first emerged at the age of 5 to 6 years (the commencement of primary school for most participants), suggesting that this period represents an early point in developmental pathways for the majority of children.

It is widely recognised that interventions during the earliest years of life are critical for the prevention of numerous emotional and behavioural problems (for example, hyperactivity, attention-regulation problems). Hence, more broad-based interventions (for example, home visiting programs), during infancy and early childhood, which aim to prevent the development of problems before they emerge, may also prove beneficial. Infants and young children whose sociodemographic and familial characteristics place them at increased risk of later developing antisocial behaviour would particularly benefit from such preventative efforts. Nevertheless, the current results suggest that when targeting the pathways to persistent antisocial behaviour, the focus should be on the early primary school years as a crucial period to intervene.

3. Persistent antisocial youth exhibit a clear profile

Individuals who went on to engage in persistent antisocial behaviour during adolescence were consistently reported to be more aggressive, more disinhibited, and more temperamentally reactive from mid-childhood onwards than individuals who later engaged in little or no antisocial behaviour. Furthermore, from late childhood, this group exhibited lower social competence, and associated more frequently with antisocial peers. Given the consistency of these findings, it may be possible to identify children who are at risk of developing persistent antisocial behaviour at quite a young age, for whom targeted interventions may be beneficial.

4. Interventions targeting experimental antisocial behaviour need to be multi-faceted and focus on the early secondary school years

Individuals who engaged in transitory antisocial behaviour during mid adolescence had shown clear signs of dysfunction from the early adolescent years, following the transition to secondary school. While they showed no signs of adjustment difficulties and were similar to the low/non antisocial group during childhood, in the early adolescent years they became more “difficult” temperamentally, more aggressive, began to experience difficulties at home and at school, and were likely to have formed friendships with youth who also engaged in antisocial behaviour. Due to the wide range of difficulties exhibited by individuals displaying experimental antisocial behaviour, interventions aimed at preventing this type of behaviour should be multi- faceted and targeted at the early secondary school years. It will be important to follow the trajectory of this group into young adulthood, to ascertain if their problems were truly transitory.

5. Precursors of antisocial behaviour are similar for males and females

When differences between antisocial groups were examined separately for males and females, differences generally emerged at the same times and in the same domains for both sexes. These findings suggest that interventions aimed at preventing the development of antisocial behaviour may be used equally well with males and females.

6. Peer relationships and their influence

The existence of friendships with other antisocial youth was one of the most powerful risk factors for both persistent and experimental antisocial behaviour identified by this study. Such friendships were evident from as early as 11-12 years of age, and prior to the onset of antisocial behaviour. Other aspects of peer relationships also appeared important. The low/non antisocial group members were more attached to their peers (had greater trust and communication), and more frequently interacted with peers in a structured setting (for example, while playing sport). The two antisocial groups, on the other hand, appeared to spend more time with peers, but their time together was more likely to be unstructured.

7. The role of family environment

There were few significant differences between the three groups on socio-demographic characteristics such as family socioeconomic status, parental education, occupational, and ethnic background, and number of children in the family. However, within-family processes, (for example, the parent-child relationship, the degree of warmth and conflict in this relationship, alienation from parents, family cohesion, and marital conflict and breakdown) were important contributors to group differences. Parenting style was also important, with parents of antisocial youth more prone to use lower supervision, less warmth and more harsh discipline. In general, family environment factors were less powerful in impact than individual child characteristics.

8. The importance of school adjustment

Clear group differences in school adjustment and school bonding were evident during the secondary school years. Both the persistent and experimental groups were observed to have more difficulties adjusting to school, and to exhibit lower levels of attachment to school, than those in the low/non antisocial group. These findings suggest that the manner in which an individual adapts to the school environment, the way in which the school accommodates the child’s individual characteristics and needs, and adolescents’ attitudes about schooling, are important predictors of adolescent antisocial behaviour.

Summary

In summary, this First Report has documented substantial group differences between adolescents who engage in high levels of antisocial behaviour and those who do not, which are evident from the early primary school years on, and increase in strength and diversity over time. The most powerful group differences emerge in intra-individual characteristics such as temperament, behaviour problems, social skills, levels of risk-taking behaviour and coping skills, and in the domains of school adjustment and peer relationships. Significant group differences in aspects of the family environment were also found. These findings have important implications for the content and timing of interventions aimed at preventing the development of antisocial behaviour.

A later report will include an examination of differences between adolescents who engage in violent antisocial acts versus those who engage in non-violent antisocial acts, an investigation of factors which may have a protective effect against the development of adolescent antisocial behaviour, and an investigation of location effects on adolescent antisocial behaviour.

1. Introduction

This report is the product of the collaborative partnership between the Australian Institute of Family Studies and Crime Prevention Victoria. The partnership began in 2001 when Crime Prevention Victoria commissioned the Institute to analyse and collect data from the Australian Temperament Project concerning the development of adolescent and young adult antisocial and criminal behaviour. This report provides valuable new information, which it is hoped will improve our understanding of the factors that place an individual at risk of engaging in this type of behaviour and, in turn, inform early intervention efforts aimed at diverting individuals away from problematic pathways.

Adolescence is a critical period for the emergence and entrenchment of antisocial behaviours (including criminal behaviours), which for some, persist into adulthood and entail substantial costs for individuals, families and the community. It is widely recognised that early intervention and prevention can curtail the development of these problems and is preferable to reacting after the problem behaviour has become established. Greater success in universal and targeted interventions is predicated on improved understanding of the genesis of such behaviour. The Australian Temperament Project, a large scale longitudinal study, with twelve waves of data spanning the first eighteen years of life, provides a valuable opportunity to investigate the precursors of antisocial behaviour among Australian adolescents, and to examine a broad range of risk factors across developmental stages and domains of functioning.

The focus of this first report is on adolescent antisocial behaviour, which includes criminal acts such as theft or the selling of drugs, and dysfunctional behaviors¹ such as running away from home or physical aggression. A brief overview of the research into adolescent antisocial behaviour is first presented, followed by a summary of the findings emerging from the current study and a discussion of their implications for the understanding of antisocial behaviour, as well as for policy development and early intervention and prevention efforts. For those interested in more a more detailed description of the findings, the appendices for this report can be obtained electronically from Crime Prevention Victoria's website, www.crimeprevention.vic.gov.au.

Nature and extent of adolescent antisocial behaviour

Information regarding the frequency and nature of antisocial behaviour among young people is typically obtained from a number of sources: (1) official statistics obtained from the criminal justice agencies (that is, police and courts), or (2) self-reported behaviour, generally obtained during the course of interviews or surveys (Rutter, Giller and Hagell 1998). Both types of information have advantages and disadvantages. Official statistics provide a measure of behaviours reported to and recorded by police. However, they provide a conservative assessment, since a high proportion of those committing antisocial acts are not apprehended, and many minor antisocial behaviours may not attract or warrant attention by authorities. Furthermore, particular groups, such as those from disadvantaged families and neighbourhoods, may be more likely to be the focus of official attention and hence have a greater likelihood of being apprehended (Rutter et al. 1998). Thus, official records provide an incomplete picture of the incidence of antisocial behaviours across different sections of the community.

Self-report has the potential to provide a more comprehensive picture and can cover a wider array of antisocial acts (not just those that are illegal), but may be affected by social desirability and other biases. It relies on the willingness of individuals to reveal potentially compromising information, and on respondents' veracity and memory. It is also reliant on the representativeness of the sample used, and researchers' ability to reach and engage the young people involved in serious antisocial acts. While recognising the advantages and disadvantages of both approaches, the current report focuses on adolescents' self-reported antisocial behaviour.

Antisocial behaviour among Australian adolescents

Studies examining rates of antisocial behaviour among Australian adolescents have found that it is very common for them to engage in some level of antisocial behaviour.

For example, in 1996, 441,234 New South Wales secondary school students in Years 7 to 12 were surveyed about their involvement in antisocial activities (Baker 1998). Close to 40 per cent of all students admitted to having attacked someone with the idea of hurting them at some time in their life, 38.6 per cent reported having purposely damaged or destroyed someone else's property, and over a fifth (22.8 per cent) had received or sold stolen goods. Significant age and gender differences were found, with rates for all types of offences peaking around Years 9 and 10 (14-16 years), and males reporting higher rates of each offence type, in each year level, almost without exception.

Further evidence of the high frequency of antisocial behaviour among Australian adolescents can be found in a Victorian survey of 8,984 Year 7, 9 and 11 students (Bond, Thomas, Toumbourou, Patton and Catalano 2000). Rates of antisocial behaviour generally increased from Years 7 to 9, but were relatively stable from Year 9 to Year 11. Consistent with Baker's (1998) findings, the peak incidence for most offences was in Year 9. The most common antisocial behaviours Year 9 students reported having been involved in during the past year were: stealing from a shop (30 per cent), engaging in graffiti (23 per cent), participating in a fight or a riot (18 per cent), carrying a weapon (18 per cent), and handling stolen property (18 per cent).

While these two studies used different time frames to assess engagement in antisocial behaviour (lifetime vs within the past year), some consistent trends are evident, particularly the escalation and peaking of antisocial behaviour in the mid teens. These trends are also reflected in official statistics.

Official statistics relating to antisocial behaviour among young people in Victoria, Australia.

• 16.5% of all persons proceeded against by Victoria Police in 2000/01 were aged between 10-16 years, compared with 38.2% who were aged 17-24 years, and 45.3% who were aged 25 years or over.
• 10-24 year-olds comprised approximately one-third of persons proceeded against for homicide and rape, half of all persons proceeded against for assault, and almost 60% of persons proceeded against for property offences.
• 15-19 years or 20-24 years are the peak ages of offending for homicide, assault, fraud, theft, burglary, vandalism and drug offences.
• 4-5 times as many young males are proceeded against by Victoria Police than young females.

Source: Victoria Police, Statistical Services Branch (2002).

Comparisons with international data

Studies that have examined rates of self-reported antisocial behaviour among adolescents in other countries have found similar patterns to those identified in Australia. That is, some degree of antisocial behaviour appears to be quite common among adolescents (Baker 1998; Rutter et al. 1998), to be more frequent among males than females, and typically peaks during midto- late adolescence. Nevertheless, there is considerable variation in rates across countries (Rutter et al. 1998). However, it should be noted that methodological differences between studies (for example, the representativeness of the sample employed, the measures of antisocial behaviour used, the time period during which participants were surveyed) make it difficult to directly compare rates of self-reported behaviour in different countries.

In summary, antisocial behaviour is common in adolescence. It ranges from relatively minor to quite serious acts, typically peaks during mid-to-late adolescence, and is more common among males than females. It has potentially serious consequences for adolescents both in the present and the future (Moffitt, Caspi, Harrington and Milne 2002), and impacts on their families and wider society (Homel et al. 1999).

Patterns of antisocial behaviour

In addition to investigating the nature and extent of antisocial behaviour in the community, considerable research has focused on differentiating between young people who exhibit distinct patterns of antisocial behaviour. Two such patterns, violent versus non-violent antisocial behaviour, and persistent versus experimental antisocial behaviour, will now be discussed.

Violent versus non-violent

Considerable research supports the notion that violent offenders are a small but distinct group from those who engage in nonviolent antisocial or criminal behaviour (Farrington and Loeber 2000; Loeber, Farrington, Rumsey, Kerr, Allen-Hagan 1998; Maughan, Pickles, Rowe, Costello and Angold 2000; Nagin and Tremblay 1999).

A comprehensive literature review undertaken by the United States Office of Juvenile Justice and Delinquency Prevention's Study Group on Serious and Violent Juvenile Offenders (Loeber et al. 1998) revealed a number of key differences between violent and non-violent offenders. These included the findings that: violent offenders are typically male; the majority of violent offenders tend to start offending earlier, and continue offending longer, than non-violent offenders; violent offenders tend to exhibit multiple problem behaviours (for example, substance use, mental health difficulties, authority conflict problems, aggression etc); and violent offenders tend to commit a range of aggressive and non-aggressive offences.

Further support for a differentiation between violent and non-violent offenders can be found in research that has attempted to chart developmental pathways to antisocial behaviour. For example, Maughan and colleagues (2000) examined the development of aggressive and non-aggressive conduct problems in a sample of 1419 American boys and girls. These authors found only a small degree of overlap between the developmental pathways for the aggressive and non-aggressive children. Similarly, a Canadian study of 1,037 males (Nagin and Tremblay 1998) identified unique developmental pathways for those who engaged in overt delinquency (for example, physical violence) and those who engaged in covert delinquency (for example, theft) during adolescence.

Nevertheless, some have argued against this distinction. Piquero (2000), for example, claims that the difference between violent and non-violent offenders is quantitative not qualitative. Piquero (2000) notes the consistent finding that violent offenders tend to commit more offences than non-violent offenders. Based on this observation, he suggests that the difference between these groups is more a matter of degree than type, in which case the correlates of one type of offence should be the same as another. Piquero (2000) tested this hypothesis on data from a sample of 987 American adolescents. After controlling for frequency of offending he found that only one variable differentially predicted violent, but not non-violent offending, namely, variation in intelligence test scores. Individuals with low intelligence scores were more likely to come into police contact for a violent offence by age 18 than those who scored highly on this measure.

While researchers should remain open to the idea that the difference between violent and non-violent offenders may be one of degree, differentiation into these subtypes appears to be a useful strategy for investigating the developmental pathways to adolescent antisocial behaviour.

Experimental versus persistent

Another distinction frequently made in the research literature relates to the stability or transient nature of antisocial behaviour.

Childhood and adolescence are periods of high experimentation, during which many young people engage in behaviours that are not pro-social (for example, shoplifting, lying, bullying, annoying peers etc) (Kelley, Loeber, Keenan, DeLamatre 1997). Nevertheless, while many young people act in an antisocial manner, this behaviour is usually transitory (Dussuyer and Mammalito 1998; Kelley et al. 1997; Moffitt and Harrington 1996). Individuals who engage in antisocial behaviour for a relatively short period of time and then desist, are often referred to as "experimenters". On the other hand, for a small group of people, antisocial behaviour is much more stable (Kelley et al. 1997; Moffitt and Harrington 1996), often beginning at a very early age and continuing well into adulthood. Those who maintain high levels of antisocial behaviour over long time periods are often labelled "persisters".

Consistent with the experimental – persistent distinction, Moffitt and colleagues (Moffitt and Harrington 1996; Moffitt, Caspi, Rutter and Silva 2001) propose two broad categories of antisocial behaviour: "life-course persistent" antisocial behaviour (which emerges early in life and persists well into adulthood); and "adolescent-limited" antisocial behaviour (which emerges alongside puberty and is transitory). According to these authors, adolescent-limited antisocial behaviour is quite common and may have few long-term deleterious consequences, whereas relatively few young people engage in life-course persistent antisocial behaviour.

Moffitt and colleagues (2001) tested this taxonomy in a sample of 922 males and females from the Dunedin Multidisciplinary Health and Development Study. They found that 200 participants fulfilled the criteria for adolescent-limited antisocial behaviour, whereas 53 met the criteria for life-course persistent antisocial behaviour. Life-course persistent antisocial behaviour was considerably more common among males than females, with approximately 10 males to every female displaying this pattern of antisocial behaviour, whereas the gender difference in adolescent-limited antisocial behaviour was small (1.5:1 males to females).

This distinction appears to be a critical one. However, relatively few studies have the requisite longitudinal data to allow the differentiation of such groups. The Australian Temperament Project, with data available at multiple timepoints from infancy onwards, has the capacity to investigate this important issue.

Theoretical models

Many models have been proposed to explain the development of antisocial behaviour. Some models propose different pathways leading to the development of antisocial behaviour. For example, Loeber and colleagues (Loeber, Wung, Keenan, Giroux, Stouthamer-Loeber, Van Kammen, and Maughan 1993) suggest that three different pathways can explain the development of antisocial behaviour in males. The first of these, the overt pathway, involves an escalation in aggressive acts (for example, minor aggression --> physical fighting --> physical violence) over time; the second, the covert pathway, involves an escalation in less overt antisocial acts (for example, minor covert behaviours --> property damage --> moderate to severe delinquency); while the third pathway, the authority conflict pathway, involves a sequence of stubborn behaviour, leading to defiance, and ultimately authority avoidance (for example, running away from home, truancy). Less serious behaviours precede more serious behaviours in these pathways and boys may proceed along more than one pathway at a time.

The Social Development model of Catalano and Hawkins (1996) emphasises the role of social learning in the development of antisocial behaviour. According to this model, children learn patterns of behaviour, whether they are prosocial or antisocial, from their family, their school, religious and other community institutions, and their peers. Hence, an individual's behaviour is determined by the predominant behaviours, norms and values held by those to whom the individual is attached. Consequently, youth attachment to prosocial individuals, developed particularly through involvement in rewarding experiences, is posited to be protective against the development of antisocial behaviours, conduct problems and substance use.

Weatherburn and Lind (2001) propose a role for economic stress in the development of criminal behaviour. According to their model, parents who experience higher levels of economic stress are more likely to neglect or abuse their children or engage in harsh, erratic and inconsistent disciplinary practices than other parents. This kind of parenting behaviour may lead a child to affiliate more strongly with their peers than their parents, making the child susceptible to the negative influence of antisocial peers. The effects of economic stress are reduced when parents have a strong social support network, but increase if such a support network is absent, or other sources of stress are present (for example, crowded household, large family, "difficult child", family conflict, parental disorder). Patterson, Reid and Dishion (1992) also suggest a critical role for parenting in the development of antisocial behaviour among males. They suggest that individuals who experience poor "basic training" as children are more susceptible to poor academic performance and peer rejection later on. These problems may lead to association with antisocial peers and engagement in antisocial acts in adolescence, and eventually, to poor adjustment in adulthood.

It should be noted that many of these "pathways" models have been developed to explain antisocial behaviour among males, with relatively little attention to antisocial behaviour among females.

Another approach to understanding the development of antisocial behaviour is the Risk Factors approach. A large body of research has been dedicated to the identification of risk and protective factors associated with the development of antisocial behaviour. Risk factors can be defined as those factors that "increase(s) the likelihood that a subsequent negative outcome will occur" (Loeber, 1990: 4), whereas protective factors operate in the context of risk and "offset risk factors and promote social development, well-being and resilience" (Bond et al. 2000: 3).

Risk and protective factors associated with the development of antisocial and criminal behaviour can occur across a number of domains. These include the characteristics of the child, the family and its experience of stressful life events, the school context, and community and cultural factors (Homel et al. 1999). The risk and protective factors that have most frequently been found to be associated with the development of antisocial and criminal behaviour are summarised in Tables 2 and 3, respectively.

Research suggests that no single risk factor can explain the development of antisocial behaviour. Rather, the more risk factors an individual is exposed to, the greater the likelihood that he or she will exhibit antisocial or criminal behaviour (Bond et al. 2000; Loeber and Farrington 2000). Similarly, the greater the number of protective factors possessed by a young person, the more likely he or she is to display resilience despite the presence of risk (Howard and Johnson 2000). Hence, the risk of a child becoming antisocial appears to be dependent upon the balance of risk and protective factors in their lives (Loeber and Farrington 2000).

Source: Homel, Cashmore, Gilmore, Goodnow, Hayes, Lawrence, Leech, O'Connor, Vinson, Najman, & Western, 1999, p136. Reproduced with the kind permission of the Commonwealth Attorney-General's Department.

This approach emphasises the accumulation of risks as critical and treats the various risks factors as of equivalent importance. The question of whether different types of risks, or clusters of risk factors, have differential impacts remains as yet unanswered.

These are only some of the models that have been proposed to explain the development of antisocial and criminal behaviour. The current report does not try to provide a comprehensive review of current theoretical models. Rather, here we focus on one of the most widely used theoretical approaches in this field, the Risk Factor approach.

The present study

It is notable that much of the research into adolescent antisocial behaviour has been conducted in the United States, with influential work also originating in the United Kingdom, Europe, Canada and New Zealand. In addition, much of the research is based on samples suffering social and economic disadvantage and focuses primarily on males. Hence, the applicability of such research to the Australian context, to individuals in the general population, and to females, is uncertain.

There are very few Australian studies which have examined the pathways to antisocial behaviour from the early years onwards (an exception being Bor, Najman, O'Callaghan,Williams and Anstey 2001) although valuable research investigating more proximal influences has been conducted (for example, Baker 1998; Bond et al 2000; Homel et al. 1999; Weatherburn and Lind 2001 etc). Thus, our understanding of the precursors of and pathways to antisocial behaviour among Australian adolescents is very limited at present.

The present study attempts to redress this by analysing data on a sample of males and females representative of the general community who have been followed from infancy (4-8 months of age) into young adulthood. The Australian Temperament Project data set provides a valuable opportunity to investigate a wide range of risk factors for antisocial behaviour. The findings will provide guidance for early intervention and crime prevention efforts.

Source: Homel, Cashmore, Gilmore, Goodnow, Hayes, Lawrence, Leech, O’Connor, Vinson, Najman, & Western, 1999, p138. Reproduced with the kind permission of the Commonwealth Attorney-General’s Department.

2. The Australian Temperament Project

The Australian Temperament Project (ATP) is a longitudinal study of the psychosocial development of a large, representative sample of children born in Victoria between September 1982 and January 1983 (see Prior, Sanson, Smart and Oberklaid 2000 for a fuller account). Twelve waves of data on the children's temperament, behavioural and emotional adjustment, academic progress, health, social skills, peer and family relationships have been collected via annual or biennial mail surveys, with the aim of tracing the pathways to psychosocial adjustment and maladjustment across the children's lifespan.

A cohort of 2443 families from urban and rural areas was initially recruited via the following process. A stratified sample of Local Government Authority (LGA) areas, drawn to be representative of the state population, was developed using census data provided by the Australian Bureau of Statistics. All families with an infant of 4 to 8 months of age who attended a Maternal and Child Health Centre in one these LGAs during the first two weeks in May 1983 were invited to take part in the project. Comparison of the demographic profile of the cohort to census data showed that the obtained cohort was representative of the state's population. Approximately two-thirds of the original cohort continue to participate in the study after 18 years (N = 1650). Although there is an over-representation of low SES and ethnic families among those no longer participating, there are no significant differences between the retained and lost/withdrawn samples on any child characteristic assessed at infancy, and the family socio-economic profiles of the original and retained sample are very similar (Prior, Sanson, Smart and Oberklaid 1999). At each survey wave, the response rates have been approximately 80 per cent from those participating at that particular timepoint.

Parents have completed questionnaires about the child's functioning and aspects of their family life at every timepoint. Primary school teachers have supplied information about the ATP child in their class at the Preparatory Grade, Grade 2 and Grade 6 stages via mail surveys assessing a range of academic and individual child characteristics. From the age of 11-12 years (Grade 6), the children have completed questionnaires about their personal functioning, relationships with others, and beliefs and attitudes. Table 4 (pp.6-11) summarises the major domains assessed at each data wave. For most domains and at most survey waves, data are available from multiple sources.

In the three latest waves of data, at ages 13-14 years, 15-16 years and 17-18 years (that is, the years 1996, 1998, and 2000), adolescents answered questions about the incidence of antisocial behaviours, via an adaptation of the Self Report Delinquency Scale (Moffitt and Silva 1988).

In the sections that follow, we report the results emerging from the study and discuss their theoretical, preventative and early intervention implications.We present descriptive data on the frequency of antisocial behaviour during the adolescent years for the total sample, and for males and females separately. Following this, we report the identification and comparison of groups of persistently antisocial, transiently antisocial and non-antisocial youth. Finally, we investigate the relative impact of different types of risk factors and report analyses investigating gender differences.

3. Frequency of antisocial behaviour

Adolescent participants were asked about their engagement in antisocial behaviours at three timepoints: 13-14 years (1996), 15-16 years (1998), and 17-18 years (2000). A summary of the questions used to assess antisocial behaviour is presented below. All questions relate to participant's behaviour within the past twelve months, with the exception of those concerning substance use, which refer to the past month.

As antisocial behaviour may be expressed in different ways at different ages (Moffitt et al. 2001), items were added to the scale at each timepoint, to accommodate age-related changes in the manner in which antisocial behaviour may be exhibited.

Australian Temperament Project assessment of antisocial behaviour

At 13-14 (1996), 15-16 (1998) and 17-18 (2000) years

• Got into physical fights with other people
• Damaged something in a public place ("on purpose" added from 15-16 years)
• Stolen something ("from a person or a house" added at 17-18 years)
• Driven a car without permission
• Been suspended or expelled from school
• Done graffiti in public places
• Carried a weapon ("e.g. gun, knife" added at 15-16 years)
• "Wagged" (skipped) school
• Frequency of cigarette use*
• Frequency of alcohol use*
• Frequency of marijuana use*
• Frequency of sniffing*
• Frequency of other drug use*

* These items relate to the past month.

Additional questions at 15-16 (1998) and 17-18 (2000) years

• Shoplifted
• Run away from home and stayed away overnight or longer
• Been in contact with, or cautioned by, police for offending
• Been charged by police
• Appeared in court as an offender
• Frequency of binge drinking*
• Frequency of drunkenness*
• Deleterious consequences of drinking alcohol*

* These items relate to the past month.

Additional questions at 17-18 (2000) years

• Sold illegal drugs
• Attacked someone with the idea of seriously harming them
• Been convicted in court of a criminal offence
• Deleterious consequences of marijuana and other drug use, including trouble with police*
• Substance dependence (alcohol, marijuana and other drugs).*

* These items relate to the past month.

Frequency of antisocial acts across the adolescent years

The frequency with which ATP participants reported engaging in each antisocial act is presented in Table 6. Data are presented for each of the three data waves in which this information was collected (that is, 13-14, 15-16, and 17-18 years). For clearer presentation of findings, antisocial acts have been grouped into five categories:

1. those involving property offences,
2. authority conflict problems,
3. violent and drug-related antisocial acts,
4. criminal justice contacts (as a consequence of antisocial acts), and
5. substance use.

The trends emerging from this table will be discussed in detail later in this section.

Figures 1-5 pictorally display the proportion of individuals in the sample who reported engaging in the antisocial act on at least one occasion during the last 12 months.

Note: * These variables relate to frequency during the past month (not year).
-- = Not assessed at this timepoint
Damaged = Damaged something in a public place on purpose; Driven = Driven a car without permission; Graffiti = Done graffiti in public places; Shoplifted = Shoplifted; Theft = Stolen something (from a person or a house); Run Away = Run away from home and stayed away overnight or longer; Skipped School = Skipped School; Suspend/Expel = Been suspended or expelled from school; Attacked = Attacked someone with the idea of seriously harming them; Fights = Got into physical fights with other people; Sold Drugs = Sold illegal drugs; Weapon = Carried a weapon (e.g. gun, knife); Charged = Been charged by police; Contact = Been in contact with, or cautioned by, police for offending; Court = Appeared in court as an offender; Convicted = Been convicted in court of a criminal offence; Alcohol = drank alcohol in last month; Cigarettes = Smoked cigarettes in last month; Marijuana = Used marijuana in last month; Other Drugs = Used other illegal drugs in last month.

Property offences

As shown in Figure 1, involvement in property offences, in particular, damaging public property and theft, was relatively common among ATP participants. Engagement in most property crimes appears to peak around 15-16 years of age in the ATP sample. The exceptions to this were driving a car without permission, which increased as participants approached the Victorian legal driving age (18 years), and property damage, which remained relatively stable between 15-16 and 17-18 years of age.

Figure 1: Property offences by age

Note: Graffiti = Graffiti drawing in public places; Shoplifted = Shoplifted; Driven = Driven a car without permission; Theft = Stolen something (from a person or a house); Damage = Damaged something in a public place on purpose.

Authority conflict problems

Skipping school was a common occurrence among ATP participants with over 40 per cent of 17-18 year olds admitting that they had skipped school at least once during the past year (See Figure 2). The other authority conflict problems were much less frequent, with fewer than seven percent of participants in each data collection wave reporting that they had run away from home, and/or been suspended or expelled from school.

Figure 2: Authority conflict problems by age

Note: Run Away = Run away from home and stayed away overnight or longer; Suspended/Expelled = Been suspended or expelled from school; Skipped School = Skipped School.

Violent and drug-related antisocial acts

As Figure 3 shows, involvement in physical fights was commonly reported. Approximately a third of the sample in 1996 (13- 14 years) and 1998 (15-16 years) reported that they had been in a physical fight with another person in the past year. However, by 2000 (17-18) the proportion of individuals engaging in physical fights had decreased quite markedly.

At 17-18 years of age, the proportion of individuals who reported having attacked someone with the intention of seriously harming them, having sold illegal drugs and having carried a weapon was very similar and quite low (between five and six percent). (Participants were not asked whether they had attacked someone or sold illegal drugs at ages 13-14 or 15-16). The proportion carrying a weapon was slightly higher at 15-16 years of age than at the other two ages.

Figure 3: Violent and drug-related antisocial acts by age

Note: Attacked = Attacked someone with the idea of seriously harming them; Sold Drugs = Sold illegal drugs; Weapon = Carried a weapon (e.g. gun, knife); Fights = Got into physical fights with other people.

Substance use

As mentioned earlier, all questions relating to substance use refer to participant's use within the past month.

As Figure 4 shows, the frequency with which participants engaged in all types of substance use increased with age. Consumption of alcohol was extremely common within the ATP sample, with almost 85 per cent of participants at 17-18 years of age reporting that they had consumed alcohol on at least one occasion during the past month. Cigarette smoking was also very common among this sample. Marijuana use was less frequent, peaking at just under 20 per cent by 17-18 years of age. The use of "hard" drugs such as amphetamines, designer drugs and opiates was very low.

Figure 4: Substance use by age

Note: Cigarettes = Smoked cigarettes in last month; Alcohol = drank alcohol in last month; Marijuana = Used marijuana in last month; Other drugs = Used other illegal drugs in last month.

Criminal justice contacts

While a considerable proportion of participants aged 15-16 and 17-18 years reported having been in contact with the police for offending; the proportion of participants who had been charged with an offence, appeared in court as an offender, or convicted of an offence, was very small. However, Figure 5 suggests a slight increase in contact with the criminal justice system with age.

Figure 5: Criminal justice contacts by age

Note: Convicted = Been convicted in court of a criminal offence; Court = Appeared in court as an offender; Charged = Been charged by police; Contact = Been in contact with, or cautioned by, police for offending.

Frequency of antisocial acts among male and female adolescents

The proportion of males and females who reported engaging in each type of antisocial act are presented in Figures 6 to 10, while a detailed description of significant group differences is contained in Appendix 1.² It should be noted that these graphs depict the relative proportion of males and females who reported engaging in each antisocial act at the separate timepoints.

Property offences

As shown in Figure 6, damaging public property (at all timepoints) and driving a car without permission (at 13-14 and 17-18 years) was noticeably more common among male ATP participants than female participants. On the other hand, a higher proportion of females than males reported engaging in graffiti in public places at 13-14 years of age. Rates of shoplifting and theft were fairly similar across males and females, except at 17-18 years, when a higher proportion of males admitted to having stolen something from a person or a house in the past year.

Figure 6: Property offences by gender

Note: Graffiti = Done graffiti in public places; Shoplift = Shoplifted; Driven = Driven a car without permission; Theft = Stolen something from a person or house; Damage = Damaged something in a public place (on purpose).

Authority conflict problems

Similar proportions of males and females reported skipping school and running away from home at each timepoint (see Figure 7). However, males were more likely to report having been suspended or expelled from school at 13-14 and 15-16 years.

Figure 7: Authority conflict problems by gender

Note: Run Away = Run away from home and stayed away overnight or longer; Suspended/Expelled = Suspended/expelled from school; Skipped School = skipped school.

Violent and drug-related antisocial acts

As Figure 8 shows, without exception, a higher proportion of males than females reported having sold illegal drugs, carried a weapon and been involved in a physical fight with others, at each timepoint. The proportion of males and females who admitted to having attacked someone with the intention of seriously harming them was fairly similar.

Figure 8: Violent and drug-related antisocial acts by gender

Note: Attacked = Attacked someone with the idea of seriously harming them; Sold Drugs = Sold illegal drugs; Weapon = Carried a weapon (e.g. gun, knife); Fights = Got into physical fights with other people

Substance use

Figure 9 shows that a similar proportion of males and females reported having smoked cigarettes, consumed alcohol, used marijuana and/or used hard drugs at all timepoints.

Figure 9: Substance use by gender

Note: Cigarettes = Smoked cigarettes; Alcohol = Drunk alcohol; Marijuana = Used marijuana; Other Drugs = Used other illegal drugs.

Criminal justice contacts

With the exception of 15-16 years, when a small but similar proportion of males and females reported having appeared in court as an offender, much higher proportions of males than females reported having been in contact with the criminal justice system at all timepoints (see Figure 10). This gender difference was most notable when comparing the proportion of males and females who had been in contact with, or cautioned by, police for offending (approximately 19 per cent males compared with 6-8 per cent females).

Figure 10: Criminal justice contacts by gender

Note: Convicted = Been convicted in court of a criminal offence; Court = Appeared in court as an offender; Charged = Been charged by police; Contact = Been in contact with, or cautioned by, police for offending.

4. Formation of persistent, experimental and low/non antisocial groups

To assist with the identification of risk factors for the development of antisocial and criminal behaviour, participants were grouped on the basis of their pattern of antisocial behaviour over three data collection points (13-14 years, 15-16 years, and 17-18 years).

First, individuals were classified as displaying high or low levels of antisocial at each timepoint. An individual was classified as displaying high levels of antisocial behaviour at a particular timepoint if he or she reported engaging in three or more different antisocial acts during the previous 12 months,³ while those who reported fewer than three different antisocial acts within the previous 12 months were classified as displaying low levels of antisocial behaviour at that timepoint.

Some acts (i.e. alcohol use, cigarette use, and skipping school), while socially undesirable and/or officially illegal at these ages, were so common within the sample as to appear almost normative. These behaviours were therefore not included in the definition of antisocial behaviour. Contact with the criminal justice system was also excluded from our definition as contact with this system is not an antisocial act in itself, but usually results from an antisocial act being detected.

Consequently, to be classified as being high in antisocial behaviour, an individual must have engaged in three or more of the following behaviours on at least one occasion during the past 12 months:

• been in physical fights with others;
• damaged something in a public place on purpose;
• stolen something (from a person or a house);
• driven a car without permission;
• been suspended or expelled from school;
• engaged in graffiti in public places;
• carried a weapon (for example, gun, knife);
• shoplifted; run away from home or stayed away overnight or longer;
• sold illegal drugs;
• attacked someone with the idea of seriously harming them;
• used marijuana (within the past month);
• used hard drugs, such as amphetamines, cocaine, designer drugs or opiates (within the past month)

Using this criterion, 12.4 per cent of participants in at 13-14 years (1996), 19.7 per cent of participants at 15-16 years (1998) and 20.0 per cent of participants at 17-18 years (2000), exhibited high levels of antisocial behaviour.

Patterns of antisocial behaviour over time

We next examined the data for the three timepoints to identify patterns of behaviour across time. For some participants, data were available for only two timepoints. These data were included if the level of antisocial behaviour was consistent across both survey waves (for example, both high, both low). Individuals who were high at one timepoint and low at the other were not included, as the absence of information from the third timepoint made it difficult to determine whether behaviour patterns were transient or stable. A further condition for inclusion was that data must be present at 17-18 years, so that participants' current levels of antisocial behaviour could be ascertained. Eight patterns of antisocial behaviour were identified (see Table 7).

Using the Experimental-Persistent typology (Kelley et al. 1997; Moffitt et al. 2001; Moffitt and Harrington 1996) as a guide, the eight groups were combined into three larger groups based upon the stability of their antisocial behaviour:

• "Low/non antisocial" – these participants displayed low or no antisocial behaviour at all timepoints that data were available for them (Group 1 in Table 7).
• "Experimental" – these participants exhibited high antisocial behaviour at one timepoint only and appeared to have desisted (Groups 2 and 3 in Table 7).
• "Persistent" – these individuals reported high antisocial behaviour at two or more timepoints, including the latest timepoint (Groups 6, 7 and 8 in Table 7).

Individuals who displayed high antisocial behaviour at only 17-18 (Group 4 in Table 7) were not included in the experimental group as, in the absence of follow-up data, it could not be determined whether their behaviour was of a transitory nature. In addition, participants who exhibited high antisocial behaviour at ages to 13-14 and 15-16 years but not 17-18 (Group 5), were excluded from the persistent group because although they exhibited persistent antisocial behaviour over two waves of data collection, they had desisted by 17-18 years. Hence, they did not fit with the other groups (6, 7, and 8) who were still actively antisocial at the last data collection wave. The three groups formed on the basis of the criteria above are summarised in the right-hand column of Table 7.

Note: Low = Low antisocial behaviour; High = High antisocial behaviour

Group characteristics

Gender

The gender composition of the three groups is shown in Table 8. It can be seen that there were more males than females in the persistent group, and somewhat fewer males than females in the other two groups.⁴ The male:female ratio in the persistent group is consistent with the New Zealand trends reported by Moffitt et al. (2001), although it is less pronounced.

While not significant, the slightly higher proportion of females in the experimental group is at odds with Moffitt's findings of slightly more males than females in the adolescent-limited group. These differences could be a result of somewhat different group selection methods; sampling differences (the ATP study includes urban and rural participants while the New Zealand study is of urban participants; and the ATP has a larger sample); time of survey effects (the New Zealand data on antisocial behaviour was collected in the mid to late 1980s while the comparative ATP data was collected in the mid to late 1990s); cultural differences; or differential attrition effects (the New Zealand study has had lower rates of attrition than the ATP study).

Table 8 also shows that ATP males, if antisocial, were much more likely to be persistently antisocial, while females were equally likely to be persistent or experimental.

Profile of antisocial behaviours in the three groups

The three groups were compared on the criteria used to define antisocial behaviour, as shown in Table 9. It can be seen that a higher proportion of the persistent group engaged in all types of antisocial behaviour at all timepoints than adolescents in the low/non antisocial group.

Furthermore, significantly more experimental than low/non antisocial individuals reported using marijuana (at 13-14), running away from home (at 15-16) and engaging in property offences (for example, property damage, theft, graffiti at 13-14 and 15- 16; driving a car without permission, shoplifting at 15-16 years). There were no significant differences between the experimental and low/non antisocial groups at ages 17-18, which is a result of the method used to select groups⁵. A more detailed description of group differences is contained in Appendix 2.

Note: * These variables relate to frequency during the past month (not year).
-- = Not assessed at this timepoint
Fights = Got into physical fights with other people; Damage = Damaged something in a public place on purpose; Theft = Stolen something (from a person or a house); Driven = Driven a car without permission; Suspend/expel = Been suspended or expelled from school; Graffiti = Done graffiti in public places; Weapon = Carried a weapon (e.g. gun, knife); Shoplift = Shoplifted; Run away = Run away from home and stayed away overnight or longer; Sold drugs = Sold illegal drugs; Attacked = Attacked someone with the idea of seriously harming them; Marijuana = Used marijuana in last month; Other drugs = Used other illegal drugs in last month.

5. Precursors of adolescent antisocial behaviour

Having examined the nature and extent of antisocial behaviour in the ATP sample, we next investigated precursors of antisocial behaviour. In order to do this, separate Multivariate Analyses of Variance (MANOVA) were performed for each data collection wave and source of report (parent, teacher, or child/teen).⁶ The strategy of examining each wave of data separately enables the identification of the age and stage of development at which significant group differences begin to emerge. Furthermore, it maximises the number of participants included in the analyses, which is of importance to maintain power.⁷ Effect sizes⁸ were used to assess the strength of group differences across the various domains.

In the following section we present analyses investigating precursors of adolescent antisocial behaviour at different ages/stages of development; across different domains of functioning; and finally, among males and females.

Predictors of antisocial behaviours across age / stage of development

Figure 11 broadly illustrates the timing and number of differences between the low/no and two antisocial groups. To facilitate identification of the onset of pathways to persistent and experimental antisocial behaviour, only differences between the persistent and low/non antisocial groups, and the experimental and low/non antisocial groups are presented here. This comparison enables the identification of the age/stage of development at which each group begins to diverge from the 'normal' developmental pathway exhibited by the low/non antisocial group. These differences will be discussed in more detail in the following sections, as will differences between the experimental and persistent groups.

Figure 11: Group differences over time

Note: P = Persistent group, L= Low/non antisocial group, E = Experimental group

As Figure 11 shows, no significant differences were found between the two antisocial groups and low/non antisocial group in the earliest years of life (infancy to 3-4 years). The first group differences emerged between the persistent and low/non antisocial groups at 5-6 years. Over the primary school years, the number of variables on which the persistent group was significantly more problematic that the low/non antisocial group rose steadily from less than 10% at 5-6 years, around 25% at 9-10 years, to over 50% at 11-12 years. The experimental and low/non antisocial groups did not differ significantly until early adolescence. The peak age at which differences were most evident was 15-16 years, with the persistent group differing from the low/non antisocial group on two-thirds of variables, and the experimental group differing from the low/non antisocial group on approximately 40% variables. Towards the end of adolescence, differences between both antisocial groups and the low/non antisocial group decreased, however, this decrease was most marked for the experimental group.

Infancy to preschool (0-4 years)

The variables included in the MANOVAs performed for this developmental stage are summarised in Table 10. The majority were assessed by parent report. There were no significant multivariate differences between the three groups (low/non antisocial, experimental, persistent) during any of the pre-school years.⁹ .

The primary school years (5-6 through 11-12 years)

A number of group differences emerged during the primary school years. These are summarised in Table 11 (over page),¹⁰ while the detailed results are displayed in Appendix 3.

At 5-6 years of age, which was the first year of primary school for most of the children in the study, significant group differences were found between the persistent and the low/non antisocial group. However, these parent-reported differences were relatively weak in terms of effect size criteria.¹¹ At 7-8 years, more consistent and powerful group differences were found for teacherreported temperament and behaviour problem measures, with the persistent group displaying more problematic temperament styles (less able to maintain attention, more volatile) and higher levels of acting out, aggressive and hyperactive behaviour problems than the other two groups. The differences found were in the small effect size range. At 9-10 years there was a trend for multivariate group differences, but the strength of these differences did not reach the adjusted significance level.¹² At 11-12 years (Grade 6 for most children), parents, teachers and children all rated the persistent group as significantly more aggressive and hyperactive than the low/non antisocial group. The persistent group was also rated as being less cooperative, lower in self-control than the low/non antisocial group; and as more difficult; having less adaptive temperament styles (more active and volatile according to parents; less persistent and able to maintain attention according to parents and teachers). They were also more likely to have friends who engaged in antisocial behaviour (although such friendships were relatively rare at this age). Additionally, the persistent group was more depressed (by their own and teacher report); and had poorer relationships with parents and were less empathic and confident by their own report, than those in the low/non antisocial group. Differences were in the small effect size range.

* p<.05; ** p<.01; *** p<.001
There were no significant differences on :
Parent report of approach-sociability, task-persistence, rhythmicity, aggression, hyperactivity, quality of parent-child relationship and socioeconomic status at 5-6 years; approach-sociability, task persistence, rhythmicity, anxiety-fearfulness, hyperactivity, quality of parent-child relationship, parent's coping skills, parent's rating of life difficulty, family stress and socioeconomic status at 7-8 years; anxiety-fearfulness, negative emotionality, shyness, sociability, activity, empathy, quality of parent-child relationship and socioeconomic status at 9-10 years; anxiety-fearfulness; approach-sociability, assertiveness, depression, responsibility, peer relationships, mother's educational level, father's educational level, father's occupation, and number of children in family at 11-12 years;
Teacher report of all temperament, behaviour problem and school readiness measures at 5-6 years; flexibility, anxiety-fearfulness, academic skills, reading ability and social skills at 7-8 years; anxiety-fearfulness, academic skills and peer relationships at 11-12 years; and
Child report of anxiety-fearfulness, self concept concerning peer relationships and quality of friendships at 11-12 years.

Overall, these results reveal clear and consistent differences between the persistent and low/non antisocial groups, from the middle childhood years to the end of primary school. Several significant differences between the persistent and experimental groups were evident over the primary school years. The domains in which such differences were found were the temperament factors of task orientation /persistence at 7-8 and 11-2 years; aggression at 7-8 and 11-12 years; hyperactivity at 7-8 years; and activity, antisocial peer affiliations, relationships with parents, self confidence and cooperation at 11-12 years. On these, the persistent group showed less positive characteristics than the experimental group.

The experimental and low/non antisocial groups were not significantly different on any domain or at any timepoint over the primary school years, suggesting that they were very similar at this developmental stage.

The secondary school years

Comparisons of groups over the secondary school years revealed numerous differences. Tables 12(a), (b), (c), and (d) summarise these differences, while more detailed results are displayed in Appendix 4. The variables included in the analyses for which no significant differences were found are listed below the tables.

At 12-13 years (Table 12a), which was the first year of secondary schooling for most young people, the persistent group was significantly more dysfunctional over a range of domains. These included temperament style (activity, volatility, and ability to maintain attention on tasks), behaviour problems (aggression, hyperactivity and depression), and social skills (particularly cooperation, empathy and self control). They also tended to have poorer relationships with parents; experienced more problems at school; and reported higher levels of eating disordered behaviour. While most differences were between the persistent and low/non antisocial groups, on several domains the persistent group was also more problematic than the experimental group.

For the first time, significant differences between the experimental and low/non antisocial groups were evident on a small number of variables: school problems (as reported by both parents and teenagers), and cooperation.

Most group differences were in the small effect size range, although on several self-report measures differences were in the medium effect size range (self- reported aggression, cooperation, self control and school problems).

At 13-14 years (Table 12b), group differences were found on the same domains as at 12-13 years. Additionally, the low/non antisocial group tended to receive better quality parenting (more supervision and monitoring: parental knowledge of their child's activities and social contacts; greater warmth: more enjoyment of each other's company and perception of closeness) than the persistent and experimental groups (according to parents), and individuals in this group reported higher attachment to their parents. Parents of persistently antisocial individuals were more likely to smoke cigarettes or drink alcohol than parents of low/non antisocial adolescents, and the persistent group was more attracted to adventurous risk-taking activities than the low/non antisocial group.

* p<.05; ** p<.01; *** p<.001
There were no significant differences on :
Parent report of: anxiety-fearfulness, approach-sociability, depression, socioeconomic status, and family stress at 12-13 years; and
Child report of: anxiety-fearfulness, assertiveness, self-concept re physical appearance, body dissatisfaction, and drive to be thin at 12-13 years.

The persistent and experimental groups were similar to each other at this age, and different from the low/non antisocial group on many measures. The domains on which both groups were faring significantly worse than the low/non antisocial group were: behaviour problems (aggression, oppositionality and hyperactivity), parental supervision and warmth, mothers' cigarette and alcohol use, school difficulties, antisocial peer affiliations, lower attachment to parents, and lower cooperation and self control.

* p<.05; ** p<.01; *** p<.001
There were no significant differences on :
Parent report of: anxiety-withdrawal, approach-sociability, task persistence, assertiveness, obedience orientation, peer involvement, mothers's use of inductive reasoning, physical punishment, socioeconomic status, and paternal unemployment at 13-14 years; and
Child report of: friendship quality at 13-14 years.

The persistent group was also more oppositional, less cooperative and more involved with antisocial peers than the experimental group. A larger number of medium and large effect size differences were found on aspects such as aggressive and oppositional behaviour problems, school difficulties, antisocial peer affiliations, cooperation and self control.

At 15-16 years (Table 12c), significant group differences were evident on most domains. About two-thirds of these differences were between the two antisocial groups, on the one hand, and the low/non antisocial group, on the other (on aspects of temperament/personality - task persistence, negative reactivity, conscientiousness, risk taking; on behaviour problems - attention problems/hyperactivity, aggression, oppositionality, depression; on the assertiveness social skills dimension; on school adjustment; on peer relationships - antisocial peer affiliations and peer involvement; and on parenting style - parental monitoring/supervision and warmth of relationship).

While in general the persistent and experimental groups were similar to each other at 15-16 years, and both "worse" than the low/non antisocial group; on some variables only the persistent group (but not the experimental group) differed from the low/non antisocial group (primarily, the activity temperament factor, conduct disorder, social competence, involvement in organised peer group activities, parental use of punishment). As in the primary school years, there were also some domains in which the persistent group was significantly lower than both the experimental and low/non antisocial groups (parent-teen relationships, the personality factor of agreeableness, and both aspects of civic mindedness). Just over half the group differences were in the medium or large effect size range. These more powerful differences were on task persistence, aggression, oppositional and hyperactive behaviour problems, attraction to risk taking, school adjustment and antisocial peer affiliations.

Thus, from 11 to 16 years (Table 12d), differences between the persistent and low/non antisocial groups strengthened, and the experimental group began to resemble the persistent group. Consistent and powerful differences between the experimental and low/non antisocial groups were also found.

The data at 17-18 years (Year 12 for most participants) suggest that a different pattern of group differences may be emerging. Differences tended to be smaller, with relatively few medium and large effect sizes.

On some variables the two antisocial groups were both significantly different from the low/non antisocial group (on aggressive and hyperactive problems, antisocial peer affiliations, tendency to cope with difficulties by using drugs, and family cohesion). On others, the persistent group was significantly different from the low/non antisocial , but not the experimental group (on negative reactivity, intellect /openness, identity clarity, and adolescent and parent report of positive and negative aspects of their relationships). However, the general pattern was for the majority of group differences to be between the persistent group and the other two groups. Differences were on aspects such as personality style (agreeableness, conscientiousness, emotional control), behaviour problems (conduct disorder and oppositionality), coping style (coping with problems by ventilating feelings), optimism about the future, and quality of parent-child relationship. It seemed that at this age, the experimental group was becoming more like the low/non antisocial group rather than the persistent group.

* p<.05; ** p<.01; *** p<.001
There were no significant differences on :
Parent report of anxiety-withdrawal, approach-sociability, mother's use of inductive reasoning, family stress, socioeconomic status and parental unemployment at 15-16 years; and
Child report of: anxiety-fearfulness, openess, friendship quality, political activity, and receiving prestige/status from school at 15-16 years.

* p<.05; ** p<.01; *** p<.001
There were no significant differences on :
Parent report of: anxiety –withdrawal, approach-sociability, extraversion, neuroticism, and socioeconomic status at 17-18 years; and
Child report of: anxiety-fearfulness, neuroticism, openness, self-esteem, attachment to peers (communication), attachment to peers (trust), attachment to peers (alienation), attachment to mother (trust), attachment to mother (monitoring), coping (support seeking), coping (via diversion), coping (wishful thinking), cope (independence), the future (desire for privacy), the future (readiness for intimacy) at 17-18 years.

Predictors of antisocial behaviour by domain of functioning

Following the description of group trends at different ages and stages of development, we now examine the extent and strength of differences between the three groups (low/non antisocial, experimental and persistent) across various domains of functioning to determine whether particular profiles may be identified for the three groups. This discussion draws on the data already presented in Tables 11 and 12 (a), (b), (c) and (d). Domains relating to personal functioning will be presented first, followed by domains related to family and peer relationships.

Temperament/personality

Clear temperamental differences were consistently observed between the persistent group and the low/non antisocial group. From mid-childhood onwards, the persistent group was frequently rated as being more reactive, volatile and emotional than individuals in the low/non antisocial group. This high reactivity continued to be evident well into adolescence. The persistent group consistently reported ongoing difficulties in their ability to control emotional responses to events. The persistent group also exhibited greater difficulties remaining focused on activities or tasks, and were frequently rated by parents as exhibiting higher activity levels than their low/non antisocial peers. During mid-to-late adolescence, the persistent group rated themselves as being more outgoing than the low/non antisocial group.

It was difficult to identify a clear temperamental profile for the experimental group. For example, at times during the primary school years, the experimental group resembled the low/non antisocial group, attending better to tasks and activities, and exhibiting lower activity levels than those in the persistent group. However, during early-to-mid adolescence, the experimental group displayed some of the temperamental characteristics of the persistent group (for example, low task persistence, high negative reactivity, low conscientiousness). Around mid-to-late adolescence, the pattern changed again, with the experimental group resembling the low/non antisocial group more closely on many personality traits (for example, more agreeable, more conscientious, more open to ideas than the persistent group).

The three groups did not differ in their levels of sociability prior to adolescence, or rhythmicity (regularity, predictability) at any time.

Sensation seeking and risk taking

The persistent group rated themselves as taking more risks, seeking more sensations and thrills, and having an interest in a broader range of activities than those who engaged in no or little antisocial behaviour. The experimental group also reported taking more risks than those in the low/non antisocial group, although they were significantly lower than the persistent group on this measure.

Behavioural and emotional problems

As with temperamental characteristics, clear differences were noted between the persistent and low/non antisocial group in the level of behavioural problems exhibited. These differences were repeatedly observed over multiple data collection waves, and by more than one source. Most notably, the persistent group was consistently observed to engage in more frequent "externalising", "acting out" behaviours (for example, aggression, hyperactivity, oppositional behaviour) than individuals in the low/non antisocial group. In addition, their parents rated them as having more attention difficulties.

Besides these differences in externalising behaviour problems, some differences in the internalising spectrum (for example, depression, anxiety) also emerged. Individuals in the persistent group consistently rated themselves as being more depressed than the low/non antisocial group (an observation also made by their teachers at 11-12 years of age), although there were no group differences in parental ratings of children's depression levels at any stage. Interestingly, despite the high co-occurrence of depression and anxiety problems noted in previous research (Cicchetti and Toth 1998), the three groups generally did not differ in their levels of anxiety (with the exception of ages 5-6 and 13-14 at which the low/non antisocial group were reported as more anxious than the persistent group).

As with temperamental factors, the pattern of behavioural problems exhibited by those in the experimental group appeared to be less severe and tended to change over time. For example, from the ages of 7-8 to 12-13, the experimental group was consistently less aggressive than the persistent group. However, from the age of 13-14 on, the experimental group exhibited many of the same behavioural problems as the persistent group engaging in more acting out behaviours (hyperactivity, oppositional behaviour, socialised aggression) and exhibiting more attention difficulties than those in the low/non antisocial group. Once again, there appears to have been a decrease in this behaviour at 17-18 years, with the experimental group engaging in less oppositional, and norm-defying behaviour than the persistent group.

School adjustment and achievement

Substantial differences were observed between the groups in their levels of adolescent school adjustment and achievement. These differences were not evident in primary school. For example, the three groups did not differ in teacher ratings of readiness for school (at 5-6 years), reading skills assessed via a standardised test (at 7-8 years) or teacher ratings of academic skills (at 7-8 and 11-12 years). But by the secondary school years, both parents and teenagers consistently reported that individuals in the two antisocial groups were experiencing more school difficulties than those in the low/non antisocial group.

The measure used to assess school adjustment in adolescence encompassed various aspects of school life, including managing school rules and routines, completing homework and assignments, understanding the work in class, achieving a satisfactory standard, and getting on with teachers. Thus, it is a broader measure than academic achievement. While parent reports suggested that the experimental and persistent groups were experiencing a similar level of difficulties, self-reports indicated that these difficulties were significantly more pronounced among the persistent group.

Clear group differences were also evident in teenagers' attitudes to school. Individuals in the low/non antisocial group reported more positive feelings towards school, more positive relationships with teachers, more confidence in their ability to succeed at school, and were more likely to perceive their schooling as relevant for future life, than those in the other two groups. The persistent group reported significantly less confidence in their ability to succeed at school, and less positive relationships with teachers than those in the experimental group.

Social competence

Levels of social competence clearly differed between the groups, particularly in the areas of cooperation and self-control. For example, parent and teenager reports frequently revealed the persistent group to be less cooperative and less self-controlled than individuals who engaged in little or no antisocial behaviour. These differences were strongest during the early adolescent years (12-14 years), and when reported by the teenagers themselves. The persistent group were also noted to be less empathic, less responsible and more assertive than those in the low/non antisocial group.¹³

Differences in the social skills of the experimental group and the other two groups are harder to summarise. In some ways the experimental group was similar to the low/non antisocial group as they rated themselves as more empathic and more selfcontrolled than those in the persistent group. However, unlike the low/non antisocial group, at times they rated themselves as quite uncooperative.

Eating behaviours

During early adolescence (12-13 years) teenagers were questioned about eating behaviours and their perceptions of their body. The three groups did not differ on their self-concept of their physical appearance, their body dissatisfaction or their desire to be thin. However, the persistent group did report engaging in more bulimic (bingeing and purging) behaviours than those in the low/non antisocial group.

Coping strategies

Significant differences in the coping strategies adopted by the three groups were observed. Both the antisocial groups revealed a stronger reliance on the use of drugs (including cigarettes and alcohol) to help them cope with difficulties than the low/non antisocial group, and individuals in the persistent group were more likely to ventilate their feelings than individuals in the other two groups. The groups did not differ in their use of other coping strategies such as support-seeking, diversion-seeking, wishful thinking, or independence.

Civic mindedness

The persistent group demonstrated lower levels of social responsibility and reported fewer prosocial behaviours than those in the experimental and low/non antisocial groups.

Future aspirations

At 17-18 years, teenagers were asked about their future aspirations. Individuals in the low/non antisocial and experimental groups reported greater optimism for the future. Individuals in the low/non antisocial group also had a clearer idea of the type of person they wished to become. The groups did not differ in their desire for privacy or readiness for intimacy.

Peer relationships

The groups differed significantly in the characteristics of the peers they associated with. Based on their own and parents' reports, individuals in the two antisocial groups were consistently more likely to socialise with peers who engaged in antisocial activities than adolescents in the low/non antisocial group. Parents of persistent and experimental teenagers also reported that their children were more involved with their peers (made friends easily, interacted with a number of peers) than were low/non antisocial individuals.

On the other hand, individuals in the low/non antisocial group were observed to participate more often in organised peer group activities than those in the persistent group (at 13-14 and 15-16 years) and to report higher attachment to their peers (at 15- 16 years). The groups did not differ in their ratings of friendship quality (at 13-14 and 15-16), or their levels of trust, communication, and alienation from peers (at 17-18 years).

Taken together, these findings suggest that, while the groups did not differ in the quality of their friendships, the low/non antisocial group members were more attached to their peers and tended to interact with them in a structured setting (for example, while playing sport). The persistent group, on the other hand, appeared to spend more time with peers, but their time together was more likely to be unstructured, and more of their peers were antisocial.

Family factors

There were few differences between the groups on family sociodemographic characteristics. The three groups did not differ significantly on parents' ages at child's birth; parents' ethnic, educational, and occupational background; parents' employment status; family socioeconomic status or the number of children in the family. Neither did they differ on child's birth weight, child's gestational age, or the child's position in the birth order.

There were also few group differences in the degree of family stress reported by parents. However, by the age of 17-18 years, those in the two antisocial behaviour groups were less likely to belong to an "intact" family unit – that is, one that has not been disrupted by the death of a parent or a breakdown in the parental relationship (66.9 per cent persistent and 69.0 per cent experimental compared with 80.7 per cent low/non antisocial) and parents of the persistent group reported higher levels of marital conflict. At this timepoint parents of persistent individuals also reported significantly higher levels of family stress, and lower family cohesion, than parents of the low/non antisocial group.

Parenting practices and characteristics

At both timepoints at which parents' reports of their parenting practices were sought (13-14 years and 15-16 years), consistent group differences emerged. These findings suggest that the low/non antisocial group received better quality of parenting than the experimental and persistent groups, in that adolescents' behaviour was supervised to a greater extent, and their relationship with their parents was warmer than that of the two antisocial groups. Furthermore, parents of persistent antisocials reported using harsh discipline more frequently than parents of the low/non antisocial and experimental groups. There were no significant differences between the groups in mothers' use of inductive reasoning (use of explanation and reasoning to control child) or physical punishment (as reported by their parents during early adolescence) or teenager's reports of maternal supervision at 17-18 years.

Parental substance use

Information was also obtained from parents about their smoking and drinking habits. Mothers and fathers of persistent individuals were more likely than parents of the low/non antisocial group to smoke and/or drink alcohol, while mothers of the experimental group were more likely than mothers of the low/non antisocial group to smoke and/or drink alcohol.

Parent-Child Relationship

Differences in the quality of the parent-child relationship were evident from late childhood onwards, with parents of those in the low/non antisocial group consistently rating their child/teenager as easier to get on with than parents of individuals in the persistent group. At the most recent data collection wave, parents of the low/non antisocial group reported less conflict with their teenager, and more positive aspects to their relationship, than parents of persistently antisocial adolescents.

Teenagers' self-reports were in agreement with these findings, with the low/non antisocial group members perceiving themselves to have a better relationship with their parents (at age 11-12), feeling more attached to parents (at 13-14 years), and reporting higher warmth, higher communication, and less alienation from their parents (at 17-18 years) than members of the persistent group.

Overall, the quality of the parent-child relationship was similar among members of the experimental and low/non antisocial groups. Only at 13-14 years did individuals in the experimental group report lower attachment to parents than low/non antisocial group members.

Gender differences in predictors of antisocial behaviour

The three groups differed significantly in gender composition. There were significantly more males than females in the persistent group, and slightly more females than males in the experimental and low/non antisocial groups. Gender is a notable risk factor for antisocial behaviour. Thus, analyses in which gender is controlled may provide a somewhat misleading picture, as they effectively cancel out the contribution of this powerful risk factor. However, for reader interest we report briefly findings from analyses in which gender was controlled, in order to examine the impact of gender differences on findings.¹⁴ The results of these analyses are shown in Appendix 5.

In brief, these analyses revealed a similar pattern of findings to those previously described. The three groups did not differ during infancy and the preschool years, and the group differences found thereafter were generally in the same domains, and at the same timepoints, as those previously outlined. However, group differences were generally less powerful, and fewer differences were observed during the early-to-mid primary school years. This suggests that the results previously described may, to a certain extent, be influenced by gender differences, presumably because boys, who were more numerous in the persistent group, also tend to have more difficult characteristics. Nevertheless, group differences were generally robust across both sets of analyses (that is, those in which the effects of gender were controlled and those in which the effects of gender were not controlled).

The question of whether the pattern of predictors of antisocial behaviour differed across males and females was next examined. Two further sets of MANOVA analyses were conducted for each sex separately. The results of these analyses are summarised in Appendix 6, and briefly described here.

As before, no significant group differences were found until the beginning of primary school (ages 5-6). After this time, group differences emerged for both sexes in many of the domains previously identified (for example, temperament, behaviour problems, social competence, relationships with parents, school achievement and adjustment, peer relationships etc). However, there were fewer group differences than in previous analyses. At most timepoints the pattern of differences was similar across the sexes in terms of domains of functioning, although at particular timepoints significant differences were evident for one sex and not the other. For example, among girls at 5-6 years but not 7-8 years, group differences were found on levels of aggression, while differences in aggression were found among males at 7-8 years but not 5-6 years.

While these sex-specific analyses suggested fewer differences between the groups, it should be noted that group sizes were very small. Hence, the power of the sex-specific analyses was considerably reduced, making it more difficult to detect group differences, and to base confident conclusions on the findings of this set of analyses.

6. Discussion of findings and implications

Based on the findings described above, the following conclusions can be drawn about the factors which characterise individuals who report involvement in antisocial behaviour.

Persistent antisocial behaviour

A number of factors clearly and consistently characterised individuals who engaged in persistent antisocial behaviour. These were:

• more 'difficult' temperamental characteristics, such as a more intense, moody, irritable style, as well as difficulties remaining focused on tasks or activities (from early primary school on);
• consistent behaviour problems, particularly aggression, hyperactivity, oppositional/defiant behaviour, (from primary school on) and depression from late childhood on;
• lower self-control and persisting uncooperativeness (from late primary school on);
• consistent school difficulties and a negative attitude to school (from early secondary school on);
• association with antisocial peers (from late primary school on); a tendency to use drugs or to ventilate feelings to cope with life difficulties (during mid adolescence); and
• an attraction to risk-taking activities (during mid adolescence).

Other less powerful predictors of persistent antisocial behaviour included:

• individual characteristics, such as higher activity levels, higher thrill and adventure seeking, lower civic-mindedness, less optimism about the future, and lower identity clarity;
• problematic eating patterns (a tendency to "binge" and "purge"); family factors, such as disruptions caused by the death of a parent or a breakdown in the parental relationship, higher family stress, and more marital conflict;
• less optimal parenting practices (such as lower parental monitoring, lower warmth, higher punishment);
• parental substance use;
• more problematic parent-child relationship, higher levels of alienation from parents, and lower family cohesion; and
• lower rates of participation in organised group activities and low attachment to peers.

Experimental antisocial behaviour

Predictors of experimental antisocial behaviour were similar to those identified for persistent antisocial behaviour, but were generally not as constant, nor as severe. Among the strongest predictors of experimental antisocial behaviour in the ATP sample were:

• "difficult" temperamental characteristics (for example, higher negative reactivity, lower conscientiousness, difficulties remaining focused on an activity or task) which first emerged during early adolescence and persisted until mid-to-late adolescence;
• behaviour problems such as hyperactivity, oppositional behaviour, aggression, attention problems, that were evident in early adolescence and continued until late adolescence;
• lower cooperativeness (during the early adolescent years);
• school difficulties and a negative attitude to school (during secondary school);
• association with antisocial peers (from the commencement of secondary school);
• a tendency to use drugs to cope with life stress (during mid-adolescence); and
• a preference for risk-taking activities (during mid-adolescence).

Other weaker predictors of experimental antisocial behaviour included:

• family factors (for example, belonging to a family that had suffered a disruption through death or relationship breakdown);
• parenting practices (such as lower parental supervision and warmth, lower punishment) and parental characteristics such as maternal substance use; and,
• lower family cohesion

Developmental pathways

The three groups (persistent, experimental and low/non antisocial) did not differ on any domain during infancy and early childhood. No significant group differences emerged until the early primary school years when the persistent group began to display more difficult temperamental characteristics (for example, lower task orientation) and more problematic behaviour (for example, higher aggression and hyperactivity) than those in the low/non antisocial group.

It is possible that our findings concerning the timing of group differences may have partly resulted from sampling factors (for example, low rates of highly disadvantaged children in this community sample), or the omission of important variables during the early years (eg. detailed parenting measures). We compared the profiles of the most highly antisocial members of the persistent group (n =26, average number of different antisocial acts = 6.40) and the remainder of the persistent group (n= 105, average number of different antisocial acts = 3.26), to investigate the possibility of infant and early childhood difficulties among the most highly antisocial sub-group.

This comparison revealed trends for differences in the early years to the disadvantage of the highly antisocial group, principally in the domains of family socioeconomic status (a composite of both parents' education and occupational status), temperament (for example, cooperativeness), and hostile/aggressive behaviour. However, the group differences were generally not significant, when the adjusted criterion was used.¹⁵ Only at 1-2 years (1984), were group differences significant, when the most highly antisocial individuals were rated as less cooperative, and their families were of lower family socioeconomic status than the remainder of the persistent group.

These findings suggest that for a small sub-group of adolescents, who consistently exhibit very high levels of antisocial behaviour throughout adolescence, there were possible indicators of difficulties in the first few years of life. However, in the present study, which utilised a representative sample of Victorian children, these early indicators were relatively weak, possibly due to the small number of highly antisocial individuals in this sample.¹⁶. Overall, for the majority of youth who displayed persistent antisocial behaviour in adolescence, substantial temperamental and behavioural differences were not apparent until the early primary school years.

Implications

Taken together, these findings have a number of important implications for understanding the development of antisocial behaviour and for early interventions aimed at preventing this type of behaviour.

1. Some degree of antisocial behaviour is "normal" in adolescence

Consistent with previous research, the findings of this study suggest that some degree of antisocial behaviour is common among adolescents. For example, at 13-14 years, one in three participants had been involved in a physical fight in the past year, and at 17-18 years, over 40 per cent reported having skipped school at least once. Substance use (especially cigarette and alcohol use) was also relatively common, especially during mid-to-late adolescence. These findings add support for the view that some level of antisocial behaviour is "normal" among this age group. However, there are distinct patterns both in the timing, the frequency, and the nature of the antisocial behaviours, which needs to be taken into consideration by prevention strategies.

2. Early intervention to divert children from pathways to persistent antisocial behaviour appears most appropriate during the primary school years

The findings have important implications for the timing of interventions. Homel et al. (1999) make the point that interventions can occur early in life, or early in a pathway. While many assume that early in a pathway means early in life, this is not always true. Few studies of the etiology of antisocial behaviour have the data from infancy and early childhood required to examine this issue, making it difficult to determine the most appropriate timing for interventions.

Targeted interventions, which aim to divert individuals from problematic pathways, cannot begin until early in the pathway, when an individual is identified as being "at-risk" of developing serious and entrenched problems. In relation to persistent adolescent antisocial behaviour, our findings suggest that the primary school years are the critical years for identifying the start of pathways to teenage antisocial behaviour for the majority of children, and hence the optimal time for targeted interventions for children identified as "at-risk". This conclusion is based on the observation that there were no differences between any groups until 5 to 6 years of age, the commencement of primary school for most participants, and that differences were consistently and increasingly apparent after that time.

It is widely recognised that interventions during the earliest years of life are critical to the prevention of numerous emotional and behavioural problems (for example, hyperactivity, attention-regulation problems). Hence, more broad-based interventions (for example, home visiting programs), during infancy and early childhood, which aim to prevent the development of problems before they emerge, may prove beneficial. Infants and young children whose sociodemographic and familial characteristics place them at increased risk of later developing antisocial behaviour (for example, low socioeconomic status, parental relationship breakdown, abusive or inept parenting) would particularly benefit from such preventative efforts (Farrington 2002; Homel et al 1999).

Nevertheless, the current results suggest that when targeting the development of antisocial behaviour, the focus should be on the early primary school years as a crucial period to intervene. This conclusion is consistent with other Australian research. For example, Bor et al. (2001) found that aggressive behaviour at this age (5 years) was the strongest predictor of later antisocial behaviour (at age 14) in a sample of 3, 792 Australian children. There is some empirical support for the potential effectiveness of interventions targeted at early primary school. Prevention programs implemented during the early primary school years have shown considerable success in reducing behaviour problems and preventing the development of later antisocial behaviour (Farrington 2002; Greenwood, Model, Rydell, and Chiesa 1998; Homel et al. 1999). These programs have generally been multifaceted, involving teacher and parent training in behaviour management skills (for example, monitoring behaviour, using effective discipline, and promoting prosocial behaviours) and child skills training (for example, social, cognitive, and problem solving skills) (Homel et al. 1999).

3. Persistent antisocial adolescents exhibit a clear profile

Another major implication is that individuals who went on to engage in persistent antisocial behaviour during adolescence displayed an identifiable profile that included behavioural difficulties and temperamental characteristics. Multiple informants consistently reported these features, from mid-childhood onwards. These individuals were consistently more aggressive, more disinhibited, and more reactive than individuals who later engaged in little or no antisocial behaviour. Furthermore, from late childhood onwards, they were reported to have lower social skills and more likely to have formed antisocial peer friendships. Contrary to previous research, early learning difficulties were not a risk factor for this pattern of antisocial behaviour in this sample, however, a broader measure of academic adjustment did differentiate this group from early adolescence onwards. Given the consistency of these findings, it may be possible for clinicians, teachers, or parents to identify children who are at risk of developing persistent antisocial behaviour, and for whom intervention may be beneficial, at quite a young age.

While some may perceive temperamental characteristics to be 'fixed', and as a consequence, not amenable to intervention, research indicates that such characteristics are moderately stable over childhood and thus not immutable (Sanson, Hemphill and Smart 2002). Furthermore, temperament's impact on a child's development depends largely on its "fit" with the environment. Hence, in striving to optimise a child's development, attempts should be made to maximise the "fit" between the child and his/her environment, matching parenting and educational practices to the characteristics of the child, and helping the child to develop strategies to best manage his/her temperamental and behavioural tendencies.

4. Interventions targeting experimental antisocial behaviour need to be multi-faceted and focus on the early secondary school years

It was not until the early secondary school years that the experimental group could be differentiated from the low/non antisocial group. At this time, the experimental group became increasingly similar to the persistent group, exhibiting more behaviour problems, more school adjustment difficulties, experiencing poorer quality parenting, and associating more frequently with antisocial peers. However, towards the end of adolescence, these individuals appeared to be desisting in their antisocial behaviour, becoming more similar to the low/non antisocial group again.

Adolescence can be a time of "stress and storm" for some, which is later outgrown. It could be argued that it is more important to focus on those who display persistent antisocial behaviour than those who "test the waters". However, recent findings from the Dunedin Multidisciplinary Health and Development Study (Moffitt, Caspi, Harrington and Milne 2002) show that individuals who engage in transitory "adolescent-limited" antisocial behaviour have poorer outcomes (for example, mental health problems, substance dependence, financial problems, engagement in property offences) at age 26 than those who do not engage in teenage antisocial behaviour. Furthermore, while the behaviour of the experimental group may be transitory, it does result in considerable personal, social and economic cost to families, schools and the wider community. Given these findings, attempts to prevent the development of experimental antisocial behaviour would appear valuable.

The current findings suggest that the period following the transition to secondary school, in which the first differences between the experimental and low/non antisocial groups were evident, would be the appropriate age to target early intervention and prevention efforts for experimental antisocial behaviour. The constellation of characteristics which appear to identify a young person at risk for such behaviour include: a distinct change in behaviour from no particular problems in childhood to acting out, oppositional, uncooperative behaviour in the early teenage years, growing alienation from and difficulties at home and school, and the development of friendships with peers who engage in antisocial activities in the early-to-mid adolescent years.

As the experimental group appeared to exhibit a wide range of difficulties in many of the same domains as the persistent group, broad intervention programs targeting similar aspects of functioning to those earlier outlined in relation to the persistent group (for example, behavioural, skills-focused) would appear appropriate. Indeed, research suggests that community-based programs that focus on a range of styles and modes of treatment (for example, cognitive and behavioural) are effective in reducing recidivism in antisocial youths whereas deterrence programs (for example, involving "shock incarceration", and "scaredstraight" techniques) have been found to produce negative effects (Greenwood et al. 1998).

5. Predictors of antisocial behaviour are similar for males and females

Predictors of antisocial behaviour were generally similar for males and females. Few other studies have investigated this issue (as the majority have focused on males) and even fewer studies have had the data from infancy necessary to examine differences in the timing of development of antisocial behaviour among males and females. When group differences were examined separately for males and females, differences generally emerged at the same times and in the same domains for both sexes. These findings are consistent with those of Moffitt and colleagues (2001) who observed that the same risk factors predicted later antisocial behaviour for both males and females in a sample of 956 New Zealand children. Taken together, these findings suggest that interventions aimed at preventing the development of antisocial behaviour may be used equally well with males and females.

6. The role of family environment

Although those in the persistent and experimental groups were less likely to belong to an intact family by late adolescence, in general, there were few significant differences between the three groups on family sociodemographic characteristics (for example, family socioeconomic status, parental education, occupation, ethnic background, and number of children in the family). There is considerable debate in the literature regarding how such factors impact on individuals' propensity to engage in antisocial acts. For example, the classical view suggests that sociodemographic characteristics such as poverty and unemployment directly impact on antisocial behaviour by motivating an individual to offend. However, more recent research suggest that sociodemographic characteristics exert their effects on antisocial behaviour in a more indirect manner, for instance, by interfering with parents' ability to appropriately discipline and/or nuture their children (Barrera et al., 2002; Scaramella, Conger, Spoth, and Simons 2002; Weatherburn and Lind 1998). The findings from the current study lend support to the latter view.

Significant group differences were observed on many aspects of the family environment often cited as important to the prediction of antisocial behaviour. For example, individuals in the experimental and persistent group reported lower attachment to parents, and poorer quality parenting (for example, less supervision of the adolescents' activities, less warmth in their relationships with their parents) than those in the low/non antisocial group. Similarly, parents of individuals in the persistent group perceived their children to be less easy to get along with than parents of children in the low/non antisocial group and also reported the family unit to be less cohesive and close.

These findings suggest that the quality of parenting a child/teenager receives, and their relationships with parents, may impact on their engagement in antisocial behaviour. While these differences were not as strong as those observed for individual characteristics, school adjustment and peer relationships (typically falling within the small effect size range), they were of a similar, if not larger, magnitude to those observed in many previous studies.

7. Peer relationships and their influence

While the groups did not differ in the quality of their friendships, the low/non antisocial group members were more attached to their peers and more frequently interacted with them in a structured setting (for example, while playing sport). The two antisocial groups, on the other hand, appeared to spend more time with peers, but their time together was more likely to be unstructured. The existence of friendships with other antisocial youth was one of the most powerful factors differentiating the antisocial and low/no groups. These findings are consistent with a large body of research which suggests that associating with antisocial peers is one of the most powerful risk factors for adolescent antisocial behaviour.

8. The importance of school adjustment

Unlike previous research, the present study did not find school achievement (as assessed by a reading test at 7-8 years and teacher ratings of academic competence at 7-8 and 11-12 years) during the primary school years to be predictive of later antisocial behaviour. However, from the beginning of secondary school clear differences in participants' levels of school adjustment (for example, ability to manage school rules/routines, understanding the work taught in class) and attitudes to schooling (for example, perceived long-term relevance, relationship with teachers, confidence in own ability) were evident. A closer examination of the individual questions that made up the elementary school adjustment and achievement measures revealed a similar pattern of findings, with some group differences being observed on items relating to school adjustment (for example, ability to follow instructions, relationships with teachers, classroom behaviour, motivation etc) rather than items pertaining to school achievement. Hence, the current findings suggest that school adjustment (both how an individual adjusts to the routines and demands of school-life and how the school accommodates the child's characteristics and needs) and school attachment (his/her attitudes and attachment to school) are of greatest salience for the development of adolescent antisocial behaviour.

7. Next phase of the research

This report by no means provides an exhaustive overview of the nature, prevalence, and development of antisocial behaviour among participants of the Australian Temperament Project. Over the coming months, the project team will be engaging in a number of tasks in an endeavour to better understand the development of antisocial behaviour in this sample.

These tasks include:

• An examination of differences between individuals who engage in violent antisocial behaviours and those who engage in non-violent antisocial acts.
• An investigation of factors which may protect against the development of adolescent antisocial behaviour among those with many risk factors.
• The collection of further self-report data from participants (who will be 19-20 years of age in 2002) concerning their engagement in antisocial acts, contact with the criminal justice system, experiences of being a victim of crime, and their perceptions of the fairness of the criminal justice system.
• The analysis of data concerning the nature and prevalence of different crimes by community area to examine neighbourhood effects on antisocial behaviour.

In completing these tasks we hope to gain a better understanding of the factors that contribute to, and protect against, the development of antisocial behaviour in young people in Australia.

References

• American Psychiatric Association (1994), Diagnostic and statistical manual of mental disorders: DSM-IV (4th edn), American Psychiatric Association, Washington, DC.
• Baker, J. (1998), Juveniles in crime Part 1: Participation rates and risk factors, Sydney, Australia: New South Wales Bureau of Crime Statistics and Research, Sydney, Australia.
• Barrera, M., Prelow, H.M., Dumka, L.E., Gonzales, N.A., Knight, G.P., Michaels, M.L., Roosa, M.W. & Tein, J. (2002), "Pathways from family economic conditions to adolescents' distress: Supportive parenting, stressors outside the family, and deviant peers", Journal of Community Psychology, vol. 30, pp. 135-152.
• Bond, L., Thomas, L., Toumbourou, J., Patton, G. & Catalano, R. (2000), Improving the lives of young Victorians in our community: A survey of risk and protective factors, Centre for Adolescent Health, Parkville, Australia.
• Bor, W., Najman, J.M., O'Callaghan, M., Williams, G.M. & Anstey, K. (2001), "Aggression and the development of delinquent behaviour in children" , Trends and Issues in Crime and Criminal Justice, May.
• Catalano, R. & Hawkins, D. (1996), "The social development model: A theory of antisocial behaviour", in D. Hawkins (ed.) Delinquency and crime: Current theories, Cambridge, New York.
• Cicchetti, D. & Toth, S.L. (1998), "The development of depression in children and adolescence", American Psychologist, vol. 53, pp. 221-241.
• Cohen, J. (1988), Statistical power analysis for the behavioral sciences (2nd edn), Lawrence Erlbaum, Hillside, New Jersey.
• Dussuyer, I. & Mammolito, M. (1998), Reoffending patterns of juveniles in Victoria: Working paper, Department of Justice, Victoria, Australia.
• Farrington, D.P. (2002), "Developmental criminology and risk-focussed prevention", in M.Maguire, R. Morgan, & R. Reiner (eds) The Oxford handbook of criminology (3rd edn), Clarendon Press, Oxford.
• Farrington, D.P. & Loeber, R. (2000), "Epidemiology of juvenile violence", Juvenile Violence, vol. 9, pp. 733-748.
• Greenwood, P.W., Model., K.E., Rydell, C.P. & Chiesa, J.R. (1998), Diverting children from a life of crime: Measuring costs and benefits,Online http://www.rand.org/publications/MR/MR699/
• Homel, R., Cashmore, J., Gilmore, L., Goodnow, J., Hayes, A., Lawrence, J., Leech, M., O'Connor, I., Vinson, T., Najman, J. & Western, J. (1999), Pathways to prevention: Early intervention and development approaches to crime in Australia, Attorney-General's Department, National Crime Prevention, Canberra.
• Howard, S. & Johnson, B. (2000), "Resilient and non-resilient behaviour in adolescents", Trends and Issues in Crime and Criminal Justice, vol. 183.
• Kelley, B.T., Loeber, R., Keenan, K. & DeLamatre, M. (1997), "Developmental pathways in boys' disruptive and delinquent behavior", Juvenile Justice Bulletin, December.
• Loeber, R. (1990), "Development and risk factors of juvenile antisocial behavior and delinquency", Clinical Psychology Review, vol.10, pp. 1-41.
• Loeber, R. & Farrington, D. P (2000), "Young children who commit crime: Epidemiology, developmental origins, risk factors, early interventions, and policy implications", Development and Psychopathology, vol. 12, pp. 737-762.
• Loeber, R., Farrington, D.P., Rumsey, E., Kerr, C.A. & Allen-Hagen, B. (1998), "Serious and violent juvenile offenders", Juvenile Justice Bulletin, May.
• Loeber, R., Wung, P., Keenan, K., Giroux, B., Stouthamer-Loeber, M., Van Kammen, W. B. & Maughan, B. (1993), "Developmental pathways in disruptive child behaviour", Development and Psychopathology, vol. 5, pp. 101-133.
• Maughan, B., Pickles, A., Rowe, R., Costello, E.J. & Angold, A (2000), "Developmental trajectories of aggressive and non-aggressive conduct problems", Journal of Quantitative Criminology, vol. 16, pp. 199-221.
• 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, vol. 14, pp. 179-207.
• Moffitt, T.E., Caspi, A., Rutter, M. & Silva, P.A. (2001), Sex differences in antisocial behaviour: Conduct disorder, delinquency and violence in the Dunedin Longitudinal Study, Cambridge University Press, Cambridge.
• Moffitt, T.E. & Harrington, H.L (1996), "Delinquency: The natural history of antisocial behaviour", in P.A Silva and W.R. Stanton (eds)From child to adult: The Dunedin Multidisciplinary Health and Development Study, Oxford University Press, Auckland, New Zealand.
• Moffitt, T.E. & Silva, P.A. (1988), "Self-reported delinquency: Results from an instrument for New Zealand, Australian and New Zealand Journal of Criminology, vol. 21, pp. 227-240.
• Nagin, D. & Tremblay, R.E. (1999), "Trajectories of boys' physical aggression, opposition, and hyperactivity on the path to physically violent and non-violent juvenile delinquency", Child Development, vol. 70, pp. 1181-1196.
• Patterson, G.R., Reid, J.B. & Dishion, T.J. (1992), Antisocial boys, Castalia Publishing Company, Eugene, Oregon.
• Piquero, A. (2000), "Frequency, specialisation and violence in offending careers", Journal of Research in Crime and Delinquency,vol. 37, pp. 392-418.
• Prior, M., Sanson, A., Smart, D. & Oberklaid, F. (2000), Pathways from infancy to adolscence: Australian Temperament Project1983-2000, Australian Institute of Family Studies, Melbourne Australia.
• Rutter, M., Giller, H. & Hagell, A. (1998), Antisocial behaviour by young people, Cambridge University Press, New York.
• Sanson, A., Hemphill, S. A., & Smart, D. (2002), "Temperament and social development" in P.K. Smith and C.H. Hart (eds) Blackwell handbook of childhood social development, Blackwell Publishing, Oxford, UK.
• Scaramella, L.V., Conger, R.D., Spoth, R. & Simons, R.L. (2002), "Evaluation of a social contextual model of delinquency: A cross- study replication", Child Development, vol. 73, pp. 175-195.
• Victoria Police, Statistical Services Branch (2002), Victoria Police 2000/2001 Provisional Crime Statistics. Updated figures obtained from Victoria Police.
• Weatherburn, D. & Lind, B (2001), Delinquent-prone communities, Cambridge University Press, Cambridge, UK.
• Weatherburn, D. & Lind, B. (1998), "Poverty, parenting, peers and crime-prone neighbourhoods", Trends and Issues in Crime and Criminal Justice, April.

Appendices

Appendices to the report are available on request - please contact us if you require any of the appendices listed below.

• Appendix 1: Chi Square Tests of Independence results - gender differences in antisocial behaviour
• Appendix 2: Chi Square Tests of Independence results - group differences in antisocial behaviour
• Appendix 3: MANOVA results - the primary school years
• Appendix 4: MANOVA results - the secondary school years
• Appendix 5: MANCOVA results - covarying for sex
• Appendix 6: MANOVA results - sex-specific MANOVAs

About the authors

Ms Suzanne Vassallo

Ms Suzanne Vassallo is a Research Officer at the Australian Institute of Family Studies. In this capacity she is currently working on the collaborative project between the Institute and Crime Prevention Victoria examining factors that influence the development of antisocial and criminal behaviour among participants of the Australian Temperament Project. Suzanne has a Masters degree in Clinical Psychology. Prior to joining the Institute, Suzanne was employed as a Research Fellow in the School of Psychology at Deakin University.

Mrs Diana Smart

Mrs Diana Smart is a Research Fellow at the Australian Institute of Family Studies, and has been the Project Manager for the Australian Temperament Project since 1988. She has conducted research for the Victorian Education Department’s Curriculum and Research Branch and the Royal Melbourne Institute of Technology’s Education Unit, and has lectured in Psychology at Rusden Teachers College. Diana has a Master of Arts degree in Psychology and Diploma of Education.

Associate Professor Ann Sanson

Associate Professor Ann Sanson is Deputy Director (Research) at the Australian Institute of Family Studies, and is the Project Director for the recently commissioned Growing Up in Australia study (the Longitudinal Study of Australian Children). She is also an Associate Professor in the Department of Psychology at the University of Melbourne, where her teaching and research have been in the areas of developmental psychology, developmental psychopathology and conflict resolution. Ann has a PhD in Psychology, is a Fellow of the Australian Psychological Society, and has been an investigator on the Australian Temperament Project since its inception in 1983.

Dr Inez Dussuyer

Dr Inez Dussuyer is Assistant Director (Research) of Crime Prevention Victoria. Prior to this she was Manager of the Criminal Justice Statistics and Research Unit, Victorian Department of Justice for five years. She is a criminologist and previously worked in criminal justice and crime prevention research in academic institutions in Canada, France and Australia. Inez has a Doctorate in Sociology, a Master of Arts degree in Psychology and a Post Graduate Diploma in Criminology.

Mr Bill McKendry

Mr Bill McKendry is Director of Crime Prevention Victoria and leads a range of program initiatives to reduce crime and improve community safety within the framework of Safer Streets and Homes, the Government’s crime and violence prevention strategy. Bill has over 20 years experience in a range of senior management, research, policy and operational roles in Victoria Police. He has developed and implemented a number of successful statewide crime prevention and public awareness strategies in partnership with communities and government. Bill has a Master of Business Administration, a Bachelor of Arts in Criminal Justice Administration and a Post Graduate Diploma in Criminology.

Associate Professor John W Toumbourou

Associate Professor John W Toumbourou is a senior researcher at the Centre for Adolescent Health, Royal Children’s Hospital, and a member of the Department of Paediatrics at the University of Melbourne. He is a founding member and the current Chair of the College of Health Psychologists within the Australian Psychological Society. John is a principal investigator on a number of studies investigating healthy youth development, including the Australian Temperament Project and the International Youth Development study (a collaborative longitudinal study with the University of Washington), and has been involved in the development of a number of youth health promotion programs.

Professor Margot Prior

Professor Margot Prior is a member of the Departments of Psychology and Paediatrics at the University of Melbourne. She is a Fellow of the Academy of the Social Sciences of Australia and of the Australian Psychological Society. She began her career in Developmental and Clinical Psychology in the 1970s, and has taught Psychology at La Trobe, Monash and Melbourne universities. From 1994 to 2002 she was Director of Psychology at the Royal Children’s Hospital. Margot has a PhD in Psychology and has been an investigator on the Australian Temperament Project since its inception.

Professor Frank Oberklaid

Professor Frank Oberklaid is the Director of the University of Melbourne’s Centre for Community Child Health, located at the Royal Children’s Hospital, and is the author of numerous publications on children’s development. Frank has received state, national and international awards, lectureships and visiting professorships, including a Medal in the Order of Australia in 1998. He has a MBBS and has been an investigator on the Australian Temperament Project since its commencement.