Childhood predictors of adult criminality: a meta-analysis drawn from the prospective longitudinal literature.

AuthorLeschied, Alan

Numerous commentaries have appeared, particularly over the past ten years, noting the important contributions of developmental criminology in adding to both our theoretical and practical understanding of the life course of crime for children and adolescents that can continue into adulthood. Farrington (1997) suggests that an understanding of developmental constructs in the context of criminal behaviour contributes to an appreciation of the interaction and implications of life events at different ages that have certain predictable outcomes, characterized as factors that either relate to desistance or exacerbation of antisocial behaviour. Le Blanc and Loeber (1998) suggest that "the application of developmental perspectives to the study of offending is likely to advance current understanding of offending's causes and courses" (115-116).

The implications of such a developmental appreciation of life course trajectories can contribute considerably to our understanding of the effectiveness of prevention and intervention. A social developmental framework that integrates the major theoretical orientations of our current psychologically informed understanding of criminal behaviour, personal, familial, and structural variables can be identified that can guide the understanding of both "causal and mediating processes hypothesized to predict behavior over the course of development" (Andrews and Bonta 2007, 95). Farrington and Welsh (2007) more recently suggest that a social developmental framework can guide the selection of risk-focused targets in prevention and intervention.

Coincidentally, while social developmental theory has contributed to an understanding of life course trajectories for antisocial behaviour, a number of longitudinal studies have appeared reporting on empirical findings related to the early life experiences for children and youth and their relationship to adult antisocial outcomes. Such data are a critical part of theory building related to risk prediction and can enhance viewing risk as a construct based on longitudinal studies within a developmental framework. This area, referred to as developmental criminology, holds the potential for refining our understanding of how risk factors may work at different ages and stages within the lives of children and their families to predict offending into the adult years. The present meta-analysis drew on the prospective, longitudinal literature in child development in identifying childhood and youth predictors for later adult offending.

Literature review

Longitudinal studies now report on the link between early child experience and subsequent antisocial behaviour. These findings suggest that parental inability to foster self-control in their children, neuropsychological disorders, a variety of negative parenting practices, coercive family interactions, and an inability of children to develop age-appropriate social skills (Lacourse, Cote, Nagin, Vitaro, Brendgen, and Tremblay 2002) appear promising in providing a basis for planning for prevention. The research on developmental disorders in children has reflected a range of findings from viewing children in isolation to understanding the impact of the social contexts that contribute to a child's potential for risk. Current findings from the longitudinal literature suggest that certain childhood disorders such as in the development of antisocial orientation will have different developmental trajectories, influenced by different systemic variables (Silk, Nath, Seigel, and Kendall 2000). And while Loeber and Farrington (2000: 746) suggest that "[t]he majority of childhood disorders reflect age normative problem behaviors which most children give up as they grow up," the challenge for developmental researchers is to identify which behaviours identified early in childhood are not transient developmental reactions but rather relate to later difficulty. Within the criminogenic risk context, early warning signs of protracted difficulty identified by a number of researchers suggest that such childhood factors as temperament, impulsivity, social withdrawal, aggression and hyperactivity associated with disruptive behaviour, family-based factors reflecting poor parenting practices, low supervision, physical punishment, neglect and poor communication, age, and gender (Loeber and Farrington; Hanish and Guerra 2002; Lacourse et al. 2002; Moffitt, Caspi, Harrington, and Milne 2002) all are factors that can play a role in the early prediction of later criminal conduct.

Relevance of theory

Social developmental theory can assist in creating a conceptual framework for identifying appropriate targets for intervention. For example, therapeutic progress is more likely to occur when theory is emphasized in intervention (Conduct Problems Prevention Group 2002; Kazdin 1997). In addition, as emphasized in the risk-predication literature by Andrews and Bonta (2007), a differential understanding of disorders that are influenced by dynamic over static characteristics and the timing or influence of systemic factors that can differentially affect childhood and adolescent outcomes will provide critical input to the understanding of prevention and early intervention.

The science of criminal conduct is now mature to the point where sufficient studies exist reporting on the long-term outcomes for children who experience early disruptions that can lead to later persistent involvement in both the youth and adult criminal justice systems. While small relative to the cross-sectional literature related to risk, the number of longitudinal studies is now adequate such that general findings drawn together through meta-analysis can provide policy makers, practitioners, and researchers with important findings on the identification of factors placing children and youth at risk. The importance of testing the predictive adequacy of early childhood experience within such a methodological framework lies in the increased reliability and validity of findings based on longitudinal studies. Acknowledging the contributions of meta-analysis, particularly within the criminal justice literature (Lipsey and Wilson 1998), this meta-analysis drew on the prospective longitudinal literature in risk prediction to examine static and dynamic risk factors that relate to children and youth who progress into the adult criminal justice system.


Studies were identified relating predictors of criminogenic risk for youths to their eighteenth birthday, or the age of majority, in determining entry into the adult criminal justice system in the country where the study took place. Only studies that were prospective and longitudinal were included. This is a methodological improvement over the use of cross-sectional studies used in most meta-analyses, which are less able to account for previous experiences of the subjects in the studies and how these might affect the findings. Longitudinal studies, particularly where concern is focused on developmental risk, will be more sensitive to within-subject comparisons. Relevant literature both published and unpublished from the major electronic databases was included: Psychinfo, ERIC, Social Work Abstracts, Medline, and Criminal Justice Abstracts. Searches were limited by date of publication from 1994 to 2004--a period during which the major prospective, longitudinal studies have appeared. The population was limited to childhood, school age, and adolescence. If the database did not allow for population limits, the keyword youth was included in each search. Thirty-seven literature searches were performed on each database. Each search reflected a variation in the combination of 17 keywords: meta-analysis, longitudinal, crime, criminality, criminal, involvement, prediction, predictors, trajectories, risk and risk factors, at risk populations, determinants, delinquency, offending, young offenders, and recidivism. The keywords were selected on the basis of reviews and meta-analyses in the area. Indexed databases that generated keywords for searches were also included. References of each study were reviewed to ensure all relevant studies were reflected by the search. Search terms generated 48 studies of which 38 contained data that were reported in a format that was amenable for the analysis.

Coding of studies

Prior to statistical analysis, data from the selected studies were coded into categories including authorship and cohort name to ensure non-duplication of the data source. Studies were also coded according to predictor variables that were assigned to one of two major categories: (1) family factors that included static risk, parental mental health, parental management, family structure, and adverse family environment; and (2) child factors including static risk, internalizing, externalizing, social interpersonal, and developmental concerns, child-specific school and learning issues, pro-social behaviour, and criminal history.

For both family and child factors, static risk was defined as one that had occurred in the past and could not be reversed or changed. For family factors this included parents' criminal history and complications during pregnancy; for child factors this included the age of onset of both the child's criminal behaviour and drug use. Dynamic risk characterizes factors in the lives of children/youth that are changeable over time and will be most amenable to intervention. In family factors, this includes parental management; for child factors, dynamic factors include lack of control and truancy.

Adult outcomes for each study were coded for official conviction or self-report. A further category focused on the age of participants prior to reaching the age of majority, and data were coded as early (birth to 6 years), mid (7 to 11 years), or late childhood/adolescence (12 to age of majority).

Demographic summary

Sample sizes were summed across the 38 studies reflecting 66,647 participants. There were 43,586 males...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT