Chapter 3: Adolescent Antisocial Behaviour and Mental Health
3.2.1 The Effects of Externalising Problems on Criminal and Antisocial
The economics, psychiatry, and criminology literatures have explored the effects of externalising problems on antisocial and criminal behaviour. For example, in the economics literature, Fletcher and Wolfe (2009) investigate the effects of ADHD at ages 5 to 12 on criminal behaviour at ages 18-26 using a subsample of 12,046 adolescents drawn from the National Longitudinal Study of Youth dataset. They measure childhood ADHD retrospectively utilising 17 ratings of ADHD symptoms collected when the respondents were between the ages of 18 and 28.56
Fletcher and Wolfe (2009) estimate logit models controlling for neighbourhood and family fixed effects to explore the effect of childhood ADHD on the probability of stealing, drug dealing, burglary, robbery, arrest, conviction, and involvement in any crime. Their findings suggest that individuals who have ADHD in childhood have a greater probability of committing any crime, stealing, robbery, and being arrested at ages 18-26. In contrast, there is no statistically significant effect of childhood ADHD on the probability of burglary or the probability of being convicted of a crime.
However, their analysis arguably suffers from a number of shortcomings. Firstly, Fletcher and Wolfe (2009) utilise retrospective reports of ADHD symptoms, which may suffer from measurement error, thus leading to attenuation bias. Secondly, they explore the effects of ADHD on criminal behaviour using a series of single-equation models. However, the error terms of the models are likely related via unobserved characteristics affecting the decisions to commit each of the crimes. Hence, the efficiency of their estimates is likely to be reduced.
Thirdly, the authors do not account for other co-morbid mental health conditions such as depression, see Anderson (1987). Finally, they are selective of the measures of adult criminal behaviour that they make use of. For example, they do not investigate the effect of ADHD on the measures of violent criminal activity reported in the National Longitudinal Study of Youth.
An important contribution from the psychiatry literature is Murray et al. (2015), who explore the effects of hyperactivity and conduct problems at age 11 on criminal behaviour at age 18. The authors use data from the Brazilian Pelotas Birth Cohort Study (sample size = 3,618) and the Avon Longitudinal Study of Parents and Children (sample size = 4,103).
56
The respondents, aged 18 to 28, were asked to recall whether they had 9 (8) symptoms of inattention (hyperactivity/ impulsivity) between the ages of 5 and 12.
77 They measure hyperactivity (conduct) problems using the hyperactivity/ inattention (conduct) problems subscales of the Strengths and Difficulties Questionnaire (SDQ). They measure nonviolent crime using a binary variable that equals 1 if in the past year the individual committed theft from shops, vehicles, or people; property damage; vehicle theft; drug dealing; burglary; handling stolen goods; or arson. Murray et al. (2015) measure violent crime using a binary variable that is equal to 1 if the individual engaged in robbery; assault; and carrying, or using, a weapon.
Making use of logit models, the authors find a positive association between hyperactivity at age 11 and participation in violent crime at age 18 for both males and females in Brazil. For the UK, their findings suggest that hyperactivity at age 11 has a statistically significant effect on whether males commit violent crime at age 18, but does not affect whether females commit violent crime at age 18. In addition, for both Brazil and Britain the authors offer no evidence to suggest that hyperactivity at age 11 increases the probability that an individual participates in nonviolent crime in both Brazil and the UK. In comparison, in Brazil and the UK, conduct problems at age 11 increase the probability that an individual engages in both violent and nonviolent crime.
Their analysis is arguably subject to a number of shortcomings. Firstly, the authors do not control for the adolescent’s social characteristics such as whether they eat evening meals with their family; and whether they stay out late, without their parents knowing their whereabouts. Sen (2010) suggests that the frequency that adolescents eat evening meals with their family is inversely associated with their likelihood of participating in ASB. In a similar vein, Iacovou (2012) suggests that adolescents who stay out late without their parents knowing their whereabouts are at an increased risk of engaging in ASB. Secondly, in common with Fletcher and Wolfe (2009), they estimate a series of single-equation models, thus reducing the efficiency of their estimates.
In the criminology literature, Dalsgaard et al. (2013) investigate the effects of childhood ADHD on adult criminality in Denmark. The authors measure childhood ADHD using clinical diagnoses of 206 children aged 4-19 at the Risskov Psychiatric Hospital, Denmark, from 1969-1989. The authors match diagnoses of childhood ADHD to criminal convictions data from 2000 from the Danish National Crime Register. To form a control group, they utilise data on the criminal convictions of the general population, drawn from the Danish National Crime Register.
Dalsgaard et al. (2013) use Cox proportional hazards models to explore the effects of childhood ADHD on whether the individual receives a criminal record as an adult.57 Dalsgaard et al. (2013) suggest that individuals who were diagnosed with ADHD as a
57 Cox proportional hazard models are a class of survival model used to explore how a set of covariates affects the time until an event occurs (e.g. receiving a criminal record), see Greene (2011).
78 child were 5.6 times more likely to receive a criminal conviction as an adult relative to the general population.
However, their analysis is subject to a number of limitations. Firstly, Dalsgaard et al. (2013) utilise data on criminal convictions in adulthood rather than data relating to participation in criminal behaviour. Arguably, ADHD may affect the likelihood of an individual who has committed criminal behaviour being caught and convicted, rather than affecting the probability that they engage in criminal behaviour per se. If this is the case, their findings may not reflect the effects of ADHD on participation in criminal