Childhood Psychopathology Predicts Adolescence-Onset Offending

Abstract

Moffitt, Caspi, Harrington, and Milne (2002) found in a follow-up study that many of the supposedly adolescence-limited offenders had committed offenses past adolescence. This finding raises the question of whether adulthood starts later or whether there are two distinct delinquency types, adolescence limited and adolescence onset, each with its own etiology. The present study tested whether the adolescence-onset group could be predicted by childhood psychopathology, as compared to the adolescence-limited and no-offender groups. In sum, 355 boys and girls were included, who reported on their emotional and behavioral problems when they were 11 to 13 years old and on their self-reported criminal behavior 8 to 13 years later. The findings lend support tothe hypothesis that adolescence-onset offenders can be distinguished from adolescence-limited offenders.

Keywords

criminal behavior psychopathology childhood predictors longitudinal

An influential theory about the classification of antisocial behavior is Moffitt’s dual taxonomy model (1993). This theory describes two types of antisocial offenders: the life-course-persistent offender, who engages in illegal behavior from a young age and continues to worsen thereafter, and the adolescence-limited offender, who engages in illegal activities only during adolescence. Both types are supposed to have different etiologies. The etiological basis of the life-course-persistent offender is thought to consist of an interaction between neuropsychological problems—verbal deficits, such as receptive listening and reading, and executive deficits, such as inattention and impulsivity—and disadvantageous environments (Moffitt, 1993). Furthermore, the presence of mental disorders—mania, schizophrenia, drug and alcohol abuse, depression and anxiety disorders (Robins & Regier, 1991)—may be associated with the continuity of antisocial behavior of the life-course-persistent offender (Moffitt, 1993). Criminal behaviors of the adolescence-limited offender are hypothesized to result from the presence of antisocial role models and a so-called maturity gap; that is, biologically mature adolescents are ready to live an adult life in which they can legally gain resources, but society holds them back. To still gain resources, adolescents copy the illegal behavior of antisocial role models (Moffitt, 1993). The existence of life-course-persistent and adolescence-limited offenders has been confirmed (e.g., Kratzer & Hodgins, 1999; Moffitt & Caspi, 2001; Raskin White, Bates, & Buyske, 2001). But other types have been found as well—for example, the adolescence-onset offenders and the adult-onset offenders (e.g., Aguilar, Sroufe, Egeland & Carlson, 2000; Kratzer & Hodgins, 1999; Lay, Ihle, Esser & Schmidt, 2005; Raskin White et al., 2001; Scholte, 1999).

As noted before, adolescence-limited offenders start and stop their criminal behavior during adolescence. However, as Moffitt, Caspi, Harrington, and Milne (2002) found in a follow-up study on males who were defined as adolescence-limited offenders, many of the supposedly adolescence-limited offenders had committed offences past adolescence. This finding raises the question of whether the adolescence-onset type is really a separate delinquency type or is still the adolescence-limited type. In the latter case, the upper age range of adolescence increases. Given their findings, Moffitt et al. suggested that the transition from adolescence to adulthood is now at the age of 25 years. Yet, there are indications that psychopathology is related to late-start delinquency (e.g., Hodgins, 1995; Moffitt, Caspi, Dickson, Silva, & Stanton, 1996). If the presence of mental disorders indeed predisposes an individual to persist in criminal behavior past adolescence, then adolescence-onset offending may not be similar to adolescence-limited offending but similar to adult-onset offending. The present study aims to shed light on whether adolescence-onset offenders constitute a different group from adolescence-limited offenders and adult-onset offenders.

The presence of a mental disorder may be a precursor of life-course-persistent offending, adolescence-onset offending, and adult-onset offending. As presented above, for the life-course-persistent offender, psychopathology may enhance criminal behavior. Psychopathology may also play a role in the criminal behavior of individuals who start offending during adolescence or adulthood and continue their criminal behavior over time. Note that Moffitt et al. (1996) suggest that adolescence-limited offenders are protected from becoming persistent offenders because, among other things, they have no mental problems. This suggestion implies that adolescence-onset offenders may continue their criminal behavior (started during adolescence) as a result of emerging psychopathology. In a follow-up study, Moffitt et al. (2002) found that about one third of the life-course-persistent and adolescence-limited men reported that they had sought treatment for a mental disorder, whereas the nonoffender group had virtually no report of mental disorder. Similarly, Hodgins (1995) suggested that adult starters among the mentally disordered begin their antisocial behavior at about the time that symptoms of their mental disorder emerge, which are thought to include cognitive–perceptual symptoms and drug abuse. Windle and Windle (1995) found that male adult-only aggressors were more likely than the control group to suffer from psychiatric conditions. Thus, psychopathology may be one of the factors that discriminate the adolescence-onset offenders and the adult starters from the adolescence-limited offenders. Psychopathology may maintain offending started during adolescence, or it may initiate offending started during adulthood. Adolescence-limited offenders then are thought to be similar to nonoffenders in that they have no psychopathology or no serious psychopathology.

Mental disorders related to delinquency include attention-deficit/hyperactivity disorder (ADHD), conduct disorder, antisocial personality disorder, schizophrenia and other psychotic disorders, drug abuse, posttraumatic stress disorder, and affective disorders (e.g., Eronen, Angermeyer, & Schulze, 1998; Hodgins, 1992, 1995; Pratt, Cullen, Blevins, Daigle, & Unnever, 2002; Vermeiren, 2003). Note that symptoms of conduct disorder and antisocial personality disorder include criminal acts and are therefore tautological to criminal behavior.

Adult mental disorders can be predicted by childhood variables, such as conduct symptoms, depressive and psychosomatic symptoms, attentional problems, social deficits, psychotic symptoms, emotional problems, and cognitive impairment (e.g., Harrington, Fudge, Rutter, Pickles, & Hill, 1990; Kagan & Snidman, 1999; Kim-Cohen et al., 2003; Poulton et al., 2000; Rutter, 1995). This implies that childhood psychopathology maybe predictive of life-course-persistent delinquency and delinquencystarted during adolescence or adulthood, which persists over time. In fact, Arseneault et al. (2003) found that childhood psychotic symptoms could predict violence by persons with schizophreniform disorder. Lay et al. (2005) observed that among males, conduct disorder and ADHD during childhood increased the risk for delinquency started in adolescence and early adulthood, as compared to no delinquency. This risk also increased by acute life events, externalizing behaviors, and drug use during adolescence. But Lay et al. state that even though childhood conduct problems could predicte late-onset offenders, such offenders were not engaged in serious criminal behavior before the age of 18; therefore, they should be seen as individuals with a long history of disruptive behavior and not as life-course-persistent offenders.

We hypothesize that the etiology of the adolescence-onset offenders differs from that of the adolescence-limited offenders by the presence of childhood psychopathology. Specifically, we hypothesize that childhood psychopathology can predict the adolescence-onset offender, the adult-onset offender, and the life-course-persistent offender, as compared to the adolescence-limited and nonoffender. The present study focuses on those aspects of childhood psychopathology that are associated with mental disorders related to criminal behavior. The focus is not on mental disorders that are, by definition, related to criminal behavior. Thus, the focal point is on psychotic-like, depressive, and anxiety symptoms and attention problems rather than on conduct disorder symptoms.

The database used is from a longitudinal study in the Netherlands (Verhulst & van der Ende, 1992). Because of the numerous participants tested and its longitudinal design, this study provides an exceptional data set to examine the above research question. During this study, the Youth Self-Report (YSR; Achenbach, 1991; Verhulst, van der Ende, & Koot, 1997) was administered. The present study uses the subscales of the YSR that refer to mental problems related to, but not a tautology to, criminal behavior—namely, Thought Problems (items refer to hallucinations, delusions, obsessions, and strange behaviors), Anxious/Depressed, Withdrawn, and Attention Problems.

Criminal behavior is based on self-reports, which many criminologists now use because they are a less biased measure (e.g., with official records, systematic arrest and conviction bias may exclude some types of offenders) and they have been shown to be valid and reliable (Hindelang, Hirschi, & Weis, 1979; Hirschi, Hindelang, & Weis, 1980; Sutterer & Karger, 1994). The present study includes boys and girls. Only a few previous studies separately studied the offender types of males and females (e.g., Aguilar et al., 2000; Kratzer & Hodgins, 1999; Moffitt & Caspi, 2001). Moffitt and Caspi (2001) argue that the origins of life-course-persistent and adolescence-limited offending are explanatory across the sexes and irrespective of sex. The present study therefore shows separate and combined analyses of boys and girls. We control for educational level because higher educational level and cognitive abilities have been found to inhibit criminal activity (Kratzer & Hodgins, 1999; Piquero & White, 2003; Redondo, Sanchez-Meca & Garrido, 1999).

Methods

Participants

The original sample consisted of participants of the Zuid-Holland prospective longitudinal study (Verhulst & van der Ende, 1992). Zuid-Holland is a Dutch province with urban and rural areas. The sample was drawn from municipal birth registers that list all residents in the Dutch province. Based on victim reports and police records, crimes reported in the Netherlands ranged from 6.2% in 1980 to 9.6% in 1998. Police records show that among adolescents aged 12 to 17, the police interviewed 2.8% in 1980 to 4.2% in 1998 (Eggen & Goudriaan, 2008). These numbers reflect current offenses. In the present study, data on lifetime offenses were collected.

The original study measured participants’ behavioral and emotional problems at five 2-year intervals, starting in 1983 (n = 2,076). From the third wave on (1987), the YSR (Achenbach, 1991; Verhulst et al., 1997) was used to measure behavioral and emotional problems. In the sixth wave (1997-1998), criminal behavior was assesed. For the present study, participants included were who completed the YSR when they were 11, 12, or 13 years old and were at least 21 years of age when they participated in the interview about criminal behavior. The first condition was to ensure that we measured childhood psychopathology; the second ensured that we would not unjustly classify a person as an adolescence-limited offender when in reality he or she was an adolescence-onset offender. As described in the section about group allocation, the cutoff age was 18 years for the adolescence-limited group (those who reported delinquency between the age of 12 and 18) and for the adolescence-onset group (those who reported delinquency between the age of 12 and 18 and past the age of 18). Setting the minimum age when criminal behavior was assessed to 21 years reduced the chance that an individual would be unjustly classified as an adolescence-limited offender when in reality he or she was an adolescence-onset offender. For example, a boy who completed the criminal behavior interview at 18 years and reported criminal behaviors at age 16 and 17 would have been defined as an adolescence-limited offender. In reality, he may have committed offenses at the age of 20, thereby making him an adolescence-onset offender. Had we included this participant, we would have misclassified him.

For individuals to be at least 21 when criminal behavior was assessed, we were limited to participants from Waves 3 and 4. A number of participants completed an YSR twice, when aged 11 to 13 (at the third and fourth waves). As such, the YSR completed at the oldest age was used for analyses, which resulted in 375 participants (177 boys, 198 girls). See Figure 1 for the selection of the final sample by wave (Waves 3 and 4). Criminal behavior was assessed at ages 21 (n = 120), 22 (n = 111), 23 (n = 107), 24 (n = 35), 25 (n = 1), and 26 (n = 1). At the time of the interview (when ages ranged from 21 to 26), 2.0% had completed primary school; 25.5%, lower general secondary education; 55.2%, higher general secondary education; 14.4%, higher vocational education; and 2.8%, university. Information on educational level was missing for 2 participants.

Figure 1.

Flowchart of Participant Inclusion for Present Study.

Measures

Youth Self-Report

The YSR (Achenbach, 1991; Verhulst et al., 1997) is a questionnaire for respondents aged 11 to 18 years. It consists of two parts: competencies and problems. For the present study, four of the eight subscales of the 112-item problem-behavior part were analyzed. Items are rated on a 3-point scale (0 = not true, 1 = somewhat or sometimes true, 2 = very true or often true), and all ratings refer to problems or symptoms experienced during the preceding 6 months. The four subscales are

Anxious/Depressed: lonely; harms self; fears impulses; needs to be perfect; feels unloved; feels persecuted; feels worthless; nervous; fearful, anxious; feels too guilty; self-conscious; suspicious; suicidal thoughts; unhappy, sad, depressed; worries

Withdrawn: would rather be alone; refuses to talk; secretive; shy; lacks energy; unhappy, sad, depressed; keeps from getting involved

Attention Problems: acts young; cannot concentrate; cannot sit still; confused; daydreams; acts without thinking; nervous; poor school work; clumsy

Thought Problems: cannot get mind of thoughts; repeats actions; hears things; sees things; stores up unneeded things; strange behavior; strange thoughts

The sum score of each subscale was used in the analyses. The Dutch version has good reliability and validity (De Groot, Koot, & Verhulst, 1996; Verhulst et al., 1997). The Cronbach’s alpha for Withdrawn, Anxious/Depressed, and Attention Problems is .65, .84, and .70, respectively; the Cronbach’s alpha of Thought Problems is rather low (.51) but is identical to that reported in the Dutch manual of the YSR (Verhulst et al., 1997). This low internal consistency can be explained by the fact that Thought Problems is a composite score of several unrelated symptoms (obsessive-compulsive symptoms and psychotic symptoms, such as hearing and seeing things that are not there).

Criminal behaviors

Criminal behaviors were assessed during an interview. A standardized questionnaire was used, an adaptation from the questionnaire used by Junger-Tas, Terlouw, and Klein (1994) and Klein (1989). The present version was adapted to make it suitable for administration to adults. Furthermore, the categories of traffic offense and fraud were added. The interview measured 49 types of criminal behaviors, divided into seven categories: problem behavior (defined as fare dodging, joyriding, and graffiti), vandalism, offenses against property, aggression, drugs, traffic offenses, and fraud. Participants were first asked if they had ever performed each behavior, which was followed by more specific questions (Donker, Smeenk, van der Laan, & Verhulst, 2003). For the present study, questions about the first and last age at which a behavior occurred (if performed at all) were used for group allocation.

Group Allocation

Several criminal and problem behaviors are rather common before the age of 15 (e.g., Sutterer & Karger, 1994). We found that when compared to the other included criminal behaviors, stealing from store and vandalism were rather common, especially during childhood. Of the total sample, 30.4% had stolen from a store, of which 54.4% had committed this act before the age of 12 (the cutoff age used in the present study for adolescence was 12 years). Similarly, 23.2% of the total sample had committed a vandalistic act, of which 24.1% had committed it before the age of 12. Therefore, stealing from store and vandalism are not regarded as criminal behaviors if they occurred before the age of 12 but are regarded as criminal behaviors if they occurred after the age of 12. (A table with the number of participants that self-reported criminal behaviors—by age category and when the behavior was first reported—is available by request.)

The following behaviors were regarded as serious criminal behaviors when conducted at any age: more than two different vandalistic acts (damaging different properties such as phone booths, windows, trees, furniture), graffiti, theft (motor vehicle, bicycle, and bags and the like; pickpocketing and burglary; from car, phone booth, or vending machine), fare dodging in train, stealing (from work, home, school, other), handling stolen goods, threat, arson, violence (directed at people, wounding with weapon), disturbances in public, carrying a weapon, drugs (using hard drugs, selling drugs), traffic violations (joyriding, not stopping after an accident), and fraud (insurance, tax, and welfare; see also, Kratzer & Hodgins, 1999; Moffitt et al., 2002; Raskin White et al., 2001). Answering yes to any of the items was sufficient to classify a participant as an offender, except for the item vandalistic acts, which was operationalized as damaging different properties, such as phone booth, windows, trees, furniture, and other (total of 12 properties). A participant had to have damaged at least two types of properties. Further, these vandalistic acts, and “stealing from store” had to have been committed past the age of 12 years in order to classify a participant as an offender. When this behavior was performed before the age of 12 or when only one type of property had been damaged, it was not used for group allocation. Based on the age in which the criminal behavior was performed, participants were allocated to one of the following groups:

Life-course-persistent offenders: started criminal behavior before the age of 12 and continued into adulthood (Moffitt, 1993); this group consisted of 38 individuals (10.1% of the total sample; 23 males, 15 females).

Adolescence-limited offenders: performed criminal behavior only during adolescence (Moffitt, 1993), operationalized as criminal behavior between the ages 12 to 18; this group consisted of 70 individuals (18.7% of the total sample; 38 males, 32 females).

Adolescence-onset offenders: started their criminal behavior during adolescence—did not report serious criminal behavior before 12 years but did report serious criminal behavior during adolescence and past the age of 18; this group consisted of 114 individuals (30.4% of the total sample; 71 males, 43 females).

Adult-onset offenders: started their criminal behavior during adulthood—that is, past the age of 18 years; this group consisted of 30 individuals (8% of the total sample; 13 males, 17 females).

No offender: had never performed criminal behavior; this group consisted of 103 individuals (27.5% of the total sample; 18 males, 85 females).

A small group of offenders (5.3% of the total sample) did not fit in the above groups and were excluded from further analyses. They were childhood-limited offenders (n = 3) and offenders who performed criminal behavior during childhood and adolescence (n = 17). Thus, the final sample comprised 355 participants who were included in the analyses.

Based on the dual taxonomy model (Moffitt, 1993) and previous studies (e.g., Hodgins, 1995; Moffitt et al., 1996; Moffitt et al., 2002), our hypothesis was that childhood psychopathology can predict the adolescence-onset offender, adult-onset offender, and life-course-persistent offender but not the no-offender and adolescence-limited offender. Therefore, the contrast group is a combination of the no-offender and adolescence-limited group and, as such, is referred to as the no-serious-offender group.

Statistical Analyses

Based on analysis of variance with Bonferroni corrections, groups were compared on Attention Problems. Because Thought Problems (skewness = 1.7, kurtosis = 2.8), Withdrawn (skewness = 1.3, kurtosis = 2.1), and Anxious/Depressed (skewness = 1.8, kurtosis = 5.2) were not normally distributed, groups were compared on these variables with the Kruskal–Wallis test. If groups differed, pairs were further compared using the Mann–Whitney U test. Educational level was missing for two participants—for one participant from the no-serious-offender group and one from the adult-onset group.

A multinomial logistic regression analysis was used in which all predictors were entered simultaneously and odds ratios (ORs) were computed for the prediction of the four offender types (adolescence onset, adult onset, life-course persistent, and adolescence limited) by Thought Problems, Withdrawn, Attention Problems, Anxious/Depressed, and sex. The no-serious-offender group was used as the contrast group. We controlled for educational level. Two further logistic regression analyses were conducted to compute the ORs for the adolescence-onset offender group by Thought Problems, Withdrawn, Attention Problems, and Anxious/Depressed, for boys and girls separately, controlling for educational level. ORs were not calculated for boys and girls separately in the life-course-persistent and adult-onset group, because these group sizes were too small (13 < n < 23). ORs indicate an increased risk (if OR > 1) or reduced risk (if OR < 1) for the four offender types, as compared to the nonoffenders.

Results

Predicting Offender Types

Table 1 shows the means and standard deviations of Thought Problems, Withdrawn, Attention Problems, and Anxious/Depressed of the four groups. Thought Problems, χ²(3) =13.3, p = .004, Withdrawn, χ²(3) = 12.4, p = .006, and Attention Problems, F(3, 351) = 5.3, p = .001, differed significantly between the four groups; Anxious/Depressed did not, χ²(3) = 5.9, p = .12. The mean scores on the four childhood psychopathology measures are comparable to those of healthy individuals (Verhulst et al., 1997).

Table 1.

Means, Standard Deviations, and Range on Each Childhood Psychopathology Measure, by Group.

	Adolescence Onset (n = 114)		Adult Onset (n = 30)		Life-Course Persistent (n = 38)		No Serious Offender (n = 173)
Subscale	M (SD)	Range	M (SD)	Range	M (SD)	Range	M (SD)	Range
Thought Problems	1.0 (1.3)^a	0–6	0.6 (1.1)^b	0–4	1.0 (1.1)^b,c	0–4	0.6 (0.9)^a,c	0–4
Anxious/Depressed	3.5 (3.3)	0–14	2.0 (2.0)	0–8	3.4 (4.3)	0–22	2.9 (3.1)	0–20
Withdrawn	2.4 (1.8)	0–8	1.5 (1.6)	0–5	2.3 (2.1)	0–9	1.6 (1.2)	0–6
Attention Problems	4.0 (2.7)^a,b	0–12	2.5 (1.9)^b	0–7	3.7 (2.4)	0–10	3.0 (2.4)^a	0–11

Note: Within each row, numbers with like superscripts are significantly different from each other (p < .05). Possible ranges: Thought Problems, 0–12; Anxious/Depressed, 0–28; Withdrawn, 0–18; Attention Problems, 0–22.

The multinomial logistic regression analysis showed that all childhood psychopathology variables were predictive of adolescence-onset offending. With every 1-point increase in Thought Problems, Attention Problems, and Withdrawn, the odds for offending starting in adolescence, as compared to no serious offending, increased by 1.4, 1.1, and 1.5, respectively. With every 1-point increase in Anxious/Depressed, the odds for adolescence-onset offending was reduced by 0.9. Males were 3.7 times more likely than females to start offending in adolescence.

The results with respect to the prediction of life-course-persistent and adult-onset offending are preliminary because many variables are included in the regression analyses relative to the number of participants in these two groups. None of the predictors significantly increased the odds for adult-onset offenders, as compared to no serious offending. With a 1-point increase in Thought Problems and Withdrawn, the odds to become a life-course-persistent offender (as compared to a nonoffender) increased by 1.5 and 1.4, respectively. Males were 3.9 times more likely than females to become a life-course-persistent-offender. Table 2 summarizes the results.

Table 2.

Odds Ratios Based on the Multinomial Logistic Regression Analyses Predicting Offender Type.

	Adolescence Onset (n = 114)	Adult Onset (n = 29)	Life-Course Persistent (n = 38)
Predictor	OR (95% CI)	OR (95% CI)	OR (95% CI)
Thought Problems	1.4 (1.1–1.8)^*	1.1 (0.6–1.8)	1.5 (1.0–2.1)^*
Anxious/Depressed	0.9 (0.8–1.0)^*	0.9 (0.7–1.1)	0.9 (0.8–1.0)
Attention Problems	1.1 (1.0–1.3)^*	0.9 (0.8–1.2)	1.1 (0.9–1.3)
Withdrawn	1.5 (1.2–1.8)^***	1.1 (0.7–1.5)	1.4 (1.1–1.9)^**
Educational level	1.0 (0.7–1.4)	1.2 (0.7–2.0)	0.9 (0.6–1.6)
Sex	4.3 (2.5–7.4)^***	1.7 (0.8–4.0)	3.9 (1.8–8.3)^***

Note: R² = .19 (Cox and Snell), .21 (Nagelkerke). The contrast group is the no-serious-offender group (n = 172). OR = odds ratio; CI = confidence interval.

p ≤ .05. ^**p ≤ .01. ^***p ≤ .001.

Predicting Offender Types in Boys and Girls Separately

The logistic regression analysis showed that for boys, a 1-point increase in Withdrawn increased the odds for adolescence-onset offending (n = 71), as compared to no serious offending (n = 55), by 1.9 (confidence interval [CI] = 1.3–2.8, p = .001). With every 1-point increase in Anxious/Depressed, the odds for adolescence-onset offending was reduced by 0.8 (CI = 0.7–1.0, p = .02); the no-serious-offending group was again the comparison group.

Among girls, with every 1-point increase in Withdrawn, the odds for adolescence-onset offending (n = 43), as compared to no serious offending (n = 117), increased by 1.4 (CI = 1.1–1.8, p = .02). There was a trend that with every 1-point increase in Thought Problems, the risk for adolescence-onset offending increased by 1.4 (CI = 1.0–2.0, p = .075).

Discussion

The aim of the present study was to test whether the adolescence-onset offenders differ from the adolescence-limited offenders by the presence of childhood psychopathology. Specifically, it tested whether the adolescence-onset offender, the adult-onset offender, and the life-course-persistent offender, as compared to the nonserious offender (adolescence-limited offenders and nonoffenders), can be predicted by the childhood psychopathology (Anxious/Depressed, Thought Problems, Withdrawn, and Attention Problems) reported by an individual at the age of 11 to 13 years. Based on the prospective measures from a longitudinal birth cohort, the results showed that higher levels of Thought Problems, Withdrawn, and Attention Problems increase the risk for adolescence-onset offenders, as compared to nonserious offenders; Anxious/Depressed, however, was protective of adolescence-onset offending. Furthermore, Thought Problems and Withdrawn were predictive for the life-course-persistent offenders but not for the adult-onset offenders. The finding that Thought Problems and Withdrawn can predict adolescence-onset and life-course-persistent offenders suggests that adolescence-onset offenders are more similar to life-course-persistent offenders than to adolescence-limited offenders. Therefore, the present findings suggest that even though the adolescence-onset offenders arose out of the adolescence-limited offenders, they should not be regarded as adolescence-limited offenders but rather as a distinct group with similarities to the life-course-persistent offenders. As suggested in the introduction, childhood psychopathology predicted both life-course-persistent offenders and adolescence-onset offenders. The group of adult-onset offenders was quite small. Future studies with more participants may reveal the predictive value of Thought Problems in this group to test the reliability of the present findings.

Moffitt and Caspi (2001) argue that the origins of life-course-persistent and adolescence-limited offending are explanatory across the sexes and irrespective of sex. The present study intended to test whether this is true for adolescence-onset and adult-onset offending. Unfortunately, the sample sizes of male and female life-course-persistent offenders and adult-onset offenders were too small for reliable analyses on each subgroup. The analyses were therefore confined to the adolescence-onset group, but the findings are still preliminary owing to small sample sizes. The findings show that Withdrawn is predictive of adolescence-onset offending in males and females, whereas Anxious/Depressed is protective in males only. As such, childhood psychopathology can predict male and female adolescence-onset offenders.

The Thought Problems subscale is a composite score of obsessive-compulsive symptoms, psychotic symptoms, and strange behaviors. Withdrawn behavior is also characteristic of, among other disorders, schizophrenia and schizoid personality disorder (American Psychiatric Association, 2000). Thought Problems and Withdrawn were not strongly related in the present study (r = .22). Thus, two independent psychopathology symptom clusters reported during childhood indicate that schizophreniform disorders are predictive of adolescence-onset offending and life-course-persistent offending. This finding is in line with that of Arseneault et al. (2003). They selected individuals with schizophreniform disorder from the Dunedian sample and found, similar to the present findings, that childhood psychotic symptoms predict later violence by individuals with schizophreniform disorder. But whereas Arseneault et al. found this to be true in individuals with schizophreniform disorder, we found that Thought Problems and Withdrawn were even in a general-population sample predictive of criminal behavior. Moreover, whereas Arseneault et al. showed that psychotic symptoms are predictive of violent behavior, the findings of the present study show that Thought Problems and Withdrawn are predictive of criminal behavior.

Withdrawn behavior is also related to depression and major depression (American Psychiatric Association, 2000). Previous studies have observed a relation between major depression and crimes (e.g., Hodgins, 1995; Swanson, Holzer, Ganju, & Jono, 1990). But other studies have found that depressed patients are less likely to be violent (Pearson, Wilmont, & Padi, 1986; Rossi et al., 1986). In the present study, symptoms of anxiety and depression protected one from becoming an adolescence-onset offender, a finding that was also observed by Moffitt et al. (2002). These symptoms possibly increase the odds for offending when assessed at a later age, when it is more likely that people have had experiences that induce more severe anxiety or depression symptoms (e.g., traumatic experiences).

Lay et al. (2005) found that the risk for adolescent and adult-onset offending increased when individuals were diagnosed with ADHD during childhood. The present study also showed that symptoms similar to those of ADHD (assessed as Attention Problems) increased the risk for adolescence-onset offenders. However, contrary to the findings of Lay et al., Attention Problems did not increase the risk for offending started past the age of 18 (adult-onset group). But note that the number of participants in the group of adult-onset offenders was quite small and only half of them were male. Previous studies have shown that symptoms of inattentive and overactive behaviors are more prevalent among boys than girls (e.g., Cantwell, 1996). Another explanation may be that reported Attention Problems during childhood are not stable over the years. For example, Hart et al. (1995) showed that hyperactivity–impulsivity declines with increasing age. However, criminal behavior may not be related to attention problems but rather to a third factor, which is related to attention problems and criminal behavior—that is, conduct problems. Attention problems are strongly associated with conduct problems (e.g., Anderson, Williams, McGee, & Silva, 1987). Because the Aggressive Behavior subscale of the YSR was assessed in the Zuid-Holland study, we were able to test this supposition. Aggressive Behavior was indeed related to Attention Problems (r = .60). And, indeed, when Aggressive Behavior was included in the multinomial regression analyses, Attention Problems was no longer a significant predictor of adolescence-onset offending.

Although previous studies have shown that life-course-persistent offending is related to ADHD (for a review, see Lynam, 1996), in the present study, the risk for life-course-persistent offending did not increase with an increase in Attention Problems. The Attention Problems subscale is a composite score of hyperactivity–impulsivity and inattention. Perhaps, these two factors work antagonistically to predict criminal behavior. Another possibility involves the small number of life-course-persistent offenders.

In line with the dual taxonomy model of Moffitt (1993), the present study found a life-course-persistent and an adolescence-limited group. Moffitt hypothesized that the life-course-persistent group is dominated by males, whereas the adolescence-limited group is highly represented by females. The findings of the present study agree with this hypothesis. The percentage of males among the life-course-persistent offenders was 60.5%; among the adolescence-limited offenders, 54%; and among the no-offender group, 17.4%. Whereas males dominate the adolescence-onset-group (62%), females dominate the adult-onset offender group (43% males).

The adolescence-onset group is quite large (30.4% of the total sample) and so may be regarded as an indication that the maturity gap is prolonged; that is, adulthood now starts later, as suggested by Moffitt et al. (2002). In that case, the upper age range of the adolescence-limited group should increase, possibly to the mid-20s. But adolescence-onset offenders were similar to life-course-persistent offenders in that the odds for these two offender types increased with every 1-point increase in Thought Problems and Withdrawn. Moreover, the risk for adolescence-onset offending increased by an increase in another childhood psychopathology measure—namely, Attention Problems. This finding may indicate that offending past adolescence, unlike adolescence-limited offending, is not a normal aspect of growing up but rather related to psychopathology. Thus, these findings suggest that the adolescence-limited and adolescence-onset types are two distinct offender types. In that case, the age range defining the adolescence-limited group should not increase. Thus, the findings of this study do not indicate that the maturity gap is prolonged and that adulthood presently starts later. To validate the present findings using different methodologies, future studies may study the duration of the maturity gap. Because the maturity gap refers to no or limited access to resources and status, the end of the maturity gap could be defined by graduation plus up to 2 years (an individual has access to recourses once he or she has a job and thus earns money). Another possibility would be to control for employment or income. It would be interesting to look more closely at the nonoffending males in future studies. These may be the males who earn from legal activities during adolescence, are thereby able to gain resources, and thus may not experience the maturity gap (see also Moffitt, 1993).

A limitation of the present study is the limited number of indicators of a mental disorder. Other possible indicators could be childhood traumatic experiences and drug use. The present study did not include neuropsychological deficits or environmental risk factors. So far, childhood neuropsychological deficits have been shown to be related to the life-course-persistent offender (e.g., Aguilar et al., 2000; Moffitt & Caspi, 2001). Therefore, it is unlikely that neuropsychological deficits assessed during childhood are predictive of criminal behavior that starts long beyond childhood—that is, during adolescence or adulthood. However, future studies may test whether Thought Problems assessed during childhood remain strong predictors of late-onset offending, when controlling for neuropsychological deficits and environmental risk factors assessed during adolescence. A further limitation was the small sample that was eventually included in the analyses—especially, the small number of males and females in the adult-onset group and life-course-persistent group and the small number of males in the no-offending group. Finally, another limitation was the retrospective nature of the assessment of criminal behavior. The strengths of the present study include its longitudinal character (10-13 years) and its focus on indications of mental disorders that are not, by definition, related to criminal behavior.

The findings of the present study are relevant not only for studies on the classification of offender types (e.g., life-course persistent, adolescence limited, adolescence onset, and adult onset) but also for studies on the etiology of offender types. So far, when studying the etiology of offender types, researchers have focused on neuropsychological deficits, environmental risk factors, and early aggressive and criminal behaviors. When mental disorders were studied in relation to the prediction of antisocial behavior at all, the focus was on conduct disorder, oppositional defiant disorder, and ADHD (but Lay et al., 2005). Moreover, most studies focused on the life-course-persistent and adolescence-limited offender. The present findings validate more studies on mental disorders as a predictor of a third group: the adolescence-onset offender.

The findings suggest that adolescence-limited and adolescence-onset offenders form two separate groups. Therefore, crime prevention programs should realize that these two distinct offender groups require different interventions. Offenders starting their criminal behavior during adolescence may be screened for attention problems and thought problems, to select those individuals with the worst prognoses regarding the continuing of their offending behavior. For males, intervention programs may focus on attention-related problems, and for females, they may focus on thought problems. However, more information is necessary on which aspects of attention problems and thought problems are predictive. Future studies may examine whether the prediction of the adolescence-onset individuals among the adolescence starters can be improved by combining measures of psychopathology with an outlook for transition events. Such transition events may include employment, military service, and/or relationships—that is, those events that provide opportunities for desistence from crime (Moffitt, 1993).

Regarding juvenile justice programming, sanctions for those adolescence-start offenders who are most likely to stop their criminal behavior by early adulthood—thus, adolescence-start offenders without psychopathology or indications thereof—may be confined to some form of deterring, such as community service. Education about the consequences of a criminal record, in terms of future career and family life, may also be helpful. Adolescence starters with indications of psychopathology may profit from sanctions, in combination with treatment for their psychopathology symptoms.

In sum, the findings of the present longitudinal study show that psychopathology—namely, withdrawn behavior, thought problems, and attention problems—is a precursor of adolescence-onset and life-course-persistent offending. The finding that the adolescence-onset offender is similar to the life-course-persistent offender, with respect to its precursors, lends support to the hypothesis that the adolescence-limited offender can be distinguished from the adolescence-onset offenders. Future studies may further test the validity of this hypothesis. Studies may test whether thought problems and withdrawn behavior still increase the risk for adolescence-onset offending when environmental and neuropsychological risk factors are taken into account. If so, studies may be aimed at unraveling what aspect of these symptoms increases the risk for adolescence-onset and life-course-persistent offending. The present study may then be a start for further studies on the etiology of adolescence-onset offending.

Footnotes

Author Biographies

Nicole Buck is a postdoctoral researcher at the Department of Forensic Psychiatry at Erasmus Medical Center in Rotterdam, Netherlands. Her current research relates to topics including domestic violence, cognitive processing in relation to aggression and sex offending, and aggression in people with schizophrenia.

Frank Verhulst is professor and director of child and adolescent psychiatry at Erasmus University Medical Center and Sophia Children’s Hospital, Rotterdam, Netherlands. His research relates to child psychiatric epidemiology, with a focus on longitudinal research.

Hjalmar van Marle is professor of forensic psychiatry at Erasmus University Medical Center and at the School of Law at Erasmus University, Rotterdam, Netherlands. His current research includes psychiatric determinants of violence and criminal careers—especially, psychopathy.

Jan van der Ende is a research psychologist at the Department of Child and Adolescent Psychiatry at the Erasmus Medical Center, Rotterdam, Netherlands. His research interests include the assessment of children’s emotional and behavioral problems, psychometrics, the development of antisocial behavior, the epidemiology of children’s mental health, and longitudinal studies investigating the precursors and consequences of children’s psychiatric problems.

References

Achenbach

T. M.

(1991). Manual for the Child Behavior Checklist/4-18 and 1991 profile. Burlington: University of Vermont, Department of Psychiatry.

Aguilar

Sroufe

L. A.

Egeland

Carlson

(2000). Distinguishing the early-onset/persistent and adolescence-onset antisocial behavior types; From birth to 16 years. Development and Psychopathology, 12, 109-132.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.). Washington, DC: Author.

Anderson

J. C.

Williams

S. M.

McGee

Silva

P. A.

(1987). DSM-III disorders in preadolescent children: Prevalence in a large sample from the general population. Archives of General Psychiatry, 44(1), 69-76.

Arseneault

Cannon

Murray

Poulton

Caspi

Moffitt

T. E.

(2003). Childhood origins of violent bahaviour in adults with schizophreniforms disorder. British Journal of Psychiatry, 183, 520-525.

Cantwell

D. P.

(1996). Attention deficit disorder: A review of the past 10 years. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 978-987.

De Groot

Koot

H. M.

Verhulst

F. C.

(1996). Cross-cultural generalizability of the youth self-report and teacher’s report form cross-informant syndromes. Journal of Abnormal Child Psychology, 24(5), 651-664.

Donker

A. G.

Smeenk

W. H.

van der Laan

P. H.

Verhulst

F. C.

(2003). Individual stability of antisocial behavior from childhood to adulthood: Testing the stability postulate of Moffitt’s developmental theory. Criminology, 41, 593-609.

Eggen

A. T. J.

Goudriaan

(2008). Criminaliteit en opsporing [Criminality and police investigation]. In Th

Eggen

Kalidien

S. N.

(Eds.), Criminaleit en rechtshandhaving. [Crime and law enforcement] (pp. 105-148). Boom Juridische Uitgevers, CBS, WODC.

10.

Eronen

Angermeyer

M. C.

Schulze

(1998). The psychiatric epidemiology of violent behaviour. Social Psychiatry and Psychiatric Epidemiology, 33, 13-23.

11.

Harrington

Fudge

Rutter

Pickles

Hill

(1990). Adult outcomes of childhood and adolescent depression. Archives of General Psychiatry, 47, 465-473.

12.

Hart

E. L.

Lahey

B. B.

Loeber

Applegate

Green

S. M.

Frick

P. J.

(1995). Developmental change in attention-deficit hyperactivity disorder in boys: A four-year longitudinal study. Journal of Abnormal Child Psychology, 23, 729-749.

13.

Hindelang

M. J.

Hirschi

Weis

J. G.

(1979). Correlates of delinquency: The illusion of discrepancy between self-report and official measures. American Sociological Review, 44(6), 995-1014.

14.

Hirschi

Hindelang

M. J.

Weis

J. G.

(1980). The status of self-report measures. In Klein

M. W.

Teilman

K. S.

(Eds.), Handbook of criminal justice evaluation (pp. 473-488). Beverly Hills, CA: Sage.

15.

Hodgins

(1992). Mental disorder, intellectual deficiency, and crime. Evidence from a birth cohort. Archives of General Psychiatry, 49, 473-483.

16.

Hodgins

(1995). Major mental disorder and crime: An overview. Psychology, Crime & Law, 2, 5-17.

17.

Junger-Tas

Terlouw

G. J.

Klein

M. W.

(1994). Delinquent behaviour among young people in the Western world: First results of the international self-report delinquency study. Amsterdam, Netherlands: Kugler.

18.

Kagan

Snidman

(1999). Early childhood predictors of adult anxiety disorders. Biological Psychiatry, 46, 1536-1541.

19.

Kim-Cohen

Caspi

Moffitt

T. E.

Harrington

Milne

B. J.

Poulton

(2003). Prior juvenile diagnoses in adults with mental disorders: Developmental follow-back of a prospective-longitudinal cohort. Archives of General Psychiatry, 60, 709-717.

20.

Klein

M. W.

(1989). Cross-national research in self-reported crime and delinquency. Dordrecht, Netherlands: Kluwer.

21.

Kratzer

Hodgins

(1999). A typology of offenders: A test of Moffitt’s theory among males and females from childhood to age 30. Criminal Behaviour and Mental Health, 9(1), 57-73.

22.

Lay

Ihle

Esser

Schmidt

M. H.

(2005). Juvenile-episodic, continued or adult-onset delinquency. European Journal of Criminology, 2(1), 39-66.

23.

Lynam

D. R.

(1996). Early identification of chronic offenders: Who is the fledging psychopath? Psychological Bulletin, 120(2), 209-234.

24.

Moffitt

T. E.

(1993). Adolescence-limited and life-course persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100(4), 674-701.

25.

Moffitt

T. E.

Caspi

(2001). Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. Development and Psychopathology, 13, 355-375.

26.

Moffitt

T. E.

Caspi

Dickson

Silva

Stanton

(1996). Childhood-onset versus adolescent-onset antisocial conduct problems in males: Natural history from ages 3 to 18 years. Development and Psychopathology, 5, 135-151.

27.

Moffitt

T. E.

Caspi

Harrington

Milne

B. J.

(2002). Males on the life-course-persistent and adolescence-limited antisocial pathways: Follow-up at age 26 years. Development and Psychopathology, 14, 179-207.

28.

Pearson

Wilmont

Padi

(1986). A study of violent behaviour among in-patients in a psychiatric hospital. British Journal of Psychiatry, 149, 232-235.

29.

Piquero

A. R.

White

N. A.

(2003). On the relationship between cognitive abilities and life-course-persistent offending among a sample of African Americans: A longitudinal test of Moffitt’s hypothesis. Journal of Criminal Justice, 31, 399-409.

30.

Poulton

Caspi

Moffitt

T. E.

Cannon

Murray

Harrington

H. L.

(2000). Children’s selfreported psychotic symptoms predict adult schizophreniform disorders: A 15-year longitudinal study. Archives of General Psychiatry, 57, 1053-1058.

31.

Pratt

T. C.

Cullen

F. T.

Blevins

K. R.

Daigle

Unnever

J. D.

(2002). The relationship of attention deficit hyperactivity disorder to crime and delinquency: A meta-analysis. International Journal of Police Science and Management, 4(4), 344-360.

32.

Raskin White

Bates

M. E.

Buyske

(2001). Adolescence-limited versus persistent delinquency: extending Moffitt’s hypothesis into adulthood. Journal of Abnormal Psychology, 110(4), 600-609.

33.

Redondo

Sanchez-Meca

Garrido

(1999). The influence of treatment programs on the recidivism of juvenile and adult offenders: A European meta-analytic review. Psychology, Crime and Law, 5, 251-278.

34.

Robins

L. N.

Regier

D. A.

(1991). Psychiatric disorders in America: The epidemiologic catchment area study. New York: Free Press.

35.

Rossi

A. M.

Jacobs

Monteleone

Olsen

Surber

R. W.

Winkler

E. L.

. (1986). Characteristics of psychiatric patients who engage in assaultive or other fear-inducing behavior. Journal of Nervous and Mental Disease, 174, 154-160.

36.

Rutter

(1995). Relationships between mental disorders in childhood and adulthood. Acta Psychiatrica Scandinavica, 91, 73-85.

37.

Scholte

E. M.

(1999). Factors predicting continued violence into young adulthood, Journal of Adolescence, 22, 3-20.

38.

Sutterer

Karger

(1994). Self-reported juvenile delinquency in Mannheim, Germany. In Junger-Tas

Terlouw

G.-J.

Klein

M. W.

(Eds.), Delinquent behavior among young people in the Western world (pp. 156-185). Amsterdam, Netherlands: Kugler.

39.

Swanson

J. W.

Holzer

C. E.

Ganju

V. K.

Jono

R. T.

(1990). Violence and psychiatric disorders in the community: Evidence from the epidemiological catchment area surveys. Hospital and Community Psychiatry, 41, 761-770.

40.

Verhulst

F. C.

van der Ende

(1992). Six-year stability of parent-reported problem behavior in an epidemiological sample. Journal of Abnormal Child Psychology, 20, 595-610.

41.

Verhulst

F. C.

van der Ende

Koot

H. M.

(1997). Handleiding voor de Youth Self- Report (YSR) [Manual for the Youth Self-Report (YSR)]. Rotterdam, Netherlands: Erasmus University, Department of Child and Adolescent Psychiatry, Sophia Children’s Hospital.

42.

Vermeiren

(2003). Psychopathology and delinquency in adolescents: A descriptive and developmental perspective. Clinical Psychology Review, 23, 277-318.

43.

Windle

R. C.

Windle

(1995). Longitudinal patterns of physical aggression: Associations with adult social, psychiatric, and personality functioning and testosterone levels. Development and Psychopathology, 7, 563-585.