and that both forms of stability are characteristic of a relatively small group of persons with extremely antisocial behavior.

Thus, in defiance of regression to the mean, a group of extremely antisocial persons remain extreme on measures taken at later ages and in different situations. Among other persons, however, temporary and situational manifestations of antisocial behavior (even to severe levels) may be quite common.

This point is vividly illustrated in a longitudinal investigation of a representative cohort of 1,037 New Zealand children born in 1972‑1973. In this sample, I compared the base rates of persistent and temporary antisocial behavior problems (Moffitt, 1991). I identified a group of boys whose antisocial behavior was rated above average at each of seven biennial assessments (ages 3, 5, 7, 9, 11, 13, and 15). The boys were also rated as very antisocial by three different reporting agents (parents, teachers, and self). Five percent of the boys in the sample met these selection criteria. As a group, their mean antisocial ratings were more than a standard deviation above the norm for boys at every age. In contrast, fully two thirds of the remaining boys were rated above average on antisocial checklists as well but at only one or two ages or by only one reporter, illustrating that stability cannot be inferred from cross‑sectional measures of extremity (Henry, Moffitt, Robins, Earls, & Silva, 1993). A disproportionate amount of the measured stability in the New Zealand sample could be attributed to the 5% of boys whose antisocial behavior was both extreme and consistent. For example, when these few boys were excluded from calculations, the 8‑year stability coefficient for teacher ratings was reduced from .28 (R² = .078) to .16 (R² = .025), indicating that 5% of the sample accounted for 68% of the sample's stability. (If antisocial behavior had been a stable characteristic throughout the sample, with most boys retaining their relative standing in the group across time, then excluding the top 5% of the sample should not have affected the stability coefficient.) In summary, there appear to be noteworthy individual differences in the stability of antisocial behavior.

I have already alluded to the small number of persons in the general population whose antisocial behavior is life‑course‑persistent. In fact, epidemiological research has shown that there is remarkable uniformity in the prevalence rates of different manifestations of severe antisocial behavior: Regardless of their age, under 10% of males warrant an "official" antisocial designation. For example, about 5% of preschool boys are considered by their parents or caretakers to be "very difficult to manage" (McGee, Partridge, Williams, & Silva, 1991). The prevalence of conduct disorder among elementary‑school‑aged boys has been found to be between 4% and 9% in several countries (Costello, 1989; Rutter, Tizard, & Whitmore, 1970). About 6% of boys are first arrested by police as preteens (Moffitt & Silva, 1988c; Wolfgang et al., 1972); such early arrest is important because it is the best predictor of long‑term recidivistic offending. The rate of conviction for a violent offense in young adult males is between 3% and 6% (Moffitt, Mednick, & Gabrielli, 1989), and about 4% of male adolescents self‑report sustained careers of serious violence (three or more violent offenses per year for 5 years; Elliott, Huizinga, & Morse, 1986). Finally, the prevalence of men with antisocial personality disorder is estimated at about 4% to 5% (Davison & Neale, 1990; Robins, 1985).

It is possible, of course, that the persons who constitute these epidemiological statistics at different ages are all different individuals. However, the longitudinal data suggest otherwise: It is more likely that the remarkable constancy of prevalence rates reflects the reoccurrence of the same life‑course‑persistent individuals in different antisocial categories at different ages. Robins (1966, 1978) has shown that there are virtually no subjects with adult antisocial personality disorder who did not also have conduct disorder as children. White, Moffitt, Earls, Robins, and Silva (1990) found notable continuity from disobedient and aggressive behavior at age 3 to later childhood conduct disorder and thence to arrest by police in the early teen years. Loeber (1982) reviewed research that pinpoints a first arrest between ages 7 and 11 as particularly important for predicting long‑term adult offending. Hare and McPherson (1984) have reported that a conviction for violence in the early 20s is characteristic of almost all men who become diagnosed with antisocial (psychopathic) personality disorder.

There are still gaps in the epidemiological database; each of the earlier cited studies connected only two or three points in the life course. Nonetheless, the consistency is impressive: A substantial body of longitudinal research consistently points to a very small group of males who display high rates of antisocial behavior across time and in diverse situations. The professional nomenclature may change, but the faces remain the same as they drift through successive systems aimed at curbing their deviance: schools, juvenile‑justice programs, psychiatric treatment centers, and prisons. The topography of their behavior may change with changing opportunities, but the underlying disposition persists throughout the life course.

Whereas a few males evidence antisocial behavior that emerges in toddlerhood and is persistent thereafter, the majority of boys who become antisocial first do so during adolescence (Elliott, Knowles, & Canter, 1981). This tidal wave of adolescent onset has been studied in the aforementioned representative sample of New Zealand boys (Moffitt, 1991). Between ages 11 and 15, about one third of the sample joined the delinquent lifestyles of the 5% of boys who had shown stable and pervasive antisocial behavior since preschool. As a group, these adolescent newcomers to antisocial ways had not formerly exceeded the normative levels of antisocial behavior for boys at ages 3, 5, 7, 9, or 11. Despite their lack of prior experience, by age 15, the newcomers equaled their preschool‑onset antisocial peers in the variety of laws they had broken, the frequency with which they broke them, and the number of times they appeared in juvenile court (Moffitt, 1991). On the basis of such common1y used indexes, of adolescent delinquency, the two delinquent groups were indistinguishable. Thus, if the sample was viewed only as an adolescent cross section, researchers would lose sight of the two delinquent groups' very different developmental histories, seeing only delinquents and nondelinquents.

Indeed, researchers and practitioners cannot yet effectively assign individual delinquent adolescents to meaningful subtypes on the basis of cross‑sectional "snapshots" of their antisocial behavior during adolescence (Loeber & LeBlanc, 1990; Moffitt, 1990a). Again, the New Zealand sample provides an example: At age 15, both the childhood‑persistent and adolescent‑onset groups had members who scored more than 5 standard deviations above the mean on self‑report delinquency, and by age 19 both groups had some members with more than 50 convictions for crimes in the New Zealand courts. Elliott and Huizinga (1984) reported similarly poor classification in a representative sample of American teens. They attempted to discriminate, at the time of first arrest, individual future career offenders from adolescence‑limited offenders. Discrimination could not be improved beyond chance by entering the kinds of information typically available to officials: type of current offense, age, sex, race, class, involvement with delinquent peers, and attitudes toward deviance. Addition of measures of the extremity of self‑reported delinquency and emotional problems improved prediction only 7% beyond chance. Earlier, I noted that the stability of antisocial behavior implies its extremity but that extremity does not imply stability, measures of the frequency or seriousness of adolescent offending will not discriminate very well between life‑course‑persistent and adolescence-limited delinquents. On the basis of their study and others, Elliott and Huizinga concluded that there is "no effective means for discriminating between the serious career offenders and nonserious offenders" (p. 98). A notable feature of the taxonomy introduced in this article is that knowledge of a subject's preadolescent behavior is required for making the differential diagnosis between the life‑course‑persistent and adolescence-limited types of antisocial teenager. Longitudinal designs are needed to collect the lifetime repeated measures that are needed to distinguish individual differences in the developmental course of antisocial behavior.[i]

I have argued in this section that juvenile delinquency conceals two categories of people. A very large group participates in antisocial behavior during adolescence. A much smaller group, who continues serious antisocial behavior throughout adulthood, is the same group whose antisocial behavior was stable across the years from early childhood. The categories remain hypothetical types, because no longitudinal study has yet repeatedly measured antisocial behavior in a representative sample of the same individuals from preschool to midlife. I describe in the next sections the two hypothetical types of antisocial youth: life‑course‑persistent and adolescence‑limited. I argue that the two groups differ in etiology, developmental course, prognosis, and, importantly, classification of their behavior as either pathological or normative. The goal of this article is to proffer a description of the two types in the form of a set of testable predictions.

Life‑Course‑Persistent Antisocial Behavior

My account of the life‑course‑persistent antisocial type follows this plan: In the first section, Continuity of Antisocial Behavior Defined, I provide a definition and description of persistent antisocial behavior. In the second section, Beginnings: Neuropsychological Risk for Difficult Temperament and Behavioral Problems, I present the hypothesis that persistent antisocial behavior has its origins in an interaction between children's neuropsychological vulnerabilities and criminogenic environments. In the third section, Maintenance and Elaboration Over the Life Course: Cumulative Continuity Contemporary Continuity, and Narrowing Options for Change, I introduce the cumulative and contemporary processes that maintain antisocial behavior across time and that expand antisocial behavior into a pervasive adult life‑style. In the fourth section, I summarize the theory's perspective on continuity, and in the fifth section, I make a case that life‑course‑persistent antisocial behavior is a form of psychopathology.

Continuity of Antisocial Behavior Defined

As implied by the label, continuity is the hallmark of the small group of life‑course‑persistent antisocial persons. Across, the life course, these individuals exhibit changing manifestations of antisocial behavior: biting and hitting at age 4, shoplifting and truancy at age 10, selling drugs and stealing cars at age 16, robbery and rape at age 22, and fraud and child abuse at age 30; the underlying disposition remains the same, but its expression changes form as new social opportunities arise at different points in development. This pattern of continuity across age is matched also by cross‑situational consistency: Life‑course‑persistent antisocial persons lie at home, steal from shops, cheat at school, fight in bars, and embezzle at work (Farrington, 1991; Loeber, 1982; Loeber & Baicker‑McKee, 1989; Robins, 1966, 1978; White et al., 1990).

The concept of behavioral coherence, or heterotypic continuity, is invoked here to extend observations of continuity beyond the mere persistence of a single behavior to encompass a variety of antisocial expressions that emerge as development affords new opportunities. Heterotypic continuity refers to continuity of an inferred trait or attribute that is presumed to underlie diverse phenotypic behaviors (Kagan, 1969). As Kagan and Moss (1962) suggested, a specific behavior in childhood might not be predictive of phenotypically similar behavior later in adulthood, but it may still be associated with behaviors that are conceptually consistent with the earlier behavior.

Examples of heterotypic continuities have been reported by Ryder (1967), who found that childhood aggression, physical adventurousness, and nonconformity were related to adult sexual behavior. Another example of coherence is provided in a 22-year follow‑up study of men and women who had been rated as aggressive by their peers in late childhood (Huesmann, Eron, Lefkowitz, & Walder, 1984). As adults, the men were likely to commit serious criminal acts, abuse their spouses, and drive while intoxicated, whereas the women were likely to punish their offspring severely. Another example of personality coherence is the finding that the developmental antecedents of erratic work histories may be found in phenotypically dissimilar attributes of difficult temperament in childhood (Caspi, Elder, & Bem, 1987). In addition, in their hallmark study, West and Farrington (1977) observed that stealing, alcohol abuse, sexual promiscuity, reckless driving, and violence were linked across the life course. The prognosis for the life‑course‑persistent person is bleak: Drug and alcohol addiction: unsatisfactory employment, unpaid debts; homelessness; drunk driving; violent assault; multiple and unstable relationships; spouse battery; abandoned, neglected, or abused children; and psychiatric illness have all been reported at very high rates for offenders who persist past the age of 25 (Farrington & West, 1990; Robins, 1966; Sampson & Laub, 1990). Thus, this theory of life‑course-persistent antisocial behavior predicts continuity across the entire life course but allows that the underlying disposition will change its manifestation when age and social circumstances alter opportunities.

Although reports of the continuity of antisocial styles from childhood to young adulthood abound, the outcomes of antisocial individuals during midlife have seldom been examined. The pattern of official crime over age (Figure 1) implies that criminal offending all but disappears by midlife,[ii] but there is no reason to expect that life‑course‑persistents miraculously assume prosocial tendencies after an antisocial tenure of several decades. Indeed, criminal psychopaths decrease their number of arrestable offenses at about age 40, but the constellation of antisocial personality traits described by Cleckley (1976) persists in male samples at least until age 69 (Harpur & Hare, 1991). As I argue in the third section of this article (Maintenance), an analysis of the cumulative developmental forces underlying the continuity of aggression from childhood to adulthood will predict continuity on into midlife as well. Beyond young adulthood, the antisocial disposition of life‑course‑persistents may be expressed in a form that is simply not yet well measured by epidemiological surveys of official crime: One such possibility is neglect and abuse of family members. Consistent with this hypothesis, Farrington and West (1990) found that half of the persistent offenders in the Cambridge longitudinal study self‑reported having hit their spouses when they were interviewed at age 32. Fagan and Wexler (1987) reviewed studies showing that spouse battery is often preceded by a history of violence against strangers. Also, crime statistics show that, whereas property crimes peak in the teen years and drop thereafter, family violence offenses show a steady increase with age (Gottfredson & Hirschi, 1986). Research is needed that follows offenders into late adulthood while measuring multiple indicators of an antisocial life‑style.

Beginnings: Neuropsychological Risk for Difficult

Temperament and Behavioral Problems

If some individuals' antisocial behavior is stable from preschool to adulthood as the data imply, then investigators are compelled to look for its roots early in life, in factors that are present before or soon after birth. It is possible that the etiological chain begins with some factor capable of producing individual differences in the neuropsychological functions of the infant nervous system. Factors that influence infant neural development are myriad, and many of them have been empirically linked to antisocial outcomes.

One possible source of neuropsychological variation that is linked to problem behavior is disruption in the ontogenesis of the fetal brain. Minor physical anomalies, which are thought to be observable markers for hidden anomalies in neural development, have been found at elevated rates among violent offenders and subjects with antisocial personality traits (Fogel, Mednick, & Michelson, 1985; E. Kandel, Brennan, & Mednick, 1989; Paulhus & Martin, 1986). Neural development may be disrupted by maternal drug abuse, poor prenatal nutrition, or pre-or postnatal exposure to toxic agents (Needleman & Beringer, 1981; Rodning, Beckwith, & Howard, 1989; Stewart, 1983). Even brain insult suffered because of complications during delivery has been empirically linked to later violence and antisocial behavior in carefully designed longitudinal studies (E. Kandel & Mednick, 1991; Szatmari, Reitsma‑Street, & Offord, 1986). In addition, some individual differences in neuropsychological health are heritable in origin (Borecki & Ashton, 1984; Martin, Jardine, & Eaves, 1984; Plomin, Nitz, & Rowe, 1990; Tambs, Sundet, & Magnus. 1984: Vandenberg, 1969). Just as parents and children share facial resemblances, they share some structural and functional similarities within their nervous systems. After birth, neural development may be disrupted by neonatal deprivation of nutrition, stimulation, and even affection (Cravioto & Arrieta, 1983; Kraemer, 1988; Meany, Aitken, van Berkel, Bhatnagar, & Sapolsky, 1988). Some studies have pointed to child abuse and neglect as possible sources of brain injury in the histories of delinquents with neuropsychological impairment (Lewis, Shanok, Pincus, & Glaser, 1979; Milner & McCanne, 1991; Tarter, Hegedus, Winsten, & Alterman, 1984).

There is good evidence that children who ultimately become persistently antisocial do suffer from deficits in neuropsychological abilities. I have elsewhere reviewed the available empirical and theoretical literatures; the link between neuropsychological impairment and antisocial outcomes is one of the most robust effects in the study of antisocial behavior (Moffitt, 1990b; Moffitt & Henry, 1991; see also Hirschi & Hindelang, 1977). Two sorts of neuropsychological deficits are empirically associated with antisocial behavior: verbal and "executive" functions. The verbal deficits of antisocial children are pervasive, affecting receptive listening and reading, problem solving, expressive speech and writing, and memory. In addition, executive deficits produce what is sometimes referred to as a comportmental learning disability (Price, Daffner, Stowe, & Mesulam, 1990), including symptoms such as inattention and impulsivity. These cognitive deficits and antisocial behavior share variance that is independent of social class, race, test motivation, and academic attainment (Moffitt, 1990b; Lynam, Moffitt, & Stouthamer-Loeber, 1993). In addition, the relation is not an artifact of slowwitted delinquents' greater susceptibility to detection by police; undetected delinquents have weak cognitive skills too (Moffitt & Silva, 1988a).

The evidence is strong that neuropsychological deficits are linked to the kind of antisocial behavior that begins in childhood and is sustained for lengthy periods. In a series of articles (Moffitt, 1990a; Moffitt & Henry, 1989; Moffitt & Silva, 1988b), I have shown that poor verbal and executive functions are associated with antisocial behavior, if it is extreme and persistent. In these studies, adolescent New Zealand boys who exhibited symptoms of both conduct disorder and attention‑deficit disorder with hyperactivity (ADDH) scored very poorly on neuropsychological tests of verbal and executive functions and had histories of extreme antisocial behavior that persisted from age 3 to age 15. Apparently, their neuropsychological deficits were as long standing as their antisocial behavior; at ages 3 and 5 these boys had scored more than a standard deviation below the age norm for boys on the Bayley and McCarthy tests of motor coordination and on the Stanford‑Binet test of cognitive performance. Contrast groups of boys with single diagnoses of either conduct disorder or ADDH did not have neuropsychological deficits or cognitive‑motor delays, but neither were their behavior problems stable over time.

In a study designed to improve on measurement of executive functions (White, Moffitt, Caspi, Jeglum, Needles, & Stouthamer‑Loeber, in press), we gathered data on self‑control and impulsivity for 430 Pittsburgh youths. Twelve measures were taken from multiple sources (mother, teacher, self, and observer) by using multiple methods (rating scales, performance tests, computer games, Q sorts, and videotaped observations). A linear composite of the impulsivity measures was strongly related to the 3‑year longevity of antisocial behavior, even after controlling for IQ, race, and social class. Boys who were very delinquent from ages 10 to 13 scored significantly higher on impulsivity than both their nondelinquent and temporarily delinquent age‑mates. Taken together, the New Zealand and Pittsburgh longitudinal studies suggest that neuropsychological dysfunctions that manifest themselves as poor scores on tests of language and self‑control‑-and as the inattentive, overactive, and impulsive symptoms of ADDH‑-are linked with the early childhood emergence of aggressive antisocial behavior and with its subsequent persistence.

Neuropsychological variation and the "difficult " infant.

Before describing how neuropsychological variation might constitute risk for antisocial behavior, it is useful to define what is meant here by neuropsychological. By combining neuro with psychological, I refer broadly to the extent to which anatomical structures and physiological processes within the nervous system influence psychological characteristics such as temperament, behavioral development, cognitive abilities, or all three. For example, individual variation in brain function may engender differences between children in activity level, emotional reactivity, or self‑regulation (temperament); speech, motor coordination, or impulse control (behavioral development); and attention, language, learning, memory, or reasoning (cognitive abilities).

Children with neurological difficulties severe enough to constitute autism, severe physical handicap, or profound mental retardation are usually identified and specially treated by parents and professionals. However, other infants have subclinical levels of problems that affect the difficulty of rearing them, variously referred to as difficult temperament, language or motor delays, or mild cognitive deficits. Compromised neuropsychological functions are associated with a variety of consequences for infants' cognitive and motor development as well as for their personality development (Rothbart & Derryberry, 1981). Toddlers with subtle neuropsychological deficits may be clumsy and awkward, overactive, inattentive, irritable, impulsive, hard to keep on schedule, delayed in reaching developmental milestones, poor at verbal comprehension, deficient at expressing themselves, or slow at learning new things (Rutter, 1977, 1983; Thomas & Chess, 1977; Wender, 1971).

Hertzig (1983) has described an empirical test of the proposed relationship between neurological damage and difficult behavior in infancy. She studied a sample of 66 low‑birth‑weight infants from intact middle‑class families. Symptoms of brain dysfunction detected during neurological examinations were significantly related to an index of difficult temperament taken at ages 1, 2, and 3 (Thomas & Chess, 1977; the index comprised rhythmicity, adaptability, approach‑withdrawal, intensity, and mood). The parents of the children with neurological impairment and difficult temperament more often sought help from child psychiatrists as their children grew up, and the most frequent presenting complaints were immaturity, overactivity, temper tantrums, poor attention, and poor school performance. Each of these childhood problems has been linked by research to later antisocial outcomes (cf. Moffitt, 1990a, 1990b). Importantly, the impairments of the children with neural damage were not massive; their mean IQ score was 96 (only 4 points below the population mean). Hertzig's study showed that even subtle neurological deficits can influence an infant's temperament and behavior, the difficulty of rearing the infant, and behavioral problems in later childhood.

Child‑environment covariation in nature: A source of interactional continuity. Up to this point, I have emphasized in this article the characteristics of the developing child as if environments were held constant. Unfortunately, children with cognitive and temperamental disadvantages are not generally born into supportive environments, nor do they even get a fair chance of being randomly assigned to good or bad environments. Unlike the aforementioned infants in Hertzig's (1983) study of temperament and neurological symptoms, most low‑birth weight infants are not born into intact, middle‑class families.

Vulnerable infants are disproportionately found in environments that will not be ameliorative because many sources of neural maldevelopment co‑occur with family disadvantage or deviance.

Indeed, because some characteristics of parents and children tend to be correlated, parents of children who are at risk for antisocial behavior often inadvertently provide their children with criminogenic environments (Sameroff & Chandler, 1975). The intergenerational transmission of severe antisocial behavior has been carefully documented in a study of three generations (Huesmann et al., 1984). In that study of 600 subjects, the stability of individuals' aggressive behavior from age 8 to age 30 was exceeded by the stability of aggression across the generations: from grandparent to parent to child. Thus, with regard to risk for antisocial behavior, nature does not follow a 2 X 2 design with equal cell sizes.

Parents and children resemble each other on temperament and personality. Thus, parents of children who are difficult to manage often lack the necessary psychological and physical resources to cope constructively with a difficult child (Scarr & McCartney, 1983; Snyder & Patterson, 1987). For example, temperamental traits such as activity level and irritability are known to be partly heritable (Plomin, Chipuer, & Loehlin, 1990). This suggests that children whose hyperactivity and angry outbursts might be curbed by firm discipline will tend to have parents who are inconsistent disciplinarians; the parents tend to be impatient and irritable too. The converse is also true: Empirical evidence has been found for a relationship between variations in parents' warmth and infants' easiness (Plomin, Chipuer, & Loehlin, 1990).

Parents and children also resemble each other on cognitive ability. The known heritability of measured intelligence (Plomin, 1990; Loehlin, 1989) implies that children who are most in need of remedial cognitive stimulation will have parents who may be least able to provide it. Moreover, parents' cognitive abilities set limits on their own educational and occupational attainment (Barrett & Depinet, 1991). As one consequence, families whose members have below‑average cognitive capacities will often be least able financially to obtain professional interventions or optimal remedial schooling for their at‑risk children.

Even the social and structural aspects of the environment may be stacked against children who enter the world at risk. Plomin and Bergeman (1990) have shown that there are genetic components to measures that are commonly used by developmental psychologists to assess socialization environments. For example, the Home Observation for Measurement of the Environment scale, the Moos Family Environment scales, and the

Holmes and Rahe scales of stressful life events all revealed the influence of heritable factors when they were examined with behavior genetic research designs (Plomin & Bergeman, 1990). Vulnerable children are often subject to adverse homes and neighborhoods because their parents are vulnerable to problems too (cf. Lahey et al., 1990).

Importantly, although examples from behavior genetics research have been cited in the previous three paragraphs, the perverse compounding of children's vulnerabilities with their families' imperfections does not require that the child's neuropsychological risk arise from any genetic disposition. In fact, for my purposes, it is immaterial whether parent‑child similarities arise from shared genes or shared homes. A home environment wherein prenatal care is haphazard, drugs are used during pregnancy, and infants' nutritional needs are neglected is a setting where sources of children's neuropsychological dysfunction that are clearly environmental coexist with a criminogenic social environment.

Problem child‑problem parent interactions and the emergence of antisocial behaviors. I believe that the juxtaposition of a vulnerable and difficult infant with an adverse rearing context initiates risk for the life‑course‑persistent pattern of antisocial behavior. The ensuing process is a transactional one in which the challenge of coping with a difficult child evokes a chain of failed parent‑child encounters (Sameroff & Chandler, 1975). The assertion that children exert important effects on their social environments is useful in understanding this hypothetical process (Bell & Chapman, 1986). It is now widely acknowledged that personality and behavior are shaped in large measure by interactions between the person and the environment (cf. Buss, 1987; Plomin, DeFries, & Loehlin, 1977; Scarr & McCartney, 1983). One form of interaction may play a particularly important role both in promoting an antisocial style and in maintaining its continuity across the life course: Evocative interaction occurs when a child's behavior evokes distinctive responses from others (Caspi et al., 1987).

Children with neuropsychological problems evoke a challenge to even the most resourceful, loving, and patient families. For example, Tinsley and Parke (1983) have reviewed literature showing that low‑birth‑weight, premature infants negatively influence the behavior of their caretakers; they arrive before parents are prepared, their crying patterns are rated as more disturbing and irritating, and parents report that they are less satisfying to feed, less pleasant to hold, and more demanding to care for than healthy babies. Many parents of preterm infants hold unrealistic expectations about their children's attainment of developmental milestones, and these may contribute to later dysfunctional parent‑child relationships (Tinsley & Parke, 1983). More disturbing, an infant's neurological health status has been shown to be related to risk for maltreatment and neglect (Friedrich & Boriskin, 1976: Frodi et al., 1978; Hunter, Kilstrom, Kraybill, & Loda, 1978; Milowe & Lowrie, 1964; Sandgrund, Gaines, & Green, 1974).

Numerous studies have shown that a toddler's problem behaviors may affect the parents' disciplinary strategies as well as subsequent interactions with adults and peers (Bell & Chapman, 1986; Chess & Thomas, 1987). For example, children characterized by a difficult temperament in infancy are more likely to resist their mothers' efforts to control them in early childhood (Lee & Bates, 1985). Similarly, mothers of difficult boys experience more problems in their efforts to socialize their children. Maccoby and Jacklin (1983) showed that over time these mothers reduce their efforts to actively guide and direct their children's behavior and become increasingly less involved in the teaching process. In a study of unrelated mothers and children, K. E. Anderson, Lytton, and Romney (1986) observed conduct‑disordered and nonproblem boys interacting with mothers of conduct‑disordered and nonproblem sons in unrelated pairs. The conduct‑disordered boys evoked more negative reactions from both types of mothers than did normal boys, but the two types of mothers did not differ from each other in their negative reactions. It may well be that early behavioral difficulties contribute to the development of persistent antisocial behavior by evoking responses from the interpersonal social environment, responses that exacerbate the child's tendencies (Goldsmith, Bradshaw, & Rieser‑Danner, 1986; Lytton, 1990). "The child acts; the environment reacts; and the child reacts back in mutually interlocking evocative interaction" (Caspi et al., 1987, p. 308).

Such a sequence of interactions would be most likely to produce lasting antisocial behavior problems if caretaker reactions were more likely to exacerbate than to ameliorate children's problem behavior. To my knowledge, students of child effects have not yet tested for interactions between child behavior and parental deviance or poor parenting, perhaps because very disadvantaged families are seldom studied with such designs. Nonetheless, some data suggest that children's predispositions toward antisocial behavior may be exacerbated under deviant rearing conditions. In the New Zealand longitudinal study, there was a significant interaction effect between children's neuropsychological deficit and family adversity on one type of delinquent act; aggressive confrontation with a victim or adversary. Among the 536 boys in the sample, the 75 boys who had both low neuropsychological test scores and adverse home environments earned a mean aggression score more than four times greater than that of boys with either neuropsychological problems or adverse homes (Moffitt, 1990b). The index of family adversity included parental characteristics such as poor mental health and low intelligence as well as socioeconomic status. Behaviorgenetic adoption studies of antisocial behavior often report a similar pattern of findings, wherein the highest rates of criminal outcomes are found for adoptees whose foster parents, as well as their biological parents, were deviant (e.g., Mednick, Gabrielli, & Hutchings, 1984). Thus, children's predispositions may evoke exacerbating responses from the environment and may also render them more vulnerable to criminogenic environments.

If the child who "steps off on the wrong foot" remains on an ill‑starred path, subsequent stepping‑stone experiences may culminate in life‑course‑persistent antisocial behavior. For life-course‑persistent antisocial individuals, deviant behavior patterns later in life may thus reflect early individual differences that are perpetuated or exacerbated by interactions with the social environment: first at home, and later at school. Quay (1987) summarized this as "this youth is likely to be at odds with everyone in the environment. and most particularly with those who must interact with him on a daily basis to raise, educate, or otherwise control him. . . . This pattern is the most troublesome to society, seems least amenable to change, and has the most pessimistic prognosis for adult adjustment" (p. 12 1).

However, inauspicious beginnings do not complete the story. In the New Zealand study, for example, a combination of preschool measures of antisocial behavior and cognitive ability was able to predict 70% of the cases of conduct disorder at age 11 but at the cost of a high false‑positive rate (White et al., 1990). The next section explores the specific interactional processes that nourish and augment the life‑course‑persistent antisocial style beyond childhood.

Maintenance and Elaboration Over the Life Course:

Cumulative Continuity, Contemporary Continuity, and

Narrowing Options for Change

In the previous section, the concept of evocative person‑environment interaction was called on to describe how children's difficult behaviors might affect encounters with their parents. Two additional types of interaction may help to explain how the life‑course‑ persistent individual's problem behavior, once initiated, might promote its own continuity and pervasiveness. Reactive interaction occurs when different youngsters exposed to the same environment experience it, interpret it, and react to it in accordance with their particular style. For example, in interpersonal situations where cues are ambiguous, aggressive children are likely to mistakenly attribute harmful intent to others and then act accordingly (Dodge & Frame, 1982). Proactive interaction occurs when people select or create environments that support their styles. For example, antisocial individuals appear to be likely to affiliate selectively with antisocial others, even when selecting a mate. Some evidence points to nonrandom mating along personality traits related to antisocial behavior (Buss, 1984), and there are significant spouse correlations on conviction for crimes (e.g., Baker, Mack, Moffitt, & Medruck, 1989).

The three types of person‑environment interactions can produce two kinds of consequences in the life course: cumulative consequences and contemporary consequences (Caspi & Bern, 1990). Early individual differences may set in motion a downhill snowball of cumulative continuities. In addition, individual differences may themselves persist from infancy to adulthood, continuing to influence adolescent and adult behavior in a proximal contemporary fashion. Contemporary continuity arises if the life‑course‑persistent person continues to carry into adulthood the same underlying constellation of traits that got him into trouble as a child, such as high activity level, irritability, poor self‑control, and low cognitive ability.

The roles of cumulative and contemporary continuities in antisocial behavior have been explored by Caspi, Bern, and Elder (1989; Caspi et at., 1987), using data from the longitudinal Berkeley Guidance Study. They identified men who had a history of temper tantrums during late childhood (when tantrums are not developmentally normative). Then they traced the continuities and consequences of this personality style across the subsequent 30 years of the subjects' lives and into multiple diverse life domains: education, employment, and marriage. A major finding was that hot‑tempered boys who came from middle‑class homes suffered a progressive deterioration of socioeconomic status as they moved through the life course. By age 40, their occupational status was indistinguishable from that of men born into the working class. A majority of them held jobs of lower occupational status than those held by their fathers at a comparable age. Did these men fail occupationally because their earlier ill‑temperedness started them down a particular path (cumulative consequences) or because their current ill-temperedness handicapped them in the world of work (contemporary consequences)?

Cumulative consequences were implied by the effect of childhood temper on occupational status at midlife: Tantrums predicted lower educational attainment, and educational attainment, in turn, predicted lower occupational status. Contemporary consequences were implied by the strong direct link between ill‑temperedness and occupational stability. Men with childhood tantrums continued to be hot‑tempered in adulthood, where it got them into trouble in the world of work. They had more erratic work lives, changing jobs more frequently and experiencing more unemployment between ages 18 and 40. Ill-temperedness also had a contemporary effect on marital stability. Almost half (46%) of the men with histories of childhood tantrums had divorced by age 40 compared with only 22% of other men.

Elsewhere, I describe in detail some of the patterns of interaction between persons and their social environments that may promote antisocial continuity across time and across life domains (Caspi & Moffitt, in press‑b). Two sources of continuity deserve emphasis here because they narrow the options for change. These processes are (a) failing to learn conventional prosocial alternatives to antisocial behavior and (b) becoming ensnared in a deviant life‑style by crime's consequences. These concepts have special implications for the questions of why life-course‑persistent individuals fail to desist from delinquency as young adults and why they are so impervious to intervention.

A restricted behavioral repertoire. This theory of life-course‑persistent antisocial behavior asserts that the causal sequence begins very early and the formative years are dominated by chains of cumulative and contemporary continuity. As a consequence, little opportunity is afforded for the life‑course-persistent antisocial individual to learn a behavioral repertoire of prosocial alternatives. Thus, one overlooked and pernicious source of continuity in antisocial behavior is simply a lack of recourse to any other options. In keeping with this prediction, Vitaro, Gagnon, and Tremblay (1990) have shown that aggressive children whose behavioral repertoires consist almost solely of antisocial behaviors are less likely to change over years than are aggressive children whose repertoires comprise some prosocial behaviors as well.

Life‑course‑persistent persons miss out on opportunities to acquire and practice prosocial alternatives at each stage of development. Children with poor self‑control and aggressive behavior are often rejected by peers and adults (Coie, Belding, & Underwood, 1988; Dodge, Coie, & Brakke, 1982; Vitaro et al., 1990). In turn, children who have learned to expect rejection are likely in later settings to withdraw or strike out preemptively, precluding opportunities to affiliate with prosocial peers (Dodge & Newman, 1981; Dodge & Frame, 1982; LaFrenier & Sroufe, 1985; Nasby, Hayden, & DePaulo, 1980). Such children are robbed of chances to practice conventional social skills. Alternatively, consider this sequence of narrowing options: Behavior problems at school and failure to attain basic math and reading skills place a limit on the variety of job skills that can be acquired and thereby cut off options to pursue legitimate employment as an alternative to the underground economy (Farrington, Gallagher, Morley, Ledger, & West, 1986; Maughan, Gray, & Rutter, 1985; Moffitt, 1990a). Simply put, if social and academic skills are not mastered in childhood, it is very difficult to later recover lost opportunities.

Becoming ensnared by consequences of antisocial behavior. Personal characteristics such as poor self‑control, impulsivity, and inability to delay gratification increase the risk that antisocial youngsters will make irrevocable decisions that close the doors of opportunity. Teenaged parenthood, addiction to drugs or alcohol, school dropout, disabling or disfiguring injuries, patchy work histories, and time spent incarcerated are snares that diminish the probabilities of later success by eliminating opportunities for breaking the chain of cumulative continuity (Cairns & Cairns, 1991; J, Q. Wilson & Herrnstein, 1985). Similarly, labels accrued early in life can foreclose later opportunities; an early arrest record or a "bad" reputation may rule out lucrative jobs, higher education, or an advantageous marriage (Farrington, 1977; Klein, 1986; West, 1982). In short, the behavior of life‑course‑persistent antisocial persons is increasingly maintained and supported by narrowing options for conventional behavior.

Interventions with life‑course‑persistent persons have met with dismal results (Lipton, Martinson, & Wilks, 1975; Palmer, 1984; Sechrest, White, & Brown, 1979). This is not surprising, considering that most interventions are begun relatively late in the chain of cumulative continuity. The forces of continuity are formidable foes (Caspi & Moffitt, in press‑a). After a protracted deficient learning history, and after options for change have been eliminated, efforts to suppress antisocial behavior will not automatically bring prosocial behavior to the surface in its place. Now‑classic research on learning shows conclusively that efforts to extinguish undesirable behavior will fail unless alternative behaviors are available that will attract reinforcement (Azrin & Holz, 1966). My analysis of increasingly restricted behavioral options suggests the hypothesis that opportunities for change will often be actively transformed by life‑course‑persistents into opportunities for continuity: Residential treatment programs provide a chance to learn from criminal peers, a new job furnishes the chance to steal, and new romance provides a partner for abuse. This analysis of life‑course‑persistent antisocial behavior anticipates disappointing outcomes when such antisocial persons are thrust into new situations that purportedly offer the chance "to turn over a new leaf."

The Reason for Persistence: Traits, Environments, and

Developmental Processes

According to some accounts of behavioral continuity, an ever-present underlying trait generates antisocial outcomes at every point in the life span (e.g., Gottfredson & Hirschi, 1990). By other accounts, antisocial behavior is sustained by environmental barriers to change (e.g., Bandura, 1979, pp. 217‑224). In this theory of life‑course‑persistent antisocial behavior, neither traits nor environments account for continuity.

True, the theory begins with a trait: variation between individuals in neuropsychological health. The trait is truly underlying in that it seldom comes to anyone's attention unless an infant is challenged by formal examinations; it is manifested behaviorally as variability in infant temperament, developmental milestones, and cognitive abilities.

Next, the theory brings environments into play. Parents and other people respond to children's difficult temperaments and developmental deficits. In nurturing environments, toddlers' problems are often corrected. However, in disadvantaged homes, schools, and neighborhoods, the responses are more likely to exacerbate than amend. Under such detrimental circumstances, difficult behavior is gradually elaborated into conduct problems and a dearth of prosocial skills. Thus, over the years, an antisocial personality is slowly and insidiously constructed. Likewise, deficits in language and reasoning are incrementally elaborated into academic failure and a dearth of job skills. Over time, accumulating consequences of the youngster's personality problems and academic problems prune away the options for change.

This theory of life‑course‑persistent antisocial behavior emphasizes the constant process of reciprocal interaction between personal traits and environmental reactions to them. The original attribute is thus elaborated on during development, to become a syndrome that remains conceptually consistent, but that gains new behavioral components (Caspi & Bern, 1990). Through that process, relatively subtle childhood variations in neuropsychological health can be transformed into an antisocial style that pervades all domains of adolescent and adult behavior. It is this infiltration of the antisocial disposition into the multiple domains of a life that diminishes the likelihood of change.

When in the life course does the potential for change dwindle to nil? How many person‑environment interactions must accumulate before the life‑course‑persistent pattern becomes set? I have argued that a person‑environment interaction process is needed to predict emerging antisocial behavior, but after some age will the "person" main effect predict adult outcomes alone? An answer to these questions is critical for prevention efforts. The well‑documented resistance of antisocial personality disorder to treatments of all kinds seems to suggest that the life-course‑persistent style is fixed sometime before age 18 (Suedfeld & Landon, 1978). Studies of crime careers reveal that it is very unusual for males to first initiate crime after adolescence, suggesting that if an adult is going to be antisocial, the pattern must be established by late adolescence (Elliott, Huizinga, & Menard, 1989).[iii] At the same time, efforts to predict antisocial outcomes from childhood conduct problems yield many errors (e.g., White et al., 1990). These errors seem to suggest that antisocial styles become set sometime after childhood.

Unfortunately, the extant longitudinal database does not provide a sound basis for conclusions. Typically, childhood behavior problems are assessed at only one time point from a single source, thereby lumping the many children who are temporarily or situationally aggressive with the few children who are on a persistent and pervasive trajectory. Outcomes are also typically assessed at a single point, often during late adolescence when temporary delinquents and future persisters are lumped together. According to my theory, such predictive designs should yield large numbers of false positives and false negatives. Analyses should ask, when between preschool and late adolescence can stable‑pervasive antisocial behavior problems best predict antisocial outcomes among adults?

Life-Course‑Persistent Antisocial Behavior as

Psychopathology

The life‑course‑persistent antisocial syndrome, as described here, has many characteristics that, taken together, suggest psychopathology. For example, the syndrome is statistically unusual; much research converges to suggest that it is characteristic of about 5% of males (Robins, 1985). Its rarity is thus consistent with a simple statistical definition of abnormality.

The theoretical syndrome is also characterized by tenacious stability across time and in diverse circumstances. This high probability response style is relied on even in situations where it is clearly inappropriate or disadvantageous (Caspi & Moffitt, in press‑b), especially if there is a very limited repertoire of alternative conventional behaviors (Tremblay, 1991). Life‑course-persistent antisocial behavior is thus maladaptive in the sense that it fails to change in response to changing circumstances.

The syndrome of life‑course‑persistent antisocial behavior described here has a biological basis in subtle dysfunctions of the nervous system (Moffitt, 1990b). (I reiterate my assertion that biological origins are in no way deterministic. Rather, individual variations in nervous system health provide raw material for subsequent person‑environment interactions.)

The syndrome is associated with other mental disorders. There is good evidence that such "comorbidity" is associated with long‑term continuity. An impressive body of research documents an overlap between persistent forms of antisocial behavior and other conditions of childhood such as learning disabilities and hyperactivity (cf. Moffitt, 1990a). Three studies (Elliott, Huizinga, & Menard, 1989; Farrington, Loeber, & Van Kammen, 1990; Moffitt, 1990a) have now shown that the presence of multiple behavioral disorders predicts persistence of illegal behavior over the course of years. This proliferation of mental disorders is common among life‑course‑persistent antisocial persons. For example, in the Epidemiological Catchment Area (ECA) study of mental disorders among 19,000 adults, over 90% of the cases with antisocial personality disorder had at least one additional psychiatric diagnosis. (Evidence of onset before adulthood is required for the diagnosis of antisocial personality disorder, confirming persistence in the ECA cases.) The comorbid conditions that disproportionately affected antisocial adults were mania, schizophrenia, drug and alcohol abuse, depression, and anxiety disorders (Robins & Regier, 1991).

Of course, no one or two of these parameters is enough to warrant the classification of life‑course‑persistent antisocial behavior as psychopathology. Nonetheless, when taken together they form a more persuasive argument that persons whose antisocial behavior is stable and pervasive over the life course may constitute a category that is distinct from persons whose antisocial behavior is short term and situational.

Adolescence‑ Limited Antisocial Behavior

My account of the adolescence‑limited antisocial type will follow this plan: In the first section, Discontinuity: The Most Common Course of Antisocial Behavior, I provide a definition and description of this ubiquitous form of antisocial behavior. In the second section, Beginnings: Motivation, Mimicry, and Reinforcement, I present three etiological hypotheses. Adolescence‑limited antisocial behavior is motivated by the gap between biological maturity and social maturity, it is learned from antisocial models who are easily mimicked, and it is sustained according to the reinforcement principles of learning theory. In the third section, I answer the question, Why doesn't every teenager become delinquent? In the fourth section, Desistence From Crime: Adolescence‑Limiteds Are Responsive to Shifting Reinforcement Contingencies, I explain how temporary delinquents come to be exempted from the processes of continuity. In the fifth section, Adolescence‑Limited Delinquency and Secular Change, I locate adolescence‑limited delinquency in its recent historical context. In the sixth section, I make a case that the antisocial behavior of adolescence‑limited delinquents is best regarded as adaptive social behavior.

Discontinuity: The Most Common Course of Antisocial

Behavior

As implied by the proffered label, discontinuity is the hallmark of teenaged delinquents who have no notable history of antisocial behavior in childhood and little future for such behavior in adulthood. However, the brief tenure of their delinquency should not obscure their prevalence in the population or the gravity of their crimes. In contrast with the rare life‑course-persistent type, adolescence‑limited delinquency is ubiquitous. Several studies have shown that about one third of males are arrested during their lifetime for a serious criminal offense, whereas fully four fifths of males have police contact for some minor infringement (Farrington, Ohlin, & Wilson, 1986). Most of these police contacts are made during the adolescent year. Indeed, numerous rigorous self‑report studies have now documented that it is statistically aberrant to refrain from crime dur‑

ing adolescence (Elliott et al., 1983; Hirschi, 1969; Moffitt & Silva, 1988c).

Compared with the life‑course‑persistent type, adolescence-limited delinquents show relatively little continuity in their antisocial behavior. Across age, change in delinquent involvement is often abrupt, especially during the periods of onset and desistence. For example, in my aforementioned longitudinal study of a representative sample of boys, 12% of the youngsters were classified as new delinquents at age 13; they had no prior history of antisocial behavior from age 5 to age 11. Between age 11 and age 13, they changed from below the sample average to 1.5 standard deviations above average on self‑reported delinquency (Moffitt, 1990a). By age 15, another 20% of this sample of boys had joined the newcomers to delinquency despite having no prior history of antisocial behavior (Moffitt, 1991). Barely into mid‑adolescence, the prevalence rate of markedly antisocial boys had swollen from 5% at age 11 to 32% at age 15. When interviewed at age 18, only 7% of the boys denied all delinquent activities. By their mid‑20s, at least three fourths of these new offenders are expected to cease all offending (Farrington, 1986).

Adolescence‑limited delinquents may also have sporadic, crime‑free periods in the midst of their brief crime "careers." Also, in contrast with the life‑course‑persistent type, they lack consistency in their antisocial behavior across situations. For example, they may shoplift in stores and use drugs with friends but continue to obey the rules at school. Because of the chimeric nature of their delinquency, different reporters (such as self, parent, and teacher) are less likely to agree about their behavior problems when asked to complete rating scales or clinical interviews (Loeber, Green, Lahey, & Stouthamer‑Loeber, 1990; Loeber & Schmaling, 1985).

These observations about temporal instability and cross‑situational inconsistency are more than merely descriptive. They have implications for a theory of the etiology of adolescence-limited delinquency. Indeed, the flexibility of most delinquents' behavior suggests that their engagement in deviant life‑styles may be under the control of reinforcement and punishment contingencies.

Unlike their life‑course‑persistent peers, whose behavior was described as inflexible and refractory to changing circumstances, adolescence‑limited delinquents are likely to engage in antisocial behavior in situations where such responses seem profitable to them, but they are also able to abandon antisocial behavior when prosocial styles are more rewarding. They maintain control over their antisocial responses and use antisocial behavior only in situations where it may serve an instrumental function. Thus, principles of learning theory will be important for this theory of the cause of adolescence‑limited delinquency.

A theory of adolescence‑limited delinquency must account for several empirical observations: modal onset in early adolescence, recovery by young adulthood, widespread prevalence, and lack of continuity. Why do youngsters with no history of behavior problems in childhood suddenly become antisocial in adolescence? Why do they develop antisocial problems rather than other difficulties? Why is delinquency so common among teens? How are they able to spontaneously recover from an antisocial life‑style within a few short years?

Just as the childhood onset of life‑course‑persistent persons compelled me to look for causal factors early in their lives, the coincidence of puberty with the rise in the prevalence of delinquent behavior compels me to look for clues in adolescent development. Critical features of this developmental period are variability in biological age, the increasing importance of peer relationships, and the budding of teenagers' self‑conscious values, attitudes, and aspirations. These developmental tasks form the building blocks for a theory of adolescence‑limited delinquency.

Beginnings: Motivation, Mimicry, and Reinforcement

Why do adolescence‑limited delinquents begin delinquency? The answer advanced here is that their delinquency is "social mimicry" of the antisocial style of life‑course‑persistent youths. The concept of social mimicry is borrowed from ethology. Social mimicry occurs when two animal species share a single niche and one of the species has cornered the market on a resource that is needed to promote fitness (Moynihan, 1968). In such circumstances, the "mimic" species adopts the social behavior of the more successful species to obtain access to the valuable resource. For example, cowbird chicks, who are left by their mothers to be reared in the nests of unsuspecting parent birds, learn to behave like the parent birds' own true chicks and thus stimulate the parents to drop food their way. Social mimicry may also allow some species to safely pass among a more successful group and thus share access to desired resources. For example, some monkey species have learned to mimic bird calls. One such species of monkeys, rufous‑naped tamarins, is able to share the delights of ripe fruit after a tree has been located by tyrant flycatchers, whose superior avian capacities in flight and distance vision better equip them to discover bearing trees. Similarly, zebras are sensitive to the social signals of impalas and gazelles and thus benefit from the latter species' superior sensitivity to approaching predators (E. 0. Wilson, 1975).

If social mimicry is to explain why adolescence‑limited delinquents begin to mimic the antisocial behavior of their life-course‑persistent peers, then, logically, delinquency must be a social behavior that allows access to some desirable resource. I suggest that the resource is mature status, with its consequent power and privilege.

Before modernization, biological maturity came at a later age, social adult status arrived at an earlier age, and rites of passage more clearly delineated the point at which youths assumed new roles and responsibilities. In the past century, improved nutrition and health care have decreased the age of biological maturity at the rate of three tenths of a year per decade (Tanner, 1978; Wyshak & Frisch, 1982). Simultaneously, modernization of work has delayed the age of labor‑force participation to ever later points in development (Empey, 1978; Horan & Hargis, 1991; Panel on Youth of the President's Science Advisory Committee, 1974). Thus, secular changes in health and work have lengthened the duration of adolescence. The ensuing gap leaves modern teenagers in a 5‑ to 10‑year role vacuum (Erikson, 1960). They are biologically capable and compelled to be sexual beings, yet they are asked to delay most of the positive aspects of adult life (see Buchanan, Eccles, & Becker, 1992, for a review of studies of the compelling influence of pubertal hormones of teens' behavior and personality). In most American states, teens are not allowed to work or get a driver's license before age 16, marry or vote before age 18, or buy alcohol before age 21, and they are admonished to delay having children and establishing their own private dwellings until their education is completed at age 22, sometimes more than 10 years after they attain sexual maturity. They remain financially and socially dependent on their families of origin and are allowed few decisions of any real import. Yet they want desperately to establish intimate bonds with the opposite sex, to accrue material belongings, to make their own decisions, and to be regarded as consequential by adults (Csikszentmihalyi & Larson, 1984). Contemporary adolescents are thus trapped in a maturity gap, chronological hostages of a time warp between biological age and social age.

This emergent phenomenology begins to color the world for most teens in the first years of adolescence. Steinberg has shown that, between ages 10 and 15, a dramatic shift in youngsters' self‑perceptions of autonomy and self‑reliance takes place. Moreover, the timing of the shift for individuals is connected with their pubertal maturation (Steinberg, 1987; Steinberg & Silverberg, 1986; Udry, 1988). At the time of biological maturity, salient pubertal changes make the remoteness of ascribed social maturity painfully apparent to teens. This new awareness coincides with their promotion into a high school society that is numerically dominated by older youth. Thus, just as teens begin to feel the discomfort of the maturity gap, they enter a social reference group that has endured the gap for 3 to 4 years and has already perfected some delinquent ways of coping with it. Indeed, several researchers have noted that this life‑course transition into high school society may place teens at risk for anti-social behavior. In particular, exposure to peer models, when coupled with puberty, is an important determinant of adolescence‑onset cases of delinquency (Caspi, Lynam, Moffitt, & Silva, 1993; Magnusson, 1988; Simmons & Blyth, 1987).

Life‑course‑persistent youngsters are the vanguard of this transition. Healthy adolescents are capable of noticing that the few life‑course‑persistent youths in their midst do not seem to suffer much from the maturity gap. (At a prevalence rate of about 5%, one or two such experienced delinquents in every classroom might be expected.) Already adept at deviance, life-course‑persistent youths are able to obtain possessions by theft or vice that are otherwise inaccessible to teens who have no independent incomes (e.g., cars, clothes, drugs, or entry into adults‑only leisure settings). Life‑course‑persistent boys are more sexually experienced and have already initiated relationships with the opposite sex.[iv] Life‑course‑persistent boys appear relatively free of their families of origin: they seem to go their own way, making their own rules. As evidence that they make their own decisions, they take risks and do dangerous things that parents could not possibly endorse. As evidence that they have social consequence in the adult world, they have personal attorneys, social workers, and probation officers; they operate small businesses in the underground economy; and they have fathered children (Weiher, Huizinga, Lizotte, & Van Kammen, 1991). Viewed from within contemporary adolescent culture, the anti-social precocity of life‑course‑persistent youths becomes a coveted social asset (cf. Finnegan, 1990a, 1990b; Jessor & Jessor, 1977; Silbereisen & Noack. 1988). Like the aforementioned bird calls that were mimicked by hungry tamarin monkeys, antisocial behavior becomes a valuable technique that is demonstrated by life‑course‑persistents and imitated carefully by adolescence‑limiteds. The effect of peer delinquency on the onset of delinquency is among the most robust facts in criminology research (Elliott & Menard, in press; Jessor & Jessor, 1977; Reiss, 1986; Sarnecki, 1986). However, is there evidence consistent with a social mimicry interpretation? I describe the evidence in the next section.

Social mimicry and the relationships between life‑course‑persistent and adolescence‑limited delinquents. One hypothesized by‑product of the maturity gap is a shift during early adolescence by persistent antisocial youth from peripheral to more influential positions in the peer social structure. This shift should occur as aspects of the antisocial style become more interesting to other teens. In terms of its epidemiology, delinquent participation shifts from being primarily an individual psychopathology in childhood to a normative group social behavior during adolescence and then back to psychopathology in adulthood. Consider that the behavior problems of the few pioneering antisocial children in an age cohort must develop on an individual basis; such early childhood pioneers lack the influence of delinquent peers (excepting family members). However, near adolescence, a few boys join the life‑course‑persistent ones, then a few more, until a critical mass is reached when almost all adolescents are involved in some delinquency with age peers. Elliott and Menard (in press) have analyzed change in peer group membership from age 11 to age 24 in a national probability sample. Their data show a gradual population drift from membership in nondelinquent peer groups to membership in delinquent peer groups up to age 17: the trend reverses thereafter. For example, 78% of 11‑year‑olds reported no or minimal delinquency among their friends. In contrast, 66% of 17‑year‑olds reported substantial delinquency on the part of the friends in their group.

The word friends in the previous sentence seems to imply a personal relationship between life‑course‑persistents and adolescence‑limiteds that is implausible. Much evidence suggests that, before adolescence, life‑course‑persistent antisocial children are ignored and rejected by other children because of their unpredictable, aggressive behavior (Coie et al., 1988; Dodge et al., 1982). After adolescence has passed, life‑course‑persistent adults are often described as lacking the capacity for loyalty or friendship (Cleckley, 1976; Robins, 1985). At first, these observations may seem contrary to my assertion that life‑course‑persistents assume social influence over youths who admire and emulate their style during adolescence. However, it is important to recall that social mimicry required no exchange of affection between the successful birds and their monkey mimics. In this theory, adolescents who wish to prove their maturity need only notice that the style of life‑course‑persistents resembles adulthood more than it resembles childhood. Then they need only observe antisocial behavior closely enough and long enough to imitate it successfully. What is contended is that adolescence‑limited youths should regard life‑course‑persistent youths as models, and life‑course‑persistent teens should regard themselves as magnets for other teens. Neither perception need involve reciprocal liking between individuals.

A modeling role would imply that measures of exposure to delinquent peers (e.g., knowledge of their delinquent behavior or time spent in proximity to them) should be better predictors of self‑delinquency than measures of relationship quality (e.g., shared attitudes or attachment to delinquent peers). Few studies have parsed peer‑delinquency effects into separate components, but two findings consistent with this prediction have been reported from the National Youth Survey, a representative sample of more than 1,500 teens. Agnew (1991) examined relationship characteristics in interaction with levels of peer delinquency. He argued that attachment to peers should encourage deviance if peers are delinquent but discourage it if they are not. Agnew's results showed that such interaction terms were good predictors. However, the results also showed that time spent with delinquent peers was a stronger unique predictor of self‑delinquency than the interaction between peer attachment and peer crime. Warr and Stafford (1991) found that the knowledge of friends' delinquent behavior was 2.5 to 5 times more important for self‑delinquency than friends' attitudes about delinquency. (This pattern has been replicated in another sample by Nagin & Paternoster, 1991.) Moreover, the effect of peer delinquency was direct; it was not mediated by influencing the respondents' attitudes to be more like those of deviant peers. These findings are not consistent with the notion that teens take up delinquency after pro‑delinquency attitudes are transferred in the context of intimate social relations. Rather, Warr and Stafford concluded that the data on peer effects are best interpreted in terms of imitation or vicarious reinforcement.

A magnet role would imply that children who were rejected and ignored by others should experience newfound "popularity" as teens, relative to their former rejected status. That is, life-course‑persistent youth should encounter more contacts with peers during adolescence when other adolescents draw near so as to imitate their life‑style. Some research is consistent with this interpretation. For example, in a study of 450 students in middle school, aggressive youths who were rejected by their peers reported that they did not feel lonely, whereas submissive rejected youths did feel lonely (Parkhurst & Asher, 1992). Similarly, aggressive seventh‑graders in the Carolina Longitudinal Study were rated as popular as often as nonaggressive youths by both teachers and themselves and were as likely as other youths to be nuclear members of peer groups (Cairns, Cairns, Neckerman, Gest, & Gariepy, 1988). In their review of peer‑relationship studies, Coie, Dodge, and Kupersmidt (1990) noted that the relationship between overt aggression and peer rejection is weaker or absent in adolescent samples compared with child samples. Findings such as these suggest that aggressive teens experience regular contacts with peers, however short‑lived. Similarly, in the Oregon Youth Study, rejection by peers at age 10 was prognostic of greater involvement with delinquent peers 2 years later (Dishion, Patterson, Stoolmiller, & Skinner, 1991). Although the Oregon researchers interpreted their results as suggesting that aggressive children seek delinquent friends, their data are equally consistent with my interpretation that aggressive youths begin to serve as a magnet for novice delinquents during early adolescence. Definitive sociometric research must follow up aggressive‑rejected children to test whether they develop networks in adolescence that include late‑onset delinquents of the adolescence‑limited type.

Researchers from the Carolina Longitudinal Study have carefully documented that boys with an aggressive history do participate in peer networks in adolescence but that the networks are not very stable (Cairns et al., 1988). Consistent with a social mimicry hypothesis, delinquent groups have frequent membership turnover. In addition, the interchanges between network members are characterized by much reciprocal antisocial behavior (Cairns et al., 1988). Reiss and Farrington (1991) have shown that the most experienced high‑rate young offenders tend to recruit different co‑offenders for each offense.

Life‑course‑persistents serve as core members of revolving networks, by virtue of being role models or trainers for new recruits (Reiss, 1986). They exploit peers as drug customers, as fences, as lookouts, or as sexual partners. Such interactions among life‑course‑persistent and adolescence‑limited delinquents may represent a symbiosis of mutual exploitation. Alternatively, life‑course‑persistent offenders need not even be aware of all of the adolescence‑limited youngsters who imitate their style. Unlike adolescence‑limited offenders, who appear to need peer support for crime, life‑course‑persistent offenders are willing to offend alone (Knight & West, 1975). The point is that the phenomena of "delinquent peer networks" and "co‑offending" during the adolescent period do not necessarily connote supportive friendships that are based on intimacy, trust, and loyalty, as is sometimes assumed. Social mimicry of delinquency can take place if experienced offenders actively educate new recruits. However, it can also take place if motivated learners merely observe antisocial models from afar.

Reinforcement of de1inquency by its "negative" consequences. For teens who become adolescence‑limited delinquents, antisocial behavior is an effective means of knifing‑off childhood apron strings and of proving that they can act independently to conquer new challenges (Erikson, 1960). Hypothetical reinforcers for delinquency include damaging the quality of intimacy and communication with parents, provoking responses from adults in positions of authority, finding ways to look older (such as by smoking cigarettes, being tattooed, playing the big spender with ill‑gotten gains), and tempting fate (risking pregnancy, driving while intoxicated, or shoplifting under the noses of clerks). None of these putative reinforcers may seem very pleasurable to the middle‑aged academic, but each of the aforementioned consequences is a precious resource to the teenager and can serve to reinforce delinquency. Bloch and Niederhoffer (1958) have offered an anthropological perspective: "It is almost as if the contemporary young person, in the absence of puberty rituals and ordeals, is moved to exclaim: If you don't care to test us, then we will test ourselves!" (p. 28).

I suggest that every curfew violated, car stolen, drug taken, and baby conceived is a statement of personal independence and thus a reinforcer for delinquent involvement. Ethnographic interviews with delinquents reveal that proving maturity and autonomy are strong personal motives for offending (e.g., Goldstein, 1990). Such hypothetical reinforcing properties have not been systematically tested for most types of delinquent acts. However, epidemiological studies have confirmed that adolescent initiation of tobacco, alcohol, and drug abuse are reinforced because they symbolize independence and maturity to youth (D. Kandel, 1980; Mausner & Platt, 1971).

In summary, in this narrative account of the etiology of adolescent‑onset delinquency I have emphasized three conditions: motivation, mimicry, and reinforcement. I have suggested that a secular change in the duration of adolescence has generated an age‑dependent motivational state. In addition, life‑course-persistent antisocial models must be available so that their delinquent behaviors can be imitated. Finally, adolescents' fledgling attempts to mimic antisocial styles will continue if they are socially reinforced by the "negative consequences" of crime.

Why Doesn't Every Teenager Become Delinquent?

The proffered theory of adolescence‑limited delinquency regards this sort of delinquency as an adaptive response to contextual circumstances. As a consequence, the theory seems to predict that every teen will engage in delinquency. Data from epidemiological studies using the self‑report method suggest that almost all adolescents do commit some illegal acts (Elliott et al., 1983). In addition, even studies using official records of arrest by police find surprisingly high prevalence rates (for a review see Farrington, Ohlin, & Wilson, 1986). Nevertheless, some youths commit less delinquency than others, and a small minority abstains completely. Unfortunately, almost no research sheds light on the characteristics of teens who abstain from antisocial behavior altogether. Speculations are thus ill-informed by empirical observations. However, some predictions may be derived from the present theory of adolescence-limited delinquency. The predictions center on two theoretical prerequisites for adolescent‑onset delinquency: the motivating maturity gap and antisocial role models. Some youths may skip the maturity gap because of late puberty or early initiation into adult roles. Others may find few opportunities for mimicking life‑course‑persistent delinquent models.

Some youths who refrain from antisocial behavior may, for some reason, not sense the maturity gap and therefore lack the hypothesized motivation for experimenting with crime. Perhaps such teens experience very late puberty so that the gap between biological and social adulthood is not signaled to them early in adolescence. For example, Caspi and Moffitt (1991) have shown that girls who do not menstruate by age 15 tend not to become involved in delinquency; in fact they evidence fewer than normal behavior problems as teens. Perhaps other abstainers belong to cultural or religious subgroups in which adolescents are given legitimate access to adult privileges and accountability. In his vivid ethnographic account of "old heads" and teenaged boys in a poor black neighborhood, Anderson (1990) described how mature community leaders drew certain boys into their own work and social lives, deliberately and publicly initiating the boys into manhood (and preventing delinquent involvement).

Some nondelinquent teens may lack structural opportunities for modeling antisocial peers. Adolescent crime rates are generally lower in rural areas than in inner‑city areas (Skogan, 1979, 1990). Teens in urban areas are surrounded by a greater density of age peers (and have readier unsupervised access to them through public transportation and meeting venues such as parks and shopping malls) than are teens in relatively isolated rural areas. For instance, Sampson and Groves (1989) determined that the strongest community‑level correlate of local rates of robbery and violence was the presence of "unsupervised groups of teenagers hanging out and making a nuisance" (p. 789). In that study, more traditional community correlates of crime, such as socioeconomic status, residential mobility, and ethnicity, were mediated by the teenaged social scene. School structures may also constrain or facilitate access to life-course‑persistent models. Caspi et al. (1993) found that early puberty was associated with delinquency in girls but only if they had access to boys through attending coed high schools. Girls who were enrolled in girls' schools did not engage in delinquency. In that study, the difference in delinquent involvement between coed and single‑sex school settings could not be explained by any personal or family characteristics that may have influenced how the girls came to be enrolled in their schools; access to delinquent role models was clearly the best explanation for the girls' behavior problems.

Youths may also be excluded from opportunities to mimic antisocial peers because of some personal characteristics that make them unattractive to other teens or that leave them reluctant to seek entry to newly popular delinquent groups. Shedler and Block (1990) found such an effect on the use of illegal drugs. They compared the personality styles of three adolescent groups: teens who abstained from trying any drug, teens who experimented with drugs, and teens who were frequent heavy drug users. Adolescents who experimented were the best adjusted teens in the sample. As expected, frequent users were troubled teens who were alienated and antisocial. However, the abstainers were also problem teens: They were "relatively tense, overcontrolled, emotionally constricted,. . . sornewhat socially isolated and lacking in interpersonal skills" (p. 618). This personality style was not a consequence of failing to try drugs. Rather, it was an enduring personality configuration. At age 7, these abstainers had been prospectively described by raters as "overcontrolled, timid, fearful and morose . . . , they were not warm and responsive, not curious and open to new experience, not active, not vital, and not cheerful" (pp. 619‑620). Similarly, Farrington and West (1990) reported that boys from criminogenic circumstances who did not become delinquent seemed nervous and withdrawn and had few or no friends. These provocative findings remind us that deviance is defined in relationship to its normative context. During adolescence, when delinquent behavior becomes the norm, nondelinquents warrant our scientific scrutiny.

In summary, this theory of adolescence‑limited delinquency suggests that adolescents who commit no antisocial behavior at all have either (a) delayed puberty, (b) access to roles that are respected by adults, (c) environments that limit opportunities

for learning about delinquency, (d) personal characteristics that exclude them from antisocial peer networks, or (e) all four. Research is needed to determine whether or not abstaining from delinquency is necessarily a sign of good adolescent adjustment.

Desistence From Crime: Adolescence‑Limiteds A re Responsive to Shifting Reinforcement Contingencies

By definition, adolescence‑limited delinquents generally do not maintain their delinquent behavior into adulthood. The account of life‑course‑persistent persons I made earlier in this article required an analysis of maintenance factors. In contrast, this account of adolescence‑limited delinquents demands an analysis of desistence: Why do adolescence‑limited delinquents desist from delinquency? This theory's answer: Healthy youths respond adaptively to changing contingencies. If motivational and learning mechanisms initiate and maintain their delinquency, then, likewise, changing contingencies can extinguish it.

Preoccupied with explaining the origins of crime, most theories of delinquency have neglected to address the massive shift in the prevalence of criminal involvement between adolescence and adulthood. Gove (1985) reviewed six of the most influential theories of deviance: labeling theory, conflict theory, differential association theory, control theory, anomie theory, and functional theory. He concluded, "All of these theoretical perspectives either explicitly or implicitly suggest that deviant behavior is an amplifying process that leads to further and more serious deviance" (p. 118). A general application of an amplifying process to all delinquency is inconsistent with the empirical observation that desistence from crime is the normative pattern.

Waning motivation and shifting contingencies. In contrast with amplifying theories, the present maturity‑gap theory does anticipate desistence. With the inevitable progression of chronological age, more legitimate and tangible adult roles become available to teens. Adolescence‑limited delinquents gradually experience a loss of motivation for delinquency as they exit the maturity gap. Moreover, when aging delinquents attain some of the privileges they coveted as teens, the consequences of illegal behavior shift from rewarding to punishing, in their perception. An adult arrest record will limit their job opportunities, drug abuse keeps them from getting to work on time, drunk driving is costly, and bar fights lead to accusations of unfit parenthood. Adolescence‑limited delinquents have something to lose by persisting in their antisocial behavior beyond the teen years.

There is some evidence that many young adult offenders weigh the relative rewards from illegal and conventional activities when they contemplate future offending. In a study of three samples, the effect of age on criminal participation was mediated by young men's expectations about whether illegal earnings would exceed earnings from a straight job (Piliavin, Thornton, Gartner, & Matsueda, 1986). Important for this theory, research shows that "commitment costs" are among the factors weighed by young adults when they decide to discontinue offending. In the criminological subfield of perceptual deterrence research, commitment costs are defined as a person's judgment that past accomplishments will be jeopardized or that future goals will be foreclosed (Williams & Hawkins, 1986). Criminal behavior incurs commitment costs if it risks informal sanctions (disapproval by family, community. or employer) as well as formal sanctions (arrest or conviction penalty). Given that very few delinquent acts culminate in formal sanctions, perceptual deterrence theories consider informal sanctions as keys to deterrence. Paternoster and colleagues have tested the proposed effects of commitment costs and informal sanctions in a follow‑up study of 300 young adults. They found that criminal offending 1 year later was best predicted by prospective indexes of commitment costs (r = ‑.23) and informal sanctions (r = ‑.40). Those variables outdid gender, perceived risk of arrest, grade point average, and peer attachment (Paternoster, Saltzman, Waldo, & Chiricos, 1983).[v]

Options for change. Consistent with this motivational analysis, the antisocial behavior of many delinquent teens has been found to decline after they leave high school (Elliott & Voss, 1974), join the army (Eider. 1986; Mattick, 1960), marry a prosocial spouse (Sampson & Laub, 1990), move away from the old neighborhood (West, 1982), or get a full‑time job (Sampson & Laub, 1990). As these citations show, links between the assumption of adult roles and criminal desistence have been observed before. The issue left unaddressed by theory is why are some delinquents able to desist when others are not? What enables adolescence‑limited delinquents to make these (often abrupt) transitions away from crime? Why do adolescence‑limited delinquents come to realize that they have something to lose, whereas life‑course‑persistent delinquents remain undeterred? Here, two positions are advanced: Unlike their life‑course‑persistent counterparts, adolescence‑limited delinquents are relatively exempt from the forces of (a) cumulative and (b) contemporary continuity.

First, without a lifelong history of antisocial behavior, the forces of cumulative continuity have had fewer years in which to gather the momentum of a downhill snowball. Before taking up delinquency, adolescence‑limited offenders had ample years to develop an accomplished repertoire of prosocial behaviors and basic academic skills. These social skills and academic achievements make them eligible for postsecondary education, good marriages, and desirable jobs.

The availability of alternatives to crime may explain why some adolescence‑limited delinquents desist later than others. (As shown in Figure 1, the desistence portion of the age‑crime curve slopes more gradually than the abrupt criminal initiation portion.) Although the forces of cumulative continuity build up less momentum over the course of their relatively short crime careers, many adolescence‑limited youths will fall prey to many of the same snares that maintain continuity among life‑course-persistent persons. Those whose teen forays into delinquency inadvertently attracted damaging consequences may have more difficulty desisting. A drug habit, an incarceration, interrupted education, or a teen pregnancy are snares that require extra effort and time from which to escape. Thus, this theory predicts that variability in age at desistence from crime should be accounted for by the cumulative number and type of ensnaring life events that entangle persons in a deviant life‑style.

Second, in stark contrast with the earlier account of life-course‑persistent offenders, personality disorder and cognitive deficits play no part in the delinquency of adolescence‑limited offenders. As a result, they are exempt from the sources of contemporary continuity that plague their life‑course‑persistent counterparts. In general, these young adults have adequate social skills, they have a record of average or better academic achievement, their mental health is sturdy, they still possess the capacity to forge close attachment relationships, and they retain the good intelligence they had when they entered adolescence. One study of girls who grew up in institutional care has illustrated that individual differences influence which adolescents are able to attain prosocial outcomes in young adulthood (Quinton & Rutter, 1988). In that study, some girls reared in institutions were able to escape adversity for advantage through marriage to a supportive husband, but a constellation of individual psychological attributes determined which girls were able to marry well.

At the crossroads of young adulthood, adolescence‑limited and life‑course‑persistent delinquents go different ways. This happens because the developmental histories and personal traits of adolescence‑limiteds allow them the option of exploring new life pathways. The histories and traits of life‑course‑persistents have foreclosed their options, entrenching them in the antisocial path. To test this hypothesis. research must examine conditional effects of individual histories on opportunities for desistence from crime.

Adolescence‑Limited Delinquency and Secular Change

I have suggested that adolescence‑limited delinquency is a byproduct of modernization, an adolescent adaptation to a maturity gap engendered by the opposing social forces of improved health and a smaller, better educated work force. If this theory is correct, then secular changes should have rendered the age-crime curve relatively steeper with increasing modernization. The theory predicts that, in contemporary preindustrial nations and in earlier historical periods, the age‑crime curve should have a flatter kurtosis; in other words, it will lack the characteristic sharp peak between the ages of 15‑18.

Empirical data support this prediction. Greenberg (1985) compared crime statistics from the mid‑1800s to 1980s in the United States, France, Norway, and Holland. He also made cross‑cultural comparisons between India and Uganda and more industrialized nations. The results show that the steepness of the age‑crime curve is indeed greatest during recent times and among modem nations. Farrington (1986) compared the relationship between age and crime for English males using British Home Office statistics from 1938, 1961, and 1983. His results, reproduced in Figure 4, show that the rate of offending by adolescents increased considerably over this historical period.

Diverse factors may be influential in accounting for the changing nature of the age‑crime curve (J. Q. Wilson, 1983). However, I suggest that many of these factors are the very features of modernization and modernity invoked in this theory of adolescence‑limited delinquency. The earlier age of puberty and the extension of the period of childhood are generally overlooked as by‑products of modernization, but they have important implications for the experience of youths. The years between 1938 and 1983, covered in the study by Farrington (1986), also witnessed an incremental displacement of sons by their mothers as the family's secondary breadwinners (Modell, Furstenberg, & Hershberg, 1976). The shift of work away from farms, trades, and small family businesses to factories and service industries has stopped adolescents from sharing the daily lives of older relatives. As Anderson (1990) has observed, fewer and fewer "old heads" are initiating young protégés into the adult world. Teens are less well‑integrated with adults than ever before. What has emerged is an age‑bounded ghetto (Schwendinger & Schwendinger, 1985) from within which it seems advantageous to mimic deviant behavior.

Important for this theory, additional data suggest that secular changes may have influenced the age pattern of some crimes but not all. A comparison of the age‑crime curve for data from the Federal Bureau of Investigation's Uniform Crime Reports for 1940, 1960, and 1980 showed that the adolescent peakedness of the curves for most crimes increased in a linear fashion over the 40‑year period (Steffensmeier, Allan, Harer, & Streifel, 1989). However, the authors noted that

the shift toward more peaked distributions is greater for some types of offenses than for others. The shifts are comparatively small for the person crimes and for those property offenses primarily involving older offenders (e.g., fraud and forgery), while the shifts are moderate to substantial for the youth‑oriented, low‑yield property offenses (e.g., robbery and burglary), public order offenses, and the substance‑abuse offenses. (p. 823)

Steffensmeier's finding of different curves for different offenses is consistent with the distinction I have made between two hypothetical types of offenders. On the one hand, life‑course-persistent offenders (with mild neuropsychological impairment, poor self‑control, pathological interpersonal relationships, weak connections to other people, and a lifelong antisocial personality configuration) should account for violence against persons as well as for crimes committed in late life. On the other hand, adolescence‑limited offenders should account primarily for crimes that serve to meet adolescents' lust for acknowledgment and privilege: theft, vandalism, public order, and substance abuse.

Adolescence‑Limited Antisocial Behavior Is Not Pathological Behavior

In an earlier section, it was contended that life‑course‑persistent antisocial behavior represented an especially pernicious and tenacious form of psychopathology. My view of adolescence‑limited delinquency is strikingly different: Its prevalence is so great that it is normative rather than abnormal. It is flexible and adaptable rather than rigid and stable: most delinquent careers are of relatively short duration because the consequences of crime, although reinforcing for youths caught inside the maturity gap, become punishing to youths as soon as they age out of it. Instead of a biological basis in the nervous system, the origins of adolescence‑limited delinquency lie in youngsters' best efforts to cope with the widening gap between biological and social maturity. Moreover, neither this theory nor the empirical evidence suggests that there are links between mental disorders and short‑term adolescent delinquency.

According to this theory of adolescence‑limited delinquency, the behavior of youths who make the transition to delinquent groups near adolescence is readily understood as a group social phenomenon, it does not represent individual‑level deviance. Quay (1987) concurred:

A second pattern . . . involves behavior of a less overtly aggressive and interpersonally alienated nature. In fact, good peer relations in the context of delinquency are at the core of this pattern. . . . There is little, if any, reason to ascribe psychopathology to youths manifesting this pattern: it may well represent an adjustive response to environmental circumstances. (p. 131)

It is my stance that individual characteristics will not predict adolescence‑limited offending; it is a product of an interaction between age and historical period. True, past studies have reported low to moderate correlations between adolescent delinquency and individual difference variables (such as IQ). However, none of these studies excluded life‑course‑persistent subjects before analysis. Thus, it remains unclear whether the obtained correlations represent linear monotonic relationships between variables or "outlier" effects of the extreme scores of life‑course‑persistent subjects. For example, in the New Zealand sample, the often‑reported 8‑point IQ difference (Hirschi & Hindelang, 1977) between delinquents and nondelinquents obtains, but it is the pooled result of a 1‑point mean deficit for adolescence‑onset delinquents and a 17‑point mean deficit for childhood‑onset delinquents. The same pattern obtains for measures of reading achievement and impulsivity (Moffitt, 1990a: White et al., in press).

The Evidence and the Alternatives

In this theory of adolescence‑limited delinquency, I have made several novel propositions. I have suggested that adolescence‑onset delinquency constitutes social mimicry of a pathological type of antisocial child. I have suggested that the motivation for such mimicry follows from a maturity gap between biological adulthood and ascribed adulthood. I have suggested that delinquent mimicry is reinforced by its own consequences while a youth is inside the maturity gap. I have suggested that those consequences lose their rewarding properties after youths age out of the gap, extinguishing delinquency. All three of the components of this theory are needed to support my assertion that adolescence‑limited delinquency is not psychopathology. Because of the newness of this set of hypotheses, there is yet no literature of studies specifically designed to test them. Nonetheless, it was possible to glean from the existing literature empirical evidence in support of most aspects of the theory.

There is some evidence for the mimicry component. A drift into delinquent peer relationships does match the timing of the maturity gap. As predicted, most teens appear to engage in delinquency because they are simply aware of delinquent peer behavior, not because they share attitudes or close friendships with delinquents. Conversely, the most experienced early‑onset delinquents do interact with other adolescents, albeit briefly‑and with their trademark antisocial style.

There is some evidence for the motivational component. The maturity gap has widened during this century, and, as predicted by the theory, the change has coincided with a differential increase in teen crime. After puberty, youngsters' thoughts do turn increasingly to proving their own adultness, and, as predicted by the theory, the particular types of crimes that increased among adolescents this century are ones that satisfy wishes for adult privileges.

There is less evidence for the reinforcement component. Research suggests that youngsters take up drug and alcohol use because it makes them feel independent, but studies of the symbolic reward value of other delinquent acts have not yet been reported. There is better evidence that the informal consequences of crime become deterrents after young adults exit the maturity gap. As predicted, young adults' desistence from crime is influenced by their expectancies of informal sanctions from family, employer, and community.

To date, almost no studies have discriminated childhood‑onset persistent delinquents from adolescence‑onset delinquents and then examined the specific correlates of delinquency in the latter group. Because the available literature mixes the two types of delinquents. it is difficult to evaluate the predictions from this theory against extant findings. However, in evaluating the empirical foundation for this theory of adolescence‑limited delinquency, it is, helpful to contrast the theory with its most favored predecessors: control theories and social learning theories.

Control theories of delinquency point to weak social controls, such as lax supervision by adults or weak bonds to parents, as the causes of burgeoning delinquency (e.g., Hirschi, 1969). The database for control theories is a cross‑sectional correlation between measures of delinquency and supervision in adolescent samples. Research has yet to demonstrate that parenting practices change before teen's interest in problem behavior begins.[vi] More critical, control theories do not explain why antisocial behavior per se is the outcome of weakened social control systems. Why do unsupervised teens not mow lawns for the elderly? Why don't weakly attached youths gather in groups to do more algebra homework? In answer, social control theories rely on the philosophical assumption that all humans are inherently antisocial; crime must thus emerge spontaneously, by default, whenever social controls are weakened. A taxonomic theory cannot afford the luxury of this philosophical premise about the universal mainsprings of human behavior. I offer instead an answer that links individual motivation for crime to its ecological context: Algebra homework does not make a statement about independence; it does not assert that a youth is entitled to be taken seriously. Crime does. How do pubescent teens come to know about antisocial behavior and its effects? I have suggested that they vicariously observe the life‑styles of the life-course‑persistent youths in their midst. Control theories assert that, in the absence of any such models, innocents would invent delinquency.

Calling on learning theory to explain juvenile delinquency, as I have done in this section, is not unique. Social learning theories have suggested that delinquency follows the learning of attitudes conducive to crime (e.g., Sutherland & Cressey, 1978). However, social learning theories of delinquency have not asked, why do so many people learn the attitudes at the same life stage? Why do they learn them so rapidly? What suddenly motivates that learning? What reinforces it? Who are the "teachers"? Why are deviant attitudes unlearned so readily a few years later? Social learning theories describe aspects of the process by which an individual acquires delinquent skills. However, without a motivational component, social learning theories do not address the inescapable epidemiological facts about adolescent delinquency. This developmental analysis of adolescence‑limited delinquency invokes the maturity gap as an explanation for the motivation and timing of adolescence‑limited delinquency. The concept of social mimicry is borrowed to explain why healthy adolescents adopt the style of youths who have been antisocial since early childhood. Thus, this narrative attempts to answer some questions begged by earlier theories.

Comparing These Two Theories With Others

Students of antisocial behavior have been blessed with a number of thoughtful theories. As a group, the theories have tended to be "general" theories of crime; each extends its causal explanation to all offenders.

General theories that summon sociological processes to explain crime and delinquency have provided valuable insights about the proximal mechanisms that promote juvenile delinquency (e.g., Becker, 1968; Cloward & Ohlin, 1960; Hagan, 1987; Hirschi. 1969; Lemert, 1967; Shaw & McKay, 1942; Sutherland & Cressey, 1978). However, sociologists have trained their lenses on the adolescent age period, when the peak prevalence of criminal involvement occurs, and when antisocial behavior is most easily studied with survey methods (Hagan, Gillis, & Simpson, 1985; Sampson & Laub, 1992). Historically, reliance on legal definitions of antisocial behavior and record sources of data kept delinquency researchers focused on the adolescent onset of illegal behavior. Consequently, many delinquency theories have failed to address the stability of antisocial behavior that begins before adolescence, during early childhood. In addition, most sociological theories invoke amplifying causal mechanisms that seem to ignore the empirical facts about the enormous amount of desistence from crime that happens soon after adolescence (Gove, 1985). Causal factors such as low social class, unemployment, cultural approval for violence, and deviant labels do not seem to remit contemporaneously with that undeniable downward shift in the prevalence of offenders during early adulthood.

General theories that invoke causal variables from personality psychology or biology have taught researchers much about how individual differences predispose toward crime (e.g., Bowlby, 1988; Buikhuisen, 1987; Cloninger, 1987; Eysenck, 1977; Gorenstein & Newman, 1980; Mednick, 1977). However, these theories, too, fail to provide a satisfying account, Because such theorists have trained their lenses on early childhood and adulthood (often to the neglect of adolescence), they have failed to anticipate the enormous surge in the prevalence of antisocial involvement that occurs during adolescence. Such theories typically rely on the stability of individual differences in traits such as impulsivity, neuroticism, autonomic nervous system reactivity, or low intelligence. Psychological theories cannot explain the onset and desistence of adolescent delinquency without positing compelling reasons for a sudden and dramatic population shift in criminogenic traits followed by return to baseline a few years later.

Despite the imperfect fit of many existing theories to the epidemiological facts, data in partial support of each theory abound. The resulting stalemate has engendered among students of crime a gentlemen's agreement to disagree. The dual taxonomy described in this article argues that this compromise may be needless. The competing theories may all be correct, but the processes they describe may fit better for different types of delinquents or may operate at different developmental stages in the natural history of antisocial behavior. Among the many mechanisms touted by this developmental taxonomy, few are brand new. What is new is the way in which many different theories of delinquency have been integrated under a taxonomic umbrella.

Indeed, this developmental taxonomy may serve to reconcile disagreements, controversies, and misunderstandings in research on antisocial behavior. For example, the developmental taxonomy may account for effects that appear, disappear, and reappear as a function of the age of research subjects. Behaviorgenetic studies have shown that childhood aggression and adult crime are heritable, whereas juvenile delinquency is much less so (DiLalla & Gottesman, 1989: Edelbrock, Rende, Plomin, & Thompson, in press). Other correlates show also strong relationships to antisocial behavior when it is measured in children and adults but only weak relationships to antisocial behavior measured during adolescence. Such age‑related fluctuations in effect size have been noticed for the associations among antisocial behavior and social class (Elliott & Huizinga, 1983), gender (Smith & Visher, 1980), and reading problems (B. Maughan, personal communication, October 1990; Murray, 1976).

These disappearing effects yield (unnecessary) controversy; they may be an inadvertent consequence of mixing apples with oranges when using adolescents as research samples. I have here proposed that the ratio of life‑course‑persistents to their social mimics will differ as a function of the age of the research sample. Samples of antisocial children and adults should contain relatively more life‑course‑persistent subjects, but in samples of delinquent teens, adolescence‑limited subjects will far outnumber their persistent peers. Consequently, effect sizes for the correlates of persistent antisocial behavior should be attenuated in adolescent samples, and developmental interpretations of cross-sectional data will be confounded. Note one implication: Juvenile delinquents may not be the best group to study if researchers wish to detect the correlates of persistent crime or antisocial psychopathology.

Strategies for Research

Epidemiological Predictions

According to the theory, natural histories of antisocial behavior should be found at predictable prevalence rates in samples followed from childhood until adolescence. Less than 10% of males should show extreme antisocial behavior that begins during early childhood and is thereafter sustained at a high level across time and across circumstances, throughout childhood and adolescence. A much larger number of males, a majority, should show similar levels of antisocial behavior during the adolescent age period but should fail to meet research criteria for a childhood history of stable and pervasive problem behavior. Teenaged males who abstain from any and all delinquency should be relatively rare. False‑positive subjects, who meet criteria for a stable and pervasive antisocial childhood history and yet recover (eschew delinquency) after puberty, should be extremely rare.

A specific research design is needed to evaluate whether these epidemiological parameters will be borne out. Samples should be representative to tap the population range of natural histories. The same individuals should be studied longitudinally to describe the trajectories of individuals as opposed to population shifts. Reports of antisocial behavior should be gathered from multiple sources to tap pervasiveness across circumstances. Antisocial behavior should be assessed repeatedly from childhood through adolescence to capture stability and change across time. Measures of antisocial behavior should be sensitive to developmental heterogeneity to tap individual differences while allowing for the emergence of new forms of antisocial behavior (e.g., automobile theft) or for the forsaking of old forms (e.g., tantrums).

If appropriate research designs fail to yield the predicted individual natural histories (or growth curves), at or near the predicted base rates, then the theory is wrong. However, if subjects are found who match the natural histories of this taxonomy, then the following hypotheses may be tested about differential predictors and outcomes,

Predictions About Differential Correlates of Life‑Course-Persistent and Adolescence‑Limited Antisocial Behavior

According to the theory, the life‑course‑persistent type has its origins in neuropsychological problems that assume measurable influence when difficult children interact with criminogenic home environments. Beginning in childhood, discipline problems and academic failures accumulate increasing momentum, cutting off opportunities to practice prosocial behavior. As time passes, recovery is precluded by maladaptive individual dispositions and narrowing life options, and delinquents are channeled into antisocial adult life‑styles. Thus, the strongest prospective predictors of persistent antisocial behavior are anticipated to be measures of individual and family characteristics. These measures include health, gender, temperament, cognitive abilities, school achievement, personality traits, mental disorders (e.g., hyperactivity), family attachment bonds, child‑rearing practices, parent and sibling deviance, and socioeconomic status, but not age.

According to the description of adolescence‑limited delinquency, youths with little risk from personal or environmental disadvantage encounter motivation for crime for the first time when they enter adolescence. For them, an emerging appreciation of desirable adult privileges is met with an awareness that those privileges are yet forbidden. After observing their antisocial peers' effective solution to the modern dilemma of the maturity gap, youths mimic that delinquent solution. Perversely, the consequences of delinquency reinforce and sustain their efforts, but only until aging into adulthood brings a subjective shift in the valence of the consequences of crime. Then such offenders readily desist from crime, substituting the prosocial skills they practiced before they entered adolescence. This narrative suggests a direct contrast with the predictions made for persistent antisocial behavior. Individual differences should play little or no role in the prediction of short‑term adolescent offending careers. Instead, the strongest prospective predictors of short‑term offending should be knowledge of peer delinquency, attitudes toward adulthood and autonomy, cultural and historical context, and age.

If life‑course‑persistent and adolescence‑limited delinquents, defined on the basis of their natural histories, do not show the predicted differential patterns of correlates, then the theory is wrong.

Predictions About Types of Offenses

According to the theory, the two types will engage in different patterns of offending. Adolescence‑limited offenders should engage primarily in crimes that symbolize adult privilege or that demonstrate autonomy from parental control: vandalism, public order offenses, substance abuse, "status" crimes such as running away, and theft. Life‑course‑persistent offenders should spawn a wider variety of offenses, including types of crimes that are often committed by lone offenders. Thus, in addition to the aforementioned crime types, they should commit more of the victim‑oriented offenses, such as violence and fraud.

If groups of life‑course‑persistent and adolescence‑limited delinquents, defined on the basis of their natural histories, do not show the predicted differential patterns of antisocial behaviors, then the theory is wrong.

Predict ions About Desistence From Crime

According to this theory, transition events in the life course are not unconditional determinants of desistence from crime. Indeed, events such as marriage, employment, or military service can provide opportunities for desistence, but such events can also provide opportunities for continuity. According to this theory, individuals' reactions to life‑transition events will vary predictably, depending on their personal antisocial histories. Adolescence‑limited delinquents can profit from opportunities for desistence, because they retain the option of successfully resuming a conventional life‑style. Life‑course‑persistent delinquents may make transitions into marriage or work, but their injurious childhoods make it less likely that they can leave their past selves behind; they should select jobs and spouses that support their antisocial style, and they should express antisocial behavior at home and at work.

If life‑course‑persistent and adolescence‑limited delinquents, defined on the basis of their natural histories, do not show the predicted differential responses to young‑adulthood transitions, then the theory is wrong.

Predictions About Teenagers Who Abstain From

Delinquency

I have proposed that adolescence‑limited delinquency does not constitute pathology. Rather, it is social activity that is normative as well as understandable from the perspective of contemporary teens. If this assertion is true, the existence of people (however few) who abstain from all delinquency during their adolescent years requires explanation. Earlier, I suggested that adolescents who commit no antisocial behavior have either (a) pathological characteristics that exclude them from peer networks, (b) structural barriers that prevent them from learning about delinquency, or (c) no experience of the maturity gap (because of late puberty or early access to adult roles).

If adolescence‑limited delinquents and abstainers, defined on the basis of their natural histories, do not differ in these predicted ways, then that part of the theory is wrong.

Predictions A bout the Longitudinal Stability of

Antisocial Behavior

I have proposed that most adults who behave in an antisocial fashion are the same individuals who began antisocial behavior in early childhood. During the peak, participation period of adolescence, those persistent individuals will be masked by the "noise" of their more numerous mimics. Following from this observation, estimates of the individual stability of antisocial behavior are expected to violate the longitudinal law which states that relationships between variables become weaker as the time interval between them grows longer (Clarke & Clarke, 1984). One study has found evidence that the longitudinal law is violated in this way when antisocial behavior is studied in the same individuals over time. Stattin and Magnusson (1984) reported that adult crime was predicted more strongly by behavior at age 10 than by behavior between ages 15 and 17. This prediction awaits additional corroboration.

Conclusions

The bulk of research, including the longitudinal research, on antisocial behavior continues to be performed on adolescent subjects. This is unfortunate. If the taxonomy introduced here has merit, then studying offenders at the peak participation age offers the least favorable prospects for understanding the sort of antisocial subject who will develop an adult career of crime and violence. Researchers will learn more about the etiology of severe, persistent antisocial behavior if they single out childhood-onset persistent cases for study and if they begin their studies during infancy, or even prenatally, and follow the same individuals to adulthood. In the past, cross‑sectional comparisons that lumped all delinquents together may have resulted in attenuated effect sizes. This probably obscured some potential causal factors from view and produced underestimates of the importance of others. Indeed, it is likely that most of the research findings cited in this article were attenuated. If the theory is correct, then the empirical footing for it could have been clearer if the distinction between persistent and temporary delinquents had been made in past research. In our past efforts to uncover the causes of persistent predatory crime, we have been studying many of the right variables but in the wrong subjects and at the wrong point in the life course.

Also unfortunate is that almost none of the contemporary theories of delinquency do a good job explaining delinquency that begins in adolescence and ends soon after. Our failure as a field to recognize the heterogeneity of adolescent delinquency may have caused us to overlook important theoretical variables, such as biological age, or structural factors in schools and neighborhoods that determine access to antisocial models. Research is needed that analyzes the roles of biological age and attitudes about maturity in the onset of teenaged delinquency. Delinquency theories are woefully ill‑informed about the phenomenology of modern teenagers from their own perspective. I fear that we cannot understand adolescence‑limited delinquency without first understanding adolescents.

References

Agnew, R. (1991). The interactive effect of peer variables on delinquency. Criminology 29, 47‑72.

American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: Author.

Anderson, E. (1990). Streetwise. Chicago: University of Chicago Press.

Anderson. K. E., Lytton, H., & Romney, D. M. (1986). Mothers' interactions with normal and conduct‑disordered boys: Who affects whom? Developmental Psychology 22, 604‑609.

Azrin, N. H., & Holz, W. C. (1966). Punishment. In W. K. Honig (Ed.), Operant behavior. Areas of research and application (pp. 390‑477). New York: Appleton ‑Century‑Crofts.

Baker. L. A., Mack, W., Moffitt, T E., & Mednick, S. A. (1989). Etiology of sex differences in criminal convictions in a Danish adoption cohort. Behavior Genetics. 19, 355‑370.

Bandura. A. (1979). The social learning perspective: Mechanisms of aggression. In H. Toch (Ed.), Psychology of crime and criminal justice (pp. 193‑236). Prospect Heights. IL: Waveland Press.

Barrett, G. V., & Depinet, R. L. (1991). A reconsideration of testing for competence rather than for intelligence. American Psychologist, 46, 1012‑1024.

Becker, G. S. (1968). Crime and punishment: An economic approach. Journal of Political Economy 76, 169‑217.

Bell, R. Q., & Chapman, M. (1986). Child effects in studies using experimental or brief longitudinal approaches to socialization. Developmental Psychology 22, 595‑603.

Bloch, H. A., & Niederhoffer, A. (1958). The gang: A study in adolescent behavior. New York: Philosophical Library.

Blurnswin, A., & Cohen, J. (1987). Characterizing criminal careers. Science, 23 7, 985‑991.

Blumstein, A., Cohen, J., & Farrington, D. P. (1988). Criminal career research: Its value for criminology. Criminology 26, 1‑35.

Borecki, I. B., & Ashton, G. C. (1984). Evidence for a major gene influencing performance on a vocabulary test. Behavior Genetics, 14, 63‑80.

Bowlby, J. (1988). Developmental psychiatry comes of age. American Journal of Psychiatry, 145, 1‑10.

Buchanan, C. M., Eccles, J. S., & Becker, J. B. (1992). Are adolescents the victims of raging hormones: Evidence for activational effects of hormones on moods and behavior at adolescence. Psychological Bulletin, 111, 62‑107.

Buikhuisen, W. (1987). Cerebral dysfunctions and persistent juvenile delinquency. In S. A. Mednick, T E. Moffitt, & S. A. Stack (Eds.), The causes of crime: New biological approaches (pp. 168‑184). Cambridge, England: Cambridge University Press.

Buss, D. M. (1984). Toward a psychology of person‑environment correspondence: The role of spouse selection. Journal of Personality and Social Psychology 47, 361‑377.

Buss, D. M. (1987). Selection, evocation, and manipulation. Journal of Personality and Social Psychology 53, 1214‑1221.

Cairns, R. B., & Caims, B. D. (1991). Adolescence in our time: Lifelines and risks. Unpublished manuscript.

Cairns, R. B., Caims. B. D., Neckerman, H. J., Gest. S. D., & Gariepy, J.‑L. (1988). Social networks and aggressive behavior: Peer support or peer rejection? Developmental Psychology, 24, 815‑823.

Caspi, A., & Bern, D. J. (1990). Personality continuity and change across the life course. In L. Pervin (Ed.), Handbook of personality theory and research (pp. 549‑575). New York: Guilford Press.

Caspi, A., Bern, D. J., & Elder, G. H., Jr. (1989). Continuities and consequences of interactional styles across the life course. Journal of Personality 57, 375‑406.

Caspi, A., Elder, G. H., & Bern, D. J. (1987). Moving against the world: Life‑course patterns of explosive children. Developmental Psychology, 23, 308‑313.

Caspi, A., Lynam, D., Moffitt, T E., & Silva, P. A. (1993). Unraveling girls' delinquency: Biological, dispositional, and contextual contributions to adolescent misbehavior. Developmental Psychology, 29, 1930.

Caspi. A., & Moffitt, T E. (1991). Individual differences are accentuated during periods of social change: The sample case of girls at puberty. Journal of Personality and Social Psychology, 61, 157‑168.

Caspi, A., & Moffitt, T E, (in press‑a). Continuity amidst change: A paradoxical theory of personality coherence. Psychological Inquiry.

Caspi, A., & Moffitt, T E. (in press‑b). The continuity of maladaptive behavior: From description to understanding in the study of antisocial behavior. In D. Cicchetti & D. Cohen (Eds.). Manual of developmental psychopathology. New York: Wiley.

Chaiken, M. R., & Chaiken, J. M. (1984). Offender types and public policy. Crime and Delinquency, 30, 195‑226.

Chess, S., & Thomas, A. (1987). Origins and evolution of behavior disorders: From infancy to early adult life. Cambridge. MA: Harvard University Press.

Clarke, A. D. B., & Clarke, A. M. (1984). Constancy and change in the growth of human characteristics. Journal of Child Psychology and Psychiatry, 25, 191‑210.

Cleckley, H. (1976). The mask of sanity (5th ed.). St. Louis, MO: Mosby.

Cloninger, C. R. (1987). A systematic method for clinical description and classification of personality variants. Archives of General Psychiatry 44, 573‑588

Cloward, R. A., & Ohlin, L. E. (1960). Delinquency and opportunity. New York: Free

Press of Glencoe.

Coie, J. D., Belding, M., & Underwood, M. (1988). Aggression and peer rejection in childhood. In B. Lahey & A. Kazdin (Eds.), Advances in clinical child psychology (Vol. 2, pp. 125‑158). New York: Plenum Press.

Coie, J. D., Dodge, K., & Kupersmidt, J. (1990). Peer group behavior and social status. In S. R. Asher & J. D. Coie (Eds.), Peer rejection in childhood (pp. 17‑59). Cambridge, England: Cambridge University Press.

Costello, E. J. (1989). Developments in child psychiatric epidemiology. Journal of the American Academy of Child and Adolescent Psychiatry, 28, 836‑841.

Cravioto, J., & Arrieta, R. (1983). Malnutrition in childhood. In M. Rutter (Ed.), Developmental neuropsychiatry (pp. 32‑5 1). New York: Guilford Press.

Csikszentmihalyi, M., & Larson, R. (1984). Being adolescent: Conflict and growth in the teenage years. New York: Basic Books.

Davison, G. C., & Neale, J. M. (1990). Abnormal psychology (5th ed.). New York: Wiley.

DiLalla, L. F., & Gottesman, I.I. (1989). Heterogeneity of causes for delinquency and criminality: Lifespan perspectives. Development and Psychopathology 1, 339‑349.

Dishion, T J., Patterson, G. R., Stoolmilier, M., & Skinner, M. L. (1990). An ecological

analysis of boys' drift to antisocial peers. From middle childhood to early adolescence. Unpublished manuscript, Oregon Social Learning Center, Eugene.

Dishion, T J., Patterson, G. R., Stoolmiller, M., & Skinner, M. L. (1991). Family, school, and behavioral antecedents to early adolescent involvement with antisocial peers. Developmental Psychology, 27, 172‑180.

Dodge, K. A.. Coie, J. D., & Brakke, N. P. (1982). Behavior patterns of socially rejected and neglected preadolescents: The roles of social approach and aggression. Journal of Abnormal Child Psychology, 10, 389‑410.

Dodge, K. A., & Frame, C. L. (1982). Social cognitive biases and deficits in aggressive boys. Child Development, 53, 629‑635.

Dodge, K. A., & Newman, J. P. (1981). Biased decision‑making processes in aggressive boys. Journal of Abnormal Psychology, 90, 375-379.

Edelbrock. C., Rende, R., Plomin, R., & Thompson, L. A. (in press), Genetic and environmental effects on competence and problem behavior in childhood and early adolescence. Journal of Child Psychology and Psychiatry.

Elder, G. H.. Jr. (1986). Military times and turning points in men's lives. Developmental Psychology, 22, 233‑245.

Elliott, D. S., Agelon, S. S., Huizinga, D., Knowles, B. A., & Canter, R. J. (1983). The prevalence and incidence of delinquent behavior: 1976‑1980 (The National Youth Survey Report No. 26). Boulder, CO: Behavioral Research Institute.

Elliott, D. S., & Huizinga, D. (1983). Social class and delinquent behavior in a national youth panel: 1976‑1980. Criminology, 21, 149‑177.

Elliott, D. S., & Huizinga, D. (1984, April). The relationship between delinquent behavior and ADM problems. Paper presented at the ADAMHA/OJJDP State‑of‑the‑Art Research Conference on Juvenile Offenders with Serious Drug. Alcohol, and Mental Health Problems, Rockville, MD.

Elliott, D. S., Huizinga, D., & Menard, S. (1989). Multiple problem youth. Delinquency, substance use, and menial health problems. New York: Springer‑Verlag.

Elliott. D. S., Huizinga, D., & Morse, B. (1986). Self‑reported violent offending: A descriptive analysis of juvenile violent offenders and their offending careers. Journal of Interpersonal Violence, 1, 472‑514.

Elliott, D. S., Knowles, B., & Canter, R. (1981). The epidemiology of delinquent behavior and drug use among American adolescents: 1976‑1980 (The National Youth Survey Project Report No. 14). Boulder, CO: Behavioral Research Institute.

Elliott, D., & Menard, S. (in press). Delinquent friends and delinquent behavior: Temporal and developmental patterns. In D. Hawkins (Ed.), Some current theories of deviance and crime. New York: Springer‑Verlag.

Elliott, D. S., & Morse, B. J. (1987). Drug use, delinquency, and sexual activity. In C. Jones & E. McAnarney (Eds.), Drug abuse and adolescent sexual activity, pregnancy, and parenthood (pp. 32‑60). Washington, DC: U.S. Government Printing Office.

Elliott, D. S., & Voss, H. L. (1974). Delinquency and dropout. Lexington, MA: Heath.

Empey, L. T (1978). American delinquency. Homewood, IL: Dorsey Press.

Erikson, E. H. (1960). Youth and the life cycle. Children Today, 7, 187-194.

Eysenck, H. J. (1977). Crime and personality. London: Routledge.

Fagan, J., & Wexler, S. (1987). Crime at home and in the streets: The relationship between family and stranger violence. Violence and Victims, 2, 5‑23.

Farrington, D. P. (1977). The effects of public labelling. British Journal of Criminology 17, 112‑125.

Farrington, D. P. (1983). Offending from 10 to 25 years of age. In K. Van Dusen & S. A. Mednick (Eds.), Prospective studies of crime and delinquency (pp. 17‑38). Boston: Kluwer‑Nijhoff.

Farrington, D. P. (1986). Age and crime. In M. Tonry & N. Morris (Eds.), Crime and Justice: An Annual Review of Research (Vol. 7, pp. 189‑250). Chicago: University of Chicago Press.

Farrington, D. P. (1989). Early predictors of adolescent aggression and adult violence. Violence and Victims, 4, 79‑100.

Farrington. D. P. (1991). Antisocial personality from childhood to adulthood. The Psychologist, 4, 389‑394.

Farrington, D. P.. Gallagher, B., Morley, L., Ledger. R. J., & West, D. J. (1986). Unemployment, school leaving and crime. British Journal of Criminology 26, 335‑356.

Farrington, D. P., Loeber, R., Elliott, D. S., Hawkins, D. J., Kandel, D. B., Klein, M. W., McCord, J., Rowe, D., & Tremblay, R. (1990). Advancing knowledge about the onset of delinquency and crime. In B. Lahey & A. Kazdin (Eds.), Advances in clinical child psychology (Vol. 13, pp. 283‑342). New York: Plenum Press.

Farrington, D. P., Loeber, R., & Van Kammen, W. B. (1990). Long‑term criminal outcomes of hyperactivity‑impulsivity‑attention deficit and conduct problems in childhood‑ In L. N. Robins & M. R. Rutter (Eds.), Straight and devious path ways. from childhood to adulthood (pp. 62-81). Cambridge, England: Cambridge University Press.

Farrington, D., Ohlin, L.. & Wilson, J. Q. (1986). Understanding and controlling, crime. New York: Springer‑Veriag.

Farrington, D. P.. & West, D. J. (1990). The Cambridge study of delinquent development: A long‑term follow‑up of 411 London males. In H. J. Kerner & G. Kaiser (Eds.), Kriminalitat (pp. 117‑138). New York: Springer‑Veriag.

Feehan, M., Stanton., W, McGee, R., Silva, P. A., & Moffitt, T E. (1990). Is there an association between lateral preference and delinquent behavior? Journal of Abnormal Psychology, 99, 198‑201.

Fergusson. D. M., Horwood, L. J., & Lloyd, M. (199 1). A latent class model of child offending. Criminal Behaviour and Mental Health. 1, 90‑106.

Finnegan. W. (1990a, September). Out there, 1. The New Yorker, pp. 5186.

Finnegan, W. (1990b, November). Out there, 11. The New Yorker. pp. 60‑90.

Fogel, C. A., Mednick, S. A., & Michelson, N. (1985). Minor physical anomalies and

hyperactivity. Acta Psychiatrica Scandinavica, 72, 551‑556.

Friedrich, W. N., & Boriskin, J. A. (1976). The role of the child in abuse. American Journal of Orthopsychiatry, 46, 580‑590.

Frodi, A. M., Lamb, M. E., Leavitt, L. E., Donovan, W. L., Neff, C., & Sherry, D. (1978). Fathers' and mothers' responses to the faces and cries of normal and premature infants. Developmental Psychology, 14, 490‑498.

Goldsmith, H. H., Bradshaw, D. L., & Rieser‑Danner, L. A. (1986). Temperament as a potential developmental influence on attachment. In J. V. Lerner & R. M. Lerner (Eds.), Temperament and social interaction during infancy and childhood (pp. 5‑34). San Francisco: Jossey‑Bass.

Goldstein, A. P. (1990). Delinquents on delinquency. Champaign, IL: Research Press.

Gorenstein, E. E., & Newman, J. P. (1980). Disinhibitory psychopathology: A new perspective and a model for research. Psychological Review, 87, 301‑315.

Gottfredson. M., & Hirschi, T (1986). The value of lambda would appear to be zero: An essay on career criminals, criminal careers, selective incapacitation, cohort studies, and related topics. Criminology 24,213‑234.

Gottfredson, M., & Hirschi, T (1990). A general theory of crime. Stanford, CA: Stanford University Press.

Gove, W. R. (1985). The effect of age and gender on deviant behavior: A biopsychosocial perspective. In A. Rossi (Ed.), Gender and the life course (pp. 115‑144). Chicago: Aldine.

Greenberg, D. F. (1985). Age, crime, and social explanation. American Journal of Sociology 91, 1‑21.

Hagan, J. (1987). Class in the household: A power‑control theory of gender and delinquency. American Journal of Sociology, 92, 788‑816.

Hagan, J., Gillis, A. R., & Simpson, J. (1985). The class structure of gender and delinquency: Toward a power‑control theory of common delinquent behavior. American Journal of Sociology, 90, 1151‑1179.

Hare, R. D., Hart, S. D., & Harpur, T J. (1991). Psychopathy and the DSM‑IV criteria for antisocial personality disorder. Journal of Abnormal Psychology 100, 391‑398.

Hare, R. D., & McPherson, L. M. (1984). Violent and aggressive behavior by criminal psychopaths. International Journal of Law and Psychiatry 7. 35‑50.

Harpur, T J., & Hare, R. D. (1991). The assessment of psychopath as as a function of age. Unpublished manuscript, University of British Columbia, Vancouver, British Columbia, Canada.

Henry, B., Moffitt, T E., Robins, L. N., Earls, F., & Silva, P. A. (1993). Early family predictors of child and adolescent antisocial behavior. Who are the mothers of delinquents? Criminal Behavior and Mental Health, 3, 97‑118.

Hertzig, M. (1983). Temperament and neurological status. In M. Rutter (Ed.). Developmental neuropsychiatry (pp. 164‑180). New York: Guilford Press.

Hirschi, T. (1969). Causes of delinquency. Berkeley, CA: University of California Press.

Hirschi, T, & Gottfredson. M. (1983). Age and the explanation of crime. American Journal of Sociology, 89, 552‑584.

Hirschi, T, & Hindelang. M. J. (1977). Intelligence and delinquency: A revisionist review. American Sociological Review: 42, 571‑587.

Hood, R., & Sparks, R. (1970). Key issues in criminology. New York: McGraw‑Hill.

Horan, P. M.. & Hargis, P. G. (1991). Children's work and schooling in the late nineteenth‑century family economy. American Sociological Review, 56, 583‑596.

Huesmann, L. R., Eron. L. D., LeNowitz, M. M., & Walder, L. 0. (1984). Stability of aggression over time and generations. Developmental Psychology, 20, 1120‑1134.

Hunter, R. S., Kilstrom, N., Kraybill, E. N., & Loda, F. (1978). Antecedents of child abuse and neglect in premature infants: A prospective study in a newborn intensive care unit. Pediatrics, 61, 629-635.

Jesness, C. F., & Haapanen, R. A. (1982). Early identification of chronic offender. Sacramento, CA: Department of Youth Authority.

Jessor, R., Costa, F., Jessor, L., & Donovan, J. E. (1983). Time of first intercourse: A prospective study. Journal of Personality and Social Psychology, 44, 608‑626.

Jessor, R., & Jessor, S. L. (1977). Problem behavior and psychosocial development: A longitudinal study of youth. San Diego, CA: Academic Press.

Kagan, J. (1969). The three faces of continuity in human development. In D. A. Goslin (Ed.), Handbook of socialization theory and research (pp. 983‑1002). Chicago: Rand McNally.

Kagan, J., & Moss, H. A. (1962). Birth to maturity. New York: Wiley.

Kandel, D. (1980). Drug and drinking behavior among youth. Annual Review of Sociology, 6, 235 ‑285.

Kandel, E., Brennan, P. A., & Mednick, S. A. (1989). Minor physical anomalies and parental modeling of aggression predict to violent offending. Acta Psychiatrica Scandanavica, 78, 1‑5.

Kandel, E., & Mednick, S. A. (1991). Perinatal complications predict violent offending. Criminology, 29, 519‑5 30.

Kazdin, A. E. (1987). Conduct disorders in childhood and adolescence. Newbury Park, CA: Sage.

Klein, M. W. (1984). Offense specialization and versatility among juveniles: A review of the evidence. British Journal of Criminology, 24, 185‑194,

Klein, M. W. (1986). Labelling theory and delinquency policy. Criminal Justice and Behavior, 13, 47‑79.

Klein, M. (1989). Watch out for that last variable. In S. Mednick, T Moffitt, & S. A. Stack (Eds.), The causes of crime: New biological approaches (pp. 25‑4 1). Cambridge, England: Cambridge University Press.

Knight, B. J., & West, D. J. (1975), Temporary and continuing delinquency. British Journal of Criminology, 13, 43‑50.

Kraemer, G, W. (1988). Speculations on the developmental neurobiology of protest and

despair. In P. Simon, P. Soubrie, & D. Widlocher (Eds.), Inquiry into schizophrenia and

depression: Animal models of psychiatric disorders (pp. 10 J ‑ 147). Basel, Switzerland:

Karger.

LaFrenier, P., & Sroufe, L. A. (1985). Profiles of peer competence in the preschool: Interrelations between measures: influence of social ecology, and relation to attachment history. Developmental Psychology, 21, 56‑69.

Lahey, B. B., Frick, P. J., Loeber. R., Tannenbaum, B. A., Van Horn, Y., & Christ, M. A. G. (1990). Oppositional and conduct disorder: I. Ameta-analytic review. Unpublished manuscript.

Lee, C. L., & Bates, J. E. (1985). Mother‑child interaction at age two years and perceived difficult temperament. Child Development, 56, 1314‑1323.

Lemert. E. M. (1967). Human deviance, social problems, and social control. Englewood Cliffs, NJ: Prentice Hall,

Lewis, D. 0., Shanok. S. S., Pincus, J. H., & Glaser, G. H. (1979). Violent juvenile delinquents: Psychiatric, neurological, psychological and abuse factors. Journal of the American Academy of Child Psychiatry, 2, 307‑319.

Lipton, D., Martinson. R., & Wilks, J. (1975). The effectiveness of correctional treatment. A survey of treatment evaluation studies. New York: Praeger.

Loeber, R. (1982). The stability of antisocial and delinquent child behavior: A review. Child Development, 53, 1431‑1446.

Loeber, R., & Baicker‑McKee, C. (1989). The changing manifestations of disruptive/antisocial behavior from childhood to early adulthood: Evolution or tautology? Unpublished manuscript, Western Psychiatric Institute, University of Pittsburgh, Pittsburgh, PA.

Locher, R., Green. S., Lahey, B., & Stouthamer‑Loeber, M. (1990). Optimal informants on childhood disruptive behaviors. Development and Psychopathology, 1, 317‑337.

Loeber, R., & LeBlanc, M. (1990). Toward a developmental criminology. In M. Tonry & N. Morris (Eds.), Crime and justice (Vol. 12, pp. 375‑473). Chicago: University of Chicago Press.

Loeber. R., & Schmaling, K. B. (1985). Empirical evidence for overt and covert patterns of antisocial conduct problems: A metaanalysis. Journal of Abnormal Child Psychology, 13, 337‑353.

Loeber, R., Stouthamer‑Loeber, M., Van Kammen, W., & Farrington, D, P. (1989). Development of a new measure of self‑reported antisocial behavior for young children: Prevalence and reliability. In M. Klein (Ed.), Cross‑national research in self‑reported crime and delinquency (pp. 203‑226). Boston: Kluwer‑Nijhoff.

Loehlin, J. C. (1989). Partitioning environmental and genetic contributions to behavioral development. American Psychologist, 44, 12851292.

Lynam, D., Moffitt, T., & Stouthamer‑Loeber, M. (1993). Explaining the relation between IQ and delinquency: Class, race, test motivation, school failure, or self‑control? Journal of Abnormal Psychology, 102, 187‑196.

Lytton, H. (1990). Child and parent effects in boys' conduct disorder: A reinterpretation. Developmental Psychology, 26, 683‑697.

Maccoby, E. E., & Jacklin, C. N. (1983). The "person" characteristics of children and the family as environment. In D. Magnusson & V. L. Allen (Eds.), Human development: An interactional perspective (pp. 75‑92). San Diego, CA: Academic Press.

Magnusson, D. (1988). Individual development from an interactional perspective: A longitudinal study: Hillsdale, NJ: Erlbaum.

Martin, N. G.. Jardine, R.. & Eaves, L. J. (1984). Is there only one set of genes for different abilities? Behavior Genetics, 14, 355‑370.

Mattick, H. W. (1960). Parolees in the army during World War II. Federal Probation, 24, 49‑55.

Maughan, B., Gray, G., & Rutter, M. (1985). Reading retardation and antisocial behavior:

A follow‑up into employment. Journal of Child Psychiatry and Psychology, 26, 741‑758.

Mausner, B., & Platt, E. S. (1971). Smoking: A behavioral analysis. Elmsford, NY. Pergamon Press.

McGee, R.. Partridge, F., Williams, S. M., & Silva, P. A. (1991). A twelve‑year follow up of preschool hyperactive children. Journal of the American Academy, of Child and Adolescent Psychiatry, 30, 224-232.

Meany, M. J.. Aitken, D. H., van Berkel, C., Bhatnagar, S., & Sapolsky. R. M. (1988). Effect of neonatal handling on age‑related impairments associated with the hippocampus. Science, 239, 766‑768.

Mednick.S.A. (1977). A bio‑social theory of the learning of law‑abiding behavior. In S, A. Mednick & K. 0. Christiansen (Eds.). Biosocial bases of criminal behavior (pp. 1‑8). New York: Gardner Press.

Mednick. S. A., Gabrielli, W. F., & Hutchings, B. (1984). Genetic influences in criminal behavior: Some evidence from an adoption cohort. Science, 224, 891‑893,

Meehl, P. E.. & Golden, R. R. (1982). Taxometric methods. In P. C. Kendall & J. N. Butcher (Eds.). Handbook of research methods in clinical psychology (pp. 127‑181). New York: Wiley.

Megargee. E. (1976). The prediction of dangerous behavior. Criminal Justice & Behavior, 3, 3‑2 1.

Millon. T. (1991). Classification in psychopathology: Rationale, alternatives, and standards. Journal of Abnormal Psychology 100, 245‑26 1.

Milner, J. S., & McCanne, T R. (1991). Neuropsychological correlates of physical child abuse, In J. S. Milner (Ed.), Neuropsychology of aggression (pp. 131‑145). Norwell, MA: Kluwer Academic.

Milowe, I. D., & Lowrie, R. S. (1964). The child's role in the battered child syndrome. Journal of Pediatrics, 65, 1079‑1081.

Modell, J., Furstenberg, F. F., & Hershberg. T. (1976). Social change and transitions to adulthood in historical perspective. Journal of Family History, 12, 7‑32.

Moffitt, T E. (1990a). Juvenile delinquency and attention‑deficit disorder: Developmental trajectories from age 3 to 15. Child Development, 61, 893‑910.

Moffitt, T E. (1990b). The neuropsychology of delinquency: A critical review of theory and research. In N. Morris & M. Tonry (Eds.), Crime and justice (Vol. 12, pp. 99‑169). Chicago: University of Chicago Press.

Moffitt, I E. (1991, September). Juvenile delinquency: Seed of a career in violent crime, just sowing wild oats-‑or both? Paper presented at the Science and Public Policy Seminars of the Federation of Behavioral, Psychological, and Cognitive Sciences, Washington, DC.

Moffitt, I E., & Henry, B. (1989). Neuropsychological assessment of executive functions in self‑reported delinquents. Development and Psychopathology 1, 105‑118.

Moffitt, T E., & Henry, B. (1991). Neuropsychological studies of juvenile delinquency and violence: A review. In J. Milner (Ed.), The neuropsychology of aggression (pp. 67‑91). Norwell, MA: Kluwer Academic.

Moffitt, T E., Mednick, S. A., & Gabrielli, W. F (1989). Predicting criminal violence: Descriptive data and predispositional factors. In D. Brizer & M. Crowner (Eds.), Current approaches to the prediction of violence (pp. 13‑34). Washington, DC: American Psychiatric Association.

Moffitt, T E., & Silva, P. A. (1988a). IQ and delinquency: A direct test of the differential detection hypothesis. Journal Of Abnormal Psychology, 97, 330‑333.

Moffitt, T E., & Silva. P. A. (1988b). Neuropsychoiogical deficit and self‑reported delinquency in an unselected birth cohort. Journal of the American Academy of Child and Adolescent Psychiatry: 27, 233-240.

Moffitt, T E., & Silva, P. A. (1988c). Self‑reported delinquency: Results from an instrument for New Zealand. Australian and New Zealand Journal of Criminology, 21, 227‑240.

Morey, L. C. (1991). Classification of mental disorder as a collection of hypothetical constructs, Journal of Abnormal Psychology 100, 289-293.

Moynihan. M. (1968). Social mimicry: Character convergence versus character displacement. Evolution, 22, 315‑331.

Murray, C. A. (1976). The line between learning disabilities and juvenile delinquency Washington, DC: U.S. Department of Justice.

Nagin, D., & Paternoster, R. (1991). The preventive effects of the perceived risk of arrest: Testing an expanded conception of deterrence. Criminology, 29, 561‑588.

Nasby, W., Hayden. B., & DePaulo. B. M. (1980). Attributional bias among aggressive boys to interpret unambiguous social stimuli as displays of hostility. Journal of Abnormal Psychology, 89, 459‑468.

Needleman. H. L., & Beringer, D. C. (1981). The epidemiology of low-level lead exposure in childhood. Journal of Child Psychiatry, 20, 496‑512.

Palmer, T (1984). Treatment and the role of classification: A review of basics. Crime and Delinquency: 30, 245‑267,

Panel on Youth of the President's Science Advisory Committee. (1974). Transition to adulthood. Chicago: University of Chicago Press.

Parkhurst, J. T, & Asher, S. R. (1992). Peer rejection in middle school: Subgroup differences in behavior, loneliness, and interpersonal concerns. Developmental Psychology, 28, 231‑241.

Paternoster, R., Saltzman, L. E., Waldo, G. P., & Chiricos, T G. (1983). Perceived risk and social control: Do sanctions really deter? Law & Society Review, 17, 457‑479.

Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia.

Paulhus, D. L., & Martin, C. L. (1986). Predicting adult temperament from minor physical anomalies. Journal of Personality and Social Psychology, 50, 1235‑1239.

Pepler, D., & Rubin, K. (Eds.). (1991). The development and treatment of childhood aggression. Hillsdale, NJ: Erlbaum.

Piliavin, I., Thornton, C., Gartner, R., & Matsueda, R. (1986). Crime, deterrence, and rational choice. American Sociological Review, 51, 101‑119.

Plomin, R. (1990). The role of inheritance in behavior. Science, 248, 183‑188.

Plomin, R., & Bergeman, C. S. (1990). The nature of nurture: Genetic influence on "environmental" measures. Behavioral and Brain Sciences, 14, 373‑386.

Plomin, R., Chipuer, H. M., & Loehlin, J. C. (1990). Behavioral genetics and personality. In L. A. Pervin (Ed.), Handbook of personality theory and research (pp. 225‑243). New York: Guilford Press.

Plomin, R., DeFries. J. C., & Loehlin, J. C. (1977). Genotype‑environment interaction and correlation in the analysis of human behavior. Psychological Bulletin, 88, 245‑258.

Plomin, R., Nitz, K., & Rowe, D. C. (1990). Behavioral genetics and aggressive behavior in childhood. In M. Lewis & S. M. Miller (Eds.), Handbook of developmental psychopathology, (pp. 119‑133). New York: Plenum Press.

Price, B. H., Daffner, K. R., Stowe, R. M., & Mesulam, M. M. (1990). The comportmental learning disabilities of early frontal lobe damage. Brain, 113, 1383‑1393.

Quay, H. C. (1964a). Dimensions of personality in delinquent boys as inferred from the factor analysis of case history data. Child Development, 35, 479‑484.

Quay, H. C. (1964b). Personality dimensions in delinquent males as inferred from the

factor analysis of behavior ratings. Journal of Research in Crime and Delinquency, 1,

33‑37.

Quay. H. C. (1966). Personality patterns in preadolescent delinquent boys. Educational and Psychological Measurements, 16, 99‑110.

Quay, H. C. (1987). Patterns of delinquent behavior. In H. C. Quay (Ed.), Handbook of juvenile delinquency (pp. 118‑138). New York: Wiley.

Quine, W. V. 0. (1977). Natural kinds. In S. P. Schwartz (Ed.), Naming, necessity and natural kinds (pp. 155‑175). Ithaca, NY: Cornell University Press.

Quinton, D., & Rutter. M. (1988). Parenting breakdown. The making and breaking of intergenerational links. Aldershot. England: Avebury.

Reiss. A. J., Jr. (1986). Co‑offender influences on criminal careers. In A. Blumstein, J. Cohen, J. A. Roth, & C. Visher (Eds.), Criminal careers and career criminals (pp. 121‑160). Washington, DC: National Academy Press.

Reiss, A. J., Jr., & Farrington. D. P. (1991). Advancing knowledge about co‑offending: Results from a prospective longitudinal survey of London males. Journal of Criminal Law and Criminology, 82, 360‑395.

Robins, L. N. (1966). Deviant children grown up. Baltimore, MD: Williams & Wilkins.

Robins. L. N. (1978). Sturdy childhood predictors of adult antisocial behaviour: Replications from longitudinal studies. Psychological Medicine, 8, 611‑622.

Robins. L. N. (1985). Epidemiology of antisocial personality. In J. 0. Cavenar (Ed.), Psychiatry (Vol. 3, pp. 1‑ 14). Philadelphia: Lippincott.

Robins, L. N., & Regier. D. A. (1991). Psychiatric disorders in America. New York: Free Press.

Rodning, C., Beckwith, L., & Howard, J. (1989). Characteristics of attachment organization and play organization in prenatally drug‑exposed toddlers. Development and Psychopathology, 1, 277‑289.

Rothbart, M. K., & Derryberry, D. (1981). Development of individual differences in temperament. In M. E. Lamb & A. L. Brown (Eds.), Advances in developmental psychology (Vol. 1, pp. 37‑66). Hillsdale, NJ: Erlbaum.

Rowe, A. R., & Tittle, C. R. (1977). Life cycle changes and criminal propensity. The Sociological Quarterly, 18, 223‑236.

Rutter, M. (1977). Brain damage syndromes in childhood: Concepts and findings.

Journal of Child Psychology and Psychiatry, 18, 1‑22.

Rutter, M. (Ed.). (1983). Developmental neuropsychiatry. New York: Guilford Press.

Rutter, M., Tizard, J., & Whitmore, K. (1970). Education, health and behaviour. London: Longman.

Ryder, R. G. (1967). Birth to maturity revisited: A canonical analysis. Journal of Personality and Social Psychology, 1, 168‑172.

Sameroff, A., &Chandler, M. (1975). Reproductive risk and the continuum of caretaking casualty. In F Horowitz, M. Hetherington, S. Scarr‑Salapatek, & G. Siegel (Eds.), Review of child development research (Vol. 4, pp. 187‑244). Chicago: University of Chicago, Press.

Sampson, R. J., & Groves, W. B. (1989). Community structure and crime: Testing social disorganization theory. American Journal of Sociology, 94, 774‑802.

Sampson, R. J., & Laub, J. H. (1990). Crime and deviance over the life course: The salience of adult social bonds. American Sociological Review, 55, 609‑627.

Sampson, R. J., & Laub, J. H. (1992). Crime and deviance in the life course. Annual Review of Sociology, 18, 63‑84.

Sandgrund, A. K., Gaines, R., & Green, A. (1974). Child abuse and mental retardation: A problem of cause and effect. American Journal of Mental Deficiency, 79, 327‑330.

Sarnecki, J. (1986). Delinquent networks. Stockholm: National Council for Crime Prevention.

Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of genotype ‑ environment effects. Child Development, 54, 424‑435.

Schwendinger, H., & Schwendinger. J. S. (1985). Adolescent subcultures and delinquency. New York: Praeger.

Sechrest, L., White, S. 0., & Brown, E. D. (Eds.). (1979). The rehabilitation of criminal offenders: Problems and prospects. Washington, DC: National Academy of Sciences.

Shaw, C. R., & McKay. H. D. (1942). Juvenile delinquency and urban areas. Chicago: University of Chicago Press.

Shedler, J., & Block, J. (1990). Adolescent drug use and psychological health. American Psychologist, 45, 612‑630.

Silbereisen. R. K., & Noack, P. (1988). On the constructive role of problem behavior in adolescence. In N. Bolger, A. Caspi, G. Downey, & M. Moorehouse (Eds.), Persons in context. Developmental processes (pp. 152‑180). Cambridge. England: Cambridge University Press.

Simmons, R. G., & Blyth. D. A. (1987). Moving into adolescence: The impact of pubertal change and school context. New York: Aldine de Gruvter.

Skogan, W. G. (1979). Crime in contemporary America. In H. D. Graham & T R. Gurr (Eds.). Violence in America. Historical and comparative perspectives (pp. 375‑391). Beverlv Hills, CA: Sage.

Skogan, W. G. ( 1990). Disorder and decline. New York: Free Press.

Smith, D. A., & Visher, C. A. (1980). Sex and involvement in deviance/crime: A quantitative review of the empirical literature. American Sociological Review, 45, 691‑701.

Snyder, J., & Patterson, G. (1987). Family interaction and delinquent behavior. In H. Quay (Ed.). Handbook of juvenile delinquency, (pp. 216‑243). New York: Wiley.

Stattin, H., & Magnusson, D. (1984). The role of early aggressive behavior for the frequency: the seriousness, and the types of later criminal offenses (report). Stockholm: University of Stockholm, Department of Psychology.

Steffensmeier. D. J., Alian, E. A., Harer, M. D., & Streifel, C. (1989). Age and the distribution of crime. American Journal of Sociology, 94, 803‑831.

Steinberg, L. (1981). Transformations in family relations at puberty. Developmental Psychology, 17, 833‑840.

Steinberg, L. (1987). Impact of puberty on family relations: Effects of pubertal status and pubertal timing. Developmental Psychology, 23, 451‑460.

Steinberg, L., & Silverberg, S. B. (1986). The vicissitudes of autonomy in early adolescence. Child Development, 57, 841‑851.

Stewart, A. (1983). Severe perinatal hazards. In M. Rutter (Ed.), Developmental neuropsychiatry (pp. 15‑31). New York: Guilford Press.

Suedfeld, P., & Landon, P. B. (1978). Approaches to treatment. In R. Hare & D. Schalling (Eds.), Psychopathic behaviour (pp. 347‑376). New York: Wiley.

Sutherland, E., & Cressey, D. R. (1978). Criminology. Philadelphia: Lippincott.

Szatmari, P., Reitsma‑Street, M., & Offord. D. (1986). Pregnancy and birth complications in antisocial adolescents and their siblings. Canadian Journal of Psychiatry, 31, 513‑516.

Tambs, K., Sundet, J. M., & Magnus, P. (1984). Heritability analysis of the WAIS subtests: A study of twins. Intelligence, 8, 283‑293.

Tanner, J. M. (1978). Fetus into man. Cambridge, MA: Harvard University Press.

Tarter, R. E., Hegedus. A. M., Winsten, N. E.. & Alterman, A. L. (1984).

Neuropsychological, personality and familial characteristics of physically abused

delinquents. Journal of the American Academy of Child Psychiatry, 23, 668‑674.

Thomas, A., & Chess, S. (1977). Temperament and development. New York: Brunner/Mazel.

Tinsley. B. R., & Parke. R. D. (1983). The person‑environment relationship: Lessons from families with preterm infants. In D. Magnusson & V L. Allen (Eds.), Human development: An interactional perspective (pp. 93‑110). San Diego, CA: Academic Press.

Tolan, P. H. (1987). Implications of age of onset for delinquency risk. Journal of Abnormal Child Psychology, 15, 47‑65.

Tremblay, R. E. (1991). Aggression, prosocial behavior, and gender: Three magic words, but no magic wand. In D. Pepler & K. Rubin (Eds.), The development and treatment of childhood aggression (pp. 71‑77). Hillsdale. NJ: Erlbaum.

Udry. J. R. (1988). Biological predispositions and social control in adolescent sexual behavior. American Sociological Review, 53, 709‑722.

Vandenberg, S. G. (1969). A twin study of spatial ability. Multivariate Behavioral Research, 4, 273‑294.

Vitaro, F., Gagnon, C., & Tremblay, R. E. (1990). Predicting stable peer rejection from kindergarten to grade one. Journal of Clinical Child Psychology, 19, 257‑264.

Warr, M., & Stafford, M. (1991). The influence of delinquent peers: What they think or what they do? Criminology, 29, 851‑866.

Warren. M. Q. (1969). The case for differential treatment of delinquents. Annals of the American Academy of Political Science, 381, 47‑59.

Weiher, A., Huizinga, D.. Lizotte, A. J., & Van Kammen, W. B. (1991). The relationship between sexual activity, pregnancy, delinquency, and drug abuse. In D. Huizinga, R. Loeber, & T Thornberry (Eds.), Urban delinquency and substance abuse: A technical report (chapter 6). Washington DC: Office of Juvenile Justice and Delinquency Prevention.

Wender, P. H. (1971). Minimal brain dysfunction in children. New York: Wiley.

West, D. J. (1982). Delinquency. Cambridge, MA: Harvard University Press.

West. D. J.. & Farrington, D. P. (1977). The delinquent way of life. New York: Crane Russak.

White, J., Moffitt, T E., Caspi, A., Jeglum, D., Needles, D., & Stouthamer‑Loeber, M. (in press). Measuring impulsivity and examining its relationship to delinquency. Journal of Abnormal Psychology.

White, J., Moffitt, T E., Earls, F., Robins, L. N., & Silva, P. A. (1990). How early can we tell? Preschool predictors of boys' conduct disorder and delinquency. Criminology, 28, 507‑533.

Williams, K. R., & Hawkins, R. (1986). Perceptual research on general deterrence: A review. Law & Society Review, 20, 545‑572.

Wilson, E. 0. (1975). Sociobiology. Cambridge. MA: Harvard University Press.

Wilson. J. Q. (1983). Crime and American culture. The Public Interest, 70, 22‑48.

Wilson, J. Q., & Herrnstein, R. J. (1985). Crime and human nature. New York: Simon & Schuster.

Wolfgang. M. E., Figlio, R. M., & Sellin, T. (1972). Delinquency in a birth cohort. Chicago: University of Chicago Press.

Wolfgang, M. E., Tbornberry, T. P., & Figlio, R. M. (1987). From boy to man, from delinquency to crime. Chicago: University of Chicago Press.

Wyshak, G.. & Frisch. R. E. (1982). Evidence for a secular trend in age of menarche. New England Journal of Medicine, 306, 1033‑1035.