What Happens to “Bad” Girls? A Review of the Adult Outcomes of Antisocial Adolescent Girls

We find data on antisocial girls in the psychiatric literature on conduct disorder and the juvenile justice literature on delinquent girls. The diagnosis of conduct disorder, which was not formalized until 1978, specifies a clustering of antisocial behaviors that persists for at least 6 months (DSM-IV). Before 1978, adolescents with antisocial behavior were categorized as having a runaway reaction, group delinquency reaction, or unsocialized aggressive reaction (DSM-II).

A girl adjudicated as delinquent has committed either a status or a criminal offense, all of which are behaviors included in the conduct disorder criteria, although girls with conduct disorder are often not classified as delinquent because they have not been caught. Many delinquent girls are arrested for both types of offenses, but the behaviors result in different consequences. Status offenses (e.g., running away) are unique to juveniles and are regarded as indications for legal supervision. Criminal offenses (e.g., assault) are behaviors that are considered illegal in any age group, but the penalties for juveniles are less severe. A girl can be adjudicated as delinquent for just one act, but most incarcerated girls meet the criteria for conduct disorder (19). In the juvenile justice research literature, girls may also be classified as delinquent on the basis of their scores on delinquency self-report inventories.

Neither conduct disorder nor delinquency is rare among girls. One review (20) reported that conduct disorder is the second most common diagnosis given to adolescent girls, and another investigation (21) revealed that 8% of 17-year-old girls met the criteria for the disorder. A large epidemiologic study of 15-year-olds (22) reported that 7.5%–9.5% of the girls met the criteria for conduct disorder, compared to 8.6%–12.2% of the boys.

Estimating the prevalence of female delinquency is more complicated. The validity of arrest data, which led to the conclusion that female delinquency was uncommon, has been questioned on the basis of gender bias in the justice system (23–26). This bias resulted in reluctance to arrest girls, coupled with a tendency toward psychiatric referrals (27). Paradoxically, this bias also resulted in harsher sentencing, with longer institutionalizations for girls who were actually convicted, in comparison with boys arrested for similar offenses (28–31). The self-report literature supports bias in the judicial system; girls report higher rates of antisocial behavior in this context than the official records indicate (32–38).

It is commonly believed that females in our society are less violent than males. However, if violent crimes against family members or same-sex peers are analyzed separately, the gender gap narrows considerably (39), and the rate of violent crimes among girls and women appears to be increasing (13–15, 40). There is also an interaction between race and gender, with black girls arrested for more aggressive acts than white girls (38, 41–44).

Two lines of evidence suggest that this belief is also the result of bias in the judicial process. First, self-reports of delinquency indicate that girls report the same patterns of antisocial behavior as boys, with the exception of sexual assault (32–38). Second, although delinquency statistics demonstrate that girls are convicted of sexual crimes (e.g., prostitution) more frequently than boys, it is important to point out that until the 1970s, arrested girls were frequently subjected to gynecologic examinations (boys did not undergo analogous examinations). Evidence of sexual activity often led to a charge of sexual delinquency, regardless of the initial offense (26). This practice was based on the notion that delinquent girls were the primary vectors for venereal disease (11, 26).

Historically, the belief was that the adult course for these girls was relatively benign (6–11). The most common pathological adult outcome for antisocial boys was adult criminality (1–5); since the prevalence of criminality in women was so low, it was assumed that most antisocial girls “outgrew” their deviance. Furthermore, marriage and childbearing, long considered indications of healthy adult adjustment for any female, were outcomes frequently found among antisocial girls, lending further support to the perception that these girls did relatively well.

The robust relationship between delinquent behavior among boys and criminal behavior among men is an excellent example of what developmental psychopathologists call “homotypic continuity,” i.e., a strong correlation between a disorder at one point in time and the same symptoms at a further point in time. In contrast, “heterotypic continuity” describes the relationship between a disorder at one point in the life cycle and continued dysfunction at another point in time, but with different signs and symptoms (45–47).

For males, the homotypic continuity between adolescent and adult antisocial behavior is stronger when one “looks backward” from adulthood to adolescence than it is in the data from prospective studies. This finding has been interpreted as evidence of desistance from pathological behavior. However, it may actually be an indication of undetected heterotypic continuity, since research has focused so heavily on studying the rate of criminality. Similarly, it is possible that the apparently benign adult course of antisocial girls may, in truth, be a reflection of heterotypic continuity, or continued dysfunction, but with different manifestations in adulthood. The purpose of this review is to summarize what is known about adult outcomes of antisocial adolescent girls, searching for evidence of homotypic continuity, heterotypic continuity, or both.

Twenty studies met the inclusion criteria. (Several studies presented relevant data but did not meet the inclusion criteria. A list of these articles is available on request.) The data on mortality, criminality, psychiatric morbidity, marriage, parenting, education, occupation, and service utilization are presented below. Table 1 and table 2 summarize the general features of the studies of girls with conduct disorder and delinquent girls, respectively. The investigations span most of this century. Since conduct disorder is a relatively recent diagnostic category, as discussed above, the earlier studies focused on delinquent girls. The most commonly used design was the uncontrolled follow-up study, and follow-up periods ranged from 1 to 41 years, with a median of 15 years. I found five cross-sectional investigations that presented data on the adolescent antisocial behavior of women who were depressed, incarcerated, or from the general population in the Epidemiologic Catchment Area (ECA) study. There is a striking paucity of black subjects across the studies of adolescents, although four of the five cross-sectional investigations included black women.

Adult mortality rates were reported in 10 studies. Five of the 10 studies reported mortality rates of 6%–11% (49, 51, 52, 56, 58), and rates of 0%–2% were found in the other five (12, 55, 61–63). Investigations conducted after 1960 reported higher death rates, except for the Kauai Longitudinal Study (55), which reported a rate of 0%.

The highest mortality rates were found for girls with conduct disorder and delinquent girls who either were in reform schools or had comorbid neuropsychiatric impair~ment. The proportions were similar, in spite of follow-up intervals ranging from 2–4 years to 30 years, suggesting that most of the deaths occurred in early adulthood.

Three studies provided comparison data from the normal population. In one study (49), 7% of a conduct disorder sample had died 30 years later (all of the deaths were natural), compared to 5% of the nonantisocial patients and 10% of the control subjects. The population-based death rate for women at the time was 7%. The 18-year mortality rate in a study of the population of all Swedish girls committed to “state-run probationary schools” (58) was 10% (77% of the deaths were violent), compared to an expected rate of 1.1%–2.6% in the general population. In a 2- to 4-year uncontrolled follow-up of girls with conduct disorder from an inpatient unit (52), it was reported that 6% of the girls had died violently. The age-matched normal population rate for violent deaths was 0.034%.

Sixteen studies reported data on adult crime. All used arrest data, except the Dunedin Multidisciplinary Health and Development Study (48), which used a self-report inventory of illegal activities. Four prospective studies of girls with conduct disorder reported adult crime rates ranging from 33% to 50% (48, 49, 51, 52). The crime rate in adulthood for delinquent girls ranged from 10% to 96%, although the rates clustered around 25%–46% in the majority of the studies (12, 55–57, 59–63). Adult crime rates were higher in all groups of antisocial girls than in any group of either normal control subjects or girls with other psychiatric problems.

The highest rates of adult criminality (71% and 96%) were found in two follow-up studies of delinquent girls (56, 57). These women were also unusual in that they committed crimes such as burglary and assault, rather than the usual “female” crimes of shoplifting, drug use, and prostitution. The lowest prevalence of adult criminality (10%–25%) was found in samples of girls on probation or institutionalized in less confining settings (12, 60–62). These girls probably were less severely antisocial at baseline. Girls with psychiatric comorbidity had the highest rates of adult criminality in two of three studies (55, 56), but the rates were similar in a third that compared criminal activity among conduct-disordered girls with and without depression (48).

In cross-sectional investigations of incarcerated women, one study (66) reported that 35% of them had been arrested as juveniles. Another study (64) found that 73% of a sample of convicted felons had histories of delinquency and that most had been charged with criminal offenses (in contrast to status offenses). However, this group was very aggressive, and all the women had histories of severe abuse. Violent female offenders in another study (65) reported that 56% of them had engaged in fights, 38% had carried weapons, and 74% had been truant.

Psychiatric morbidity, as defined by outcomes varying from commitment to an institution to formally diagnosed disorders, was investigated in 12 studies; three studies (48–50) compared girls with conduct disorder and normal girls. Fourteen percent to 60% of the girls with conduct disorder had adult psychiatric problems, compared to 0%–40% of the normal girls, depending on whether disorders or hospitalizations were counted. Another study (55) reported that 13% of a group of delinquent girls with comorbid psychiatric disorders had adult symptoms, but that none of the girls who were only delinquent developed psychiatric problems.

Six uncontrolled follow-up studies (12, 51, 52, 56, 59, 62) reported that 23%–38% of girls with conduct disorder and 3%–90% of delinquent girls demonstrated psychopathology as adults. Four studies provided data on the continuity between adolescent antisocial behavior and antisocial personality disorder. Robins's study (49) demonstrated that antisocial girls as women had higher rates of antisocial personality disorder than either normal control subjects or other types of former patients. This relationship was confirmed in a birth cohort study (48), where mean scores on an antisocial personality disorder symptom scale were significantly higher for young women with baseline conduct disorder than for women who were normal or depressed as adolescents. In another study (50), 35% of the girls with conduct disorder met the criteria for antisocial personality disorder as adults, compared to 0% of the girls without conduct disorder. An uncontrolled follow-up (51) identified antisocial personality disorder in 23% of women with histories of conduct disorder. Examination of the ECA data (54) revealed strong continuity between conduct disorder and externalizing disorders.

Investigators have also studied the rates of adult hysteria, substance abuse, and depression in antisocial girls. Robins (49) reported that 21% of the referred antisocial girls became women with hysteria. Others (50) reported that 42% of girls with conduct disorder developed a “dramatic” personality disorder as young women; 55% of these also met the criteria for antisocial personality disorder. A cross-sectional study of female criminals (66) revealed that 65% of them met criteria for sociopathy and 40% met criteria for hysteria.

Four studies (49, 51, 52, 56) demonstrated that 40%–70% of girls with conduct disorder or delinquency develop substance abuse problems as women. Suicidal behavior was the most commonly used measure of depression, and rates were as high as 90% (52, 56). When depression was formally measured (48), rates were much lower. In light of these data, the suicidal behavior may have resulted from either antisocial personality disorder or substance abuse, rather than an affective disorder. In support of this interpretation, Rowe et al. (53) reported that 62% of depressed women had no history of conduct disorder, 25% had one or two symptoms, and only 13% had three or more symptoms. A history of conduct disorder had no effect on the course of depression or response to treatment.

Three studies presented data on parenting skills. In Robins's study (49), 36% of the offspring of mothers with histories of conduct disorder were placed outside the home, a rate higher than that for either normal women or other former patients. These mothers had sons with higher arrest rates than the sons of other women or the sons of men who had been antisocial adolescents.

Mothers in the Kauai Longitudinal Study (55) who had been delinquent had a higher rate of family court involvement than other women. Thirty-three percent of the women with delinquency plus psychiatric problems had a family court record, compared to 8% of the delinquency-only group and 4% of the control subjects. Similarly, in a sample of neuropsychiatrically impaired delinquent girls (56), 15 of 21 became pregnant, and 80% of them could not provide safe, stable environments for their children.

The proportions of antisocial girls who married ranged from 19% to 100%, depending on the length of follow-up and the age of the subjects (12, 48–50, 55, 56, 66). There was a trend toward early marriage. In one sample (49), 21% of the antisocial girls married before the age of 17, compared to 9% of the normal control subjects and 8% of the nonantisocial former patients. Thirty-three percent of female criminals with past histories of delinquency reported that they had married before they were 18 years old (66). In another study (48), even those who were not married had a higher rate of early cohabitation.

Four studies presented data on the quality of these relationships. In one investigation (49), antisocial girls developed into women with higher rates of divorce and extramarital sexual activity than either normal control subjects or other types of patients. Sixty-five percent of the antisocial group had marital problems, including many women who were married to abusive or alcoholic men. This finding was replicated in a sample of neuropsychiatrically ill delinquent girls (56); 10% of these subjects were divorced by their early 20s, and 62% of those living with a partner were in violent relationships. Girls with conduct disorder at 21 years of age were 3.9 times more likely to have been involved in a mutually violent relationship than either normal or depressed control girls (48). None of the delinquent girls in the Kauai Longitudinal Study (55) were happy in their marriages; one-third of them ranked their marriage as their biggest concern, compared to 14% of the normal control subjects.

Academic achievement and occupational success were assessed by six studies each. In the Kauai Longitudinal Study (55), 40% of the delinquent subjects did not go beyond high school, compared to 9% of the group without problems. Similarly, in the Dunedin Multidisciplinary Health and Development Study (48), girls with conduct disorder were 3.8 times more likely to have no school certificate (similar to a high school diploma) than the healthy control subjects, although depressed girls were 5.8 times more likely to demonstrate this outcome. In two uncontrolled follow-ups of clinical samples (52, 56), only 10%–29% of the subjects had completed high school.

The data on occupational outcomes are conflicting. One study (49) reported that 15% of the women with antisocial histories were unemployed, compared to 7% of the other former patients and 0% of the normal control subjects. Eleven percent of the antisocial group reported 10 or more jobs in the previous 10 years, compared to 1% of the other former patients and 0% of the normal control subjects. In another follow-up of more impaired delinquent girls (56), only 29% had any job training, with histories of moving from one low-paying job to another. However, an uncontrolled follow-up of delinquent girls on probation (60) found that 48% of them had “good or very fair records” of work. Zoccolillo et al. (50) reported that only 7% of women with conduct disorder had a history of occupational problems, similar to the rate of 9% for women without conduct disorder. There were no differences in employment outcomes between women in the conduct disorder, depressed, and normal groups of the Dunedin Multidisciplinary Health and Development Study (48).

Service utilization outcomes included welfare, involvement with social services agencies (e.g., child protection agencies), and medical care. Social services were required by 55% of women with antisocial histories, compared to 35% of nonantisocial patients and 10% of normal control subjects (49). The antisocial group also had the highest rate of physician utilization: 23% versus 0% for the nonantisocial group and 10% for the normal control subjects. Women with conduct disorder histories in the Dunedin Multidisciplinary Health and Development Study (48) were 3.7 times more likely to use multiple sources of welfare than healthy control subjects or depressed girls, a finding similar to results from two other studies (56, 61).

With a simple summary measure of adult functioning, Robins (49) found that nearly one-half of antisocial adolescent girls developed into adults with poor adjustment in multiple domains. Similarly, the girls with conduct disorder in the Dunedin Multidisciplinary Health and Development Study (48) averaged nearly three adult adjustment problems, compared to one for the normal control subjects and two for the depressed girls. The delinquent girls in the Kauai Longitudinal Study (55) had more problems coping with adult life. Women with histories of delinquency plus mental health problems had the worst outcome, 42% of them having two or more coping problems, compared to 12% of the normal women, 25% of the delinquency-only group, and 26% of the girls with only psychiatric problems.

Zoccolillo et al. (50) found that a past history of conduct disorder was not strongly associated with any one specific adverse outcome, but that 62% of the girls with conduct disorder had difficulties in two or more domains of adult functioning, compared to 9% of the girls without conduct disorder. Two uncontrolled follow-ups of girls with conduct disorder (51, 52) reported that nearly 20% of each sample adjusted poorly to adulthood.

The data clearly support a model of homotypic continuity. At least 25%–50% of antisocial girls engage in adult criminal behavior, and recent delinquency follow-up studies suggest that the rates may be even higher in contemporary cohorts. The cross-sectional studies report that 60%–80% of women with criminal records have histories of antisocial behavior as teenagers. Taken together, these findings imply that antisocial behavior in adolescence is the initial step in the predominant pathway to female adult crime.

Do these girls have an antisocial lifestyle that simply persists into adulthood? Do they start with minor infractions, progressing to more severe criminal behavior? Is there a pattern of misconduct followed by a quiescent period, then emergence of a different type of crime in adulthood? Anecdotal and retrospective data suggest that there are at least two paths between antisocial deviance in girls and such behavior in women (14, 64, 65, 67, 68). The first is a pattern of persistent, aggressive antisocial behavior beginning during latency and continuing throughout adolescence into adulthood, with a steady increase in the severity and aggressiveness of offenses. The second path is characterized by norm violations starting in the teen years, escalating to more severe adult criminal activity in the context of substance abuse or continued association with deviant peers. These paths are similar to some of those described in boys (4, 5, 69–71), although the findings need to be replicated in controlled prospective studies of girls.

Antisocial behavior among adolescent girls was not associated just with adult crime. Regardless of the outcome measure used, the studies reported that substantial percentages of antisocial girls did poorly in adulthood. This multiplicity of adverse outcomes may be a reflection of 1) latent heterogeneity in the population, with subgroups of girls who at baseline may have different prognoses (susceptibility bias) or 2) a large proportion of antisocial girls who display problems in multiple domains of adult life (heterotypic continuity). The data indicate that both conditions may be in operation.

It is quite probable that the population of antisocial girls contains subgroups defined by baseline characteristics that may have prognostic significance, for example, aggression, psychiatric comorbidity, history of abuse, or family dynamics. If a correction for this heterogeneity is not made in the sampling process or in the data analysis, adverse outcomes may show up in every domain. However, this would be a reflection of susceptibility bias, with each of the subgroups developing the outcomes unique to them, rather than a reflection of the majority of the girls having widespread pathology.

In support of this explanation, several studies examined the effects of baseline aggression or psychiatric comorbidity and reported that groups of girls with these characteristics each had sets of outcomes different from those of the girls without them (12, 48, 52, 55, 56, 62). Future cross-sectional studies of antisocial girls and women should be designed to identify subgroups with potentially different outcomes. Similarly, longitudinal stud~ies should include measurements of the prognostic effects of aggression and baseline psychiatric comorbidity as well as explore the effects of variables such as history of abuse, family psychiatric history, and family functioning.

Although susceptibility bias could explain the variety of untoward outcomes, six studies (48–52, 55) reported that one-half of the girls had serious problems in multiple domains of their adult lives, data that clearly support a model of heterotypic continuity. There are two possible mechanisms for heterotypic continuity: 1) there may be a core biological or psychological deficit underlying symptoms in both stages of life, or 2) adolescent antisocial behavior may derail the normal developmental process so significantly in these girls that it seriously compromises their ability to cope with adulthood.

A core biological or psychological deficit would generate a picture of heterotypic continuity by producing maladaptive behavior across all adult domains. Several biological or psychological abnormalities have been associated with antisocial behavior (primarily in males) and could potentially be the mechanism(s) for heterotypic continuity, but I will discuss only the two most likely deficits—prefrontal lobe dysfunction and poor attachment.

Prefrontal lobe dysfunction has been suggested as a potential biological deficit in antisocial behavior (72, 73). Abnormalities in one or all of the three prefrontal lobe circuits responsible for executive functions, mood regulation, and motivation would result in women characterized by impulsivity, short attention span, difficulty planning or delaying gratification, inability to learn from experience, mood instability, and poor motivation, all of which would impair normal adult functioning. Research into the relationship between prefrontal lobe function and antisocial behavior has predominantly focused on executive functions in deviant males (73). Two studies that included antisocial girls (74, 75), however, reported that girls with inattention, distractibility, and impulsivity had higher rates of psychosocial impairment. We have no data on any other aspects of prefrontal lobe function and no research using more sophisticated diagnostic techniques such as imaging. Further research should be done on the role of prefrontal lobe dysfunction in antisocial females.

A core psychological deficit in attachment has long been postulated as an etiologic factor in antisocial behavior (76) but could also be the mechanism for the heterotypic continuity reported here. Poor attachment has been associated with the inability to form and maintain stable adult relationships with partners, children, peers, and co-workers. This could certainly explain the pervasive dysfunction in women with histories of adolescent antisocial behavior. Most of the studies on the effects of attachment are cross-sectional, use self-report delinquency measures, and have been done on males, but several (77–79) have studied female adolescents and reported that attachment difficulties are associated with antisocial behavior. We clearly need to investigate the developmental effect of poor attachment as a factor in adverse adult outcomes.

The second possible mechanism explaining heterotypic continuity is interference with normal adolescent development. This derailment may occur because antisocial behavior prevents girls from learning the social and psychological skills needed for adulthood, or because antisocial conduct may expose girls to situations in which each behavioral choice leads to further deviance. There are no data with which to test this hypothesis, but prospective studies of development in girls with other types of problems, such as early puberty, histories of institutionalization, or depression (80–83), indicate that such experiences can derail normal development and lead to multiple poor outcomes in adulthood. We need to collect detailed longitudinal data from antisocial females about the impact of their behaviors on their daily lives. To determine whether antisocial deviance has a uniquely deleterious effect on development, comparison data should also be collected from normal girls and girls with other psychiatric disorders.

Although the results of this review are provocative, it is important to point out limitations of the studies. The most significant is the absence of data on the nonwhite female population. This is particularly troubling given the higher arrest rate for black females. Few of the studies used control groups, many samples were small, and larger samples were often heterogeneous with regard to type of antisocial behavior. A variety of data collection methods were used, ranging from public registries to clinical interviewing. There was an assortment of outcome measures, and cohort effects were of particular concern in the definition of conduct disorder, changes in the definition of sexual deviance, and changes in institutionalization practices over the past century.

In spite of these limitations, this review establishes that female antisocial behavior has important individual and societal consequences. The studies reviewed here suggest that if we do a 10-year follow-up on Jennifer, described at the beginning of this article, we are likely to find that she has not graduated from high school, has had multiple, unstable relationships, is using drugs and alcohol, uses aggression to solve conflicts, has received psychiatric and social services, has been in jail, and has had difficulties caring for her children. Our understanding of the developmental trajectories of antisocial girls and women is so limited, however, that as policy makers or clinicians, we do not know how to prevent or treat such outcomes.

Several lines of research should be developed to resolve these issues. In all types of studies, particular emphasis should be placed on including antisocial females from nonwhite populations, and comparison data should be collected not only from normal females but from psychiatric groups as well. The first type of further research needed is phenomenological studies of antisocial girls and women. This would allow us to assess the validity of our current diagnostic criteria for conduct disorder in girls (20) as well as identify variables in adolescence that may have long-term prognostic value, such as aggression, psychiatric comorbidity, and type of delinquent behavior. Second, we need cross-sectional studies of the rates of prefrontal lobe dysfunction and poor attachment in antisocial females to determine whether these are important factors in the trajectories to adult pathology. Third, the next generation of longitudinal studies on antisocial adolescent girls should be designed with multiple data collection points, should track subgroups of antisocial girls with potentially different prognoses, and should assess brain and psychological changes over time. Data from these three types of studies will give us the knowledge necessary to develop treatment and secondary prevention strategies for girls like Jennifer.