Correctional Service Canada
Symbol of the Government of Canada

Common menu bar links

The Impact of Experiencing and Witnessing Family Violence during Childhood: Child and Adult Behavioural Outcomes

Warning This Web page has been archived on the Web.

Aggressive Behaviour by Witnesses and/or Victims in Adolescence

Violence in general

The evidence for a relationship between child experiences and violent behaviour in later years is mixed. For example, Widom (1991) conducted a prospective study of the criminal sequelae associated with childhood neglect and physical or sexual abuse. She compared 908 abused individuals and 667 matched controls with respect to criminal behaviour exhibited in adolescence. Those in the abused group had come to the attention of the courts between 1967 and 1971 as a result of having been neglected by parents or having been physically or sexually abused. Matches were found for about three-quarters of these children; the control group was comprised of children matched by gender, race, date of birth and hospital of birth (for children under school age) or class in elementary school (for children of school age). Widom found that abused/neglected individuals were indeed more likely than non-abused comparisons to come to the attention of justice authorities as juvenile offenders (26.0% and 16.8% arrest rates respectively). The abused and neglected adolescents also had more offences recorded and were significantly younger at the time of their first offence. Yet with respect to violent offences, there were no significant overall differences between the arrest rates of the abused and control groups (4.2% vs. 2.8%). Separate analyses by gender indicated that abused/neglected males were not significantly different from control males but surprisingly, abused/neglected females were marginally more likely than control females to have been arrested for a violent crime (1.9% vs. .3%).

In contrast, Truscott (1992) tested the cycle of violence hypothesis with an all-male sample consisting of 65 young offenders and 25 grade 10 students. Subjects were asked whether they had ever witnessed or experienced violence by their parents (50 out of 90 replied in the affirmative) and whether they had themselves been violent with others (42 out of 90 had been). Among those who admitted to behaving violently, the majority (69%) came from a violent family. Violent adolescent behaviour was significantly associated with being physically as well as verbally aggressed against by the father4 but there was no relationship with maternal verbal or physical aggression nor with paternal or maternal violence witnessed. The discrepancy in the results of these two studies might be due to the use of self-report rather than official abuse records in the Truscott study. As pointed out earlier, the latter technique greatly underestimates the incidence of child abuse, which would in turn suppress between-group differences.

4 R2=.21 and R2=.20 respectively.

Some researchers have found that under-reporting of abuse incidents even occurs in self report studies (Della Femina, Yeager & Lewis, 1990; Stein & Lewis, 1992). In a follow-up investigation of formerly incarcerated juvenile offenders, 69 of the original 97 subjects were reinterviewed nine years later (see Lewis, Shanok, Pincus & Glaser, 1979). When asked about abuse suffered in childhood, 26 respondents provided information that was inconsistent with that obtained in the original study; eighteen subjects denied or minimized abuse experiences whereas the remaining eight described abuse incidents that they failed to reveal during the first study. In an effort to clarify the discrepancies, the investigators requested another interview with these 26 people. Eleven respondents agreed to be interviewed again, eight of whom had denied abuse in the follow up and 3 of whom revealed abuse for the first time. When asked to account for the discrepancies in their reports, all eleven subjects maintained that they had been abused and gave some reason for withholding the information. The explanations given for denying abuse included embarrassment, a wish to protect parents, a conscious wish to forget the past and a lack of rapport with the interviewer (Della Fernina et al., 1990). The findings of this study are especially interesting given the tendency for the general public to expect that delinquent and/or criminal populations might be motivated to exaggerate childhood abuse in an effort to elicit sympathy. In reality, the respondents' concealment of abuse experiences may have resulted in a muting of differences between groups, thus hindering the researchers' efforts to determine the criminal effects of experiencing family violence.

When considering the connection between child maltreatment and later violent criminal behaviour by adolescents, Widom (1989b) found that some research supported the violence begets violence hypothesis while other investigations did not. She concluded that although the majority of abused children do not become violent, there was a definite relationship between the two variables (Widom, 1991). A similar statement would characterize the studies summarized in this report. Furthermore, examination of the studies reviewed here reveals that methodological problems continue to plague the research, impacting on the interpretation of findings. One serious problem with investigations of adolescents' aggressive behaviour is that they often rely on official records to identify violent delinquent behaviour. It is possible that the effects of childhood exposure to violence are thus confounded with the effects of processing by the justice system.

Violence directed at dating partners

Recently, it has been recognized that a phenomenon analogous to spousal violence is not entirely uncommon among adolescent dating couples (Roscoe & Callahan, 1985; O'Keeffe, Brockopp & Chew, 1986). Research on courtship violence is important because it may provide glimpses of interaction patterns that are carried forward into adult intimate relationships (Makepeace, 1981). Carlson (1990) recruited adolescents aged 13 to 18 years from 4 residential treatment centres and a youth shelter and asked about violence directed at dating partners. Respondents from non-violent families were compared with those who had witnessed marital violence (almost 1/2 reported seeing their fathers hit their mothers, 26% saw mothers hit fathers). Counter to expectation, no differences were found between witnesses and non-witnesses with respect to attitudes about the appropriateness of using violence against a boyfriend/girlfriend or in the actual use of violence against a dating partner. In short, the levels of dating violence reported were similar to those found in research with non-clinical samples of high school students. However, estimates suggest that those levels are actually fairly high, with one quarter to one half of high school students having experienced an abusive relationship 5 (Myers Avis, 1992; Bergman, 1992).

5Although no national data exists regarding the extent of physical violence in high school dating relationships, a recent national survey conducted with Canadian university and college students indicates that 22.3% of women had been physically assaulted by a partner within the last 12 months, while 35% had been so victimized in the years since they left high school (DeKeseredy & Kelly, 1993).

Carlson proposed that the adolescents in her sample had experienced numerous other stressors in their lives which were not accounted for, e.g., physical abuse, sexual abuse, divorce and/or substance abuse by one or both parents, possibly overshadowing the effect of witnessing violence. In order to test this hypothesis, she conducted another study with the same sample, this time differentiating between 4 groups: those who had only experienced child abuse (n=6), those who had only witnessed marital violence (n=12), those who had both experienced abuse and witnessed marital violence (n=50) and those who had experienced neither type of abuse (n=25; Carlson, 1991). Composite measures were created from a small number of open-ended questions tapping approval of violence and use of violence. There was a main effect of gender wherein males were more likely to approve of violence. Males also had marginally higher scores for use of violence. However, there was no significant effect related to abuse or witness status, even when gender was statistically controlled. This finding is surprising given earlier studies demonstrating that teens involved in dating violence had higher rates of maltreatment in childhood (Roscoe & Callahan, 1985; Reuterman & Burcky, 1989).

Carlson listed a number of possible reasons for her non-significant findings, including the dearth of information on the nature and extent of the violence that subjects experienced/ witnessed and the confounding of family violence variables with other potential risk factors. Additional methodological problems that Carlson did not consider were the unequal (and small) cell sizes and the unacceptably low internal consistencies achieved for the violence "scales" she constructed (a=.56 and a=.64 for approval of violence and use of violence respectively). Both these factors could have obscured significant relationships. Studies using a scale with better psychometric properties (Child Witness to Violence Interview; Jaffe et al., 1989) have indicated that children exposed to wife battering are indeed more likely to condone violence as a means of resolving conflict in interpersonal relationships. Unfortunately, no studies have attempted to follow up this effect into adolescence. A recent summary of the research describing long-term consequences of physical abuse revealed that the bulk of adolescent studies dealt with the question of delinquency6 rather than dating relationships (Malinosky-Rummell & Hansen, 1993).

6The majority of these studies have already been reviewed by Widom (1989b).