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Warning This Web page has been archived on the Web.

Introduction

In recent years, researchers in the area of family violence have provided alarming statistics on the incidence of both wife battering and child abuse. Within the general population, the most accurate estimate regarding the extent of battering in Canada is found in the Violence Against Women survey (Statistics Canada, 1993). In this survey, conducted with a random sample of over 12,000 women nation-wide, the proportion of ever-married women disclosing physical abuse by a male partner was 29%. Fifteen percent of those currently in a marital relationship reported violence. Within an offender population, the prevalence of violence against female partners is at least as high as in the larger community. In the first national study of family violence perpetrated by federal offenders, Robinson & Taylor (1994) found references to physical abuse of a female partner in 29% of offender files. However, it must be noted that most of the recorded incidents had resulted in official charges. Because wife assault cases seldom come to the attention of the criminal justice system -the probability of detection has been estimated at 6.5% (Dutton, 1987) - it seems reasonable to assume that the true rate of domestic violence among offenders is in fact higher than the review of files indicates. Research conducted with the female partners of offenders confirms that domestic violence is often perpetrated by men who have no record of such behaviour in their files (Dutton & Hart, 1992b).

Children are thought to be present during 68% (Leighton, 1989) to 80% (Sinclair, 1985) of wife assault incidents. As high as these percentages are, they may still underestimate the proportion of children who witness marital violence since information regarding observation by children is frequently collected from mothers. When children themselves are questioned, they often divulge that they are aware of more abuse than their mothers believed them to be (Rosenberg & Rossman, 1990). A study of school children residing in London, Ontario revealed that 23% have witnessed the assault of their mothers by a male partner (Jaffe, 1990; cited in Myers Avis, 1992).

As for the number of children who are direct victims of physical abuse by parents, there are no national statistics available in Canada. According to American research, between 5% (Parke & Collmer, 1975) and 14% of children in the general population are victims of child abuse (Straus, Gelles & Steinmetz, 1980). Robinson & Taylor's (1994) file review study of federal offenders revealed that quite a substantial proportion (34.6%) had been physically abused in childhood. The rate of child abuse perpetrated by this sample of offenders was 3.1%, but as mentioned in the context of partner abuse, the majority of family violence incidents in the offenders' files were those that resulted in official charges. One suspects that there may have been many more unreported incidents of child abuse perpetrated by these men. The secrecy surrounding child abuse and the dependency of children on the abusers means that information gathered from official sources (e.g., police and hospital records) greatly underestimates the extent of child abuse (Canadian Centre for Justice Statistics, 1994).

The phenomena of wife battering and child abuse appear to be correlated although the estimates of overlap are variable. According to Jaffe, Wolfe & Wilson (1990), in one-third of families where there is marital violence there is also child abuse occurring. Walker (1984) reports that over 50% of batterers also physically abuse their children. Bowker, Arbitell & McFerron (1988) found that batterers also abused their children in 70% of families and that the more severe and frequent the wife abuse was, the more severe the child abuse that occurred. The association between marital violence and child abuse is more often attributable to a batterers directing aggression toward his children than to a battered woman's displacing anger at her partner onto her children. A recent review of the literature indicates that when wife assault occurs in a family, children are at greater risk for abuse from the batterer than they are from the battered woman (Saunders, 1994). Wolfe (1987) agrees that although much research on child abuse focuses on female perpetrators, men are more likely to physically abuse children than are women.

Given the substantial proportion of children who are witnesses to marital violence and/or direct victims of physical abuse by their parents, it is important to ascertain the impact that such experiences might have on these individuals. Recently, reliance on anecdotal evidence regarding the consequences of family violence has given way to empirical research on the question. Studies of children in battered women's shelters that consisted of informal descriptions provided by shelter workers have been replaced by quasi-experimental designs comparing witness and/or victimized children with controls on standardized instruments. Much of the research conducted demonstrates that children who see or experience violence in their families are at risk for adjustment difficulties in emotional, psychological and cognitive domains (Goodman & Rosenberg, 1987; Jaffe, Wolfe & Wilson, 1990). For example, the child witness to marital violence may experience sleep disturbance, poor impulse control, poor academic performance, difficulty in concentrating (Hurley & Jaffe, 1990), low self-esteem, lack of self-confidence, insecurity, fear and vulnerability resulting in anxiety (Hughes, 1986) and internalizing symptoms such as depression, withdrawal, passivity, and feelings of hopelessness (Jaffe, Wolfe & Wilson, 1990). Individual responses can vary considerably, there is no one "typical reaction" to family violence (Jaffe, Wolfe & Wilson, 1990; Kashani, Daniel, Dandoy & Holcomb, 1992). Some of the problems described, e.g. depression, anxiety, can still be present in adolescence and adulthood (Carlson, 1990; Forsstrom- Cohen & Rosenbaum, 1985).

Despite the serious nature of the above-noted correlates of abuse, the issue that has likely received the most research attention is the link between violence in the family of origin and later aggressive behaviour by the witness/victim. Even in childhood, those who were abused by parents often display more aggressive behaviour than comparison groups (Dodge, Bates & Pettit, 1990; Truscott, 1992). Children who witness physical violence between their parents but are not themselves physically abused may also be more apt to be aggressive (Fantuzzo & Lindquist, 1989). As with internalizing behaviours, it may be possible to trace aggressive outcomes into adulthood. Some studies have uncovered an increased likelihood of arrests for violent crime among family violence victims (Widom, 1989a; Widom, 1991). McCord (1983) found that parental conflict and violence in an individual's history were predictive of serious crimes against the person (e.g., assault, kidnapping, sexual assault and murder) committed in adulthood. Investigators have also explored the issue of domestic violence in the witness/observer's own adult family.

The idea that "violence begets violence" has a great deal of intuitive appeal. Yet a recent review of the empirical support for this hypothesis has revealed numerous methodological shortcomings in the research (Widom, 1989b). These methodological problems may have led investigators to overestimate the extent to which violent behaviour is exhibited by former abuse victims/witnesses. For example, the results of a child abuse study by Hunter and Kilstrom (1979) illustrate how two common procedures in the cycle of violence research, i.e. use of a retrospective design and a focus on clinical or agency-identified samples, tip the balance in favour of finding that victims become victimisers. A sample of 255 families with a premature or ill new-born was recruited at the time of birth from a new-born intensive care unit. Within a year, 10 families had been reported to social service agencies for abusing their children. During an interview conducted at study intake, nine of these abusive parents (90%) revealed that they had themselves been abused by their parents. But if one considers all the parents in the sample who reported having been victimized in childhood (n=49), the nine who were later reported for child abuse represent only 18% of the subsample. Clearly, a much higher rate of intergenerational transmission for child abuse would have been inferred if only the parents reported for abuse were questioned about their family of origin than if the (actual) prospective study of "high risk" families were conducted.

Widom's (1989b) survey of the literature on the cycle of violence identified several additional methodological problems that characterized research in the area:

  • criteria for abuse and neglect across studies were inconsistent, sometimes including unsubstantiated cases;
  • many investigations relied on retrospective or second-hand information which may be questionable with respect to accuracy;
  • most studies could not speak to the issue of prediction because they utilized a retrospective design;
  • convenience or opportunity samples were often used;
  • correlational designs precluded determinations of causality;
  • abused and neglected individuals were not always differentiated;
  • many studies failed to include appropriate comparison groups or to consider statistical base rates for violent behaviour;
  • delinquency studies often described more generalized delinquent behaviours instead of concentrating on violent criminal behaviour;
  • few studies assessed the long-term consequences of abuse and/or neglect into adulthood.

After outlining these limitations, Widom still concluded that violence in the family of origin increased an individual's risk for becoming violent in adulthood. However, she cautioned that "it cannot be said that the pathway is straight or certain." (Widom, 1989b; p. 24). For example, she estimated that 30% of child abuse victims would go on to abuse their own children; a figure that is smaller than some might expect. Widom's estimate is consistent with numbers provided by other family violence researchers (Kaufman & Zigler, 1987; Oliver, 1993). It should be noted that emphasis on the rate being lower than expected is not intended to downplay the importance of the problem of family violence in any way. Rather, the intention is to underscore the fact that violent behaviour by victims and/or witnesses is not a foregone conclusion; the presence of violence in the family of origin does not mean that one is destined to become an abusive partner or parent.

The purpose of this report is to provide an update on the "cycle of violence" research published in the years since Widom's summary 1, with an eye toward determining whether the problems described in the earlier review are addressed in subsequent investigations. Although the present review is intended to be comprehensive, it does not purport to be exhaustive. Furthermore, it should be made clear that some researchers use the phrases "cycle of violence" and "intergenerational transmission of violence" to refer specifically to former victims/witnesses engaging in family-directed violence whereas others use those terms more broadly to encompass violent acts in general. Both forms of the cycle of violence hypothesis are of interest to the Correctional Service of Canada (CSC) because both intrafamilial and extrafamilial violence may contravene the Canadian Criminal Code.

1 Selected studies published prior to or concurrent with Widom's review will also be examined.

The main focus of this review will be on studies pertaining to the extent of intergenerational transmission of child physical abuse and wife battering, although extrafamilial aggression will also be discussed. Of special interest is research exploring the issue of family violence among offenders, especially since abuse histories are said to be the norm in such populations -- one investigator claims that 90% of prison populations have a history of sexual, emotional and/or physical abuse (Miller, 1990). Others suggest that there are numerous background variables that are common to both criminality and family violence, making offenders an important group to study (Dutton & Hart, 1992b). For the purposes of this report, research on the violent sequelae associated with viewing and experiencing abuse will be described first for children and adolescents, setting the stage for a more detailed discussion of the aggressive behaviour displayed by adult witnesses and/or victims.