Correctional Service Canada
Symbol of the Government of Canada

Common menu bar links

The Incidence of Family Violence Perpetrated by Federal Offenders: A File Review Study

Warning This Web page has been archived on the Web.

Correlates of Family Violence

In this section we examine a number of variables which may be correlates of the perpetration of family violence. We have selected characteristics which might differentiate offenders who engaged in family violent behaviour from offenders who did not engage in such behaviour. As Tolman and Bennett (1990) have pointed out, the identification of the distinguishing characteristics of family violent men increase our potential for predicting and controlling this undesirable behaviour. In addition, the identification of such distinguishing features may illuminate factors which cause or maintain abusive behaviour.

It should be clear that we made no attempt to establish causal links between offender characteristics and family violent behaviour in this study. The file review method of the study precludes definitive conclusions about the causes of family violence in this population of men. For example, with the exception of childhood victimization, the temporal sequence vis à vis the occurrence of family violence could not be specified for many of the characteristics we examined (e.g., substance abuse, employment stability, etc.). Nevertheless, the exercise of identifying offender characteristics that correlate with family violence furnishes helpful information for identifying which offenders are likely to be perpetrators of family violence. In addition, the establishment of correlates is an important first step in the research process aimed at determining causality.

Childhood Victimization of Offenders by Family Members

According to both research and theory (for example, see Appleford, 1989, Tolman & Bennett, 1990), experience of family violence during childhood is one of the most important risk markers of later violence against family members. Because of the importance of this variable we provide detailed descriptive information about childhood victimization in this sample. We then show evidence of the correlation between childhood victimization by family members and later reports of perpetration of family violence.

Table 10 shows the weighted rates of childhood victimization of offenders by the major types of abuse that were recorded in the files for this sample of offenders admitted to federal institutions. According to the file information, slightly more than half of the offenders (50.2%) had experienced childhood abuse in the form of physical, sexual, or psychological abuse, neglect, or had witnessed the abuse of other family members. Over one-third of the offenders (39.6%) in this sample were physically or sexually abused by family members, usually a father. One-quarter (24.6%) had witnessed the abuse of siblings or parents.

It is likely that the file review method we have employed resulted in an underestimation of the number of offenders who were victimized by family members as children or, witnessed the abuse of others. It is difficult to estimate the extent of under-estimation that occurred. However, the majority of files contained information about family background and it was possible to code the level of contact the offenders had with their parents during childhood. For example, information on contact during early childhood (0 - 5 years) was unavailable in only 1.9% of files with respect to mother contact and 2.4% of files with respect to father contact. For later age periods (i.e., 6 - 16 years), missing parental contact information ranged form 3.3% to 4.3% of cases. Overall, we judge that the level of under-reporting of childhood victimization resulting from missing parental contact information to be low. Nevertheless, there may have been other file maintenance factors that posed threats to reliable recording of childhood victimization information.

Table 10 Childhood Victimization of Offenders (N=935)

Abuse Types
Sexual Abuse
Physical Abuse
Psychological Abuse
Witnessed Abuse
Sexual and/or Physical Abuse
Any Childhood Victimization

Table 11 through Table 15 present more detailed information with respect to age of abuse, characteristics of perpetrators, and type of abuse witnessed. Table 11 through 13 provide information on the sub-sample of 436 offenders for which file evidence was available to suggest a history of childhood victimization by members of their families. The majority of those who were abused were first abused between 0 and 5 years of age (Table 11). According to the available information, abuse appears to be relatively constant across the years of 0 to 16 for those offenders who were abused. Most of these offenders were abused by their fathers (75.5%). However, as Table 12 demonstrates, victimization by mothers was not uncommon (42.4%). Abuse by parents included biological, adoptive, foster parents, and guardians. A notable proportion (5.6%) of this group suffered abuse at the hands of an institutional authority figure. In terms of the types of abuse committed by the parents of offenders who were victimized, fathers were more identified as being physically abusive than mothers. On the other hand, neglect was more often associated with mothers than with fathers. Table 13 provides further details on types of abuse by gender of parents.

Table 11 Experience of Abuse by Age (n=432)

Age Period
Any Abuse
Age at First Abuse

Table 12 Perpetrators of Offenders who Were Victimized (N=432)

Other Family
Institutional Authority

Table 13 Types of Abuse Perpetrated by Parents
of Offenders who were Abused

Father (n=327)
Physical Abuse
Sexual Abuse
Psychological Abuse
Mother (N=183)
Physical Abuse
Sexual Abuse
Psychological Abuse

While not all files were complete, it was possible to collect information concerning the nature of abuse witnessed by offenders. We examined the information for the 230 offenders who had, according to the files, witnessed abuse. Table 14 shows that fathers were most often the perpetrators of abuse witnessed by offenders, while Table 15 indicates that mothers or an adult female were most often the victims of abuse that was witnessed. However, it must be noted that this group of offenders also witnessed a high rate of abuse of other children (63.5%). The available information, presented in Table 16, shows that physical abuse was the most common form of abuse offenders witnessed.

Table 14 Perpetrators of Abuse Witnessed by Offenders (N=222)

Other Family

Table 15 Victim of Abuse Witnessed by Offenders (N=222)

Father (or Adult Male)
Mother (or Adult Female)

Table 16 Types of Abuse Witnessed by Offenders (N=222)

Types of Abuse Witnessed
Neglect or Finanial

Childhood victimization of offenders varied significantly by region (Figure 5) and the differences did not correspond exactly to the regional differences in rates of perpetration of family violence. Ontario region had the lowest proportion of files containing evidence of childhood victimization, while Quebec region had the largest proportion. There was also a statistically significant difference in rates of childhood victimization by Aboriginal status. Figure 6 shows that Aboriginals were much more likely to have experienced abuse as children (64.9%).

Figure 5
Childhood Victimization by Region

Figure 6
Childhood Victimization by Aboriginal Status

As expected, childhood victimization by family members was found to be significantly correlated with later perpetration of violence against family members. Figure 7 shows the rates of perpetration of family violence, including physical or sexual abuse of family members, by any evidence of childhood victimization of offenders ***. While not all offenders who had been identified in the files as childhood victims of abuse showed evidence of perpetration of family violence, those who were identified were almost twice as likely (1.8 times) to be perpetrators than those who were not victimized. Of files containing evidence of childhood victimization, 42.4% also contained evidence of perpetration of abuse against family members. The corresponding figure for files lacking evidence of childhood victimization was 23.9%.

*** We exclude psychological abuse from all of the correctional analyses presented in this report because of the low frequency of occurrence of this variable.

Figure 7
Perpetrators of Family Violence by Childhood Victimization

More detailed information concerning the link between abuse of family members and childhood victimization is displayed in Table 17. Appendix C also includes more specific statistical information with respect to cell counts and measures of associations. We present the rates of perpetration of any family violence, abuse against female partners, and abuse of children for four categories of childhood victimization: witnessing abuse, any victimization excluding witnessing abuse, any victimization including witnessing, and any victimization and witnessing abuse. All types of childhood victimization were significantly related to the three perpetration variables, with the exception that witnessing of abuse was unrelated to perpetration of abuse against children. The data corroborate earlier results regarding the relationship between victimization and perpetration in offenders samples (Dutton & Hart, 1992) and are consistent with the more general body of research literature on wife-batterers (Tolman & Bennett, 1990).

Table 17 Childhood Victimization Correlates of Family Violence

Correlates Any Family
Violence %
Violence Against
a Female Partner %
Violence Against
Childern %
Witnessed Any Abuse
Yes 45.2*** 37.8*** 16.2
No 28.4 24.3 11.4
Experienced Abuse
Yes 42.4*** 33.7*** 20.0***
No 23.9 22.6 6.4
Both Witnessed and Experienced Physical/Sexual Abuse
Yes 47.1*** 37.9** 20.2**
No 29.2 25.3 10.7
Physical and/or Sexual Abuse
Yes 42.5*** 33.8** 22.2***
No 25.8 23.7 6.1

While all three categories of childhood victimization were predictive of perpetration of family violence by the offenders, witnessing of abuse appears to be the most important victimization correlate of abuse against female partners. While we can only report the occurrence of a trend in this data, it is clear that witnessing of abuse was at least as important as experiencing abuse in terms of the perpetration of family violence as an adult. **** Previous evidence regarding the relative importance of witnessing abuse over experiencing abuse with respect to wife-battering has been reported by Hotaling and Sugarman (1986).

**** The reader with an interest in statistical procedures will note that we report phi coefficients to indicate the magnitude of association between our independent and dependent variables in the tables of correlates included in Appendix C.

Other Correlates of Family Violence

We examined a number of additional variables which were viewed as potential predictors or correlates of the perpetration of family violence based on our review of the literature (see, for example, Tolman & Bennett, 1990). Included among the variables we examined were demographic variables (age, marital history, education, employment stability), substance abuse problems and mental disorders, and criminal history indicators. Appendix C provides detailed breakdowns of rates of family violence by these correlates including cell frequencies and measures of association.

Table 18 shows the rates of occurrence of physical and sexual violence against family members generally and violence against female partners and children by demographic factors. The most notable finding was that age was positively and significantly correlated with the perpetration of family violence.

Table 18 Demographic Correlates of Family Violence

Correlates Any Family
Violence %
Violence Against
a Female Partner %
Violence Against
Childern %
Under 30 23.2*** 23.9 2.8
30 and Over 41.7 30.7* 18.9
High School Diplomas
Yes 26.6 20.7** 9.6
No 34.9 30.8 13.9
Employment Instability
Yes 30.4* 26.7 11.5
No 35.6 29.4 14.1
Number of Marriages
Under 3 29.3*** 25.8* 11.1*
3 and over 48.7 35.3 18.0

Older men (i.e., men 30 years and over) were more likely to have evidence of family violence on their files than younger men. This association held for both violence against female partners and children. The most likely explanation for this finding is that younger men have been at risk to offend for fewer years than older men in the sample. In addition, it is likely that many younger men who have been violent have not yet been detected. Another possibility is that the age effect may be a function of sentencing and admission patterns associated with older offenders who are currently being convicted of incest crimes committed when they were younger. It is likely that both time at risk and sentencing factors play a role in accounting for the higher rates of family violence among older men in the sample.

Men under age 30 comprised approximately 50% of the current sample. Dutton (1988) estimated that 31 years is the age when men are most at risk for the commission of wife assault. This suggests that as the younger men in our admission sample age, they will show more evidence of engaging in family violence.

Other demographic factors related to family violence included a significantly higher rate of abuse against female partners among men who had not received high school diplomas. Men who had three or more marriages were also more likely to have assaulted their female partners and children then men who had fewer than three marriages.

Rates of family violence associated with substance abuse problems and the presence of mental diagnoses are presented in Table 19. Consistent with previous data on the link between substance abuse and violent behaviour among federal offenders (Robinson, Porporino, & Millson, 1991), alcohol problems were associated with family violence while drug problems were unrelated. In fact, offenders with drug problems had lower rates of violence against family members than offenders who had no drug problems. It should be noted that the presence of alcohol problems was one of the strongest correlates of family violence in the current data, although not as strong as the correlation observed for childhood victimization. Interestingly, alcohol problems were significantly correlated with abuse against female partners but not assaults against children.

Table 19 Substance Abuse and Mental Disorder Correlates of Family Violence

Disorder Any Family
Violence %
Violence Against
a Female Partner %
Violence Against
Childern %
Alcohol Problems
Yes 37.7*** 34.0*** 2.2***
No 23.2 16.6 13.5
Drug Abuse
Yes 16.3*** 15.4** 5.8
No 35.3 130.1 13.6
Diagnosis of Mental Illness
Yes 41.2* 36.2*** 19.7***
No 28.2 24.1 9.2
Yes 46.8* 30.3 20.8
No 31.8 27.8 12.3
Yes 5.2*** 6.8*** 22.1**
No 30.3 26.5 11.1
Personality Disorder
Yes 41.1*** 38.1*** 19.1**
No 29.1 24.3 10.2

A diagnosis of mental illness was also related to family violent behaviour in this sample. The latter variable was measured using psychological reports on file. Because there was previous evidence of a link between personality disorder and the perpetration of family violence (Dutton & Hart, 1992), we conducted more specific analyses aimed at examining the link between type of diagnosis and family violent behaviour. We included the categories of psychotic, anxiety/mood disorders, and personality disorders in the analyses. Unfortunately, our data was not sufficiently detailed to allow for a categorization for borderline personality disorder *****. Family violence perpetration was related to each of the mental diagnoses variables, suggesting little differentiation on the basis of diagnosis. We caution, however, that the classification of diagnoses was not based on a systematic diagnostic scheme such as the DSM. Rather, we coded any reference to diagnostic categories based on psychology reports on file.

***** According to the DSM-III-R definition, the essential features of borderline personality disorder are: "... pervasive pattern of instability of self-image, interpersonal relationships, and mood, beginning by early adulthood and present in a variety of contexts(American Psychiatric Association, 1987, 346).

In Table 20 we present a series of breakdowns of family violence perpetration by criminal history variables. The variables included number of convictions, violation of community supervision, prior federal admissions, sentence length, assaults against non-family members, and major offence type for the current admission. For the most part, criminal history was unrelated to abuse against female partners with the exception of number of past convictions. Men who had 15 or more convictions were slightly more likely to have assaulted their female partners than men who had fewer than 15 convictions. Criminal history was related to the occurrence of child abuse, however. Men with fewer convictions were more likely to have file evidence of child abuse than men with more extensive criminal histories. Men who had not violated previous community supervision conditions and who had not assaulted non-family members were also significantly more likely to have been identified as abusive toward their children. This may reflect the tendency for incest offenders to have less previous criminal justice involvement than other types of offenders.

Table 20 Criminal History Correlates of Family Violence

Correlates Any Family
Violence %
Violence Against
a Female Partner %
Violence Against
Childern %
Number of Past Convictions (15 or More)
Yes 31.5 31.3* 6.1***
No 33.4 24.6 18.6
Mandatory Supervision or Parole Revoked/ Suspended
Yes 29.3* 27.3 4.3***
No 34.6 28.2 17.7
Prior Admission
Yes 31.5 27.4 13.0
No 33.3 28.2 12.5
Aggregate Sentence
3 years and over 32.7 28.3 12.2
Under 3 years 32.4 27.3 13.4
Assaults Against Non- Family Members
Yes 31.0 29.5 6.7***
No 35.6 24.8 23.8
Offence Type
Property 33.8 27.9 14.5
Drugs 31.5 27.7 12.5
Assault 31.3 28.5 12.3
Sexual 32.6 25.7 9.5
Other 36.1 29.2 14.8