Introduction

The Co-occurrence of Family Violence

Presence of intimate partner violence in the family home has been shown to be a significant risk factor for various forms of child abuse and neglect (Browne 1993; Browne and Hamilton 1999; Cox et al. 2003; Rumm et al. 2000; Tajima 2000). Children living with partner violence are at risk of being the direct victims of separate incidents of maltreatment by the parent/s and/or getting caught up in the parental violence. Appel and Holden’s (1998) review demonstrates while children living in the context of spouse abuse are at high risk of physical abuse themselves, the prevalence rates fluctuate dramatically across studies due to methodological issues. For example, retrospective studies conducted with representative community samples provided an estimate base rate of co-occurrence at 6% in the USA, while retrospective reports from clinical samples of abused women or children, using a conservative definition of child abuse, demonstrate a median co-occurrence rate of 40% (Appel and Holden Op.Cit.). Edelson’s (1999) review also reports co-occurrence rates between 30–60% for the majority of the 31 studies considered. Despite methodological discrepancies across studies, the literature clearly demonstrates a considerable overlap between partner and child maltreatment. In addition to child physical abuse, links between wife abuse and child sexual abuse have also been established (Farmer and Pollock 1998; Goddard and Hillier 1993; Hester and Pearson 1998).

Women are often seen as the primary victims of partner violence due to higher injury levels and the initiation of violence for self defence purposes (Saunders 2002). However, approximately 100 research studies have documented rates of partner violence to be equal for both men and women (e.g. Archer 2000, 2002; Fiebert 2001) and Archer (2000) in his meta-analytic review demonstrates that while women are injured more often that men, men constitute approximately one third of those injured. Therefore, when exploring the links between partner and child maltreatment, it is important to consider the role that both mothers and fathers play in the violent interaction. It is evident that the child may be victimised by the perpetrator of partner abuse because that is their usual mode of interpersonal control with all family members (Salzinger et al. 2002). Indeed, McCloskey et al. (1995) found that children living with women who were abused by their male partner are at serious risk of sexual assault from that partner. Additionally, considering the woman as a possible perpetrator of partner maltreatment, Ross (1996) reported that women who abused their male partners were twice as likely to abuse their children.

However, it is not only perpetrators of partner abuse who maltreat their child/ren. Victims of partner abuse must also be considered. Indeed, Straus and Gelles (1990) showed that women abused by their partners were twice as likely to physically abuse their children, than non-abused women. Similarly, Salzinger et al (2002) demonstrated that the presence of partner abuse in addition to existing family stress, increased the chances of child abuse occurring by more than 2.5 times. Examination of who aggressed against the child revealed that both perpetrators and victims of partner abuse were abusive. Indeed, according to mothers’ self reports, they were more likely to be physically aggressive to their child than the domestically violent fathers. However, research needs to differentiate between mothers involved in reciprocal partner abuse and child maltreatment and those who were uni-directional victims of partner abuse and perpetrators of child maltreatment. Indeed, Dixon and Browne (2003) distinguish between models of family violence that detail the mother as a victim of spouse abuse (Paternal and Hierarchical family violence) and those where she is being actively involved in reciprocal spouse maltreatment (Reciprocal family violence). This distinction needs to be considered to gain a more detailed explanation of the link between partner violence and child maltreatment.

Given the co-occurrence between partner and child abuse and the potential involvement and effects on all family members, adopting a holistic approach to the aetiology, maintenance and intervention with violent families is necessary to reduce all forms of abuse and intergenerational transmission of maltreatment. Indeed, some professionals have judged families with co-occurring child and partner abuse to be a higher risk to children, in terms of injury severity and are thus deemed to be more in need of services (Beeman et al. 2001; Browne and Hamilton 1999).

Discriminating Between Perpetrators

Despite the high co-occurrence of child and partner abuse documented in the family violence literature, little attention has been attributed to identifying characteristics that discriminate between those perpetrators who only abuse their adult partner and those who also abuse their child. Research has shown the co-occurrence of partner and child maltreatment to be associated with low socio-economic status, larger household size, higher numbers of family stressors, maternal distress, psychopathology, more caregiver alcohol or drug problems, maternal childhood abuse and poor quality of parent–child relationships, especially with the father (Coohey 2004; Hartley 2002; O’Keefe 1995). Hartley (2002) distinguished co-occurrence in terms of physical child abuse and neglect. Descriptive analysis found that families with spouse abuse and child neglect had significantly fewer biological fathers and more maternal substance abuse and mental health problems, in comparison to families with child neglect only. In addition, families with spouse abuse and child physical abuse had significantly higher paternal substance abuse, mental health problems and criminal convictions/incarceration in comparison to families with child physical abuse only.

To address these issues the present study investigated variables that have been previously shown to be associated with both child and intimate partner abuse in the research literature. It has been argued that family violence is caused and maintained by a number of diverse mechanisms associated with biological, psychological and sociological theories of abusive behavior (Browne and Herbert 1997). Previous typology research has differentiated between men who abuse their female partner using a variety of different theoretically driven etiological variables such as psychopathology, early childhood and peer experiences, attachment styles, impulsivity and attitudes toward violence (Holtzworth-Munroe and Stuart 1994). These factors have also largely been associated with a high risk of child maltreatment (Dishion et al. 1991; Morton and Browne 1998).

Previous research using theoretical approaches derived from social learning and developmental psychopathology needs to be incorporated into an integrated model to explain co-occurrence of partner and child maltreatment and family violence in general (Appel and Holden 1998). This study draws on the literature from these theoretical approaches to explore the presence of variables associated with pathological parenting. Variables monitoring the extent of parents’ childhood maltreatment, juvenile delinquency, adult criminality, history of abusive relationships, psychopathology, adult substance abuse, factors associated with adult mental health problems and high risk parenting are investigated and compared between groups of parents who perpetrate concurrent partner and child maltreatment in comparison to parents who only maltreat their child.

Method

Participants

One hundred and five child maltreatment cases were examined, providing psychological report information on 164 parents (75 men and 89 women) who were individually assessed by a forensic psychology consulting service on their suitability to parent, following allegations of child maltreatment. Participants lived in the English Midlands or South Wales and were assessed by the service between June 1996 and June 2003. All clients provided their written consent for data derived during their assessment to be used anonymously for research purposes.

Psychological reports were available for both parents from the same family in 59 cases and for one parent only in 46 cases. Thus, in total 105 families were considered. The ages of parents ranged from 17–56 years (mean age = 30; SD = 8.14). The age of the index child ranged from 1 month to 15 years (mean age = 4.6, SD = 4.26). Information on ethnicity was only available for 58 (35.4%) parents. Of this sub-sample, 53 (91.4%) parents were classified as white UK, 1 (1.7%) Asian, 1 (1.7%) African–Caribbean and 3 (5.2%) African–Caribbean/White UK.

Ninety seven (59.1%) parents were perpetrating, or looking after their child who was receiving physical and/or sexual abuse and 67 (40.9%) parents were neglecting or looking after their child who was neglected.

Procedure

Psychological reports gave detailed information on childhood, criminal and romantic relationship histories, mental health problems and parenting factors (see Appendix 1). Thus, data are based on the psychological report of each individual client. This report is constructed from interviews with the client and cross-verification of client self report with additional sources, such as medical records, social services, school and police reports and reports from witnesses and family members. In addition, psychometric tests assessing psychopathology (Millon Clinical Multiaxial Inventory—MCMI-III; Millon et al. 1997) and parenting stress (Parenting Stress Index—PSI; Abidin 1995) were available. The ‘Index of Need’ checklist (Browne 1989a, b, 1995; Browne and Saqi 1988) was also completed from the available file information. These tools are described in the measures section.

Parents were deemed to be partners if a level of romantic/intimate attachment was discussed in the report and/or parents were married, cohabiting or living separately. In cases where one or more children were considered to be at risk of child maltreatment, parenting information relating to the child involved in the most recent incident of maltreatment (index child) was considered for the sake of clarity.

In cases where a child or partner suffered multiple forms of abuse or neglect, the most active form of abuse was designated to define abuse type (Browne and Herbert 1997). Thus, sexual abuse overrides physical and neglect; and physical abuse overrides neglect. This follows the ‘coexistence of different forms of maltreatment model’ presented by Browne and Herbert (1997, p11). For the purpose of this study, cases of physical and sexual child abuse are concatenated into one active category of ‘physical and/or sexual child abuse’. These cases may have suffered multiple forms of abuse, including neglect. Cases of neglect were classified as passive ‘child neglect’; in these cases the child will not have suffered any other forms of abuse.

Content analysis of the psychological reports was conducted using a standardised proforma, designed to extract specific and reliable information. Three independent raters extracted information associated with a risk of family violence. Additional demographic information was also collected. Variables were recorded as present or absent. With regard to the psychometric measures, MCMI-III sub-scales were grouped into three clusters of personality disorder outlined by DSM-IV [American Psychiatric Association (APA) 2000]. These clusters are ‘odd or eccentric’ (Cluster A); ‘dramatic or emotional’ (Cluster B); ‘anxious or fearful’ (Cluster C) and a severe clinical syndrome scale which included the presence of thought disorder, major depression or delusional disorder. As per MCMI guidelines, any profile that was not valid (according to the validity scale) was disregarded and a Base Rate score of 75 was used as the criteria to indicate the presence of personality traits and a severe clinical syndrome.

PSI subscales were also recoded to represent the presence or absence of a score elevated above the 75th percentile (i.e. the cut-off for clinical significance) on the ‘Child Domain’ and ‘Attachment’ subscale. The ‘Child Domain’ score provides a representation of the parent’s perception of the child characteristics. The ‘Attachment’ sub-scale is indicative of the type of attachment between the parent and child.

To ensure the reliability of data collection, variables were systematically extracted from reports using definitions outlined in the coding dictionary (Appendix). Inter-rater and intra-rater reliability was measured to assess the validity and reliability of data obtained using the standardised proforma. Each rater completed the proforma for the same two parents at two different points in time. Agreement between the three researchers reached a 100% concordance for inter-rater reliability for each variable measured. Agreement within each individual rater over time also met a 100% concordance rate for intra-rater reliability.

Measures

Millon Clinical Multiaxial Inventory (MCMI-III)

The MCMI-III (Millon et al. 1997) is a self report, 175 item questionnaire. This psychometric measure is based on the DSM-IV classification system (APA 2000) and provides clinicians with information on 14 personality disorders (11 clinical personality patterns and 3 severe personality pathology) and 10 clinical syndromes (7 clinical syndromes and 3 severe syndromes), for adults undergoing psychological or psychiatric assessment or treatment. In addition, three modifying indices and one validity index are incorporated into this test. A base rate score of 75 indicates the presence of a personality trait or clinical syndrome, a score of 85 or above indicates the presence of a personality disorder or prominence of a clinical syndrome.

Parenting Stress Index (PSI; Abidin 1995)

The PSI is a parent self-report, 101-item questionnaire, designed to identify potentially dysfunctional parenting and parent child interactions. An optional 19-item Life Events stress scale is also provided. This instrument measures two areas: child domain and parent domain. The child domain is divided into distractibility/hyperactivity, adaptability, reinforces parent, demandingness, mood and acceptability. The parent domain is divided into competence, isolation, attachment, health, role restriction, depression and spouse. Scores above the 75th percentile are considered to represent clinical significance.

The Index of Need

The ‘Index of Need’ (depicted in Appendix) is a weighted checklist that measures the presence or absence of 14 risk factors for child maltreatment within the family (Browne 1989a, b, 1995; Browne and Saqi 1988). A total score is derived from presence and absence of each factor. Scores of 6 and above are considered to reflect at risk parenting.

Definitions of Acronyms

Parents were designated to specific groups for analytic comparison. Group titles are referred to by acronyms, which are defined below;

  • Father perpetrator of child maltreatment and perpetrator of intimate partner violence (Paternal Family—PF)

  • Mother perpetrator of child maltreatment and perpetrator of intimate partner violence (Maternal Family—MF)

  • Father perpetrator of child maltreatment only (Paternal Child—PC)

  • Mother perpetrator of child maltreatment only (Maternal Child—MC)

  • Father victim of intimate partner violence and perpetrator of child maltreatment (Paternal Victim—PV).

  • Mother victim of intimate partner violence and perpetrator of child maltreatment (Maternal Victim—MV)

  • Father did not maltreat the child but lived with the mother who did (Paternal Non-abusive Carer—PNC)

  • Mother did not maltreat the child but lived with the father who did (Maternal Non-abusive Carer—MNC)

Results

Group Membership

All cases of intimate partner violence were characterised by physical abuse, with the exception of two cases, in which the mother received psychological abuse only. These cases were not included in further analysis to ensure consistency in abuse type across cases. Number of parents classified by each of the stipulated groups is depicted in Fig. 1.

Of the 22 parents classified as Non-abusive Carers, 3 (42.9%) of the PNC cases were characterised by the father physically abusing the mother, 1 (14.3%) by reciprocal partner violence and the remaining 3 (42.9%) by the mother maltreating the child only. Of the MNC cases, 6 (40%) were characterised by the mother being abused by the father, 1 (6.7%) by the mother perpetrating partner violence against the father, 2 (13.3%) by reciprocal partner violence and 6 (40%) by the father maltreating the child only. Demographic information and the type of child maltreatment perpetrated by parents characterised by each group is shown in Table 1.

Fig. 1
figure 1

The number of parents classified into groups of ‘Family Maltreatment’, ‘Child Maltreatment’, ‘Victims of Intimate Partner Violence’ and ‘Non-abusive Carers’ (n = 162)

Full size image
Table 1 Demographic information for each group of family violence
Full size table

Rates of Concurrent Family Violence

Examining results from the perspective of the individual perpetrator, 66 parents perpetrated violence to both their partner and child, providing a 40.7% co-occurrence rate within this sample. However, from a holistic family perspective 104 (64.2%) parents experienced partner and child maltreatment in their home, either as a result of them or their partner conducting both types of maltreatment concurrently within the family or because both parents conducted one type of maltreatment (intimate partner violence or child maltreatment) each within the family.

Group Comparisons

Examining sex differences

Bivariate analysis was conducted to examine if fathers (n = 75) were significantly more likely to be assigned to a group in comparison to mothers (n = 87; using an alpha criterion = 0.0125 to correct for type 1 error across 4 tests). Fathers were significantly more likely to conduct concurrent forms of maltreatment within the family (PF; n = 43, 57.3%) than mothers (MF; n = 23. 26.4%), (\( \chi ^{2}_{1} = 7.104 \), p = 0.008). In addition, mothers were significantly more likely to be classified as victims of intimate partner violence (MV; n = 23, 26.4%) than fathers (PV; n = 2, 2.7%; \( \chi ^{2}_{1} = 17.438 \), p = 0.000). No significant differences resulted between non-abusive carers (PNC; n = 7, 9.3% and MNC; n = 15, 17.3%) or perpetrators of child maltreatment only (PC; n = 23, 30.7% and MC; n = 26, 29.9%).

Examining Group Differences Between Mothers

Demographic Information

No significant differences were found between the ages of parents or index child in MF, MC, MV and MNC groups.

Significant differences in the marital status within each group were found (using an alpha value of 0.016 to correct for Type 1 error across 3 tests). MF and MV mothers were significantly more likely to cohabit than live separately (\( \chi ^{2}_{1} = 13.8 \), p = 0.000). MNC mothers were significantly more likely to cohabit than live separately and be married than live separately (Fishers Exact = 0.010, p < 0.016). No other significant differences resulted.

No significant differences were found between number of MF, MC, MV and MNC mothers who cohabited, were married or lived separately from their partner (using an alpha value of 0.008 to correct for Type 1 error across 6 tests).

Type of Child Maltreatment Perpetrated

As MNC mothers did not perpetrate child maltreatment this group was not included in the analysis. The frequency with which mothers in MF, MC and MV groups perpetrated active or passive forms of child maltreatment within each group was examined. MC and MV mothers were significantly more likely to neglect the index child than physically/sexually abuse him or her (\( \chi ^{2}_{1} = 7.692 \), p = 0.006 and \( \chi ^{2}_{1} = 7.043 \), p = 0.008 respectively). MF mothers did not differ significantly in their form of passive or active abuse.

Frequency with which parents perpetrated active or passive forms of child maltreatment between MF, MC and MV groups was also analysed. No significant differences were found (using an alpha value of 0.016 to correct for inflated type 1 error across 3 tests).

Comparison of Group Characteristics

Prevalence of characteristics for MF, MC, MV and MNC mothers are shown in Table 2. Where 4 × 2 Chi square tests could not be conducted, due to low expected frequencies in cells, Fisher exact tests were ran to determine if any significant differences between groups existed, using an alpha criterion = 0.008 to correct for type 1 error across 6 tests. Where this is necessary the range of Fisher Exact tests are presented in the Test Statistic column of Table 2.

Table 2 Characteristics differentiating mothers in Maternal Family (MF), Maternal Child (MC), Maternal Victim (MV) and Maternal Non-abusive Carer (MNC) groups
Full size table

Significant differences resulted between groups for ‘current relationship difficulties’ (\( \chi ^{2}_{3} = 25.80 \), p = 0.000), ‘residing with a violent adult’ (\( \chi ^{2}_{3} = 20.60 \), p = 0.000) and ‘single parenthood’ (Fisher Exact range = 0.007–1.000). Further post hoc analysis demonstrated that MF and MV mothers have a significantly higher prevalence for current relationship difficulties than MC mothers (\( \chi ^{2}_{1} = 18.32 \), p = 0.000 and \( \chi ^{2}_{1} = 16.05 \), p = 0.000 respectively). MF and MV mothers were significantly more likely to live with a violent adult (\( \chi ^{2}_{1} = 16.61 \), p = 0.000 and \( \chi ^{2}_{1} = 8.846 \), p = 0.003 respectively) in comparison to MC mothers. Additionally, MC mothers were significantly more likely to be a single parent than MNC mothers (Fishers Exact = 0.007, p < 0.008).

Examining Group Differences Between Fathers

PV and PNC groups were not included in further statistical analysis as they were deemed unsuitable due to their small sample size. Thus, comparisons of PF and PC groups were conducted.

Demographic Information

No significant differences were found between the ages of parents or index child in PF and PC groups.

Significant differences in the marital status within each group were found (using an alpha value of 0.016 to correct for Type 1 error across 3 tests). PF fathers were significantly more likely to cohabit than live separately (\( \chi ^{2}_{1} = 16.298 \), p = 0.000) and to be married than to live separately (\( \chi ^{2}_{1} = 11.972 \), p = 0.001). PC fathers were significantly more likely to cohabit than live separately (\( \chi ^{2}_{1} = 13.8 \), p = 0.000 respectively). No other significant differences resulted.

No significant differences were found between the number of PF fathers who cohabited, were married or lived separately from their partner in comparison to PC fathers.

Type of Child Maltreatment Perpetrated

Frequency with which fathers perpetrated active or passive forms of child maltreatment within each group was examined. PF fathers were significantly more likely to physically and/or sexually abuse the index child than neglect him or her (\( \chi ^{2}_{1} = 20.512 \), p = 0.000). PC fathers did not differ significantly in their form of passive or active abuse.

No significant differences were found in the frequency with which fathers perpetrated active or passive forms of child maltreatment between PF and PC groups.

Comparison of Group Characteristics

Prevalence of characteristics for PF and PC fathers are shown in Table 3. PF fathers have a significantly higher prevalence for a childhood abuse history (\( \chi ^{2}_{1} = 7.07 \), p = 0.008), factors associated with juvenile delinquency (juvenile substance abuse; \( \chi ^{2}_{1} = 10.53 \), p = 0.001), criminal history (convictions for violent/sexual offence; \( \chi ^{2}_{1} = 11.85 \), p = 0.001 and convictions for non-violent criminal offences; \( \chi ^{2}_{1} = 9.16 \), p = 0.002) adult substance dependency (\( \chi ^{2}_{1} = 5.21 \), p = 0.022), current relationship difficulties (\( \chi ^{2}_{1} = 34.0 \), p = 0.000), mental health factors (dramatic emotional personality cluster; \( \chi ^{3}_{1} = 10.63 \), p = 0.001), residing with a violent adult (\( \chi ^{2}_{1} = 26.71 \), p = 0.000) and index of need score which reflects ‘at risk’ parenting (\( \chi ^{2}_{1} = 26.71 \), p = 0.000). PC fathers have a significantly higher prevalence of having a physically/mentally disabled child (Fishers Exact = 0.029, p < 0.05) and scores above the 75th percentile on the PSI subscales of child domain and attachment (\( \chi ^{2}_{1} = 5.01 \), p = 0.025 and \( \chi ^{2}_{1} = 5.64 \), p = 0.018 respectively).

Table 3 Characteristics differentiating fathers who perpetrate family maltreatment and child maltreatment (PF v PC)
Full size table

Sex Comparisons

Frequency of Uni-directional Intimate Partner Violence

For the majority of MF cases (n = 21; 91.3%) reciprocal partner maltreatment occurred, with only two cases (8.7%) characterised by female uni-directional abuse. However, in PF cases 19 (44.2%) were characterised by reciprocal abuse and the remaining 24 cases (55.8%) by male uni-directional violence. PF fathers were significantly more likely to administer uni-directional partner abuse in their relationship than MF mothers (\( \chi ^{2}_{1} = 13.934 \), p = 0.000).

Perpetrators of Family Maltreatment (PF and MF)

Statistical analysis examined characteristic differences between PF and MF and PC and MC groups, thus criterion alphas were lowered to 0.025, using the Bonferroni correction procedure, to correct for type 1 errors across two tests.

Demographic Information

No significant differences were found between the age of parents, index child or marital status of PF and MF parents.

Type of Child Maltreatment Perpetrated

Frequency with which parents perpetrated active or passive forms of child maltreatment between PF and MF parents was analysed. No significant differences were found.

Comparison of Group Characteristics

Table 4 demonstrates the prevalence of characteristics for each group and highlights significant differences. PF fathers showed a significantly higher prevalence of factors associated with juvenile delinquency (juvenile substance abuse; \( \chi ^{2}_{1} = 5.19 \), p = 0.023; fighting with peers at school; \( \chi ^{2}_{1} = 7.44 \), p = 0.006) and criminal history (convictions for violent/sexual offence; \( \chi ^{2}_{1} = 8.13 \), p = 0.004 and convictions for non-violent criminal offence; \( \chi ^{2}_{1} = 11.35 \), p = 0.001) in comparison to MF mothers. MF mothers displayed a significantly higher prevalence of mental health factors (previous suicide attempt/ideation; \( \chi ^{2}_{1} = 9.33 \), p = 0.002 and presence of a severe clinical syndrome; Fishers Exact = 0.024) and the parenting risk factor of feelings of isolation (Fishers Exact = 0.014).

Table 4 Characteristics differentiating mothers and fathers who perpetrate family maltreatment (PF v MF) or child maltreatment (PCvMC)
Full size table

Perpetrators of Child Maltreatment (PC and MC)

Demographic Information

No significant differences were found between the age of parents, index child or marital status of PC and MC parents.

Type of Child Maltreatment Perpetrated

Frequency with which parents perpetrated active or passive forms of child maltreatment between PC and MC groups was analysed. PC fathers showed a trend for being more likely to physically and/or sexually abuse their child than MC mothers (\( \chi ^{2}_{1} = 4.469 \), p = 0.035).

Comparison of Group Characteristics

MC mothers had a significantly higher prevalence of childhood physical and/or sexual abuse (\( \chi ^{2}_{1} = 6.53 \), p = 0.011), mental health factors (‘odd eccentric’ personality cluster; \( \chi ^{2}_{1} = 6.29 \), p = 0.012 and presence of a severe clinical syndrome; \( \chi ^{2}_{1} = 5.94 \), p = 0.015) and parenting risk factors (single parenthood; Fishers Exact = 0.002 and Total Index of Need score; t 45 = −2.85, p = 0.007) in comparison to PC fathers. PC fathers were significantly more likely to have no biological relation to the index child that they maltreated (Fishers Exact = 0.004).

Discussion

This study considers intimate partner violence and child maltreatment from the perspective of the individual perpetrator. It was found that two in five parents (40%) perpetrated both partner and child maltreatment within the family unit, corroborating previous literature, which demonstrates high co-occurrence rates for both forms of violence (Appel and Holden 1998; Cox et al. 2003; Edleson 1999; Falshaw and Browne 1997). However, examination of rates by gender showed that fathers were significantly more likely to conduct both forms of maltreatment than mothers (57 and 26% respectively) and mothers were significantly more likely to be victims of partner violence than men. This demonstrates that while fathers and mothers do commit both forms of maltreatment within the family, fathers do so with greater frequency.

Group Comparisons

As significant differences between mothers were centered on relationship difficulties, it is possible that relationship factors need to be measured more closely when considering variables associated with mother’s concurrent family violence. However, use of relationship difficulties as a discriminating group factor is limited, as it is unknown whether relationship difficulties encouraged partner abuse or resulted as a consequence of the abuse. Additionally, mothers who abused their child only (MC) were significantly more likely to be a single parent in comparison to mothers who did not abuse their child but lived with a parent who did (MNC). Thus, in accordance with previous research it could be postulated that the stress of being a single parent for MC mothers increases the chances of them maltreating their child, while a two parent family acts as a protective factor for MNC mothers (Browne and Saqi 1988; Cerezo et al. 1996; Crouch et al. 2001; Dixon et al. 2005).

Fathers who perpetrated both partner and child maltreatment (PF) were significantly more likely to have experienced factors associated with developmental psychopathology and criminogenic lifestyle in comparison to fathers perpetrating child maltreatment only (PC). Of particular interest, PF fathers were significantly more likely to present with ‘dramatic/emotional’ personality traits. Thus, PF fathers demonstrate similar personality characteristics to the Generally Violent/Antisocial and Dysphoric/Borderline subtypes of partner abusing men proposed by Holtzworth-Munroe and Stuart (1994). This is consistent with previous research, which has found the ‘Dysphoric/Borderline’ subtype to have the highest child abuse potential (Herron and Holtzworth-Munroe 2002). However, unlike Holtzworth-Munroe and Stuart (1994), this study does not distinguish between antisocial and borderline personality traits and only considers the presence of a trait rather than a disorder; thus, it is not possible to draw conclusions about specific forms of psychopathology. Additionally, PC fathers had a significantly higher prevalence of parenting stress factors in comparison to fathers who maltreated both child and partner. Thus, father’s negative perceptions and insecure attachment to their child is associated with child maltreatment more frequently for these fathers.

Sex Comparisons

PF fathers are significantly more likely to engage in physical and/or sexual child maltreatment than neglect, while MC and MV mothers are significantly more likely to neglect him or her. Thus, fathers who maltreat both their partner and child within the family unit are likely to do so in a physically aggressive manner.

Characteristics that significantly distinguish PF and MF groups are PF father’s higher prevalence of factors associated with an antisocial lifestyle and MF mother’s higher prevalence of factors associated with mental health problems and feelings of isolation. Therefore, findings demonstrate that men characterised by high levels of antisocial/criminal behavior are most likely to engage in concurrent forms of family violence in addition to extra-familial aggression. Furthermore, discrimination of PC and MC parents found that MC mothers also had a significantly higher prevalence for characteristics associated with mental health problems and a childhood abuse history. Thus, in this study, maternal mental health is associated with the perpetration of child maltreatment for mothers classified by both co-occurring and child maltreatment only families, in comparison to fathers from the same family pattern.

Additionally, MC mothers were significantly more likely to be single parents than PC fathers. Thus, mothers were more likely to report that they reared their child alone, despite having a romantic partner. As previously stated, this factor is consistently associated with child maltreatment in the literature (Browne and Saqi 1988; Cerezo et al. 1996; Crouch et al. 2001; Dixon et al. 2005).

Practical Implications

It is evident from this study that both mothers and fathers can aggress against their partner, child or both. Therefore, this lends support for the need to explore violent families from a more holistic perspective in both research and practice, considering the overlap of child and partner maltreatment and the effects of intimate partner violence upon all members of the family rather than exclusively considering the violent man.

An integrated perspective of child and partner abuse will increase interagency collaboration and integrative treatment for the family. As Osofsky (2003): states “the necessary integration of this perspective into the work of law enforcement, the judicial system and social service providers has not yet occurred” (p161). Indeed, research examining police recognition of the links between spouse and child abuse demonstrated a lack of referral between Child Protection Units and Domestic Violent Units (Browne and Hamilton 1999), highlighting a partnership gap. The Police are in a position to aid the prevention and intervention of child maltreatment by providing child protection professionals with information on the criminal background of a parent who has aggressed against their intimate partner. In relation to the findings of this study, a father who aggresses against his female partner and has a serious history of developmental psychopathology and criminogenic lifestyle would be at high risk of physical child maltreatment, and thus this information should be passed on to child protection for further investigation.

Furthermore, findings showed that mothers who maltreated their child were more likely to have mental health problems than fathers. Although child maltreatment is not an inevitable product of parental mental illness, evidence suggests that some parents cannot meet the needs of their children due to mental health problems, which may be associated with partner abuse (Browne and Herbert 1997). Consequently, these parents have a greater risk of their child being removed into care (Sheppard 1997). Therefore, results highlight the need for practitioners to be aware of (a) the role that parental mental health problems have in increasing the risk for child maltreatment and (b) the need for interagency collaboration between adult and child mental health and social services (Falkov and Davies 1997; Jolley and Maitra 2000). Indeed, Reder and Duncan (1999), in their study of fatal child abuse, emphasize need for such collaboration, encouraging liaison both within and between health and social services, rather than encouraging specialists to focus on meeting the needs of one specific client group. Additionally, the Royal College of Psychiatrists (2002) has acknowledged that care in the community has resulted in an increasing number of adults who are treated for psychiatric disorders while living with their families and children and thus emphasize the need for psychiatrists to work closely and effectively with other services. For example, health visitors working in primary care are well placed to determine parental risk profiles for child maltreatment and family dysfunction and consequently carry out early intervention and/or refer families for more detailed assessment or prevention work such as to community mental health teams.

Implications for considering the overlap of family violence extend to custody cases and visitation rights during the legal proceedings of abuse allegations or relationship breakdowns. Examining domestic violent offenders within the context of the family as a whole is important if cycles of aversive family interactions are to cease. For example, it is important to accurately understand and assess the risk that a spouse-abusing male will pose to his children or the risk that a victimised female will pose to her children post-separation from the violent partner.

Finally, an integrated approach will empirically inform the design of prevention and treatment programmes for men and women who abuse within the family. This study shows the importance of examining an offender within the context of their family, in order to understand the aetiology and maintenance of violence. Using this approach, it should be apparent that parents who maltreat their child can have very different treatment needs. For example, just fewer than fifty percent of MF and PF parents were in a reciprocally violent relationship with their partner in addition to maltreating their child. In such families the mother’s partner violence needs to be addressed, in addition to the father’s, rather than simply viewing her as a victim of his violence. For those couples that wish to stay together, intervention may focus on relationship counselling or family therapy in addition to parenting skills and programmes that will address their aggression, such as anger management. This is in contrast to parents who maltreat their child only. This study found factors of single parenting, negative perceptions of the child and insecure attachments with the child to be associated with their maltreatment. Therefore, it is plausible that intervention focused on an increase in social support and parenting skills would better address their child maltreatment. The reliable identification of risk factors associated with perpetrators of child and partner abuse or child abuse only is necessary to inform such practice.

Methodological Considerations

Present findings are based on cross-sectional, non-randomised data, making generalisations to the wider population difficult. Indeed, the nature of the sample can moderate the findings of studies of family violence. Populations selected with high rates of male aggression are likely to report extreme male violence in comparison to community samples or young dating couples (Archer 2002; Johnson 1995). In addition, the lack of control groups limits the interpretation of the present findings. Groups of ‘non-maltreating/at risk of child maltreatment’ and ‘non-maltreating/not at risk of child maltreatment’ parents are needed as comparison groups to accurately determine group differences.

In this study, severity or context of the partner violence was not known and it was not possible to determine who the main perpetrator of intimate partner violence was. It is plausible that mothers were acting in self-defence or were less severe in their actions (Archer 2002). Additionally, frequency of mothers involved in performing acts of physical violence against a partner is determined by self-report of the client and where possible corroborated by additional evidence. Thus, parents may have exaggerated the presence of aggressive acts by their partner, especially if they had a vested interest to present to the courts in a favourable light in order to gain rights over access to a child. Indeed, research has shown that people are more likely to report partner violence than their own violence (Riggs et al. 1989).

Conclusion

Findings of this study demonstrate the importance of adopting a holistic perspective to family violence, considering the effects of partner violence upon all members in a violent family, rather than exclusively considering the abusive man, who has been the primary focus of research examining the perpetration and prevention of domestic violence. While the study supports the high co-occurrence of partner and child maltreatment in violent families, and demonstrates that fathers are significantly more likely to perpetrate concurrent forms of abuse than mothers, it is evident that mothers do aggress against their partner, child or both.

These findings support researchers who assert that general samples can provide evidence of both men and women being physically aggressive in intimate relationships (Archer 2002; Johnson 1995; Straus 1997). The present study has extended this concept to the wider family. It is demonstrated that mothers who perpetrate or are victims of partner violence may also maltreat their child, using active or passive forms. However, claims of mutual abuse must be interpreted with a full understanding of women’s use for violence (Renzetti 1999), as exploration of perpetration by females often ignores the context and consequences of these assaults. Indeed Straus (1995) found that the injury women receive requires them to seek medical attention seven times more often, while other research has demonstrated that wives usually instigate aggression for self defence purposes (Dobash et al. 1992: Saunders 1986). However, as Archer (2002) asserts, considering women as victims of partner violence is too narrow and addressing the issue of female violence does not need to detract from the intervention and prevention of abuse against women (Archer 2002).