Introduction

Exposure to intimate partner violence (IPV) is a widespread problem that has significant impact on the health and well-being of children [43]. A recent meta-analysis found that the effect of exposure to IPV on children’s externalizing difficulties strengthened over time, rather than decreased [43]. Externalizing problems such as aggressive behavior may interfere with school-aged children’s peer relations and academic performance at school, and may also increase the risk for dating violence [21]. Given that children exposed to IPV are at greater risk for aggressive behaviour overall, it would seem logical that they are also at greater risk for aggression with their siblings.

Aggressive behaviour between siblings in childhood is commonplace [10]. As children grow older, physical aggression tends to decline, although a recent national survey in the United States found nearly half (46%) of 6–9 year olds engaged in some form of sibling victimization [41]. Chronic sibling aggression has serious negative consequences, such as trauma-related symptoms [15] and depressive symptoms [7]. Given the pervasiveness of sibling aggression and its potential for deleterious outcomes, it is surprising that it has not been studied more extensively in children exposed to IPV [22, 44]. This important topic has begun to be addressed with preschoolers [28] and more recently, with school-aged children [23, 32, 42]. Therefore, our main purpose was to investigate aggression by school-aged siblings exposed to IPV in more depth.

Multi-Informant Approach

We explored similarities and differences in multi-informant reports of aggression in siblings exposed and not exposed to IPV. To date, most research on aggression by children exposed to IPV has been limited to a single reporter, although the source of reporting appears to make a difference in the strength of association between exposure to IPV and children’s externalizing behavior. For example, a recent meta-analysis found there was a stronger association when information about child exposure and aggression was obtained from the same reporter, usually the mother [43]. Single reporter bias may be of particular concern when mothers are the sole reporters of children’s problem behavior due to potential mental health issues such maternal depression, which may negatively bias their perceptions [12]. We addressed this methodological issue by comparing maternal perspectives with those of observers, as well as with each child’s perspective of their own aggressive behavior. To our knowledge, no other work to date has included each child’s unique perspective on aggression in siblings exposed to IPV. In addition, we also investigated risk and protective factors that may influence child aggression.

Risk and Protective Factors

Developmental psychopathology provides a useful theoretical framework for understanding risk and protective factors that may influence child aggression, as it focuses on the origins and pathways of individual patterns of behavior maladaptation [11]. This perspective is uniquely suited to explain how some children exposed to IPV may be more aggressive by giving equal attention to risk factors that increase the likelihood of negative outcomes, as well as protective factors that either lessen or eliminate adjustment difficulties. Both may operate differently within differing contexts, as the degree and nature of their influence is often context dependent [11]. Developmental pathways can be seen as an unfolding process of how risk and protective factors emerge, interact and fade in importance at differing times and across differing contexts. We focused on the context of exposure to IPV, which is known to significantly increase the likelihood of child aggression [16]. To this end, we explored the extent to which child depressive and trauma-related symptoms functioned as potential risk factors, and warmth in mother–child and sibling relationships functioned as potential protective factors for children exposed and not exposed to IPV.

Depressive and Trauma-Related Symptoms

Depressive and trauma-related symptoms are considered to be significant risk factors for aggressive behavior in the general population [14]. Many studies have found a high degree of co-morbidity between depressive symptoms and externalizing problems such as aggression in children exposed to IPV [43]. While sibling aggression in childhood has been associated with later mental health difficulties [3], to our knowledge no work to date has investigated if depressive symptoms are concurrently associated with aggression by siblings exposed to IPV. Therefore, we explored if and how children’s self-reported depressive symptoms were a risk factor for sibling aggression. Strong links have also been found between teacher-reported child trauma-related symptoms and externalizing problems in preschool-aged children exposed to IPV [29] and chronic sibling aggression has been associated with trauma-related symptoms in children aged 2–17 years [15]. In this study, caregivers reported on trauma-related symptoms for younger children, while older children provided self-reports; however, other researchers found self-reports by school-aged children who were both exposed to IPV and abused in the home were not associated with behavior problems [47]. Differences in findings across studies may be due, in part, to different informants. We used maternal reports to explore if and how child trauma-related symptoms were a risk factor for aggressive behavior in the present study.

While a large body of work has demonstrated a consistent relationship between maternal mental health difficulties and child externalizing problems in IPV-affected families [16], very little work has addressed potential linkages between sibling mental health and child externalizing behavior [10]. We explored this understudied phenomenon by investigating if and how younger and older siblings’ depressive and trauma-related symptomatology was a potential influence on their sibling—a perspective that has been recommended but seldom implemented [13].

Maternal and Sibling Warmth

Positive family relationships are widely regarded as a key protective factor children exposed to IPV [20] and maternal warmth in particular has been associated with less aggression in children exposed to IPV [16]. For example, Graham-Bermann and her colleagues found that higher levels of maternal warmth helped to distinguish asymptomatic children from those with more externalizing difficulties in a sample of school-aged children exposed to IPV [17]. Higher maternal warmth is associated with less aggression in non-exposed children as well [27], although comparative work has suggested exposed children are at greater risk [19]. Unfortunately, maternal warmth is not often measured using child report [46], and including the perspectives of more than one child in a family is particularly rare. Therefore, we explored if children’s reports of maternal warmth served as a protective factor for aggressive behavior in both groups, carefully including the perspective of each sibling.

Building upon this body of work, we also assessed sibling warmth. It is well established that higher levels of sibling warmth are related to fewer externalizing difficulties in children not exposed to IPV [13]. However, researchers have cautioned that sibling warmth can sometimes be associated with more externalizing problems for later-born children, particularly between brothers, who may be more likely to model the negative behavior of their older sibling in warmer relationships [13]. Overall, sibling warmth has been shown to be a significant protective factor for sibling aggression in childhood [6], but to our knowledge has not yet been examined in children exposed to IPV. Therefore, we examined the potential for sibling warmth to serve as a protective factor for aggression in both groups.

Summary

The present study had two main objectives. First, we compared multi-informant perspectives on child aggression in families with and without a history of IPV; we expected that exposed siblings would engage in more aggressive behavior than non-exposed siblings overall. Next, we assessed potential risk and protective factors for aggressive behavior. Given the exploratory nature of these analyses, we did not hypothesize differences in the potential role these factors may play for siblings exposed and not exposed to IPV.

Methods

Recruitment

Following Research Ethics Board (REB) approval, 47 families exposed to IPV and 45 families without such a history were recruited from the community using newspaper adverts, posters and mail flyers. Participants were screened on the following criteria: (1) At least 2 siblings of school-age (5–18 years of age) who were willing to participate, and (2) All family members spoke English fluently. For families exposed to IPV, there were two additional criteria: (1) Mothers self-identified as having a history of IPV, and (2) Mothers had or were currently receiving counselling concerning their abuse as required by the REB.

Participants

Mothers

Mothers who were exposed (n = 47) and not exposed (n = 45) to IPV were asked a variety of demographic questions. Those who were and were not exposed self-identified as 62 and 82% European-Canadian, 30 and 11% Indigenous, and 8 and 7% multiracial respectively. Most of those exposed to IPV had a high school education or less (57%) while only 36% of those not exposed did so. Sixty-six percent of family exposed to IPV lived below the urban low-income cut-off (66%) set for a of three or more persons [37] while 42% of those not exposed did so. Most of the non-exposed mothers (64%) but only 47% of exposed mothers were lone parents defined as separated, divorced, widowed, or never married. Mothers in both groups were similar in age (see Table 2).

Mothers who self-reported a history of IPV completed the 8-item Physical Aggression scale of the Conflict Tactics Scale (CTS) [40] concerning their own and their partner’s violent behaviours that occurred within the context of a conflict in the past 12 months. Items were rated on a 7-point scale. Overall, 66% of mothers reported an intimate partner had directed at least one violent behavior towards them in the past year and 68% reported directing at least one violent behavior towards their partner; these behaviours included pushing, grabbing or shoving (66 and 62% respectively), kicking, biting or hitting (50 and 34% respectively), and threatening with a weapon (28 and 15% respectively).

Children

Forty-seven sibling pairs exposed to IPV and 45 who were not exposed were recruited with their mothers. Average ages for older and younger siblings in both groups are reported in Table 2. Age ranges and sex composition for older (range 6–17; 29 male and 19 female) and younger siblings (range 5–15 years; 27 male and 20 female) in the exposed group were similar to those of older (range 6–19; 19 male, 26 female) and younger siblings (range 5–16; 21 male, 24 female) in the non-exposed group. Most mothers (64%) with IPV histories indicated that their children had received some counseling in the past.

Materials and Procedure

Mothers were screened over the phone to ensure eligibility. If criteria were met, mothers chose one of four locations for data collection. Mothers provided written informed consent and each child provided oral assent. Family members were interviewed separately and privately by a female interviewer. Families were paid $75 and given information about community resources.

Observed Sibling Aggression

Children were asked to “wait” in a living-room like setting while their mothers were interviewed. They were provided with a snack and with a variety of materials, including a Trouble Game©, Lego©, Barbie doll©, drawing materials, chalkboard and hand-held videogame. Their unstructured interaction was video-taped for 30 min, and physical and verbal aggression were coded. Physical aggression was defined as any behavior that included hitting, punching, kicking, serious wrestling (pinning someone down), throwing an object at someone but missing and physical threat (e.g., raised fist). Verbal aggression was defined as any behavior that included yelling, name calling, swearing, insulting, or hostile teasing, jokes or comments. Interactions between siblings were coded jointly, not independently. Each coding interval was 30 s long. Inter-rater reliability was assessed by two independent observers coding 25% of the sample of videotapes; coefficient kappa was .81 for physical aggression and .80 for verbal aggression.

Child Reports of Sibling Warmth and Aggression

Both siblings independently completed the Sibling Relationship Interview (SRI) originally developed by Stocker and McHale [39]. It was revised to include behaviours both initiated and received by each sibling [38]. All items consisted of a 4-point Likert scale, ranging from 1 (not ever) to 4 (a lot). Warmth subscales included 5 items initiated by the child and 5 items received from the sibling including mutual play, sharing, affection, sharing secrets, and comforting. The two aggression subscales included 4 items initiated by the child and 4 items received from the sibling including starts fights, mean, hits, and scared, the only item added for the purposes of the present study. Cronbach alphas for warmth subscales ranged from .84 to .86, and for aggression subscales from .71 to .85.

Maternal report of Child Aggressive Behavior

The Externalizing Disorder scale of the Child Behavior Checklist (CBCL) is widely used for assessing externalizing difficulties in children aged 4–18 years [1]. It was completed by mothers twice, once for each sibling. Maternal perceptions of child aggressive behaviour were measured by the aggression subscale, consisting of 20 items on a 3-point scale, ranging from 0 (not true) to 2 (often true). Coefficient alphas ranged from .88 to .92.

Child Report of Maternal Warmth

Children’s perceptions of maternal warmth were assessed by a subscale of the Parental Acceptance Rejection Questionnaire [33]. The Perceived Parental Warmth and Affection Subscale consisted of 20 4-point items ranging from 1 (almost never true) to 4 (almost always true). Coefficient alphas ranged from .91 to .98.

Child Trauma-Related Symptoms

Wolfe and colleagues developed the 20-item Child Behavior Checklist—PTSD (CBCL-PTSD) scale used to assess child trauma-related symptoms [45]. It has good concurrent validity with clinical PTSD status; however, results concerning discriminant validity distinguishing those with and without a clinical PTSD diagnosis was not as strong [34]. Each item uses a 3-point scale with 0 as “not true”, 1 as “sometimes true”, and 2 as “often true”. Mothers completed the scale twice, once for each sibling. It should be noted that 3 of these items overlapped with the Aggression subscale (argues a lot; stubborn, sullen or irritable; sudden changes in moods, feelings); therefore, analyses were conducted with two versions of this trauma scale: one which included these 3 items and one which did not. Cronbach alphas for both versions were similar and ranged from .77 to .84. More details are provided in the results section.

Child Depressive Symptoms

Children under the age of 14 completed the Child Depression Inventory (CDI) [25] while children aged 14 and older completed the short form of the Beck Depression Inventory (BDI) [2] Footnote 1. The CDI has 27-items designed to assess depressive symptoms within the past 2 weeks. Each item is rated on a 3-point scale ranging from 0 (not at all) to 2 (all the time). In the present study, coefficient alpha for the total score ranged from .75 to .85. The short form of the BDI consisted of 13 items rated on a 4-point scale ranging from 0 (not at all) to 3 (all the time). Cronbach alphas ranged from .75 to .83.

Data Analytic Plan

We compared multi-informant reports of aggression in children exposed and not exposed to IPV using correlational analyses and mean comparisons. We investigated potential risk and protective factors for aggressive behavior using regression analyses. In order to maintain the unique perspective of each informant, separate regression analyses were conducted for each informant in both the exposed and non-exposed groups.

Results

Comparison of Multiple Informants

Correlations between continuous variables were assessed by Pearson Product Moment correlations, and correlations between categorical variables were assessed by Spearman correlations.

Children Exposed to IPV

Of the 45 sibling pairs who participated in the unstructured observation, 33 (73%) did not display physical aggression and 28 (62%) did not display verbal aggression during the observation period. Verbal and physical aggression was highly interrelated (r(45) = 0.73, p = .0001) but did not occur frequently (standardized M = 0.89, SD = 2.60 for physical aggression and standardized M = 1.89, SD = 3.93 for verbal aggression); therefore, these variables were summed for further analyses. It should be noted that when combined, 24 (53%) dyads engaged in either physical and/or verbal aggression. Observer reports of aggression were significantly related to maternal reports of younger sibling aggression (see Table 1).

Table 1 Correlations between multiple informants reports of aggression
Full size table

Child reports included aggression they initiated as well as received from their sibling. Reports of received aggression were strongly correlated with reports of initiated aggression for both siblings (younger siblings r(47) = .63, p = .0001; older siblings r(47) = .73, p = .0001). Mean levels of received aggression (M = 62.77, SD = 18.75 for younger siblings; M = 57.05, SD = 14.49 for older siblings) and initiated aggression (M = 64.49, SD = 18.39 for younger siblings; M = 62.05, SD = 16.81 for older siblings) were also similar and were therefore summed for further analyses. Sibling reports of aggression with each other were not related. Maternal reports of aggression were strongly related (r(45) = .68, p = .0001). Interestingly, maternal reports of older sibling aggression were significantly related older sibling reports, but their reports of younger sibling aggression were not related to younger sibling reports. Overall, observer reports were independent of family members’ reports with one exception, and sibling reports were independent of each other.

Children Not Exposed to IPV

Of the 45 sibling pairs who participated in the unstructured observation, 36 (80%) did not display physical aggression and 36 (80%) did not display verbal aggression during the observation period. Observed verbal and physical aggression was highly related (r(45) = .72, p = .0001) but did not occur frequently (M = 0.74, SD = 2.58 for physical aggression and M = 1.41, SD = 4.25 for verbal aggression); therefore, these variables were summed for further analyses. It should be noted that when combined, 12 (27%) dyads engaged in physical and/or verbal aggression. Observer reports were significantly related to older sibling reports of aggression only (see Table 1).

Child reports of aggression received from their sibling was strongly related to reports of initiated aggression (younger siblings r(45) = .57, p = .0001; older siblings r(45) = .81, p = .0001). Mean levels of received aggression (M = 60.0, SD = 18.19 for younger siblings; M = 60.97, SD = 12.37 for older siblings) and mean levels of initiated aggression (M = 63.33, SD = 16.46 for younger siblings; M = 63.01, SD = 12.56 for older siblings) were similar. Therefore, received and initiated aggression was summed for further analyses. Overall, sibling reports of aggression were inter-related, while observer reports were independent of family reports with one exception.

Comparing Aggression in Children Exposed and Not Exposed to IPV

A comparison of the proportion of sibling dyads exposed and not exposed to IPV who engaged in aggression (53 and 27% respectively) was significant (Z = 2.17, p = .03); more siblings exposed to IPV were aggressive.

Before conducting mean level comparisons of aggression across the two groups, they were first compared on demographic characteristics using Chi square statistics for categorical variables and independent t tests for continuous variables. Results indicated that mothers exposed to IPV were significantly less educated, had lower incomes, and were more likely to self-identify as a person of color than non-exposed mothers (see Table 2). Given these differences, an analysis of covariance controlling these variables was conducted to compare mean differences in reported aggression between the two groupsFootnote 2. Standardized means with a range of 0-100 are reported in Table 1. Results indicated that maternal reports of older (F(7,81) = 2.33, p = .03, Ω2 = .06) and younger sibling (F(7,81) = 2.65, p = .02, Ω2 = .01) aggression were significantly lower in exposed families than non-exposed families.

Table 2 Correlations between risk and protective factors
Full size table

Exploring Risk and Protective Factors

We explored potential risk and protective factors with stepwise multiple regression (SMR) analyses assessing first order main effects. Because these risk and protective factors had not been previously examined with this population, and because of the constraints of our smaller sample size, we chose a regression approach that minimized the number of degrees of freedom used (MAXR in SAS) and maximized R2 by comparing all possible combinations of predictor variables in the model until the most optimal parsimonious model was found. These SMR analyses were conducted separately for each reporter of aggression in the exposed and non-exposed groups. We set our level of significance at p < .05; however, given the exploratory nature of our analyses, we identified models and predictors at the level of p < .10, noting this constraint. Depressive and trauma-related symptoms were explored separately in children exposed and not exposed to IPV (see Table 3). Trauma-related symptoms were significantly related in exposed (r(46) = .59, p = .0001) and non-exposed siblings (r(45) = .50, p = .0005). Depressive symptoms were significantly related between exposed siblings (r(45) = .59, p = .0001) but not between non-exposed siblings (r(45) = .20, p = .19).

Table 3 Comparisons of family demographic characteristics
Full size table

We assessed potential protective factors in a similar fashion (see Table 3). Sibling reports of maternal warmth (r(45) = .85, p = .0001) were significantly interrelated, but not reports of sibling warmth (r(45) = .12, p = .40) in children exposed to IPV. In contrast, sibling reports of maternal warmth were unrelated (r(45) = .01, p = .96) but sibling reports of sibling warmth were significantly associated (r(45) = .42, p = .004) in non-exposed children. The two groups differed significantly on maternal education, income, and ethno-cultural background; however, none of these characteristics were significantly related to reports of aggression and were therefore not included in regression analyses (see Table 3).Footnote 3

Children Exposed to IPV

Four of the five regression models were significant (see Table 4). Twenty-three percent of the variance in the Observed Aggression model was accounted for; maternal warmth reported by younger siblings was a significant risk factor. Younger sibling trauma-related symptoms and warmth reported by older siblings tended to be protective at the p < .10 level only. The model for Younger Sibling Reports of Sibling Aggression accounted for 24% of the variance. Younger sibling depressive symptoms were a significant risk factor for this model, as was maternal warmth reported by older siblings. Younger sibling trauma-related symptoms tended to be a risk factor for Older Sibling Reports of Aggression at the p < .10 level only. Maternal Reports of Aggression by Younger and Older Siblings were both significant; each accounted for 59 and 42% of the variance respectively. Younger sibling trauma symptoms and older sibling depressive symptoms were significant risk factors for Maternal Reports of Aggression by Younger Siblings, while older sibling trauma symptoms were a significant risk factor for Maternal Reports of Aggression by Older Siblings. Sibling warmth reported by younger siblings was a significant protective factor in this model.

Table 4 Regressions testing first-order effects of child depressive symptoms, child trauma symptoms and maternal and sibling warmth on reports of aggression in families exposed to IPV
Full size table

Children Not Exposed to IPV

Four of the five regression models were also significant (see Table 5). Only 9% of the variance in the Observed Aggression model was accounted for by older sibling reports of sibling warmth at the p < .10 level only. The model for Younger Sibling Reports of Aggression accounted for 45% of the variance; younger sibling reports of sibling warmth were a significant protective factor, while younger sibling depressive symptoms and older sibling trauma-related symptoms were a significant risk factor in this model. Older sibling trauma-related symptoms tended to be a risk factor for Older Sibling Reports of Aggression at p < .10 only.

Table 5 Regressions testing first-order effects of child depressive symptoms, child trauma symptoms and maternal and sibling warmth on reports of aggression in families not exposed to IPV
Full size table

Models of Maternal Reports of Aggression by Younger and Older Siblings were both significant; each model accounted for 46 and 42% of the variance respectively. For Maternal Reports of Aggression by Younger Siblings, younger sibling trauma-related symptoms were a significant risk factor while maternal warmth reported by younger siblings tended to be protective at p < .10 only. For Maternal Reports of Aggression by Older Siblings, older sibling trauma-related symptoms were a significant risk factor, while younger sibling reports of sibling warmth tended to be protective at p < .10 only.

Discussion

Sibling aggression is a serious issue [10] but is especially concerning for children exposed to IPV, as it may exacerbate their risk for peer aggression at school as well as later mental health difficulties [3]. Using a cross-sectional multi-informant approach, we investigated if children exposed to IPV engaged in more aggressive behavior with their sibling than non-exposed children, and found that reports differed greatly across informants. We also explored potential risk and protective factors for aggressive behavior, and found that depressive symptoms and trauma-related symptoms were significant risk factors in both exposed and non-exposed groups. The role of maternal and sibling warmth was more complex—serving as a protective factor from the perspective of some informants, but as a risk factor from the perspective of others.

Comparing Reporters of Sibling Aggression

A greater proportion of sibling dyads exposed to IPV were observed to engage in aggressive behavior than siblings who were not exposed, although both groups engaged in a similar frequency of aggressive acts on average. Waddell and colleagues [44] also found no differences in observed aggressive behavior of siblings who were and were not exposed to IPV. Using a multi-informant approach, we found that children’s perceptions of the frequency of sibling aggression did not differ across groups while in contrast, mothers of children exposed to IPV reported significantly less aggressive behaviour than mothers of non-exposed children. An important caveat here is that mothers reported on overall child aggression, which included but was not limited to sibling aggression, while children and observers reported on sibling aggression specifically. Our results may have differed if mothers had focused on siblings, given the ubiquitousness of sibling aggression [31].

Our findings are in striking contrast to previous work showing IPV-affected mothers tend to report more negative child behavior than other informants [30]. Maternal mental health issues such as depression may contribute to their more negative outlook; however, maternal mental health was not assessed in the present study. More negative maternal perceptions have been found in shelter-based samples, while our community-based sample included mothers had experienced counselling concerning IPV. More work is needed on the nature of maternal perceptions of sibling aggression in families affected by IPV.

There was a marked degree of independence across informants in the present study, underscoring the importance of including and comparing each unique perspective on child aggression. It is interesting to note that sibling perspectives on aggression were not related in children exposed to IPV, which directly contrasts with siblings who were not exposed, as well as with the literature on siblings in general [13]. This finding suggests that these children perceived a lack of contingency–that their sibling would be aggressive towards them whether they were aggressive or not, which may mirror their IPV experiences. Recent work has shown that some children exposed to IPV have maladaptive perceptions of violent behavior [18]; it is currently unknown how this may apply to sibling relationships. Future work should address this issue more closely.

Depressive and Trauma Symptoms

Drawing upon a developmental psychopathology framework, we investigated depressive and trauma-related symptoms as potential risk factors for aggressive behavior. It is important to emphasize that our analyses were not causal in nature, given our cross sectional design and the concurrent measurement of all variables. Younger sibling depressive symptoms were a significant risk factor for their reports of sibling aggression in both the exposed and non-exposed groups. These findings support an “irritable depression” model, in which depressive symptoms contribute to behavioral difficulties [24]. Depression and aggression do share several common determinants that may account for their association, such as rumination and internalized anger [14]. Future investigation of these determinants may help inform interventions for children exposed to IPV.

Trauma-related symptoms were also a significant risk factor for maternal reports of aggression in both exposed and non-exposed families. Single reporter bias may have accounted for part of these results; however, trauma-related symptoms also predicted child perspectives on sibling aggression. Older sibling trauma-related symptoms were a significant risk factor for both younger and older sibling reports of sibling aggression in non-exposed families. As older siblings are typically more dominant, it makes sense that their trauma-related symptoms would serve as a greater risk factor for aggression than their younger counterparts, given that these symptoms often include negative mood and a low threshold of arousal. These findings parallel previous work linking older sibling hostile temperament, also characterised by irritability and a low threshold for arousal, with greater agonism between siblings [4]. While it may appear surprising that trauma-related symptoms were significant for non-exposed children, it serves as an important reminder that the majority of children in the United States are estimated to experience some form of trauma during childhood, and these experience have been linked to a greater risk for externalizing problems [9].

In families exposed to IPV, maternal reports of trauma-related symptoms were a significant risk factor for maternal reports of aggression by both younger and older siblings. As noted earlier, this finding may be due in part to shared method variance. However, it also supports previous work in which teacher reports of preschool-aged child trauma-related symptoms were significantly associated with teacher reports of aggression and, in fact, mediated the effects of exposure to IPV on aggression [29]. Since common trauma-related symptoms include outbursts of anger, it may be the case that children exposed to IPV engaged in more reactive aggression, often defined as an angry, emotionally dysregulated response to threat or frustration [8]. Taking these forms of aggression into account would be an important avenue for future research.

It is unclear why trauma symptoms served as a significant risk factor for maternal but not child perceptions of aggression in IPV-affected families. Children exposed to IPV may be more likely to experience recurring or chronic trauma, thereby desensitizing them to a greater extent than their non-exposed counterparts. In support of this notion, other work also found that trauma-related symptoms in school-aged children exposed to IPV were not related to externalizing behavior [47]. Both studies used a multi-informant approach that included maternal and child report, as well as a wide age range. Our multi-informant approach highlighted the importance of including children’s perspectives, particularly in families affected by IPV. These children may be more likely to perceive aggression as normal, acceptable or deserved [18], particularly with a sibling. Future work addressing whether exposed children perceive sibling aggression differently than non-exposed children is needed.

Maternal and Sibling Warmth

We also investigated relationship warmth as a potential protective factor. Overall, mother–child warmth appeared to play a greater role in families affected by IPV, but surprisingly as a risk factor rather than as a protective factor. For example, in IPV-affected families, warmth between mothers and older siblings was a significant risk factor for younger sibling reports of aggression, and warmth between mothers and younger siblings was a significant risk factor for observed sibling aggression. This suggests that children exposed to IPV may be more sensitive to parental differential treatment (PDT), in that perceptions of warmth between mother and the other sibling may have contributed to feelings of jealousy or resentment [5]. Previous work has highlighted the importance of adult explanations in reducing the negative effects of PDT [26]; it is currently unknown if these occur in families affected by IPV. In non-exposed families, greater mother-younger sibling warmth was a significant protective factor for maternal reports of younger sibling aggressive behavior. This finding supported previous work also using maternal report [35]. A warmer mother-younger sibling relationship may reduce the likelihood of aggressive behavior through more frequent monitoring, more secure parent–child attachment, or other protective mechanisms [36].

Sibling warmth was expected to play a protective role for both children who were and were not exposed to IPV, but this was moreso the case for non-exposed siblings. For example, sibling warmth reported by older siblings was an important protective factor for observed aggression between non-exposed siblings, but not for exposed siblings. In addition, sibling warmth reported by younger siblings was a significant protective factor for younger but not older sibling reports of aggression in non-exposed families. Therefore, warmer sibling relationships were protective for aggression between non-exposed siblings. Interestingly, warmth reported by younger siblings exposed to IPV was a protective factor for maternal but not children’s reports. This finding again highlights the importance of a multi-informant approach, and the unique perspective of children exposed to IPV. A longitudinal approach that tracks how exposed children’s perceptions of sibling warmth and aggression may differ from non-exposed children is clearly needed.

Conclusions

Our multi-informant approach highlighted the importance of comparing unique perspectives on aggression. More siblings exposed to IPV engaged in observed aggression; children reported similar levels of sibling aggression across groups but exposed mothers reported less aggressive behavior. Depressive symptoms increased the risk for younger sibling reports of aggression in both exposed and non-exposed groups, and trauma-related symptoms increased the risk for maternal reports of aggression in exposed and non-exposed families. Sibling warmth played very different roles across both groups; however, differences in children’s perceptions were perhaps the most striking. Younger siblings exposed to IPV appeared more sensitive to mother-older sibling warmth than non-exposed siblings, and this was linked to sibling aggression. Overall, sibling warmth was not as protective for exposed children as it was for non-exposed children.

Limitations

These findings should be interpreted within the limitations of our study, including a cross-sectional approach which did not capture the direction of effect or causal relations. Exposure to IPV was measured with a heavy focus on physical violence; other forms of IPV and child abuse were not assessed [36]. Other risk factors known to contribute to child aggressive behavior, such as maternal mental health issues, parenting style, or exposure to other forms of violence in the home, community, school or media were not addressed [28]. Although children in the two groups were matched on age and sex, the groups differed on income, education and ethno-cultural background, which may have influenced our results. Our community sample of exposed mothers had attended counselling, which may have influenced our results. Finally, given the smaller sample size and wide span of ages of children included in the present study, our findings were not sensitive to gender or developmental differences in aggressive behavior, and potential interactions between risk and protective factors were not able to be tested. Despite these caveats, the advantages of our multi-informant approach, particularly our inclusion of the perspectives of both younger and older siblings and our observation of sibling interaction, provided a meaningful step towards better understanding aggression between siblings exposed to IPV. These findings can be used to inform interventions about the important influence of sibling relationships in children’s adjustment.

Summary

Significant differences were found between multi-informant reports of aggression by siblings in families with and without a history of IPV. More exposed siblings engaged in observed aggression, while child reports of sibling aggression did not differ across groups. Exposed mothers reported significantly less aggression than non-exposed mothers. Depressive and trauma-related symptoms were risk factors for both exposed and non-exposed children. Child perceptions of mother-sibling warmth increased the risk for sibling aggression in exposed families, while child perceptions of maternal warmth were protective in non-exposed families. Sibling warmth was protective in non-exposed families, but not in exposed families.