The Influence of Internalizing Symptoms and Emotion Dysregulation on the Association Between Witnessed Community Violence and Aggression Among Urban Adolescents

Abstract

The current cross-sectional study examined whether internalizing (i.e., anxiety and depressive) symptoms and/or emotion dysregulation moderated the association between witnessed community violence and aggressive behavior. Participants were 180 predominantly African American adolescents (62% girls; M age = 15.87 years, SD = 1.19 years) from a high school located in an urban community in the United States. Approximately 95% of adolescents reported having witnessed at least one violent act during their lifetimes, with many endorsing repeated exposure to severe acts of community violence. Results indicated that emotion dysregulation exacerbated the association between witnessed community violence and aggression. A quadratic effect of anxiety symptoms also moderated this association, such that witnessed community violence was linked to aggression at low and high, but not moderate, levels of anxiety symptoms. In contrast, a quadratic effect of depressive symptoms was uniquely related to aggression, regardless of witnessed community violence. Directions for future research and implications for practice are reviewed.

Introduction

Community violence has long been regarded as a major public health problem in the United States [1] and in countries across the world [2]. Several decades of research have demonstrated that community violence exposure, whether experienced directly or indirectly, is associated with myriad negative mental health outcomes among children and adolescents, including internalizing and trauma-related symptoms [3], academic problems [4], substance use [5], and suicidal ideation [6]. Extant evidence also indicates that violence tends to beget more violence; that is, witnessed community violence has been consistently linked to higher levels of aggression across developmental periods, including early [7] and middle childhood [8,9] as well as adolescence [10,11]. Yet, despite this clearly established association, relatively little is known regarding what malleable factors may heighten risk for engagement in aggressive behavior among youth who witness high levels of community violence. The current study sought to advance this literature by examining whether internalizing (i.e., anxiety and depressive) symptoms and/or emotion dysregulation would exacerbate the association between witnessed community violence and aggression during adolescence.

Community violence disproportionately impacts urban communities in the United States. Available estimates indicate that since 2010, approximately 90% of all violent crimes have occurred in metropolitan areas [12]. Previous research has revealed that urban youth in particular are exposed to violence in their communities early in life, as alarmingly high rates of community violence exposure have been reported among preschool-age children [13]. For example, one investigation found that 78% of 3 ½- to 4 ½-year-old inner-city children had witnessed or been the victim of at least one violent act in their community [14]. Further, prevalence rates and frequencies of witnessed community violence tend to increase with age. By adolescence, a substantial proportion of urban youth have witnessed severe acts of community violence, including stabbings (i.e., rates range between 18% to 56%) and/or shootings (i.e., rates range between 20% and 66%) [15]. Fowler and colleagues [3] reported that “violent communities have been compared to war zones in which there is no foreseeable end to the combat…this violence can pervade the lives of everyone in the neighborhood, causing youths to continually hear about and witness the victimization of family, friends, and neighbors” (p. 248).

Witnessed community violence is a robust predictor of aggressive behavior among children and adolescents [4,6–11,16]. The progression from witnessed community violence to aggression may be explained from a social information processing (SIP) theoretical framework. According to a reformulation of the SIP model [17], individuals are constantly engaged in a reciprocal series of steps during their social interactions that involve the encoding and interpretation of situational and internal cues, selection of goals or desired outcomes for the situation, access of possible responses to the situation from memory, evaluation and selection of the possible responses, and behavioral enactment of the response(s). Negatively-biased patterns of SIP at each of these steps may lead to various forms of social maladjustment, including aggression [17]. Indeed, extant evidence indicates that SIP deficits play a central role in the development of aggression among children and adolescents exposed to community violence. That is, previous research has demonstrated that witnessed community violence is associated with hostile attribution biases, such that youth may become hypervigilant to cues of threat and automatically attribute hostile intent to others in ambiguous situations [10,18]. Further, witnessing high levels of violence in the community serves as a powerful learning context for children and adolescents, which may lead them to believe that aggression is an appropriate, justifiable, and normative response to provocation [8–11]. Witnessed community violence has also been linked to aggressive response generation [10], cognitive rehearsal of aggressive actions [8], and positive outcome and efficacy beliefs for aggression [9,19]. Of note, Guerra and colleagues [8] have suggested that, in the context of community violence, certain individual-level factors may heighten the likelihood that youth will experience psychosocial maladjustment; however, it is not yet clear what co-occurring risk factors may exacerbate the association between witnessed community violence and aggression.

A growing body of research has considered the role of internalizing symptoms in the development of aggression. In particular, anxiety has been shown to be highly comorbid with aggression among children and adolescents [20,21]. Although symptoms of anxiety may inhibit aggressive behavior in some contexts [22], Granic [21] outlined a theoretical model detailing the causal role of anxiety in the development of aggression, which posits that symptoms of anxiety may lead to hypervigilance and the tendency to interpret ambiguous situations as threatening while simultaneously depleting individuals’ capacity to inhibit aggressive impulses. During adolescence, this pattern may be particularly pronounced for girls, as they tend to demonstrate greater reactivity to emotional stimuli in their environments than boys [23]. Indeed, girls are more likely to receive comorbid anxiety and externalizing diagnoses than boys [24,25]. One recent study provides initial support for the notion that anxiety primes youth to act out aggressively in the context of interpersonal stressors. Specifically, high levels of anxiety were found to exacerbate the tendency for children exposed to peer victimization to engage in higher levels of physical aggression over time [26]. It is likely that similar patterns occur in the context of environmental stressors such as witnessed community violence, which poses a clear threat to youth’s physical safety.

Depressive symptoms have also been identified as a risk factor for aggressive behavior [27]. The explanatory styles of these two constructs seem somewhat contradictory (i.e., depressed individuals tend to attribute blame for negative events to themselves, whereas aggressive individuals tend to attribute blame for negative events to others); however, depressive symptoms and aggression have been linked in both clinical [28] and community-based [29] samples, with some evidence indicating that this association is stronger among girls than boys [30]. Two theoretical models may explain this link. First, Dutton and Karakanta [27] suggested that the rumination (i.e., intrusive, unhelpful, repetitive thinking) associated with depression, which is focused on attributing negative events to the self, can turn into rumination that is focused on attributing blame to an external entity; this progression from “sadness” to “anger” may serve as a precursor to aggression. Girls, in particular, tend to rely on ineffective ruminative coping relative to boys [31–33]. Second, irritability is a common symptom of depression in youth [34,35] that can be expressed behaviorally through aggression [30,36]. Taking into account this cycle of rumination along with the associated irritability, it is possible that depressive symptoms may interact with witnessed community violence to heighten the risk of youth engaging in aggression.

Finally, recent findings have highlighted the central role that emotion regulation processes play in the development of aggression across developmental periods [37–40], especially emotion dysregulation. Emotion dysregulation can be defined as “difficulty modulating emotion experience and expression in response to contextual demands and controlling the influence of emotional arousal on the organization and quality of thoughts, actions, and interactions” (p. 85) [41]. Prior work has demonstrated that difficulties managing negative emotions, particularly anger, directly impact patterns of SIP by priming aggressive scripts, focusing one’s attention on hostile cues, and compromising the reappraisal process that is involved in the decision to respond aggressively [40]. There is also evidence to suggest that some individuals use aggression in an attempt to repair, terminate, or avoid uncomfortable emotional states [42,43]. Indeed, several longitudinal investigations have demonstrated that children and adolescents who exhibit greater emotion dysregulation are at risk for engaging in higher levels of aggressive behavior over time [44–46], with some evidence indicating that this association is stronger among girls than boys [47,48]. As youth enter adolescence, gender differences in emotion regulation strategies become more pronounced, with girls endorsing less frequent anger suppression [49] and exhibiting more externalizing emotions relative to boys [50]. Commensurate with this trend, emerging psychophysiological data indicate that the stress-response systems that facilitate an approach-coping (e.g., aggressive) response in the context of perceived threat (e.g., witnessed violence) may be more pronounced in girls than boys [51,52]. Previous research indicates that youth’s ability to manage their emotions may interact with interpersonal risk factors at both the family [53] and peer [38] levels to influence subsequent aggressive behavior. Therefore, in line with Guerra and colleagues’ [8] assertion, emotion dysregulation may exacerbate the link between witnessed community violence and aggression.

The current study sought to advance the literature on witnessed community violence and aggression by examining internalizing (i.e., anxiety and depressive) symptoms and emotion dysregulation as moderators of this association. Considering previous research suggesting that the effects of depressive symptoms and emotion dysregulation on aggression may be stronger for girls than boys [30,47,48], and girls are more likely to receive comorbid anxiety and externalizing diagnoses than boys [24,25], potential gender differences were also considered. Based on available theory and evidence, it was hypothesized that witnessed community violence would be more strongly related to aggression at high levels of anxiety symptoms, depressive symptoms, and emotion dysregulation, especially for girls.

Method

Participants

Participants were 180 adolescents (61.7% girls) in grades 9 (35%), 10, (27.8%), 11 (23.9%), and 12 (13.3%) from a high school located in a large urban school system in the Southeastern region of the United States. All students at the school were given the opportunity to participate in a larger study examining the impact of hopeful, goal-directed thinking on youth’s psychosocial adjustment (n = 420). Caregiver consent was obtained by sending letters home during the fall semester. Consent forms were returned by 78.1% of the families (n = 328); of these, 98.2% of caregivers granted permission for their child to take part in the study (n = 322). Adolescents were also asked to provide written assent at the onset of the project. The overall participation rate for students in the current study was 42.9%; data were completely missing for 39 students who had transferred to another school prior to data collection, 100 students who were absent on the days in which data collection took place, one student who declined assent, and two students who did not complete any of the study measures in the allotted time.

The final sample consisted of 69 boys and 111 girls who ranged in age from 14 to 19 years (M = 15.87, SD = 1.19). Participants identified as 86.3% African American, 5.1% Latino, 2.9% Asian, 2.3% biracial/multiracial, and 3.4% other. Approximately 69% of adolescents reported living in single-parent homes, and school records indicated that 93% of study participants were eligible for free or reduced-price lunch. At the time of data collection, the high school was located in a community in which the poverty rates were almost four times higher than the national average, with 42.7% of families living below the federal poverty line [54]. Further, the violent crime rates in the community where the school was located were also almost four times higher than the national average [12].

Measures

Witnessed Community Violence.

The Community Experiences Questionnaire (CEQ-WIT) is a 14-item self-report measure that assesses lifetime exposure to community violence by witnessing among children and adolescents [9]. Respondents were asked to rate the frequency of each item on a 4-point Likert scale ranging from 1 (Never) to 4 (Lots of Times) while excluding incidents that involved family members as well as those they had seen on television or in movies. Total scores were created by summing all items, with higher scores indicating higher levels of lifetime witnessed community violence. Strong psychometric properties have previously been found in adolescent samples [9], and the CEQ-WIT demonstrated excellent internal consistency in this study (α = .93)

Anxiety symptoms.

The Multidimensional Anxiety Scale for Children-10 (MASC-10) is a 10-item self-report questionnaire designed for use with children and adolescents between the ages of 8 and 19 years [55]. The scale measures physical symptoms of anxiety, harm avoidance, social anxiety, and separation anxiety. Respondents were asked to rate items on a 4-point Likert scale ranging from 0 (Never True About Me) to 3 (Often True About Me). Total scores were created by summing all items, with higher scores indicating greater symptom severity. Strong psychometric properties have previously been reported [55], and the MASC-10 demonstrated adequate internal consistency in this study (α = .78).

Depressive symptoms.

The Beck Depression Inventory, 2nd Edition (BDI-II) is a 21-item self-report measure that assesses the presence and severity of depressive symptoms in adolescents and adults between the ages of 13 and 80 years [56]. Respondents were asked to rate the extent to which each item described how they had felt during the past 2 weeks on a 4-point Likert scale. Total scores were created by summing all items, with higher scores indicating greater symptom severity. Strong psychometric properties have previously been reported in adolescent samples [57], and the BDI-II demonstrated excellent internal consistency in this study (α = .94).

Emotion dysregulation.

The Difficulties in Emotion Regulation Scale (DERS) is a 36-item self-report measure that assesses awareness and understanding of emotional experience, acceptance of emotions, ability to modulate emotional arousal, and effective action in the presence of intense emotions [58]. Respondents were asked to rate items on a 5-point Likert scale ranging from 1 (Almost Never) to 5 (Almost Always). Total scores were created by summing all items, with higher scores indicating greater difficulties managing emotions. Although originally developed for use among adults, strong psychometric properties have previously been reported in adolescent samples [59], and the DERS demonstrated excellent internal consistency in this study (α = .93).

Aggression.

The Revised Peer Experiences Questionnaire (RPEQ) Aggressor Version is a 9-item self-report measure that assesses the use of physical and relational aggression among adolescents [60,61]. Respondents were asked to rate how often each item describes their typical experiences with peers on a 5-point Likert scale ranging from 1 (Never) to 5 (Several Times a Week). In the current study, physical and relational aggression were strongly correlated (r = .86), sharing approximately 74% of their variance; thus, total scores were created by summing all items, with higher scores indicating higher levels of aggressive behavior. Strong psychometric properties have previously been reported in adolescent samples [60], and the RPEQ Aggressor Version demonstrated excellent internal consistency in this study (α = .95).

Procedure

The larger project on which the current study is based was approved by the researchers’ Institutional Review Board (IRB) as well as by school administrators and the school district’s IRB. Data collection occurred 3 weeks prior to the end of the spring semester during students’ 50-minute study hall periods. Trained research assistants were present in each classroom to provide standardized instructions and answer participants’ questions. No school staff or nonparticipating students were present in the classrooms in order to maintain confidentiality and facilitate accurate responding. Survey packets were administered over the course of 1 week in order to avoid making any alterations to the standard school schedule. All students received a t-shirt as compensation for their participation in the study.

Data Analytic Plan

Preliminary analyses were conducted within SAS (University Edition) [62] to provide descriptive statistics and to test the assumptions of multiple linear regression using PROC REG. Given that several of these assumptions were violated, as described in greater detail below, subsequent analyses were conducted within Mplus (Version 7.4) [63] using full information maximum likelihood estimation with robust standard errors (MLR). Note that MLR estimation was used to accommodate the minimal amount of missing data for the four participants who completed one or more, but not all, self-report measures.

A hierarchical approach was employed to test study hypotheses. Aggression was initially regressed on all independent variables and covariates (i.e., gender and grade) in order to examine main effects (Model 1). Model 1 was fully saturated (i.e., 0 degrees of freedom), as it included a covariance between each of the predictor variables, which resulted in a perfect fit to the data; therefore, model fit statistics are not relevant. Next, witnessed community violence by anxiety symptoms and witnessed community violence by anxiety symptoms² (Model 2), witnessed community violence by depressive symptoms and witnessed community violence by depressive symptoms² (Model 3), and witnessed community violence by emotion dysregulation (Model 4) product terms were added independently to the model. Gender differences were then evaluated for anxiety symptoms (Model 5), depressive symptoms (Model 6), and emotion dysregulation (Model 7); specifically, aggression was independently regressed on the corresponding three-way interaction(s) along with the embedded lower-order two-way interactions. All continuous independent variables were standardized prior to computing the interaction terms and conducting analyses in order aid in the interpretation of effects; covariates were centered such that boys and students in the ninth grade were treated as the reference group. Significant interactions were interpreted by evaluating the model when it was conditioned to represent associations at low (−1 SD), moderate (mean), and high (+1 SD) levels of the moderator according to standard procedures [64].

Results

Preliminary Analyses

As shown in Table 1, participants endorsed high levels of lifetime witnessed community violence. Initial inspection of the aggression outcome variable indicated that its skewness (2.19) and kurtosis (4.62) fell below the recommended values of 3 and 10, respectively [65]. However, results from preliminary multiple regression models conducted within SAS using PROC REG revealed that assumptions of normality (Shapiro-Wilk W-test = 0.93, p < .001) and homoscedasticity (White test = 44.52, p = .01) were violated. Further, examination of the partial residual plots indicated that the relations between internalizing symptoms and emotion dysregulation and aggression were potentially nonlinear.

Table 1.

Percentage of Adolescents Who Endorsed Witnessed Community Violence Items

Item	% youth reporting 1 occurrence	% of youth reporting ≥ 2 occurrences
1. How many times have you seen or heard somebody else get threatened?	80.0%	59.4%
2. How many times have you seen somebody else get chased by gangs, other kids, or adults?	69.1%	52.6%
3. How many times have you seen somebody trying to break in or force their way into somebody else’s home?	45.1%	27.4%
4. How many times have you seen somebody get hit, punched, or slapped?	84.0%	69.7%
5. How many times have you seen somebody get robbed or have something stolen from them by force (like a person beating somebody up and then taking their money)?	49.7%	32.6%
6. How many times have you seen somebody carrying a gun or other weapon (besides police, military, and security guards)?	69.7%	49.7%
7. How many times have you seen or heard gunshots?	84.0%	73.1%
8. How many times have you seen somebody try to hurt another person with a knife or other sharp object?	58.3%	38.9%
9. How many times have you seen somebody get hit with a stick, bat, pole, or club?	56.0%	39.4%
10. How many times have you seen somebody have a bottle, rock, or other hard object thrown at them?	76.0%	49.7%
11. How many times have you seen somebody get arrested or taken away by the police?	86.3%	71.4%
12. How many times have you seen a dead body (besides at funerals, wakes, or burials)?	41.7%	22.9%
13. How many times have you seen or heard somebody trying to use force or threats to get another person to do something they didn’t want to do?	64.6%	46.9%
14. How many times have you seen somebody get killed?	30.9%	18.3%

Subsequent analyses were conducted within Mplus using MLR estimation, which is robust against deviations from normality and homoscedasticity [63]. Potential nonlinear relations were first tested independently using polynomial regression models. Results indicated that the quadratic effects of anxiety symptoms (β = .34, SE = .10; b = 1.84, SE = .43, p < .001) and depressive symptoms (β = .29, SE = .12; b = 0.69, SE = .22, p = .002) on aggression were significant. In contrast, there was not a significant quadratic relation between emotion dysregulation and aggression (β = .16, SE = .09; b = 1.17, SE = .68, p = .09); accordingly, this effect was excluded from later multiple regression models. Correlations among study variables were then estimated and are presented in Table 2 along with descriptive statistics.

Table 2.

Descriptive Statistics and Bivariate Correlations Among Study Variables

	1	2	3	4	5	6	7	8	9	10
1. Gender	−−
2. Grade	.22*	−−
3. Witnessed Community Violence	−.10	−.13	−−
4. Depressive Symptoms	−.04	−.02	.14	−−
5. Depressive Symptoms2	−.11	.03	.06	.77*	−−
6. Anxiety Symptoms	.17*	.02	.14	.42*	.29*	−−
7. Anxiety Symptoms2	−.03	.00	.18	.26	.40	.36*	−−
8. Emotion Dysregulation	−.07	−.16*	.20*	.44*	.18*	.29*	.22*	−−
9. Emotion Dysregulation2	−.03	.00	.01	.04	.06	.02	.11	.20	−−
10. Aggression	−.10	−.21*	.37*	.39*	.41*	.25*	.39*	.43*	.24*	−−
M	0.62	1.56	32.67	7.42	−−	9.54	−−	79.60	−−	13.52
SD	0.49	1.05	11.06	10.05	−−	5.90	−−	25.07	−−	7.67

Note.

^*p < .05; Gender (0 = boys, 1 = girls); Grade (0 = 9^th, 1 = 10^th, 2 = 11^th, 3 = 12^th).

Main and Interactive Effects

Results from Model 1 (see Table 3) indicated that grade level was negatively associated with aggression, whereas witnessed community violence and emotion dysregulation were positively associated with aggression. After controlling for covariates and other independent variables in the model, the quadratic effect of depressive symptoms on aggression also remained significant (see Figure 1). Follow-up regions of significance testing [66] revealed that depressive symptoms were negatively associated with aggression at values less than or equal to −1.2 SD and positively associated with aggression at values greater than or equal to +1.2 SD. The main effects model resulted in a total R² = .43.

Table 3.

Main Effects Model Predicting Aggression

	Standardized Estimates		Unstandardized Estimates
	β	SE	b	SE	p
Gender	.01	.06	0.17	.99	.87
Grade	−.15	.05	−1.09	.37	.003
Witnessed Community Violence	.26	.07	1.98	.55	< .001
Depressive Symptoms	−.10	.13	−0.75	.97	.44
Depressive Symptoms2	.36	.13	0.85	.23	< .001
Anxiety Symptoms	.01	.08	0.08	.58	.89
Anxiety Symptoms2	.16	.11	0.84	.51	.10
Emotion Dysregulation	.29	.07	2.23	.61	< .001

Note. Gender (0 = boys, 1 = girls); Grade (0 = 9^th, 1 = 10^th, 2 = 11^th, 3 = 12^th).

Figure 1.

Quadratic effect of depressive symptoms on aggression.

When the product terms were independently added to the model, the quadratic effect of anxiety symptoms significantly moderated the association between witnessed community violence and aggression (Model 2; β = .22, SE = .12; b = 0.62, SE = .30, p = .04; ΔR² = .01). As illustrated by Figure 2a, the association between witnessed community violence and aggression was stronger at low (β = .26, SE = .09; b = 1.96, SE = .72, p = .007), as compared to high (β = .20, SE = .10; b = 1.55, SE = .77, p = .04), levels of anxiety symptoms, and it was not significant when levels of anxiety symptoms were moderate (β = .15, SE = .08; b = 1.14, SE = .66, p = .09). Taking into account the nonlinear relation between anxiety symptoms and aggression, this model was reconditioned to aid in the interpretation of the interaction. As illustrated by Figure 2b, the quadratic effect of anxiety symptoms on aggression was significant at high (β = .20, SE = .11; b = 1.04, SE = .53, p = .048), but not low (β = −.04, SE = .11; b = −0.21, SE = .58, p = .72) or moderate (β = .08, SE = .09; b = 0.42, SE = .47, p = .37), levels of witnessed community violence. In contrast, the quadratic effect of depressive symptoms did not interact with witnessed community violence to influence aggression (Model 3; β = −.08, SE = .15; b = −0.11, SE = .21, p = .59; ΔR² = .00). Emotion dysregulation did, however, significantly interact with witnessed community violence to influence aggression (Model 4; β = .15, SE = .07; b = 1.19, SE = .56, p = .03; ΔR₂ = .02). As illustrated by Figure 3, emotion dysregulation exacerbated the link between witnessed community violence and aggression; specifically, this association was stronger at high (β = .41, SE = .11; b = 3.17, SE = .95, p = .001), as compared to moderate (β = .26, SE = .07; b = 1.99, SE = .54, p = < .001), levels of emotion dysregulation, and it was not significant when levels of emotion dysregulation were low (β = .11, SE = .07; b = 0.80, SE = .55, p = .14). No significant gender differences were observed when three-way interactions were included in Models 5 through 7 (ps ranged from .25 to .88). Finally, a post-hoc analysis was conducted to examine the alternate direction of effect, wherein aggression predicted witnessed community violence. After controlling for covariates, internalizing symptoms, and emotion dysregulation, only aggression was positively associated with witnessed community violence (β = .34, SE = .08; b = 3.71, SE = .83, p = < .001; R² = .15).

Figure 2 (a-b).

Interaction between witnessed community violence and anxiety symptoms² in the prediction of aggression.

Figure 3.

Interaction between witnessed community violence and emotion dysregulation in the prediction of aggression.

Discussion

A substantial body of research has demonstrated that witnessed community violence is linked to higher levels of aggression among children and adolescents [4,6–11]. Although there is some evidence to suggest that parenting practices, friend support, and coping styles may impact this association [67–69], relatively little is known regarding what malleable individual-level factors might moderate risk for aggression among youth exposed to community violence. The current study advanced this literature by evaluating whether internalizing (i.e., anxiety and depressive) symptoms and/and or emotion dysregulation would exacerbate the relation between witnessed community violence and aggression in a sample of urban adolescents. Specific findings, directions for future research, and implications for practice are detailed below.

Consistent with previous work that has documented the disproportionate impact of community violence on youth living in urban communities in the United States [3,12,15,70,71], participants in the present investigation endorsed alarmingly high levels of witnessed community violence. That is, approximately 95% of adolescents reported having seen at least one violent act during their lifetimes. Although this prevalence rate was substantially higher than a national sample of adolescents in the United States (i.e., 38%) [72], it was unfortunately not surprising given that given that an estimated 90% of all violent crimes occur in metropolitan areas, and the violent crime rates in the community where the school in this study was located were almost four times higher than the national average [12]. Further, this rate is comparable to other samples of adolescents living in urban communities (i.e., 93–99%) [73,74]. The reported rates of having witnessed severe acts of community violence in the current study, including physical assault (84%), robbery (50%), seeing a dead body (42%), and murder (31%), are also comparable and/or somewhat lower than those reported in previous samples [73,75–80]. Of note, many adolescents endorsed having witnessed acts of violence on multiple occasions, indicating that this is a chronic stressor affecting their lives.

In line with expectations, a quadratic effect of anxiety symptoms moderated the association between witnessed community violence and aggression among both boys and girls. This is the first investigation to our knowledge to demonstrate that the association between anxiety symptoms and aggression may be nonlinear in nature. That is, witnessed community violence was associated with aggressive behavior at low and high, but not moderate, levels of anxiety symptoms. Current findings build on prior work and suggest that whereas moderate levels of anxiety may inhibit aggressive behavior [22], high levels of anxiety may heighten youth’s risk for aggression when they are faced with environmental and/or interpersonal (i.e., peer victimization) [26] stressors. Interpreted differently, the quadratic effect of anxiety symptoms on aggression was only significant at high levels of witnessed community violence. At low levels of anxiety, witnessed community violence might be related to aggression via SIP deficits, such as hostile attribution biases, aggressive response generation, cognitive rehearsal of aggressive actions, and positive outcome and efficacy beliefs for aggression [8–11,19]. In the presence of moderate levels of anxiety, however, inhibitory factors related to threat avoidance might prevent aggressive behavior (i.e., flight from threat). Yet, in the context of violent neighborhoods, higher levels of anxiety might contribute to hypervigilance and hostile attribution biases while simultaneously depleting the capacity to inhibit aggressive impulses [21]. Thus, severe perceived threat in their environments may prompt highly anxious youth to act out aggressively as a means of self-preservation and protection (i.e., fight against threat).

Support was not found for the hypothesis that symptoms of depression would interact with witnessed community violence to influence aggressive behavior among girls or boys. Rather, a quadratic effect of depressive symptoms was uniquely related to aggression after taking into account witnessed community violence, anxiety symptoms, and emotion dysregulation. To our knowledge, this is also the first investigation to demonstrate that this association may be nonlinear in nature; the current findings build on previous research conducted in clinical [28] and community-based [29] samples and suggest that depressive symptoms are independently associated with aggression at low and high, but not moderate, levels. It is thought that the irritability [36] and the cycle of sadness to anger rumination [27] associated with depression may prime individuals to act out aggressively. Conversely, others have posited that the attributional biases associated with depressive symptoms (e.g., self-blame) and aggression (e.g., hostile attributions directed to others) would support these being mutually exclusive phenomena [27]. Our results seem to provide support for both explanatory models, suggesting that a moderate level of depressive symptoms–potentially characterized by self-blame–might disrupt youth’s hostile attributional biases and decrease their engagement in aggressive behavior. Conversely, adolescents experiencing high levels of depressive symptoms may be more likely to “act out” or express their irritability in interpersonal relationships.

Moreover, emotion dysregulation was found to exacerbate the association between witnessed community violence and aggression among both boys and girls. Specifically, witnessed community violence was related to aggression at moderate to high, but not low, levels of emotion dysregulation. Previous research has demonstrated that difficulties managing negative emotions can lead to the SIP deficits associated with aggression (e.g., activation of aggressive scripts, attention to hostile cues, impairment of the reappraisal process) [40], and individuals may use aggressive behavior in order to repair, terminate, or avoid uncomfortable emotional states [42,43]. Current findings build on previous research and support the notion that, in the context of both environmental and interpersonal (i.e., parental psychological control; peer victimization) stressors [38,53], “individual differences in emotional responsivity and emotion regulation may predispose some children to be more at risk [for aggression] than others” (p. 1561) [8].

Limitations and Directions for Future Research

Findings from this investigation should be interpreted in the context of its methodological limitations. First, the cross-sectional nature of the data precludes conclusions regarding the direction of effects. Previous research suggests the association between witnessed community violence and aggression is bidirectional, such that aggressive youth are at increased risk for witnessing community violence [81,82]. Indeed, this alternate direction of effect was supported in a post-hoc analysis. Additional work is therefore needed to examine whether internalizing symptoms and emotion dysregulation moderate the reciprocal links between witnessed community violence and aggression. Note that the current study aims were investigated during adolescence, considering meta-analytic findings demonstrating that the effects of witnessed community violence on externalizing behavior (including aggression) are likely to develop as a result of cumulative exposure and to be strongest during this developmental period [3]. Still, it would be informative for future investigations to examine how anxiety symptoms and emotion dysregulation influence the prospective associations between community violence and aggression across the lifespan.

Shared method variance may also have impacted our results given that all study constructs were assessed using self-reports. Although adolescents are considered to be the most reliable reporters of their exposure to community violence [83] and their emotional functioning [84], future work would benefit from a multi-informant assessment of aggressive behavior that spans the home, school, and community contexts by including parent-, teacher-, and peer-reports. Moreover, the current measure of aggression was not able to distinguish between specific functions, which refer to the underlying motivation behind the behavior. Previous research has shown that symptoms of anxiety [20] and emotion dysregulation [85] are more closely related to reactive aggression (i.e., impulsive aggression that occurs in response to perceived provocation or threat) than proactive aggression (i.e., instrumental, goal-oriented aggression); thus, additional research is needed to investigate how these individual-level factors may differentially impact the relations between witnessed community violence and functions of aggression.

Summary

In summary, the current findings suggest that low and high levels of anxiety symptoms and moderate to high levels of emotion dysregulation may exacerbate risk for engagement in aggression among urban adolescents who witness violence in their communities. In contrast, low and high levels of depressive symptoms may independently contribute to youth’s aggressive behavior. Despite the aforementioned limitations, the present study may have important implications for practice. Cooley-Strickland and colleagues [71] note that “[although] the need for psychosocial treatment and prevention is particularly critical in inner-cities… there is a paucity of empirically validated, published treatment intervention studies for community violence related distress” (p. 148–149). In addition to targeting posttraumatic stress, which is the strongest correlate of community violence exposure [3], it is imperative to address the cycle by which violence begets more violence. School-based prevention programs represent one avenue to reduce risk for aggression among children and adolescents living in communities affected by violence. Existing group-based [86] and ecological [87] interventions have demonstrated success at targeting the SIP deficits associated with aggression (i.e., hostile attribution biases, beliefs about the acceptability of aggression) and promoting the ability to control aggressive impulses by teaching effective coping skills and problem-solving strategies. The current findings highlight the need for such interventions to also attend to other co-occurring individual-level risk factors for aggression, including anxiety symptoms and emotion dysregulation, among youth who have witnessed high levels of violence in their communities.