Both peer-targeted and dating aggression are fairly common in adolescence, with an estimated 15% to 35% of adolescents reporting peer-targeted aggression perpetration (Modecki et al., 2014) and 13% to 25% of adolescents reporting physical dating aggression perpetration (Wincentak et al., 2017). Research on patterns of aggression perpetration has heavily focused on in-person forms of aggression, such as physical and relational or psychological aggression. There is limited empirical evidence on patterns of peer-targeted and dating aggression perpetration that includes cyber aggression, despite cyber aggression representing a distinct form of aggression (e.g., Lu et al., 2021; Mehari & Farrell, 2018). Peer and dating relationships represent different contexts in which youth may perpetrate aggression, with some adolescents engaging in multiple forms of aggression across relationships (e.g., Van Ouytsel et al., 2017). The purpose of this study was to conduct person-centered analyses to explore whether youth were distinguishable by forms of aggression they perpetrate (physical, relational or psychological, or cyber) or by the relational context of the perpetration (dating or peers). Patterns of aggression are differentially associated with mental health (e.g., anxiety; Garthe et al., 2021) and behavioral outcomes (e.g., delinquent peer association; Whiteside et al., 2013). Identifying youth with different patterns of aggression can help inform individualized approaches to reducing youth violence.

Forms of Aggression in Peer Relationships

Distinct forms of peer-targeted aggression include physical, relational, and cyber aggression. Perpetration of physical aggression involves engaging in physical acts toward one’s peers that may result in bodily harm (e.g., kicking and shoving; Olweus, 1993). In contrast, relational aggression is conceptualized as targeting relationships, image, or reputation through behaviors such as rumor spreading and exclusion (e.g., Crick & Grotpeter, 1995). In a nationally representative sample of sixth- through tenth-grade students, one study found that 27% of youth endorsed perpetrating relational aggression, and 13% endorsed perpetrating physical aggression. Cyber aggression, or cyberbullying, involves perpetrating aggression using electronic communication technologies (e.g., sending threatening texts and posting embarrassing photos; Kowalski & Limber, 2007). Although some researchers have provided theoretical arguments that electronic communication provides another medium for aggression, rather than cyberbullying representing a new form of aggression (e.g., Law et al., 2012; Mehari et al., 2014), most empirical research has supported the conceptualization of cyberbullying as a distinct form of aggression, parallel to physical and relational aggression (e.g., Farrell et al., 2020; Mehari & Farrell, 2018). Prevalence rates among youth are an estimated 16% for cyberbullying based on results from a meta-analysis (Modecki et al., 2014). Although research on verbal aggression (e.g., name-calling) also exists (e.g., Card et al., 2008), results from factor analyses indicate that verbal aggression does not represent a distinct form of aggression (Farrell et al., 2020). Because of this, this study includes only physical, relational, and cyber peer-targeted aggression.

There are positive concurrent relations among distinct forms of peer-targeted aggression. A meta-analysis found that cyberbullying was significantly associated with in-person aggression perpetration, with a medium effect size (r = 0.45; Kowalski et al., 2014). One limitation of the meta-analysis is that it did not examine the relation between cyberbullying and specific forms of in-person aggression, such as physical and relational aggression. This is important given the findings that, like cyberbullying, physical and relational aggression are distinct forms of aggression (Farrell et al., 2020). Research that has examined the relation between cyberbullying and physical and relational aggression has found that cyberbullying is positively associated with both forms of aggression (e.g., Barlett & Wright, 2018; Mehari & Farrell, 2018). However, fewer studies have examined the extent to which youth who perpetrate aggression in peer relationships, particularly cyberbullying, also perpetrate aggression in dating relationships.

Forms of Aggression in Dating Relationships

Forms of aggression toward dating partners can include physical, psychological, and cyber aggression. Physical dating aggression involves causing or threatening to cause physical harm to a current or former dating partner. It includes acts such as hitting, shoving, kicking, or throwing something at the dating partner (Centers for Disease Control and Prevention [CDC], 2020). Findings of a meta-analysis suggest that 13% to 25% of youth aged 13 to 18 perpetrate physical dating aggression (Wincentak et al., 2017). In contrast to physical aggression, psychological aggression involves emotionally manipulating one’s dating partner through behaviors such as demeaning or yelling at one’s partner and controlling access to social support (e.g., Stets, 1991). Cyber dating aggression involves harming or threatening to cause harm to one’s partner using electronic technologies (e.g., Borrajo et al., 2015), such as spreading private information, monitoring location and communications, and pretending to be someone else to deceive one’s partner. Prevalence rates range from 20 to 25% for adolescent psychological dating aggression and 13% to 54% for cyber dating aggression, according to findings of a systematic review (Stonard et al., 2014). Although relational aggression, which can involve rumor spreading, social exclusion, and confrontational behaviors (e.g., Crick & Grotpeter, 1995; Murphy & O’Leary, 1989), also represents a form of dating aggression, relational and psychological aggression have overlapping features (e.g., Linder et al., 2002) and are not well differentiated by existing dating aggression measures (Persram et al., 2021). This may be due to psychological aggression being considered a specific example of relational aggression, which represents a broader construct (Linder et al., 2002). Both relational and psychological aggression target individuals’ sense of belongingness within relationships. For these reasons, this study assesses only physical, psychological, and cyber dating aggression.

Similar to forms of peer-targeted aggression, forms of dating aggression are also interrelated. One study found that among an ethnically diverse sample of middle and high school students, adolescents who perpetrated cyber dating aggression were also likely to perpetrate physical and psychological dating aggression (Zweig et al., 2013). In a similar sample of high school students, Lu and colleagues (2021) found that cyber dating aggression was moderately correlated with physical and psychological dating aggression. In addition, one study found that cyber dating aggression was moderately correlated with multiple types of physical dating violence in a sample of high school and university students from Chile (Lara, 2020). Taken together, these findings suggest that adolescents who engage in one form of dating aggression are likely to concurrently engage in other forms of dating aggression.

Relations among Forms of Peer-Targeted and Dating Aggression

Adolescents who perpetrate aggression within one relationship context may also perpetrate aggression in another relationship context. According to attachment theories (e.g., Bowlby, 1982), attachment styles in friendships (e.g., Furman, 2001) and romantic relationships (e.g., Steinberg et al., 2006) are similar to attachment styles developed with caregivers. Related to the notion of shared working models across relationships, the developmental-contextual theory of romantic relationships indicates that romantic relationships emerge from and are influenced by peer relationships (Brown, 1999). As such, it is possible that adolescents behave similarly across peer and dating relationships. For example, multiple studies have found that adolescents who perpetrate physical peer-targeted aggression also perpetrate physical dating aggression (e.g., Connolly et al., 2000; Farrell & Vaillancourt, 2019). Zych and colleagues’ (2021) meta-analysis of 20 studies found that peer-targeted aggression perpetration was positively associated with dating aggression perpetration. However, relations were most robust between peer-targeted aggression and psychological dating aggression and between peer-targeted aggression and cyber dating aggression. This meta-analysis was limited in that it did not assess distinct forms of peer-targeted aggression perpetration, including cyberbullying. Given these findings, it is possible that adolescents concurrently perpetrate aggression across peer and dating relationships.

Patterns of Peer-Targeted and Dating Aggression

A person-centered approach can be useful in exploring groups of individuals that are characterized by specific patterns of aggression, which may not be captured by variable-centered approaches (e.g., McCutcheon, 1987). Studies have identified patterns of adolescents’ peer-targeted and dating aggression, but these analyses have not included cyber aggression. For example, one study identified three patterns of adolescents aged 14 to 18: one group had high levels of perpetration of peer-targeted aggression, one group had high levels of perpetration and victimization of peer-targeted aggression, and one group had high levels of perpetration and victimization of peer-targeted and dating aggression (Whiteside et al., 2013). They found that relations with outcomes (e.g., delinquent peer association and substance use) varied by class membership. A second study identified three classes of youth using an ethnically diverse sample of seventh-grade students: one group had low levels of perpetration of peer-targeted and dating aggression, one group had high levels of perpetration and victimization of peer-targeted aggression, and one group had high levels of perpetration and victimization of peer-targeted and dating aggression (Garthe et al., 2021). They found that youth in the peer and dating violence class reported higher levels of depression and anxiety compared with youth in the low violence class. Similar classes were identified in a predominantly African American sample of sixth-, seventh-, and eighth-grade students (Sullivan et al., 2022). It is unclear whether the patterns would vary when cyberbullying and cyber dating aggression are included; it is possible that form rather than relationship context would more effectively differentiate between youth or that youth would be differentiated in some combination of form and relationship context of aggression.

Purpose of the Present Study

The purpose of this study was to conduct person-centered analyses to explore whether youth were distinguishable by forms of aggression they perpetrate (physical, relational or psychological, or cyber) or by the relational context of the aggression (peers or dating). Prior findings identified patterns of youth distinguishable by relational context (e.g., Whiteside et al., 2013). However, these studies did not assess cyberbullying or cyber dating violence. Research indicates that cyberbullying (e.g., Mehari & Farrell, 2018) and cyber dating violence (e.g., Lu et al., 2021) represent counterparts to other forms of aggression. As such, our aim to examine whether adolescents would be clustered into profiles that are differentiated by relational contexts (dating or peers) or by forms of aggression (cyber, physical, relational, or psychological) was exploratory. Understanding the overlap between peer-targeted and dating aggression can inform strategies for use in violence prevention and intervention programs as distinct patterns of aggression are differentially associated with mental health and behavioral outcomes in youth.

Method

Participants

Participants were 185 ninth-grade students attending one high school in the southeastern United States who endorsed dating someone in the past three months, representing 65% of the larger sample. The high school is located in an under-resourced community with about one-third of city residents living in poverty (U. S. Census Bureau). Students who did not endorse dating someone in the last three months (N = 100) were excluded from analyses. The racial and ethnic composition of the included participants was 91.8% Black or African American, 2.7% Other, 2.3% White or Caucasian, 1.2% Latinx or Hispanic, 1.6% American Indian or Alaska Native, and less than one percent Asian or Pacific Islander. The sample was approximately evenly divided by gender (47.2% male, 51.7% female, and 1.1% nonbinary). Participants ranged in age from 13 to 17 years old, with an average age of 14.31 years (SD = 0.56). Of the 96% of students who reported on their family structure, 40% lived with their mother only; 27% lived with both parents; 12% lived with their mother and step-father; 6% lived with other relative only; 6% lived with their mother and other relative; 5% lived with foster or adoptive parent; 2% lived with their father only; and 2% lived with their father and step-mother.

Procedures

All procedures were approved by the university’s Institutional Review Board, by the administration in the school district’s central office, and by the specific school’s administration. An opt-out (passive) consent process was used for parent/caregiver informed consent. Letters were sent home to parents/guardians, and they were given the opportunity to opt out of the study.

Research staff informed students that the survey was confidential rather than anonymous. Staff explained that identification numbers would be entered at the start of the survey in place of students’ names. Students were assured that no one outside of the research team would have access to the master sheet that linked students’ names and identification numbers. Each participant provided active written assent prior to participation. Youth were eligible to complete the survey if they were in regular education classes; their parent/guardians did not opt-out; the youth provided active assent; the youth could read or speak English; and the youth was present during one of the two days of data collection. None of the parents opted their child out of the study. Of the 306 eligible students, 21 students (6.8%) did not provide assent. Ninety-three percent of eligible ninth-grade students participated in the study.

Participants completed surveys in August of 2019 as part of a violence prevention program originally developed for ninth-grade students. The transition between early to middle adolescence and middle to high school tends to occur in ninth grade, which is also when youth are at an increased risk for problem behaviors (e.g., Liao et al., 2013; Matuszka et al., 2017). Only the baseline data were used for this study (prior to any intervention implementation). Surveys were administered via Qualtrics in classrooms with about 25 students per room. Students were seated far enough apart to ensure privacy. Research assistants were available to answer any questions about the survey and to keep participants on task. The surveys took about 30 min to complete, and participants were given $5 as a thank-you for their time and effort.

Measures

Demographics. Age, gender, race and ethnicity, and family structure were based on youth reports. Age was measured using the question, “How old are you?” Gender was measured using the question, “What is your gender?” Response options included boy, girl, and non-binary. Race was measured using the question, “What is your race or ethnicity?” Response options included Black or African American, Latino or Hispanic, White or Caucasian American, American Indian or Alaska Native, and Asian or Pacific Islander. Family structure was measured using the question, “Who lives with you ALL or MOST of the time? You can check more than one.”

Peer-targeted Aggression. The Problem Behavior Frequency Scale-Adolescent Report (PBFS-AR; Farrell et al., 2018) assessed physical, relational, and cyberbullying. The validity of the PBFS-AR is supported by its correlations with code violations from school office referrals (Farrell et al., 2020) and teachers’ ratings of students’ problem behavior (Farrell et al., 2018). Youth reported on their experiences during the past 30 days using a 6-point scale, ranging from 1 (never) to 6 (20 or more times). Five items assessed physical aggression (e.g., “Hit or slapped someone” and “Thrown something at someone to hurt them,” Cronbach’s alpha = 0.80). Six items assessed relational aggression (e.g., “Spread a false rumor about someone” and “Told someone you wouldn’t like them unless they did something you wanted,” Cronbach’s alpha = 0.72). Five items assessed cyberbullying (e.g., “Used cell phone pictures to make fun of someone” and “Used text-messaging to threaten to hurt someone physically,” Cronbach’s alpha = 0.76).

Dating Aggression. An adapted version of the Dating Violence Scale (Foshee et al., 1996) assessed physical and psychological dating aggression. Participants were first asked, “In the past 3 months, have you had a boyfriend or girlfriend (someone that you dated or gone out with)?” (Yes/No). If they endorsed “yes,” participants then reported on their experience in the past 3 months using a 4-point scale ranging from 1 (never) to 4 (10 or more times). Five items assessed physical dating violence perpetration (e.g., “Threw something at him/her that could hurt” or “Kicked him or her,” Cronbach’s alpha = 0.85). Four items assessed psychological dating violence perpetration (e.g., “Damaged something that belonged to him or her” and “Said things to hurt his or her feelings on purpose,” Cronbach’s alpha = 0.71). Although evidence indicates that physical and psychological forms of dating aggression are not well differentiated using a confirmatory factor analysis (Goncy et al., 2016), it is possible that this differentiation exists when using a latent profile analysis. Differentiating between physical and psychological forms of dating aggression in the current study would allow for examination of whether youths’ engagement in these forms of aggression is distinguishable from one another when using a latent profile analysis.

Three items assessing cyber dating aggression (e.g., “Wrote nasty things about him or her online” and “Use his or her social networking account without permission,” Cronbach’s alpha = 0.75) were adapted from previous studies (Picard et al., 2007; Zweig et al., 2013). Cyber dating aggression is a fairly new construct, and there is currently no best practice measure of cyber dating aggression. Validity of the cyber dating perpetration items are supported by correlations with all forms of peer-targeted and dating aggression, as evidenced in the current study.

Analytic Approach

Latent profile analyses (LPA) were conducted to identify profiles of youth with different patterns of peer-targeted and dating aggression using Mplus version 8.6 (Muthén & Muthén, 2017). Latent profile analysis was used because of the use of differences in means to calculate profiles, which differs from latent class analysis and the use of items to calculate classes (Oberski, 2016). Latent profile analyses were conducted to test a series of models, with each model identifying an increasing number of profiles. The number of profiles was determined based on theory, profile group size considerations, and model fit indices. Model fit was evaluated using information criterion indices, including the Vuong-Lo-Mendell-Rubin (VLMR) likelihood ratio test, Akaike information criterion (AIC), Bayesian information criterion (BIC), sample-size adjusted Bayesian information criterion (saBIC), and entropy. Smaller information criterion indices indicate better model fit. The VLMR likelihood ratio test was used to establish the point at which increasing the number of profiles no longer significantly improved the model fit, with a significant VLMR likelihood ratio test indicating improved model fit. Emphasis was placed on the BIC given findings that it is the most effective fit index in establishing the number of profiles (Nylund et al., 2007), with smaller values indicating better fit (Wang & Wang, 2012). Entropy identifies the number of participants who can be assigned to a profile, with scores near 1 indicating a better fit (Ramaswamy et al., 1993).

Data were missing on the following items: “Punched or hit him or her with something that could hurt” (physical dating aggression, 0.5%); “Hit or slapped someone” (physical peer-targeted aggression, 0.5%); “Thrown something at someone to hurt them” (physical peer-targeted aggression, 0.5%); “Shoved or pushed someone” (physical peer-targeted aggression, 0.5%); “Not let someone be in your group anymore because you were mad at them” (relational peer-targeted aggression, 0.5%); “Used cell phone pictures to threaten to hurt someone physically” (cyber peer-targeted aggression, 1.1%); and “Used text-messages to make fun of someone” (cyber peer-targeted aggression, 0.5%). Mean scores were calculated only if at least three items within each subscale had valid values.

Results

Descriptive Statistics

The most prevalent form of peer-targeted aggression perpetrated at least once in the past 30 days was physical aggression (84%), followed by relational aggression (73%), and cyberbullying (54%). The most prevalent form of dating aggression was psychological aggression (52%), followed by physical aggression (35%), and cyber aggression (22%). There was some overlap between peer-targeted and dating aggression perpetration. About 15% of youth who perpetrated cyberbullying also perpetrated cyber dating aggression, 33.5% of youth who perpetrated peer-targeted physical aggression also perpetrated physical dating aggression, and 41.6% of youth who perpetrated peer-targeted relational aggression also perpetrated psychological dating aggression.

Means, standard deviations, and correlations for all variables are reported in Table 1. Correlations among the peer-targeted aggression variables ranged from moderate (i.e., r = 0.46 between physical and relational peer-targeted aggression) to large (i.e., r = 0.65 between physical aggression and cyberbullying). Correlations among the dating aggression variables were large (e.g., r = 0.81 between physical and psychological dating aggression). Correlations among the three peer-targeted aggression variables and the three dating aggression variables ranged from low (i.e., r = 0.27 between relational peer-targeted aggression and cyber dating aggression) to moderate (i.e., r = 0.43 between physical peer-targeted aggression and physical dating aggression).

Table 1 Means, standard deviations, and correlations for peer-targeted and dating aggression measures
Full size table

Latent Profile Analyses

A series of latent profile analyses was conducted to test the primary aim of the study: to explore the extent to which different profiles of youth would emerge based on the form of aggression they perpetrated or the relationship context of their aggression. The fit indicators for models with one- to five-profile solutions are reported in Table 2. A two-profile solution improved the fit over a one-profile solution according to the VLMR likelihood ratio test (445.70; p = 0.04). Adding a third profile did not significantly improve the model fit based on the VLMR likelihood ratio test. However, a three-profile solution was selected for further analysis due to smaller information criterion indices (i.e., BIC, saBIC, and AIC) for a three-profile solution compared with a two-profile solution, and based on theory informed by previous findings (e.g., Garthe et al., 2021; Whiteside et al., 2013). Although evidence from the VLMR likelihood ratio test indicated a two-profile solution, more weight was given to the BIC due to findings that the BIC is more suitable with smaller sample sizes and is more reliable compared with the VLMR (e.g., Nylund et al., 2007). The entropy for a three-profile solution was 0.98, indicating that youth could accurately be assigned to profiles. Although the information criterion indices continued to become smaller with additional profiles and entropy was strong, adding a fourth and fifth profile did not create meaningfully different profiles and resulted in small profiles for the four-profile (i.e., 3.2% for the smallest class) and the five-profile (i.e., 1.6% for the smallest class) solutions. Biological sex was included as an auxiliary variable and it did not predict profile membership.

Table 2 Fit statistics for 1- through 5-profile models
Full size table

The profiles displayed in Fig. 1 depict the probability of members in each profile reporting perpetration of each form of peer-targeted and dating aggression. With regard to characterizing and describing the profiles, labels were selected to reflect differences across profiles rather than similarities. Visual examination of profiles suggests that Profile 1 was characterized by low levels across the different forms of peer-targeted and dating aggression (the low frequency aggression profile). Among the youth in the low frequency aggression profile (85.3%), youth had some perpetration of peer-targeted physical aggression but had negligible perpetration of any other form of aggression in any relationship context. Profile 2 was characterized by high levels of dating aggression and moderate to high levels of peer-targeted aggression (the high frequency dating aggression profile). The high frequency dating aggression profile (6.6%) was characterized by high levels of all forms of dating aggression, high levels of peer-targeted physical aggression, but moderate levels of peer-targeted relational and cyber aggression. Profile 3 was characterized by high levels of peer-targeted aggression and low levels of dating aggression (the high frequency peer-targeted aggression profile). The high frequency peer-targeted aggression (8.1%) profile was characterized by high levels of physical, relational, and cyber peer-targeted aggression but low levels of physical, psychological, and cyber dating aggression. Overall, results support the notion that youth are distinguishable by the relationship context of their aggression.

Fig. 1
figure 1

The 3-Solution latent profile analysis representing the probability of engaging in the different forms of aggression for each profile

Full size image

Discussion

The purpose of this study was to conduct person-centered analyses to explore whether there were distinct profiles of adolescents based on different forms of aggression (physical, relational or psychological, or cyber) or relational contexts (peers vs. dating partners). Previous studies have been fairly siloed, such that very little research has examined multiple aggression perpetration contexts (form and relationship) simultaneously and studies typically did not include cyber forms of aggression. Results of the LPA indicated that youths’ patterns of aggression were generally distinguishable by the relational context of aggression rather than by the form of aggression they perpetrated. Three profiles of youth were found in this study, including (a) a low frequency aggression profile, (b) a high frequency dating aggression profile, and (c) a high frequency peer-targeted aggression profile. Overall, these findings suggest that youth who use one form of aggression (e.g., physical aggression) in specific relational contexts (e.g., towards peers) also are likely to use other forms of aggression (e.g., cyberbullying) in those same contexts. The pattern does not, however, suggest that youth who use one form of aggression indiscriminately perpetrate that form of aggression across relational contexts. This study contributed to the existing literature by using latent profile analysis to identify profiles of youth involved in various forms of peer-targeted and dating aggression perpetration among a sample of ninth-grade students.

The finding that adolescents’ patterns of aggression were mostly distinguishable by the relational context is consistent with previous findings among middle school students (i.e., Garthe et al., 2021; Sullivan et al., 2022) and older adolescents (i.e., 14 to 18; Whiteside et al., 2013). For instance, among samples of youth from under-resourced communities, Garthe and colleagues (2021) and Whiteside and colleagues (2013) identified patterns of adolescents that were differentiated by violence involving peers and dating partners. Notably, the inclusion of cyber aggression did not shift the profiles from what has been found when cyber forms of aggression were not included. Adolescents’ may restrict their use of aggression to a specific context, which may be explained by social acceptability of aggression in certain relationships. For instance, in a sample of sixth- and ninth-grade students, one study found that dating violence was viewed as more acceptable than peer-targeted physical aggression (Cerda-Smith et al., 2022). It is likely that acceptability of violence within one relational context extends across multiple forms of violence within that context. It is also likely that adolescents’ reports of peer-targeted aggression overlap with reports of dating aggression, as dating partners often emerge from peer relationships (Brown, 1999). Although youth mostly interact with same-sex peers during childhood, they begin to interact with opposite-sex peers during adolescence (Furman, 2002). As a result, romantic relationships tend to form from those existing opposite-sex peer networks (Furman, 2002). As such, the distinction between these relationship contexts may not be as clear from the youths’ perspective, however, although the context may not be discrete, the experience of peer-targeted versus dating aggression help differentiate strategies used to cope (e.g., who the adolescent seeks out for support). More research is needed to examine if these patterns of peer-targeted and dating aggression are differentially associated with mental health and behavioral outcomes.

Despite theoretical and empirical support for distinct forms of aggression (e.g., Mehari et al., 2014), except for physical aggression, profiles of youth were not differentiated based on the form of aggression they perpetrate. This finding may be explained by the fairly high co-occurrence of cyber and in-person bullying (e.g., physical and relational; Kowalski et al., 2014) and cyber, physical, and psychological dating aggression (e.g., Lu et al., 2021). In the current study, cyberbullying was highly correlated with peer-targeted physical and relational aggression (i.e., r = 0.64 to 0.65) and cyber dating aggression was highly correlated with physical and psychological dating aggression (i.e., r = 0.55 to 0.65). In contrast, similar forms of peer-targeted and dating aggression (e.g., cyber) were not as highly correlated (i.e., r = 0.29 to 0.33). The notion that youth engage in multiple forms of aggression within one context is supported by findings that adolescents characterized as cyberbullies are part of a group of adolescents who also perpetrate other forms of bullying (e.g., Wang et al., 2012). This suggests that more overlap may exist across forms within a specific context rather than within forms across a specific context.

Importance of the Relationship Context

Peer and dating relationships represent distinct relational contexts with different functions and expectations. Early attachment in parent–child relationships contribute to attachment in peer and romantic relationships (Bowlby, 1982). However, research indicates that parental attachment is differentially associated with attachment in peer and romantic relationships (e.g., Doyle et al., 2009). Parents’ modeling of behavior, as well as adolescents’ relationships with their parents, may set the stage for peer and dating relationships. Findings from systematic reviews indicate that exposure to parental domestic violence is positively related to youths’ dating aggression perpetration (Taquette & Monteiro, 2019), whereas parental warmth is inversely related to youths’ peer-targeted aggression perpetration (Nocentini et al., 2019). Similar to parent relationships, peer relationships, which begin in early childhood, often serve as a training ground for social skills that will be used in later relationships (e.g., Ellis et al., 2009). Peer relationships are an integral part of autonomy and identity development, as youth seek acceptance and a sense of belonging (Zimmer-Gembeck, 2002). Dating relationships, which often begin in adolescence, are emotionally intense relationships involving different expectations around commitment and exclusivity (e.g., Branje et al., 2007), and expressions of affection, including physical intimacy (Collins, 2003). Given these differences, it is not surprising that adolescents may also have different motivations for using aggression in their peer and dating relationships. Adolescence is characterized by an increase in competition among peers (Steinberg, 2011). Adolescents may initially direct aggression towards peers to increase their social status, to distinguish between in-group and outgroup, to defend against threats, or to respond to provocation (e.g., Farrell et al., 2010; Volk et al., 2015). These motivations for peer-targeted aggression may be especially true for youth exposed to violence as a potential consequence of living in under-resourced communities (e.g., Richards et al., 2015). In romantic relationships, however, aggression may be used to gain or maintain power, such as to control a partner’s behavior or as a reactive response when expectations are not met (Giordano et al., 2016; Hester & Donovan, 2009). Taken together, findings that youths’ aggression perpetration patterns are distinguishable based on relational context may be explained by differences in the function of peer and dating relationships and in the role of aggression within those relationships.

Implications for Prevention

These findings highlight the need for prevention and intervention programs targeting adolescents to explicitly address the relational contexts of aggression (peers vs. dating partners) rather than focusing exclusively on the form of aggression that is perpetrated (e.g., cyber). In addition, individual characteristics related to the perpetration of aggression (e.g., beliefs about aggression, empathy, perspective-taking, and goals) may be specific to the relational context. Due to the different functions and expectations in peer versus romantic relationships, youth might not generalize skills learned in one relationship context to other relationship contexts. This may be especially true for adolescents who are not currently dating. Adolescents who are not currently dating report high levels of anxiety about romantic relationships (Paulk et al., 2011), which might prevent them from pursuing future romantic relationships. When adolescents do eventually begin to date, skills learned in one context (e.g., peer relationships) may not have been retained for use in another context (e.g., romantic relationships). Of note, peer-targeted aggression (54% to 84% for physical, relational, and cyber aggression) was more prevalent than dating aggression (22% to 52% for physical, psychological, and cyber aggression) in the current sample, suggesting that late middle school to early high school would be an optimal time to engage in preventative strategies. School personnel may help youth generalize skills across relational context by modeling behavior at school. Specifically, school personnel could model respect, boundary-setting, and problem-solving skills with colleagues and with students.

Given that youth who perpetrate one form of aggression within one relational context are likely to perpetrate other forms of aggression within that context, it is possible that similar risk and promotive factors may exist within rather than across relational contexts. This is supported by the development of separate programs broadly addressing youth violence (e.g., Second Step; Frey et al., 2005) and specifically addressing dating violence (e.g., Safe Dates; Foshee et al., 2005). However, implementing multiple programs may place an unnecessary burden on schools, that have multiple academic demands (e.g., student evaluations and standardized testing). Given that adolescents’ behavior differs by context, school personnel should identify the overlapping components for broad violence prevention programs and those for dating violence (e.g., norms or beliefs about aggression, proactive peer bystander approaches; Joseph & Kuperminc, 2020). Recognizing this overlap could aid school personnel in choosing violence programs that encourage youth to have both positive peer and dating relationships. Components of existing programs can also be adapted to cover situations that are similar across various relational contexts. For instance, intervention components focused on increasing norms or beliefs supporting nonviolent behaviors could include norms specific to both peer and romantic relationships. In addition, components that include role playing or vignettes can include examples from peer and romantic relationships. Integrating both peer and romantic relationships within broader violence prevention programs might increase the likelihood that youth will generalize skills across different relational contexts.

Limitations

The current study has several limitations. Peer-targeted and dating aggression perpetration were assessed using youth-report. Although appropriate when measuring perceptions of youths’ behavior, self-report measures are at-risk for social desirability effects (e.g., DeVellis, 2011). Another limitation is response bias, as some adolescents may have been reluctant to share the true frequency of their peer-targeted and dating aggression perpetration. The measure used to assess peer-targeted aggression did not specify the target of the behavior. Given that dating partners tend to form from existing peer groups and the potential overlap of peer-targeted and dating aggression, findings should be interpreted with caution. This study only included adolescents who endorsed dating someone in the past three months. It is possible that adolescents excluded from the sample perpetrate dating aggression towards former dating partners or towards peers prior to them becoming dating partners. Aggression victimization represents another way in which youth can experience violence. Future research should examine whether these patterns exist when both aggression perpetration and aggression victimization are taken into consideration. Physical and psychological dating aggression were examined as distinct forms of aggression, which is inconsistent with findings indicating that they represent one construct (Goncy et al., 2016). Although distinguishing between these two forms of aggression was appropriate for latent profile analyses, it is possible that patterns would differ if those forms of aggression were treated as one broad construct.

Verbal peer-targeted and dating aggression and relational dating aggression were not assessed in the current study. It is possible that the patterns of aggression identified in this sample would be different if those forms of aggression were included. It is also possible that the patterns may vary as a function of sexual orientation and gender, which were also not assessed in this study. Sexual orientation and gender may be particularly relevant given that sexual and gender minority youth experience high rates of dating violence victimization (Miller et al., 2018). The study had a predominantly African American sample of ninth-grade students who attended a high school in the southeastern United States, limiting the generalizability of the research findings. More research with preadolescents and older adolescents is needed to examine the degree to which these results generalize across developmental stages. It is possible that children may be less differentiated in the contexts in which they perpetrate aggression and that aggression becomes more siloed over time. It is also possible that as youth get older, the likelihood of generalizing coercive methods from one context to another increases, especially if the coercive methods result in desired outcomes.

Conclusion

These findings indicate that among a predominantly African American sample of ninth-grade students from under-resourced communities, youth are distinguishable by the relational contexts (peers vs. dating) in which they perpetrate aggression. Specifically, latent profile analysis identified three patterns of youth: 1) a low frequency aggression profile; 2) a high frequency peer-targeted aggression profile; and 3) a high frequency dating aggression profile, highlighting the importance of using person-centered approaches to better understand aggression perpetration. Given the difficulties associated with transitioning to ninth-grade (e.g., Liao et al., 2013; Matuszka et al., 2017) and the lower prevalence of dating violence compared to peer-targeted violence during this school year, implementing violence prevention programs prior to high school may be beneficial. Future research may explore how these profiles change over time and how these profiles differentially relate to mental health and behavioral outcomes.