Bullying, defined as undesired and intentionally aggressive interpersonal behaviors that involve a power imbalance between the aggressor and victim and that cause, or threaten to cause, harm to the victim (Gladden et al., 2014; Olweus, 1997), is a significant concern for adolescents. Bullying behaviors can occur both in-person and online and include psychological and verbal aggression (e.g., spreading rumors, teasing, and name-calling) and physical aggression (e.g., punching and shoving; Wang et al., 2009). Bullying is both a normative and a troubling aspect of childhood and adolescence (Kennedy, 2021), with recent data suggesting 20% of middle school youth report perpetrating some form of bullying (Da Silva et al., 2020). The negative psychosocial consequences of both perpetration and victimization are well-documented (Hawker & Boulton, 2000; Man et al., 2022; Wolke & Lereya, 2015). Given the developmental demands and challenges of early adolescence (i.e., ≈11 to 15 years old; Lerner, 2021; Pepler et al., 2006), and negative consequences of perpetrating bullying in early adolescence for functioning in later adolescence and beyond (Farrington et al., 2011; Kretschmer et al., 2017; Zych et al., 2015), it is especially important to understand how bullying perpetration at this age is associated with other harmful outcomes and behaviors, such as Teen Dating Violence (TDV).

TDV consists of verbally, physically, relationally, or sexually aggressive or abusive behaviors directed toward romantic partners (Wolfe et al., 2001). Physical TDV perpetration refers to using (or trying to use) physical force against a romantic partner (e.g., punching, pushing, slapping), and sexual TDV perpetration refers to forcing (or attempting to force) a romantic partner to engage in sexual behaviors without their consent (Center for Disease Control and Prevention, 2024). Understanding the factors that increase risk for physical and sexual TDV are particularly important because these TDV subtypes can result in serious health outcomes in youth, such as sexually transmitted diseases (including HIV; Decker et al,, 2005), physical injury (Tharp et al., 2017), pregnancy, and suicidality (Silverman et al., 2001). A meta-analytic evaluation of adolescents (aged 13 to 18 years) found that 13% of boys and 25% of girls had ever perpetrated physical TDV, while 10% boys and 3% of girls had ever perpetrated sexual TDV (Wincentak et al., 2017). In a study of past year TDV prevalence rates of 10^th graders, 9% of youth who had dated reported perpetrating physical TDV (Haynie et al, 2013). While TDV among middle schoolers has received less attention, evidence suggests rates of physical TDV in these youth are also high—ranging from 15-17% (Niolon et al, 2015; Simon et al., 2010). One large study of high-risk middle schoolers found that among those who had dated, 32% perpetrated physical TDV and 15% perpetrated sexual TDV over their lifetime (Niolon et al., 2015). These data highlight the importance of research to identify factors that can be targeted in TDV prevention efforts (e.g., bullying)—especially for middle school youth.

Numerous studies indicate that bullying perpetration and TDV perpetration tend to co-occur in the same individuals (Niolon et al., 2015; Vivolo-Kantor et al., 2016; Yahner et al., 2015). As a recent meta-analysis attests, bullying perpetration among adolescents is consistently associated with TDV perpetration (Zych et al., 2021). In high school samples, this association has been observed cross-sectionally (Yahner et al., 2015) and longitudinally (Espelage et al., 2012, 2015, 2018; Rivas-Koehl et al., 2023), with bullying earlier in adolescence predicting subsequent TDV perpetration. The association between bullying and TDV exists in samples that focus exclusively on middle school students as well (Cutbush et al., 2016; Miller et al., 2013; Foshee et al., 2014;). Indeed, multiple longitudinal studies suggest that bully perpetration predicts TDV perpetration in middle schoolers (Foshee et al., 2014; Cutbush et al., 2016; Humphrey & Vaillancourt, 2020). However, studies thus far assume that bullying is a developmental precursor to TDV, without evaluating the potential bidirectionality of these associations. Thus, the purpose of the current study is to evaluate bidirectional longitudinal associations between bullying perpetration and TDV perpetration in middle school youth.

Multiple conceptualizations and theories have been used to explain the connection between bullying and TDV (Espelage et al., 2014; Humphrey & Vaillancourt, 2020;). The Social Learning Theory suggests youth learn aggression via modeling and then continue to engage in aggression because it is adaptive or reinforcing (Bandura, 1976, 1977; Hong et al., 2018; Pusch, 2024). Therefore youth who have learned that aggressive behavior is effective for goal attainment in one relationship (e.g., friendships; Hong et al., 2018), may be reinforced to use aggression to achieve their goals in other relationships (e.g., dating relationships; for a review see Pusch, 2024). Similarly, the Social Interaction Theory (Patterson, 1982; Patterson et al., 1984) suggests that aggressive behaviors stem from conflictual behaviors in the home, with the home functioning as a training ground for aggressive behaviors. These behaviors are then applied to peers and—later—dating partners (Espelage et al., 2014). Alternatively, the phenomenon of heterotypic continuity purports that multiple different types of aggression stem from an underlying propensity of aggression, which results in diverse behavioral manifestations depending on context and developmental stage (Moffitt, 1993; Humphrey & Vaillancourt, 2020). All of these conceptualizations of interpersonal aggression support that aggression in one context may generalize to other relationships. This is supported by research indicating youth often engage in multiple types of violence (Hamby & Grych, 2012; Taquette & Monteiro, 2019) and that different forms of violence stem from common risk factors (Catalano et al., 2012; Foshee et al., 2016).

Due to the assumed developmental timing of bullying and TDV, many researchers have proposed bullying is a developmental precursor to TDV (Espelage et al., 2022; Joseph & Kuperminc, 2020; Zych et al., 2021). Indeed, longitudinal studies of the association between bullying and TDV focus exclusively on the bullying-to-TDV pathway (Zych et al., 2021). However, consistent with Social Learning Theory, it is possible that TDV behaviors may precede bullying behavior. Indeed, some adolescents may be first exposed to aggressive behaviors through their dating relationships or they may be introduced to new friend groups through their dating partners in which aggressive behavior is normative (Furman & Shaffer, 2003); and then begin practicing aggressive behaviors in-turn (Norona et al., 2017). However, the authors were unable to identify any studies that examined the degree to which TDV behaviors predicted bullying behaviors. A better understanding of the potentially bidirectional nature of associations between bullying and TDV perpetration may have important implications for violence interventions for youth. Violence interventions for middle school youth tend to focus exclusively on peer relationships (Joseph & Kuperminc, 2020); however, a better understanding of TDV in this population may indicate interventions should incorporate a multi-pronged approach which addresses TDV as well.

Missing data ranged from 21.8-28.0% at 7^th grade Fall, 27.0-27.1% at 7^th grade Spring, and 24.5%-25.7% at 8^th grade Fall. Mplus (Version 8.4) software was used to conduct analyses using a panel design with cross-lagged and autoregressive paths (Muthén & Muthén, 2017). Consistent with previous literature, scores on bullying and TDV perpetration were dichotomized due to a high amount of zero inflation (see mean scores in Table 1; Norris et al., 2022) as well as a limited amount of response variance (for Bullying 85%-92% of non-zero mean scores ≤ 1; for TDV 92%- 95% of non-zero mean scores ≤ 1; Espelage et al., 2020; Shorey et al., 2018). Thus, models are evaluating bullying and TDV at the level of the population. Models were estimated using full information maximum likelihood (FIML) and Montecarlo Integration. FIML accounted for missing data at 7^th Grade Spring and 8^th Grade Fall. Participants who were missing bullying and TDV data at 7^th Grade Fall were dropped from analyses (4%; n =76). A main effects model was estimated to examine cross-lagged associations between bullying perpetration and TDV perpetration from 7^th Grade Fall to 7^th Grade Spring and from 7^th Grade Spring to 8^th Grade Fall. Effect sizes were reported as Odds Ratios. Although the current study includes a large sample of middle school youth, the amount of TDV perpetrated at each time point was relatively low (n = 34-61), thereby limiting our ability to detect effects. Therefore, we focus our interpretation on the effect sizes. Similarly, gender and sexual orientation are not addressed in the models due to the high amount of missingness we had on these variables—which would have further reduced our ability to detect effects.

Correlations, means, and standard deviations of all variables are reported in Table 1. Percentages of youth who endorsed having dated, bullying perpetration, and TDV perpetration at each time point are reported in Table 2. Paired samples t-test indicated bullying rates were significantly different from 7^th Grade Fall to 7^th Grade Spring, t(1057) = .53, p <.01, and from 7^th Grade Spring to 8^th Grade Fall, t(1097)=.51, p<.01. TDV rates were significantly different from 7^th Grade Fall to 7^th Grade Spring, t(1067)=.24, p =.02, but not from 7^th Grade Spring to 8^th Grade Fall, t(1086)=.24, p =.80.

The current study evaluated bidirectional associations between bullying perpetration and TDV perpetration across 3 time points over 1 year (7^th Grade Fall, 7^th Grade Spring, and 8^th Grade Fall). Findings indicated that bullying and TDV perpetration consistently predicted themselves throughout the course of the study, though rates of perpetration varied. Bullying perpetration consistently predicted subsequent TDV perpetration at all time points and TDV perpetration predicted subsequent bullying perpetration from 7^th Grade Fall to 7^th Grade Spring, but not from 7^th Grade Spring to 8^th Grade Fall. Taken together, these findings suggest bullying and TDV perpetration bidirectionally predict one another over time.

Aligning with hypothesis 1, bullying perpetration consistently predicted itself across time points—aligning with previous research on the stability of bullying perpetration in middle school youth (Crapanzano et al., 2011; Espelage et al., 2001; Zych & Farrington, 2021). Rates of bullying perpetration in our sample at 7^th Grade Fall and 8^th Grade Fall are consistent with other research in adolescents indicating rates of perpetration range from 20% to 35% (Da Silva et al., 2020, Modecki et al., 2014). However, the rate of perpetration at 7^th Grade Spring was surprising (51.4%). The 7^th Grade Spring perpetration rate reflected a marked increase in the number of youth who reported engaging in bullying from 7^th Grade Fall (35.3%) and was followed by a decline at 8^th Grade fall (34.6%). The spike in bullying behavior at the end of 7^th grade adds to the mixed literature regarding the trajectory of peer aggression/bullying in early adolescence. For example, in a sample of aggressive youth, Fite et al. (2008) found peer aggression peaks as youth transition into middle school (i.e., 6^th grade) and then steadily declines. In contrast, Hatigan and Vaillancourt (2014) found aggressive youth steadily increased their engagement in peer aggression throughout middle school, while other youth remain steadily low or unengaged in peer aggression. Taken together, findings support that bullying is a substantial problem for middle school youth, with up to half of youth engaging in bullying perpetration at the end of 7^th grade.

Aligning with hypothesis 1, TDV perpetration also consistently predicted itself across all study time points. Rates of TDV perpetration in the current sample (2.5-4.6%) were low compared to other research on physical TDV in middle school samples, which have found prevalence rates of 15-17% (Simon et al., 2010; Niolon et al, 2015). The differences in our findings may be because Simon et al. (2010) and Niolon et al. (2015) specifically examined high risk samples or because these studies examined lifetime prevalence of perpetration, whereas our study examined perpetration prevalence in the past 6 months. Interestingly, rates of TDV perpetration substantially increased from 7^th Grade Fall (2.5%) to 7^th Grade Spring (4.6%), with perpetration rates nearly doubling despite rates of youth endorsing that they had a dating partner remaining the same during this time (41.6-41.8%). Unlike bullying perpetration, the spike in TDV perpetration in 7^th Grade Spring was not followed by a decline in 8^th grade fall. This suggest that 7^th grade may be an important year where youth are at increased risk of initiating TDV and then maintaining TDV. While longitudinal research on TDV in middle school youth is sparse, extant work indicates perpetration is stable during this time (Cutbush et al. 2016; Temple et al., 2021). A better understanding of the developmental trajectory of TDV perpetration in middle school youth is an important area for future research.

Aligning with hypothesis 2, bullying perpetration at 7^th Grade Fall and 7^th Grade Spring predicted TDV perpetration at subsequent time points (medium effect sizes). Youth who perpetrated bullying at 7^th Grade Fall were approximately 4 times more likely to perpetrate TDV at 7^th Grade Spring, while youth who perpetrated bullying at 7^th Grade Spring were approximately 3 times more likely to perpetrate TDV at 8^th Grade Fall. Findings support past research proposing bullying increases future risk for TDV (Espelage et al., 2022; Joseph & Kuperminc, 2020) and longitudinal work indicating bullying perpetration predicts TDV in middle schoolers (Cutbush et al., 2016; Foshee et al., 2014; Humphrey & Vaillancourt, 2020).

However, partially aligning with hypothesis 3, TDV perpetration at 7^th Grade Fall also predicted bullying perpetration at 7^th Grade Spring (medium effect size), but TDV at 7^th Grade Spring did not substantially predict bullying perpetration at 8^th Grade Fall (small effect size). Youth who perpetrated TDV at 7^th Grade Fall were almost 4 times more likely to perpetrate bullying at 7^th Grade Spring. Notably, TDV perpetration in adolescence can be bidirectional within couples, with individuals both being victimized by and perpetrating violence towards their partner (Norris et al, 2022; Reidy et al., 2016; Wekerle & Wolf, 1999). Therefore, consistent with the Social Learning theory (Bandura, 1976, 1977), previously non-aggressive youth may learn aggressive behaviors through the coercive pattern of victimization and perpetration that emerges within violent dating relationships (Wekerle & Wolf, 1999) and then translate this behavior towards their peers. It was somewhat surprising that TDV perpetration at 7^th Grade Fall did not substantially predict bullying at 8^th Grade Fall. It is possible that this small and non-significant effect may be due to the small number of youth who endorsed TDV perpetration at time 2 (n=61), which limited the ability for our analyses to detect small effects. Findings may also indicate that developmental timing or contextual factors are important to consider, with TDV perpetration as a significant risk factor for future bullying within the same school year, but not as youth transition grades. Findings in the current study are important given previous longitudinal evaluations of bullying and TDV perpetration have focused exclusively on a bullying-to-TDV pathway due to assumptions about the developmental timing of aggression (Zych et al., 2021). Results indicate further research should be conducted to understand the trajectory of TDV in middle schoolers and how early perpetration of TDV increases youth risk for peer violence and maladaptive peer relationships.

Bidirectional results also support research demonstrating bullying and TDV perpetration tend to co-occur in the same youth (Niolon et al., 2015; Vivolo-Kantor et al., 2016; Yahner et al., 2015) and that multiple types of aggression may stem from the same risk factors (Catalano et al., 2012; Foshee et al., 2016; Gladden et al., 2014). However, in contrast to conceptualizations rooted early childhood exposure to aggression (e.g., Social Interaction Theory; Patterson, 1982) or developmental context (phenomenon of heterotypic continuity; Moffit, 1993), findings suggest all peer relationships (including friends and dating partners) may serve as training grounds for each other. That is, aligning with the Social Learning Theory (Bandura, 1976, 1977), youth who learn violence perpetration is effective in dating may transfer this behavior to friendship. Findings support the need for cross-cutting violence interventions that target both bullying and TDV simultaneously. Existing prevention efforts for middle school youth tend to focus exclusively on peer relationships (Joseph & Kuperminc, 2020), with limited focus on the intersections between co-occurring forms of harm.