The current study uses high school follow-up data from a randomized controlled trial (RCT) of Dating Matters: Strategies to Promote Healthy Teen Relationships^® (DM), implemented during middle school, to evaluate its long-term effects on TDV and other relationship behaviors in 9th-11th grade. DM is a comprehensive prevention model that includes unique prevention programs for middle school youth in 6th, 7th, and 8^th grade, as well as three unique programs for parents of 6th, 7th, and 8th graders, with each program designed to be developmentally relevant and building on content delivered in earlier grades. In addition, the model includes training for school staff, a youth communications program, and policy and data activities implemented in the community to promote healthy relationships and prevent TDV and related risk behaviors.¹

DM was evaluated using a longitudinal, comparative effectiveness, multi-site cluster RCT in four US cities. Although some research suggests adolescents living in economically and socially disadvantaged neighborhoods may be at higher risk for TDV (Wincentak et al., 2017), they have been underrepresented in the TDV prevention literature (Teten Tharp et al., 2011). To address this gap, schools included in the Dating Matters RCT were located in urban neighborhoods identified by local health departments as having above average rates of crime and poverty. In middle school, DM was found to prevent TDV perpetration and victimization and reduce the use of negative conflict resolution strategies (ways of resolving a conflict that are ineffective or may lead to higher risk of aggression) (Niolon et al., 2019) and a host of secondary outcomes including other forms of interpersonal aggression and risk behaviors (DeGue et al., 2021; Estefan et al., 2021; Vivolo-Kantor et al., 2021) when compared to an evidence-based, standard-of-care (SC) TDV prevention program implemented in 8th grade only (Safe Dates) (Foshee et al., 1998). No effects were found during middle school on positive relationship behaviors (Niolon et al., 2019). Effectiveness during middle school was examined among the two cohorts of students (Cohorts 3 and 4) who had the opportunity for full exposure to DM in 6th, 7th, and 8th grade and, for dating violence outcomes, only among those who reported dating at any point in middle school (Niolon et al., 2019).

The current study follows the same two full-exposure cohorts into high school and examines long-term effects (assessments in grades 9, 10, and 11 after the intervention ended in 8th grade) on the primary outcomes of the RCT, specifically TDV perpetration and victimization, negative conflict resolution strategies, and positive relationship behaviors, among students who reported dating at any point in middle or high school. Because the multiple components of DM focused on helping middle school students develop healthy relationship skills, we hypothesized that the effects of DM would last beyond the intervention and continue to be protective as youth mature and engage in more intimate romantic relationships. Specifically, we hypothesized that students in DM schools, as compared to students in SC schools, would continue to report less TDV perpetration, less TDV victimization, less use of negative conflict resolution strategies, and higher use of positive relationship skills in high school.

This study examined the longitudinal effects of the Dating Matters (DM) middle school intervention on several TDV-related outcomes in high school. Findings suggest that, on average, the Dating Matters comprehensive prevention model, implemented in middle school, continued to be more effective at reducing the risk of TDV perpetration, TDV victimization, and use of negative conflict resolution strategies in high school than an evidence-based comparison program implemented in 8th grade only, although the results were less consistent across cohort and sex than was true in the middle school evaluation. Results for teen dating violence perpetration, teen dating violence victimization, and negative conflict resolution skills extend the short-term results that were found in middle school (Niolon et al., 2019). However, although no differences were found during middle school on the use of positive relationship skills, significant program effects for this outcome were found in 10th and 11th grades. These findings demonstrate long-term impacts of Dating Matters on our primary outcomes of interest. Although the patterns of these effects were not consistent across groups (i.e., sex, cohort, grade), positive effects were found for both cohorts and sexes on all outcomes (see Fig. 6 for a summary of relative risk reductions for all four outcomes). In general, more significant effects were found among Cohort 4 students than among Cohort 3 students, which is notable considering that Cohort 4 was only followed through 10th grade while Cohort 3 was followed to 11th grade. Overall, significant effects were found for both cohorts, for males and females, and across grades 9 and 10. Only Cohort 3 was assessed in 11th grade, but only one significant effect was found in grade 11 (i.e., PRS for females only), indicating that effects of the program may be waning over time.

For TDV perpetration, significant program effects were found for both cohorts of males in 9th grade and one cohort of males in 10th grade and for both cohorts of females in 10th grade. It is not clear why effects on TDV perpetration were only detected for males in 9th grade, while the effect for females did not emerge until 10th grade. However, the fact that 3 of the 4 groups demonstrated significant program effects in 10th grade on perpetration of TDV following an intervention that ended in 8th grade is noteworthy. Significant average relative risk reductions for TDV perpetration ranged from 33 to 59%, with an average of 19% across all waves and groups (this average includes both significant and nonsignificant effects); the sizes of these average relative risk reductions are also notable considering that Dating Matters was being compared to an evidence-based TDV prevention program delivered in 8th grade. However, the lack of effects in 11th grade suggests that the long-term impacts of the intervention may wane over time, and booster or follow-up programming in high school might be helpful in continuing the effects on TDV perpetration into late adolescence and young adulthood.

For TDV victimization, significant program effects were found for both cohorts of males in 9th grade and one in 10th grade; effects were found for both cohorts of females in 10th grade and one in 9th grade. Overall, the effects suggest the middle school intervention was successful in reducing TDV victimization into high school; this finding is particularly notable given that many of the middle schools in our study did not feed into one high school but sent students to many different high schools, meaning that many of the students in our sample went to high school with peers who had not received Dating Matters in middle school. Significant average relative risk reductions for TDV victimization ranged from 32 to 56%, with an average of 24% risk reduction for DM students compared to SC students across all waves and groups. As with TDV perpetration, no significant program effects for TDV victimization were found for Cohort 3 in 11th grade, indicating the potential for waning effects over time.

For use of negative conflict resolution strategies, a different pattern emerged. Effects were found for Cohort 3 in 9th grade only and for Cohort 4 in 10th grade only. Although potential reasons for this pattern are unclear, the findings suggest that Dating Matters also continued to reduce the use of negative conflict resolution strategies in dating relationships into high school. Significant average relative risk reductions for use of negative conflict resolution strategies were smaller than those for TDV perpetration and victimization, ranging from 11 to 22% with an overall average of 7% risk reduction for DM students compared to SC students across all waves and groups. As with TDV perpetration and victimization, no effects were detected in Cohort 3 in 11th grade for either male or female students.

Significant program effects were found for the use of positive relationship skills for all groups in 10th grade except for Cohort 3 males and for Cohort 3 females only in 11th grade. These effects are particularly interesting given that no effects were found on this outcome in middle school, where sizeable ceiling effects were identified (Niolon et al., 2019). Increased use of positive relationship skills may reflect the developmental trajectory toward having more serious romantic relationships as youth age, providing additional opportunities to use relationship skills gained in the Dating Matters programs, making detection of these effects more likely. Significant average relative risk reductions for lack of ⁵ positive relationship skills were smaller than those for TDV perpetration and victimization, ranging from 9 to 24% with an overall average risk reduction of 3% for DM relative to SC students across all waves and groups.

Overall, a slightly higher number of significant program effects were detected in Cohort 4 than in Cohort 3 (11 significant effects v. 7 in 9th and 10th grades; see Table 1), and it is possible that Cohort 4, having started Dating Matters in 6th grade during the second year of implementation (as opposed to starting 6th grade in the first year of implementation as with Cohort 3), may have received the Dating Matters components delivered with better fidelity to the intended intervention, especially in their first (6th grade) year. It is difficult to determine whether this is the case; anecdotally, we know that sites improved in their administration of all components of Dating Matters after their first year and in each subsequent year and Cohort 4 demonstrated better program effects on some of these outcomes in middle school as well (Niolon et al., 2019).

In sum, evidence of continued effectiveness through high school for this middle school intervention, particularly on TDV outcomes, is promising and adds to existing evidence on the potential for creating sustained change in risk for dating and intimate partner violence through early prevention efforts. These findings add Dating Matters to the relatively short list of TDV prevention programs that have sustained effects beyond the end of program implementation (Foshee et al., 2004a, b; Wolfe et al., 2009). Evaluations of other adolescent risk behavior interventions have found that when effects on primary and secondary outcomes are sustained beyond the intervention, the effect sizes tend to be small and often fade over time (Ellickson et al., 1993; Hale et al., 2014). Despite the relatively robust findings and relative risk reductions in 9th and 10th grade, the lack of significant findings in 11th grade in the one cohort with available data suggests that Dating Matters may be similar to other interventions in terms of waning effects and that implementation during middle school alone may not be sufficient to sustain significant risk reduction through later adolescence. It is notable that Dating Matters was compared to an existing evidence-based intervention, Safe Dates, in this comparative effectiveness trial—thus, any effects measured were those that were above and beyond the anticipated effects of Safe Dates. Safe Dates, implemented in 8th grade in both conditions of this RCT, has previously been found to have long-term effects on teen dating violence in a four-year follow-up through high school (Foshee et al., 2004a, b). Overall, the current findings suggest the need for additional violence prevention efforts in high school to boost and extend skills developed through middle school intervention. Further, to achieve primary prevention, Dating Matters was intentionally designed to be delivered in middle school, before young people typically engage in the more intimate romantic relationships that are common in later adolescence. However, it is possible that older adolescents may need additional education and opportunities for skill-building in high school prior to the drop-off in effects observed in 11th grade. It should be considered that the effects observed in middle school and the earlier years of high school may translate into reduced risk for other outcomes in later adolescence and adulthood; for example, prevented TDV victimization in 9th grade could mean reduced sexual risk behavior or reduced mental health problems in late adolescence, even if effects on TDV victimization wane by 11th grade. Further research could examine such associations.

The findings from this study highlight the benefit of early and comprehensive TDV prevention efforts that address risk and protective factors among youth but also with their parents or caregivers, schools, and communities. Our results, overall, suggest that this multi-year, comprehensive prevention model is more effective than the single-year, single program model to which we compared it in, at least, the short and medium term and may help set the stage for long-term risk reductions through adulthood, especially if combined with additional prevention approaches through early adulthood. Further research should examine component-specific effects to help determine whether all components are necessary to maintain program effects and to examine mechanisms of change where program effects exist. Implementation research should investigate the feasibility of implementing Dating Matters with fidelity in the field, given the resources necessary to implement its multiple components and multi-year design. Prior research with Dating Matters also points to the potential for effectiveness beyond the primary intended outcomes of TDV and relationship behaviors to other forms of violence and risk behavior (DeGue et al., 2021; Estefan et al., 2021; Vivolo-Kantor et al., 2021); additional research should assess whether those effects are sustained as well. The identification of effective comprehensive prevention models that address cross-cutting risk and protective factors for violence, such as healthy communication skills and social-emotional development, can reduce the burden on communities to implement multiple prevention strategies addressing different but related health outcomes. Early comprehensive prevention efforts can improve the overall health and well-being of youth, and their communities, throughout their lifetimes.