Longitudinal Data on Teen Dating Violence: Meeting Summary

June 7-8, 2011
Washington, D.C.

Welcome and Introductions

The June 7 discussion opened with comments from Katrina Baum, Division Director, Violence and Victimization Research Division, NIJ (who spoke in place of Phyllis Newton, Director, Office of Research and Evaluation, NIJ). Baum said that all would agree “teen dating violence [TDV] is a critical public health issue.” High prevalence rates, associations and a range of negative correlates for both victims and perpetrators make this an issue that must be examined more closely, Baum said.

Baum continued by noting that NIJ’s Violence and Victimization Research Division of the Office of Research and Evaluation is fully committed to advancing knowledge and disseminating information about TDV. NIJ has actively researched violence against women for about 15 years, but only in the last 5 years has there been a recognized need to start examining adolescence (and earlier in childhood) to understand the factors that put individuals at risk in abusive romantic relationships, Baum noted. She said that NIJ began looking at teen violence with a small workshop in 2005, and that the Federal Interagency Workgroup sprang from that workshop and now meets regularly. The workgroup comprises 19 agencies across the Departments of Health and Human Services, Justice, Education and Defense. She pointed out that NIJ has nearly a dozen active grant awards in the area of adolescent relationship abuse and that much progress has been made in understanding TDV. Baum noted important forays into preventing such violence, but acknowledged “there is much remaining we do not know.” Baum said that the interagency workgroup has stated a need for longitudinal research to identify the path from childhood experiences (e.g., exposure to violence) and behavior problems (e.g., bullying, aggression) to adolescent victimization and perpetration of TDV and to later victimization and perpetration of intimate partner violence among young adults.

This meeting, Baum said, is about identifying gaps in knowledge and “maximizing our investments and best utilizing existing resources.” She then identified four groups selected to attend this conference:

  1. Representatives from research groups using existing longitudinal data sets collected for other primary purposes, but where TDV outcomes were collected that have not yet been analyzed or have been underutilized
  2. Representatives from research groups using longitudinal data sets that begin or continue into middle childhood/early adolescence where TDV measures could be added in subsequent waves of data collection
  3. Experts in TDV research
  4. Federal partners

Baum expressed optimism about the potential of what could be discovered in this meeting and how alliances could be forged on the path ahead. The Violence and Victimization Research Division “is increasingly looking toward taking a lifespan developmental perspective in all of our work as we seek to examine the causes and correlates of violence and victimization over time, particularly how early victimization experiences relate to future violent behavior and/or re-victimization,” Baum stated. She expressed hope that this meeting and subsequent efforts would serve as a model for the future.

Dara Blachman-Demner and Carrie Mulford of NIJ then asked the meeting participants to briefly introduce themselves.

What We Know from Longitudinal Research on Teen Dating Violence

Dorothy L. Espelage of the University of Illinois, Urbana-Champaign, began the substantive discussion with an assessment of the current state of knowledge about TDV based on what is known from longitudinal (published) studies.

Espelage set forth the goals of this brief review paper:

  • Summarize what we already know from existing longitudinal research about predictors and consequences of TDV.
  • Identify existing longitudinal data sets where TDV measures could be added in subsequent waves of data collection. The most appropriate data sets would be those with data already collected from participants in young and/or middle childhood who are currently in fourth through ninth grades.
  • Identify existing longitudinal data sets collected for a primary purpose other than the understanding of relationship abuse and violence, but where TDV outcomes were collected. The most appropriate data sets would be those with data collected during adolescence and young adulthood, where there are data available on TDV that have not yet been analyzed or have been underutilized.

This review, according to Professor Espelage, adopted a social-ecological framework that expanded on Foshee and Matthew (2007) and their review of 12 longitudinal studies:

  • Arriaga and Foshee, 2004
  • Bank and Burraston, 2001
  • Brendgen et al., 2001
  • Capaldi and Clark, 1998
  • Capaldi et al., 2001
  • Foshee et al., 2001
  • Foshee et al., 2005
  • Gorman-Smith et al., 2001
  • Lavoie et al., 2002
  • Ozer et al., 2004
  • Simons, Lin and Gordon, 1998
  • Wolfe et al., 2004

Working from a slide presentation, Espelage summarized the following conclusions:

  • Current data on the consequences of dating violence stem largely from cross-sectional or longitudinal studies with major compromises to internal validity, questionable sampling techniques and assessments/questionnaires with weak psychometric properties (Library of Congress, 2011).
  • Why the lack of longitudinal studies?
  • We need to consider how IRB constraints might be contributing to this.
  • Surveillance studies assess perpetration and victimization with single-item indicators, which will never allow for a complex understanding of TDV.
  • More observational studies are needed, particularly studies with romantic partners, and social network analysis is necessary to assess peer influence.
  • Many theories remain untested given the dearth of longitudinal studies focused specifically on TDV (for review, see Shorey, Cornelius and Bell, 2008).
  • TDV is a gendered interaction, yet very little longitudinal work has adopted a gendered framework. Simply adding TDV indicators to existing data sets without close scrutiny of the study design, sample and measurement might not advance these theories; new prospective studies may need to be proposed instead.

For more information, please see the final review paper Commissioned Paper for National Institute of Justice Research Meeting on Longitudinal Data on Teen Dating Violence (pdf, 40 pages)

Response and Discussion

Several TDV experts were asked to react and add to the conversation about what is known about adolescent relationship abuse from existing longitudinal data. The experts were also asked to comment on which key questions still need to be answered from longitudinal data. Some of the key points raised during this discussion were:

  • It is important to specify the types of violence and to use the information in analyses. Frequency scales mostly get collapsed into yes/no categories and we lack information about severity, motives and context. Determining how violence is different in different types of relationships requires more specificity. Similarly, our understanding of TDV across relationships (why an individual might be abusive with one partner, but not another) is limited.
  • It is important to understand the implications of online relationships and the role of technology in TDV. We don’t really know what is normative and nonabusive in communication via technology.
  • Studying gender differences between boys’ and girls’ experiences with TDV is important, and you need to collect data from boys and girls to make comparisons about their experiences. The use of, response to and impacts of the use of violence in relationships are areas in which more research comparing boys and girls could be beneficial.
  • More research is needed in the following areas: biological predictors, replication of predictors, consequences of TDV perpetration and victimization, gay and lesbian relationship abuse, attitudes of parents and peers, educational outcomes as predictors and/or outcomes, the role of emotion and emotion regulation in TDV, demographic patterning, social networking data, and sociometric predictors.
  • There is little theoretical work in the field.
  • Definitions of dating and abuse among adolescents are variable from one study to another. Consolidating the measures of TDV into a compendium would be helpful.
  • Longitudinal research needs to be collected for longer periods of time and at frequent intervals. Trajectories at 2-3 years are very different from those that have longer time periods. Similarly, the more frequently the sample group is asked, the more frequently physical abuse/violence is reported. If long periods of time pass between data points, the sample tends to forget or fails to mention instances of violence that may have occurred.
  • In need of further research is understanding how TDV reporting is affected by relationship status and breakups.
  • Data collected outside of the school setting may be desirable. Schools are difficult to access and there are limitations of data that come from school samples (e.g., missing group of youth most at risk, less likely to include sexual victimization or information about sexual orientation). Relatedly, more information is needed about the impact of community- and neighborhood-level variables.

Existing and Current/Future Data Set Overview and discussion

In advance of the meeting, most of the participants were asked to come prepared to discuss the data set(s) that participants identified as fitting into one of two categories: 1) existing data that included TDV measures but either have not been analyzed or have been underutilized; or 2) data collections with younger children who are moving into the adolescent years where TDV measures could be added. Participants were asked to describe the purpose of the study, the data set, TDV measures that were collected, analyses related to TDV that have been done and plans for the future. These presentations and subsequent discussions comprised the meeting.

Time and Funding for Data Analysis

As part of this discussion, participants were asked what they would need to analyze existing data. Almost all responded that time and lack of funding were the primary factors that prevented them from using the TDV data. Several participants made the point that there is a problem when data are collected but there is no funding for data analysis. Some also indicated that they didn’t realize the level of interest in that data prior to this meeting.

The federal representatives and the researchers agreed that the lack of funding is a concern. A suggestion was made to create a roster of post-docs who are interested and knowledgeable about adolescent relationship abuse. There were also discussions about possible mechanisms for collaboration. For some topics, such as the intergenerational transmission of partner violence, no one study had a sufficient sample size to conduct analyses with adequate power. However, there was interest in a snapshot approach, in which individual investigators could use their data to answer one question, and the separate results could be combined into one paper or special issue. Another possibility involved pooling data when the measures were sufficiently similar and conducting meta- or mega-analyses.

A Need for Definitions

The start of the substantive discussion on the first day sparked a question that became a key theme for the entire meeting:

Have the participants and their research effectively — and accurately — defined the operative terms “relationship” and “dating”?

Several participants agreed that there needs to be a close examination of what is meant by the term “relationship,” as dating has changed drastically even over the last few years. One participant pointed out that many young people customarily view relationships as trysts that may occur over 25 minutes in a public restroom. People send naked pictures of each other online now, even though they may have never met in person. If these new types of relationships are dates, then we need to redefine dating and dating violence. “What is dating for a fifth grader?” It is unclear whether a fifth grader would answer this question in the same way that one might have 30 years ago. If the norms have changed, then the research can become highly suspect.

Other participants, however, were not so troubled about formulating new definitions of dating, saying that, while corsages may not be involved, kids are still dating because they have feelings for each other and are spending time together. Following the suggestion for the need for more formative research in this area, Carrie Mulford spoke of an ongoing multiagency project that will use concept mapping to understand how adults and youth view relationships in disparate ways. It was also suggested that collaborating with developmental researchers who are examining normative romantic relationships could be quite useful.

On a related point, when the question was raised about which measure researchers should use if they wanted to add TDV to existing data, a long discussion ensued about the need to develop a better measure of TDV that incorporates critical contextual constructs such as impacts, fear and injury. The decision was made to table the discussion and note that a separate meeting was needed to address the issues of definition and measurement.

Lessons Learned: Panel Discussion

A lunch panel on the first day of the meeting comprising experts in the field who had collected longitudinal data on TDV was asked to shed light on what others who want to move into collecting TDV data should know. The panelists were asked to offer advice and to mention things they would have done differently. Here is a highlight of some of the tips:

  • Examine mutual violence in a more thoughtful way by focusing less on a count of the acts of violence and more on the consequences of the violence.
  • Ask about relationship violence early in adolescents’ lives.
  • Consider emotional processes (e.g., anger) in addition to traditional focus on power and control.
  • High-risk samples may be necessary to fully understand the complexities of relationship dynamics involved in TDV.
  • Ask more about girls’ perpetration of violence and norms related to girls’ use of violence against boys.
  • Steer away from jargon and slang.
  • Add more measures on healthy and respectful aspects of relationships.
  • Make sure to distinguish self-defense and play fighting from other types of aggression.
  • Assess injury, impact and fear for a more comprehensive picture of relationship violence.

Surveillance Data Perspectives

A lunch presentation on the second day by Sherry Hamby of Sewanee: The University of the South provided a national surveillance perspective on TDV. Several key points are summarized below:

  • Nationally representative surveillance data are critical for documenting the extent of a problem as well as for making decisions about programming and resources and for providing clinical norms.
  • Prior to the National Survey of Children’s Exposure to Violence (NatSCEV) fielded in 2008, few surveys focused on youth victimization with the types of offenses most relevant in childhood. In addition, there was no nationally representative survey coverage for children younger than age 12, and most surveys only focused on a small number of violence types.
  • NatSCEV is a national random-digit dial telephone survey of 4,549 youth ages 1 month to 17 years conducted between January and May 2008 (caregiver interviews for youth ages 0-9) that assessed more than 40 types of victimization. There was an oversampling of African American, Hispanic and low-income youth.
  • Poly-victimization (e.g., more than one type of victimization) is the rule rather than the exception, such that even studies that focus on one type of violence (e.g., TDV, child maltreatment) are likely studying many of the same youth. In NatSCEV, there was not a single victim of TDV who did not also report at least one other type of victimization. In addition, in general, trauma symptoms seem to be more related to variety and number of exposures than to specific victimization types.
  • TDV showed an interesting pattern of gender differences that highlight the importance of considering emotions such as fear in our assessment of TDV. For males, TDV rated second to last out of 43 types of victimizations with respect to fear inducement. For females, it was number 13.

Meeting Wrap-up

After 2 intense days, NIJ moderators Dara Blachman-Demner and Carrie Mulford expressed optimism and said that the meetings allowed for more focus on pursuing the most useful path going forward. They summarized five major themes/ideas that emerged from the meeting:

  1. Several topics have been examined in numerous studies and may lend themselves to the snapshot activity concept (e.g., although one study may not provide enough data or rigor to answer questions on each topic, pooling results could be useful). Specific topics could include:
    • The role of emotions, such as anger
    • Intergenerational transmission of violence
    • Biological measures, such as genetic information and salivary cortisol (a stress hormone)
    • Cognitive-/perception-/appraisal-related assessments
    • Relationship context (e.g., quality, duration)
    • Neighborhood influences
    • Peer norms and behaviors
  2. Similarly, given the number of studies that have collected data examining predictors of young adult intimate partner violence (primarily using the Conflict Tactics Scale), a meta-analysis could be a contribution to the field.
  3. Measurement continues to be an area in which more work is necessary. In addition to the issues summarized above regarding definitions and context, more research is necessary to understand the role of meaning/interpretation of behaviors and how that may vary by gender as well as the role of culture. In addition, exploring the increasing use of observational measures, “experience sampling” techniques and social network analyses are all promising areas. Some believe that a compendium of existing measures would also be useful.
  4. We need to be sure to bring a theoretical perspective to all of our work in TDV.
  5. The role of technology in dating violence needs to be considered analytically. We need to be thoughtful about how we see it fitting into our overall theories and develop specific hypotheses instead of simply doing descriptive work. For example, is it similar to the availability of guns and drugs in that it creates more opportunity for abuse because it is increasingly accessible? Or is the abuse perpetrated by technology somehow qualitatively distinct, perhaps because of its potential for more far-reaching implications?

The meeting was adjourned and participants agreed they would look forward to future opportunities for discussion and collaboration.


Arriaga, X.B., and V.A. Foshee. “Adolescent Dating Violence: Do Adolescents Follow in Their Friends’ or Their Parents’ Footsteps?” Journal of Interpersonal Violence 19 (2004): 162-184.

Bank, L., and B. Burraston. “Abusive Home Environments as Predictors of Poor Adjustment During Adolescence and Early Childhood.” Journal of Community Psychology 29 (2001): 195-217.

Brendgen, M., F. Vitaro, R.E. Tremblay and F. Lavoie. “Reactive and Proactive Aggression: Predictions to Physical Violence in Different Contexts and Moderating Effects of Parental Monitoring and Caregiving Behavior.” Journal of Abnormal Child Psychology 29 (2001): 293-304.

Capaldi, D.M., and S. Clark. “Prospective Family Predictors of Aggression Toward Female Partners for At-Risk Young Men.” Developmental Psychology 34 (6) (1998): 1175-1188.

Capaldi, D.M., T. Dishion, M. Stoolmiller and K. Yoeger. “Aggression Toward Female Partners by At-Risk Young Men: The Contribution of Male Adolescent Friendships.” Developmental Psychology 37 (2001): 61-73.

Foshee, V.A., F. Linder, J.E. MacDougall and S. Bangdiwala. “Gender Differences in the Longitudinal Predictors of Adolescent Dating Violence.” Preventive Medicine 32 (2001): 128-141.

Foshee, V.A., S.T. Ennett, K.E. Bauman, T. Benefield and C. Suchindran. “The Association Between Family Violence and Adolescent Dating Violence Onset: Does It Vary by Race, Socioeconomic Status, and Family Structure?” Journal of Early Adolescence 25(2005): 317-344.

Foshee, V.A., and R.A. Matthew. “Adolescent Dating Abuse Perpetration: A Review of Findings, Methodological Limitations, and Suggestions for Future Research.” In The Cambridge Handbook of Violent Behavior and Aggression, ed. D.J. Flannery, A.T. Vazsonyi, and I. D. Waldman. New York, NY: Cambridge University Press, 2007.

Gorman-Smith, D., P.H. Tolan, A.J. Sheidow and D.B. Henry. “Partner Violence and Street Violence Among Urban Adolescents: Do the Same Family Factors Relate?” Journal of Research on Adolescence 11(2001): 273-295.

Lavoie, F., M. Hebert, R. Tremblay, F. Vitaro, L. Vezina and P. McDuff. “History of Family Dysfunction and Perpetration of Dating Violence by Adolescent Boys: A Longitudinal Study.” Journal of Adolescent Health 30 (2002): 375-383.

Library of Congress. “Teen Dating Violence: An Annotated Bibliography.” A report prepared by the Federal Research Division, Library of Congress, under an interagency agreement with the Violence and Victimization Research Division, National Institute of Justice, April 2011.

Ozer, E.J., J.M. Tschann, L.A. Pasch and E. Flores. “Violence Perpetration Across Peer and Partner Relationships: Co-occurrence and Longitudinal Patterns Among Adolescents.” Journal of Adolescent Health 34 (2004): 64-71.

Shorey, R.C., T.L. Cornelius and K.M. Bell. “A Critical Review of Theoretical Frameworks for Dating Violence: Comparing the Dating and Marital Fields.” Aggression and Violent Behavior 13 (2008): 185-194.

Simons, R.L., K. Lin and L.C. Gordon. “Socialization in the Family of Origin and Male Dating Violence: A Protective Study.” Journal of Marriage & Family 60 (1998): 467-478.

Wolfe, D.A., C. Wekerle, K. Scott, A. Straatman and C. Grasley. “Predicting Abuse in Adolescent Dating Relationships Over 1 Year: The Role of Child Maltreatment and Trauma.” Journal of Abnormal Psychology 113 (2004): 406-415.

Date Created: October 12, 2011