Vagi KJ, O’Malley Olsen E, Basile KC, Vivolo-Kantor AM. Teen Dating Violence (Physical and Sexual) Among US High School StudentsFindings From the 2013 National Youth Risk Behavior Survey. JAMA Pediatr. 2015;169(5):474-482. doi:10.1001/jamapediatrics.2014.3577
National estimates of teen dating violence (TDV) reveal high rates of victimization among high school populations. The Centers for Disease Control and Prevention’s national Youth Risk Behavior Survey has provided often-cited estimates of physical TDV since 1999. In 2013, revisions were made to the physical TDV question to capture more serious forms of physical TDV and to screen out students who did not date. An additional question was added to assess sexual TDV.
To describe the content of new physical and sexual TDV victimization questions first administered in the 2013 national Youth Risk Behavior Survey, to share data on the prevalence and frequency of TDV (including the first-ever published overall “both physical and sexual TDV” and “any TDV” national estimates using these new questions), and to assess associations of TDV experience with health-risk behaviors.
Design, Setting, and Participants
Secondary data analysis of a cross-sectional survey of 9900 students who dated, from a nationally representative sample of US high school students, using the 2013 national Youth Risk Behavior Survey.
Main Outcomes and Measures
Two survey questions separately assessed physical and sexual TDV; this analysis combined them to create a 4-level TDV measure and a 2-level TDV measure. The 4-level TDV measure includes “physical TDV only,” “sexual TDV only,” “both physical and sexual TDV,” and “none.” The 2-level TDV measure includes “any TDV” (either or both physical and sexual TDV) and “none.” Sex-stratified bivariate and multivariable analyses assessed associations between TDV and health-risk behaviors.
In 2013, among students who dated, 20.9% of female students (95% CI, 19.0%-23.0%) and 10.4% of male students (95% CI, 9.0%-11.7%) experienced some form of TDV during the 12 months before the survey. Female students had a higher prevalence than male students of physical TDV only, sexual TDV only, both physical and sexual TDV, and any TDV. All health-risk behaviors were most prevalent among students who experienced both forms of TDV and were least prevalent among students who experienced none (all P < .001).
Conclusions and Relevance
The 2013 TDV questions allowed for new prevalence estimates of TDV to be established that represent a more complete measure of TDV and are useful in determining associations with health-risk behaviors among youth exposed to these different forms of TDV.
Although there has been research on teen dating violence (TDV) for several decades, the subject has only received attention as a public health concern in recent years.1- 3 The Centers for Disease Control and Prevention’s (CDC’s) definition of TDV comprises several forms of violence including physical, sexual, psychological, and stalking behaviors,4 although researchers rarely measure all types in the same study.5 Associations exist between various forms of TDV victimization and a range of both short- and long-term negative health outcomes. For example, cross-sectional research indicates that TDV victimization is associated with increased alcohol and tobacco use, depressive symptoms and suicidality, internalizing behaviors, eating disorders, and risky sexual behaviors (eg, not using condoms and multiple sexual partners).6,7 Longitudinal studies8- 11 have demonstrated that TDV victimization has lasting effects because those persons entering adulthood with a history of TDV victimization report negative consequences such as depression, substance use, suicidality, and future intimate partner violence victimization.
Over time, prevalence estimates of physical TDV victimization from the CDC’s national Youth Risk Behavior Survey (YRBS) (first measured in 1999) have remained around 9% with similar rates among female and male students.12 Until recently, there have been no ongoing national studies of sexual TDV to our knowledge, but one study from the 2005 National Survey of Adolescents3 found higher lifetime prevalence rates of sexual TDV for girls than for boys. In addition, results from studies that used convenience samples found rates of sexual TDV varying from 8% to 25% for females and from 5% to 11% for males, depending on the questions used and populations assessed.13- 15
The 1999 through 2011 national YRBS included a single TDV victimization item (“During the past 12 months, did your boyfriend or girlfriend ever hit, slap, or physically hurt you on purpose?”) that only assessed physical behaviors. This question combined less serious acts of aggression (eg, slapped) with more serious acts (eg, hurt you on purpose). Also, the location of “on purpose” made it unclear if it was referring to “physically hurt” or to all the types of aggression included in the question. As a result, this item may have also captured play-fighting or “horseplay,” which has been shown to be common among adolescent dating partners16,17; researchers have pointed out that the boundaries between play-fighting/flirting and actual aggression in adolescence are not always clear.18- 20 The previous question also only asked about violence perpetrated by a “boyfriend or girlfriend,” which may have inadvertently included only serious relationships and excluded more casual dating relationships. Notably, the previous YRBS measure did not differentiate between those who reported dating and those who did not, and the only response options to the question were “yes” and “no,” rather than assessing the frequency of TDV. Using “yes” and “no” response options limits variance, which not only can hinder efforts to detect meaningful differences but can also fail to provide a sense of burden, particularly for those youth who experience violence more frequently. Furthermore, research shows that many teens experience TDV more than once and that the violence can be stable within a given relationship.21
To address these weaknesses, the CDC engaged in a formal process with TDV experts within the CDC’s Division of Violence Prevention and with external researchers and practitioners to reexamine the items for the 2013 administration. For the 2013 YRBS, the CDC revised the physical TDV question by stating “physically hurt you on purpose” prior to listing potential physical dating violence types to indicate that any harm experienced was intentional, including only more serious acts in the question such as injuring with an object or weapon, rewording “boyfriend or girlfriend” to more inclusive language used in other reliable and valid TDV measures (ie, “dating or going out with”),22,23 having students identify whether they have dated during the past 12 months, and asking about the frequency of the violence. In addition, the CDC created 1 new survey question to assess sexual TDV victimization, which others have called for in TDV measurement5 and which, to our knowledge, does not exist on any ongoing national survey of adolescents. The specific question wording and response options are described in our analysis.
By including questions on both physical and sexual TDV, we are able to look at those youth who experienced physical TDV only, sexual TDV only, both physical and sexual TDV, any TDV (ie, those who experienced physical TDV only, sexual TDV only, or both physical and sexual TDV), and none. These distinctions may be particularly important when investigating health outcomes associated with different types or combinations of TDV because some health-risk behaviors have been shown to be associated with certain types of TDV but not others.24 By measuring 2 different types of TDV, we can determine whether there is variation in association with health-risk behaviors by type of TDV.
The purpose of our study is to describe the content of the new 2013 physical and sexual TDV questions; to present the updated prevalence estimates for TDV, including the first-ever published “both” and “any” national estimates and the frequency of any TDV among both female and male students; and to examine differences in health-risk behaviors by type of TDV victimization. We hypothesize that the 2013 physical and sexual TDV questions will have strong and nuanced associations with selected health-risk behaviors (eg, suicide ideation and attempts, violence and bullying, alcohol and other drug use, and sexual risk behaviors).
The CDC developed the Youth Risk Behavior Surveillance System to monitor priority health-risk behaviors among youth. The national school-based YRBS is a cross-sectional survey that has been conducted biennially since 1991. In each survey year, an independent 3-stage cluster-sample design is used to obtain a nationally representative sample of public and private school students in grades 9 through 12 in the 50 states and the District of Columbia. Student participation in the survey is anonymous and voluntary, and local parental permission procedures are used. Students record their responses directly on a self-administered computer-scannable questionnaire. A weighting factor is applied to each record to adjust for nonresponse and the oversampling of black and Hispanic students. An institutional review board at the CDC approved the national YRBS. More details regarding sampling strategies and the psychometric properties of the YRBS questionnaire are reported elsewhere.25,26
Physical TDV was assessed with the following question: “During the past 12 months, how many times did someone you were dating or going out with physically hurt you on purpose? (Count such things as being hit, slammed into something, or injured with an object or weapon).” The new sexual TDV question was: “During the past 12 months, how many times did someone you were dating or going out with force you to do sexual things that you did not want to do? (Count such things as kissing, touching, or being physically forced to have sexual intercourse).” Response options for both items were: “I did not date or go out with anyone during the past 12 months,” “0 times,” “1 time,” “2 or 3 times,” “4 or 5 times,” and “6 or more times.” Because of skewed frequency levels and for ease in interpreting the ratios, the responses for both TDV variables were dichotomized into 0 times and 1 or more times for all bivariate and multivariable analyses.
We used the physical and sexual TDV questions to create a 4-level (ie, “physical TDV only [physical TDV: ≥1 times, sexual TDV: 0 times],” “sexual TDV only [physical TDV: 0 times; sexual TDV: ≥1 times],” “both physical and sexual TDV [physical TDV: ≥1 times, sexual TDV: ≥1 times],” and “none [physical TDV: 0 times, sexual TDV: 0 times]”) and a 2-level combined TDV measure, which was dichotomized into “any TDV” and “none.” The YRBS prevalence of physical and sexual TDV have been published elsewhere27; however, the physical and sexual TDV measures presented in the present study describe students who experienced only physical TDV or only sexual TDV (rather than students who experienced physical or sexual TDV regardless of whether they also experienced the other). Students who responded that they did not date or go out with anyone during the 12 months before the survey and students who have missing data for either TDV survey question were excluded from both variables.
We examined associations between each type of TDV and various health-risk behaviors. These behaviors, selected because they have been shown to be associated longitudinally with TDV,9,24 included suicide ideation and attempts, violence and bullying, alcohol and other drug use, and sexual risk behaviors (ie, multiple sex partners and currently sexually active). For this analysis, all health-risk behaviors were dichotomized into “no/0 days/0 times” or “yes/1 or more days/1 or more times.” More information regarding the survey questions is available at http://www.cdc.gov/healthyyouth/yrbs/pdf/questionnaire/2013_xxh_questionnaire.pdf (accessed May 9, 2014).
The national YRBS used 2 questions to assess race and ethnicity. Students were classified as white, non-Hispanic (referred to as “white”), black or African American, non-Hispanic (referred to as “black”), and Hispanic or Latino (referred to as “Hispanic”). The numbers of students from other racial/ethnic groups were too small for meaningful analysis.
All analyses were conducted in SUDAAN version 10.0.1 (Research Triangle Institute) to account for the complex sample design of the national YRBS, and all prevalence estimates reported herein reflect weighted estimates. The significance level was set at 5%. Because female and male students experience TDV differently,3,22,28 all bivariate and multivariable analyses were stratified by sex, and no overall estimates are reported. Bivariate associations were tested using overall χ2 tests. Multiple logistic regression models were used to separately assess the association between TDV and each health-risk behavior, controlling for race/ethnicity and grade in school; these associations are reported as adjusted prevalence ratios with 95% CIs29 with “none” as the referent group. Because these are cross-sectional survey data, adjusted prevalence ratios are more appropriate than adjusted odds ratios; adjusted prevalence ratios are mathematically identical to adjusted risk ratios and can be interpreted in a similar way.
The 2013 national YRBS had a student response rate of 88%, a school response rate of 77%, and an overall response rate of 68%. Of the 13 633 completed questionnaires, 50 failed quality control and were excluded from the data set, leaving a total of 13 583 usable questionnaires; of those, 13 097 (96.4%) were valid responses for both TDV questions. The analytic sample was 50.9% female, 56.2% white, 15.3% black, and 20.8% Hispanic, and roughly 25% of the students were in each grade (ninth through 12th).
Table 1 shows the 2013 prevalence of TDV among students who dated during the 12 months before the survey by demographic subgroups. Among the 75.0% of female students who dated during the past 12 months, the prevalence of TDV was 6.6% for physical only, 8.0% for sexual only, 6.4% for both physical and sexual, and 20.9% for any TDV; among the 72.8% of male students who dated during the past 12 months, the prevalence of TDV was 4.1% for physical only, 2.9% for sexual only, 3.3% for both physical and sexual, and 10.4% for any TDV. The distribution of TDV differed by sex (P < .001). The prevalence of TDV also varied by race/ethnicity, with black and Hispanic students generally experiencing more TDV than white students, but did not vary by grade.
Table 2 shows the 2013 prevalence of TDV among students who dated during the 12 months before the survey broken down by frequency of victimization. Among all subgroups, the vast majority of students did not report experiencing TDV, but most students who experienced TDV experienced more than 1 incident. For example, among female students, 4.8% reported 1 incident of physical TDV, while more than 8% reported 2 incidents or more, and 5.7% reported 1 incident of sexual TDV, while more than 8% reported 2 incidents or more. Similarly, among male students, 2.3% reported 1 incident of physical TDV, while more than 5% reported 2 incidents or more, and 2.0% reported 1 incident of sexual TDV, while more than 4% reported 2 incidents or more.
Table 3 shows bivariate associations between TDV and selected health-risk behaviors, and Table 4 shows the adjusted associations. For both male and female students, every risk behavior was most prevalent among students who had experienced both physical and sexual TDV and least prevalent among students who experienced no TDV (all P < .001). For both male and female students, the adjusted rates were higher for every risk behavior among students experiencing physical TDV only, compared with students experiencing none, and among students experiencing both physical and sexual TDV, compared with students experiencing none. The associations were not consistent among students who experienced sexual TDV only. Female students who experienced sexual TDV only were more likely than those who experienced none to seriously consider attempting suicide, make a suicide plan, attempt suicide, get in a physical fight, carry a weapon, be electronically bullied, and report current alcohol use and binge drinking. Male students who experienced sexual TDV only were also more likely than those who experienced no TDV to experience these same health-risk behaviors, as well as to have had sex with 4 or more people and to be currently sexually active.
The purpose of the present study was to describe the content of and findings from new physical and sexual TDV victimization questions first administered in the 2013 YRBS. The CDC updated the existing physical TDV question for the 2013 YRBS to account for more serious forms of violence, to allow students to indicate that they did not date during the past 12 months, and to measure the frequency of physical TDV. The 2013 YRBS also included a new sexual TDV question. The results of our study suggest that both physical and sexual TDV are prevalent among high school students and that significant sex differences exist in both outcomes. Approximately 1 in 5 female students and 1 in 10 male students have been victims of physical and/or sexual TDV during the past 12 months. Female students had double the prevalence of any form of TDV than male students. Consistent with other research, most victims of physical or sexual TDV reported more than 1 incident, suggesting that TDV is not usually an isolated incident. Physical and sexual TDV victimization was associated with several health-risk behaviors.
Our analyses also indicate that, although health-risk behaviors were prevalent among those who experienced any form of TDV, for female students, it is especially important that we can now look at sexual TDV in the YRBS. For male more than female students, a combined physical and sexual TDV measure produces stronger associations with the health-risk behaviors than physical or sexual TDV alone. For example, compared with students who experienced either physical or sexual TDV, female students who experienced both forms of TDV were approximately twice as likely to attempt suicide, and male students who experienced both forms of TDV were roughly 3 times as likely to attempt suicide. These findings suggest that, consistent with previous research,6 there may be different health risks related to the type of violence experienced and that there may be a cumulative negative effect for victims experiencing both forms of TDV.
Furthermore, the literature suggests differential variation by sex depending on the form of TDV.22,30 Some studies15,22 suggest that female and male students report physical TDV victimization at similar rates but that female students report more incidents of sexual TDV victimization than do male students. As Hamby and Turner5 point out, the studies that have shown sex parity typically show it for physical (but not sexual) aggression. These studies often combine measures of severe physical acts with less severe acts (eg, pushing and shoving) that are less likely to result in serious injury.31,32 The findings in the present study are consistent with previous studies examining more severe forms of TDV by sex (eg, hitting, hurting with a weapon, and forced sexual penetration) that have shown that female adolescents report more physical and sexual victimization than male adolescents.3,28
Although the data cannot directly answer this question, the lower physical TDV prevalence estimates among male students in 2013 compared with previous administrations of the YRBS may be due, in part, to the fact that the new question clarifies the intent to physically harm and eliminates confusion around whether or not to include play-fighting; Foshee et al33 have found that teens often use physical contact (eg, scratching and twisting arms) as a form of flirting in dating relationships. The change to include only those who dated in the denominator is an important clarification that allows for a more accurate understanding of TDV victimization. Enabling students to indicate that they have not dated during the time period of interest increases the accuracy of the measures by excluding acts that may have been perpetrated outside the context of a dating relationship.2,34 In addition, sexual violence is often a component of the TDV victimization experience.5,22 Even though the YRBS does not afford the space to measure TDV comprehensively, the new items represent an expansion and fine-tuning of the TDV measurement.
These results present broader implications for TDV prevention efforts. Although female students have a higher prevalence than male students, male and female students are both impacted by TDV, and prevention efforts may be more effective if they include content for both sexes. Our study findings also suggest that prevention efforts can take a comprehensive approach to preventing TDV and health-risk behaviors. Because TDV victimization was associated with a constellation of health-risk behaviors, it is possible that implementing TDV prevention programming may also affect the rates of these behaviors.
There are a few limitations of our study. First, because of the limited space for questions in the YRBS, we were unable to assess all aspects of TDV, including psychological aggression and stalking, nor were we able to include several behaviorally specific items for each type of violence, which is recommended to increase disclosure.5,35 We also could not measure fear or injury associated with TDV to get further context around the violence, which has been suggested by Hamby and Turner.5 In addition, the YRBS data are cross-sectional and can only provide an indication of associations between TDV and the selected health-risk behaviors. These data are only generalizable to students who attend school and may not be representative of all people in this age group. In 2009, approximately 4% of people in the United States 16 to 17 years of age were not enrolled in a high school program and had not completed high school.36 Finally, these data are self-reported. Although the extent of underreporting or overreporting of TDV on this survey cannot be determined, the YRBS questions assessing other risk behaviors have been shown to have good test-retest reliability.26
We believe that we now have a more relevant and robust estimate of TDV, by focusing the physical TDV item on more serious aggression and adding sexual TDV, and the first nationally representative rate of sexual TDV from an ongoing survey. As a result, the field has new national prevalence estimates of TDV for high school students who experienced physical or sexual TDV or both. We further demonstrated that those who experience different forms of TDV are at risk for multiple other health-risk behaviors. Future work should examine in more detail the frequency of physical and sexual TDV and the effect that a higher frequency of TDV has on negative health outcomes.
Accepted for Publication: December 1, 2014.
Corresponding Author: Kevin J. Vagi, PhD, Division of Violence Prevention, Centers for Disease Control and Prevention, 4770 Buford Hwy NE, MS F-63, Atlanta, GA 30341 (firstname.lastname@example.org).
Published Online: March 2, 2015. doi:10.1001/jamapediatrics.2014.3577.
Author Contributions: Ms O’Malley Olsen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: O’Malley Olsen, Vivolo-Kantor.
Study supervision: Vagi.
Conflict of Interest Disclosures: None reported.
Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.
Additional Contributions: We thank the CDC’s Division of Violence Prevention and Division of Adolescent and School Health who helped with the wording of the new physical and sexual TDV questions. No compensation was received from a funding sponsor.