Taking Stock of Behavioral Measures of Adolescent Dating Violence

Abstract

The past 2 decades have witnessed an increase in dating violence awareness and research. As the field evolves, it is critical to examine the definition and measurement of adolescent dating violence. This article summarizes the behavioral measures of adolescent dating violence used in the field. Based on a review of the literature and federally funded studies, we identified 48 different measures. The most commonly used measures were the Conflict Tactics Scale–2, the Safe Dates Scale, and the Conflict in Adolescent Dating Relationship Inventory, which all examine aspects of psychological, physical, and sexual violence. Researchers also adapted or created their own measures. This article concludes with a discussion of developments for consideration as the field moves forward.

Adolescent dating violence is a significant public health problem linked to a variety of negative short- and long-term consequences, ranging from impaired functioning to chronic disease and death (Ackard, Eisenberg, & Neumark-Sztainer, 2007; Campbell, 2002; Exner-Cortens, Eckenrode, & Rothman, 2013). These negative sequelae, paired with an emerging focus on primary prevention, have shifted adolescent dating violence to the forefront of public health injury control efforts. The past two decades have witnessed an increase in adolescent dating violence awareness, policy change, funding, and research. For example, 25 states have introduced or passed adolescent dating violence prevention policies (National Conference of State Legislatures, 2013). Further, the number of recent programs with evidence of effectiveness in preventing some forms of dating violence has increased, reflecting a significant step toward fostering healthy and safe relationships among our youth and in preventing partner violence in subsequent adult relationships. However, much work remains, as the surge in awareness, resources, and research has yet to translate into widespread decreases in the prevalence of dating violence. Thus, as the field continues to evolve, it is critical to examine the foundation of this work: definition and measurement of adolescent dating violence.

Although it is generally well-accepted that adolescent dating violence is defined as any psychological, physical, or sexual violence or stalking perpetrated by a current or former dating partner either in person or electronically (Centers for Disease Control & Prevention [CDC], 2014a), the measurement of adolescent dating violence does not always parallel this conceptualization. For the field to determine the magnitude of the problem (and to have confidence in the validity and reliability of estimates), conduct ongoing surveillance, and evaluate interventions, it is imperative that all measurements of this important public health problem match this definition. For example, although the definition of adolescent dating violence includes psychological and sexual violence and stalking, some of our most cited estimates of the prevalence of adolescent dating violence are based only on reports of physical violence, sometimes measured by just one item (e.g., CDC, 2011). Additionally, researchers in the prevention field overwhelmingly agree that the prevention of dating violence perpetration is essential (Whitaker, Murphy, Eckhardt, Hodges, & Cowart, 2013). However, many measures of adolescent dating violence used in evaluation studies either focus on victimization (e.g., Whitaker et al., 2013) or knowledge, attitudes, or intentions rather than behavior (e.g., Ting, 2009). Similarly, although the CDC (2014a) definition of dating violence is gender-neutral, acknowledging that males and females can report perpetration or victimization, some studies evaluate program effects based only on male dating violence perpetration and female victimization (e.g., Miller et al., 2012). These discrepancies in definition and measurement lead to challenges with the development and implementation of prevention strategies and miscalculation of the magnitude of the problem and of the effects of prevention efforts.

THIS STUDY

The goals of this review are twofold. First, we provide a summary of the behavioral measures of adolescent dating violence that are currently being used in research and evaluation. In so doing, we examine the most frequently used measures, common adaptations to these measures, and gaps between the definition of dating violence and forms of dating violence measured. Second, we identify important next steps and considerations for measurement of adolescent dating violence.

Behavioral Measures of Adolescent Dating Violence

To achieve our first goal, we conducted a review of behavioral measures of adolescent dating violence. To ensure that we captured the most current measures in use by researchers, we employed a novel review method that accounted for research findings not yet published as well as studies that were not primarily designed to assess dating violence. First, via the Federal Interagency Workgroup on Teen Dating Violence, we queried current and former federally funded researchers who were conducting studies related to adolescent violence, including projects discussed as part of an adolescent dating violence longitudinal data meeting hosted by the National Institute of Justice (NIJ) (Espelage, n.d.). Then, we supplemented this scan of researchers with a review of the literature, using keyword searches in a variety of databases (Academic Search Premier, PsychINFO, ProQuest Criminal Justice Periodicals, Criminal Justice Abstracts, and SocIndex). When multiple publications on adolescent dating violence stemmed from the same research project, we only counted the dating violence measure once. Of the measures received and identified, we included items or scales tapping any type of adolescent dating violence. We focused on studies that used a population drawn from the United States or Canada, and, with the exception of college-only samples,¹ we included all behavioral measures administered to youth between the ages of 10 and 24 to cover all periods of adolescence (Gutgesell & Payne, 2004; Vagi et al., 2013). Therefore, this review was intended to be comprehensive and reflective of the field, but not exhaustive.

Through these methods, we identified 48 different behavioral measures used in the 130 studies that were reviewed.² Eighteen measures were developed specifically for the purposes of the study in which they were used, and the majority of these applied narrow definitions that captured just one or two types of adolescent dating violence (i.e., physical, psychological, sexual) and often only measured victimization (e.g., Michigan Study of Adolescent Life Transitions, Zweig, Barber, & Eccles, 1997; Tween Relationship Study, Glauber, 2008; Teen Assessment Project, Rodgers & Small, 1999). In other cases, the measures did not specify the type of violence (e.g., Ackard & Neumark-Sztainer, 2002) or did not distinguish between victimization or perpetration behaviors (e.g., Understanding Risk and Protective Factors in Sexual Minority Youth, Elze, 2002). Although these measures were not generally used beyond the study for which they were created, they were often employed in conjunction with more established measures (e.g., Stop the Violence Project, McDonnell, 2012).

Of the studies reviewed (see Table 1), 54% of the measures (n = 26) used to examine adolescent dating violence were preexisting and established; we were unable to determine if the remaining four measures were original or preexisting. Further, some of the established measures used to assess adolescent dating violence were developed for adults, including the Conflict Tactics Scale (CTS; Straus, 1979); the Revised Conflict Tactics Scale (CTS–2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996); the Sexual Experiences Survey (SES; Koss & Gidycz, 1985; Koss & Oros, 1982); the Psychological Maltreatment Inventory (PMI; Kasian & Painter, 1992; Tolman, 1989); the Abusive Behavior Inventory (ABI; Shepard & Campbell, 1992); the Interpersonal Control Scale (ICS; Stets, 1991); and the Domestic Conflict Index (DCI; Margolin, Burman, John, & O’Brien, 1990). This raises concerns about the applicability and developmental appropriateness of using these measures with adolescent populations.

TABLE 1.

Measurement Comparison Relative to Conceptualization of Dating Violence

Measure	Developer(s)	Type of Violence Measureda				Mode
		Physical	Psychological	Sexual	Stalking	Electronic	In-Person
Revised Conflict Tactics Scale	Straus et al. (1996)	B	B	B	—	—	Yes
Safe Dates Psychological and Physical Abuse Scale	Foshee et al. (1996)	B	B	B	—	—	Yes
Conflict Tactics Scale	Straus (1979)	B	B	—	—	—	Yes
Conflict in Adolescent Relationships Inventory	Wolfe et al. (2001)	B	B	B	—	—	Yes
Youth Risk Behavior Survey	CDC,	V	—	b	—	—	Yes
Sexual Experiences Survey	Koss & Oros (1982); Koss & Gidycz (1985)	—	—	B	—	—	Yes
Psychological Maltreatment Inventory	Tolman (1989); Kasian & Painter (1992)	—	B	—	—	—	Yes
Juvenile Victimization Questionnairec	Hamby et al. (2005)	V	V	V	—	—	Yes
Date and Family Violence Abuse Scale	Symons et al. (1994)	V	V	V	—	—	Yes
Adjustment with Partner Questionnaire	Kessler (1990)	B	B	—	—	—	Yes
Abusive Behavior Inventory	Shepard & Campbell (1992)	V	V	—	—	—	Yes
Sexual Harassment in Schools Survey	AAUW Educational Foundation (1993)	—	—	B	—	—	Yes
Dyadic Adjustment Scale	Spanier (1976)	—	B	—	—	—	Yes
The Psychological Abuse Index	Molidor (1995)	—	B	—	—	—	Yes
Partner Interaction Questionnaire	Capaldi (1991)	V	V	—	—	—	Yes
Interpersonal Control Scale	Stets (1991)	—	B	—	—	—	Yes
Dating Violence Perpetration Acts Scale	Rothman et al. (2011)	P	P	P	P	Yes	Yes
Lifetime Trauma and Victimization History Survey	Widom et al. (2005)	V	V	V	V	—	Yes
Domestic Conflict Inventory	Margolin et al. (1990)	B	B	—	—	—	Yes
Behavioral Affect Rating Scale	Conger (1989)	V	V	—	—	—	Yes
WHO Multicountry Survey	Garci-Moreno et al. (2006)	V	V	V	—	—	Yes
Michigan Violence Against Women Survey	DCH (1997)	B	B	B	—	—	Yes
General Violence Questionnaire	Holtzworth-Munroe et al. (2000)	V	V	—	—	—	Yes
The Sexual Aggression Scale	Pirog-Good & Stets (1989)	—	—	V	—	—	Yes

^aV measures victimization only; P measures perpetration only; B measures both victimization and perpetration.

^bSince 2013, the YRBS also includes one item measuring sexual violence.

^cOnly one question specifically asks about boyfriend/girlfriend physical dating violence. All other types of violence are general with follow-up questions to determine victim–offender relationship. Information based on survey for those ages 17 and younger.

Only 35% (n = 9) of the 26 preexisting measures that examined adolescent dating violence were developed specifically for use with adolescents. Some of these dating violence measures, including the Youth Risk Behavior Survey (YRBS; CDC, 2009), the Juvenile Victimization Questionnaire (JVQ; Finkelhor, Turner, Ormrod, Hamby, & Kracke, 2009), the Lifetime Trauma and Victimization History Youth Version (LTVH; Widom, Dutton, Czaja, & DuMont, 2005), and the Sexual Harassment in Schools Survey (American Association of University Women Educational Foundation [AAUW], 1993), were part of larger scales or studies designed to capture information about a range of adolescent experiences. The four measures mentioned here reflect narrow definitions of dating violence, albeit to varying degrees. The LTVH is the most comprehensive in capturing all forms of adolescent dating violence (stalking and electronic abuse are separate modules that can be added to the base survey), but it only includes victimization. The YRBS has previously examined only physical violence victimization, but a new question assessing sexual violence in a dating context was added in 2013 (CDC, 2014b). The JVQ just measures victimization, although it includes physical, psychological, and sexual violence.³ The Sexual Harassment in Schools Survey is the sole measure to include both perpetration and victimization, but it only measures sexual violence.

The other five scales that have been developed specifically to measure dating violence among adolescents are the Psychological Abuse Index (PAI; Molidor, 1995), the Safe Dates Psychological and Physical Dating Abuse Scales (Safe Dates; Foshee et al., 1996), the Conflict in Adolescent Dating Relationship Inventory (CADRI; Wolfe et al., 2001), the Date and Family Violence and Abuse Scale (DFVAS; Symons, Groer, Kepler-Youngblood, & Slater, 1994), and the Dating Violence Perpetration Acts Scale (DVPAS; Rothman et al., 2011, 2012). The DVPAS captures all forms of adolescent dating violence (physical, psychological, sexual, stalking, and electronic), and the majority of the other measures (Safe Dates, CADRI, DFVAS) assess physical, psychological, and sexual dating violence. Only the CADRI and the Safe Dates, however, measure both victimization and perpetration behaviors. See Table 2 for a description of these measures.

TABLE 2.

Adolescent Dating Violence Measures

Measure	Developers	Description	Example Items
Safe Dates Psychological Abuse Scale (Safe Dates)	Foshee et al. (1996)	14 items measuring 4 categories of psychological abuse: threatening behavior, monitoring, personal insults, and emotional manipulation/fear.	Scratched me. Damaged something that belonged to me.
Safe Dates Physical Violence Scale		18 items measuring physical abuse in dating relationships. 2 of the items measure sexual violence.	Physically twisted arm. Forced partner to have sex.
Conflict in Adolescent Dating Relationships Inventory (CADRI)	Wolfe et al. (2001)	35-item measure of adolescent dating violence with 5 subscales: physical violence, verbal abuse, threatening behavior, sexual violence, relational aggression and positive conflict resolution skills.	I kicked, hit, or punched him. I spread rumors about her.
The Psychological Abuse Index (PAI)	Molidor (1995)	15 items of psychological abuse with subscales: isolation, monopolization, economic abuse, degradation, rigid sex-role expectations, psychological destabilization, and emotional withholding.	My dating partner told me my feelings were crazy or irrational. My dating partner withheld affection from me.
Dating Violence Perpetration Acts Scale (DVPAS)	Rothman (1997)	24 items measuring physical, psychological, sexual violence; stalking; and electronic abuse. Based on Safe Dates physical violence scale. Added items to measure psychological and electronic abuse.	Broke into their email or cell phone. Made them feel afraid.
Date and Family Violence Abuse Scale (DFVAS)	Symons et al. (1994)	30 items measuring dating and family violence. 23 measures assess dating violence with the following subscales: physical violence, sexual violence, verbal abuse.	How often has a date raped you? How often has a date humiliated you in public?

Most frequently used measures

Even though we identified a large number of measures used to assess adolescent dating violence, the frequency of use varied considerably. To find the most frequently used adolescent dating violence measures, we looked at the total number of times each measure was used across the different studies. We found the most frequently used measure—the CTS–2—was used in 24% (n = 31) of the studies, followed by the Safe Dates (22%, n = 28) and the CADRI (15%, n = 20).⁴ All three of these measures examine psychological, physical, and sexual victimization and perpetration as well as assess a wide range of behaviors. For example, the CADRI measures relational abuse and positive conflict resolution skills, and the CTS–2 includes assessment of negotiation and injuries associated with dating violence. Additionally, the Safe Dates and the CTS–2 use gender-neutral language, whereas the CADRI language is gender-specific.

Although the CTS–2, Safe Dates, and the CADRI assess the same forms of adolescent dating violence, there is variability in how each type of violence is operationalized. For example, the Safe Dates scales measure sexual violence with only two items (one item on forced sex; one item on other forced sexual acts). The CTS–2, on the other hand, includes a greater number of sexual violence items, but it only assesses more severe forms of violence, namely forced or coerced sexual intercourse and oral sex. Of these three most commonly used measures, only the CADRI includes less severe forms of sexual violence (e.g., unwanted kissing) that might be more closely aligned with the sexual experiences of younger adolescents.

Measurement modifications

When we took a closer look at the specific studies that used the CTS–2, the Safe Dates, or the CADRI measures, it became clear that researchers often modified or adapted the measures for use in the field. Common adaptations included shortening the length (n = 43), changing response categories (n = 19), and modifying the language (n = 10) to be gender-neutral and more developmentally or culturally appropriate. Shortening of scale length was often accomplished in one of two ways: using select items from one or more subscales (e.g., Middle School Success Project, Leve, 2003; Healthy Teens Longitudinal Study, Orpinas, Horne, Song, Reeves, & Hsieh, 2013) or combining multiple items into fewer items (e.g., Coaching Boys to Men, Miller et al., 2012; SafERteens, Cunningham et al., 2010). Researchers noted that measures were often shortened because of time constraints (a consistent challenge of doing research in school settings [e.g., Healthy Passages, Windle et al., 2004]), because adolescent dating violence represented just one construct assessed as part of a larger survey (e.g., Add Health, Killeya-Jones, 2013), or because the dropped items were not age-appropriate (e.g., Juvenile Justice Girls Study, Leve, Chamberlain, & Reid, 2005). Interestingly, it was noted that when the CTS–2 was shortened, both the negotiation and injury scales were omitted more than 70% of the time; the sexual coercion scale was also frequently omitted (58%).

Change to the response categories was the second most common adaption. This modification often occurred when the measures were supplemented with measures from other scales that had different response categories. In most cases, this adaption resulted in less variation in the response categories than what was provided for in the original scale (e.g., Project D.A.T.E., Reppucci et al., 2013). The most drastic changes occurred when response categories were altered to be dichotomous (e.g., first three waves of Dating it Safe: A Longitudinal Study on Teen Dating Violence, Temple, 2012; A MultiLevel, Cohort-Sequential Study of Rural Adolescent Dating Violence Victimization and Perpetration, McDonnell, 2012).

Additionally, our scan revealed three primary types of language adaptations. First, and only applicable to the CADRI, researchers replaced gender-specific language with gender-neutral language (e.g., Perpetration of Partner Violence among Adolescents from Violent Homes, Jouriles, Platt, & McDonald, 2009; Project D.A.T.E., Reppucci et al., 2013); this was the most common change made to the CADRI. Researchers further modified language to be more culturally appropriate (e.g., Family-Based Dating Abuse Prevention for Latino Teens, V. Foshee, personal communication, July 19, 2012) or age appropriate (e.g., Adolescent Dating Violence Victimization and Psychological Well-Being, Callahan, Tolman, & Saunders, 2003) for the target participants.

Supplemental measures

We also examined whether measures from different scales were used in combination. This was an important observation in that the types of supplemental measures selected might reflect a missing dimension in the commonly used adolescent dating violence scales. Of the studies that utilized the Safe Dates, the CTS–2, or the CADRI measures, 43% (n = 32) also used adolescent dating violence measures that were drawn from other scales. Measures of sexual violence were most often used to supplement these measures (53% of studies that included supplemented measures), followed closely by measures of electronic (50%), physical (44%), and psychological adolescent dating violence (44%). Six percent of the studies included supplemental stalking measures. Researchers also supplemented with measures of conflict resolution (16%), relational abuse (9%), and injury (9%).

Sixty-three percent (n = 20) of the studies that supplemented the CTS–2, the Safe Dates, and the CADRI measures with additional items increased the number of different types of adolescent dating violence measured. This suggests that researchers were aware of the disconnection between the operationalization and conceptualization of adolescent dating violence. However, only three (15%) studies incorporated all forms of adolescent dating violence occurring both in person and electronically.

Approximately two thirds of the supplemental items were drawn from established scales. Although many of these established scales were developed for an adult population (e.g., ABI, SES, CTS–2, PMI, SAS, DCH), researchers also used supplemental items from scales developed specifically for adolescents (e.g., Safe Dates, CADRI, YRBS, FDB, JVQ, SHS). The remaining 25% of the supplemental items were project-developed. Although these measures accounted for only one quarter of the supplemental measures, they comprised 50% of the electronic dating violence measures and 31% of the sexual violence supplemental measures. In contrast, less than 15% of the physical and psychological supplemental variables were project-developed.

Because a large proportion of electronic and sexual violence measures were project-developed, we took a closer look at the content of these items. It was noted that there are few established measures of electronic violence, so the large number of supplemental measures that were project-developed was not surprising. Despite this fact, however, most of the measures tapped similar behaviors. Examples of the types of victimization and perpetration behaviors assessed include showing or posting private or embarrassing pictures, being verbally abusive on networking sites, keeping tabs through various electronic mediums (e.g., cell phone, email, etc.), and spreading rumors electronically.

For the supplemental sexual violence measures that were project-developed, many of the behaviors captured were measures that examined a broad range of sexual dating violence behaviors. For example, Ybarra (personal communication, July 26, 2012) developed a sexual aggression measure that included language such as unwanted kissing and unwanted touching, which was developmentally appropriate for her sample (ages 10–15). Other researchers (e.g., Miller et al., 2012) incorporated developmentally appropriate behaviors for older adolescents, such as convinced to have sex. Similarly, Zweig, Sayer, Crockett, and Vicary (2002) developed sexual violence measures that captured both forced and coerced sexual experiences and included the circumstances surrounding victimization (e.g., victim was so drunk or stoned that he or she was unaware of what was going on; victim felt obligated).

DISCUSSION

This review aimed to summarize the behavioral measures of adolescent dating violence that are currently being used in research and evaluation, including common adaptations to these measures, and gaps between the definition of dating violence and forms of dating violence measured. Of note, the variety of measures used to assess dating violence and variation among them highlights measurement issues that are similar to those faced in many areas of social science inquiry, particularly other forms of violence and abuse (Follingstad & Ryan, 2013). However, it is imperative that adolescent dating violence research progresses to a point where there is a better match between the conceptualization agreed on by the field and the behavioral measures used by researchers. Otherwise, it will be difficult to have confidence in the validity and reliability of estimates to determine the magnitude of the problem, to conduct ongoing surveillance, and to accurately evaluate interventions. Next, we summarize key highlights of our scan and address a few considerations for future research aiming to capture dating violence in adolescent relationships.

Through our multimodal search, we identified 48 behavioral measures that have been used across studies to measure adolescent dating violence. Of these, researchers most frequently used the original or adapted versions of the Safe Dates Scale, the CTS–2, or the CADRI. These three measures all assess some aspects of psychological, physical, and sexual victimization and perpetration. Although these measures capture the same forms of adolescent dating violence, our ability to compare studies that use these measures is hampered for three reasons. First, there is great variation in how these measures operationalize each form of adolescent dating violence. Even regarding physical violence, which tends to suffer from less conceptual disagreement than other forms of violence, there are differences in how the three commonly used scales define some concepts. For example, “threw you around,” “threw you onto furniture,” and getting “slammed into a wall” are all examples of items from the three scales that measure similar behaviors but might result in somewhat different responses. Second, these measures are often adapted for use in the field. Common adaptations include shortening the length, changing response categories, and changing language to be gender-neutral and more developmentally or culturally appropriate. Third, researchers within our search frequently supplemented the most commonly used measures with additional items from other scales. Generally, the additions that captured physical and psychological violence were taken from established scales, whereas items added to measure electronic and sexual violence were frequently project-developed.

There is also a lack of consensus about whether electronic abuse is a unique type of violence requiring a separate measurement scale, or whether technology just provides another means for perpetrating psychological and sexual violence and stalking. Current qualitative work being conducted as part of a concept mapping study of adolescent relationship characteristics suggests that adolescents view the part of their relationships that occurs in digital space as highly integrated into all other aspects of their relationships (Goldman, Mulford, & Blachmen-Demner, in press), indicating that it is more of another means for perpetrating relationship violence rather than a separate form of violence. That said, the field is just beginning to explore this question empirically, and much more qualitative and quantitative research is needed in this area.

Another unresolved issue has to do with the potential impact of phrasing of the survey questions. Most established scales (including the three identified herein) are designed to gather information specifically about a dating partner. However, other measures (e.g., National Survey of Children’s Exposure to Violence, Finkelhor et al., 2009) first ask whether a particular action or behavior occurred and then ask who the perpetrator was. Although some work (e.g., Hamby & Turner, 2012) has begun to consider the impact of these various phrasings on responses and prevalence rates, more research is needed to better understand the implications of such subtle wording changes on our measurement of dating violence. Perhaps in-depth cognitive testing of these various options would be warranted.

Our review highlighted two notable gaps with regard to behavioral measurement of adolescent dating violence: the measurement of sexual violence and stalking. Our scan indicated that when sexual violence items were added to scales (often in the form of project-developed measures or items), typically only victimization and not perpetration was assessed. The rationale for excluding perpetration was often not made clear by researchers. Because much of the adolescent dating violence research occurs in school settings, it is likely that part of the problem is that often, in our experience, school administrators are uncomfortable allowing researchers to ask questions about sexual behaviors, which is an issue particularly common with younger samples (e.g., middle school youth). This critical omission, however, hampers our understanding of both the prevalence and etiology of sexual perpetration in dating relationships, as well as understanding program effects. This suggests a need to either move some of the research on dating violence out of school settings or to raise schools’ awareness about the prevalence and consequences of sexual dating violence among youth. Indeed, the few studies that have measured sexual dating violence among middle school youth suggest that early adolescence is a critical developmental period for addressing this behavior (e.g., Taylor, Stein, Mumford, & Woods, 2013).

As previously noted, stalking measures were also frequently excluded from adolescent dating violence studies. The inclusion of stalking behaviors in the definition of dating violence is fairly recent, and this might have contributed to its exclusion in most measures. For an adolescent audience that frequently interacts in digital space, there might be some difficulty disentangling what constitutes stalking from other forms of electronic abuse. The standard definitions of stalking that are used in legal and research contexts (e.g., The National Intimate Partner and Sexual Violence Survey, Black et al., 2011) define stalking as a pattern of behaviors that are unwanted and cause fear. Therefore, as stalking items are added to instruments in the future, using simple frequency counts of individual behaviors might be inadequate. Considerations of how these acts are interpreted by the victim are critical to accurately measuring stalking behavior.

Of note, when we compared the measures identified in this review to the CDC compendium of assessment tools for partner violence (Thompson, Basile, Hertz, & Sitterle, 2006), we identified a variety of established scales that were not being used in the research we reviewed. This is likely due to the fact that some scales in the compendium might not be developmentally appropriate for adolescents (e.g., Measure of Wife Abuse; Rodenburg & Fantuzzo, 1993); however, some scales are potentially applicable (e.g., Courtship Persistence Inventory; Sinclair & Frieze, 2000) and, as such, the compendium could serve as a resource for researchers who wish to measure sexual violence and stalking using established scales. Established scales in the compendium could also serve as alternatives to project-developed scales and should be considered for inclusion in dating violence studies by researchers.

With few exceptions (e.g., Safe Dates, CADRI), the majority of the 48 scales identified in our scan were developed for adult rather than adolescent populations. This raises concerns about the developmental appropriateness of measures that are often used. Critical to the measurement of all adolescent behaviors is the consideration of the unique cognitive, biological, social, and emotional developmental changes that occur throughout adolescence. For example, measures must be interpretable in light of the cognitive abilities and limitations that characterize this developmental time period (e.g., level of concrete rather than abstract thinking, world experience) and must elicit responses that are relevant in the rapidly changing physical, socioemotional, and cognitive context of adolescence (Pfeifer & Blakemore, 2012).

An important consideration for future research on the measurement of adolescent dating violence is the extent to which changing established scales affects the psychometric properties of those scales. To our knowledge, no research has been done to determine how adaptations might affect the validity and reliability of dating violence measures. For example, one can imagine that adapting a scale to make the items gender neutral, while seemingly harmless and desirable, changes the nature of what is being measured. Likewise, altering or omitting items or changing the response categories to make the scale more developmentally appropriate for younger adolescents could compromise the validity of the measure. This does not mean that adaptations should be avoided, but rather when doing so, investigators should consider the impact of the change on the psychometric properties of the measure.

Limitations

This review was limited by several factors. First, our multimodal search was intended to identify the scales that are most commonly used to assess adolescent dating violence. It was not intended to be a systematic or exhaustive review of every measure or every study that used each measure; as such, the numbers and percentages included in the results are intended to be illustrative. As measurement evolves over time, we also placed an emphasis on measures that have been used most recently, which is why we first reached out to current or recent grantees who were receiving federal funding to study dating violence. Therefore, the information in the tables is best described as study exemplars rather than a comprehensive list.

Second, we attempted to include studies that were designed to examine dating violence as well as studies that were created to measure other youth risk behaviors but also assessed dating violence. This tactic might be the reason for the large proportion of project-specific dating violence measures. Studies designed to measure other youth behaviors likely used a limited number of items to capture dating violence, whereas studies specifically designed to measure dating violence most likely used more comprehensive and validated measures. Third, because we queried researchers about measures they were using in their studies, this review captures measures that might or might not also be reported in study publications, given that researchers often assess more behaviors than they ultimately translate into publications. Therefore, our results might differ from other reviews that included only published work.

CONCLUSIONS

As we move forward into the next phase of adolescent dating violence measurement, two trends in the field bear mentioning. First, the conceptualization of dating violence has started to expand as our understanding of how coercive control can manifest in relationships. For example, Miller and colleagues (2011) began to explore the critical role of pregnancy coercion and reproductive control as a unique form of sexual violence. This review did not include these measures, but this important work points to the complexity of the dynamics involved in sexual violence and promising directions to guide future efforts.

Second, there has been a recent shift in public health from a focus on disease prevention to one of health promotion, which is mirrored by a change in adolescent dating violence prevention whereby programs emphasize promotion of healthy relationships in addition to preventing violence (Robert Wood Johnson Foundation, 2013; Tharp et al., 2011). A focus on healthy relationships extends beyond preventing violence in relationships and includes other aspects of adolescent sexual and reproductive health, such as preventing pregnancy, sexually transmitted infections, and HIV (Tharp et al., 2013). National dating violence prevention initiatives, such as the CDC’s Dating Matters^™: Strategies to Promote Healthy Teen Relationships and Robert Wood Johnson Foundation’s Start Strong: Building Healthy Teen Relationships, are examples of programs focusing on middle school youth that aim to prevent dating violence by building skills for healthy relationships. Although measurement of adolescent dating violence was a primary focus of this review, it is important to note that for measurement of dating violence to remain in sync with the next generation of prevention strategies, we must also take into account measurement of healthy relationships. Unlike dating violence, which has generally agreed-on elements (psychological, physical, sexual, and stalking), there is no consensus for how to define or measure healthy relationships. This gap allows for significant variation across studies. Although this is to be expected in emerging fields of research, future research on adolescent relationships must assess the extent to which variation across studies reflects measurement inconsistencies or complexities in adolescent dating relationships.