Do context and emotional reaction to physical dating violence interact to increase the likelihood of disclosure in 13 year-old British adolescents?

Objectives: The aim of the current study is to examine the prevalence and nature of adolescent dating violence (ADV), and to understand how perceptions about, and consequences of, ADV influence the likelihood of adolescents disclosing their experiences of dating violence. Design: The sample comprised 3,604 adolescents aged 13 years (51.6% female; 91% White) from the Avon Longitudinal Study of Parents and Children (ALSPAC). A cross sectional design was employed to examine the self-reported prevalence of dating violence and the association between its context, the emotional reactions it gave rise to and the act of disclosing its occurrence. Results: Within the sample, 10.5% reported using violence and 11.7% reported experiencing violence in romantic relationships, with males more likely to identify themselves as victims, and girls more likely to be perpetrators and perpetrator/victims. Girls however, were more likely to disclose their experiences. Males were more likely to be victimized due to anger and jealousy, and perpetration was explained as being part of a game or done in self defense. For females, victimization was experienced in the context of a game, and perpetration was initiated due to anger or in response to being annoyed by their partners. Associations were found for each gender in relation to the context of the violence and the resulting emotional response to the experience. Conclusions: These findings need to be considered in relation to educational strategies that may be required in order to ensure that adolescents can enjoy healthy dating relationships.


Introduction
During adolescence -typically defined as the age period between 10 and 19 years-a key developmental milestone is the formation of romantic relationships (Collins, 2003). Romantic relationships serve to facilitate the development of self-identity, and also provide a potential training ground for relationship attitudes and behaviours of relevance to adult marital-type relationships (Wekerle & Wolfe, 1999).

Page 19
Research indicates that the majority of adolescents have experienced some form of romantic relationship. For example, it has been reported that 50% of young adolescents (aged 11 or 12 years) reported having a boyfriend or girlfriend in the past three months (Carver, Joyner, & Udry, 2003;Laursen, Furman, & Mooney, 2006;Simon, Miller, Gorman-Smith, Orpinas, & Sullivan, 2010), and by age 16 years this figure increases to close to 90% in British studies (Barter, McCarry, Berridge, & Evans, 2009). Although many adolescents enjoy healthy positive relationships, some also experience unhealthy relationships, where verbal, psychological and physical aggression are common features of adolescent dating relationships (e.g., Banyard & Cross, 2008;Burman & Cartmel, 2005;Danielsson, Blom, Nilses, Heimer, & Högberg, 2009;Krahé & Berger, 2005;Reed, Silverman, Raj, Decker, & Miller, 2011). Of concern, this population is generally unwilling to seek help when experiencing relationship violence (Ashley & Foshee, 2005;Black & Weisz, 2003;Foshee et al., 1996). To date, research has not specifically examined adolescent dating violence (ADV) and help seeking behaviours in early adolescence, as research of this nature tends to also include older adolescents, e.g., ages 12-18 years (Burman & Cartmel, 2005) and ages 14-20 years (Reed et al., 2011). Therefore, the extent to which young adolescents (i.e. aged 12-14 years) experience these behaviours is not clear. In addition to this, while researchers have examined the extent of ADV, very few have studied the situational context of ADV, and those that have tend to find that context might be related to help-seeking behaviours (e.g., Bowen et al., 2013). Nevertheless, to date, this has not been explored within early adolescent relationships. Therefore, the aim of this paper is to address these gaps in current knowledge, and to build on existing findings, by assessing prevalence of ADV in a sample of 13 year olds. The current study will also examine contextual antecedent correlates and consequences of adolescents' involvement in dating violence, and explore how these might influence their help-seeking behaviours.
ADV comprises a vast array of different types of violent, aggressive and abusive behaviours. In academic terms, the body of research literature on ADV does not offer a uniform definition of such violence. Generally, the term ADV has been used to describe all the forms of violent behaviour that may occur in a dating relationship (Teten, Ball, Valle, Noonan, & Rosenbluth, 2009). This mainly encompasses three forms of violent behaviour: (i) emotional/psychological/verbal; (ii) physical; and (iii) sexual. Such behaviours are defined using various terminologies and in several ways that vary in their comprehensiveness (Glass et al., 2003). However, the commonly cited definition of ADV from the Centers for Disease, Control and Prevention (2013) reads that ADV is 'physical, sexual, or psychological/emotional violence within a dating relationship, as well as stalking. It can occur in person or electronically and may occur between a current or former dating partner.' Data accumulated during the last decade indicate that one-third to a half of adolescents have experienced either physical or sexual violence, within a dating relationship (Arriaga & Foshee, 2004;Foshee et al., 1996;Williams & Martinez, 1999). Most of the empirical evidence on ADV prevalence is derived from US studies (Barter et al., 2009). For example Arriaga and Foshee (2004) reported in their study of 12 -17 year old American adolescents, that 12% had perpetrated moderate ADV and 17% had experienced moderate ADV. When gender was examined, more females (17%) than males (6%) reported perpetrating ADV, and more males (23%) than females (11%) reported experiencing ADV. Females who had been involved in ADV were more likely (50%) than males (29%) to be identified as victim-perpetrators. Indeed, O'Keefe (1997) found that experiencing ADV was a strong predictor of using ADV for boys and girls aged between the ages of 14 and 20 years.

Page 20
ADV has also been examined across other cultures, although to a lesser extent.  carried out a comprehensive international review of the prevalence of ADV across Europe and North America, and found that rates of physical ADV ranged between 10 and 20% of the general population samples, with similar rates for boys and girls. Across studies however, victimisation was reported to be higher for girls than boys. Overall, the authors identified two general trends, despite methodological variations between studies: psychological ADV is more prevalent than physical ADV, and prevalence rates are similar for girls and boys in the majority of the reported studies across all forms of ADV, except for sexual violence, where there is typically twice the number of female than male victims .
It has been noted that, generally in the UK, little is known about adolescents' experiences of dating violence (Barter, 2009). However, Fox, Corr, Gadd andButler (2013) found in a survey of 1,143 British 13 to 14 year-olds (82% of whom had been in a dating relationship), that 44% of boys and 46% of girls reported having ever experienced a form of ADV, with 38% of the sample reporting experiencing emotional abuse, 17% reporting physical abuse, and 14% reporting sexual abuse. Sex differences in victimisation were only found for sexual abuse, with 17% of girls and 10% of boys identified as victims. For perpetration, 25% of boys and 24% of girls reported having used ADV. Twenty percent of the sample reported using emotional abuse, 7% physical abuse and 4% sexual abuse. No sex differences in perpetration rates were reported. It has only been since 2011 that European and British policy makers have extended formal definitions of domestic abuse to include 'those under 18' (Council of Europe, 2011, p.) and those 'from 16 years' (Home Office, 2012, p. 19). Although this is a move in the right direction, worryingly, it does not protect those younger who, as has been evidenced, are also experiencing ADV.
In relation to prevalence levels, it is also worth noting that it is common for ADV to be mutual (e.g., Barter et al., 2009;Cyr, McDuff, & Wright, 2006;Giordano, Soto, Manning, & Longmore, 2010). Giodano et al. (2010) found that of 234 adolescents who reported dating violence victimisation, perpetration or both, 49% reported symmetry. In relation to psychological violence these rates appear to be higher, with Cyr, McDuff, and Wright (2006) finding that of 126 females aged 13-17 years who reported experiencing offensive or degrading verbal behaviour, 90% also reported that this was reciprocal. Indeed, based on findings from the Youth Risk Behaviour Survey, Swahn et al. (2008) concluded that those who reported being perpetrators of ADV were 12 time more likely to also report being victims of ADV, than those who reported no dating violence at all. Consequently, the most prevalent profile of involvement in dating violence is that of mutuality, with up to 66% of those involved reporting being both perpetrator and victim (Chiodo et al., 2012;Lewis & Fremouw, 2001).
These data imply that young adolescents across different countries and cultural contexts can be involved in violence in romantic relationships. In the current paper however, it is within the cultural context of the UK that ADV will be examined. Within this cultural context (i.e., in England and Wales) it is important to understand violence and abuse in groups younger than policy currently acknowledges. However, due to the lack of research specifically in the UK, and to the range of ages reported across all previous studies, it is not clear the extent of experience with these behaviours in young adolescents aged 12 -14 years , or indeed whether the estimates provided are unduly influenced by the older age-groups. This is important to understand, given that evidence suggests adolescents in [Erica Bowen, Coventry  Page 21 this age group do endorse the use of violence in dating relationships in certain contexts .
Few studies have examined the situational context of ADV. Follingstad, Wright, Lloyd, and Sebastian (1991) found in their sample of 495 college students (mean age 20 years) that female perpetrators were more likely than male perpetrators to report using violence through anger, and to resolve emotional pain. Males were more likely to report using violence through jealousy, and in retaliation for being hit first. Female victims were more likely than male victims to report that their victimisation was a means of being controlled or in retaliation to them hitting their partners first. Males reported that their victimisation was due to their partners being angry. Female victims were also more likely than male victims to report negative emotional consequences after their experiences. In a sample of 900 high school students, O' Keefe and Treister (1998) found that female victims were more likely to believe that their victimisation arose from their partner's jealousy and anger. Male victims believed that their victimisation was also due to jealousy, or their partner getting back at them. Female victims were also most likely to identify fear and emotional hurt as consequences of their experiences, whereas males were likely to report amusement and anger as their main emotional responses. These two studies highlight that perceptions of the causes and consequences of dating violence are gendered, at least in samples aged between 14 and 20 years.
There is further evidence which suggests that context may also be linked to help-seeking. In a recent qualitative study of Northern European adolescent attitudes towards dating violence, Bowen et al. (2013) found that help-seeking is likely to be gendered, and also influenced by perceptions of the acceptability of the violence used. This, in turn, was a function of the perceived severity and emotional significance of the violence, part of which related to the context within which the violence occurred. For example, adolescents reported that violence in the context of a game or as a joke was unlikely to be severe, which as a consequence, made it more legitimate and less likely to be something you would tell other people about, or seek help for. In addition, violence by girls was not viewed as violence and therefore, again, was viewed as both legitimate and insignificant. This may explain why males are typically less likely than females to seek help if experiencing ADV (Jackson, Cram, & Seymour, 2000;Watson, Cascardi, Avery-Leaf, & O' Leary, 2001).
Aside from gender, few studies have attempted to identify other potential correlates of help-seeking behaviour. One notable exception is the study conducted by Black et al. (2008), in which 57 high school students completed a modified version of the Conflict Tactics Scale (CTS2: Straus, Hamby, Boney-McCoy, & Sugarman, 1996), and then responded to a range of questions concerning their response to the 'worst physical or violent behaviour' experienced. It was found that if someone had witnessed the violence, or the victim had assigned emotional significance to the experience, they were more likely to seek help. Consequently, drawing on the findings of Bowen et al. (2013), it is possible that the originating context of the violent episode determines to some extent the emotional significance placed on the experience, which in turn will influence an individual's appraisal of the severity of an incident which may lead to disclosure and/or help-seeking. Therefore, it might be expected that both context and emotional response interact to predict disclosure of the experience and/or use of violence.
The aim of the present study is to provide additional data concerning the nature and prevalence of dating violence, within a UK population of younger adolescents i.e., 13-year old children. A further

Page 22
aim is to explore whether perceptions and consequences of dating violence interact to influence the likelihood of adolescents disclosing their experiences. Due to the findings of Jackson et al. (2000) and Watson et al. (2001), it is expected that girls will be more likely than boys to report experiencing dating violence, both in terms of receiving and using such behaviours. Gender differences in the motivating context (e.g. jealousy, anger, game) are expected, and it is also expected that girls will be more likely than boys to report negative consequences (e.g. anger, sadness, fear) from dating violence. It is further expected that disclosure will be more prevalent in girls than boys. Drawing on the literature reviewed, it is expected that disclosure of violence experience will be a function of the interaction between the context and emotional consequences of the violence experienced.

Method Participants
The data upon which the present study is based has been taken from a cohort of 13 year old children born to mothers who were recruited into the Avon Longitudinal Study of Parents and Children (ALSPAC). ALSPAC recruited 14,541 pregnant women living in Avon, UK, with expected dates of delivery from 1st April 1991 to 31st December 1992, through pre-natal appointment. The cohort was developed to comprise a representative population sample of pregnant women in the UK, although comparisons with the 1991 Census data revealed that the mothers of infants in Avon were slightly more likely than those in the rest of Britain to live in owner occupied accommodation, to have a car available to the household, and to be less likely to have one or more persons per room, and be non-white. Out of the initial 14,541 pregnancies, all but 69 had known birth outcome. Of these 14,472 pregnancies, 195 were twin, three were triplet and one was a quadruplet pregnancy meaning that there are 14,676 fetuses in the initial ALSPAC sample. Note that of these 14,676 fetuses, 14,062 were live births and 13,988 were alive at 1 year.
The sample for the current study comprised those children who attended a data collection clinic in response to an invitation at the approximate age of 13.5 years, which ran between January 2005 and September 2006 (N = 10,580). This was the first age at which dating violence was assessed. Of those who attended, only those who reported ever having had a romantic relationship, regardless of length, were eligible to complete the dating violence survey. For the present study, only data provided by children who were aged 13 years at the time of data collection was examined. In total, the sample for the present study comprised 3,604 children, which represent 29% of the invited sample. Of the final sample, 1,838 (51%) were female.

Procedure
Ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee, and the Local Research Ethics Committees. The dating violence interview was conducted within an interview session held on university premises. Interviews were conducted by experienced staff who had been interviewing the children on an annual basis for the previous six years. Consequently, they were staff with which the children had developed a long-term relationship.The interview format was deemed to be most appropriate by the ethics committee due to the sensitive nature of the topic, as it was felt that, should a child become distressed, the clinic staff -with whom all children have a good rapport formed through attending such clinics since age eight -would be able to provide appropriate support. Informed consent was obtained from the children and their parents/guardians, and all children were informed of their right to refuse to answer and to withdraw their participation at any time. In addition, the child protection conditions under which notification to parents and/or external agencies would take place, were made clear to the children prior to the interview.

Measures
The present study examines the data obtained from one interview-administered measure of dating violence conducted when the children were, on average, 13.5 years of age. The dating violence interview consisted of a number of items taken from a revised version of the Conflict Tactics Scale (Straus, 1990) that was employed by Arriaga and Foshee (2004) in their study of 12 to 17 year olds. Due to ethical concerns regarding asking young adolescents about serious and sexual violence, the items used reflected less serious elements of dating violence.
Participants were asked by an adult interviewer whether they had ever intentionally used or experienced any of seven behaviours in the context of a dating or romantic relationship. Following the wording used by Arriaga and Foshee (2004), participants were asked 'How many times have you done the following behaviours to a dating partner or someone you have 'gone out' with?' The behaviours included: 'scratched', 'slapped', 'kicked', 'bent fingers', 'pushed/grabbed/shoved', 'thrown something', 'hit with their fist' or 'other' violence. In addition, participants were asked 'How many times has any person that you have been on a date with or 'gone out' with done the following things to you?' The same behaviours were asked about in relation to victimisation.
Participants responded either 'yes' or 'no' to each item as well as providing a quantitative estimate of the number of times they had used or experienced each behaviour. All responses were recorded by the interviewer on a response sheet. Skewed response distributions for each item led to the decision to classify a participant as a perpetrator they had to respond 'yes' to having used one of the behaviours listed. Similarly, for a participant to be classified as a victim they had to respond 'yes' to having experienced one of the behaviours listed.
Participants were further asked whether they had disclosed their experience (as perpetrator and victim) to anyone ('yes'/'no'), and to whom they had disclosed it. Participants were asked if they had told: 'teacher', 'parent', 'friend', 'relative', 'sibling'. To each item participants responded either 'yes' or 'no'. In addition, to measure the 'context' within which violence was being used, participants were asked which, of a range of motives, they believed led to their experience of violence (love, anger, jealousy, part of a game, intimidation, relationship break-up, friends do it, normal to do it, self-defense, other person annoyed them). Finally, to understand 'response' to ADV, participants were asked what was the dominant emotion they felt after the experience of violence (angry, anxious, depressed, better, guilty, unhappy/sad, shocked, not bothered). To all questions, participants answered 'yes' or 'no'.
Age was controlled in this study with all participants being aged 13 years. On the basis that there is no empirical evidence indicating that disclosure is linked to other socio-demographic characteristics, no other control variables were included in the analysis.

Sample Bias
In order to examine the possibility of bias within the sample, those children who provided data analysed in this paper (n = 3,604) were compared with the remainder of the original birth cohort (n = 7,944) on a variety of socio-economic indicators (including: young maternal age at time of first pregnancy, inadequate housing, low maternal education, financial difficulties and large family size during pregnancy. Data available on request). It was found that the sample that provided these data Page 24 analysed in this paper, differed systematically on these variables from the sample that did not, being characterised by less adversity than the group that did not provide data. Table 1 presents the prevalence of each form of violence, total prevalence and gender differences for both the perpetration and experience of dating violence.  Table 1 shows that girls were more likely than boys to report perpetrating all forms of violence, and that girls were more than three times more likely to be identified as perpetrators than were boys. With regard to the reported experience of violence in dating relationships, boys were more likely than girls to identify themselves as a victim, in particular with reference to having been scratched and slapped in relationships. Indeed the odds of reported victimisation were 68% greater for boys than girls.

Prevalence of Adolescent Dating Violence
A significant association between gender and role in violence was also found (χ 2 2 = 186.85, p < .001; V = .24), with girls more likely than boys to be identified as perpetrators (123, 7.3% girls; 19, 1.2% boys) and perpetrator/victims (120, 7.2% girls; 60, 3.8% boys). In contrast, boys were more likely than girls to be identified as victims (169, 13.4% boys, 33 2.0% girls). Consequently the first hypothesis, that there would be sex differences in the perpetration and experience of dating violence, is supported.  Table 2 shows that there were considerable variations in the frequencies of motives identified for perpetration and victimisation, and that in some cases these appeared to be gendered. For example, girls were nearly four times more likely than boys to cite that their use of violence occurred in the context of anger. However, boys were nearly twice as likely as girls to identify violence in the context of a game, and nearly ten times more likely than girls to cite their use of violence as self-defense. In contrast, for victimization, boys were nearly four times more likely than girls to cite that their victimization occurred in the context of a relationship break-up, and were more than four times more likely to cite a context of jealousy. Anger and annoying their partner was also more likely to be identified by boys than girls as a reason for their victimisation. Girls were more likely than boys to perceive that their victimization occurred in the context of a game.

Prevalence of Disclosure
In total, 37.9% of young people reported their use of violence, and 39.1% reported their experience of violence, to someone else. Females were significantly more likely than males to inform someone else ([Perpetration: Males 21.8%, Females 43.1%; χ 2 = 11.34, p < .001, OR = 2.72], [Victimisation: Males 32.7%, Females 48.9%; χ 2 = 9.36, p = .002, OR = 1.92]). Moreover, when the identity of the confidante was examined, disclosure of violence perpetration and instigation was most often to friends (36% and 35% respectively). As a consequence, hypothesis two, that there would be sex differences in the disclosure of involvement in dating violence, was supported.

Interaction Between Context, Emotional Response and Disclosure
A series of 3-way loglinear analyses were conducted between context of violence (anger, jealousy, game, victim annoyed perpetrator) and emotional response (anger, unhappy/sad, shocked, not bothered), which were the categories with sufficient cell counts suitable for such analyses. Table 3 shows that in the majority of cases, no higher-order interactions were significant, although most second-order interactions were significant. The exception to this rule was the interaction between Game context, Shocked emotional response and Disclosure. Follow up Chi Square analyses were conducted which revealed that in the Non-Disclosure group, there was a marginally non-significant association between Game context and Shocked emotional response (Shocked: 0.0%; Not-Shocked: 37.3%; χ 2 = 3.520, p = .044). In the Disclosure group there was a non-significant association between Game context and Shocked emotional response (Shocked: 27.3%; Not-Shocked: 22.1%; χ 2 = .159, p = .466). This indicates that if violence occurred in the context of a game, young people were not shocked by it, and as a consequence of not being shocked were less likely to disclose it.
Follow up Chi Square analyses were conducted for each of the 15 groups of variables for which a significant second-order interaction was identified. For the sake of brevity, only the significant results are reported here.

Discussion
The findings of the present study mirror the general pattern of those previously reported. Arriaga and Foshee (2004) reported that of their sample of 12 -17 year olds, 12% were perpetrators of moderate dating violence (based on the items used in the present study), which included 6% of males and 17% of females. In the present study, 10.5% were perpetrators including 5% of males and 14.5% of females. Arriaga and Foshee (2004) reported that 17% of the sample were victims of dating violence, which included 23% of males and 11% of females. In the present study, the overall victimisation rate was lower at 11.7%, which included 14.4% of males and 9.1% of females.
The present findings, however, stand in contrast to those reported by Barter et al. (2009), as the prevalence of violence experienced in the ALSPAC cohort was lower, and more males than females reported being a victim of physical violence than reported by Barter et al (2009). In line with the findings of Barter et al'sstudy , albeit at a lower level, more females than males reported perpetrating physical force. It is possible that age group and methodological variations account for these differences.

Page 28
However, in comparison with a study by Fox et al. (2013), who studied dating violence in a sample of 1,143 school children of comparable ages (13-14 years) to those in the current study, there were still contrasting results in relation to dating violence perpetration and victimisation prevalence. That is, perpetration prevalence rates were lower (7% vs 10.5%) but victimization rates were higher (17%vs 11.6%), and interestingly there were no gender differences in the reporting of either perpetration or victimisation. These findings may be attributed to the methodologies used, in particular that Fox et al. based their findings on two questions (experiences of (i) pushed, slapped or grabbed and (ii) punched, kicked choked or beaten up), whereas the current study asked about the experiences of seven specific behaviours in the context of dating violence.
The additional finding from the present study, that females are more likely to have both perpetrated and experienced physical violence by the age of 13 within a dating relationship, whereas males are most likely to be victims only, also replicates the findings of Arriaga and Foshee (2004). These data may also lend support to the notion that violence against females is not socially accepted, and that societal norms of chivalry and the protection of women Felson, 2000) prevail from as early as adolescence. As a consequence, the greater use of violence in self-defense reported by the males, and the higher levels of violence perpetration reported by females, may reflect normative beliefs that women should be protected. It is possible that such findings may also reflect response bias with males more likely to under-report their use of violence. If this is the case, then this would be more evidence for the normative belief that violence against women and girls is wrong (Felson, 2000). Indeed, recent research has found that adolescents view female violence to males as more acceptable and appropriate than male violence towards females Sears, Byers, Whelan, & Saint-Pierre, 2006). However, as Ashley and Foshee (2005) note, the systematic examination of response bias within adolescent dating violence is yet to be conducted. Consequently it is unclear the extent to which this is an issue in the present study.
There was some evidence to support the notion that the perceived context of violence use and experience was gendered, although the patterns are not completely consistent with prior research (O'Keefe & Treister, 1998). Males were more likely than females to report that their victimisation arose out of a context of anger or jealousy, due to the relationship breaking up or because they had annoyed their partner. Females were more likely than males to consider that their victimisation arose out of a context of a game. Females in the present study were more likely than males to believe that their use of violence arose out of anger, or was in response to their victim annoying them. Conversely, boys were more likely than girls to identify their use of violence as occurring as part of a game and as selfdefense, suggesting that they were responding to violent romantic partners. This again highlights the use of violence by female adolescents, and suggests that for some, it may pre-empt their own victimisation.
Consistent with the findings of previous research, the majority (62%) of those who had experienced or used violence in dating relationships had not disclosed their experiences to anyone. This figure reflects non-disclosure estimates in the international literature which range from 30% -79%, and as expected, girls were more likely than boys to have disclosed their experiences, and were most likely to do so to their friends rather than adults (e.g., Ashley & Foshee, 2005;Barter et al., 2009;Boldero & Fallon, 1995;Jackson et al., 2000;Molidor & Tolman, 1998).

Page 29
The data examined herein, identified a number of associations between the context in which violence occurred and the resulting emotional response. Specifically, a context of anger was associated with victims feeling unhappy/sad, and they were less likely to feel 'not bothered' as a consequence of their experience. In addition, jealousy was associated with an increase in the likelihood feeling unhappy/sad and shocked as a consequence, and also with an increased likelihood of reporting feeling 'not bothered'. This suggests that for some young people who experience violence in the context of jealousy, such jealousy may act as an aggravating context, whereas for others, it may act as a mitigating situational context. What is unclear from the data examined, however, are the precise appraisals of young people's understanding of the situation which gave rise to jealousy and consequently violence, and how this links to the apparently inconsequential emotional impact (feeling not bothered) of their experience.
It has been reported in the general help-seeking literature that perceiving a problem as severe, increases the likelihood of disclosure (Fallon & Bowles, 1999;Timlin-Scalera, Ponterotto, Blumberg, & Jackson, 2003), although this has previously not been found to be the case with dating violence (Tishby et al., 2001). However, if we accept emotional response to victimisation as an indicator of perceived problem severity (based on the previous discussion), the findings in the present study provide some evidence to support this idea in the context of dating violence disclosure. Feeling angry, unhappy and shocked was associated with increasing the likelihood of disclosure. In contrast, feeling 'not bothered' was associated with a reduced likelihood of disclosing victimisation. Of interest, only one higher order interaction was found between context, emotional response and disclosure. If violence occurred in the context of a game, and young people were not shocked by its occurrence, and were therefore less likely to disclose.. This may be attributed to cognitive distortions (Beck, 1963) that eliminate shock, and therefore create the thought process that there is, in fact, nothing really to disclose. Unfortunately, the small numbers of participants in each group precluded sex-based analyses of these data in the present study.
Whilst this study provides new information regarding the nature of dating violence and predictors of disclosure in a young adolescent cohort, the results need to be interpreted within the context of the study limitations. The greatest limitation concerns the use of dichotomised variables which may reduce the measured variability and limit the statistical power (Altman & Royston, 2006). However, where continuous data had been collected (e.g. number of times the behaviour was experienced) the distributions were so severely skewed that more robust analyses were not deemed appropriate. An observation to be gained from this, however, is that whilst physical dating violence was reported, it was typically a low-frequency occurrence which was reported by a minority of the sample. The estimated prevalence was lower than that reported elsewhere, which may reflect the methodological variations across studies alluded to previously. A further consideration, perhaps, is that the data were collected in 2006, at a time in the UK before formal responses to domestic violence were willing to accept its occurrence in such a young age-group. Consequently, it is possible that the broader social context influenced responding and that the data examined here represent something of a cohort effect.
Although participants were asked if they told anyone about their use or experience of violence, they were not asked whether by doing so they had formally asked for help from the people to which they had disclosed. Therefore it is unclear the extent to which the findings translate to the active helpseeking literature, despite the obvious parallels identified, and the consistency of findings in relation to this literature. Moreover, the study focused only on physical and not non-physical forms of ADV. A more specific measure which examined the nature of the disclosure, and the intent behind the decision to disclose across the range of ADV behaviours, would help to clarify this issue.
A further limitation is the use of forced choice categories from which context and consequences were specified. Giving choices as a response means that certain other relevant answers were likely omitted. Another issue that arises from using forced choices is seen, for example, in the item 'violence took place as part of a game', as this is likely to include a considerable amount of measurement error as the nature of the game is not specified, and as such is left open to interpretation. However, what is counterintuitive is that the phrasing of the items was such that the violence had been intentionally used towards someone they were 'going out with'. These data, therefore, suggest that there may be some contexts in which such behaviours are part of a game that is played in the context of romantic relationships. This is notionally consistent with the idea of 'horseplay' which has been identified as a significant feature in romantic relationships (Gergen, 1990). However, the presence of horseplay in adolescent relationships was also positively associated with the presence of non-playful aggression, suggesting that horseplay may be a risk marker for dating violence and aggression, rather than a nonsignificant mitigator of these behaviours. This possibility, and the potential impact on the meaning assigned to violence, requires additional empirical exploration.

Implications for Policy and Practice
Despite these limitations, the findings of the present study have important implications for current policy and practice development. Based on the prevalence rates found in the current study and in other research, intervention and prevention is perhaps best introduced to all children within their curriculum at school, at an early age. Policy needs to acknowledge ADV in populations younger than those it currently specifies, and this can be used to inform prevention. Adolescents are reporting that they are dating at an earlier age, and within these relationships there is evidence of a range of abusive behaviours. In the current study this was age 13, and this would therefore suggest policy needs to include this age (and younger), and that intervention is required prior to this. Changes in policy can raise awareness, as can the introduction of intervention programmes. By raising awareness and education, this will enable adolescents to learn how to develop healthy relationships, and understand that any form of violence and aggression, regardless of the context and emotion attached to it, is wrong.
The gender-equal presentation of ADV found in this study, needs to alert practitioners to enquire about the behaviours of both partners within any ADV experienced, and to make a careful examination of the extent of all the different behaviours used by both parties. In order to reduce physical and psychological violence with dating partners, this needs to be addressed for both males and females, as both victims and perpetrators, or those mutually involved. In doing this, educational strategies need to be informed by the reality of the reciprocal nature of some violence in adolescents, as well as the potential use of violence by girls which might pre-empt their own victimisation. Indeed, recent research has highlighted the fact that the content of current primary interventions is perceived by young people as 'sexist' (Fox et al., 2013), due to representing ADV as a gendered issue. It also appears that as girls use violence in contexts charged with anger, emotion management needs to feature within educational strategies focusing on dating violence. The present study has shown that boys are more likely to report using violence in the context of a game, and that girls acknowledge that their victimisation occurs in this context too. As such, this indicates that an analysis of this game context is required in order to determine what meaning is assigned to it by both parties, and how it might serve to minimise the seriousness of the violence by both parties involved. Moreover, these data also suggest that interventions may need to include an examination of 'horseplay' as a precursor to less playful aggression in relationships. The data presented here and previously by Bowen et al. (2013), support the game context as minimising perceptions of violence severity, but it is unclear how this context is used in order to justify violence, and therefore whether such games are genuinely cause for little concern. Consequently, interventions should also address the cognitive strategies used to permit the use of violence in early dating relationships, i.e. they need to focus on challenging attitudes and cognitive strategies that condone and mitigate the seriousness of ADV.
The findings of the present study indicate that prevention and education strategies need to also focus on helping peers to help their friends safely. Moreover, should teachers be placed with the responsibility of actively seeking to help victims, training regarding the nature and dynamics of violence in adolescent relationships is required, building on the findings of the present and previous studies. Such training, however, should not be based on adult domestic violence but should be based specifically on adolescent research findings, in order to make it resonate with the experiences of young people.
In summary, therefore the implications for practice are:  Policy needs to acknowledge ADV in populations younger than those it currently specifies;  Interventions need to reflect the gender-equal presentation of ADV in order to be relevant to young people;  Interventions need to focus on challenging the attitudes and cognitive strategies that condone and mitigate the seriousness of ADV, and promote disclosure and help-seeking.