Introduction

The formation of romantic relationships during early and middle adolescence is a normative developmental task (van de Bongardt et al. 2015), and the quality of these relationships is critical for adolescents’ psychological adjustment and well-being (Viejo et al. 2015). However, research has shown substantial prevalence rates ranging from 30 to 51% for any form of teen dating violence experienced (Exner-Cortnes 2014; Vagi et al. 2015; Ybarra et al. 2016). This substantial variation in prevalence depends on how violence is defined (Winstok 2007), differences in methodological approach, such as sampling design and assessment procedures, as well as characteristics of the populations studied (Vagi et al. 2015). Moreover, recent research has identified different trajectories with distinctive types of victimization. While some adolescents report only experiences of psychological abuse, others experience multiple forms of dating violence victimization (especially emotional abuse in conjunction with physical violence; Sabina et al. 2016).

Although research and prevention programs on teen dating violence have increased over the last decades, some vulnerability factors have received only peripheral attention. For example, while both attachment styles and conflict resolution strategies have been widely associated with an increased vulnerability to victimization in college-age couples (Bonache et al. 2016a; Cornelius et al. 2010; Yarkovsky and Fritz 2014), only a few studies have examined these factors in relation to violence among adolescent partners (Burk and Seiffge-Krenke 2015; Messinger et al. 2012). Moreover, findings on adults’ experience of intimate partner violence cannot be generalized to adolescents because research has noted different patterns in both developmental periods (Johnson et al. 2015). For instance, the formation of romantic relationships emerges as a new developmental task in adolescence (van de Bongardt et al. 2015), and partners become attachment figures (Exner-Cortnes 2014). Although dating relationships provide an opportunity for building constructive conflict resolution strategies (Simon and Furman 2010), there are individual differences in interpersonal skills (Reese-Weber 2000). Moreover, adolescents are involved in risk behaviors, such as alcohol or drug abuse, and violence more frequently than adults (Mahalik et al. 2013), especially when they are insecurely attached (Letcher and Slesnick 2014). These risk behaviors and the lack of communication skills tend to make teens more vulnerable to dating victimization than adults, which seems to be reflected in the peak of violence observed during adolescence (Brooks-Russell et al. 2013; Orpinas et al. 2013).

Because of these characteristics of adolescence as a developmental period, findings based on college-age samples cannot be generalized to adolescents, and empirical studies are required to address the association between attachment styles, conflict resolution strategies, and dating victimization in this age group. Furthermore, identifying vulnerabilities in victims may be useful for developing evidence-based prevention programs, which does not imply victim blaming (Hamby and Grych 2016). Additionally, research highlights the need to incorporate skill-building components to increase the effectiveness of teen dating violence prevention programs (De la Rue et al. 2016). Based on this evidence, the present study examined the role of romantic attachment styles and conflict resolution strategies (self-reported and attributed to the partner) as predictors of psychological abuse and physical violence victimization in a sample of adolescents.

Attachment Styles and Conflict Resolution Strategies

According to attachment theory, experiences in early close relationships with caregivers lead to the formation of internal working models about the self and the other (Mikulincer and Shaver 2012). These working models guide cognition, affect, and behavior through different attachment styles, which emerge from two underlying dimensions: anxiety about abandonment and avoidance of intimacy (Mikulincer and Shaver 2012). While securely attached individuals (low in attachment anxiety and attachment avoidance) tend to show independence and comfort with intimacy, insecurely attached people tend to show a strong need for intimacy and fear of being rejected by their attachment figures (high anxious attachment) or emotional detachment and self-sufficiency (high avoidant attachment). Moreover, attachment styles have been related to emotion regulation strategies. In particular, anxiously attached individuals experience high levels of negative affect and hypersensitive proximity-seeking behaviors, whereas those with high levels of avoidant attachment are characterized by employing strategies to deactivate their negative emotions. By contrast, securely attached individuals tend to be able to deal properly with both positive and negative affect (Maas et al. 2011).

As mentioned above, attachment theory has also been useful to explain some aspects of dating relationship dynamics, such as conflict resolution strategies. However, research on this topic is limited in adolescents even though they report higher levels of conflicts with their romantic partners than in their other close relationships (Furman and Shomaker 2008). In adult samples, the use of destructive conflict resolution strategies has been related to insecure attachment, with different underlying processes for anxious and avoidant attachment (Fowler and Dillow 2011; Mikulincer and Shaver 2012). Specifically, individuals scoring high on anxious attachment tend to report conflict engagement, based on the desire to secure attention, care, and support. At the same time, they tend to report withdrawal strategies based on the fear of being rejected by their partners. Individuals with higher levels of avoidant attachment tend to report evasive communication, avoidance of disagreements, and withdrawal from conflicts (Bonache et al. 2016a; Fowler and Dillow 2011). However, when arguments escalate, avoidantly attached individuals may resort to conflict engagement as a way of distancing themselves from their partner (Mikulincer and Shaver 2012). By contrast, secure attachment has been linked to positive social skills and constructive conflict resolution strategies in adolescents (Exner-Cortnes 2014; Tan et al. 2016). Although adolescents tend to employ positive conflict resolution styles more often than destructive strategies (Simon et al. 2008), some of them also use withdrawal and conflict engagement strategies to manage arguments with their partners (Shulman et al. 2006), which may be explained by limited emotion regulation abilities (Zimmermann and Iwanski 2014). Poor affect regulation, which has been related to insecure attachment styles, was shown to play a key role in interpersonal conflicts. For instance, adolescents who report more emotion dysregulation also report higher levels of conflicts with their parents (Moed et al. 2015). These findings suggest that insecure attachment styles may be related to destructive conflict resolution strategies in adolescents in a similar way as in adults.

Communication patterns tend to be developed within romantic relationships, and conflict resolution strategies used by both partners are interrelated during disputes. Some of these dyadic communication patterns have been consistently linked to violence in adult couples, especially the demand/withdrawal pattern (Fournier et al. 2011). Researchers have also shown that attachment styles affect perceptions of the partner’s conflict resolution strategies. For instance, stressful situations, such as conflicts with the partner, tend to activate both negative emotions (Mikulincer and Shaver 2012) and schemas related to insecure attachment (Furman and Shomaker 2008), making a negative interpretation of the partner’s behaviors more likely (Beck et al. 2014). Therefore, anxiously attached individuals perceive a higher frequency of both withdrawal and conflict engagement behaviors in their partners compared to avoidantly attached individuals (Collins et al. 2006), and they also tend to report escalation (conflict engagement by one partner predicting conflict engagement by the other partner) during their arguments (Bonache et al. 2016a; Exner-Cortnes 2014).

Regarding avoidant attachment, the findings are less consistent. While some studies have shown that avoidantly attached adults perceive their partners as less responsive and supportive than do non-avoidant individuals (Beck et al. 2014; Segal and Fraley 2015), others did not find an association between avoidant attachment and the perception of partner support (Karantzas et al. 2014) or between avoidant attachment and the attribution of destructive conflict strategies to the partner (Bonache et al. 2016a). Based on attachment theory, it may be assumed that avoidantly attached individuals see their partner’s behavior as conflict engagement because they perceive it as undermining their desire for independence.

Conflict Resolution Strategies and Teen Dating Violence Victimization

Effective communication is a challenge for many teens, as it requires emotion regulation and interpersonal skills that may still be insufficiently developed, making different forms of victimization more likely than in later developmental stages (Espelage et al. 2015). Although some adolescents manage conflicts with their partner by using compromise strategies, others tend to ignore them to preserve the relationship (Shulman et al. 2006) or may be prone to using conflict engagement and withdrawal strategies in trying to handle arguments (Furman and Shomaker 2008). In addition, evidence has shown that conflict resolution strategies of one partner depend on the strategy used by the other partner (Paradis et al. 2017). For example, adolescents may display a downplaying pattern, in which both partners tend to minimize and avoid confrontational arguments, or a conflictive pattern that is characterized by active confrontation and reciprocal escalation of conflicts by both partners (Fernet et al. 2016; Shulman et al. 2006). The demand/withdrawal communication pattern has rarely been explored in teen dating relationships, even though its use in adult relationship conflicts has been demonstrated (Fournier et al. 2011; Siffert and Schwarz 2011).

Although negative communication patterns may produce adverse relational outcomes, such as dating violence (Schrodt et al. 2014), they have been much less studied in adolescents than in adults. In addition, conflict resolution strategies have received more attention in relation to teen dating violence perpetration than to victimization (Burk and Seiffge-Krenke 2015; Fernet et al. 2016), and some of those studies refer to negative styles of resolving conflict based on a composite score that encompasses engagement and withdrawal strategies, both by the self and seen in the partner (Paradis et al. 2017; Russell et al. 2014). The few studies that have analyzed them separately found different findings for conflict engagement and withdrawal. Female adolescents’ physical victimization was predicted by self-reported conflict engagement and conflict engagement attributed to the partner (Messinger et al. 2012). Regarding withdrawal from conflicts, self-reported withdrawal was found to be unrelated to victimization in college students (Bonache et al. 2016a), whereas other studies have found a significant association between both factors in college females (Katz and Myhr 2008) and in adolescents (Bonache et al. 2016b). No association was found between withdrawal strategies attributed to the partner and victimization (Messinger et al. 2012), suggesting that conflicts in adolescent couples do not lead to physical or psychological victimization if the partner shows, or is perceived to show, withdrawal behaviors. This finding is consistent with the literature on adult couples (Bonache et al. 2016a; Katz et al. 2008).

Indirect Paths from Attachment Styles to Teen Dating Violence Victimization via Conflict Resolution Strategies

Studies demonstrating an association between attachment styles and victimization by a romantic partner have mostly focused on adult couples (Bonache et al. 2016a; Yarkovsky and Fritz 2014). Nevertheless, there is some evidence that more anxiously and avoidantly attached adolescents are more likely to experience abuse in dating relationships (Capaldi et al. 2012). Likewise, longitudinal research has shown that insecure attachment styles predicted both teen dating violence victimization and perpetration over a 4-year period (Miga et al. 2010).

Few studies have included attachment styles and conflict resolution patterns in the same analysis to explain teen dating violence victimization. For example, Pepler (2012) suggested that insecurely attached adolescents, who are prone to using destructive conflict resolution strategies, tend to choose partners with similar attachment styles. This may promote a negative communication pattern, increasing the vulnerability to victimization in romantic relationships. In line with this reasoning, it has been found that insecure attachment and higher rates of conflict predict reciprocal aggressive behaviors in teen dating relationships (Burk and Seiffge-Krenke 2015).

In a similar vein, research on adult couples has demonstrated that vulnerability to intimate partner violence is high in partners with an anxious attachment style (Péloquin et al. 2011), especially when they report destructive conflict strategies (Bond and Bond 2005). Thus, an indirect path from anxious attachment to victimization has been shown via self-reported conflict engagement strategies and conflict engagement perceived in the partner (Bonache et al. 2016a). In accordance with the demand/withdrawal pattern, two further indirect effects may be expected. As noted above, individuals high on avoidant or on anxious attachment may show more withdrawal strategies, at the same time they may attribute more conflict engagement to their partner. This dyadic communication pattern may be related to higher vulnerability to victimization among teen partners. However, more research is necessary to clarify the role of insecure attachment on patterns of conflict resolution and teen dating violence victimization.

Gender Differences

Although there is evidence that both females and males may be victimized by dating partners, findings on the role of gender in teen dating violence victimization are mixed (Brooks-Russell et al. 2013; East and Hokoda 2015; Paradis et al. 2017), and only few studies have examined possible gender differences in attachment styles, conflict resolution strategies, and their relationship with teen dating violence victimization. According to the literature on romantic attachment, boys show higher attachment avoidance and lower attachment anxiety compared to girls (Furman and Simon 2006). Anxious attachment has been linked to increased dating victimization among females (Grych and Kinsfogel 2010), but little comparative evidence is available for males (Sandberg et al. 2016). Moreover, past research on conflict resolution strategies in adolescents has found inconsistent findings with regard to gender differences. Some studies found no differences between females and males (Furman and Shomaker 2008; Paradis et al. 2017), and others have shown that the use of specific conflict resolution strategies depends on who generated the conflict topic regardless of gender (Laurent et al. 2008). Therefore, gendered links between attachment styles, conflict resolution strategies, and teen dating violence victimization need to be further examined.

The Current Study

The aim of the current study was to address the aforementioned gaps in the literature by examining the associations between attachment styles, conflict resolution strategies, and victimization by a dating partner among adolescents. More specifically, we examined whether insecurely attached adolescents would be more likely to use destructive strategies and perceive these strategies in their partner when dealing with relationship conflict and whether these dysfunctional conflict resolution strategies would be linked to an increased vulnerability to physical and psychological victimization in a dating relationship. A structural equation model, displayed in Figs. 1 and 2, was tested to explore these associations in a sample of adolescents.

Fig. 1
figure 1

Paths from adult attachment style to psychological abuse (multigroup model by gender, paths freely estimated; age as covariate). N = 1289 (652 males; 637 females). Chi2(df = 6) = 19.45; p = .004, CFI = .994; SRMR = .014, RMSEA = .059 (C.I. .031 −.089). First coefficients refer to males, second coefficients refer to females. Boxed coefficients differ significantly. *p < .05; **p < .01; ***p < .001

Full size image
Fig. 2
figure 2

Paths from adult attachment style to physical violence victimization (multigroup model by gender, paths freely estimated; age as covariate). N = 1289 (652 males; 637 females). Chi2(df = 6) = 19.45; p = .004, CFI = .994; SRMR = .014, RMSEA = .059 (C.I. .031 −.089). First coefficients refer to males, second coefficients refer to females. Boxed coefficients in bold differ significantly. *p < .05; **p < .01; ***p < .001

Full size image

According to attachment theory, we expected that insecure attachment styles (characterized by higher scores on the anxious or avoidant attachment dimensions) would be positively related to both psychological abuse and physical violence experienced in teen dating relationships (Hypothesis 1). We further proposed that insecure attachment styles would be positively related to self-reported conflict engagement (Hypothesis 2a) and withdrawal (Hypothesis 2b). Additionally, research has noted that attachment styles relate to perceptions of the partner’s conflict resolution strategies (Beck et al. 2014; Mikulincer and Shaver 2012). Building on this line of research, we hypothesized that the higher the avoidant attachment, the more conflict engagement adolescents would attribute to their partner, because this strategy tends to be perceived as undermining their desire for independence (Hypothesis 3a). The higher the anxious attachment, the more conflict engagement (Hypothesis 3b) and withdrawal strategies (Hypothesis 3c) would be perceived in their partner.

Based on previous research on communication patterns describing escalation of conflicts, demand/withdrawal patterns, and downplaying patterns when couples handle disagreements, we expected that both self-reported conflict engagement and withdrawal strategies would be positively related to conflict engagement and withdrawal attributed to the partner (Hypothesis 4). Moreover, destructive conflict resolution strategies may increase the vulnerability to victimization (Bonache et al. 2016b; Messinger et al. 2012). Specifically, while both self-reported conflict engagement and withdrawal were predicted to be positively related to the two forms of victimization (Hypothesis 5a), we expected only partner-attributed conflict engagement to be positively associated with victimization (Hypothesis 5b).

In accordance with the theoretical and empirical evidence mentioned above, a focus of our analysis was on the indirect paths from attachment styles to victimization via destructive conflict resolution strategies. Thus, we predicted that anxious attachment would be indirectly related to both forms of teen dating violence victimization through conflict engagement behaviors, shown by the self and perceived in the partner (Hypothesis 6). Additionally, anxious and avoidant attachment were predicted to be indirectly linked to psychological and physical victimization through self-reported withdrawal, which in turn should predict conflict engagement attributed to the partner (Hypothesis 7). These indirect paths were assumed to reflect the escalation of conflict and demand/withdraw patterns found in adult couples.

Finally, since little research has tested gender differences in relation to attachment styles, conflict resolution strategies, and teen dating violence victimization, further analysis are needed to shed light on gendered patterns. Hence, this study explored commonalities and differences between male and female adolescents in the associations of these factors.

Method

Participants and Procedure

Participants were recruited from eight public high schools located in urban, rural, and tourist areas within a Spanish region (The Canary Islands), which ensured that they represented a variety of socioeconomically strata. All students aged from 13 to 18 years enrolled in these schools (M = 15.41, SD = 1.11) completed the questionnaires (N = 1917). Only those students who had had at least one opposite-sex romantic partner were included in the study. The final sample included 1298 adolescents (638 females, 660 males) who had an opposite-sex romantic partner at the time of the study (37.4%), or had been in at least one romantic relationship in the past (62.6%). All participants were born in Spain and of European ethnic origin.

The study was approved by the Institutional Review Board of the first author’s university. In addition, permission was obtained from the heads of the participating high schools, participants’ parents, and from the participants themselves. Participants were told that participation was voluntary and unpaid (only 0.3% refused to participate). Confidentiality and anonymity of their responses was guaranteed. Data were collected by trained research staff who administered paper-and-pencil questionnaires during school hours.

Instruments

Attachment styles

A Spanish adaptation by Fernández-Fuertes et al. (2011) of the Experiences in Close Relationships−Revised Scale (ECR-R; Fraley et al. 2000) was used to measure attachment styles. This 18-item scale comprises nine items to assess anxiety about abandonment (e.g., I often wish that my partner’s feelings for me were as strong as my feelings for him or her) and nine items to measure avoidance of intimacy (e.g., I get uncomfortable when a romantic partner wants to get very close). The response scale ranged from 1 (strongly disagree) to 7 (strongly agree). The internal consistency, measured by Cronbach’s alpha, was high with .87 for anxiety and .85 for avoidance.

Conflict resolution strategies

Destructive conflict resolution strategies were measured with the Spanish adaptation by Bonache et al. (2016b) of the Conflict Resolution Styles Inventory (CRSI; Kurdek 1994). This version is a 13-item self-report scale designed to assess patterns of behavior during couples’ attempts to resolve disagreements. For this study, two of the three conflict resolution styles that have proven to discriminate between victims and non-victims of dating partners’ aggression were selected. Specifically, Conflict Engagement, which includes criticizing, attacking, and losing self-control (e.g., Launching personal attacks), and Withdrawal, which includes becoming silent, refusing to discuss the topic, and avoiding the problem (e.g., Staying silent for long periods of time). Adolescents were asked to rate the extent (1 = never; 5 = always) to which they (CRSI-Self) and their partners (CRSI-Partner) generally used these behaviors in conflict situations. Cronbach’s alphas for CRSI-Self and CRSI-Partner, respectively were .73 and .75 for conflict engagement, and .70 and .70 for withdrawal.

Teen dating violence victimization

Psychological abuse was assessed through a subscale developed by Safe Dates-Psychological Abuse Victimization (Foshee et al. 1998). It consists of 14 items (e.g., said things to hurt my feelings on purpose, told me I could not talk to someone of the opposite sex). In addition, three items from a shortened version of the Conflict Tactics Scale (CTS; Straus and Douglas 2004) were used to measure physical violence victimization (pushing, hitting, and causing injury). In both scales, responses were made on a scale from 0 (never) to 3 (very often). Cronbach’s alphas were .73 and .75 for psychological abuse and physical violence victimization, respectively.

Results

Preliminary Analyses

Means and standard deviations for all study variables are presented in Table 1. A series of one-way analyses of variance was conducted to explore gender differences, adopting an alpha level of .05/8 = .006 to correct for multiple testing. Only three significant gender differences were found: Boys scored significantly higher than girls on avoidant attachment, F (1, 1189) = 15.54, p < .001, and on physical violence victimization, F (1, 1185) = 12.47, p < .001, and lower on self-reported withdrawal, F (1, 822) = 12.63, p < .001.

Table 1 Bivariate correlations and means between teen dating violence victimization, perception of conflict resolution styles of self and partner, and attachment styles
Full size table

Zero-order correlations among all variables are also shown in Table 1, separately for males and females. The two forms of victimization were positively correlated with all factors in both gender groups. Self-reported and partner-attributed withdrawal and conflict engagement strategies were highly correlated, suggesting a destructive dyadic communication pattern. Avoidant and anxious attachment styles were significantly, but moderately correlated with each other (r = .12, p < .01 among males and r = .26, p < .001 among females), supporting their conceptual distinctiveness. Both styles also correlated significantly with all conflict resolution strategies among boys, and with all but one (self-reported conflict engagement) among girls. Few correlations with age were found. Age correlated negatively with attachment anxiety and avoidance among females. Among males, age was negatively correlated with avoidant attachment and positively correlated with perceived partner withdrawal. To account for these correlations, age was included as a covariate in the path models.

Hypothesis Testing

The hypotheses were tested by structural equation modeling using the Mplus 7.1 software (Muthén and Muthén 2012). The predicted pathways for the two forms of teen dating violence victimization were tested in a single model. Missing data were handled using a robust Full Information Maximum Likelihood (FIML) estimator implemented in Mplus (Muthén and Muthén 2012). To test the proposed indirect paths, we inspected the bootstrapped confidence intervals with 10,000 replications, using the ML estimator in Mplus. Indirect paths were considered significant at p < .05 if the 95% confidence intervals did not include zero and at p < .01 if the 99% confidence intervals did not contain zero.

We first specified a multigroup model by gender in which all path coefficients were constrained to be equal for males and females. This constrained model showed an acceptable fit with the data, χ 2 (df = 39) = 135.25, p < .001, CFI = .954; TLI = .916, SRMR = .078; RMSEA = .062, 95% CI = (.051; .073). In the next step, we compared this model against an unconstrained model in which all paths were freely estimated for each gender group, χ 2 (df = 6) = 19.45, p < .004, CFI = .994, TLI = .924, SRMR = .014, RMSEA = .059, 95% CI = (.031; .089). A significant chi-square difference test, diff χ 2 (df = 33) = 115.80, p < .001, indicated that the unconstrained model provided a significantly better fit with the data than the constrained model. Therefore, the unconstrained multigroup model was adopted as the final model, and individual coefficients were compared between the gender groups using the DIFF test option in Mplus. The model is presented in Fig. 1 for psychological abuse as the outcome variable and in Fig. 2 for physical violence victimization as the outcome variable. Splitting the results in this way is done for clarity of presentation. The path models were estimated including both outcomes in a single model. The association between the two forms of dating violence victimization was .68 for males and .34 for females, both ps < .001, with the gender difference being significant at p < .01.

The results largely supported our hypotheses about direct and indirect paths between the factors, although some gender differences deserve attention. Consistent with Hypothesis 1, we found that anxious and avoidant attachment styles were significantly related to psychological abuse in both gender groups. For physical violence victimization, the path from anxious attachment was significant for males, but not for females, whereas the path from avoidant attachment was significant only for females, lending only qualified support to our prediction.

The links proposed between attachment and self-reported conflict resolution strategies were also confirmed with some exceptions in the two gender groups. Specifically, in partial support of Hypothesis 2a, anxious attachment predicted self-reported conflict engagement in males and females, but the path from avoidant attachment to conflict engagement was significant for males only. Consistent with Hypothesis 2b, anxious and avoidant attachment predicted self-reported withdrawal as a conflict resolution strategy among both males and females.

Our predictions concerning the links between insecure attachment and the attribution of destructive conflict resolution to the partner were partially confirmed. Against our predictions in Hypotheses 3a and 3b, the paths from anxious and avoid attachment to partner’s perceived conflict engagement strategies were non-significant among both males and females. However, in line with our prediction in Hypothesis 3c, higher anxious attachment was associated with higher withdrawal behaviors perceived in the partner in both gender groups.

In accordance with the communication patterns proposed in Hypothesis 4, self-reported conflict strategies were related to conflict strategies attributed to the partner, although the results varied by gender. In both gender groups, self-reported conflict engagement and withdrawal were significantly related to the respective strategies perceived in the partner. In addition, self-reported conflict engagement predicted withdrawal perceived in the partner for females, but not for males, whereas self-reported withdrawal predicted conflict engagement perceived in the partner for males, but not for females.

Contrary to our prediction in Hypothesis 5a, self-reported withdrawal was a significant negative predictor of both forms of victimization among girls; for boys, the path was non-significant. Likewise, the path from self-reported conflict engagement to psychological abuse was significant only for girls, while the path to physical victimization was non-significant in both gender groups. In line with Hypothesis 5b, the perceived use of conflict engagement by the partner was a significant predictor of both forms of teen dating violence victimization in males and females.

The DIFF test indicated some significant gender differences on individual paths in our model. As noted above, the association between psychological abuse and physical aggression victimization was closer for males than for females. In addition, physical violence victimization was more closely linked to anxious attachment in males than in females; in fact, the path for females was non-significant. Perceived conflict engagement by the partner was more strongly linked to physical violence victimization among males than among females, whereas the association between anxious and avoidant attachment was stronger for females than for males.

All hypothesized indirect paths from insecure attachment styles to teen dating violence victimization were found to be significant in males, as shown in Table 2. Consistent with Hypothesis 6, anxious attachment was indirectly linked to experiences of psychological abuse and physical violence through self-reported conflict engagement and perceived conflict engagement by the partner. Additionally, significant indirect paths were found from both anxious and avoid attachment via self-reported withdrawal and conflict engagement attributed to the partner to psychological abuse and physical violence victimization, as predicted in Hypothesis 7. For females, a different picture emerged: Anxious attachment was indirectly linked to both forms of victimization via the escalation pattern of conflict engagement by the self, which in turn predicted conflict engagement perceived in the partner, consistent with Hypothesis 6. The routes from anxious and avoidant attachment via self-reported withdrawal predicting perceived conflict engagement by the partner were not significant in the female sample, failing to support Hypothesis 7.

Table 2 Proposed indirect paths from attachment styles to dating violence victimization for the models in Figs. 1 and 2 (bootstrapped confidence intervals)
Full size table

Significant indirect paths were found in both gender groups from self-reported conflict engagement via conflict engagement perceived in the partner to physical violence victimization and psychological abuse, for males: .30 (99% C.I. .112; .169) for physical violence victimization and .35 (99% C.I. .203; .534) for psychological abuse victimization, for females: .17 (95% C.I. .037; .504) for physical violence victimization and .38 (99% C.I. .152; .599) for psychological abuse victimization.

Discussion

On the basis of attachment theory, the present study developed a model describing the relationships between romantic attachment, conflict resolution strategies employed by adolescents and attributed to their partner, and teen dating violence victimization in the form of psychological abuse and physical violence. Some previous studies have analyzed the role of attachment styles or conflict resolution skills (Burk and Seiffge-Krenke 2015; Capaldi et al. 2012; Fernet et al. 2016) to explain dating violence in teen romantic relationships, although without integrating those factors in the same design from the victim’s perspective. Only one previous study has studied the combination of these factors, testing a similar model in college student couples (Bonache et al. 2016a). Although the present findings for adolescents largely confirm those found in a sample of college students, there are also differences, which suggest that the findings on intimate partner violence among adults cannot be generalized to adolescents (Johnson et al. 2015; Mahalik et al. 2013). The direct paths found in our adolescent sample were comparable with findings published in other parts of Europe, Canada, and the United States (Exner-Cortnes 2014), which provides support for the potential generalizability of our results in the age group of adolescents. Additionally, the current study extends prior knowledge by examining gender differences. The analysis identified some differences between males and females that suggest a gendered approach to increase the effectiveness of intervention programs.

Research on attachment has shown strong links between anxious attachment and different forms of victimization in adult samples, especially among women (Yarkovsky and Fritz 2014). By contrast, avoidant attachment has been linked to victimization to a lesser extent (Bonache et al. 2016a). In the current study, the higher adolescents scored on both insecure attachment styles, the more psychological abuse victimization they reported. For physical violence victimization, a gender difference emerged in that attachment anxiety was a significant predictor in boys and attachment avoidance was a significant predictor in girls. In line with attachment theory, these results reflect that avoidantly attached girls may demand autonomy and independence from their partners, while anxiously attached boys may claim attention and care. These behaviors are contrary to traditional gender roles, which are learned in the family and reinforced by other socialization agents (Montañés et al. 2012; van de Bongardt et al. 2015). Adolescents have been shown to be highly accepting of sexist attitudes (Ferragut et al. 2016) and may therefore react negatively to behavior by their partner that is incongruent with gender role expectations. In this sense, physical aggression may be more likely when partners behave differently than expected according to gender roles. As pointed out by England (2015), the expectations partners have of each other because of their respective social positions are constraints that make the social become personal by influencing relationship dynamics. Our finding that for females self-reported conflict engagement predicted withdrawal perceived in the male partner, whereas for males self-reported withdrawal predicted conflict engagement perceived in the female partner is in line with the gender stereotype that males are more likely to avoid communication about relationship conflicts, which may trigger pressure from their female partners.

Based on previous findings in adults (Bonache et al. 2016a; Fowler and Dillow 2011), we expected that anxious and avoidant attachment styles would be related to destructive conflict resolution strategies as reported for the self and perceived in the partner. The results largely confirmed our hypotheses, although three exceptions were found. First, the association between avoidant attachment and self-reported conflict engagement strategies was only found in boys. Evidence suggests that avoidantly attached individuals may use conflict engagement strategies when arguments escalate to maintain independence and relational distance from their partners (Mikulincer and Shaver 2012). Males tend to score higher on avoidant attachment than do females (Furman and Simon 2006), which has also been confirmed in the current study. Some research has found that hostile dominance mediated this association in men (Lawson and Brossart 2013). Therefore, future studies should test if hostile dominance explains the gendered pathways found in our study. Second, both anxious and avoidant attachment styles were unrelated to perceived conflict engagement by the partner in both males and females. This lack of association is consistent with the finding that some anxiously attached individuals are involved in romantic relationships with secure partners, who usually employ positive conflict resolution strategies during arguments (Sierau and Herzberg 2013). To address this possibility, future research should assess both partners’ attachment styles.

The significant associations found in our study between self-reported strategies and strategies attributed to the partners during conflicts are consistent with destructive interaction patterns. In line with previous research, our results suggest that the escalation of conflict pattern is displayed in teen romantic relationships as well. Additionally, the downplaying pattern, consisting of both partners avoiding confrontation and retreating from the conflict situation (Fernet et al. 2016; Shulman et al. 2006), was also supported by the significant link between self-reported withdrawal strategies and withdrawal perceived in the partner.

The predicted relationship between conflict engagement attributed to the partner and psychological abuse and physical violence victimization was found in both gender groups. For physical violence victimization, it was stronger in boys than in girls. The assumption of a direct path from self-reported conflict engagement and both types of victimization was only partially supported. Females, but not males, who reported more conflict engagement behaviors experienced higher levels of psychological abuse, while self-reported conflict engagement was not directly related to physical violence victimization in either gender group. However, these findings are qualified by the significant indirect paths in both gender groups from self-reported conflict engagement via conflict engagement perceived in the partner to physical abuse and psychological violence victimization.

Contrary to the expected direction in the paths, those female adolescents who employed more withdrawal strategies reported less psychological abuse and physical violence victimization. This is a notable finding because it may suggest that withdrawal behaviors could be a protective factor against teen dating violence for girls. Better understanding the context in which this negative association occurs and corroborating the gender differences would help to optimize interventions. For instance, in addition to promoting healthy relationships, it would suggest providing adolescents with skills to leave relationships without incurring the risk of violent confrontations (Wolfe et al. 2009).

Beyond examining the direct paths, a main objective of our study was to demonstrate indirect pathways by which individual differences in attachment style are related to teen dating violence victimization. As hypothesized, anxious attachment was indirectly linked to experienced psychological abuse and physical violence through an escalation of conflicts through self-reported conflict engagement that predicted conflict engagement perceived in the partner. Among boys, anxious attachment was additionally linked to victimization through self-reported withdrawal that predicted conflict engagement perceived in the partner. Different results also emerged for boys and girls regarding the link of avoidant attachment to victimization. Among boys, avoidant attachment was related to both types of victimization through self-reported withdrawal strategies and conflict engagement behaviors attributed to their partner. No parallel indirect effect of avoidant attachment was found for girls. One explanation could be that boys in our sample were more avoidantly attached than girls, so the level of avoidant attachment among girls might have been too low to play a role in the communication dynamics that predicting victimization. This reasoning is compatible with the finding that the paths from avoidant attachment to both self-reported conflict engagement and conflict engagement attributed to the partner were non-significant for girls.

These results provide further useful information for intervention programs. For example, raising awareness of the link between insecure attachment and destructive conflict resolution patterns may help teens to improve their communication in response to disagreements with a romantic partner. Programs should also focus on emotion regulation to address conflict communication patterns and attachment schemas to reduce dating violence victimization. This may be specifically helpful for adolescents, as they often show a lack of effective emotion regulation strategies compared to adults (Zimmermann and Iwanski 2014), which is related to insecure attachment (Maas et al. 2011) and poor interpersonal skills (Espelage et al. 2015).

Altogether, our findings emphasize the importance of jointly analyzing both attachment styles and conflict resolution strategies as predictors of psychological abuse and physical violence victimization in teen dating relationships. The gender differences observed in several associations underline the necessity to examine gender-specific pathways to teen dating violence victimization. For example, the direct effect suggested that withdrawal behaviors may be a protective factor against psychological abuse for girls. For boys self-reported withdrawal emerged as a vulnerability factor for psychological and physical victimization by predicting conflict engagement attributed to the partner.

The current study has some limitations that need to be addressed in future research. First, information gathered to develop the model comes from only one of the partners. Therefore, it would be necessary to test the model using the responses of both partners. Second, data collection was conducted in high schools in Spain, which calls for replication of these findings in other populations. Another limitation of this study is the cross-sectional design that does not permit any conclusions about causality. Previous longitudinal research with adults and adolescents has found that changes in the perception of the partner are related to attachment style (Segal and Fraley 2015) and to experiences of dating violence (Calvete et al. 2016). Hence, longitudinal studies would provide insight into possible causal relations among attachment style, conflict resolution dynamics, and vulnerability to teen dating violence victimization.

Conclusion

The current study extends past research on adolescents’ experience of dating violence by testing a model that conceptualizes the associations of romantic attachment style and conflict resolution strategies in predicting psychological abuse and physical violence victimization. Our study provides evidence of how insecure attachment and teen dating violence victimization are indirectly linked, identifying gender-specific paths. Specifically, anxious attachment was related to victimization via conflict engagement strategies (self-reported and perceived in the partner) among boys and girls, whereas both anxious and avoidant attachment styles were indirectly linked to victimization via self-reported withdrawal and conflict engagement perceived in the partner only among boys. This study has implications for the design of prevention programs aimed at facilitating healthy relationships through training constructive conflict resolution strategies. Moreover, it highlights the role of individual differences in attachment style in relation to dysfunctional conflict resolution strategies.