From Family Dynamics to Online Interactions: Exploring the Predictors of Aggression in Social Media Settings

1. Introduction

The ever-changing environment of technology and internet relationships is every day more widespread, especially among adolescents and young people [1]. Nowadays, online interactions are normalized in youth, facilitating contact and connection, while also encompassing risks related to online control and monitoring [2,3]. Cyber dating violence, also known as cyber intimate partner violence (CIPV), is one of the most extreme but also common of those risks, entailing distinct forms of abuse in interpersonal relationships, which vary depending on their specific characteristics and whether it is perpetrated or experienced [4]. Cyber-aggression perpetration, on one hand, refers to hostile or threatening behaviors enacted through digital means, such as insults, humiliation, or spreading rumors [5]. Cyber-control perpetration, on the other hand, involves invasive or coercive monitoring of a partner’s online activities—such as demanding access to passwords or checking messages without consent [6]. Online users may also be on the receiving end of such behaviors, reflected in cyber-victimization, which includes experiencing online aggression from a partner, and received cyber-control, which refers to being digitally monitored, pressured, or restricted in one's online autonomy [5,7]. As mentioned above, this specific kind of violence can vary greatly depending on the developmental stage in which it is both experienced and perpetrated, which raises concerns on its study and prevention [8].

This is particularly problematic during adolescence, a period ranging from ages 10 to 19 according to the World Health Organization that already entails severe psychosocial and emotional changes, as it follows the person while they transition from child to adult [11]. This stage, marked by adolescents’ heightened emphasis on peer relationships and engagement in unsupervised online activities, provides the context for the internalization and normalization of violent dating behaviors, which may then be reproduced within their own intimate relationships [5,7,12]. In adolescence, CIPV is an increasingly debated, researched and worrisome topic, since it is accompanied by the enormous social and technological changes that define the current era and affect young people the most [3,9]. Mirroring patterns of offline abuse within the digital sphere, CIPV engages in aggression, control, harassment, and surveillance within romantic relationships in online contexts, providing enhanced feelings of immunity towards the perpetrator and vulnerability in the victim [7,10]. This phenomenon has a strong gender factor: literature on cyber dating violence has observed a close connection between sexting (sending or receiving messages of sexual nature) and CIPV, and this relationship is stronger for boys than for girls during adolescence [13]. Additionally, girls appear to be more likely to face negative consequences of CIPV and boys have been observed to be more engaged in control or harassment behaviors over girls in online settings, suggesting that gender inequality could be a driver of this kind of violence in adolescence [4], although gender roles in CIPV remains a debate in scientific literature [14]. From this perspective, what other individual traits could be playing a role in the perpetration or experience of cyber intimate partner violence?

Literature indicates that individual traits can be associated with being the responsible or victim of certain kinds of violence, especially those associated with empathy and aggressiveness such as psychopathic traits, antisocial and law-violating behavior, and child-to-parent violence [15,16,17]. Psychopathic traits, particularly when related to callousness, impulsivity, and interpersonal manipulation, have been linked to reduced empathy and increased likelihood of engaging in manipulative or aggressive behaviors in intimate contexts like romantic relationships [17,18]. Interestingly, psychopathic traits do not only associate with cyber dating violence perpetration, but also to victimization–having experienced this kind of cyber violence in the past can also be associated with Dark Tetrad traits, which include, subclinical narcissism, Machiavellianism, subclinical psychopathy, and everyday sadism [14]. These traits may contribute to the normalization of controlling or harmful behaviors within adolescent relationships, both online and offline. Antisocial and law-violating behavior, including a broad spectrum of norm-violating actions such as aggression or rule-breaking, has similarly been identified as a risk factor for CIPV [19], suggesting that adolescents who display antisocial tendencies may be more likely to engage in digital forms of harassment, threats, or coercive control. Additionally, child-to-parent violence, which refers to physical, psychological, or financial abuse directed at parents or caregivers, has been observed to work as a predictor of patterns of family dysfunction and aggression [20]. There is a consensus in literature suggesting that child-to-parent violence presents higher rates when directed towards the mother than the father, especially when perpetrated by boys [21]. In any case, adolescents who engage in these types of behaviors often show difficulties in emotional regulation and conflict resolution, which may also appear in peer and romantic relationships, increasing the likelihood of their involvement in CIV as either perpetrators or victims [22,23]. Furthermore, literature supports a traditional belief in psychology studies: past behavior predicts future behavior; therefore, previous experiences of dating violence, online or offline, have been found to strongly predict CIVP in adolescents [24]. These individual risk factors often interact with intimate relational dynamics to shape adolescents' experiences and behaviors within their digital romantic interactions [4,25]. However, these individual traits do not exist in a vacuum – social and cultural contexts play a significant role in the way adolescents establish patterns of communication, regulate their emotions or engage in violent behaviors [26].

In that sense, Spanish adolescents have been observed to be at a particular risk of CIPV, given their extensive use of social media – it is estimated that they spend over an hour daily in social media apps such as TikTok and Instagram, making these part of their everyday lives [27]. Online spaces associated with violence–especially those linked to gender, such as dating violence–are increasingly becoming a part of mainstream discourse and daily use, fueling aggression and control in intimate relationships among Spanish adolescents [28]. Despite this urgency to address violence in online settings among youth in Spain, few studies have been conducted on this topic that combine the individual risk factors mentioned above [12]. Moreover, most of them have been conducted through linear methodologies such as hierarchical regression models (HRM), likely missing key information about how these variables interact in the prediction of CIPV. Thus, the use of non-linear methodologies such as fuzzy set Qualitative Comparative Analysis (fsQCA) in combination with traditional linear analyses could provide a more comprehensive understanding of the complex and configurational nature of CIPV, identifying multiple pathways and combinations of factors that lead to similar outcomes.

To this purpose, the current study aims to analyze the impact of psychopathic traits, antisocial behavior, child-to-parent violence, and dating violence on cyber intimate partner violence through a mixed-methods design. We hypothesize that (1) psychopathic traits, antisocial behaviors, child-to-parent violence, and dating violence will be positively associated with CIPV; (2) Cyber-aggression and cyber-control perpetration will be predicted by psychopathic traits, antisocial behaviors, child-to-parent violence, and dating violence, and (3) CIPV will be linked to high levels of psychopathy, high levels of antisocial behaviors, high child-to-parent violence towards both the mother and the father, and high levels of perpetrated or experienced dating violence.

2. Materials and Methods

2.1. Participants

This study included a sample of 223 Spanish adolescents between the ages of 14 and 18, with a mean age of 16.18 years (SD = 1.52). Most participants were female (65.90%), and 94.20% held Spanish nationality. Most students (55.60%) were enrolled in Compulsory Secondary Education (ESO), while 27.40% were in Bachillerato, and the remaining 17.00% were attending university. Informed consent was obtained from both the participants and their parents.

2.2. Variables and Instruments

Sociodemographic variables, including gender, age, nationality, and academic level, were collected using a custom-designed measure. Additionally, various validated questionnaires, proven to have strong psychometric properties in similar populations, were used to assess key variables. The selection of these instruments was carefully made to ensure a thorough and reliable evaluation.

• Violence in Adolescent Relationships on Social Media (e-VPA): this instrument consists of 20 items divided into four subscales, assessing violent behaviors towards a partner and experiences of victimization through social networks. It includes two main subscales, each containing 10 items: e-victimization and e-violence. Additionally, these subscales provide specific scores for cyber-violence perpetrated, cyber-control perpetrated, cyber-victimization, and cyber-control received. Previous studies have demonstrated good internal consistency and an adequate factorial structure [6].
• Psychopathy: was assessed using the Psychopathy Content Scale (P-16), originally developed by Salekin et al. [29]. This scale was derived from the Millon Adolescent Clinical Inventory (MACI), a 160-item self-report measure designed to evaluate personality traits and psychopathology in adolescents. The P-16 was constructed by selecting items from the MACI that theoretically aligned with the Hare Psychopathy Checklist-Revised (PCL-R) and corresponded to the psychopathy models proposed by Cooke & Michie [30] and Frick et al. [17].
• The final version of the P-16 consists of 16 dichotomous items (True/False), which are categorized into three subscales: callousness, egocentricity, and antisociality. The total score, obtained by summing the subscale scores, was used in this study. In the original validation, the internal consistency for the total scale was α = .86, while the reliability values for the subscales were .62 (callousness), .61 (egocentricity), and .56 (antisociality) [29].
• Antisocial and law-violating behaviors: the Antisocial and Criminal Behavior Scale in Adolescents (ECADA) was used to assess antisocial and law-violating behaviors [31]. This scale consists of 25 True/False items that measure different forms of delinquent conduct. The items are distributed across five key dimensions: early antisocial tendencies, vandalism, property-related offenses, violent acts, and substance use (alcohol and drugs). For this study, a total score was computed by summing all subscales, with higher scores reflecting a greater engagement in antisocial and law-violating behaviors. The ECADA has demonstrated strong psychometric properties, with internal consistency reliability coefficients ranging from α = .82 to .86 [31,32].
• Child-to-parent violence: was evaluated using the Conflict Tactics Scales (CTS2) – children-to-parents version [33,34]. For this study, the adaptation developed by the Lisis Group [35] was applied. The instrument consists of 10 items, answered separately for mother and father, using a five-point Likert scale ranging from 0 (Never) to 4 (Many times). The scale provides both a global index of child-to-parent violence and scores for three distinct dimensions: verbal violence, physical violence, and economic violence. In this study, the total score was obtained by summing the subscale scores. This version of the scale has demonstrated satisfactory psychometric properties, with internal consistency indices ranging from α = .66 to .85 across subscales [35].
• Dating violence: was assessed using a brief version of the Conflict in Adolescent Dating Relationships Inventory (CADRI) [36,37].This study employed the adaptation by the Lisis Group [38], which consists of 34 items, evenly divided between violence perpetrated (17 items) and violence received (17 items). The items are categorized into three dimensions: relational violence, verbal-emotional violence, and physical violence. Responses are recorded on a four-point scale, ranging from 0 (Never) to 3 (Frequently, six or more times). In this study, only received physical violence and received verbal-emotional violence were analyzed to assess dating violence. The original version of the scale reported an internal consistency of α = .83 [37], while the Spanish adaptation obtained α = .86 [36].

2.3. Procedure and Design

This study employed a convenience sampling method, selecting adolescents from various educational institutions in the Valencian Community based on their availability and willingness to participate. The sample included individuals aged 14 to 18 years, ensuring diversity while maintaining a manageable group size.

Before data collection, researchers identified appropriate assessment instruments and contacted educational centers through an introductory letter explaining the study. Follow-up communication was conducted via phone and email to discuss participation with school administrators. Upon obtaining their approval, further meetings were arranged to provide detailed information about the research. Necessary authorizations, including informed consent from the Regional Ministry of Education and parental permissions, were secured to allow students to complete the surveys.

The surveys were administered in a single one-hour session, after which responses were systematically recorded in a database for later statistical analysis. The results were then processed and interpreted to draw relevant conclusions.

To ensure methodological rigor, a pilot study was conducted with a group of university students, who assessed the clarity of the questions and estimated completion time. This process allowed researchers to refine and optimize the final version of the questionnaire. This study had a cross-sectional design.

The study followed ethical research standards, complying with the principles outlined in the Helsinki Declaration, including participant rights to informed consent, privacy, data confidentiality, non-discrimination, voluntary participation, and the option to withdraw at any point. Ethical approval was obtained from the Ethics Committee of the Universitat de València (REF 2024-PSILOG-3592610).

2.4. Analysis

Descriptive analyses were conducted to characterize the sample and determine calibration values. Additionally, Pearson correlations were performed. For predictive analyses, hierarchical linear regression models and fuzzy set Qualitative Comparative Analysis (fsQCA) were applied. The analyses were carried out using SPSS v28 and fsQCA 3.2.

Fuzzy set Qualitative Comparative Analysis (fsQCA) is a configurational analytical approach that explores how different conditions work together to bring about a particular outcome [39]. In contrast to conventional regression techniques—which typically assume linear, additive, and independent effects—fsQCA allows for the identification of multiple, equally valid combinations of variables (equifinality) that can lead to the same result. This makes it especially valuable in social science research, where variable relationships are often complex, non-linear, and influenced by context. In the present study, fsQCA was used alongside hierarchical regression to gain a deeper insight into the factors contributing to CIPV in Spanish adolescents.

Regarding this analysis, missing data were first removed, and recalibration was performed based on three threshold settings: 0% (low level/exclusion), 50% (intermediate level/partial membership), and 90% (high level/full inclusion). Next, the necessary and sufficient conditions were calculated. Necessary conditions are those that must always be present for an outcome to occur, while sufficient conditions can lead to an outcome but are not always required.

QCA models evaluate explained variance, measure the proportion of cases where the model applies (coverage), and assess model accuracy (consistency). A condition is considered necessary if its consistency is ≥.90, while a model is deemed sufficiently informative if its consistency is around or above .74. In the fsQCA analysis, both the intermediate and complex solutions were combined to account for core and peripheral conditions, ensuring a more comprehensive interpretation of the configurational pathways leading to cyber intimate partner violence. Unlike conventional regression techniques, this approach emphasizes the combinations of conditions that lead to specific outcomes rather than isolating the independent effect of each variable.

3. Results

3.1. Correlational Analysis

Moderate, positive, and statistically significant linear associations were found among all the studied variables. However, child-to-parent violence towards the father showed no significant relationship with cyber-violence in romantic relationships, nor did psychopathy correlate with cyber-violence victimization in relationships. Overall, different forms of violence (directed towards parents and romantic partners), psychopathy, and delinquent behaviors were interrelated. More information can be found in Table 1.

3.2. Predictive Analysis

3.2.1. Hierarchical Regression Models

The study aimed to predict cyber dating aggression (including both cyber-aggression and cyber-control) and cyber victimization (cyber-victimization and received cyber-control) using a hierarchical model. The predictors were introduced in three steps: antisocial and law-violating behaviors and psychopathy in the first step, child-to-parent violence (towards both mother and father) in the second step, and intimate partner violence (both perpetrated and experienced) in the third step.

The prediction model for cyber-aggression accounted for 23% of the variance, with child-to-parent violence towards both parents and experienced intimate partner violence emerging as significant predictors. Meanwhile, the cyber-control perpetration model explained 40% of the variance, primarily influenced by psychopathy and perpetrated intimate partner violence.

Regarding cyber-victimization, the model explained 34% of the variance in the final step, with child-to-parent violence towards both parents and experienced intimate partner violence as key contributors. Lastly, the prediction model for received cyber-control accounted for 36% of the variance, with child-to-parent violence towards the father and experienced intimate partner violence as significant predictors. Table 2 presents the detailed results of the hierarchical regression models.

3.2.2. Fuzzy Set Qualitative Comparative Fuzzy Set Analysis (fsQCA)

Analysis of necessary conditions

First, the primary descriptive statistics and calibration values for the study variables are presented (Table 3). The analysis revealed that none of the examined conditions met the threshold for being considered necessary for predicting high or low levels of cyber dating aggression (including both cyber-aggression and cyber-control) or cyber victimization (cyber-victimization and received cyber-control). This conclusion is since the consistency values remained below 0.90 in all cases (Table 4), as recommended by Ragin (2008).

Additionally, it was not possible to calculate the cyber-aggression perpetration model using QCA due to the lack of variability in this variable. The homogeneity in responses prevented the identification of sufficient diversity in causal conditions, making it unfeasible to extract meaningful configurational patterns for cyber-aggression. However, the other subfactors of cyber-control perpetrated, cyber-victimization, and received cyber-control were analyzed as planned.

Analysis of sufficiency conditions

In the sufficiency analysis, the combinations of conditions that led to high and low levels of cyber-control perpetrated, cyber-victimization, and received cyber-control were identified (Table 5). Following the criteria established in fsQCA, where a model is considered informative when consistency values are around or above 0.74 (Eng & Woodside, 2012), all the models met this threshold, ensuring robust findings.

For high levels of perpetrated cyber-control towards the partner, five pathways were identified, collectively explaining 72% of the cases (overall consistency = .77; overall coverage = .72). The most significant configuration involved the presence of psychopathy, perpetrated intimate partner violence, and experienced intimate partner violence, accounting for 62% of the cases. Another relevant pathway, explaining 54% of the cases, was characterized by experienced and perpetrated intimate partner violence combined with child-to-parent violence towards the mother. A third configuration, which explained 35% of the cases, included psychopathy, antisocial and law-violating behaviors, experienced intimate partner violence, and the absence of child-to-parent violence towards the father.

For low levels of perpetrated cyber-control towards the partner, five pathways were identified, explaining 87% of the cases (overall consistency = .77; overall coverage = .87). The most relevant configuration considered solely the absence of experienced intimate partner violence, accounting for 75% of the cases. Another key pathway, explaining 73% of the cases, was defined by the absence of perpetrated intimate partner violence. Additionally, a third pathway, explaining 51% of the cases, involved the absence of psychopathy and antisocial and law-violating behaviors.

For cyber-victimization, a total of 13 pathways were identified, collectively explaining 96% of the variance (overall consistency = .74; overall coverage = .96). Among these, one of the most relevant configurations, explaining 46% of the cases, was defined by the absence of psychopathy and child-to-parent violence towards both parents. Another configuration, which also accounted for 46% of the variance, combined the presence of psychopathy, antisocial and law-violating behaviors, and both perpetrated and experienced intimate partner violence. Additionally, another meaningful pattern, explaining 44% of the cases, was characterized by the absence of delinquent behaviors, absence of child-to-parent violence towards both parents, and absence of experienced intimate partner violence.

The analysis of low levels of cyber-victimization revealed eight pathways, which together accounted for 77% of the variance (overall consistency = .87; overall coverage = .77). A prominent configuration, explaining 49% of the cases, was defined by the presence of perpetrated intimate partner violence and the absence of experienced intimate partner violence. Another relevant combination, explaining 38% of the cases, included the absence of antisocial and law-violating behaviors, absence of experienced intimate partner violence, and presence of child-to-parent violence towards the father. A further pattern, explaining 36% of the cases, was characterized by the presence of psychopathy, the absence of antisocial and law-violating behaviors, the presence of child-to-parent violence towards the mother, and the presence of experienced intimate partner violence.

Regarding the experienced cyber-control, three pathways explained 51% of the variance (overall consistency = .81; overall coverage = .51). The first configuration, which accounted for 35% of the cases, was characterized by the presence of psychopathy, antisocial and law-violating behaviors, perpetrated and experienced intimate partner violence, and the absence of child-to-parent violence towards the father. A second pathway, explaining 33% of the cases, included a similar combination but with the absence of child-to-parent violence towards the mother. The third configuration, explaining 31% of the cases, involved child-to-parent violence towards both the mother and the father, perpetrated and experienced intimate partner violence, and the absence of antisocial and law-violating behaviors.

For low levels of experienced cyber-control, seven pathways were identified, collectively explaining 72% of the cases (overall consistency = .88; overall coverage = .72). The most relevant configuration was characterized by the absence of antisocial and law-violating behaviors, absence of perpetrated intimate partner violence, and absence of experienced intimate partner violence. Another key pathway, accounting for 46% of the cases, involved the absence of antisocial and law-violating behaviors, absence of child-to-parent violence towards the mother, and absence of experienced intimate partner violence. Additionally, a third relevant pattern, explaining 44% of the cases, was defined by the absence of child-to-parent violence towards both the mother and the father, as well as the absence of both perpetrated and experienced intimate partner violence.

It is important to highlight that the perpetrated cyber-aggression model could not be estimated using fsQCA due to a lack of variability in this variable. The homogeneity of responses resulted in insufficient diversity in causal conditions, preventing the identification of meaningful configurational patterns. However, the analyses for cyber-control perpetrated, cyber-victimization, and received cyber-control were conducted as planned, providing valuable insights into the mechanisms underlying cyber intimate partner violence.

4. Discussion

The present study aimed to analyze the impact of psychopathic traits, antisocial behavior, child-to-parent violence, and dating violence on cyber intimate partner violence through a mixed-methods approach. This research presents a strong contribution to research by filling the gap in literature on CIPV and its predictors, analyzing them in combination and finding the specific configurations in which they jointly operate. Psychopathic traits, antisocial behavior, child-to-parent violence, and dating violence have unique relationships with CIPV, while also interacting in distinct ways that create multiple pathways to both perpetration and victimization, highlighting the multifactorial nature of adolescent cyber intimate partner violence. Furthermore, this study provides a novel contribution by combining hierarchical regression models (HRM) and fuzzy set Qualitative Comparative Analysis (fsQCA) to explore predictors of CIPV among adolescents, offering both linear and configurational perspectives.

Our first hypothesis stated that psychopathic traits, antisocial behaviors, child-to-parent violence, and dating violence would be positively associated with CIPV. Correlation and HRM analyses partially support this hypothesis. Psychopathic trats were related to all forms of CIPV except cyber-violence victimization in relationships. This aligns with previous research indicating that there exists a relationship between psychopathic traits and all dimensions of CIPV, both linked to perpetration and experience of dating online violence [14,18]. Similarly, antisocial and law-violating behaviors have also been found to correlate with all kinds of CIPV, aligning with previous research on the topic [19]. These associations might be due to the normalization of aggression and violence in Spanish adolescents, in which lower levels of empathy could play a significant role in facilitating both perpetration and experience of CIPV. Regarding child-to-parent violence, it presented a significant relationship with CIPV when directed towards the mother, but not directed towards the father. This result is consistent with previous literature indicating that there is a significant difference between the violence perpetrated towards mothers of fathers, being the former more frequent and probably linked to other kinds of violence, such as CIPV [21]. Additionally, both perpetrated and experienced intimate partner violence were positively associated to all dimensions of CIPV, suggesting that cyber violence in these settings might mirror offline forms of intimate partner violence for adolescents, as previous literature also observed [24]. These findings support the idea that CIPV is embedded within a broader pattern of aggression and transgressive behaviors in adolescence, though not all variables equally contribute across all forms of cyber-violence.

Our second hypothesis indicated that cyber-aggression and cyber-control perpetration would be predicted by psychopathic traits, antisocial behaviors, child-to-parent violence, and dating violence. Our results partially support this hypothesis, expanding on the correlations with all dimensions of CIPV previously found. On the first step, the model including psychopathy and antisocial behavior was significant overall, with antisocial behavior being a significant predictor on all dimensions of CIPV and psychopathy only being significant to forms of cyber control, but not aggression, both perpetrated and experienced. Psychopathy being linked to a specific form of intimate partner violence and not others has been addressed in literature, but the topic is still debated–contrary to previous research, psychopathy seems to be more associated to manipulative and controlling forms of violence rather than direct aggression in Spanish adolescents [14].

In the second step, adding child-to-parent violence added a significant amount of variance in perpetrated and experienced forms of violence, but not control, being violence towards the mother a positive predictor and towards the father, negative. This might indicate that Spanish adolescents who engage in aggressive forms of CIPV tend to be more violent towards their mothers and have more respect for their fathers. Previous research has addressed these distinct adolescents’ gender perceptions of their caregivers: traditional gender roles are strongly linked with sexism in domestic settings, in which mothers become the main caregivers and attachment figures and fathers adopt a more detached position [21]. This, together with increasing discourse of male supremacy and anti-feminism found in online settings (as well as in family dynamics) could lead adolescents who are familiarized with CIPV to be violent towards their mothers while respecting, and even idolizing, their fathers [4]. This suggests a potential line of research exploring whether the fathers’ attitudes towards their female spouses could mediate these gender-based dynamics.

The final model, tested in the third step, provided the highest amount of explained variance by adding perpetrated and experienced dating violence as predictors. As observed in previous literature, perpetrated violence in intimate settings was a predictor of perpetrated CIPV, and experienced violence strongly predicted experienced CIPV, the latter being the strongest predictor [24]. These results suggest that adolescents who have been victims of offline dating violence may be more vulnerable to being involved in online forms of violence, either as a result of trauma, normalization of controlling behaviors, or reciprocation in the digital space. This highlights the importance of understanding CIPV not in isolation, but as part of a continuum of violence that spans offline and online interactions.

The third and last hypothesis posited that CIPV would be linked to high levels of psychopathy, high levels of antisocial behaviors, high child-to-parent violence towards both the mother and the father, and high levels of perpetrated or experienced dating violence, as indicated by previous research. Our results partially support this hypothesis: while HRM provided useful information about the linear relationships between CIPV and its predictors, fsQCA analyses shed light on non-linear associations, adding important nuance and explaining these relationships in a different, richer paradigm.

Starting with perpetrated cyber-control, the strongest predictor across pathways was the presence of experienced intimate partner violence, closely followed by perpetrated intimate partner violence, supporting previous research on the topic [7,10]. These results suggest that engaging in controlling forms of violence towards partners is linked to a profile of adolescents who both engage in and endure violence in their relationships. The presence of psychopathic traits also emerged as a strong predictor in two of the three main pathways for this type of violence [14]. Together with the previous predictors, this suggests a reciprocal and instrumental dynamic of intimate partner violence, where control is used deliberately, perhaps to preempt or retaliate against perceived threats. Furthermore, being the perpetrator of CIPV was linked to profiles of both presence of child-to-parent violence towards mothers and absence of this type of violence towards fathers [22,40]. This further supports our previous hypothesis that Spanish adolescents who are violent towards their romantic partners might have a profile rooted in sexist ideas of men and women, facilitating contempt towards their mothers and respect towards their fathers. Future research is encouraged to explore this phenomenon in more depth.

Adding to these profiles, lower levels of perpetrated cyber aggression were linked to the absence of variables explained above: experienced intimate partner violence, perpetrated intimate partner violence and psychopathy and antisocial and law-violating behaviors, further supporting our established hypotheses. Due to variability issues, perpetrated cyber-aggression could not be analyzed.

As per experienced cyber-aggression, pathways shown in fsQCA analyses are different across variables, but patterns can be identified. Multiple pathways were identified, reflecting the heterogeneous nature of victim profiles, which indicates that adolescents may become victims of CIPV through multiple trajectories. One important configuration included the absence of psychopathy and child-to-parent violence, indicating a more vulnerable victim profile—individuals who are not themselves aggressive but may be involved in asymmetric or coercive relationships. Conversely, another equally relevant pathway featured the presence of psychopathy, antisocial behavior, and both perpetrated and experienced dating violence, in line with previous research [17,41]. This suggests a profile of mutual violence or "victim-perpetrators", consistent with research showing bidirectional dynamics in adolescent relationships [5,26]. Perhaps the most consistent trait of the main profiles was the absence of child-to-parent violence towards both their mother and father, suggesting that they might be less experienced with conflictual dynamics, less likely to recognize early signs of coercion, or more inclined to adopt passive or compliant roles in romantic relationships.

Finally, regarding experienced cyber-control, analyses included various combinations of psychopathy, antisocial behavior, dating violence, and child-to-parent violence, suggesting again a bidirectional or mutually violent relationship structure. The overlap of both perpetrated and experienced dating violence in all configurations aligns with findings of perpetrated cyber control, and reinforces the idea that control in relationships is often reciprocated or normalized, particularly when both partners show high-risk traits [10,12]. Results on lower experienced cyber-control mirror these findings: the absence of antisocial behavior, dating violence, and child-to-parent violence emerge in the most relevant configurations, suggesting that adolescents embedded in non-violent relational environments and with low-risk individual profiles are significantly less likely to experience controlling behaviors from their partners, possibly due to healthier relationship models, greater assertiveness, and mutual respect within their romantic interactions. These patterns highlight the protective nature of non-violent relational contexts and low-risk personality profiles, suggesting potential lines of intervention to prevent CIPV in Spanish adolescents.

Research and clinical implications

The present study offers important insights for the prevention and intervention of CIPV among adolescents. The results emphasize the need for professionals working with youth to assess not only romantic relationship dynamics, but also online communication habits—particularly in contexts of conflict, control, or jealousy [24]. Family dynamics can also be a potential line of intervention according to our findings, as child-to-parent violence is strongly associated with cyber aggression and control in romantic relationships in Spanish adolescents. Literature suggests that designing interventions based on attachment insecurity or parental styles could target the potential roots of behaviors and attitudes towards violence in children and adolescents, preventing that maladaptive relationship dynamics could be transferred from the household to intimate relationships, especially in online settings [42]. These programs should also include gender-based perceptions of caregivers, as research consistently finds connections between aggressive or controlling behaviors in adolescents and their treatment of their parents [16,22]. Targeting underlying beliefs of sexism when addressing adolescents’ violent behaviors towards their mothers could prevent other forms of aggression in their intimate relationships, ultimately protecting both adolescents and female caregivers from these forms of violence [20,23,40].

These programs could also include high-risk individual factors that place adolescents in particular vulnerability towards CIPV. These could focus on teaching healthy communication skills, and raising awareness of how individual traits like impulsivity or low empathy can shape dating behaviors. Since both psychopathic traits, antisocial behavior and previous experiences of violence (within the family or romantic contexts) emerged as key risk factors, interventions should target emotion regulation, empathy-building, and conflict resolution—especially in youth with behavioral or relational difficulties [10,18].

Importantly, interventions should also address harmful online practices, including exposure to websites or social media content that promote sexist beliefs, normalize violence in relationships, or reinforce toxic gender roles [4,25]. Adolescents frequently engage with online spaces that may encourage hostility, jealousy, and control under the guise of romantic interest—increasing the risk of CIPV [42]. Literature points at the strong connection between online and offline violence, suggesting that despite healthy family dynamics, adolescents are still exposed to harmful discourse about intimate and romantic relationships [43]. Thus, these findings could direct interventions to focus on debunking myths about digital control, teaching about ethic online practices and encouraging adolescents to engage in positive communication with their intimate partners, ultimately preventing cyber dating violence perpetration and experience.

Limitations and future research

The present study is not without limitations. First, the study used a cross-sectional design, which limits the ability to make conclusions about cause and effect. Longitudinal studies would be useful to further understand associations between variables, since cross-sectional data does not allow to confirm the direction of these relationships. This kind of research could also explore how different types of violence appear and change throughout adolescence, helping us design better interventions adapted to specific life stages. Additionally, future studies could examine the role of other factors (e.g., attachment styles, trauma or family dynamics) as potential mediators or moderators in these processes.

The second limitation addresses the characteristics of the sample. Convenience sampling was conducting, therefore the results should be interpreted with caution. All participants came from educational settings, where the levels of dating violence and child-to-parent violence (CPV) tend to be moderate or low, potentially biasing results. More diversity is advised for future research to ensure generalizability: Including cross-cultural include young people from other environments, such as juvenile justice facilities, would ensure representation of adolescents showing more serious or high-risk behaviors.

Third, we relied on self-reported questionnaires, introducing the possibility of bias. Although this method is useful for capturing adolescents’ personal experiences, it can be affected by social desirability or memory errors. Including reports from other sources, such as parents, teachers, or professionals, could help verify and enrich the data, offering a fuller view of the adolescents’ behavior and context.

Finally, one notable limitation of the present study is the insufficient variability in the perpetrated cyber-aggression subfactor, which prevented its inclusion in the fsQCA analyses. Several factors may account for this lack of variability. First, sample-related biases such as the relatively small sample size and the possibility of a restricted range in responses due to social desirability bias—especially given the sensitive nature of admitting to violent behaviors—may have led participants to underreport perpetration. In contrast, hierarchical regression models (HRM) are able to accommodate limited variability and still produce interpretable results, whereas fsQCA requires sufficient distribution across conditions to identify meaningful configurational patterns. Furthermore, it is worth considering whether the instrument used to assess cyber-aggression perpetration was sufficiently sensitive or whether alternative or complementary measures might capture a broader range of behaviors more effectively. Finally, this limitation could also reflect an inherent characteristic of the phenomenon itself, in which cyber-aggression perpetration among adolescents may indeed occur less frequently or be less readily acknowledged, particularly in community-based samples. Future research should aim to replicate findings with larger and more diverse samples and explore the utility of other instruments or qualitative methods to capture the full spectrum of perpetrated behaviors in adolescent digital relationships.

Reflecting on these limitations could direct further research on this important topic, especially when designing intervention programs for adolescents with high-risk traits and behaviors who are more likely to engage in CIPV.

5. Conclusions

In conclusion, this study explored the predictors of cyber intimate partner violence (CIPV) in adolescents through both hierarchical regression models (HRM) and fuzzy set Qualitative Comparative Analysis (fsQCA). The findings highlight the complex interplay between individual traits—such as psychopathy and antisocial behavior—and relational experiences, including child-to-parent and dating violence, in shaping patterns of both perpetration and victimization in the digital sphere. The combination of linear and non-linear methodologies provided valuable nuance to the interpretation of our results, reaching conclusions that would have been unattainable relying solely on linear analyses. The configurational approach provided by fsQCA analyses revealed that multiple risk pathways can lead to similar outcomes, emphasizing the need for flexible, tailored prevention actions.

These results underscore the importance of developing early intervention programs that address not only individual behavior but also family dynamics and the quality of adolescents’ romantic relationships. Schools, families, and community organizations must work together to foster empathy, emotional regulation, and healthy relationship skills among young people. By doing so, we can help create safer digital and offline spaces for adolescents—spaces where trust, respect, and connection take the place of control and harm. There is still much work to be done, but by listening to young people and acting early, we can help change their stories—for the better.