Introduction

Intimate partner violence (IPV) is a major public health concern and a serious violation of human rights, with far-reaching physical, mental, social, and economic consequences (Gunarathne et al., 2023; Laskey et al., 2019). Although traditionally conceptualized as male-perpetrated violence driven by power and control over a female partner—a perspective widely adopted in research, policy, and interventions (Bates, 2016; Langhinrichsen-Rohling et al., 2012; Laskey et al., 2019; Mennicke & Wilke, 2015)—research has increasingly documented bidirectional patterns in which both partners perpetrate and experience violence (Langhinrichsen-Rohling et al., 2012; Lysova et al., 2024; Machado et al., 2024). One of the most influential contributions to this model came from Johnson (1995, 2006), who proposed an IPV typology that included four distinct categories: intimate terrorism (IT), violent resistance (VR), situational couple violence (SCV), and mutual control (MC). While IT aligns with the classical patriarchal and unidirectional perspectives (Bates, 2016), other types suggest a bidirectional pattern in which both partners engage in IPV (Bates, 2016; Johnson, 2006).

Research has consistently suggested that bidirectional IPV is the most prevalent form of IPV across different samples (e.g., clinical, community, and students) and continents (Langhinrichsen-Rohling et al., 2012; Lysova et al., 2024; Machado et al., 2024). Although public perceptions often downplay the severity of bidirectional violence (BV), viewing it as involving lower offender risk, less harm to victims, and greater victim responsibility (Ahmed et al., 2024), research suggests that BV can result in more severe consequences than unidirectional violence, including heightened psychological and physical harm (Behnken et al., 2018; Hellmuth, Jaquier, Swan, & Sullivan, 2014; Palmetto et al., 2013; Ulloa & Hammett, 2016).

Despite the growing recognition of BV, literature remains marked by conceptual inconsistencies and methodological limitations. While some studies conceptualize BV as symmetrical in terms of frequency or intensity, others emphasize gender-based asymmetries in intimate partner conflicts (Hamel et al., 2022; Langhinrichsen-Rohling et al., 2012). BV often reflects complex dynamics, including physical differences between men and women, divergent motivations and causes of violence, varied consequences (Palmetto et al., 2013; Ridings et al., 2021), subjective experiences of victimization, and severity of inflicted injuries—particularly when one partner acts primarily in self-defense (Hamel et al., 2022; Langhinrichsen-Rohling et al., 2012; Mennicke & Wilke, 2015). Further complicating the study of BV is the persistent inconsistencies in its definition and operationalization (Cunha et al., 2025; Lysova et al., 2024; Machado et al., 2024; Santiago et al., 2025; Stover et al., 2025). This phenomenon has been labeled in different studies with a panoply of terms such as mutual violence, symmetrical violence, reciprocal violence, victim-offender overlap, and situational violence, each reflecting slightly different theoretical assumptions (Hines & Douglas, 2018; Machado et al., 2024; Muftić et al., 2012). This terminological fragmentation contributes to conceptual confusion and hinders the accumulation of systematic knowledge.

To address this gap, a scoping review of the scientific literature on BV in intimate relationships was conducted to understand the key issues and challenges related to its definition and operationalization. Based on this analysis, an integrative definition of the phenomenon will be proposed, and its implications for research, clinical practice, and public policy will be discussed.

Method

Protocol and Registration

This scoping review followed the framework proposed by Peters et al. (2022) and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and extensions specific to scoping reviews (PRISMA-ScR; Tricco et al., 2018). This review has been registered on the OSF REGISTRIES [Project DOI: 10.17605/OSF.IO/TYHK9].

Eligibility Criteria

Studies were eligible if they included an explicit definition or conceptualization of bidirectional IPV or related constructs (e.g., mutual violence, reciprocal violence, and SCV), regardless of whether they articulated a broader theoretical framework. Studies that did not include an explicit definition of bidirectional IPV or a related construct were excluded. The inclusion criteria were primary or original studies, reviews, systematic reviews, commentaries, reports, and guidelines written in Portuguese, Spanish, or English and published between 1995 and January 2025. Reviews, systematic reviews, commentaries, reports, and guidelines were included because they provide explicit conceptual definitions of BV and related constructs, which are central to the aims of this scoping review. These types of papers often synthesize and critically examine definitions across multiple sources, offering a broader perspective on conceptualizations that may not be captured in empirical studies alone. Their inclusion therefore allowed for a more comprehensive mapping of how bidirectional IPV has been defined and conceptualized over time.

Information Sources and Search Process

The following equation was used: (“mutual violence” OR “bidirectional* violence” OR “symmetrical violence” OR “perpetrator-victim role” OR “situational partner violence” OR “victim-perpetrator overlap” OR “couple violence” OR “reciprocal violence” OR “mutual control”) AND (defin* OR classif* OR Concept*). In January 2025, two reviewers conducted a search using the equation across five electronic databases, PubMed, B-on, Web of Science, and SCOPUS, to examine titles, abstracts, and keywords.

Study Selection

All titles and abstracts retrieved through the search strategy were imported into Rayyan (Ouzzani et al., 2016). During the initial screening phase, the two reviewers independently assessed the titles and abstracts to determine their eligibility. The same two reviewers independently evaluated the full text of the selected studies. Any disagreements between reviewers were resolved through discussion until a consensus was reached. If no agreement was reached, a third reviewer resolved the disagreement. The level of agreement between the raters was strong, with a Cohen’s kappa coefficient of .87.

Data Extraction and Analysis

A detailed codebook was developed to guide data extraction from all included papers. Extracted information included study identification (e.g., title, author(s), and year of publication), sample characteristics (e.g., gender, sample size, age, and study setting), study design (e.g., methodology), research objectives, and the explicit conceptual or operational definitions of BV or related constructs provided by the authors. Definitions were extracted verbatim from each article.

Multiple definitions were extracted only when explicitly articulated within the same article, primarily in theoretical or review papers, but studies were classified according to the definition most clearly aligned with the study’s primary conceptual framing. In empirical studies, coding focused on the definition explicitly linked by the authors to the operationalization or measurement of bidirectional violence, rather than on all conceptual discussions presented in the text. Two reviewers independently extracted data, including definitions. Discrepancies between the two raters were first reviewed and resolved through discussion. A third reviewer was available to intervene when consensus could not be reached. Based on the extracted definitions, an analysis grid to categorize studies according to the terminology used to describe BV and the presence of an explicit typological framework. Although multiple terminological or conceptual references may co-occur within the same article, categorization was based solely on the definitions explicitly provided by the authors, without reinterpretation or inference. When definitional excerpts were similar across studies, but labeled differently (e.g., mutual violence, bidirectional violence, reciprocal violence), classification followed the terminology explicitly used by the authors. Johnson’s typologies were coded only when explicitly defined and applied by the authors, rather than when referenced solely for contextual purposes. No classifications or inferences were made based on measurement strategies, analytic methods, operationalization, or observed patterns of violence in the absence of an explicit definition; only definitions explicitly stated by the authors were included. Table S1 should be understood as a transparent documentation of the extracted definitional excerpts used to inform categorization, rather than as a comprehensive representation of each article’s broader conceptual orientation.

Quality Assessment

The quality of the included studies was assessed using the Quality Assessment Tool for Studies with Diverse Designs (QATSDD; Sirriyeh et al., 2012), a 16-item instrument designed to evaluate the methodological quality of qualitative, quantitative, mixed-methods, and multi-method studies. Each item is rated on a scale of 0 to 3, with higher scores reflecting stronger methodological rigor. The maximum total score is 42 for purely qualitative or quantitative studies and 48 for mixed-methods research. The total score is then converted into a percentage, providing an overall quality rating. Based on the final score, studies are classified as low quality (≤50%), moderate quality (51%–74%), or high quality (≥75%; Hines & Douglas, 2019). While the tool was originally developed for primary research, several of its criteria are applicable to review and commentary papers. For items that were not relevant to non-empirical studies, we noted them as not applicable, and these items were excluded from the total score calculation. In line with the aims of this scoping review, the quality assessment was conducted for descriptive and contextual purpose only and was not used as a criterion for study inclusion or to evaluate the adequacy or validity of the definitions identified.

Results

Before presenting the results, it is important to note that the categories used in this review reflect dominant definitional tendencies within the literature rather than mutually exclusive or empirically uniform constructs. Several definitions of bidirectional IPV share substantial conceptual overlap, and distinctions between categories often reflect differences in emphasis (e.g., symmetry, control, temporal framing) rather than fundamentally different phenomena. Accordingly, the classifications presented should be interpreted as heuristic tools to organize a fragmented field, rather than as definitive taxonomies. Additionally, in several studies, there is a lack of alignment between how concepts are defined and how they are operationalized in empirical analyses. When highly similar definitional formulations appeared across studies classified into different categories, this reflected differences in how authors explicitly framed their work within distinct conceptual or terminological traditions, rather than indicating substantive differences in the phenomena described.

Selection of Evidence Sources

In total, 6,640 articles were identified from the selected databases. After removing duplicates, 1,356 articles remained for screening based on titles and abstracts, and 47 additional studies were identified through manual searching. Subsequently, 1,298 articles were excluded because they did not meet the eligibility criteria. A full-text review of 105 articles led to the exclusion of 23. Ultimately, 82 articles were deemed eligible and included in the scoping review (see Figure 1).

Quality Assessment

This scoping review included articles classified as low (n = 15), acceptable (n = 54), and high (n = 13) quality based on the QATSDD scoring procedures and thresholds (see Table 1). The highest quality scores were achieved by Hines and Douglas (2019) and Chen and Chan (2021), with scores of 88% and 85%, respectively. Eleven studies received scores around 70% (see Table 1). Among the low-quality studies, only one (Straus & Gozjolko, 2016) scored below 40%.

Study Characteristics

The included studies were published between 1995 (Johnson, 1995) and 2025 (Stover et al., 2025), with a notable increase in publications after 2010. Almost half of the included studies were conducted in the United States (n = 41, 50%), while the remaining studies were conducted in a variety of other countries (see Table 1). The sample sizes in quantitative studies ranged from 68 (Crane et al., 2014) to 25,778 participants (Afifi et al., 2012). Approximately half of the studies used mixed-gender samples (n = 32; 56.14%), with a quarter focusing exclusively on women (n = 16; 27%), and a residual number included only men (n = 7; 12.5%). The reported mean age of participants ranged from 15.5 years (Jennings et al., 2012) to 43.89 years (Hines & Douglas, 2018, 2019), although 15 studies did not report this information. Data were collected primarily in community settings (48%), with smaller proportions obtained from clinical and forensic contexts. The Conflict Tactics Scale (CTS/CTS2) was the most common measure.

Conceptual Definitions of Bidirectional Violence in Intimate Relationships

This review identified two predominant conceptual approaches to intimate violence involving both partners: one focusing on the typological analysis of interactive violence dynamics, and another employing specific concepts to describe the co-occurrence of violent behaviors (see Table S1, Supplemental Files).

Typological Perspectives on Violence Involving Both Partners: The Johnson’s Typology

A significant proportion of studies (n = 40) conceptualized violence involving both partners through frameworks that, while clearly describing this dynamic, did not use consensual or systematic terminology to explicitly classify this pattern of violence. Within this approach, Johnson’s (1995, 2006) typology emerged as the predominant framework for describing violence dynamics. The definition most frequently applied and detailed by these articles was “Situational Couple Violence” (SCV; 40 papers; see Table S1). SCV is described as cases in which both partners resort to violent behaviors during marital conflicts, in the absence of an ongoing pattern of coercive control characteristics of Intimate Terrorism (IT). Such violence typically results from situational stressors, emotional escalation, or ineffective conflict management and is generally of lower severity and episodic in nature.

Although SCV emerges as the most frequently applied category among studies referencing Johnson's typology, other categories, particularly Violent Resistance (VR) and Mutual Violent Control (MVC), also meaningfully contribute to the conceptualization of violence involving both partners. VR appeared in 16 papers (see Table S1) and refers to instances in which one partner uses physical aggression as a reactive strategy in response to a sustained pattern of coercive control, typically exercised by the other partner. In such cases, violence is not symmetrical in either motivation or intent. Rather, it emerges as defensive behavior, often employed to de-escalate immediate threats or assert physical autonomy in the context of persistent victimization. For instance, several articles describe patterns in which the respondent identifies both victimization and perpetration but situates their own use of violence within narratives of fear, retaliation, or protection (Stith et al., 2011; Straus & Gozjolko, 2014). When the relational context is not considered, particularly in studies relying solely on self-report instruments such as CTS, this type of asymmetrical dynamic risk is misclassified as mutual or BV. MVC, in contrast, refers to relationships in which both partners simultaneously exert control over one another through violence. In this configuration, each individual engages in coercive strategies, often within a volatile and escalating relational pattern. A few studies included in this review (15) report findings consistent with this conceptualization, namely, the presence of reciprocal controlling behaviors, high levels of aggression, and a chronic escalation of conflict (see Table S1).

The absence of explicit terminology in these 40 studies did not imply ambiguity in the dynamics described. On the contrary, there is a clear characterization of the overlap of roles between victim and perpetrator, with statistical coding of participants as “victim-perpetrators” or “bidirectional involvement” being common, even without the use of a specific theoretical term (Arteaga et al., 2015; Behnken et al., 2018).

Specific Concepts and Their Distinctions for the Co-Occurrence of Violence

Four concepts emerged across the literature that explicitly label violence involving both partners: bidirectional violence, mutual violence, victim–perpetrator overlap, and reciprocal violence (see Table 2). Each term reflects a specific analytical focus, temporal frame, and relational dynamics, although these distinctions are not consistently applied across studies.

The most frequently employed term was “bidirectional violence” (BV), identified in 26 studies (see Table S1), typically defined as the co-occurrence of violent behaviors perpetrated by both members of an intimate couple, with each partner, at different times or episodes, acting as both perpetrators and victims. This type of violence may occur in a single incident or throughout the relationship, but it does not necessarily imply symmetry in terms of frequency, severity, motivation, or impact (Arteaga et al., 2015; Capinha et al., 2022). Across studies, BV is generally used as a broad descriptive term, with relatively limited specification regarding interactional processes or contextual factors.

Secondly, “mutual violence” (MV) appeared in six studies (see Table S1), generally referring to the presence of violent acts by both partners, with each being both a victim and a perpetrator. Across the six studies that explicitly employed the term “mutual violence,” definitions often described the co-occurrence of violent acts by both partners, in some cases within the same incident or conflict episode. However, this characterization was not consistently specified across all studies. In several cases, definitions appeared to implicitly assume symmetry, without clearly specifying differences in severity, motivation, or impact between partners (Afifi et al., 2012; Chen & Chan, 2021). At the same time, other studies explicitly problematize assumptions of symmetry and emphasize the importance of relational dynamics (Dokkedahl & Elklit, 2019). In most cases, these studies did not include an explicit empirical assessment of symmetry beyond the co-occurrence of acts. Overall, mutuality is sometimes described in terms of event-level simultaneity, although this emphasis is not consistent across all uses of the term (Afifi et al., 2012; Ahmed et al., 2024). Importantly, this characterization reflects tendencies in how MV is defined in the reviewed literature, rather than a uniform or consistently operationalized construct across studies.

The term “victim-offender overlap” was used in six studies (see Table S1) and refers to the phenomenon in which an individual experiences both the roles of victimization and perpetration of IPV, either within the same relationship or across different relationships and time points. In the six studies using the concept of victim–perpetrator overlap, authors consistently defined it at the individual level, referring to persons who have both experienced victimization and perpetrated IPV. Some studies conceptualized this overlap as occurring sequentially over time (i.e., role shifts), whereas others explicitly allowed for the simultaneous occupation of victim and perpetrator roles (Jennings et al., 2012; Tillyer & Wright, 2013; Whitaker et al., 2007). It was often studied with broader psychosocial or criminological factors and has been associated with an increased frequency and severity of IPV (Jennings et al., 2011; Whitaker et al., 2007).

Finally, another term found was “reciprocal violence,” in four studies (see Table S1). In these studies, reciprocal violence was described as a patterned, interactional process in which violent acts are linked through retaliation, escalation, or emotional reactivity across multiple episodes (Bates, 2016; Langhinrichsen-Rohling et al., 2012; Pu et al., 2022). These definitions tend to emphasize relational interdependence between acts of violence, rather than simultaneity within a single incident. However, as with other concepts, the extent to which these interactional dynamics are empirically assessed varies across studies. This characterization reflects the definitional framing adopted by these authors, rather than a uniform empirical assessment of escalation processes.

Key Conceptual Distinctions

When examining explicitly defined concepts (BV, MV, reciprocal violence, and victim–perpetrator overlap), several key distinctions emerge, although these should be interpreted as general tendencies rather than strict or universally applied boundaries. First, the level of analysis varied. While BV, MV, and reciprocal violence generally refer to the dyadic or relational level (i.e., the couple; Babcock et al., 2019; Capaldi & Kim, 2007), victim–perpetrator overlap is primarily individual-centered, focusing on how one person may simultaneously or sequentially occupy both the roles of victim and perpetrator (Jennings et al., 2012; Muftić et al., 2012).

Second, these concepts differ in their temporal framing. MV is often described as temporally narrow, referring to violence occurring within the same incident or conflict episode; however, some studies use the term more broadly without explicitly restricting it to a single event (Afifi et al., 2012; Dokkedahl & Elklit, 2019). In contrast, BV and reciprocal violence may refer to patterns unfolding across multiple episodes (Crane et al., 2014; Pu et al., 2022; Whitaker et al., 2007). The victim–perpetrator overlap extends further, encompassing role shifts that occur over time and potentially across different relationships (Jennings et al., 2012; Tillyer & Wright, 2013). This temporal distinction aligns with the episodic nature of the previously detailed SCV, which often manifests in multiple instances rather than a single isolated event.

Third, the relational dynamics implied by each term vary in terms of complexity. BV indicates the presence of aggression from both partners but does not specify interaction or intent (Arteaga et al., 2015; Capaldi & Kim, 2007; Capinha et al., 2022). MV may emphasize co-occurrence, sometimes within the same incident, but does not consistently capture broader relational patterns (Ahmed et al., 2024). In contrast, reciprocal violence tends to emphasize an interactive and potentially escalating process in which each partner’s aggression contributes to a cycle of conflict (Pu et al., 2022; Whitaker et al., 2007). This makes reciprocal violence the most relational dynamic of the four (Langhinrichsen-Rohling, 2010). This focus on the interdependence of violent acts is a fundamental characteristic of SCV dynamics (Ali et al., 2016; Johnson, 2006).

Fourth, assumptions regarding symmetry vary across concepts and studies. MV tends to be assumed to be symmetrical, although this assumption is increasingly being questioned in different studies in terms of severity, motivation, and impact (Ahmed et al., 2024; Capaldi et al., 2018; Dokkedahl & Elklit, 2019; Muftić et al., 2012). Other concepts, especially reciprocal violence and victim–perpetrator overlap, more explicitly recognize the possibility of asymmetry between partners. This critical distinction resonates strongly with the findings in the previous section, where studies employing Johnson's typology consistently highlight the frequent asymmetry of violence in SCV, even when both partners are involved (Bates, 2016; Johnson, 2006).

Finally, the use of different terms across disciplines and research designs differs. BV and MV are common in population-based IPV studies (Capinha et al., 2022; Cunha et al., 2023). Reciprocal violence is more prominent in relational or therapeutic research (Babcock et al., 2019; Pu et al., 2022), while victim–perpetrator overlap is often used in criminology, developmental studies, or trauma-informed approaches (Arteaga et al., 2015; Jennings et al., 2012; Tillyer & Wright, 2013).

Conceptual Debates and Controversies

The literature also highlights several conceptual and methodological challenges that continue to shape scholarly debates surrounding violence involving both partners in intimate relationships. One of the most salient issues is terminological inconsistency: terms such as BV, MV, reciprocal IPV, victim-perpetrator overlap, and gender symmetry are often used in overlapping or inconsistent ways, despite reflecting distinct theoretical frameworks and relational dynamics (Ahmed et al., 2024; Machado et al., 2024). This inconsistency is further exacerbated by a significant number of studies (n = 40) that describe violence involving both partners using typological frameworks, primarily Johnson’s, without explicitly employing these specific terms.

A second critical concern in many empirical studies is the limited assessment of the relational context. Several authors argue that any meaningful interpretation of interpersonal violence involving both partners must account for contextual variables, such as motivational intent, control dynamics, escalation patterns, and differential harm, without which classifications such as “mutual” or “bidirectional” risk becoming overly simplistic or even misleading (Graham et al., 2012; Stover et al., 2025). This emphasis on context aligns with the detailed descriptions found in studies that use Johnson's typology, which consistently highlights the importance of understanding underlying motivations (e.g., reactive vs. coercive control) and the often asymmetrical impact of violent acts, moving beyond a simple count of perpetration to truly grasp the relational dynamics. Furthermore, high rates of violence involving both partners have been documented in specific populations, including individuals with substance use disorders, mental health conditions, LGBTQIA+ identities, and families with young children (Cunha et al., 2023; Holmes et al., 2019; Suomi et al., 2019).

Challenges in the Operationalization of Violence Involving Both Partners

Beyond conceptual heterogeneity, the operationalization of violence involving both partners also presents significant challenges that directly impact the comparability and synthesis of empirical findings. These challenges stem primarily from the diversity of measurement instruments and the methodological approaches used to identify and quantify these dynamics.

One of the main sources of inconsistency is the heterogeneity of measurement procedures. This review revealed that although the CTS and its various versions (e.g., CTS2) were predominant, the diversity of instruments used leads to the phenomenon being measured inconsistently. This lack of standardization may contribute to situations in which studies that, on the surface, appear to address the same phenomenon capture distinct facets of violence, making the direct comparison of prevalence rates or dynamics particularly challenging.

Additionally, methodological approaches to identifying bidirectionality also vary considerably, affecting its operationalization. Some studies simply report the co-occurrence of perpetration and victimization based on individual reports without a more in-depth dyadic analysis (Afifi et al., 2012; Ahmed et al., 2024). Others use more sophisticated dyadic designs (Pu et al., 2022; Whitaker et al., 2007), but even within these, the criteria for classifying instances as “bidirectional” or “mutual” diverge. For instance, while some studies may consider “bidirectional” any report of aggression by both partners (Cunha et al., 2023; Santiago et al., 2025), others require a certain frequency or a specific reactive pattern (Dokkedahl & Elklit, 2019; Frías, 2017).

Finally, the design type has an impact on operationalization. While cross-sectional studies are useful for providing prevalence data, they limit the ability to capture the dynamic and cyclical nature inherent in concepts such as reciprocal violence. In contrast, longitudinal studies allow for a better understanding of temporal patterns and role shifts (e.g., victim-perpetrator overlap over time), thereby shaping how bidirectionality is operationalized to reflect these temporal aspects. Similarly, qualitative studies often provide rich, nuanced descriptions of relational dynamics that are difficult to capture through quantitative operationalization, highlighting the gap between conceptual understanding and the empirical measurement of the phenomenon.

Discussion

This scoping review provides a synthesis of how violence involving both partners in intimate relationships has been conceptually defined and operationalized across the literature. While interest in BV has grown markedly since the early 2000s (Cunha et al., 2025; Lysova et al., 2024; Machado et al., 2024), the field remains conceptually and methodologically heterogeneous. The findings highlight persistent inconsistencies in terminology, definitional boundaries, and measurement approaches, which complicate the accumulation and integration of knowledge.

One of the most salient issues is terminological inconsistency. Our review identified two predominant conceptual approaches for describing the bidirectional patterns of IPV. The first centers on a typological analysis of interactive violence dynamics, primarily based on Johnson’s (1995) typology. This approach, while providing rich descriptions of dynamics such as the SCV, contributes to the absence of a unified nomenclature. Within Johnson’s framework, other subtypes, such as VR and MVC, demonstrated that violence involving both partners cannot be adequately captured through a singular lens or by relying solely on the frequency of violent acts. Instead, these typologies highlighted the importance of assessing motivational intent, control dynamics, and relational history to accurately classify and understand the patterns of aggression within intimate partnerships.

The second approach draws on four specific terms—BV, MV, victim–perpetrator overlap, and reciprocal violence—to describe the co-occurrence of violent behaviors. Although these terms are often used interchangeably or inconsistently, the present review suggests that they reflect different conceptual emphases and analytical lenses rather than clearly delineated or empirically distinct forms of violence (Ahmed et al., 2024; Machado et al., 2024), differing, for example, in their level of analysis (e.g., person-centered, couple), temporal framing, relational dynamics, interactional processes, symmetry, and contextual emphasis. Critically, the present review does not assume that categories such as MV, BV, and reciprocal violence represent empirically distinct phenomena. Rather, these labels often refer to highly similar operational definitions, with differences reflecting authors’ terminological preferences and theoretical framing rather than substantively different forms of violence. The lack of definitional clarity complicates both the synthesis of empirical findings and accumulation of theory-driven knowledge.

Rather than advancing a novel typology or privileging a specific theoretical perspective, the present review contributes an integrative, minimal definition of bidirectional IPV grounded in the common elements observed across existing conceptualizations. This definition is intended to reduce terminological ambiguity while preserving theoretical flexibility for future research and practice. Therefore, drawing on the findings of this review, we propose the following working definition of bidirectional IPV intended to capture the core elements consistently present across existing conceptualizations. BV is defined as a pattern of IPV in which both members of an intimate partnership engage in perpetration and victimization of aggressive behaviors (e.g., physical, verbal, emotional, sexual, or controlling acts) within the same relationship, regardless of whether such behaviors occur symmetrically or simultaneously. This definition deliberately avoids assumptions about symmetry, simultaneity, or equivalence between partners. Moreover, this definition is not intended to imply moral, motivational, or functional equivalence between partners. As highlighted in prior research (Babcock et al., 2019; Langhinrichsen-Rohling et al., 2012), the presence of violence by both partners does not necessarily imply symmetry in intent, impact, or responsibility. Rather, the proposed definition delineates a minimal conceptual core that can accommodate heterogeneity in motivations, severity, and relational dynamics observed across studies. Within this broad definitional framework, several authors—including those adopting systemic or relational perspectives—interpret how BV unfolds in practice, highlighting patterns of escalation, interactional processes, and contextual factors such as emotional regulation and communication difficulties (Ahmed et al., 2024; Conroy, Griffin, et al., 2024). However, these interpretative frameworks should be understood as theoretical lenses applied to bidirectional IPV, rather than as defining features of the construct itself.

Adopting a systemic interpretation of BV can enable more context-sensitive interventions, focused on transforming relational patterns rather than assigning individual blame. Within this framework, the couple is understood as a dynamic and mutually influential system in which violence is both a symptom and a maladaptive mechanism of interaction. Crucially, this perspective represents one possible lens through which BV may be understood, rather than a defining feature of the construct itself. These conceptual and methodological challenges have significant real-world implications for research, practice, and policy, and are further explored below (Tables 3 and 4).

Implications for Research and Prevalence Estimates

Terminological and operational inconsistencies directly compromise the ability to conduct robust research and to generate reliable prevalence estimates. The diversity of measurement instruments, particularly the varied use of the CTS and its subscales as well as different recall periods and administration methods, limits the comparability of findings across studies (Dokkedahl & Elklit, 2019; Langhinrichsen-Rohling et al., 2012). This lack of standardization leads to a fragmented evidence base in which findings cannot be easily aggregated or meta-analyzed, hindering the cumulative advancement of knowledge (Ahmed et al., 2024). Consequently, the reported prevalence rates of BV may vary widely depending on the specific operational definition and methodology employed (Cunha et al., 2025; Frías, 2017), making it difficult to establish accurate epidemiological trends and identify at-risk populations.

Implications for Reporting and Intervention

The confusion surrounding BV also affects how individuals report their experiences, and how professionals intervene. For victims of VR who have used violence in self-defense, a lack of distinction between their actions and their abusers can create a barrier to seeking help, as they may fear being labeled as perpetrators (Babcock et al., 2019; Bates, 2016). From an intervention standpoint, the systemic definition proposed in this paper—one that moves beyond a simplistic victim-perpetrator dichotomy—is crucial for tailoring responses (Cunha et al., 2023). A systemic approach encourages practitioners to focus less on unilateral blame and more on the dysfunctional relational patterns and emotional dysregulation that fuel the cycle of violence without creating moral equivalence between fundamentally different behaviors (Hardesty et al., 2024).

Implications for Surveillance Systems, Prevention, and Policy

At a broader level, the lack of a consistent definition affects surveillance systems, prevention strategies, and policies. Public health and criminal justice systems rely on clear definitions to track trends, allocate resources, and design effective preventive campaigns. An oversimplified definition of BV can obscure the nature of violence, potentially underestimating its severity or misallocating resources away from the most vulnerable populations (Johnson, 2006; Langhinrichsen-Rohling, 2010). A policy framework that fails to distinguish between different types of violence involving both partners may implement “one-size-fits-all” interventions that may be less effective or insufficiently targeted. This review advocates a policy shift toward a more nuanced understanding of IPV, one that recognizes the systemic and relational factors at play, to develop targeted and effective prevention and intervention strategies (Lysova et al., 2024).

Limitations and Future Directions

While this review offers a comprehensive synthesis of how violence involving both partners is conceptualized in the literature, several limitations must be acknowledged. These limitations concern not only the methodological design of the review but also broader epistemological and empirical challenges within the field.

First, the scope of this review was limited to articles published in English, Spanish, or Portuguese, which may have led to the exclusion of relevant studies published in other languages or cultural contexts. Furthermore, the exclusion of gray literature, such as doctoral theses, technical reports, or institutional documents, may have precluded the inclusion of emerging or practice-based insights into dyadic violence. A second limitation is that many studies rely on individual self-reports and cross-sectional designs, limiting their capacity to capture relational dynamics, escalation patterns, and shifts in perpetrator-victim roles over time (Dokkedahl & Elklit, 2019; Leonard et al., 2014). Without dyadic data or longitudinal follow-up, it becomes difficult to determine whether violence occurs simultaneously, sequentially, or in cycles of retaliation, or to assess contextual factors such as emotional dysregulation, coercive control, or trauma (Langhinrichsen-Rohling, 2010). A third limitation concerns the methodological quality of the included studies. Of the 82 manuscripts, only 13 were rated as high-quality, while 15 were rated as low-quality. The predominance of manuscripts with low or merely acceptable quality may undermine the reliability of the data and, consequently, the robustness of our findings.

Finally, certain epistemological tensions in the field may also contribute to the inconsistencies observed in the conceptualization of BV. Divergent theoretical frameworks, particularly feminist, systemic, and criminological approaches, frequently offer conflicting interpretations of similar empirical patterns. These tensions influence not only how violence is labeled (e.g., mutual, reciprocal, or bidirectional) but also what aspects are foregrounded or excluded in the analysis, such as structural gender dynamics versus relational or contextual factors. Addressing these epistemological divides is crucial for advancing a pluralistic and nuanced understanding of IPV involving both partners.

A further limitation concerns the limited representation of diverse populations in the reviewed literature. Notably, the literature reviewed rarely includes sexual and gender minority individuals, older adults, individuals with disabilities, or people in non-heteronormative or non-cohabiting relationships, limiting the applicability of current conceptualizations of bidirectional IPV across diverse populations.

Given these limitations, future research should address these gaps by adopting more nuanced and integrative approaches. In particular, the field would benefit from the development of standardized measurement tools that move beyond frequency counts to include motivational asymmetry, emotional regulation, and the relational context. Longitudinal and dyadic designs are also essential for tracing the evolution of roles within violent dynamics and distinguishing reactive behaviors from patterns of coercion.

Conclusion

This review identified significant conceptual inconsistencies in the definition and operationalization of BV in intimate relationships. By systematically synthesizing studies, this study highlights the need for a more integrated and context-sensitive understanding of BV. The proposed systemic definition advances this field by providing a minimal conceptual core, while systemic theory offers a compelling lens to interpret BV as a dynamic and relational phenomenon. Such an approach challenges the notion of moral and functional symmetry, advocating instead for context-sensitive analysis that considers motivational asymmetries, power dynamics, and the emotional intensity underpinning violent episodes. Recognizing that BV frequently arises in relationships characterized by dependency, unresolved trauma, and poor conflict resolution skills, systemic theory provides a compelling lens through which to design both research and intervention. This conceptual shift has the potential to improve research precision, guide tailored interventions, and inform policies that better reflect the complexity of violent relational dynamics.