M. P. Johnson’s control-based typology of violence was first introduced to the scholarly intimate partner violence (IPV) literature in 1995, with the aim of bringing closure to the longstanding debate between family violence and feminist scholars regarding the nature of IPV in general and the role of gender in IPV in particular. Johnson (1995) posited:

This debate has been structured as an argument about the nature of family violence, with both sets of scholars overlooking the possibility that there may be two distinct forms of partner violence, one relatively gender balanced (and tapped by the survey research methodology of the family violence tradition), the other involving men’s terroristic attacks on their female partnersFootnote 1 (and tapped by the research with shelter populations and criminal justice and divorce court data that dominates the work in the feminist tradition). (p. 285)

In other words, Johnson’s typology was developed with the underlying assumption that IPV is not a unitary phenomenon, and with the assertion that research rooted in different theoretical perspectives – and different methods – ultimately captured two distinct types of IPV. In particular, Johnson proposed that large-sample survey data most often gathered by family violence scholars largely captures situational couple violence, and the agency sample data most often utilized by feminist scholars primarily captures intimate terrorism.Footnote 2 According to Johnson, situational couple violence is best understood as violence that occurs in the context of specific conflicts and is perpetrated by women and men at comparable rates, whereas intimate terrorism refers to the ongoing, systematic use of coercive control, paired with physical violence used in the service of coercive control, and is disproportionately perpetrated by men against women.Footnote 3

Despite Johnson’s (1995) initial position – and several attempts thereafter to clarify (e.g., Johnson, 2005, 2006, 2011, 2017; Johnson & Ferraro, 2000; Kelly & Johnson, 2008) – that there are two distinct types of IPV characterized by notable gender differences, the scholarly debate he aimed to resolve persists today. According to Johnson (2017), there are three major issues underlying the ongoing gender debate despite empirical evidence that gender is a key contributing factor to making distinctions between types of violence and their effects: (i) “ideologically motivated misrepresentations and errors of logic,” wherein scholars argue that women are at least as violent as men in intimate relationships (p. 157), (ii) “neglecting (either through ignorance, laziness, or ideological bias) to make distinctions between types of IPV,” occurring when research findings are affected by sampling biases but such biases are not acknowledged (p. 157), and (iii) using cluster analysis to determine whether there is a pattern of coercively controlling behavior – information necessary for determining whether the IPV is situational couple violence or intimate terrorism – in a sample that represents only one type of IPV.Footnote 4 Anderson (2008) identified another issue plaguing the literature on Johnson’s typology of violence: Johnson’s work has been broadly cited in the research literature and used to inform practice, jumping to real world application without sufficient assessment of the validity of the typology. Together, these critiques of the IPV literature suggest that Johnson’s typology has potentially been both misrepresented and prematurely applied without sufficient empirical testing to support, refute, or otherwise modify the typology.

Despite the ongoing impasse in the scholarly literature regarding gender and IPV, alongside the potentially premature acceptance of Johnson’s typology by some scholars, there has been no systematic evaluation of the literature on Johnson’s typology until this special issue. To address this gap, Conroy et al. (2022) systematically reviewed and evaluated the quality of peer-reviewed, empirical literature on Johnson’s typology and synthesized findings across studies for the first time, focusing solely on literature deemed to accurately represent Johnson’s typology and to have tested, directly or indirectly, at least one of Johnson’s core assumptions about IPV. By focusing on a subset of articles deemed to have accurately represented Johnson’s typology, conceptually and operationally, Conroy et al. (2022) and the studies reviewed therein provide valuable insight into the process of hypothesis testing and theory development that Johnson (2008) himself has called for.

However, in the original version of Conroy (2022) submitted for review, we had extracted data and reviewed results for empirical studies testing Johnson’s typology, including studies that were deemed to be misuses of Johnson’s typology. In doing so, our first round of analysis was largely inconclusive; findings were mixed, with relatively equal amounts of support for and against the typology, and difficult to reconcile due to methodological inconsistencies and errors. Based on reviewer feedback and our focus on the empirical testing of Johnson’s assumptions to determine its validity and utility, we ultimately removed the misuses of Johnson’s typology from the systematic review. Doing so allowed us to provide a fair assessment of Johnson’s typology, as the included studies were deemed to accurately represent (conceptually) and appropriately apply (methodologically) Johnson’s typology. Importantly, but perhaps not surprisingly, by focusing on this subset of 44 articles, the widespread scholarly debate about the role of gender and distinctions between types of violence was not represented. Instead, we found that Johnson’s theoretical assumptions about the effects of sampling and measurement, the role of gender, and distinctions between situational couple violence and intimate terrorism were largely supported, with few exceptions, in this subset of literature.

Yet, in limiting our assessment of Johnson’s typology to this high-quality subset of literature, we had to forego our secondary aim of better understanding the characteristics of research contributing to the continued impasse among IPV scholars regarding the issue of gender and IPV today. As such, in the current paper, we aim to provide greater clarity about how and why the gender debate persists today, despite Johnson’s ongoing refinement and clarification of the typology and empirical support of its validity (see Conroy et al., 2022). To that end, we systematically review the conceptual misrepresentations and methodological misapplications of Johnson’s typology in the empirical IPV literature. Tudge et al. (2009, 2016) provided a similar type of review, evaluating the uses and misuses of Bronfenbrenner’s bioecological theory of development and the implications for theory development itself. Like Tudge et al., we believe researchers should:

Make explicit their theoretical foundations, accurately represent that theory, and use appropriate methods to test it, so that their work can aid understanding of the theory either by providing appropriate supporting evidence or by calling into question some or all of the theory. (Tudge et al., 2016)

Failure to do so ultimately precludes the ability to understand and sufficiently test the validity and utility of a theory and is a threat to research integrity (Tudge et al., 2009, 2016). Therefore, we review the misrepresentations and misapplications of Johnson’s typology in empirical studies testing Johnson’s typology and its assumptions. In doing so, we hope to bring greater conceptual clarity to Johnson’s typology and identify key issues affecting progress toward validating and/or refuting the typology after more than 25 years of research after its initial development.

Johnson’s Typology: A Brief History

Johnson’s typology was introduced to the scholarly literature in 1995, as part of a critical review of IPV research. We have organized this section by Johnson’s four main arguments presented in his original paper and as summarized in his future work (e.g., Johnson, 2006, p. 1004).

Intimate Partner Violence is Not a Unitary Phenomenon

Johnson (1995) first argued that two distinct forms of IPV existed, but family violence and feminist scholars had long failed to make such distinctions. More specifically, Johnson proposed the following types of violence: Patriarchal terrorism, wherein the use of violence is motivated by the perpetrator’s desire to exert general control over their partner (i.e., coercive control), and common couple violence, characterized by the use of violence in the context of specific conflicts without the intent to control one’s partner (i.e., without the use of coercive control). Johnson (1995; Johnson & Ferraro, 2000) explains that coercive control is about the systematic use of a variety of control tactics against one’s partner with the goal of generally controlling that partner. Coercive control is distinct from attempts to gain momentary, situation-specific control like winning an argument, as well as other, more isolated use of emotional abuse. In future iterations of the typology, Johnson modified the terminology from patriarchal terrorism to intimate terrorism (e.g., Johnson, 2000; Johnson & Ferraro, 2000) or coercive controlling violence (e.g., Hardesty et al., 2015; Kelly & Johnson, 2008), and from common couple violence to situational couple violence (e.g., Johnson, 2005; Johnson & Leone, 2005). Johnson (2008) explains that although the definitions of these types have remained the same, he modified the language for the following reasons:

I soon abandoned [patriarchal terrorism] because it begs the question of men’s and women’s relative involvement in this form of controlling violence. It also implies that all such intimate terrorism is somehow rooted in patriarchal structures, traditions, or attitudes. . . However, it is clear that there are women who are intimate terrorists in both heterosexual and same-sex relationships. . . Furthermore, it is not necessarily the case that all intimate terrorism, even men’s, is rooted in patriarchal ideals and structures. With regard to “common couple violence,” I abandoned it in favor of “situational couple violence” because the former terminology implies to some readers that I feel that such violence is acceptable. I also prefer the new terminology because it more clearly identifies the roots of this violence in the situated escalation of conflict. (p. 115)

With regard to frequency, severity, and escalation of violence, Johnson has also suggested that situational couple violence typically includes “‘minor’ forms of violence, and more rarely escalating into serious, sometimes even life-threatening, forms of violence” (1995, p. 285). Although sometimes misrepresented in the literature, Johnson never suggested that situational couple violence is trivial nor necessarily infrequent, nor has he suggested that high rates of violence inevitably indicate intimate terrorism. Johnson (2000; Johnson & Ferraro, 2000) clarifies that intimate terrorism is rooted in a general pattern of coercive control and may include relatively low levels of violence, and situational couple violence, although not embedded in a context of control, may nonetheless be so severe as to result in homicide. Overall, frequency, severity, and escalation of violence are not sufficient for making distinctions between types. Instead, information about motives of both partners is necessary for classifying violence types (Johnson & Ferraro, 2000).

Johnson (2000; Johnson & Ferraro, 2000) later expanded his typology to include the following types of violence: violent resistance, violence enacted by the victim of intimate terrorism against their perpetrator, whether as a self-protective measure or retaliation, and mutual violent control, where both partners perpetrate violence and coercive control against one another (i.e., two intimate terrorists).

Different Sampling Strategies Capture Different Types of IPV

Johnson (1995) further suggested that sampling biases in research by family violence and feminist scholars ultimately capture different types of IPV. More specifically, the feminist approach to studying IPV (i.e., qualitative data gathered from agencies or clinical settings) overwhelmingly captures intimate terrorism, whereas the family violence approach to studying IPV (i.e., large, community sample survey research) largely captures situational couple violence. According to Johnson (1995, 2006; Kelly & Johnson, 2008), perpetrators and victims of intimate terrorism, unlike those in relationships characterized by situational couple violence, would be highly unlikely to participate in survey research given the social undesirability of perpetrating such abuse and the terror associated with experiencing it, respectively. On the other hand, victims of situational couple violence would be less likely to want or need services provided by shelters, emergency rooms, and courts and, unlike victims of intimate terrorism, are likely underrepresented in such samples. As Johnson explains “both major sampling designs in domestic violence research are seriously biased, and those biases account for the fact that each side of the debate has been able to marshal empirical evidence to support its position” (Johnson, 2006, p. 44), thus contributing to the persistent impasse in the literature. As a resolution, Johnson suggests that scholars on both sides are not wholly right nor wrong; rather, they are capturing two different types of violence that are qualitatively different and calls for such distinctions to be made moving forward. However, Johnson has never stated that sample type alone can be used as a proxy for classifying types of violence. It is possible for community sampling strategies to capture intimate terrorism, particularly when ex-partners and past relationships are reported on (Johnson et al., 2014), and for situational couple violence victims to utilize resources more typically used by intimate terrorism victims, particularly if they experience severe, frequent, and/or otherwise fear-inducing acts of violence (Johnson, 2008).

Intimate Partner Violence is Gendered, Particularly in Intimate Terrorism

Johnson (2008) also suggested that gender is the best predictor of IPV in heterosexual relationships. More specifically, Johnson (e.g., 1995, 2006, 2008, 2010, 2017) has long argued that intimate terrorism is perpetrated overwhelmingly by men against women, whereas situational couple violence is perpetrated at more comparable rates by men and women. Similarly, because women are disproportionately the victims of intimate terrorism, they also make up the majority of people using violent resistance against their perpetrators. Unlike situational couple violence, which is initiated by men and women at comparable rates and may be bidirectional in a given relationship, Johnson estimates mutual violent control to be exceedingly rare.

Johnson (1995) initially conceptualized the term intimate terrorism as being “terrorized by systematic male violence enacted in the service of patriarchal control” (p. 283) and “terroristic control of wives by their husbands” (p. 284) involving violence paired with the systematic use of coercive control. In 2010, he also asserted, “The feminists are right. Gender is central to the analysis of intimate partner violence, and the coercive controlling violence that most people associate with ‘domestic violence’ is indeed perpetrated primarily by men against their female partners” (p. 217). Nonetheless, Johnson has also acknowledged that empirical evidence suggests that intimate terrorism is not exclusive to heterosexual relationships but also exists in gay and lesbian relationships (Johnson, 1995, 2008; Johnson & Ferraro, 2000). Johnson (1995) also acknowledged his own work with male victims using shelter services after experiencing intimate terrorism perpetrated by female partners. Kelly and Johnson (2008) also state, “It is important not to ignore female-perpetrated Coercive Controlling Violence. Although it may represent only one-seventh or so of such violence... it is necessary that we recognize it for what it is when we make decisions about interventions” (p. 482). Even in Johnson’s earliest iteration of the typology, he did not claim that intimate terrorism was solely perpetrated by men against women, even in heterosexual relationships. Instead, Johnson argued that the prevalence of intimate terrorism was less likely to be gender symmetrical as compared to that of situational couple violence, with men typically perpetrating intimate terrorism at higher rates than women in heterosexual relationships. Relatedly, Johnson suggested that intimate terrorism is much less likely to be bidirectional (i.e., mutual violence control) than situational couple violence.

Moreover, despite a feminist perspective guiding his own IPV research, Johnson and Ferraro (2000) argued that “a full understanding of partner violence must go beyond this feminist analysis to ask questions about the role of control in the use of violence that may have little to do either with patriarchal traditions and structures or with individual patriarchal motives” (p. 995). Johnson’s position has always been that coercive control is at the heart of making distinctions and identifying intimate terrorism and is sometimes rooted in – but cannot be solely explained by – systemic and/or individual patriarchal motives. Similarly, he has long critiqued the limits of feminist scholarship on IPV (e.g., Johnson, 1995, 2000, 2005) alongside that of family violence scholars, each for their contributions to the notion that IPV is a unitary phenomenon.

Johnson (2008, 2010) also posited that dominant notions of gender symmetry or mutuality of violence are flawed given the focus on mere prevalence. For example, Johnson (e.g., 2000, 2008, 2010; Johnson & Leone, 2005) provides empirical evidence suggesting that women are more likely than men to experience negative outcomes regardless of IPV type, and that intimate terrorism (with women as the primary victims) is more likely to result in significantly more negative psychological and physical health outcomes than situational couple violence, particularly because intimate terrorism is more likely to be frequent, severe, and escalate over time. In this way, the role of gender in IPV cannot be solely understood by comparing rates of men and women’s violence (Anderson, 2005, 2008).

“The Task Ahead: Developing a Theoretical Framework”

Johnson (1995, 2006) has acknowledged that the initial argument and supporting evidence in his 1995 literature review was, to a degree, speculative because the distinctions in IPV that he was proposing had not yet been empirically tested. Because the proposed distinctions between types of IPV rely on the underlying motivations, i.e., coercive/high control or low/no control, Johnson (1995, 2008; Johnson & Ferraro, 2000) has always called on others to produce research to empirically investigate such motives and test the distinctions he proposed in the name of theory development. He has also empirically tested it himself (e.g., Hardesty et al., 2015; Johnson, 2006; Johnson & Leone, 2005; Johnson et al., 2014; Leone et al., 2004, 2007; McKay et al., 2020), and his 2008 book has provided the most comprehensive description of and methodological recommendations for empirically testing the typology. To do so fairly and effectively, however, necessitates accurate understanding and representation of the typology, paired with theoretically fitting measures and analysis.

Current Study

In this systematic review, we aim to provide insight to the ongoing impasse between feminist and family violence scholars regarding the validity and utility of Johnson’s typology. To that end, we systematically review and critically analyze the conceptual misrepresentations and methodological misapplications of Johnson’s typology of violence in the empirical IPV research. To ensure the systematic and comprehensive nature of our search, the fourth author, an Information and Instruction Librarian, assisted in the development of inclusion criteria, search method and database selection, and study selection process, and he completed the database searches.

In our analysis, we identify the primary aims of included studies and review errors in conceptual definitions of key concepts in Johnson’s typology, misrepresentations of Johnson’s core assumptions, and errors in measurement and/or analysis. We focus on these types of misuses because testing a theory, model, or typology necessitates a clear linkage between one’s theoretical framework, methods, and analyses (Tudge et al., 2009).Footnote 5 The process of theory-building also requires that the theory is explicitly articulated as the central framework and its conceptual definitions and assumptions are accurately represented. Failure to do so runs the risk of perpetuating misunderstanding of the theory and precludes the opportunity to fairly evaluate – and therefore support, refute, or otherwise modify – the theory (Tudge et al., 2009, 2016).

Method

Our review process was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Page et al., 2021).

Inclusion Criteria

Our inclusion criteria were: (a) peer-reviewed, empirical, and original research, (b) studies with the aim of testing Johnson’s typology and/or assumptions, (c) published 1995 or later, and (d) qualitative and quantitative studies written in English.

Search Method and Study Selection Process

We searched the following databases: PsycInfo, PsycArticles, Pubmed, Web of Science, Genderwatch, Sociological Abstracts, and Social Services Abstracts. We searched all fields using terminology from Johnson’s typology, including: intimate terrorism, coercive controlling violence*, situational couple violence*, common couple violence*, violent resistance, mutual violent control, and patriarchal terrorism, excluding records published before 1995 and non-English records. The final database search was conducted on March 31, 2021, yielding 881 total records. The first and third authors reviewed the records for deduplication in Zotero. After deduplication, 502 unique records remained.

Next, the first and fourth authors utilized Rayyan for preliminary screening of titles and abstracts to exclude studies using the following criteria: studies unrelated to and/or did not implement Johnson’s typology (e.g., ‘violent resistance’ to political violence; n = 203); empirical research not published in scholarly, peer-reviewed journals (conference abstracts, theses, dissertations; n = 27); papers published in peer-reviewed journals that did not constitute original, empirical research (commentaries, literature reviews, personal narratives, book reviews, erratum; n = 75); and books and book chapters (n = 17). We made preliminary screening decisions independently and agreed on all exclusion decisions. In cases where decisions were unclear based on the title and abstract, the first and second author reviewed the full-text record using this preliminary exclusion criteria.Footnote 6 This excluded 322 records.

The first and second author then conducted full-text reviews of the remaining 180 articles to determine which studies met our criteria of empirical research articles focused on testing Johnson’s typology and/or related theoretical assumptions. If there was disagreement or clarification was needed to determine inclusion or exclusion, the third author reviewed the record, and we engaged in discussion until consensus was reached. We excluded empirical studies that: referenced Johnson’s typology but did not focus on testing Johnson’s typology or related assumptionsFootnote 7 (n = 93); non-empirical (clinical and case law) applications of Johnson’s typology (n = 5); and studies with alternative uses of Johnson’s typologyFootnote 8 (n = 4). This resulted in 78 articles remaining.

Next, the first, second, and third authors independently reviewed and classified the remaining articles into accurate uses and misuses of the typology. Papers were deemed misuses if: IPV types were conceptually defined and/or operationally classified by something other than coercive control (e.g., violence severity or frequency or sampling strategy used) and/or the classification process was not described; if the conceptual and/or operational definitions of a violent relationship included nonviolent behaviors to classify violence victimization/perpetration [given Johnson’s (2008) emphasis on physical aggression]; the conceptualization and/or operationalization of coercive control did not align with Johnson’s conceptual definition; and/or core assumptions were otherwise misrepresented (i.e., inaccurate statements about the typology that could be refuted by Johnson’s published work). We deemed 44 studies to be appropriate uses of Johnson’s typology; these studies are reviewed in Conroy et al. (2022). This resulted in 34 recordsFootnote 9 meeting the inclusion criteria for this systematic review of misuses of Johnson’s typology.

Data Extraction and Classification of Misuses

The first and second authors again independently reviewed the 34 articles meeting inclusion criteria for the purpose of data extraction. We used a shared online spreadsheet for data extraction, wherein we recorded study aims, types of misrepresentations/ misapplications of the typology, and detailed notes about our decision-making processes. In instances of disagreement or when clarification was needed to determine classification decisions, the third author reviewed the record, and we engaged in discussion until consensus was reached.

We classified studies as misuses of Johnson’s typology as follows: Misrepresentation includes errors in (i) conceptual definitions of key concepts, including types of violence [situational couple violence (SCV), intimate terrorism (IT), violent resistance (VR), mutual violent control (MVC)] and/or coercive control, and (ii) Johnson’s related assumptions about the effects of sampling, gender, and the nature of violence types, as discussed above. We also note when researchers use outdated or modified terminology and/or reference only Johnson’s, 1995 article after new terminology, more recent empirical data, or other clarification by Johnson are available. Misapplication refers to errors in measurement and/or analyses that often paralleled the conceptual misrepresentations, including but not limited to classifying violence types based on sample type; severity, frequency, or escalation of violence, and/or something other than coercive control level.

Analysis and Discussion

Of the 34 included studies, one (2.9%) article was published between 1995 and 2000, five (14.7%) were published between 2001–2005, eight (23.5%) between 2006–2010, 10 (29.4%) between 2011 to 2015, and 11 (32.4%) were published in 2016 or later. Articles by Hines and Douglas (n = 6); Bates et al., Bates & Graham-Kevan, and Graham-Kevan and Archer (n = 3); Emery et al. (n = 2); Friend et al. (n = 2); Rosen et al. (n = 2); and Straus and Gozjolko (n = 2) account for 50% of the articles. The remaining articles were written by unique authors. More than half of the studies (n = 19; 55.9%) include data collected in the United States.Footnote 10

Overall, each of the 34 included articles had at least one conceptual misrepresentation of Johnson’s typology, all but one of which also had at least one methodological misapplication (n = 33; 97.1%). Of note, the one article without any identifiable methodological misapplications (Nybergh et al., 2016) had only one, relatively minor conceptual misrepresentation and is arguably an otherwise useful contribution to the empirical literature testing the utility and validity of Johnson’s typology.Footnote 11 The nature and implications of misuses are described below. While the examples discussed below are not exhaustive, the aims and specific types of misuses for each study are identified in Table 1.Footnote 12

Table 1 Summary of included studies, aims, and types of misrepresentation/misapplication (n = 34)
Full size table

Conceptual Misrepresentations

Errors in Defining Coercive Control & Types of IPV

The most common conceptual misrepresentations related to definitions of violence types and the role of coercive control in the typology. In particular, authors most often erroneously suggested that distinctions between SCV and IT are based on severity, e.g., that IT could be classified by the presence of severe violence and/or that SCV included only low-level violence (n = 22; 64.7%). Similarly, others erroneously provided explanations of the typology as though distinctions are made by the frequency (n = 14, 41.2%) and/or escalation (n = 5, 14.7%) of violence. However, these statements do not align with Johnson’s conceptual distinction between types, which relies solely on the pattern (or absence) of coercive control rather than the characteristics of physical violence. Additionally, descriptions did not always align with Johnson’s intention for the typology to represent forms of physical violence, with seven (20.6%) articles having suggested that nonviolent, aggressive acts absent the use of physical violence can be used to classify IPV types. For example, Esquivel Santoveña et al. (2016) described SCV perpetrators as using “low levels of any form of coercive control and/or physical violence,” which suggests that nonviolent, noncontrolling relationships can be classified as SCV, and that IT, VR, and MVC include coercive control “with/or in the absence of physical violence” (p. 103; emphasis added). Although Johnson (2008) acknowledges the importance of nonviolent coercive control as an exceptionally harmful pattern of abusive behavior and potential precursor to IT, he is clear that the typology is one of physical violence.

Moreover, eight articles (23.5%) erroneously used terms like ‘psychological aggression’ or ‘emotional abuse’ interchangeably with or in lieu of ‘coercive control,’ despite such acts not rising to the pervasive, threatening nature of coercive control. For example, Hines and Douglas (2011a) explained that “IT involves emotional abuse” (p. 113), without explaining the distinctions between emotional abuse and the tactics used to establish a general pattern of control that is central to coercive control as a concept. In some cases, coercive control was not referred to at all (e.g., Burgess-Proctor, 2003). Although psychological aggression and other emotional abuse tactics are often part of the pattern of coercive control, these tactics alone are not enough to “regulate and dominate an intimate partner’s daily life and restrict personal liberties” as is the case with coercive control (Hardesty et al., 2015, p. 834). Moreover, failure to name or clearly and accurately define coercive control erroneously suggests to readers that any emotionally hurtful behavior paired with physical violence is sufficient for classifying a violent relationship as IT.

Not only do these errors misrepresent the typology, but theoretically and practically, inconsistency in defining the different types of IPV in Johnson’s typology – and in definitions of IPV more generally throughout the literature – make it difficult to parse out why rates and types of IPV differ across studies. Such inconsistencies also have practical implications, such as an inability to accurately determine the scope of the problem and allocate sufficient resources which, in turn, precludes the assessment of program efficacy for reducing IPV (Centers for Disease Control & Prevention, 2015).

Errors in Describing Assumptions of Sampling & Gender

Studies also contained misrepresentations of Johnson’s assumption regarding the effects of sampling on IPV types (n = 3; 8.8%). For example, Hines and Douglas (2019) claimed that Johnson asserts “that population-based samples capture only SCV, while agency samples capture only IT,” (p. 192). Similarly, Walby and Towers (2018) state that “Johnson claims surveys cannot capture the experiences of the most abused women” (p. 10). However, as noted above, Johnson asserts that there is inherent sampling bias for both community and agency samples, but he does not assert that they are mutually exclusive from capturing IT and SCV, respectively. Such errors fail to capture the nuance of Johnson’s position regarding sampling and the likelihood that SCV and IT will be captured by certain strategies. This misrepresentation may also result in some of the methodological misapplications discussed below, such as researchers using sample type to classify the type of violence rather than measures of coercive control.

Johnson’s assumptions related to gender and experiences of physical violence and coercive control are also misrepresented in 11 articles (32.4%). In some cases, researchers implied or explicitly stated that “Johnson asserts that IT... can be solely explained by patriarchal theory” (Hines & Douglas, 2019, p. 180) or is “the sole domain of men” (Hines & Douglas, 2010a, p. 287). Relatedly, Straus and Gozjolko (2014) stated that Johnson’s typology is not gender inclusive, particularly that “there is no category in the typology for IT solely by women” (p.51), and that women are only considered potential victims in IT relationships (Straus & Gozjolko, 2016). Olson (2002) similarly suggested that Johnson’s theoretical foundations for IT preclude men from being IT victims. However, Johnson (1995, 2000, 2008) has not suggested that patriarchal sociocultural traditions and individual motives wholly explain relationship control dynamics. Instead, he has called for more general theories, beyond a feminist analysis of control, to be tested in relation to typological differences, thus contradicting the claims that men, masculinity, and patriarchy are considered the sole root of IT. Moreover, no type of violence in the typology is solely reserved for men or for women; anyone can be classified as using SCV, IT, or VR based on the use of physical violence and the pattern of coercive control.

Terminological Errors

Nineteen (55.9%) articles included terminological errors. For example, researchers sometimes used outdated terminology, such as “patriarchal terrorism” (e.g., Burgess-Proctor, 2003) and/or “common couple violence” (e.g., Bates & Graham-Kevan, 2016), despite the intentional shifts to “intimate terrorism” (Johnson & Ferraro, 2000) and “situational couple violence” (Johnson & Leone, 2005; Leone et al., 2004), respectively. As discussed above, Johnson (2006, 2008) made these terminological changes to clarify his intentions for readers and researchers, as the terms had come to be misunderstood and misrepresented in the literature. Researchers also implemented other variations of the terminology or terms coined by other IPV researchers, including “situated couple violence” (Walby & Towers, 2018), “situated violence” (Pietri & Bonnet, 2017), “situationally violent” (Friend et al., 2011, 2017), “characterological violence” (for IT; Friend et al., 2011, 2017), “mutual violence” (for MVC; Yingling et al.,  2015); and “mutually controlling violence” (Esquivel Santoveña et al., 2016). For the purpose of testing a model or theory, precision in representing that model or theory is important, otherwise conceptual confusion may result.

Other Conceptual Misrepresentations

Twenty-two studies (62.9%) included several other misrepresentations of Johnson’s typology, assumptions underlying the typology, and his work more generally. For example, additional definitional errors included defining violent resistance as self-defense (e.g., Hines & Douglas, 2019; Ross & Babcock, 2009; Yingling et al., 2015) and having suggested that SCV, by definition, is mutual or bidirectional (e.g., Hines & Douglas, 2012). Ross and Babcock (2009) also erroneously stated that types of violence can be classified by level of symmetry (i.e., bidirectionality or similar levels of violence or control used), with SCV being described as the comparable use of violence and control by partners. However, SCV is the unidirectional or bidirectional use of violence absent the use of coercive control, and not all acts of violent resistance meet legal definitions of self-defense as it can also include premeditated acts of retaliation (Johnson, 2008; Johnson & Ferraro, 2000).

In other cases, authors paid little or no attention to defining coercive control, despite being a central element of the typology (e.g., Hines & Douglas, 2011a) or only briefly referenced the typology in the literature review, with little to no description of each type of violence (e.g., Melander, 2010). Similarly, some authors did not situate their results using Johnson’s terminology, instead referring more generally to violent relationships with high/low control, despite their stated aims of comparing rates of SCV to IT (e.g., Krantz & Vung, 2009; Simpson et al., 2007). Pietri and Bonnet (2017) had a single citation of Johnson’s, 1995 article in the literature review, with no other reference to Johnson’s work, despite patriarchal (intimate) terrorism and “situational violence” being explicitly incorporated in the hypotheses, analyses, results, and discussion of the study. In doing so, it is difficult to determine how findings from these studies support or refute Johnson’s typology.

Various authors also misrepresented Johnson’s work in other ways. For example, authors erroneously suggested that Johnson sees SCV as “of being of little social concern” (Bates et al., 2014, p. 44) and not “of any serious consequence” (Bates et al., 2016, p. 5), or that “intimate terrorism is more harmful” than SCV (Walby & Towers, 2018, p. 23). These claims misrepresent Johnson’s longstanding and more nuanced position that “all family violence is abhorrent, but not all family violence is the same” (1995, p. 293), and that SCV can be frequent and even lethal (Johnson, 2008; Johnson & Leone, 2005).

Authors also misrepresented Johnson’s position on men’s victimization and women’s perpetration. For example, Bates and Graham-Kevan (2016, p. 8) suggested that Johnson considers men’s IPV victimization to be trivial. However, such a claim is not an accurate representation of Johnson’s (2008) position that men are less likely than women in heterosexual relationships to be seriously injured or effectively terrorized. Some researchers also posited that women’s use of severe violence and/or coercive control indicate that Johnson’s theorized gender distinctions between types of violence are unsupported altogether (e.g., Bates & Graham-Kevan, 2016; Bates et al., 2014). However, Johnson (2008) does not argue that women do not use severe violence nor coercive control; instead, he argues that men are more likely to perpetrate IT in heterosexual relationships. Johnson further states that numerical symmetry based on prevalence alone is not sufficient for claiming gender symmetry in IPV. Rather, Johnson explains that IPV “(even situational couple violence) is not gender-symmetric,” when the nature of violence and its outcomes, such as injury and fear, are considered. That is, numbers alone cannot capture the qualitatively different experiences of men and women experiencing IPV.

Researchers also attempted to discredit Johnson’s work by suggesting that his work is rife with intentional bias, in pursuit of a particular feminist agenda. For example, Hines and Douglas (2010b) asserted that Johnson “never called for” research on men’s victimization, and that he “preselected samples that conformed to his ideas that IT perpetrated by women was rare and was, therefore, able to conclude again that it could be explained exclusively by patriarchal theory” (p. 39). Bates et al. (2014) similarly suggested that Johnson (1995) found gender differences in IPV due to “using samples selected for a high proportion of male-to-female aggression (e.g., women’s shelter samples)” and “a national violence against women survey that cannot be regarded as an unbiased sample of violence by both sexes” (p. 44). Moreover, Bates and colleagues implied that their findings of women using more physically aggressive and controlling behaviors compared to men “are undoubtedly different to those that would be found in more ‘biased’ sample such as shelter or prison samples” (p. 53). By doing so, they suggest that Johnson’s sampling strategies were inherently –even intentionally – biased, yet fail to acknowledge the sampling bias inherent to the Western, undergraduate sample used in their own study. Hines and Douglas (2010a) similarly state that a major limitation to the typology is that Johnson “used only shelter samples of battered women, and men mandated into batterer treatment programs” in its development (p. 287). However, Johnson and colleagues had published findings in support of the typological differences between SCV and IT using data from mixed sample and general population surveys (e.g., Johnson, 2006; Johnson & Leone, 2005; Leone et al., 2004, 2007). These statements also fail to acknowledge that even Johnson’s earliest published works explicitly articulate both the need for studying men’s victimization and IPV in same-sex relationships, as well as the bias inherent to clinical/agency and community samples, the latter of which is a point clearly articulated as a central element of Johnson’s argument that distinct types of IPV are being captured by different samples.

Overall, the insufficient specificity, clarity, and/or accuracy in representing the key concepts and core assumptions of Johnson’s typology (or any theory) undermines the integrity of research and the collective understanding of IPV for scholars, professionals, and students alike. With each of these conceptual misrepresentations, limited explanations, and reliance on older iterations of Johnson’s typology, researchers perpetuate misrepresentations and a limited or flawed understanding of the typology. This, in addition to insufficiently situating one’s research in the context of the extant literature on Johnson’s typology, impedes progress toward conceptually and methodologically sound theory building. In turn, conceptual confusion may result in errors in measurement and analysis, which ultimately preclude the ability to test the validity of the typology. Finally, assertions that Johnson minimizes any type of IPV or men’s victimization as well as claims that Johnson’s research is not sound due to ideological bias can easily be refuted by referencing Johnson’s published work.

Methodological Misapplications

In all but one article (n = 33; 97.1%), conceptual errors were paralleled by methodological misapplications, including errors in measurement, analysis, and classification of IPV.

Classifying Types of IPV by Severity, Frequency, & Escalation

The most common misapplication, found in 24 studies (70.6%), was the erroneous classification of low/high control categories and/or SCV/IT types based on something other than a coercive control measure. One type of classification error included classifying violence types based on the severity (n = 14; 41.2%), frequency (n = 10; 29.4%), and/or escalation (n = 1; 2.9%) of violence. For example, Drumm et al. (2009) regressed items associated with IT (control, emotional abuse, fear, PTSD symptoms) on measures of escalating violence, sexual violence, and potentially lethal violence to identify patterns of IT. Although Frías (2022) distinguished between levels of coercive control, they ultimately classified types of violence by severity, with moderate/severe violence + no coercive control or moderate violence + moderate coercive control classified as SCV, and any violence + high coercive control or severe violence + moderate coercive control classified as IT. In this case, SCV and IT are characterized by the same level of coercive control, which does not align with Johnson’s conceptualization of SCV and IT being distinct phenomenon based on low and high control, respectively.

Classifying Types of IPV Based on Sample

Others classified types of violence based on the type of sample used (n = 9; 26.5%). For example, Graham-Kevan and Archer (2003) presupposed that violent male prisoners and an agency sample of women in refuges captured IT, whereas student and nonviolent prisoners were assumed to represent SCV. Using sample type as a proxy for coercive control, they tested Johnson’s assumptions about distinctions between SCV and IT by comparing the samples on coercive control, physical violence, and outcome measures. Hines and Douglas (2010b) similarly assumed that a help-seeking sample of men represented IT victims and that a community sample of men represented SCV victims, without classifying IPV type by a measure of coercive control. This data set and sample-derived classification of IT served as the basis for several other studies by Hines and Douglas (2010a, 2011a, b, 2012). Other researchers also presupposed that their national survey samples represented SCV without measuring coercive control, thus precluding the ability to classify SCV and IT types in line with Johnson's prescribed methods (e.g., Grandin & Lupri, 1997; Olson, 2002; Sillito, 2012).

Absence of Coercive Control Measure

Relatedly, coercive control was not measured in some studies (e.g., Friend et al., 2017; Kamimura et al., 2017; Olson, 2002). For example, Burgess-Proctor’s study (2003) classified violence as IT if participants had been subject to one or more of the following acts by their partners: beaten or choked, forced into sexual activity, and/or threatened with or had weapons used; those answering ‘no’ to those items were classified as SCV. In other instances, the authors suggested that they assessed coercive control, but the measures used were not sound operationalizations of the concept. For example, Krantz and Vung (2009) coded exposure to a single controlling behavior as constituting coercive control, which does not capture the pervasive pattern of control necessary for being coercive. Straus and Gozjolko (2014, 2016) utilized the Psychological Aggression subscale of the CTS2 (Straus at el., 1996) as a proxy measure for coercive control. However, the items of the Psychological Aggression subscale do not capture the same tactics of isolation, monitoring, and efforts at general control that more fully represent the pervasiveness of IT. In each of these misrepresentations, types of violence were classified using a strategy other than what is prescribed for the typology, i.e., based on the level of coercive control in physically violent relationships, regardless of the nature of the violent acts.

Inclusion of Nonviolent Aggressive Behaviors

Researchers also incorporated measures of nonviolent aggressive behaviors (e.g., psychological aggression) in classifying a relationship as violent (n = 11, 32.4%; e.g., Próspero, 2008; Rosen et al., 2005). For example, Friend et al. (2011) used a screening tool that captures “relationship behaviors found to be highly correlated with incidence of IPV” (p. 554) to classify violence types, even for participants who reported no physically violent behaviors on the CTS2 (Straus et al., 1996). In Emery et al.’s (2015, 2017) studies, intimate terrorism was conceptualized as synonymous with nonviolent coercive control and was not paired with the measure of physical violence in the analysis. In Olson’s (2002) study, nonviolent, minor acts of verbal aggression, were used to classify SCV.

Other Misapplications

Other methodological misapplications not captured by the abovementioned categories were present in 13 studies (38.2%). For example, Rosen et al. (2005)’s qualitative coding scheme excluded situationally violent relationships if participants expressed fear of injury because they interpreted fear to only exist in response to IT. However, SCV can also include severe and injurious violence (Johnson, 2008) and may result in fear of negative outcomes, particularly for female victims with violent male partners (Anderson, 2005). Moreover, Rosen et al. (2005) classified some experiences as VR although they determined there were no experiences of IT present in the sample, despite VR being defined by Johnson as an IT victim’s use of violence against their violent and controlling partner; i.e., VR cannot exist without IT.

Some of the included studies also used cluster analysis to distinguish between low and high levels of control in community samples, which was then used to classify violence as SCV, IT, VR, and/or MVC (e.g., Bates et al., 2014; Bates & Graham-Kevan, 2016; Próspero, 2008; Ross & Babcock, 2009). This error is identified as one of three major mistakes identified by Johnson (2008, 2017) that contributes to the ongoing gender debate in IPV research. In short, cluster analysis in such a sample is affected by the inherent sampling bias, where a “high control” group will be identified (and used to classify IT), even though the data likely represent little to no IT. In other words, the threshold for meeting the coercive control characteristics of IT is artificially lowered in such samples, resulting in the overidentification of IT. Without acknowledging such limitations, empirical evidence seemingly supports the notion that women and men perpetrate IT at comparable rates but is ultimately a false premise. In other cases, authors failed to report on or lacked clarity regarding their method for classifying types of IPV altogether (e.g., Melander, 2010; Pietri & Bonnet, 2017; Yingling et al., 2015).

Study Limitations and Contributions

Due to our focus on peer-reviewed, empirical research studies testing Johnson’s typology, this review does not capture the extent of the scholarly debate regarding gender and IPV, in general, nor how broadly Johnson’s typology has been utilized in the IPV literature. However, the early stages of our systematic review process ultimately resulted in examining over 200 additional records of IPV research integrating Johnson’s typology. In some cases, articles referenced Johnson’s work very briefly, such as a single citation without any explicit reference to the typology; brief reference to the typology as part of the broader IPV literature; or relying on the typology as a central framework for justifying the need for validated coercive control measures. Such uses may reflect Anderson’s (2008) concern that Johnson’s typology and position on gender and IPV have been accepted at face value, without sufficient empirical testing to validate his assumptions. Johnson’s work has also been the subject of many commentaries and literature reviews, often with the aim of arguing for or against the use of Johnson’s typology or gender theories of IPV more generally. Of note, many of the articles we initially reviewed in each such category also misrepresent Johnson’s typology in various ways, further contributing to conceptual confusion in the IPV literature beyond what is described above.

Overall, our review of the empirical IPV literature of direct and indirect tests of Johnson’s typology suggests that such research is characterized by a variety of conceptual misrepresentations and methodological misapplications. Among the included studies, the degree of error varies broadly, with some studies characterized by definitional errors that miss some of the nuance and specificity of Johnson’s typology, and other studies include a variety of misrepresentations of Johnson’s core assumptions and key concepts, alongside errors in measurement and analysis. In any case, such errors preclude fair evaluation of Johnson’s typology and, at times, arguably include ideologically motivated misrepresentations of Johnson’s work that contribute to the ongoing gender debate in IPV research.

Because the included studies were deemed misuses of Johnson’s typology, we do not systematically review their findings. However, the research reviewed herein ultimately included studies that provide support for and, as was the case for the majority of studies, challenges to Johnson’s typology. This contrasts with a corresponding systematic literature review of 44 studies deemed appropriate uses of Johnson’s typology (Conroy et al., 2022), where there was overwhelming support for Johnson’s typology, with very few exceptions. These preliminary findings suggest that a gendered perspective of IPV is more strongly supported, and that the great gender debate that has persisted may be due, at least in part, to the volume of research with flawed methodological foundations represented in the scholarly literature. Nonetheless, more methodologically sound research is needed to test specific assumptions underlying Johnson’s typology and to explore the practical implications of making distinctions between violence.

While it is beyond the scope of this review to evaluate the implications of the typology for practice and policy, failure to conduct conceptually and methodologically sound IPV research – when testing Johnson’s typology and in general – undermines research integrity and the process of theory-building. Our findings, alongside Conroy et al. (2022), suggest that 43.6% of empirical tests of the typology fall into this category. Furthermore, supporting, refuting, or otherwise modifying a theory based on inaccurate representations and inappropriate methods results in flawed evidence, which, in turn, has real-world implications that may ultimately undermine the safety of survivors. Considering the role that theory and research play in service delivery, including but not limited to informing the development of risk assessment protocol and intervention programming, as well as policy development, researchers have a responsibility to test Johnson’s typology accurately and in good faith. Moreover, we encourage students, researchers, and peer-reviewers to consider the misrepresentations and misapplications discussed above when engaging with any scholarly discussion about Johnson’s typology of violence, in order to discern between fair, valid assessments and flawed, arguably invalid assessments of the typology.

Conclusion

To gain insight into the longstanding debate about the role of gender in IPV that persists in the scholarly literature today, we systematically and critically reviewed peer-reviewed, empirical research studies testing the theoretical assumptions of and types of violence identified in Johnson’s typology of violence, a typology proposed as a means of resolving the impasse between IPV scholars with different theoretical orientations. Our review resulted in the identification of 34 empirical research studies being classified as misuses of Johnson’s typology, including conceptual misrepresentations and methodological misapplications. Interestingly, the majority of the studies included in this review had findings that challenged the validity of Johnson’s typology, unlike the 44 studies reviewed in Conroy et al. (2022) that were deemed theoretically and methodologically sound and provided overwhelming support for the typology. These findings provide some insight to why the ideological impasse about the role of gender in IPV persists between feminist and family violence scholars despite the introduction of Johnson’s control-based typology over 25 years ago.

In order to evaluate the utility and validity of Johnson’s typology, direct tests of Johnson’s typology and related assumptions are necessary. However, across the 34 included studies, we identified varying degrees of two types of misuses: conceptual inaccuracy (misrepresentations) and errors in operationalization and analysis (misapplications). Such errors ultimately preclude fair testing of the typology and hinder potential theoretical advances. In turn, as Johnson (2005) noted, “errors of theory lead to potentially life-threatening errors of intervention strategy and general policy” (p. 1129). As such, IPV researchers intending to test Johnson’s typology have a responsibility to do so in good faith, with accurate representations and the appropriate methods necessary for supporting, refuting, or otherwise modifying the theory.