Abstract
Intimate partner violence (IPV), including intimate partner homicide (IPH) and femicide, raises issues for general theories of crime, such as control and opportunity theories, that see close relationships among friends and family as barriers to interpersonal crimes. Crime-specific studies of both correlates and trends in IPV, including recent interrupted trend studies that examine the effects of COVID restrictions, often test opportunity theories absent considerations of theoretically driven images of actors. Review of empirical research on IPV and IPH reveals strong compatibility between the predictions of modern control theory and consistent findings from trend data. Barriers to understanding of the explanatory power of general theories of crime (including, for example, control theories and feminist perspectives) in contemporary research include use of poor definitions of intimacy, misspecification of age effects, failure to consider the versatility of offending behavior, neglecting the importance of trends in analogous behaviors, neglecting the role of situational factors in violence, and the limitations in the measurement of repetitive victimization. Theories such as routine activity and situational crime prevention that fail to explicitly include characteristics of actors can go only a limited way in providing meaningful policy. Research supports the potential policy effects of investments in early childhood and attention to situational barriers (including limitations on alcohol use and firearm availability) to reduce IPV. Although modern control theory is used to illustrate these issues, other general theories, like feminist theories, can make similar arguments.
Keywords
Introduction
Research focused on trends in violence against women, including intimate partner violence (IPV), presents major theoretical and policy issues for criminology. A now substantial body of contemporary research, including both national and cross-national studies, focuses on trends and correlates for IPV (including intimate partner homicide [IPH] and femicide), scholarship that has great relevance for both causal theory and methods of intervention and prevention. At the policy level, this concern is reflected in the identification of violence against women as a global public health issue and as a violation of women’s human rights (Caman et al., 2017). Both specialized and general theories have been proposed to help explain IPV (cf. Campbell et al., 2007; Chapple & Hope, 2003; R. Felson & Lane, 2012; Finkel et al., 2009; Frías, 2023; H. Johnson et al., 2019; I. D. Johnson & Lewis, 2023; Nofziger, 2009; Payne et al., 2010; Piquero et al., 2021). Recent studies examining violence during the pandemic using before/after or interrupted time series designs have received considerable attention both in the academic literature and in media accounts (Aebi et al., 2021; Lopez & Rosenfeld, 2021; Nivette, Ribeaud, et al., 2021; Piquero et al., 2021).
IPV, particularly IPH, presents issues for many putative general crime theories, as they struggle to explain violence among persons who have ongoing interpersonal relationships. This is certainly the case for both control theories (e.g., Gottfredson & Hirschi, 1990, 2020; Hirschi, 1969) and lifestyle perspectives (Gottfredson, 1981, 2021a; Hindelang et al., 1978), both of which see attachments to significant others as creating informal and personal controls (social and self-controls) that inhibit causing harm to others. Along with routine activity theory (RAT), these general theories include the idea of opportunity as a causal element (Gottfredson, 2018; Gottfredson & Hirschi, 2020). The concept of opportunity, as deployed by these theories, includes consideration of the features of settings that are either conducive to or which impede the presence of monitors that serve as barriers to the use of force or fraud used to pursue immediate goals. Included is the idea that settings away from the restraints provided by close friends and family are more likely to involve aggressive or violent acts. Although IPH are a relatively rare type of crime, these theoretical images in some ways mirror public or media interest in violence and homicide among intimate partners (see Fairbairn & Dawson, 2013; Richards et al., 2011). Such instances are often sensationalized in the media because they raise the question “how could anyone do something like that to someone they love?” It is highly problematic for both theory and public sentiment to understand or explain serious violence among close friends and family.
In contemporary criminology, general theories like routine activity, opportunity, and lifestyle theories (Clarke, 2018; Cohen & Felson, 1979; Hindelang et al., 1978) have been actively researched as explanations for trends in crime and violence (e.g., M. Felson et al., 2020; Nivette, Ribeaud, et al., 2021; Peitzmeier et al., 2022) and, most recently, for understanding the effects of COVID lifestyle restrictions on IPV (Aebi et al., 2021; Piquero et al., 2021). In fact, studying crime and violence trends interrupted by the COVID responses has been said to present an especially good opportunity to use a quasi-experimental method to study the role of routine activities in the causation of victimization (M. Felson et al., 2020; Piquero et al., 2021).
It is not surprising that these perspectives have received considerable research attention associated with COVID restrictions. There have been important advances in recent years in the prediction and explanation for crime patterns and trends using variation in opportunity as a casual mechanism for many forms of crime, including interpersonal violence (see Clarke, 2018; Gottfredson & Hirschi, 2020 for reviews). Opportunity theories are often selected in this research because of a preference for focusing on specific types of crime or violence, as opposed to a focus on general violence or general crime rates. Studies of specific types of violence, such as IPV, are frequently undertaken with the view, sometimes implicit and other times explicit, that general theories or general prevention mechanisms are less useful, or are simply wrong, when it comes to explanation and for development of practical methods of reducing levels of harm caused by such violence. It may be argued that differences between IPV and other types of violence outweigh the similarities in producing useful causal explanations and for the design of useful prevention or intervention methods. But commonalities among types of crimes may also be very useful for identifying common causes, causes that may be useful for policy. As such, this article explores both substantive and methodological issues raised by these assumptions, principally from the point of view of one type of general theory, modern control theory (MCT) (Gottfredson, 2021b; Gottfredson & Hirschi, 1990, 2020). Our focus is the study of corelates and trends in IPV, including IPH, over time, with an aim of inferring recommendations about intervention and prevention. The logic and analytic perspective adopted here are meant to be illustrative rather than limited to only control theories, as many of these issues pertain to the applicability of a range of general theories that may be applied to IPV, a point we support with a feminist framework.
Scope: IPV, Femicide, and Crime Specific Theory
Before we begin to explore the meaning of the findings for IPV research and the methods used to determine them, it is important to consider briefly common definitional issues and terminology and how these pertain to explanations offered by general theory. Although there is much scholarship on intimate partner violence (IPV) and on its most serious outcome, IPH, there are important differences in the definition and scope of the activities that are studied. In general, IPV refers to abusive behavior, both coercive and physical, that occurs by one or both intimate partners. The literature includes research on IPV among persons related, well known, or previously known to one another, and is sometimes restricted by measures assumed to describe intimacy (e.g., marriage or cohabitation; the term “boyfriend,” and so forth). IPH refers to the murder of one’s current or former partner and either explicitly or implicitly views the person who murders the other within the relationship as the primary aggressor, meaning self-defense or self-protection is not considered. Femicide most often refers to the murder of women by their male partners within an intimate relationship. 1
In IPV research, sometimes, only homicide or femicide is selected for study, frequently on measurement grounds, under the assumption that reliance on homicide data limits some major nonreporting biases in official data, compared with other offenses. Both survey and official record data have been the basis for trend studies, yielding a large body of research with considerable complexity and variability. Also, it is important to note that research has used vastly differing age-ranges in the study of IPV, sometimes including (or focusing on) adolescents, other times restricting study to data including only adults, or, even studying only older adults with extensive prior records. As will be clear, these preferences and associated data limitations have considerable theoretical and policy implications.
Variation exists among studies not only in the acts under study, but also in the categorization of acts by the relationship or previous relationship between the victim and offender. In some research, definitions include only instances with specific motives; in other research, any offense committed by particular victim/offender relationships are included. For example, using data on offenses reported to the police, the National Institute of Justice defines intimate relationships in victimization surveys as those involving current or former spouses, boyfriends, or girlfriends; similarly, the U.S. census bureau defines intimate relationship violence to includes rape, sexual assault, robbery, aggravated assault, and simple assault committed by an offender who is the victim’s current or former spouse, boyfriend, or girlfriend. These classification and data issues exist globally as well. There are many studies focusing specifically on femicide, but differences in how researchers use the term. Popularized by Radford and Russell (1992) femicide is defined as “the misogynist killing of women by men” (p. 3). Contemporary usage of the term emphasizes that many female victims of homicide are perpetrated by their male partners and point to a general tolerance for gender violence by the state (Frías, 2023).
In contrast, the term intimate partner violence is often used to encompass a wide variety of offenses and abuses, including those with differing specific motives. As Aebi et al. (2021, p. 61) points out violence against women has led to an increase not only in the terms used to refer to the murder of a woman, but also in the definitions of these terms. In the case of femicide, these range from etymological interpretations—all murders in which the victim is a woman—to definitions that require that a current or recent male partner is the perpetrator, continuing through different kinds of combinations of the relationship between the perpetrator and the victim.
Expanded definitions include the “killing of a female family member by a male family member, such as cases of gendered dishonor upon the family” (Messerschmidt, 2017, p. 76). As they also point out, this definitional or typological diversity among data can affect data comparability in significant ways.
Such considerations led Aebi et al. (2021) to focus attention on homicide with female victims, to allow for cross-societal comparability. Absent mechanisms to permit conceptual consistency and comparable data, this stance is clearly appropriate. Moreover, using a general definition for victimization and for homicide without restriction to specific motivation is consistent with the logic of general theories, such as MCT, that seek first to explain variation in crime and victimization under the assumption that all criminal motivations have something in common, such as, in the case of MCT, perceived momentary self-interest.
Of course, not all feminist theories have the same emphases with respect to explanatory motivations for IPH. In a systematic review, Graham et al. (2022) identified feminist theories that address IPH, including several that use abstract concepts similar to other general theories, clearly applicable to a wide variety of offenses and settings (e.g., theory of intersectionality, ameliorative hypothesis, exposure reduction hypothesis, and economic marginalization hypothesis). 2 As another example, general strain theories could enumerate many types of strain or frustration, which could “motivate” specific acts (see Piquero et al., 2021 for a lengthy list of potential strains during COVID lockdowns, as examples). While specific offenses may have different “motives” according to each of these general theories (jealousy, power, lack of fear of the criminal justice system, place-bound frustrations, as examples), general theories seek a level of abstraction that allows explanation for a wide range of offenses in addition to IPV.
Crime specific theories, however, have trouble when otherwise similar offenses have markedly different “motives” and may struggle to explain the enormous variation in motives for offenses. In fact, a crime can be found (or imagined) for nearly every specific potential motive that can be enumerated. For MCT, and other general theories, to the extent that a specific form of crime shares correlates or trends similarly with other offenses, the distinctions between types of crime may be less useful to consider than what they have in common, both for theory and for public policy.
Brief Review of Some Basic Research Findings for IPV and IPH Pertinent to All General Theories
Many IPV scholars focus on official statistics for homicide because among personal crimes it is the most likely to come to the attention of authorities and to be recorded. A focus on IPH in the United States begins by the observation that “the criminology literature correctly refers to homicide as a rare event and that among these, IPH homicide, which accounts for 7% of all homicides, is very rare” (Roberts, 2009, p. 68). Even though this observation must be guided by the fact that IPV and IPH are known to be substantially under reported in both official and survey data, nevertheless the general frequency of IPH creates problems for theory and for multivariate statistical analysis. Of course, we agree with Roberts (2009), that “this level of rarity does not negate the importance of understanding as much as possible about IP homicide to develop and apply policy alternatives focused on eliminating it altogether” (p. 68). But the ability to predict the relative frequency of an offense is a first order problem for theoretical adequacy. Control theories have focused on this problem, by emphasizing conformity rather than deviance and by utilizing the idea of probabilistic propensities (rather than determinants) associated with decision-making about which people differ. Thus, even those with relatively low self-control and relatively weak social bonds are not committing crimes and violence most of the time, consistent with the findings that very harmful offending is relatively infrequent. Of course, some individuals engage in crimes and violence even as they have relatively high social and self-control.
It is also known that the vast majority of cases of homicide have male victims and male offenders (estimates vary between 75 and 90). Also, as Caman et al. (2017) note, IPH can be best “characterized by gender asymmetry: while female victims are more likely to be victimized by an intimate partner or family member, male victims are more likely to be victimized by acquaintances or strangers” (p. 14). Taylor and Jasinski (2011) note that about one-fourth of murders committed annually are perpetrated against women and that according to the most recent Bureau of Justice Statistics figures, about one third of female homicide victims are killed by an intimate partner. In fact, Campbell et al. (2007) show that in police data women are murdered by intimate partners (married and nonmarried) or former partners, much more often (about 9 times) than by strangers. These data show that 30% of murdered American women (42% of those with a known perpetrator) are killed by an intimate partner (e.g., husband, ex-husband, or boyfriend), compared with 5.5% of men killed by an intimate partner (Campbell et al., 2007, p. 246; see Fox and Zawitz, 2007 for trends in U.S. Bureau of Justice Statistics reports). Thus, the overwhelming male participation as offenders in homicide, including but not restricted to women victims, is an important feature also of IPH.
Globally, women share the circumstance that violence against women is normalized and major barriers exist to investigating femicides, to developing prevention strategies, and to advocating for improved policies (Smith, 2018). A history of prior violence characterizes many IPV and IPH incidents. IPH is deeply gendered and very often takes place after substantial, ongoing violence by men in the relationship (Taylor & Jasinski, 2011). They show this finding persists over lengthy periods of time; research on IPH during the past 30 years has demonstrated that the vast majority of IPHs (70% of cases where the female partner is killed, 75% of cases where the male partner is killed) are preceded by known intimate partner violence (IPV) against the female partner (citations omitted) “. . . making IPV the most important risk factor for [IPH]” (Taylor & Jasinski, 2011, p. 247). In the United States, as many as two-thirds to three-quarters of women killed by partners were physically abused before their deaths by the same partner who killed them (Campbell et al., 2007). Costa et al. (2015) undertook an analysis of 25 prospective longitudinal studies of IPV, finding that prior family violence, (including child and adolescent abuse experiences), child and adolescent problem behaviors, including prior violence, are consistent predictors of IPV. Spencer and Stith’s (2020) meta-analysis also identified several previous types of IPV (e.g., threats, nonfatal strangulation, forced sex, or stalking) as well as substance abuse (which includes both drug and alcohol abuse) as correlates for IPH. These findings of repeated use of violence, even of extreme violence, against a partner, and the circumstance of a strong relation between the prior misconduct of the offender and IPV and IPH stands as critical facts of great relevance for theoretical explanation (see also Roberts’ review, 2009). The substantial association between ongoing or prior abuse and violence in IPV and IPH raise a significant conceptual question: how can the attachment between partners be considered “intimate,” given continuing use of physical harm?
Researchers have also noted a relatively consistent association across many studies between IP homicide involving husbands and wives with a history of estrangement (Campbell et al., 2007). Catalano (2012) presents typical data from the NCVS on IPV, defining intimate relationships as involving current or former spouses or boyfriends, and finds that both females and males who were separated or divorced faced the greatest risk of nonfatal intimate partner violence while persons who were married or widowed reported the lowest risk of violence. This is one fact among others that seriously questions many of the commonly used definitions of “intimacy” on the research on IPV and femicide—although this term may categorize a relationship—or, importantly, a former relationship—separation, estrangement, divorce, and the like would not suggest the type of emotional attachment referenced by MCT—rather it signals the opposite. In addition to these commonly found correlates there are of course many other important empirical findings from the extensive body of research on IPV. Other commonly cited findings for IPV include gun use, prior threats to kill and threats with a weapon, stepchild in the home if a female victim and the presence of alcohol (Roberts, 2009; Spencer & Stith, 2020). Any theory purporting to offer valid explanations for IPV should surely be expected to provide plausible accounts for these major patterns. Before examining the trend data with a similar objective, we can look first to the interpretation of MCT vis a vis these important consistent correlates of IPV.
MCT and IPV
The general theory of crime presented by Gottfredson and Hirschi in 1990 has been extensively researched and the assumptions underlying it long discussed in the literature. In 2020, Gottfredson and Hirschi adopted the term “Modern Control Theory” as a descriptor for their general theory of crime and made explicit the links between social and self-control theories (see Gottfredson & Hirschi, 2020; Gottfredson, 2021a, 2021b). They have depicted the essential elements of their theory as including (a) a disciplinary-free assumption of human nature; (b) a focus on harms to self and others (as opposed to standard definitions of “crimes”; (c) casual influence of both age and settings for crimes; (d) causal influence of both self-control and social bonds; and (e) a public policy focus on childhood, prevention, and the limits of criminal sanctions (see Gotfredson, 2021a, 2021b; Gottfredson & Hirschi, 2020b for elaboration).
What, then, does MCT assume about IPV? The following several general inferences seem reasonable: (a) IPV is decidedly within the intended scope of the theory, since it fits the definition of “crime” used by MCT; (b) MCT assumes that it can “explain” major empirical facts about the distribution of crime, both over time and cross-sectionally—not, of course, providing the “sole causes,” but providing substantively important and internally consistent explanations for each; (c) MCT clearly predicts that IPV is more likely when individuals are in association with others who have relatively low levels of self and social control and correspondingly less likely, the stronger the affectional bond among people and higher levels of self-control; (d) the theory predicts that, generally, victimization is more likely perpetrated by individuals who also have undertaken other, analogous problem behaviors, indicative of lower self-control, certainly including, but not restricted to, other prior offenses and prior violence; (e) MCT predicts that many of the events labeled “intimate partner violence” are not actually among intimates, as defined by attachment and other social bonds; (f) MCT assumes age effects for IPV are similar to other forms of interpersonal violence and risky behaviors—that is, to have peak rates in the late adolescence and early adulthood and to decline significantly over time with age, when implicit controls for age are not included in the definition of the offenses as is sometimes the case when researchers rely only on police or court data; (g) MCT assumes significant “situational” or “opportunity” effects, central to causation, can be identified for IPV (and potentially controlled); (h) MCT assumes that there will be both general causes (distal) and specific causes (proximate) for IPV (see, generally, Gottfredson & Hirschi, 2020; Gottfredson, 2021a, 2021b).
For MCT, the neglected harmful consequences of behaviors for the offender include both immediate harm to self (emotional or physical injury), and more remote consequences, such as criminal justice sanctions, loss of respect, and loss of affection. Bullying (see, Nofziger, 2001), repeated use of threats and force, and placing a host of self-interested motivations above the interests of the partner are more often in evidence for persons with relatively low self and social controls (Gottfredson & Hirschi, 1990, 2020). Thus, for MCT, cessation of arguments, envy, revenge, jealousy, hate, lack of attention, pursuit of sexual relations, bullying and dominance, resentment, and so forth are all proximate motives for IPV as each can be seen in significant ways as pursuit of narrow self-interest. A considerable body of contemporary research supports these general ideas (see Gottfredson & Hirschi, 2020; Gottfredson, 2021a, 2021b for reviews of relevant research; see Moffitt et al., 2011; Vazsonyi et al., 2017 Moffitt et al., 2013 for general reviews and meta-analysis of expectations from the theory).
For those with low self and social controls, the use of substances may further reduce inhibition; weapons may seem to assist in achieving short term ends but can themselves increase the violence of specific incidents when available. In general, individuals with lower self-control seek vulnerable targets, targets that either do not or cannot impede immediate goals (Gottfredson & Hirschi, 1990). Factors outside of the scope of the theory, of course, may also have explanatory value, but larger, first-order effects are argued to be consistent with the theory whatever the time, people and harm under consideration and models that test other specific causes are said to be mis-specified absent a consideration for the known effects of MCT. It stresses that social bonds (especially attachment to others and belief in the moral validity of laws) and self-control (the tendency to attend, or not, to the harmful consequences of acts), age, and opportunities (characteristics of settings) to be the most important first-order explanations for variability in crime and victimization and as a result the plausible foci for effective interventions for policy. Because the theory does not depend on legal definitions of crimes, the versatility of behaviors causing harm to others (and long-term harm to self) are included in the scope of the theory. Thus, trends in criminal violence are predicted to follow trends in other interpersonal and property crimes, and to follow trends in other problems behaviors, such as many accidents, drug and alcohol use, school problems, and the like.
When applied to the facts of IPV, including IPH, the expectations of MCT seem generally to be well met. There is much agreement in the literature that femicide offenders tend (not in every case, but tend) to have substantial prior records, not only for violence (but including violence), to be “generalists,” in the sense of having records of harm to others in a variety of ways, to have poor social relations generally, to have patterns of school and employment difficulties, to have histories of substance abuse problems, and to be repeat offenders within the relationship ending in femicide. Nevertheless, the research literature presents several important challenges for the application of MCT and other general theories, the resolution of which could enhance our understanding of the causes and prevention and of IPV.
The explanation of these patterns by general theories is not, of course, the sole province of control theories. For example, there are a number of overlaps between a feminist framework and MCT. MCT recognizes that the socialization process, including gender socialization, has substantial impact on behavior. Similarly, under the idea of hegemonic masculinity, this socialization process may lead some individuals to adopt patriarchal values and renders certain characteristics as appropriate for men or women. Similar to the relationship between early development of social bonds and self-control, how one views and engages in intimacy begins in the family and converges later in their intimate relationships. According to hooks (2004) boys are taught that “. . . there is only one emotion that patriarchy values when expressed by men; that emotion is anger” (p. 18). For men, this includes the externalization of emotions (i.e., anger and aggression) and placing high value on risk taking behavior, both indicators of low self-control. There is also strong evidence of misogyny in other extreme forms of violence, such as mass shootings, drawing further connections between self-control and the social context under which it operates. Finally, MCT and a feminist framework share an understanding of intimacy as a complex, and dynamic, rather than static, process.
Some Key Issues in the Explanation and Prevention of Intimate Partner Violence From the Point of View of General Theory
MCT Has a Different Concept of “Intimacy” Then Do Some Definitions Used in Research on IPV
In a discussion of the limitations of RAT in accounting for the sometimes-conflicting results of research for IPV and femicides due to COVID restrictions, Aebi et al. (2021) importantly focus critical attention on the lack of an image of the actor in RAT and on the problems prior research has had in definitions of the critical idea of intimacy. As they indicate (2021): “. . . the definitions of femicide usually include two elements: (unbalanced) power and kinship. The latter implies some sort of affection, which should serve as a regulator of aggressive impulses. Nevertheless, affection does not play a major role in the criminological explanations of femicide.” (p. 631)
This attention to the meaning of intimacy is often absent in literature about both IPV and IPH, but it is central to MCT. Clearly, not all cohabitants or marriages, let alone previous such relationships, involve similar levels of intimacy. In control theory terms, “attachment is the affectional component of the social bond (caring . . . “about the wishes and expectations of other people”) (Hirschi, 1969, p. 18). There is ample evidence that the attachment bond is relatively weak in cases of intimate partner violence—repeated assaultive victimization, arguments, absences, estrangement and separation and other indicators of weak bonds are frequently discovered correlates among the participants. Direct measures of intimacy are typically not available in much of the IPV research. But, as indicated above, what evidence exists indicates weak social bonds. It may well be that a large part of the answer to the theoretical and commonly expressed public question “how could someone do this to someone they cared about” is that it is wrong to assume they did “care,” in the sense control theories use the term attachment. Having weaker bonds than most “intimate partners” they are less constrained about using threats and actual force, more likely to seek satisfaction of their own interests unfettered by consequences to others. The use of the label “intimate” is thus misleading, at least for theoretical purposes, and perhaps also for considerations of policy. On the other hand, continuing or increasing behavior that “works” to achieve momentary goals through coercive power is consistent with MCT expectations. Thus, there can be a degree of “normalization” of these behaviors within the relationship.
MCT Includes Age as a Major Factor in the Causation of Crime, Including Interpersonal Violence, and Cautions Against (for Theory and for Policy) Definitions of Crime or Samples of Offenders That Implicitly Control for Age
MCT views age as a causal variable for acts of interpersonal violence, and emphasizes the generally similar age distribution for all crimes, including IPV. In most crime data, violence offenses peak later than property, drug, and driving offenses, but show the rapid and monotonic declines with advancing age (see, Gottfredson & Hirschi, 2020, Ch.3 for a recent discussion of age effects in crime). It is important to note that many studies of IPV and IPH incorporate age into the definitions of the dependent variables, via restrictions that involve studying persons who are married or cohabiting, or formerly so—all of which are clearly age-related categories. Sometimes only “adult” offenses are studied in official data, with the same result. This can yield the appearance that interpersonal violence among those known to one another has an older, sometimes much older, age profile than do other examples of low self- control. Yet, inclusion of directly analogous relationships at younger ages would likely yield a more normal age/crime relationship, masking the critically important fact that interpersonal violence tends (not exclusively, of course), like other risky behaviors, to disproportionally be a circumstance of youth. Because ages of adult jurisdiction vary over time and place, caution needs to be exercised in the study of age distributions in official data.
Consider the findings of Campbell et al. (2007), using NCVS data: Intimate partner violence rates differ greatly based on the age of the victim, consistent with other offenses. While the overall per capita rate of intimate partner violence against women was 5.8 victimizations per 1,000 in 1999, among females age 16-24, it was 15.6 per 1,000. The importance of this is clear: among other things, it has broad implications for both trend studies and for the importance of early intervention as a prevention strategy.
With respect to age effects on violent crime rates, recent empirical work substantiates the role of age cross nationally. Santos et al. (2019) note that “. . . age has been established as one of the best predictors of deviant and criminal activity” (p. 37) and provide an extensive analysis of homicide data internationally that shows that many countries have experienced the substantial increases followed by substantial declines in homicide rates over the modern period.
As Santos et al. (2019), report, “[e]vidence presented in the current study supports the conclusion that the ageing of populations globally has been a driver of the international homicide decline, and that the percentage of a country’s population aged 15 to 29 is strongly associated with global homicide trends since the 1960s” (p. 18).
The generality of the effect leads them away from explanations centered on domestic policies and social events within individual countries (similar to Gottfredson, 2021b; Tonry, 2014) to a broader global phenomenon; in the end, they argue that changes in country-level age structure is a “key factor” in understanding global homicide trends over the past six decades, a pattern relevant to IPV, as IPV also shows a strong relationship to age (e.g., Roberts, 2009).
The Importance of the Ideas of “Versatility” or “Generality” and of Self-Control in MCT Leads to Caution About Inferences From Crime-Specific Studies
Versatility in expression of low self-control is fundamental to the meaning of self-and social control (Gottfredson & Hirschi, 2020), as their definition of self-control derives in substantial part from the versatility or generality of crime findings, thoroughly documented in criminology (see Gottfredson & Hirschi, 2020, 1990). As a result, MCT expects both similarities and differences in the causes of specific forms of crimes, but anticipates that, properly measured and conceived, the causal elements of the theory—self and social control, age, and elements of the situation—will each be a substantial part of the causal picture and be important factors to consider for public policy (Gottfredson & Hirschi, 2020, ch. 11). For one thing, the theory expects that offenders will tend (not always, but tend) to have engaged in a range of crimes and delinquencies. For another, a broad versatility notion underscores the importance of the ideas of social and self-controls and helps distinguish MCT from “motivation- specific” explanations. In addition, as a general theory, MCT sees these causes of crime and victimization as applying to all people, regardless of age, gender, race, ethnicity, culture, or period, an expectation with considerable empirical support (for reviews, see, for example, Gottfredson & Hirschi, 2020; Vazsonyi et al., 2017).
Of course, the finding that offenders are versatile and do not specialize in a particular form of delinquency is not the sole province of control theory: it is consistent with arguments put forth by feminist theorists as well as proponents of the term femicide. Femicide requires an acknowledgment of the gendered aspect of this form of violence. However, it does not mean that this violence should be studied in a vacuum or examined separately from other forms of delinquency. Doing so fails to consider that men have a “virtual monopoly on the use of violence in other social contexts” (Dobash et al., 1992, p. 72). Although there are feminist theories which provide motivation-specific explanations of IPV, a feminist framework is not limited to explaining those motivations. For example, not all men, even those who hold patriarchal beliefs, engage in violence against their female partners. Just like not all men who commit femicide are, or would, consider themselves as holding patriarchal beliefs. But all men, and all people, exist under patriarchy. An interpretation of feminist theories which seeks to explain motives to engage in this specific form of violence are limited. Instead, a feminist framework supports this assumption put forth by control theories. It does not negate men’s motives, but it places it within a broader social context.
Direct tests of the relationship between self-control and IPV offending are frequently reported in the research literature (for commentary on the operationalization of measures for multivariate testing of MCT, see Gottfredson & Hirschi, 2020, esp. Ch. 5, 6). Reviews of self- control (or as is sometimes the case in psychological literature, “self-regulation”) typically show important relationships. For example, Payne et al. (2010) review results from studies they argue are particularly relevant in the consideration of the relationship between self-control and partner violence, finding that self-control was indeed significantly associated with courtship or dating violence (see Chapple & Hope, 2003). Similar results about the role of self-control for review of research for IPV are reported by Spivey and Nodeland (2021). In a wide-ranging review, Finkel et al. (2009) argue that “many acts of IPV are immediately precipitated by perpetrators acting upon gut-level violent impulses that conflict with their more deliberative and self-controlled preferences for nonviolent conflict resolution. From this perspective, many acts of IPV are caused in large part by momentary failures in self-regulation” (p. 483).
In general, self-report data routinely demonstrate important self-control relationships to IPV as well as other common offenses (see, for example, Turanovic, 2022).
MCT Sees Common Causes for the Prevalence of Many Problem Behaviors and Thus Generally Expects Common Trends in These Behaviors Over Time, Which Can Make Crime-Specific Trend Study Potentially Misleading
Long-term trend data for the United States analyzed by Roberts (2009, p. 67) shows decreases for both men and women in IPH incidence over the past three decades, generally consistent with the overall decrease in homicide rates (see also, Aebi & Lind, 2014). Catalano (2012) shows similar data, with the declines somewhat greater for males relative to females, but, from 1994 to 2010, the overall rate of intimate partner violence in the United States declined by more than 60% for both males and females. Similar results are reported by Caman et al. (2017) for U.S. homicide offenses, which increased in the 1960s, followed by a steady decline from the early 1990s. Trend studies of interpersonal violence over significant time periods show considerable correlation between trends for IPH and other forms of homicide in the U.S. Trend studies of interpersonal violence over significant time periods show considerable correlation between trends for IPH and other forms of homicide in the United States (Roberts, 2009; for similar results for Australia, see Forrest, 2022). Along with other forms of homicide, there is evidence that IPH has decreased significantly over a several decade period (although perhaps more for male victims than for female victims). Although modest differences exist in these trends, the overall similarity is consistent with substantial common causes.
Eisner et al. (2016) show that trends in interpersonal violence are not only similar for different types of violence, but are also similar for property crime, teenage pregnancy rates, and alcohol use, smoking, and the use of illicit drugs other than cannabis among adolescents (for comparable findings, see also, Mishra & Lalumière, 2009). Although Eisner et al. (2016), report that local or national policy decisions do not seem to matter very much, the underlying consistency in trends despite variability in policy and practice suggests strongly to them that “. . . some broad underlying factor has shaped the shared trend across high-income countries” (p. 71). Their inferences are consistent with the results of international comparisons provided by Ball et al. (2023) who studied data on trends in adolescent risk behaviors (12-16) for a variety of developed countries. They show that smoking, drinking, drug use, underage sex, and juvenile crime all show similar, large declines throughout this period. For example, declines in heavy drinking ranged from 40% to 65% during the period, rates of juvenile offending between 40% and 80%, with similar declines for other risky behavior. They indicate that one major feature of these data is that they have occurred across demographic groups (with some variation in timing). They conclude their survey with the observation that a “unitary trend hypothesis” has considerable theoretical and empirical support, with growing international evidence that decreasing unstructured in-person socializing with friends may be a common underlying driver.
The consistency of these long-term trends, across many problem behaviors, is consistent with the expectations of MCT (Gottfredson, 2018; Gottfredson & Hirschi, 2020). Because the theory does not depend on legal definitions of crimes, the versatility of behaviors causing harm to others (and long-term harm to self) are included in the scope of the theory and thus, trends in criminal violence are predicted to follow trends in other interpersonal and property crimes, and to follow trends in other problem behaviors, such as many accidents, drug and alcohol use, school problems, and the like. Other putative causes, especially criminal justice and school sanctions are seen as either unlikely or at best secondary considerations in harm reduction, to be evaluated once elements of the theory are accounted for.
For the United States, data indicate that not only have differing forms of interpersonal violence substantially co-varied over lengthy periods of time and for separate jurisdictions, but so too have both official and survey measured rates of a wide variety of illegal behaviors. And, rates of many measures of risky behavior show similar long-term trends for adolescence. For example, in the past three decades, government agencies report substantial declines in consumption of by teens for alcohol, drugs, and smoking, and in the teen birthrate (see, generally, Centers for Disease Control and Prevention [CDC], 2021; National Institutes of Health, 2015). Accounting for declines in interpersonal violence in the same way as for declines in other problem behaviors seems more consistent with the data than are theories or expectations based on crime-specific models. In this sense, IPV and other crimes, delinquencies, and problem behaviors must have some similar causes, as MCT claims (Gottfredson, 2021a; Gottfredson & Hirschi, 1990).
Given the relatively low incidence of IPH in national data, and the substantial reporting biases that exist for both survey and official data for IPV, it needs to be stressed that these trends may have greater reliability for longer periods of time than for the often very brief time periods involved with studies of interventions associated with COVID lockdowns. The focus in much of the “crime-drop” literature in the United States on variation in criminal justice sanctions is a good example of how shorter-term trends can be highly misleading (Gottfredson, 2021a).
The evidence for “treatment” effects in IPV for COVID policies, are mixed and difficult to interpret. Nivette, Zahnow, et al. (2021) studied the effects for COVID restrictions in many countries, finding that the summary effect for assault “suggests that the implementation of stay-at-home restrictions was associated with a 35% reduction in daily assaults.” . . . but also “. . . substantial heterogeneity in the effect sizes across cities and crime outcomes” (870). Their homicide data suggest a small decline in the number of daily homicides following the implementation of stay-at-home restrictions, but generally a lack of a statistically significant decline among the cities studied. Piquero et al. (2021) conducted a review of 18 studies and concluded there was a “moderate increase in IPV” and speculated about a variety of causes. However, Peitzmeier et al. (2022) examine the immediate pre-COVID period and find that the prevalence of IPV in the immediate pre-COVID and during COVID periods did not differ significantly.
One hypothesis to explain such mixed effects is the lack of documentation for what treatment literature refers to as the integrity of the treatment: the assumption that substantial changes in lifestyle patterns occurred consistently, across individuals. In fact, evidence suggests that adherence to these restrictions very likely interacted substantially with level of self-control, for example. The research literature seems to agree: Nivette, Zahnow et al.’s (2021, abstract) review of compliance with COVID restrictions worldwide concludes that “non-compliance was higher in young adults who had previously scored high on indicators of ‘antisocial potential,’ including low acceptance of moral rules, pre-pandemic legal cynicism, low shame/guilt, low self-control, engagement in delinquent behaviors, and association with delinquent peers.” Furthermore, they conclude that such problems complicate the inferences about most of the before/after designs for this research. In addition, since much of the data for these studies comes from police data, and since these mostly depend on victim reports, there is the additional measurement problem that increased presence of the offender could affect the likelihood of reporting the violence to others, causing an artificial decrease in some rates.
Tests of RAT as an explanation of crime rate variability due to COVID restrictions using before/after designs confront the additional problem of determining how much exposure to motivated offenders actually changed, or changed in a theoretically meaningful way, due to the restrictions. Direct measures are rarely available and it is likely that considerable exposure already existed prior to the lockdowns. Exposure to violence in the home and availability of services which assist victims have been found to impact femicide rates. Both of these factors may have been significantly impacted during COVID (Gillespie & Reckdenwald, 2017). Because of these measurement problems, it is likely that the argument that COVID presents “the largest criminological experiment in history” (Stickle & Felson, 2020, p. 534) requires considerable context.
MCT Predicts Important Effects for Situational Factors as Causes for IPV: Guns and Alcohol
MCT argues that aspects of situations or settings have direct effects in causing crime and violence (Gottfredson, 2021a, 2021b; Gottfredson & Hirschi, 2020). The presence of “capable guardians” (monitors, parents or teachers, others in the household, security personnel, etc.) in some settings (schools, homes, playgrounds, etc.) can affect the likelihood that individuals with very low self-control will act on their immediate self-interest with force or fraud and the presence of drugs or alcohol and of weapons can greatly alter the situational determinants. Access to weapons can be especially dangerous in the hands of persons with low self- control in disputes, and the consumption of alcohol can relax whatever personal controls are available. In many domestic settings, these aspects of the situation enhance the likelihood that violence will be invoked. The data about IPV consistently documents the influence of alcohol and weapons, especially guns in the United States, as aggravating factors in the likelihood of IPV. “A remarkable 70% of the male perpetrators were using drugs and/or alcohol at the time of the homicidal incident” (Campbell et al., 2007, p. 253) and for approximately two-thirds of female victims the presence of alcohol or drugs was reported in about 43% of all incidents of nonfatal intimate partner violence (see also, lengthy citations for alcohol use in IPH [50%–75% of offenses] in Campbell et al., 2007). As Roberts (2009, abstract) puts it, “An overwhelming proportion of intimate partner (IP) homicide perpetrators are under the influence of substances when the crime occurs, and alcohol consumption is a strong predictor of intimate terrorism of women.”
The role of guns is substantially documented in the United States in IPH. According to Caman et al. (2017, p. 70), “Guns are the agents of homicide in the majority of IP homicides, with the percentages varying according to gender and marital status (citations omitted).” Ex-husbands and ex-wives are the most frequently killed with guns (87% and 78%, respectively), with husbands and wives at 70% and 68%, and boyfriends (46%) and girlfriends (57%). Caman et al. (2017, p. 70), shows that for IPH, “. . . female victims are twice as likely to die from a gunshot wound as from stabbing, strangling, or other methods; and firearm ownership is shown to increase the likelihood of IP homicide [substantially].” Such strong associations, routinely found in U.S. data, provide ample cause for interventions targeted to these situational causes.
Measurement Issues and Counting Rules for Repetitive Victimization are Problematic in IPV Research
Both victim surveys and data based on official records fundamentally count and classify crimes or victimizations based on the concept of an incident or an event, with definable parameters in space and time. These parameters provide a basis for counting, or measuring crimes and comparing their characteristics with other, similar events. Focusing on crimes as events has considerable advantages for causal analysis, advantages that have provided empirical criminology with powerful explanatory principles (see, Gottfredson, 2021a, for elaboration of this view). But it has been clear since the beginning of victim surveys that this approach also has significant limitations, limitations that have not received the attention they deserve, perhaps particularly with respect to interpersonal violence among persons known to one another. For example, early proponents of self-reported victimization surveys cautioned that highly victimized respondents, those who reported so-called “series” victimizations, were not well understood and that the counts of their experiences in these surveys were seriously undercounted and misleading. This suggested to Biderman (1975) that they should be viewed as separate phenomena to other crimes, and should not be understood as “events,” but rather more akin to enduring conditions. In a study of personal victimization and lifestyle in the British Crime Survey, Gottfredson (1984, pp. 30–33) argued for a stronger focus on the relatively small portion of the sample who suffer repeated victimization and who thus bear a disproportion of the burden from crime. There remains a need to understand much better this phenomenon, referred to by Farrell and Pease (2007) as “the sting in the tail” of victimization distributions. As they have stressed, most counting procedures for victim surveys and for data relying on police reporting and recording procedures restrict markedly the number of “events” for reporting these offenses. Given the marked skew in victimization counts associated with either official or survey counts of crimes, the experiences of very frequently victimized persons is not well represented in either data and as a consequence of this strong disproportionality of victimization for some victims, both their experiences and policy efforts directed to them are worth special focus (for comprehensive discussion, see Pease et al., 2018). From a theory point of view, studies of highly repetitive offending can provide insight into offender characteristics and their “decisions” about target selection, while a focus on reducing multiple victimization could well be an effective strategy for a reduction in these offenses that could materially affect the overall level of IPV. In addition, femicide is frequently the result of multiple incidents of violence that occur over time and increase in severity (see Dugan et al., 2003). In many cases of femicide, there is evidence of escalating instances of dominance, coercion, and violence throughout the relationship, such that when femicide occurs, these repeated attempts to dominate their partner through physical violence are typically “progressively persistent and severe” (Campbell et al., 2007).
General Theory and General Prevention: Policy Implication
The substantial harm caused by IPV to women presents important challenges to develop strategies to reduce the personal social harm it causes. According to Campbell et al. (2007, p. 253) some research claims that improvements in IPV laws and arrest policies in the United States helped to decrease the incidence of IP homicide beginning in the mid-1970s. Given the longer term trends shown in the empirical research, and the evidence from a large body of evaluation research demonstrating effectiveness of early childhood prevention strategies in reducing the likelihood of engaging in problem behaviors in adolescence and beyond, a focus on enhancement of early-life prevention support could help reduce IPV substantially. The imperative to focus resources on prevention to ameliorate IPV harms has considerable recent support. Eisner et al. (2016, p. 71) put it this way: [The data] suggest a common factor or some factors that has/have affected the common underlying trend across space and between behavior domains research indicates high comorbidity across developmental psychopathologies, a lack of offender specialization, and the subordinate role of domain-specific risk factors . . . Overall, these findings underline the significance of prevention programming that broadly supports a healthy child and youth development rather than focusing overly on specific subtypes of violence.
On the basis of their substantial study of the connections between interpersonal violence and age in international data over lengthy periods of time, Santos et al. (2019, pp. 19–20) argue persuasively that . . . our research also encourages a focus on evidenced based policies and programs which provide support to at-risk youth as a strategy for violence prevention . . . numerous other initiatives have been demonstrated to be effective in interrupting deviant trajectories and in reducing the vulnerability of high-risk youth even before the onset of serious violent behavior [citations omitted].
Similarly, Costa et al. (2015, p. 271) have argued that much research on IPV allows strong implications for policy and interventions targeted to prevention. They point to the appeal of targeting significant resources to at-risk families to provide support in early life (for similar recommendations, see also Peterman et al., 2015). Recent reviews of early childhood programs report a variety of positive results, with effectiveness for particular programs depending on age (e.g., Tehrani et al., 2023). These programs are similar to early prevention programs with favorable outcomes and highly favorable benefit/cost ratios cited by Gottfredson and Hirschi (2020, ch. 11; Gottfredson, 2018) as a preferred alternative to reliance on criminal justice interventions and sanctions.
A stronger focus on “distal” causes for IPV and IPH does not obviate the need to also provide proximate services and interventions. For example, for the United States, where handguns predominate as a means of injury for femicide, interventions that restrict handgun availability for those with previous records of partner violence is strongly indicated, and given the age distribution of these offenses, at a minimum until age 25. Ellyson et al. (2023), document the important role that guns play in IPV generally in the United States as well as the important role that legal prohibitions can play in reducing access to them for those with prior histories of IPV (see also Spencer & Stith, 2020). Similarly, limitations on alcohol use, for those with recent and prior IPV offenses, until a similar age are consistent with the data on age, and both IPV and IPH. Targeting supportive services to victims with prior relationship violence is also consistent with a focus on the individuals likely to suffer a disproportionate amount of IPV, thereby reducing a disproportionate amount of social/interpersonal harm.
Conclusion
General theories, such as MCT, appear to be reasonably consistent with much of the contemporary empirical work on the correlates and trends in IPV and IPH. Additional causes outside the scope of MCT are, of course, plausible and certainly IPH has specific as well as common causes. Long-term general trends for interpersonal violence, including IPV are consistent with expectations from MCT, as are other “crime-free” indicators of problem behaviors stipulated by theory. Research showing trends in violence associated with the age-distribution of crime and for other problem behaviors are consistent with the expectations of MCT over lengthy periods. The findings for indicators of adolescent behaviors such as substance use, teen birth rates, driving accidents, and school problems, together with historical studies of indicators of self-control over substantial historical periods, help explain the rises and declines in crime during the past several decades, at least in North America and Europe.
However, study of IPH using brief periods, limited jurisdictions, and small samples limit inferences that can be made on the basis of both official and survey data about the effects of COVID lockdowns. Normal threats to the validity of these before/after designs, most notable problems of selection (e.g., not all individuals “locked down” behave similarly with respect to these restrictions), and variability or integrity of the “treatments,” make the “mixed” results difficult to interpret.
With respect to policy, MCT clearly predicts a general lack of effectiveness for criminal justice sanctions for IPV generally. Along with the attendant collateral costs of such sanctions, sensible public policy favors early childhood support as a general prevention strategy. Such strategies have now demonstrated considerable empirical support (Gottfredson and Hirschi (2020, ch. 11; Gottfredson, 2018). Restrictions on alcohol and gun availability, actively enforced especially for individuals with recent prior violent records, are strongly indicated by the empirical research. Data also support the idea that such restrictions are most necessary for youth (e.g., prior to 25–30 years of age, after which risk declines appreciably, but not completely, of course). Arguably, those with prior violent interactions with partners may forfeit considerations not to be treated as potential false positives given a substantially greater likelihood of repeated, specific violence. In the case of the United States, prohibitions for alcohol use and for possession of firearms are possible for individuals on restricted release—such as pretrial, probation and parole.
A key area for policy focus, also related to the validity of both self-reported and official data about IPV, centers on the notification to others of violence among persons known to one another. Whether notification to police or to others, the data have been clear for decades—well over half of interpersonal violence incidents (surely much more, given known problems with nonreporting of repetitive or serial victimizations) are not reported. The reasons are multiple (I. D. Johnson & Lewis, 2023), but prior experiences with the lack of effective response, trust in police and the criminal justice system, and fear of reprisal are significantly reported to survey interviewers. Focused attention on the reporting issues, which must be followed by rapid and effective support, particularly but not exclusively focused on repetitive victims, remains an essential need.
The available data can also be read as consistent with several other general theories, provided they specify general characteristics of perpetrators similarly consistent with the available empirical research. RAT and situational opportunity theories continue to suffer shortcomings against available research by failing to explicitly identify the indicators for “motivated offenders,” despite an abundance of evidence. Absent an actor, characteristics of situations can go only a limited way in predictions and create ambiguity about even the direction of effects the theory suggests—witness the alternative explanations available to “account for” differing results from the COVID-era research.
There is no apparent necessary dispute between MCT and feminist perspectives which also sees common causes for the variety of specific motivations often expressed for IPV. As H. Johnson et al. (2019) put it, [t]he context in which a great many femicides take place is one of male dominance and control which is manifested in possessiveness, extreme jealousy, attempts to isolate the women, threats of suicide, and threats to kill that are often triggered by loss of control due to impending separation or real or imagined infidelity. (p. 7)
As the self-control component of MCT might put it: targets are selected (by men, most often) in substantial part because they believe that through coercion and violence, they can achieve immediate goals with impunity—a perceived power differential consistent with selection by those with very low self-control. Weapons exacerbate the harm and can be used impulsively, without regard to the consequences to the victim or, in the longer term, to the offender.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.