Influence of Intimate Terrorism, Situational Couple Violence, and Mutual Violent Control on Male Victims

Abstract

The goal of the current study is to test – among samples of men – some previous findings relative to Johnson’s (2008) typology of partner violence (PV). Among samples of exclusively women, Johnson has found that the frequency and severity of PV – and victims’ mental health – are worse for female victims of intimate terrorism (IT) than of situational couple violence (SCV). However, such findings have rarely been tested in male victims of PV. Furthermore, although Johnson posits that mutual violent control (MVC) is relatively rare, other research suggests that it is about as common as IT and may be associated with worse PV and health. We examined the relative frequency and severity of various forms of PV, and the relative health and mental health between men who (a) experienced IT versus SCV, and (b) experienced IT versus MVC. Our two samples were 611 men who sought help for PV victimization and 1,601 men from a population-based sample. Our analyses showed that across samples, men who were victims of IT had significantly worse mental health than men who experienced SCV, and IT victims experienced more severe and frequent PV, including physical, sexual, and non-physical forms of PV. Male victims of IT had worse mental and physical health than either men who perpetrated IT or experienced MVC, but men involved in MVC were involved in relationships with more types of PV than either male IT perpetrators or victims. Results are discussed in terms of their implications for Johnson’s typology, future research, and service providers.

Johnson’s (1995; 2008) typology of partner violence (PV) is cited as the explanation for the differences in two divergent lines of PV research: one that shows that men predominate as the perpetrators of PV and a second that shows that men and women are equally likely to perpetrate PV. Specifically, Johnson asserts that research on agency samples (e.g., domestic violence agencies, law enforcement, hospitals) show that men vastly predominate as perpetrators of more severe forms of PV – called intimate terrorism (IT) – which can be explained by patriarchal theories (e.g., Dobash & Dobash, 1983). Such theories, Johnson says, assert that men use PV within a patriarchal system to maintain dominance within their relationship. Men are trying to exert control over their wives/girlfriends, a control to which they feel entitled and is supported by patriarchal norms. On the other hand, Johnson (1995; 2008) contends that community samples show that men and women are equally likely to perpetrate more minor forms of PV – called situational couple violence (SCV) – that is due to arguments occasionally escalating to the point of typically minor violence. SCV is not due to the male partner trying to maintain dominance over his female partner.

Johnson (1995; 2008) also acknowledges that female victims of IT sometimes do use violence, but, he says, it is in an effort to defend themselves and not meant to control their partners. This is known as violent resistance, and it happens when a woman is faced with a partner who is both violent and controlling. Johnson further asserts that on very rare occasions, a couple’s violent relationship might be characterized as mutual violent control (MVC); this is when both the man and the woman use IT in a battle for control (Johnson, 1995, 2008).

As previous analyses have shown, Johnson’s underlying theoretical explanations for his typology are based upon research with inherent sampling biases, and his conclusions that only men perpetrate IT have been contradicted with empirical evidence (Ehrensaft, Moffitt, & Caspi, 2004; Graham-Kevan & Archer, 2003; Hines & Douglas, 2016; Laroche, 2008). In addition, his conclusions regarding MVC being rare have also been questioned (Ehrensaft et al., 2004; Graham-Kevan & Archer, 2003; Hines & Douglas, 2016; Straus, 2011). This line of research indicates that we need to investigate whether hypotheses drawn from Johnson’s typology can also be applied to victims of PV who are not heterosexual women.

The extant research on IT and SCV victims shows that female IT victims suffer more frequent and severe physical PV and more severe physical and mental health problems in comparison with female SCV victims (e.g., Frye, Manganello, Campbell, Walton-Moss, & Wilt, 2006; Johnson & Leone, 2005; Leone, Johnson, Cohan, & Lloyd, 2004). Few studies have focused on the relative frequency and severity of PV in relationships where only one partner is a perpetrator of IT versus where both partners are (i.e., MVC; Ehrensaft et al., 2004; Straus, 2011), and none examine the physical and mental health of people involved in relationships with IT versus MVC. The goals of the current study are to apply Johnson’s typology to samples of heterosexual men (a population-based sample and a male PV victims sample), and to examine the relative frequency and severity of various forms of PV, and the relative health and mental health between men who (a) experience IT versus SCV, and (b) experience IT versus MVC.

Differences in Abuse and Consequences of IT versus SCV

One area that Johnson has researched in depth on female victims of PV is the distinct abuse dynamics and potential consequences of IT, as compared to SCV. For example, in analyses of the NVAWS female sample data, Johnson and Leone (2005) found that in comparison the SCV victims, IT victims sustained significantly more frequent and severe physical PV, and were significantly more likely to be injured. Similarly, in Johnson’s original study of women in Pittsburg (a study that combined two samples: female victims who sought help for PV victimization and a sample of community women), of the women experiencing IT, 76% were injured, compared to 28% of the SCV women (Johnson, 2006). Thus, evidence suggests that the severity and frequency of abuse and its consequences (i.e., injury) are different for IT and SCV victims, but two questions remain: (1) Do these differences extend to men as victims of female-perpetrated IT and SCV?, and (2) Do these differences in severity and frequency of abuse extend to other forms of PV, such as sexual PV and other non-physical forms of PV, such as psychological aggression? We hypothesize that physical and non-physical forms of PV will be more severe and frequent among male IT victims than among male SCV victims.

There is also evidence that the physical and mental health consequences of PV differ for victims of IT versus SCV. For example, in the NVAWS female sample, victims of IT were significantly more likely to display post-traumatic stress disorder (PTSD) symptoms (Johnson & Leone, 2005), and in a study of women in a poor neighborhood in Chicago, victims of IT reported poorer general health, greater likelihood of visiting the doctor, and more psychological distress (Leone et al., 2004). Frye and colleagues (2006) suggest that because of social and personal power-level differentials among men and women, it is likely that violence and control tactics would have different impacts on men as they do on women. Yet, existing evidence casts doubt on that hypothesis. For example, Laroche (2008) found that among both men and women in the Canadian GSS sample, the physical and psychological consequences of PV were concentrated among severe IT victims. These consequences included receiving medical care, consulting a mental health practitioner, having everyday activities disrupted, and fearing for one’s life. Another purpose of this study is to investigate whether the potential physical and mental health consequences differ among male IT and SCV victims. Given Laroche’s empirical findings, we hypothesize that male IT victims will have worse physical and mental health in comparison with male SCV victims.

Intimate Terrorism and Mutual Violent Control

Johnson’s (2005) theory specifically states that IT is not gender symmetric. He says that it is possible, although rare, for there to be two intimate terrorists in a relationship. As mentioned above, this is known as MVC (Johnson, 2008), and it is so rare, he says, that Johnson’s only commentary on this form of PV is that we need more research to understand it better. However, several researchers have noted that the majority of the tests of Johnson’s typology have used data on only victimization – and mostly on only women’s victimization – thus, the degree of mutuality of IT could not be assessed (Capaldi & Kim, 2007; Straus, 2011). Researchers who have measured both victimization and perpetration have found evidence of a much higher rate of MVC. For example, McDonald and colleagues’ (2009) study of a shelter sample of women found that 67% had engaged in severe violence toward a partner. Graham-Kevan and Archer (2003) found that 31% of a sample of male prisoners reported MVC. Similarly, in a study of a birth cohort in New Zealand, Ehrensaft et al. (2004) found that IT was much more likely to be bidirectional than was SCV, which was predominantly female-perpetrated.

In a previous analysis of the dataset for the current article, we found that MVC and IT were equally prevalent in the population-based sample, but that IT predominated in the male victims sample (Hines & Douglas, 2016). However, it is unknown whether IT victimization or MVC victimization has a worse impact on its victims. In a review of 91 empirical studies of PV that provided information on both minor and severe PV perpetrated by both men and women, Straus (2011) showed that risk of injury and the probability of violence continuing or escalating was greatest when both partners were violent (Straus & Gozjolko, 2014; Whitaker, Haileyesus, Swahn, & Saltzman, 2007). We tested whether MVC was associated with more frequent and severe PV, greater injury rates, and greater likelihood of other health problems among the men, in comparison to men who experienced unidirectional IT. Given this research that there is greater injury and mental health problems in relationships characterized by bidirectional violence, we hypothesized that men involved in MVC will have worse mental and physical health problems than either perpetrators or victims of IT, and that relationships characterized by MVC will be characterized by more severe PV.

Summary and Hypotheses

This study represents a test of several components of Johnson’s (1995, 2008) typology of PV on a combined sample of men: one of men who sought help because their female partner was violent (male victims sample) and a population-based sample who reported on their intimate relationships with female partners. We hypothesized that:

1. Physical and non-physical forms of PV will be more severe and frequent among male IT victims than among male SCV victims.

2. Male IT victims will have worse physical and mental health in comparison with male SCV victims.

3. Men involved in MVC will have worse mental and physical health problems than either male perpetrators or male victims of IT, and relationships characterized by MVC will be characterized by more severe PV.

Methods

Participants and Procedure

Two separate samples of male participants were recruited for this study: a help-seeking sample of physical PV victims and a population-based sample. For both samples, the men had to speak English, live in the U.S., and be between the ages of 18 and 59 to be eligible; they also had to have been involved in an intimate relationship with a woman lasting at least one month in their lifetimes. In addition to the above criteria, to be eligible for the help-seeking sample of male victims, the men had to have sustained a physical assault from their female partner at some point in their relationship, and they had to have sought assistance for their partner’s violence from at least one of the following sources: medical doctor or dentist, domestic violence agency, domestic violence hotline, the Internet, a lawyer, the police, a clergy member, a family member, a friend, or a mental health therapist.

We recruited the help-seeking sample of male victims (n = 611) from a variety of sources. We posted advertisements on our research webpage and Facebook page, and we posted ads on webpages and Facebook pages of agencies that specialize in male victims of PV, the physical and mental health of men and minority men, fathers’ issues, and divorced men’s issues. We also sent out announcements to a database of researchers, practitioners, and other interested parties who signed up to be on our e-mailing list through our research webpage, which has been in existence since 2008. The advertisement stated that we were conducting “a study on men who experienced aggression from their girlfriends, wives, or female partners.” The ad then provided a link to the anonymous online questionnaire. After providing consent, the next two pages of the survey contained questions to assess the above screening criteria. Men who were eligible were allowed to continue the survey. Men who did not meet the eligibility requirements were thanked for their time and were redirected to an “exit page” of the survey. Demographics of the male victims sample can be found in Table 1.

Table 1.

Demographics of the Population-Based and Male Victims Samples

	Population-Based Sample (n = 1,601)	Male Victims Sample (n = 611)	χ2 or t
	% or M (SD)	% or M (SD)
Male Participant Demographics
Age	41.77 (11.35)	43.89 (9.18)	4.52***
White	76.5%	75.5%	0.28
Black	10.2%	4.1%	21.09***
Hispanic/Latino	11.8%	4.9%	23.57***
Asian	1.9%	4.3%	10.16***
Native American	1.4%	2.9%	5.54*
Income (in thousands)	48.5 (27.6)	47.7 (27.7)	0.63
Educational Status1	3.68 (1.83)	4.71 (1.63)	12.90***
Female Partner Demographics
Age	40.28 (11.60)	40.77 (9.53)
White	75.5%	67.4%	14.76***
Black	8.1%	4.1%	10.74***
Hispanic/Latina	9.9%	9.7%	0.02
Asian	4.0%	5.7%	3.10
Native American	1.4%	1.0%	0.71
Income (in thousands)	36.8 (23.5)	43.9 (29.6)	5.14***
Educational Status1	3.79 (1.78)	4.17 (1.77)	4.40***
Relationship Demographics
Currently in a Relationship	86.5%	26.3%	730.93***
Relationship Length (months)	150.09 (122.86)	112.33 (87.62)	8.05***
Time since relationship ended (in months)	6.55 (29.91)	45.17 (54.33)	16.63***
Minors Involved in the Relationship	41.6%	67.7%	118.83***
# of Minors involved in Relationship	0.79 (1.12)	1.12 (1.03)	6.58***

¹Educational Status: 1 = Less than high school, 2 = High school graduate or GED, 3 = Some college/trade school, 4 = Two-year college graduate, 5 = Four-year college graduate, 6 = At least some graduate school.

^*p < .05,

^**p < .01,

^***p < .001.

Participants also included a population-based sample of 1,601 men. Their data were collected by the Internet survey research firm, Knowledge Networks (KN). KN offers the only Internet research panel of about 43,000 adults that is representative of the U.S. population. Panel members are chosen through an intensive, list-assisted random digit dial methodology, supplemented by traditional mailing addressed-based sampling to reach cell-phone only populations. They are invited to participate in the Web panel, and those who agree (~56%) are enrolled in the panel. Those who do not have Internet access are sent an Internet appliance and are provided with Internet access through KN. As incentives, panelists are enrolled in a points program where they accumulate points by completing surveys and then trade them in for prizes.

To increase the likelihood of the panel members’ participation in our study, KN provided extra incentives and sent reminder emails three times during the month of data collection. KN’s email was sent to male panel members between the ages of 18 and 59, and it informed them about a study, supported by the National Institutes of Health (NIH), on how well men and women get along, and men’s health. Of the 3,536 men who were invited to participate, 2,174 (61.5%) entered the survey; 90% of them consented to participate, and of those who consented, 82.5% were eligible. Demographic information on this sample can be found in Table 1.

The methods for this study were approved by the boards of ethics at our institutions of higher education. All participants were apprised of their rights as study participants. All of the men in the male victims sample participated anonymously. Participants in the population-based sample participated confidentially. KN links the data from each survey to the demographic information that it maintains on each participant. However, KN did not release any identifying information to the investigators on this project. Participants were informed that their responses would remain confidential, their confidentiality was protected with a Certificate of Confidentiality obtained from the NIH, KN would not release any identifying information to the investigators, and they could not be personally identified in any reports that resulted from their participation. Also, steps were taken to ensure all participants’ safety: At the completion of the survey, the participants were given information about obtaining help for PV victimization or psychological distress, and on how to delete the history on their Internet web browser.

Measures

Both the male victims and population-based samples were given the same questionnaires regarding demographics, aggressive behaviors that they and their female partners may have used, their mental health, their physical health, various risk factors for IPV, and if applicable, their children’s witnessing of IPV, their children’s mental and physical health, and other risk factors for their children. Men in the victims sample were asked to report on their relationship with their abusive female partner, whereas men in the population-based sample were asked to report on their most recent female partner. Only the questionnaires used in the current analyses are described here.

Demographic information

Men were asked basic demographic information about both themselves and their partners, including age, race/ethnicity, personal income, and education. Men were also asked about the current status of their relationship, the length of their relationship with their partners, how long ago the relationship ended (if applicable), whether they parented any minor children with their abusive female partner, and how many minor children they parented together.

Revised Conflict Tactics Scales (CTS2)

We used the CTS2 (Straus, Hamby, Boney-McCoy, & Sugarman, 1996) to measure the extent to which the men perpetrated and sustained severe psychological, physical, and sexual aggression, and injuries in their relationships. The items used for this study included 4 items assessing severe psychological aggression (e.g., threatening to hit or throw something at partner, calling partner fat or ugly), 12 items assessing physical aggression (e.g., slapping, beating up), 6 items assessing injuries (e.g., having a small cut or bruise, broken bone, passing out), and 6 items assessing sexual aggression (e.g., insisting on, threatening, or using force to have sex when the partner did not want to). For the physical aggression, injury, and sexual aggression items, we further divided the behaviors into subcategories, according to Straus et al. (1996) and Johnson (1995). Severe physical aggression was aggression that had a higher likelihood of causing an injury (e.g., punching, kicking). Very severe physical aggression was considered even more life-threatening (e.g., beating up, using a knife or gun). Severe injury was an injury that needed medical attention (e.g., broken bone, passing out from being hit on the head). Minor sexual aggression was insisting on vaginal, oral, or anal sex when one’s partner did not want to, whereas severe sexual aggression was threatening or forcing one’s partner to engage in vaginal, oral, or anal sex.

Consistent with our previous research on male victims (e.g., Hines & Douglas, 2010a; Hines & Douglas, 2010c, 2011c, 2012; Hines & Douglas, 2013), we supplemented the CTS2 with nine items from the Psychological Maltreatment of Women Inventory (PMWI; Tolman, 1995) that focused on controlling behaviors and could be applied to men as victims. A factor analysis (Hines & Douglas, 2010c) showed that these items represented a unique factor that was distinct from the severe psychological aggression items of the CTS2. We also added twelve items to measure men’s perpetration and victimization from legal/administrative aggression, which included both threats to and actually: making false accusations to authorities that the partner physically or sexually abused the other, making false accusations to authorities that the partner physically or sexually abused the children, leaving and taking the children away, leaving and taking all the money and possessions, ruining the partner’s reputation at work, and ruining the partner’s reputation in the community. Previous analyses of this subscale assessing victimization of both actual and threatened legal/administrative aggression with the current samples showed that it has excellent psychometric properties, including good construct validity and good alpha reliability; analyses with the perpetration items, however, should be viewed with caution given less desirable validity and reliability (Hines, Douglas, & Berger, 2014).

Participants responded to items depicting each of the conflict tactics by indicating the number of times these tactics were used by the participant and his partner. Participants indicated on a scale from 0 to 7 how many times they experienced each of the acts, 0 = never; 1 = 1 time in previous year; 2 = 2 times in previous year; 3 = 3–5 times in previous year; 4 = 6–10 times in previous year; 5 = 11–20 times in previous year; 6 = more than 20 times in previous year; 7 = did not happen in the previous year, but has happened in the past.

For the current analyses, each subscale of the CTS2 (i.e., perpetration and victimization of each type of PV) was scored in two ways: (1) Whether any of the types of aggression ever happened (dichotomous yes/no variable), and (2) The number of different acts of each type of aggression that ever happened (e.g., there were a total of 12 items of physical aggression, so participants could be victimized by up to 12 types of physical aggression). This method of scoring is called variety scores and is recommended by Moffitt and colleagues (1997), who showed that it provides a reliable and valid assessment of the severity and frequency of the various forms of IPV, without violating statistical assumptions, and they have stronger reliability and predictive validity than frequency or seriousness measures (Ehrensaft et al., 2004).

The CTS2 has been shown to have good construct validity and good reliability (Straus et al., 1996). Alpha reliability statistics for the current samples ranged from .69 (perpetration of severe psychological aggression) to .94 (victimization from physical aggression).

Posttraumatic stress disorder symptoms

The PTSD Checklist (PCL; Weathers, Litz, Herman, Huska, & Keane, 1993) is a 17-item, self-administered instrument for assessing the severity of PTSD symptomatology. Items cover three symptom clusters: re-experiencing, numbing/avoidance, and hyperarousal. Participants indicate on a 5-point scale (1 = not at all, 5 = extremely) the extent to which they were bothered by each symptom in the previous month in relation to their worst argument with their female partner (e.g., “I have repeated disturbing memories, thoughts, or images of that argument”). The PCL has been used to evaluate PTSD symptomatology in a variety of populations, including female sexual assault victims (Blanchard, Jones-Alexander, Buckley, & Forneris, 1996) and male victims of PV (Hines & Douglas, 2011c). The PCL has demonstrated excellent reliability, with alpha coefficients above .90 (Blanchard et al., 1996; Lang, Laffaye, Satz, Dresselhaus, & Stein, 2003; Weathers et al., 1993) and test-retest reliability of .96 (Weathers et al., 1993). The measure has also shown strong convergent and divergent validity (Blanchard et al., 1996; Ruggiero, DelBen, Scotti, & Rabalais, 2003). Based on pilot testing with male PV victims from previous research (Hines & Douglas, 2011c), we eliminated one item from the original PCL (“feeling as if your future will somehow be cut short”) because the men reported not understanding the item. Cronbach’s alpha for the 16-item scale for the current sample was .97.

Depression symptoms

The Center for Epidemiologic Studies Depression (CES-D; Radloff, 1977) scale was used to measure depressive symptomatology. The CES-D contains 20 questions about feelings and behaviors from the past week (e.g., “I was bothered by things that usually don’t bother me”). Response options range from 0 (rarely or none of the time) to 3 (most or all of the time). The CES-D has high internal consistency and adequate test-retest reliability. Cronbach’s alpha for the current sample was .95.

Physical health symptoms

Physical health symptoms were assessed with the Cohen-Hoberman Inventory of Physical Symptoms (CHIPS; Cohen & Hoberman, 1983). Participants indicated on a 6-point scale, ranging from 0 (never) to 5 (>4 times/week) the frequency with which they experienced each of the 30 symptoms listed, including sleep problems, fatigue, and various aches and pains. The CHIPS has been used successfully in clinical samples of women who have sustained PV (Sutherland, Sullivan, & Bybee, 2001), with internal consistencies above .90. For the current sample, Cronbach’s alpha was .96.

Poor Health

Poor health was measured with the SF-4, a 4-item measure of the limitations that physical or emotional problems may have placed on work, physical, and social activities, and general levels of energy and pain. Participants were asked to rate aspects of their health on a 6-point scale (0 = very poor; 5 = excellent), or indicate on a 5-point scale how much their health limitations interfered with various aspects of their life (0 = not at all; 4 = could not do). An example item is, “During the past 4 weeks, how much did physical or mental health problems limit your usual physical activities (such as walking or climbing stairs)?” The SF-4 is a shortened version of both the SF-36 (McHorney, Ware, Lu, & Sherbourne, 1994) and the SF-8, widely used measures of general health that have shown excellent reliability and validity. The SF-8 is recommended for use in general population-based research surveys to reduce participant burden. We shortened it to four items to ease participant burden. Items were summed, for a potential range of scores from 0–17. Cronbach’s alpha for this 4-item scale was .88.

Classification of Participants into Johnson’s Typology

To determine whether participants were perpetrators and/or victims of the various types of PV identified in Johnson’s (2008) typology, we first determined whether the participants experienced PV. If participants indicated on the CTS2 that they either perpetrated and/or sustained any of the 12 acts of physical PV, they were classified as being involved in a PV relationship. As shown in Table 2 and as expected given the recruitment criteria for the male victims sample, none of the men in the male victims sample indicated that they were in a non-violent relationship; on the other hand, 74.4% of the men in the population-based sample indicated that they were in a non-violent relationship.

Table 2.

Percent and number of men in each type of relationship delineated by Johnson’s Partner Violence Typology

Category of Violent Relationship	Combined Sample	Population-Based Sample	Male Victims Sample
	N = 2172	N = 1583	N = 589
	% of Total (n)	% of Total (n)	% of Total (n)
Both Male Participant and Partner were Non-Violent	54.2% (1177)	74.4% (1177)	0.0% (0)*
Male Participant Perpetrates SCV; partner non-violent	1.8% (40)	2.5% (40)	0.0% (0)*
Female Partner Perpetrates SCV; male non-violent	10.5% (227)	7.8% (124)	17.5% (103)*
Mutual SCV	10.4% (225)	10.0% (158)	11.4% (67)
Male Participant Perpetrates IT; partner non-violent	0.1% (3)	0.2% (3)	0.0% (0)
Male Participant Perpetrates IT; female partner uses VR	0.5% (10)	0.4% (7)	0.5% (3)
Female Partner Perpetrates IT; male non-violent	10.8% (234)	0.8% (12)	37.7% (222)*
Female Partner Perpetrates IT; male uses VR	9.0% (195)	1.3% (20)	29.7% (175)*
MVC (i.e., both perpetrated IT)	2.8% (61)	2.7% (42)	3.2% (19)

Note. SCV = situational couple violence; IT = intimate terrorism; MVC = mutual violent control; VR = violent resistance. According to a chi-square test, the population-based sample and the male victims sample significantly differed in the categorization of Johnson’s typology, χ² (8) = 1424.03, p < .001. An * in the male victims sample column indicates that there was a significant difference in that specific category, according to post hoc z-tests comparing proportions, with a Bonferonni correction.

Of the men who indicated at least one act of physical PV, we further subdivided them into Johnson’s typology based on the level of controlling behaviors reported in the relationship. Analyses of Johnson’s (2008) typology have consistently shown that people who use three or more different types of controlling behaviors within a violent relationship can be classified as IT perpetrators (Frye et al., 2006; Johnson, 2008; Johnson & Leone, 2005; Laroche, 2008). These categories were derived in previous research using cluster analyses, and this cut-off of three or more corresponds to two standard deviations above the mean for controlling behaviors in population-based samples; thus, Johnson (2008) suggests using two standard deviations above the mean for the number of types of controlling behaviors in population-based samples as the cut-off between IT and SVC. These results were found using analyses of the NVAWS data (Johnson & Leone, 2005) and the Canadian GSS (Laroche, 2008), and this scoring method was also adapted by Frye et al. (2006) in their analyses.

Similarly, we used our population-based sample to establish the cut-off point for the controlling behaviors variety score. The mean and standard deviation were higher for men’s reported victimization (M = 0.46, SD = 1.32) than for perpetration (M = 0.36, SD = 1.21). Using Johnson’s (2008) argument that the cut-off for IT should correspond to two standard deviations above the mean, we established the cut-off for both victimization (3.10) and perpetration (2.78) and then averaged the two. In our population-based sample, that equals 2.94 for the controlling behaviors variety score, and therefore, we also used the cut-off of three or more different types of controlling behaviors in order to classify participants and their female partners as IT perpetrators/victims.

Based on the presence of physical PV and the level of controlling behaviors, we classified participants into eight potential configurations of Johnson’s typology: (1) male participant perpetrates SCV/female partner is nonviolent; (2) female partner perpetrates SCV/male participant is nonviolent; (3) mutual SCV; (4) male participant perpetrates IT/female partner is non-violent; (5) male participant perpetrates IT/female partner uses violent resistance; (6) female partner perpetrates IT/male participant is nonviolent; (7) female partner perpetrates IT/male participant uses violent resistance; and (8) MVC (i.e., both male participant and female partner perpetrate IT). Table 2 presents the number and percent of participants in each category overall and separately by sample type (see Hines & Douglas, 2016, for further discussion of this table).

Results

Preliminary Analyses

We checked the data assumptions relevant to the MANCOVA analyses we performed to test our hypotheses. These assumptions include: normality of the dependent variate, absence of extreme scores, and homogeneity of the covariance matrices, and the data did not violate these assumptions. In addition, we tested for a sufficient correlation among the dependent variables using Bartlett’s Test of Sphericity. The correlations among the dependent variables, separated by sample type, are in Table 3, and Bartlett’s Test suggested a sufficient correlation (p < .001) to proceed with analyses. For significant MANCOVAs, our follow-up tests included one-way ANCOVAs, with Tamhane’s T2 post hoc tests. Bonferonni corrections were used where indicated.

Table 3.

Correlations Among the Dependent Variables, by Sample Type

	CES-D Score	PCL Score	CHIPS Score	SF-4 Score
CES-D Score	--	.55*	.58*	.58*
PCL Score	.73*	--	.43*	.37*
CHIPS Score	.70*	.65*	--	.62*
SF-4 Score	.73*	.57*	.70*	--

Note: Population -based sample correlations are above the diagonal, while the male victims sample correlations are below the diagonal.

^*p < .001

Hypothesis 1: Differences in Abuse Severity and Frequency Between Intimate Terrorism and Situational Couple Violence

Our first hypothesis stated that physical and non-physical forms of PV would be more severe and frequent among male IT victims than among male SCV victims. We conducted a MANCOVA with victimization group as the independent variable (no victimization, SCV victims, IT victims) and all forms of non-physical, sexual, and physical PV as the dependent variate. MANCOVA analyses were conducted separately for the victimization dependent variate and the perpetration dependent variate. We also used the following demographics as potential covariates: men’s age, racial/ethnic minority status (minority vs. non-minority), education, income, whether they reported on a current relationship, the length of the relationship, and whether they had minor children. For the victimization dependent variate, all covariates contributed significantly to the dependent variate. In addition, the independent variable of victimization type was a significant predictor, Wilks’ Δ = .108, F (22, 4122) = 382.73, p < .001, partial η² = .671. Follow-up ANCOVAs showed that victimization type was a significant predictor of every victimization dependent variable. For the perpetration dependent variate, men’s income was not a significant covariate and was removed from the model. Similar to the victimization dependent variate, victimization type was a significant predictor of the perpetration dependent variate, Wilks’ Δ = .636, F (22, 4116) = 47.51, p < .001, partial η² = .203. Follow-up ANCOVAs also showed that victimization type was a significant predictor of all of the perpetration dependent variables.

Table 4 presents the results of follow-up ANCOVA analyses looking at the differences between victimization groups in both their victimization and perpetration of various forms of PV. For ease of interpretation, we presented the non-adjusted means. The top portion of the table focuses on victimization. Consistent with our hypothesis, for all forms of non-physical, sexual, and physical aggression, IT victims experienced significantly more variety types than SCV victims. Similarly, for perpetration, IT victims perpetrated significantly more variety types of almost all forms of PV, with the exception of minor psychological, minor sexual, and minor physical aggression. In addition, for almost all forms of PV, both SCV and IT victims perpetrated and were victimized by more variety types of PV than men who experienced no physical PV victimization. The one exception was that male SCV victims did not perpetrate more types of severe injuries than men who experienced no violent victimization.

Table 4.

Differences in Partner Violence Variety Scores and Health Indicators Across Johnson’s Victimization Subtypes, according to follow-up ANCOVA tests of significance

	No Violent Victimization	SCV Victimization Only	IT Victimization	F	p
	N = 1223	N = 466	N =493
	M (SD) or %	M (SD) or %	M (SD) or %
Male Participant’s Victimization (Variety Types)
Non-Physical Aggression
Minor Psychological Aggression	1.70 (1.39)a	3.19 (0.98)a	3.76 (0.49)a	444.08	<.001
Severe Psychological Aggression	0.14 (0.41)a	1.33 (1.21)a	3.07 (1.05)a	1427.41	<.001
Controlling Behaviors	0.17 (0.55)a	0.75 (0.85)a	5.34 (1.88)a	2927.47	<.001
Threatened Legal/Administrative Aggression	0.07 (0.34)a	1.27 (1.76)a	3.75 (1.90)a	890.36	<.001
Actual Legal/Administrative Aggression	0.03 (0.24)a	0.84 (1.52)a	2.43 (2.01)a	275.38	<.001
Sexual Aggression
Minor Sexual Aggression	0.09 (0.33)a	0.26 (0.54)a	0.73 (0.81)a	169.58	<.001
Severe Sexual Aggression	0.01 (0.15)a	0.12 (0.45)a	0.74 (1.24)a	178.58	<.001
Physical Aggression
Minor Physical Aggression	0.00 (0.00)a	2.33 (1.41)a	3.89 (1.24)a	2562.61	<.001
Severe Physical Aggression	0.00 (0.00)a	0.89 (1.18)a	2.83 (2.07)a	712.31	<.001
Injury
Minor Injury	0.00 (0.04)a	0.49 (0.78)a	1.33 (0.84)a	634.36	<.001
Severe Injury	0.00 (0.06)a	0.17 (0.57)a	0.90 (1.17)a	243.43	<.001
Male Participants’ Perpetration (Variety Types)
Non-Physical Aggression
Minor Psychological Aggression	1.66 (1.36)a,b	2.69 (1.18)a	2.64 (1.27)b	170.87	<.001
Severe Psychological Aggression	0.13 (0.43)a	0.53 (0.82)a	0.80 (1.16)a	161.03	<.001
Controlling Behaviors	0.13 (0.50)a	0.43 (0.78)a	1.15 (1.98)a	167.35	<.001
Threatened Legal/Administrative Aggression	0.02 (0.15)a	0.12 (0.41)a	0.35 (1.02)a	72.45	<.001
Actual Legal/Administrative Aggression	0.00 (0.06)a	0.06 (0.31)a	0.13 (0.45)a	26.12	<.001
Sexual Aggression
Minor Sexual Aggression	0.18 (0.46)a,b	0.28 (0.56)a	0.28 (0.61)b	18.29	<.001
Severe Sexual Aggression	0.02 (0.17)a	0.05 (0.28)a	0.21 (0.80)a	63.67	<.001
Physical Aggression
Minor Physical Aggression	0.05 (0.28)a,b	0.98 (1.24)a	1.10 (1.44)b	326.73	<.001
Severe Physical Aggression	0.00 (0.09)a	0.20 (0.60)a	0.58 (1.49)a	124.81	<.001
Injury
Minor Injury	0.00 (0.07)a	0.20 (0.50)a	0.37 (0.69)a	149.03	<.001
Severe Injury	0.00 (0.06)a	0.05 (0.25)b	0.28 (0.85)a,b	83.13	<.001
Mental and Physical Health Indicators
CES-D Depression Score	8.38 (8.35)a	15.92 (12.60)a	27.10 (14.91)a	294.13	<.001
PCL PTSD Score	18.64 (5.64)a	26.21 (13.52)a	43.09 (25.78)a	504.55	<.001
CHIPS Score	19.42 (17.54)a	30.92 (24.00)a	51.60 (31.25)a	242.99	<.001
SF-4 Poor Health Score	2.65 (2.68)a	4.48 (3.70)a	6.85 (4.03)a	198.08	<.001

Note. Means in the same row sharing subscripts are significantly different from each other, according to Tamhane’s T2 post-hoc test, which we used due to violations of the homogeneity of variance assumption.

Hypothesis 2: Health Differences Between Intimate Terrorism and Situational Couple Violence Victims

Our second hypothesis was that male IT victims would have worse physical and mental health in comparison with male SCV victims. We conducted a MANCOVA with victimization type (no violent victimization, SCV victimization, IT victimization) as the independent variable and the four health indicators (CES-D score, PCL score, CHIPS score, and SF-4 score) as the dependent variate. We controlled for significant demographic covariates (income, whether they were currently in a relationship, age, education, and racial/ethnic minority status). The MANCOVA was significant, Wilks’ Δ = .662, F (8, 4244) = 121.62, p < .001, partial η² = .187. The bottom portion of Table 4 shows the results of the follow-up ANCOVA analyses comparing the groups on the health indicators. For ease of interpretation, we presented the unadjusted means. Consistent with this hypothesis, male IT victims had significantly worse mental and physical health on all measures than male SCV victims. In addition, both IT and SCV victims had significantly worse mental and physical health than men who experienced no violent victimization.

Hypothesis 3: Differences in Abuse Severity and Health Between Mutual Violent Control and Intimate Terrorism

Our final hypothesis was that men involved in MVC would have worse mental and physical health problems than either perpetrators or victims of IT, and relationships characterized by MVC would be characterized by more severe PV.

A series of MANCOVA analyses were conducted to test this hypothesis. In the first MANCOVA, we compared the IT perpetration, IT victimization, and MVC groups on the perpetration of the various forms of PV, while controlling for significant demographic covariates (age and education). The MANCOVA was significant, Wilks’ Δ = .326, F (22, 960) = 32.84, p < .001, partial η² = .429, with follow-up ANOCVAs showing that the differences were significant for all of the variables that comprised the dependent variate, except actual legal/administrative aggression. Table 5 presents the results of the follow-up ANCOVAs. For ease of interpretation, we presented the unadjusted means. The top portion shows differences in men’s perpetration of various forms of PV across IT perpetration, IT victimization, and MVC groups. When we compare the MVC group to the IT perpetration group, we see that men involved in MVC perpetrated significantly more types of each form of PV, with the exception of actual legal/administrative aggression and minor sexual aggression, where there were no statistical differences. Similarly, in comparison to male IT victims, men involved in MVC perpetrated significantly more types of each form of PV, except actual legal/administrative aggression.

Table 5.

Differences in Partner Violence Variety Scores and Health Indicators Within IT Relationships, according to follow-up ANCOVA tests of significance

	Perpetrates IT with non-violent or VR partner	Victim of IT, but not violent or uses VR	Mutual Violent Control
	N = 13	N = 429	N = 61	F	p
	M (SD) or %	M (SD) or %	M (SD) or %
Male Participants’ Perpetration (Variety Types)
Non-Physical Aggression
Minor Psychological Aggression	3.23 (0.73)a	2.50 (1.27)a,b	3.57 (0.78)b	14.03	<.001
Severe Psychological Aggression	0.77 (1.01)a	0.56 (0.90)b	2.44 (1.48)a,b	82.81	<.001
Controlling Behaviors	3.46 (0.66)a	0.54 (0.85)a	5.36 (2.48)a	393.18	<.001
Threatened Legal/Administrative Aggression	0.75 (0.97)a	0.15 (0.48)b	1.77 (2.12)a,b	71.88	<.001
Actual Legal/Administrative Aggression	0.08 (0.28)	0.11 (0.39)	0.31 (0.76)	2.86	.058
Sexual Aggression
Minor Sexual Aggression	0.77 (0.73)a	0.15 (0.41)a,b	1.21 (0.91)b	92.68	<.001
Severe Sexual Aggression	0.31 (0.63)a	0.04 (0.24)b	1.43 (1.76)a,b	93.37	<.001
Physical Aggression
Minor Physical Aggression	2.00 (1.15)a	0.80 (1.13)a	3.23 (1.62)a	87.13	<.001
Severe Physical Aggression	0.46 (0.66)a	0.26 (0.73)b	2.85 (2.90)a,b	93.14	<.001
Injury
Minor Injury	0.31 (0.63)a	0.28 (0.59)b	1.03 (0.95)a,b	26.62	<.001
Severe Injury	0.15 (0.55)a	0.12 (0.40)b	1.39 (1.80)a,b	61.32	<.001
Male Participants’ Victimization (Variety Types)
Non-Physical Aggression
Minor Psychological Aggression1	3.08 (1.19)	3.76 (0.48)	3.79 (0.55)	4.12	.017
Severe Psychological Aggression2	1.54 (1.27)a,b	3.06 (1.05)a	3.13 (1.08)b	9.09	<.001
Controlling Behaviors	1.42 (0.90)a	5.19 (1.81)a	6.25 (2.13)a	35.44	<.001
Threatened Legal/Administrative Aggression2	2.17 (1.99)a,b	1.88 (0.09)a	2.06 (0.27)b	4.89	.008
Actual Legal/Administrative Aggression	0.69 (1.25)a	2.58 (1.98)a,b	1.36 (1.88)b	6.01	.003
Sexual Aggression
Minor Sexual Aggression	0.23 (0.44)a,b	0.69 (0.79)a	0.95 (0.88)b	5.16	.006
Severe Sexual Aggression	0.08 (0.28)a	0.63 (1.10)a	1.39 (1.76)a	14.09	<.001
Physical Aggression
Minor Physical Aggression	1.62 (1.50)a	3.83 (1.26)a	4.26 (1.03)a	22.15	<.001
Severe Physical Aggression	1.08 (1.26)a	2.65 (1.95)a	3.95 (2.44)a	14.52	<.001
Injury
Minor Injury2	0.62 (0.96)a,b	1.31 (0.85)a	1.43 (0.81)b	4.04	.018
Severe Injury	0.23 (0.60)a	0.79 (1.02)a	1.64 (1.75)a	20.34	<.001
Health Indicators
CES-D Depression Score2	14.68 (10.31)a	27.49 (14.93)a	23.77 (14.38)	8.75	<.001
PCL PTSD Score2	27.09 (10.89)a	44.20 (16.58)a,b	34.95 (17.27)b	11.64	<.001
CHIPS Score	22.28 (15.40)a,b	52.13 (30.87)a	47.13 (34.28)b	10.08	<.001
SF-4 Poor Health Score	3.38 (2.59)a,b	6.93 (4.03)a	6.23 (4.14)b	7.35	.001

Note. Means in the same row sharing subscripts are significantly different from each other, according to Tamhane’s T2 post-hoc test, which we used due to violations of the homogeneity of variance assumption.

¹Although the follow-up ANCOVA was statistically significant, post-hoc Tamhane T2 tests to correct for violations of the homogeneity of variance assumption found no differences among the means of the three groups.

²Tukey’s post hoc tests were used to determine where there were significant differences because these analyses did not violate the homogeneity of variance assumption.

The MANCOVA comparing IT groups on the victimization dependent variate, controlling for significant covariates (income, presence of minor children, relationship length, and whether they reported on a current relationship), was also significant, Wilks’ Δ = .711, F (22, 880) = 7.45, p < .001, partial η² = .157. Follow-up ANCOVAs showed that the differences were significant across all victimization variables in the dependent variate. The middle portion of Table 5 presents the unadjusted means of the dependent variables, along with the results of the follow-up ANCOVAs. For victimization, men involved in MVC were victimized by significantly more types of each form of PV, in comparison to male IT perpetrators, with the exception of minor psychological aggression and actual legal/administrative aggression, for which there were no differences. Moreover, male IT perpetrators were the victims of significantly more threatened legal/administrative aggression than men involved in MVC. When comparing men involved in MVC to male IT victims, we see a similar pattern in that men involved in MVC were victimized by significantly more types and forms of PV, with the exception of minor psychological aggression, severe psychological aggression, threatened legal/administrative aggression, minor sexual aggression, and minor injury. In addition, male IT victims sustained significantly more actual legal/administrative aggression than men involved in MVC. Thus, it appears that on the whole and consistent with our hypothesis, men involved in MVC are involved in relationships with more types of PV than either male IT perpetrators or victims.

In our third MANCOVA, we used the same independent variable, with the four health indicators (CES-D score, PCL score, CHIPS score, and SF-4 score) as the dependent variate. We entered significant demographic covariates (income, whether they reported on a current relationship, age, and education) as control variables. The MANCOVA was significant, Wilks’ Δ = .937, F (8, 920) = 3.79, p < .001, partial η² = .032. Follow-up ANCOVAs, presented in the bottom portion of Table 5, showed that the results were significant for all health indicators. Men involved in MVC did not significantly differ from male IT perpetrators in either PTSD or depression symptom severity. However, they did have significantly poorer physical health, as indexed by the CHIPS and SF-4. In comparison to male IT victims, men involved in MVC did not differ in depression symptom severity or either index of physical health. However, male IT victims had significantly more symptoms of PTSD and were significantly more likely to reach the clinical cut-off on the PTSD measure. The only consistent significant differences were between the IT perpetrators and victims, in that IT victims had significantly poorer mental and physical health on all measures. Moreover, IT victims had significantly worse PTSD symptom severity than men involved in MVC, which is not consistent with our hypothesis.

Discussion

The current study tested three sets of hypotheses regarding Johnson’s (2008) typology of domestic violence. We used a sample of male victims and a population-based sample of men, and we assessed both victimization and perpetration, to test whether: IT victimization was associated with greater severity and frequency of various types of PV and worse mental health among men, and whether, in comparison to IT, MVC was associated with greater severity and frequency of various types of PV and worse mental health among men. Overall, we found mixed support for our hypotheses.

Intimate Terrorism versus Situational Couple Violence

We found support for Johnson’s (2008) assertion that because IT and SCV are distinct forms of PV, then they should have different dynamics and consequences. Moreover, we extended these findings by showing that the differences extend to men as victims and to other forms of PV as well.

Consistent with research on female PV victims (Graham-Kevan & Archer, 2003; Johnson, 2006; Johnson & Leone, 2005), we found that in comparison to male victims of SCV, male victims of IT reported greater levels of victimization from minor physical aggression and severe physical aggression, and they reported significantly more types of both minor and severe injuries. Thus, it seems that we can broaden Johnson’s (2008) assertion about the PV experienced by victims of IT to male victims as well. This finding of greater victimization extended to other forms of PV as well, in that victims of IT also reported significantly more types of sexual aggression (both minor and severe), psychological aggression (both minor and severe), controlling behaviors, and legal/administrative aggression. It appears that victims of IT not only experience more physical aggression and injuries, but they experience more sexual and non-physical forms of PV as well. We recommend that this finding should also be tested among female IT and SCV victims.

Also extending Johnson’s work, we found that in comparison to male SCV victims, male IT victims also perpetrated a significantly greater number of types of all forms of PV, with the exceptions of minor physical, minor psychological, and minor sexual aggression. Thus, it appears that not only are the IT victims in more danger, but perhaps their perpetrators are as well. This greater level of violent perpetration by the IT victims could be due to the male IT victims’ greater use of violent resistance in order to protect themselves from an abusive partner, but it could also be due to victims also being perpetrators of IT (i.e., MVC). Because on average, the number of types of controlling behaviors used among the male victims was 1.15, the majority of this violence is likely violent resistance. However, our findings on MVC (discussed below) suggest that a minority of it is also due to MVC. Specifically, of the 490 men who reported IT victimization, 12.4% reported MVC. This finding of greater PV perpetration for most types of PV should be tested among female IT and SCV victims as well. Future studies on women should include information regarding their perpetration of PV, data that is commonly missing from studies of PV in women, particularly in agency samples (Straus, 2011).

Consistent with our hypotheses derived from Johnson’s (2008) typology, we found that male IT victims had significantly worse mental and physical health than male SCV victims. This is consistent with the literature on female victims (Johnson & Leone, 2005; Leone et al., 2004), one previous study of male IT victims (Laroche, 2008), and with studies that focused on male victims of PV who seek help (Hines & Douglas, 2011c). The latter study found that male victims of PV who seek help have rates of PTSD (57.9%) that are exponentially greater than men from a population-based sample who experienced physical PV (8.2%), and that the male victims’ rates were similar to those of agency samples of female victims. Our results replicate those findings in that we found significantly greater levels of PTSD symptom severity among male IT victims than among male SCV victims. In addition, we found significant differences between IT and SCV victims in depression symptom severity and indicators of physical health.

Although Frye and colleagues (2006) suggest that because of social and personal power-level differentials between men and women, violence and control tactics would necessarily have differential impacts, our data suggest that this is not likely the case. In fact, almost half of the male IT victims exceeded the clinical cut-off on our measure for PTSD, and almost three-quarters on our measure for depression, with 62.1% of these men reporting on relationships that occurred on average three years previously. Although a true test of Frye et al.’s supposition would involve longitudinal analyses of both men and women, the fact that our findings on men are consistent with the extant literature on women suggests that both men and women greatly suffer both mentally and physically from IT victimization.

Intimate Terrorism versus Mutual Violent Control

Johnson (2008) contended that MVC is very rare, whereas others have shown that as the severity of violence increases, the likelihood that the violence is bidirectional increases (e.g., Straus, 2011). Indeed, when researchers measure both victimization and perpetration among men and women, MVC is much more common than Johnson theorized (e.g., Ehrensaft et al., 2004; Graham-Kevan & Archer, 2003b; Hines & Douglas, 2016). However, few researchers have tested whether – other than physical PV – various forms of PV are more frequent and severe in relationships characterized by MVC versus IT. We hypothesized that consistent with the research on physical PV (e.g., Straus, 2011), other forms of PV would also be more frequent and severe in MVC relationships, and on the whole, the results were consistent with our hypothesis. Men involved in MVC were involved in relationships with more types of PV than either male IT perpetrators or victims. This is likely due to escalating abuse – when one violent partner abuses the other, the likelihood that the other violent partner would be abusive in return is high. In relationships were one partner is non-violent, the likelihood of the non-violent partner retaliating with abuse is much lower.

We also hypothesized, based on research showing that people in bi-directionally violent relationships suffer more injuries and mental health problems than victims of uni-directional violence (Straus & Gozjolko, 2014; Whitaker et al., 2007), that men involved in MVC would have more problematic physical and mental health indicators than men involved in IT. Our hypothesis was not supported. In fact, men who were the victims of uni-directional IT had the worst health indicators. In comparison to perpetrators, they suffered more severe depression symptomatology, PTSD symptomatology, and physical health problems, and in comparison with MVC victims, they suffered more severe PTSD symptomatology. No other differences were significant.

The reasons why male IT victims suffer worse health than male MVC victims should be the subject of future research. Based on our findings, these differences cannot be due to differences in PV severity, because men involved in MVC experienced more frequent and severe PV overall. The health differences might be due to perhaps feelings of helplessness in that one is not able or willing to fight back, and may be unable to get the help one needs to prevent or flee the abusive situation. Indeed, research shows that male victims of PV have a difficult time finding the help they need – they are often turned away, accused of being the real abuser, blamed for the abuse, and even ridiculed when seeking help (Douglas & Hines, 2011; Hines & Lysova, 2016). These reactions often come from the very services designed to help PV victims -- such as DV agencies, DV hotlines, and the police – and such negative reactions from helping professionals can lead to poorer mental health (Douglas & Hines, 2011). Future research should investigate whether the lack of resources available to male IT victims contributes to their overall poor health in comparison with male MVC victims, and whether such findings also apply to female IT victims.

Limitations

Future research on Johnson’s (1995, 2008) typology should take into consideration the limitations of our study. As discussed in Hines and Douglas (2016), our sampling procedure suffers from the same sampling bias that Johnson had, in that we sampled only men and one of our samples was comprised of men who sought help due to their female partner’s violence. Thus, our sample was limited in that it did not include women, nor did it include the partners of the men. We chose not to include the partners of the men in the victims sample because of safety concerns, and therefore, researchers should consider how we can obtain data from both members of the dyad without risking the safety of one or both members.

The limitations of our male victims sample also need to be considered in future research because of potential limitations on generalizability. We specifically recruited our sample of male PV victims so that it would be comparable to the majority of studies on battered women, which typically recruit battered women who sought help for PV victimization. Thus, we also required that the male PV victims sought help. This limits generalizability because it is likely that the majority of male PV victims do not seek help. It is important to note that the research on male help-seeking for a variety of mental and physical health concerns shows that men have to overcome several societal and internal barriers to seek help, which are compounded when the problem is viewed to be non-normative by society and something that men should be able to handle themselves (Addis & Mahalik, 2003), as would be the case for male PV victims. Indeed, Laroche’s (2008) analysis of gender differences in help-seeking among victims of severe IT show that male victims are significantly less likely to seek help. Related to this issue of generalizability and sampling method, the male victims had to have seen our advertisement on the Internet or been alerted to our study by a service provider who saw our advertisement online. In addition, they had to complete the study online; male victims without access to the Internet were excluded. Future studies should aim to recruit men who may have sought help from other sources of support or who may not have sought help at all to investigate any possible differences in their experiences.

It is also important to understand that although we tested whether some of the predictions from Johnson’s (1995, 2008) typology can extend to men as victims, those tests were based on the underlying assumption that the typology itself is actually valid. Future research may need to use more sophisticated person-centered analyses and taxometric approaches (e.g., latent class analysis, mixture modeling) to determine whether these groups truly emerge as distinct subgroups in the population. Another option would be to use an actor-partner interdependence model to capture the dyadic nature of PV. Further, this research needs to examine the typology in both genders simultaneously to determine whether there is a significant gender by type interaction. These analyses may allow for an understanding on how men and women are similar or different in their perpetration and/or victimization of PV.

Another limitation is that this study was solely based on the self-reports of the men, which can lead to two potential problems: (1) shared method variance, which may cause inflated correlations because the same person reported on PV and men’s health; and (2) inaccurate reporting of PV victimization and perpetration. For the former issue, it is possible that male PV victims who report negative behaviors by their partners are likely to also report/notice negative health in themselves in comparison to men who do not experience PV. For the latter issue, research shows that the typical pattern is under-reporting of one’s own use of undesirable behavior, but not of one’s partner’s undesirable behavior (Woodin, Sotskova, & O’Leary, 2013). However, even for the partner’s behavior, under-reporting is common because victims tend to feel embarrassed or humiliated by being abused (Follingstad & Rogers, 2013). Nonetheless, it is likely that the current study underestimated the men’s use of all forms of PV, and therefore, future studies should strive to obtain information about PV from multiple informants, including both partners, to gain a more accurate understanding of the dynamics of the relationship.

There is also a need for additional research examining motivation for physical PV and controlling behaviors. Research shows that men and women who perpetrate PV have similar underlying motivations (Medeiros & Straus, 2006). This is particularly necessary given the category of VR, which assumes that the underlying motive is self-defense. Because the present study and other studies testing Johnson’s theory (e.g., Johnson, 2008) did not have a measure of motivation for violence, the VR type is solely theoretical. There is no way to know whether someone’s use of physical PV was due to resistance based solely on their level of controlling behaviors.

The use of modified instruments, specifically the PCL and the SF-4, may also be a limitation of this study. Although the scales did have excellent internal consistency, eliminating items from the original scale impacts our ability to compare our findings to other studies using the original scales and perhaps alter its validity to some extent. A final limitation is that our study is cross-sectional; therefore, we cannot establish the sequence of events of PV in this study. We need longitudinal studies to establish the temporal ordering of PV victimization and the health indicators assessed in this study. Without such studies, causal inferences between PV victimization and mental health cannot be made.

Implications

Despite these limitations, this study has important implications for both policy and practice related to PV. Our study adds to accumulating evidence that we can no longer consider patriarchal theory as the principal theory to explain PV or IT, or to guide policy and practice. Patriarchal theory, as applied to Johnson’s (2008) typology, contends that men are the predominant perpetrators of IT and use IT to maintain dominance in their relationship with their female partner, a dominance to which they are entitled based on patriarchal norms. Research on both domestic violence agencies (Hines & Douglas, 2011a) and male victims who sought help (Douglas & Hines, 2011) show that help is not readily available for male PV victims; indeed, they are often turned away. Johnson (2008) noted that “once women decide to seek help, effective help is often the case” (p. 43). Unfortunately, the same is not true for male PV victims, ostensibly because the underlying theory guiding policy and practice in this field is still patriarchal theory (Dutton & Corvo, 2006). Because our study shows that male IT victims have poorer mental and physical health in comparison to SCV victims, men involved in MVC, and non-victims, it is imperative that PV services become readily available to them.