Abstract

Violence Types and Injuries

In a previous study of police reports of men’s IPV, the most common acts of violence reported were slapping/pushing and punches to the face or body (Dobash & Dobash, 1984). For men and women arrested for IPV, they were equally likely to commit at least one act of severe aggression (e.g., punching, choking, kicking), with rates of 87% for men and 92% for women, although men were more likely to use a higher number of severe violent tactics in the IPV incidents compared to women (Busch & Rosenberg, 2004).

A further consideration of the developmental systems model is the impact of aggression toward a partner. Gender differences have been found in the likelihood of injury. For men and women arrested for IPV, women were as likely as men to inflict injuries, but more women than men reported being victimized or injured (Busch & Rosenberg, 2004). For patients seeking emergency medical services for IPV injuries, women reported more frequent and severe injuries in the past year and lifetime than men (Phelen et al., 2005). Further, in the United States, women are more likely to be killed by intimate partners than by any other type of perpetrator (Campbell et al., 2003). Across studies involving IPV arrests, 19% – 25% of the sample had injuries that required medical attention (Busch & Rosenberg; Duncan et al., 1999). In a study of police reports, Duncan et al. found that 17.4% of the incidences resulted in injuries, and of those injured, 90% were women and 10% were men. About one half of the injuries involved bruises, scratches, or pain, with other common injuries including swelling and redness. Severe injuries (e.g., broken bones, loss of consciousness) were relatively rare. Less than 1% of men and women arrestees have been found to inflict serious injuries in IPV incidents (Martin, 1997). One cause of differing findings regarding severity of injuries and arrests among studies of police reports is that arrest policies differ by state in the United States, with some having mandatory arrest policies and some not. As more arrests are likely to occur because of mandatory arrest policies, the severity of injuries might be expected to be lower for IPV arrest incidents in these areas. The State of Oregon, where the majority of couples in the current study resided, has a mandatory arrest policy.

Mutual Versus One-Sided Aggression

There is substantial evidence that much physical violence in couples is bidirectional or mutual. Overall, prevalence rates for IPV toward a partner among young adults have been found to be relatively high and similar for men and women, although slightly higher for women (e.g., Archer 2000; Fergusson, Horwood, & Ridder, 2005). Perpetration rates are estimated to range from about 36% to 51% for young women and 22% to 43% for young men (Moffitt & Caspi, 1999). Similar to nonphysical patterns of mutual negativity, such as contempt/belligerence and reciprocated aversive behavior observed in the interactions of aggressive couples (e.g., Jacobson et al., 1994), much IPV is best described as mutual, particularly in young couples (e.g., Moffitt, Caspi, Rutter, & Silva, 2001; Stets & Straus, 1990; Whitaker, Haileyesus, Swahn, & Saltzman, 2007). Aggressive couples reported mutual physical aggression in about one half of the incidents in both the National Family Violence Survey (Stets & Straus) and on the National Longitudinal Survey of Adolescent Health at ages 18–28 years (Whitaker et al., 2007). Men and women in mutually aggressive couples report about equal frequency and severity of physical aggression perpetrated and sustained (Gray & Foshee, 1997; Henton, Cate, Koval, Lloyd, & Christopher, 1983). Within a birth cohort of young adults, women were more frequently the perpetrators in relationships defined as nonclinically abusive whereas both men and women were the perpetrators in those defined as clinically abusive, involving aggression linked to psychopathology (Ehrensaft, Moffitt, & Caspi, 2004). Physical violence is frequently initiated by women according both to self-reports (Morse 1995; Whitaker et al., 2007) and to behavior observed during problem-solving interactions (Capaldi et al., 2007). Reports of relative levels of initiation of physical aggression by men and women range from about equal responsibility (Henton et al.) to more initiation by the women found in a birth cohort of 25 year olds, where 58% of the women compared to 32% of the men reported that they initiated aggression as perpetrators (Fergusson et al., 2005).

There is some evidence that even more frequent and serious aggression, which has been hypothesized to be one sided (Johnson, 1995) also may be mutual. Capaldi and Owen (2001) found that the proportion of couples where both partners were frequently aggressive (defined as a cutoff score of 19 or more acts) was 6 times higher than expected by chance. In addition, injuries were likely to have occurred for both partners at three times the rate expected by chance for such mutually aggressive couples. Findings from Whitaker and his colleagues (Whitaker et al., 2007) also support the argument that injuries are more likely to occur when aggression is mutual rather than one sided. Further, McDonald, Jouriles, Ramisetty-Mikler, Caetano, and Green (2006) found that 67% of women in domestic violence shelters (who had to have experienced at least one act of partner physical violence in the past year) had engaged in severe violence toward a partner according to self-reports. Thus, there is evidence of a high prevalence of mutual violence even among couples where more serious violence occurs.

The Current Study

The current study presents a unique opportunity to examine hypotheses related to arrests for IPV and aggression toward a partner outside of arrest occasions in a community sample of at-risk men and their romantic women partners. The study focused on IPV incidents that resulted in the arrest of the OYS young man or his partner. The IPV arrest incidents were expected to be predominantly arrests of the man (in keeping with the prevalence of such arrests for men and women nationally; Greenfeld et al., 1998) and, therefore, to reflect more violence by the men and a greater likelihood of injury for the women. However, it was also expected that the types of injury would be varied, with the majority being minor, similar to the findings from survey studies. We also hypothesized that there would be an association between IPV arrests and aggression toward a partner outside of the arrest incident. It was predicted that couples experiencing an IPV arrest incident would show involvement by both the men and women in IPV (i.e., psychological aggression and physical aggression) assessed by self- and partner reports and observational data, thus indicating some mutual engagement in aggression. It was also expected that both men and women in couples experiencing an arrest incident would show involvement in severe aggression toward a partner (assessed by self- and partner reports). However, given that the men arrested in the IPV incidents were expected to be some of the more severely aggressive men, and as men show higher levels of antisocial behavior than women in general (Blumstein, Cohen, Roth, & Visher, 1986), the men were expected to have significantly higher levels of physical, psychological, and severe aggression toward a partner outside of the arrest occasion, compared with the women. Sexual coercion was not within the scope of the current study. For a study regarding sexual coercion for the OYS men see Teten, Hall, and Capaldi (2008).

Method

Measures

Ratings of the IPV police reports

A rating system was created on the basis of information available in the first 12 police reports reviewed (Capaldi et al., 2005). Items included in the current study were (a) the prevalence of each type of violence perpetrated, (b) injuries experienced for both the men and women, and (c) the mutuality of the physical fight and who appeared to have made the first physical contact. Two trained research staff who were blind to research hypotheses of the study independently coded each incident. Disagreements were resolved by rereading the report and a discussion between the two coders.

Coding of the interaction tasks

The Family and Peer Process Code (FPPC; Stubbs et al., 1998) was used to code the couples’ interaction. All coders were professional research assistants who received approximately 3 months of training. The behavioral and emotive contents of each couple's interaction were recorded in real time. To assess coder reliability at each wave, two coders independently coded a randomly selected minimum of 14% of the tasks. The overall content and affect kappas (Cohen, 1960) ranged from .73 to .82 across the three time points, the kappas for psychological aggression ranged from .63 to .69 and for physical aggression from .53 to .65. Physical aggression occurred less often in interactions at the older ages; therefore, reliability of the codes dropped somewhat. For reports of the prevalence of physical aggression in the interaction tasks, see Capaldi et al., (2007).

Constructs

As described by Patterson, Reid, and Dishion (1992), the strategy at the outset of the OYS was to develop several indicators for each theoretical construct from multiple agents and methods. In building scales and constructs, items constituting the indicator or scale had to show internal consistency (i.e., an alpha of .6 or higher and an item-total correlation of .2 [p approximately .05] or higher). Second, items had to converge with the other indicators designed to assess the construct. These criteria ensured that items and indicators assessing the theoretical construct were used for the psychological aggression construct. However, for physical aggression, these criteria were not considered appropriate. It is not necessarily the case that if one type of physical aggression is present for a couple (e.g., punching) that another is likely to be present (e.g., hitting with an object), but there is no doubt that both of these behaviors are acts of physical aggression. Using the alpha criteria above, some acts of physical aggression that are more rare could be excluded from the construct inappropriately. Similarly, the occurrence of physical aggression during the problem-solving discussions was relatively infrequent and, therefore, showed lower convergence with other measures of physical aggression. Again, it was considered to be significant and worthy of inclusion in the measure of physical aggression. Technical reports providing additional detail are available from the Oregon Social Learning Center upon request.

Physical aggression

The physical aggression constructs for the men and women were composed of self- and partner reports, coder ratings, and the observed interactions. Each indicator was standardized using a 0–3 scale and then averaged to produce construct scores. Sample items and internal consistency information for each instrument (e.g., Adjustment with Partner Questionnaire) assessing the construct are described in Table 1. For observed physical aggression, the overall rates per minute of two content codes (Physical Aggression and Physical Attack) in four affects (neutral, distress, aversive, and sad) were used from the FPPC coding of the interaction task. The observed Physical Aggression Code reflected items on the Conflict Tactics Scale (CTS; Straus, 1979) that were considered to be of low severity (e.g., grabbing, slapping, shoving). The observed Physical Attack Code included moderate and severe aversive physical contact (e.g., kicking, punching, hitting with an object with force) in only neutral or negative affect (i.e., occurrences in positive affect were excluded). Severity of behaviors in the two combined codes ranged from a slight shove to hard hits (e.g., shoving the partner on the shoulder, poking with a pencil, hand slaps, kicks, hitting across the head).

Table 1

Constructs and Scales

For the analyses conducted on severe physical aggression only, items reported by the men and the women to their own and to their partner’s frequency of kicking, beating up, choking and burning from the Adjustment with Partner Questionnaire were used. Thus, both self- and partner reports were used for each of their behaviors.

Psychological aggression

This construct was measured using the same assessment agents and modes as for physical aggression (Table 1). Each indicator was standardized using a 1–5 scale and then averaged to produce construct scores. Observed scores for psychological aggression included the rate per minute of three behaviors in neutral or negative affect: Negative Interpersonal Behavior (i.e., verbal expressions of disapproval of the partner), Verbal Attack (i.e., name calling, threats, and specific humiliation of the partner), and Coerce (i.e., demand for behavior change coupled with a threat of personal injury). Further details regarding the measures for the physical and psychological aggression constructs are provided in Table 1.

Results

IPV Arrests

Police records for 47 IPV incidents involving an OYS man and any woman with whom he was in a romantic relationship were located. Of these arrests, 85% or 40 involved an arrest of the OYS young men, 6% or 3 involved an arrest of women, and 9% or 4 involved a dual arrest of OYS men and partners. The 44 IPV arrests of the OYS men were attributed to a total of 28 men -- 19 men with 1 IPV arrest and 9 men with 2 or more IPV arrests (2 men had 3 arrests, 1 man had 4 arrests, and 1 man had 5 arrests). The women arrested for an incident with an OYS man were each arrested one time for such an incident. The N for the analyses of IPV arrests was the 47 incidents. Because of nonindependence in the data because (a) the 47 incidents involved 31 men and 30 women and (b) multiple types of violence and injury could occur during one incident, significance tests were not conducted on the IPV arrest data.

Types and severity of violence

The prevalence and the type of any violence occurrence reported in the IPV incidents is shown in Table 2. The most common type of violence included pushing, grabbing, shoving, and slapping, which would be considered less severe and minor violence. Men appeared to be more likely than women to participate in violence in the incidents. There were some types of violence such as hitting partner with an object, restraining or choking partner, and burning partner that were only carried out by the men. Examples of violence by the men included several cases of pinning the woman down, banging the woman with a door or hitting with a car door, pushing her so she hit her head on a shelf, dragging her across a street, holding a hand over her mouth or mouth and nose, twisting a wrist, and threatening with a gun.

Table 2

Prevalence of Any Violence and Types of Violence Reported to be Perpetrated in the Domestic Violence Incidents^*

Injuries

During the IPV incidents, a minority of the men and a majority of the women were injured. The prevalence of any injury and each type of injury is shown by gender in Table 3. The most common injuries reported included bruising, cuts, and scrapes, and the rates for these injuries were two to four times greater for the women than the men. There were some injuries, such as being knocked down and knocked unconscious, that only occurred for women. There were no cases of broken bones, and only two women received medical treatment. For example, one case involved a severe laceration to the woman’s foot when the man ran over it with his car when trying to leave the house during an argument.

Table 3

Prevalence of Any Injuries and the Types of Injuries Sustained^*

Mutuality of the fight and first physical conflict

The relative balance of the men’s and women’s contributions to the physical fight during the incidents was coded on a 7-point scale from all or almost all the man (7) to all or almost all the woman (1). The physical aggression was coded as all or almost all from the man in 51% of the cases, with a further 30% being more the man than the woman. The remaining 19% was split between 11% appearing mutual, 2% all or almost all the woman, and 6% being predominantly the woman. Regarding first physical contact, the man was rated as making the first physical contact in 66% of the cases, the woman in 15% of the cases, in 17% of the cases it was ambiguous, and in 2% not available.

Association Between IPV Arrests and Aggression Toward a Partner

The prevalence of physical and psychological aggression toward a partner, as well as severe physical aggression toward a partner, was examined for men and women who had been involved in an IPV arrest versus those who had not. All couples in both groups had engaged in psychological aggression toward a partner, at least to the extent of endorsing one such item. Regarding physical aggression, 74% of the men and 87% of the women who had experienced any IPV arrest were reported or observed to use some physical aggression in their relationship, compared to 43% of the men and 60% of the women in couples that had not experienced an IPV arrest. Regarding severe physical aggression, 44% of the men and 39% of the women who had experienced any IPV arrest were reported to use severe physical aggression, compared to 5% of the men and 10% of the women in couples that had not experienced an IPV arrest. It was hypothesized that there would be an association between arrests for IPV and aggression toward a partner outside of the arrest incident, and it was also predicted that couples involved in an IPV arrest incident would show involvement of both the men and women in IPV outside the arrest incident, but that the men would show higher levels of aggression than the women. To test these hypotheses, levels of physical and psychological aggression overall were examined by arrest group and by gender, and levels of severe physical aggression (e.g., kicking, punching, choking) were also examined. First, a multivariate analysis of variance was conducted with the men’s and women’s physical and psychological aggression construct scores. The two between-participants factors were IPV arrest group (couples with no IPV arrests versus couples with one or more arrests of the man, a woman, or both) and gender. Shown in Table 4, Panel I are the mean levels of aggression toward the partner by IPV arrest group and gender. The multivariate IPV arrest group effect, Pillais-Barlett trace (2, 177) F = 6.88, p < .001, and the gender effect, Pillais-Barlett trace (2, 177) F = 8.84, p < .001, were significant. The IPV arrest group by gender effect, however, was not significant, Pillais-Barlett trace (2, 177) F = 0.07, p = .929. As expected, there was an association between IPV arrests and aggression toward a partner -- couples involved in one or more IPV incidents had higher levels of physical and psychological aggression than couples with no IPV incidents. Women had higher levels of physical and psychological aggression than the men overall. Men that were arrested did not have higher levels of aggression toward a partner overall compared to the women involved in the incidents. Second, an analysis of variance was conducted for the men’s and women’s mean frequencies of severe physical aggression (each being a combination of self- and partner reports). Again, the two between-participants factors were IPV arrest group and gender. Findings (Table 4, Panel II) indicated that the IPV arrest group effect, F (1) = 37.48, p < .001, was significant. However, neither the gender effect, F(1) = 1.35, p = 0.247, nor the IPV arrest group by gender effect, F(1) = 0.10, p = .750, were significant. Thus, whereas the IPV arrest group engaged in higher mean levels of severe physical aggression than the no-arrest group, there was no significant difference between the men and women in levels of severe physical aggression toward a partner for either of the groups.

Table 4

Mean Levels of Men’s and Women’s Aggression by IPV Incident Groups

Discussion

Violence and injuries during IPV arrest incidents, as recorded in police reports, were examined for both men and women in an at-risk community sample of couples. Types and severity of IPV arrest incidents were examined and compared with nonofficial aggression toward a partner using self- and partner reports and observational data. Of the approximately 200 men in the larger OYS, 28 men were involved in 47 incidents of IPV resulting in the arrests of 1 or both of the partners (including 4 arrests of women only and 3 dual arrests). For the men arrested, one third were multiple IPV offenders (cf., Feder & Henning, 2005). Although significance tests could not be conducted, twice as many arrest incidents involved some violence by the men compared to women, and four times as many women as men were reported to be injured.

There was convergence between official IPV and nonofficial aggression toward a partner: couples with IPV arrests had elevated levels of physical and psychological aggression by both partners compared with couples not involved in an arrest incident. A particularly important finding of the study was that, although men were more likely to be arrested in the context of the official IPV, nonofficial data indicated that the men in the group experiencing an official IPV incident were not significantly higher than were their women partners in physical and psychological aggression or in mean levels of more severe physical aggression. In fact, the women were more likely to show higher levels of overall physical and psychological aggression (although not of severe physical aggression) in the context of nonofficial aggression toward a partner.

According to the police reports of the IPV arrest incidents, men were likely to initiate physical contact than women and the physical force used in the IPV incidents ranged from minor to severe. The distribution of IPV acts is consistent with reports that IPV incidents resulting in arrest do not necessarily involve severe violence (e.g., Duncan et al., 1999). There was a relatively high prevalence of minor physical contact (e.g., pushing and shoving, slapping) and a lower prevalence of severe types of violence, as is reflected in the injuries that resulted. As predicted, the men used more violence in general and more severe violence than the women during the IPV arrest incident.

The view that official IPV incidents involve much more severe injuries overall than incidents described in community samples was not supported. Any injury during the IPV incident was reported for 21% for the men and 83% of the women, indicating that some type of injury is a common impact of IPV incidents involving the police. The prevalence of severe injuries and injuries requiring medical attention, however, were relatively low. There was one case involving loss of consciousness and no cases of broken bones. Only two women and none of the men visited a physician or hospital for their injuries, which is lower than 19% – 25% for some other studies involving IPV arrests (Busch & Rosenberg, 2004; Duncan et al., 1999). With only severe injuries brought to the attention of the medical services, police reports can be a source of information about less severe injuries and how these injuries occur (Duncan et al.). Injuries reported in the IPV incidents were similar in type and severity to those described by Capaldi and Owen (2001) using self-reports at about age 21 years for the men and their partners and to those reported at the two later ages included in the current study (not detailed here).

As expected, women experienced more injuries than the men during the arrest incident. This would be expected from the greater total number of violent acts perpetrated by the men during the incidents. In addition, physical differences in size and strength between men and women contribute to the gender differences in injury rates (Felson, 1996). Men are generally heavier and more muscular and likely to cause greater harm than women, even when men and women engage in similar acts of IPV. Men and women arrested for IPV inflict similar types of severe injuries by using different violent acts. Busch and Rosenberg (2004) found that women were more likely to inflict severe injuries on their partners when they used a weapon or object, but men were able to inflict similar types of severe injuries without using weapons or objects.

Both physical and psychological aggression toward a partner using construct scores incorporating self-report, partner report, and observational data were significantly associated with the IPV arrests groupings (any arrest versus no arrest) for both the men and the women, as was reported severe physical aggression. Because of their higher likelihood of arrest for the incident, higher levels of aggression toward a partner were expected for the men than for the women in the IPV arrest group. Contrary to prediction, there was no interaction with gender in these associations. Overall, women in both the arrest and no-arrest groups had higher levels of aggression (but not of severe physical aggression) toward a partner than the men. Thus, the men arrested for IPV were involved in relationships with high levels of physical and psychological aggression by both partners. The view that those involved in official IPV incidents would be couples predominantly characterized by one-sided male-to-female aggression, or “patriarchal” or “intimate terrorism” (Johnson 1995; Johnson & Leone 2005), was therefore not supported. These findings suggest that aggression toward a partner in many of these couples’ relationships can be described as dyadic behavior, with predominantly mutual involvement of men and women.

Whereas the couples involved in any IPV arrest showed an overall pattern of mutual aggression, the IPV that led to police calls and subsequent arrests involved more violent behavior by the men than by the women. This suggests the possibility that IPV incidents resulting in police calls often represent the worst incidents that occur among couples with a pattern of predominantly mutual aggression. It is also possible that the women are more likely to seek police assistance in situations that involve male initiated and unilateral violence than in situations with bidirectional aggression, although in only about one quarter of the incidents was the woman the caller. In the current study, the police were called to the situation by the women in 23% of the incidents, by another person in 32%, by the men in 2%, and by unknown sources for the rest of the cases. In other IPV studies, women also were more likely to call the police than were the men (e.g., Phelen et al. 2005). The police reports are somewhat limited by the information the men and women are willing to report, but the police records may also reflect some bias in recording information and making arrest decisions.

As the physical and psychological aggression of both partners was associated with risk for an IPV arrest for at least one member of the couple, interventions that focus on a couple’s behavior may have a greater effect on reducing IPV than intervention or treatment for men only. The equivalent rates of aggression toward a partner for the men and women also emphasize that aggression and violence in couples’ involve problems shared by both partners. Treating one partner without the other is not likely to reduce the mutual aggression or violence. Recent evidence further suggests that dyadic-based conjoint group treatment can be an alternative intervention model to reduce IPV in couples. Preliminary work evaluating couple-focused IPV programs indicate that this work can effectively reduce IPV, although not placing women at increased risk for injury (e.g., Stith, Rosen, & McColloum, 2003). This suggests that dyadic-based programs could be a viable option for many young couples.

Some limitations of the study should be noted. First, the number of men and women arrested for one or more IPV incidents was relatively low for hypothesis testing. Second, the purposes of the police report are generally to document that an offense occurred, rather than to collect data for research purposes. Thus, some pertinent information may not have been recorded and the reports may have underrepresented the violence that took place (cf., Harris, Dean, Holden, Carlson, 2001). Third, acknowledging that each method of assessment may have biases, combining data from multiple sources by using constructs defined by multiple indicators for most purposes can provide a comprehensive understanding of the behavior of interest and a more reliable strategy than single data sources (e.g., Cronbach & Meehl, 1955; Patterson & Bank, 1986). Like arrests, self-reports of violence and aggression also have limitations, such as difficulties with recall, and social desirability biases limiting reporting of behavior that is negative and illegal, such as underreporting violence toward a partner (e.g., Archer, 1999). The current study, however, involved partner reports and observational data in addition to arrest data.

Despite these limitations, the current study made a unique contribution to our understanding of the contexts and impacts of IPV incidents involving arrests, particularly through the juxtaposition of detailed police report information and data on the same couples from an ongoing longitudinal study of romantic relationship adjustment. The association between the types of engagement in violence -- namely the male-dominated violence in the IPV incidents and the mutual aggression toward a partner for these couples more generally -- raises the question of whether different groups of couples indeed engage in distinctive types of violence that have unique origins, developmental courses, impacts, influences, and implications for treatment (Johnson, 2006). The comprehensive consideration of violence and injuries during IPV arrest incidents, and the predominant patterns of mutual aggression for couples who experienced such incidents, produced insights that may be used to inform improvements in prevention and treatment efforts.

Article information

References