A Qualitative Analysis of Physically Aggressive Conflict Episodes among a Community Sample

Abstract

Intimate partner aggression among community couples has been conceptualized as representing an occasional, situationally-provoked response to a conflict. Yet, relatively few studies have considered the situational factors that contribute to the occurrence of an aggressive episode. The present study used thematic analysis to understand how episodes of physical aggression come about, from the participants' perspective. We examined narrative descriptions of relationship conflicts that included physical aggression to gain insight into the types of aggressive incidents experienced, the reasons and motives behind the aggression, and the meaning of these events. Married and cohabiting couples (ages 18 – 45 at baseline) were recruited from the community to participate in a longitudinal study of relationships. Heavy drinking couples were oversampled. At Wave 3, participants were asked to describe the most severe conflict they had experienced over the past year and to answer questions about severity and impact. Narratives that described use of partner physical aggression by one or both partners were subject to thematic analysis. Using narratives provided by 27 male and 29 female respondents (representing 51 different couples), we identified three primary motives or reasons for aggression: Expression, Instrumental, and Punishment. Narratives suggested as a primary theme that participants view partner violence as unusual, undesirable, and hence, meaningful. This was particularly true for male-to-female violence, which appeared objectively and subjectively different from female-to-male violence. Findings provide unique insight into the function and meaning of partner violence, including its gendered nature, within a community sample. Implications for measurement of partner aggression are also discussed.

Intimate partner aggression (IPA), defined as intentional physical aggression toward a relationship partner, is prevalent among the general population (Desmarais, Reeves, Nicholls, Telford, & Martin, 2012). IPA is associated with a range of negative psychological, health, and economic consequences (Foran & Phelps, 2013), particularly for women (Sillito, 2012). Individuals with certain characteristics, such as psychopathology, history of family violence, or substance use disorders (Birkley & Eckhardt, 2015; Foran & O'Leary, 2008; Schumacher, Feldbau-Kohn, Slep, & Heyman, 2001), or in relationships characterized by distress and low satisfaction (Stith, Smith, Penn, Ward, & Tritt, 2004) are more likely to perpetrate IPA. However, less is known about the situational factors that contribute to the occurrence of episodes of physical aggression (Wilkinson & Hamerschlag, 2005). Married and cohabiting couples experience many conflicts, but the majority of them do not involve physical aggression (Testa & Derrick, 2014). Even individuals prone to aggression do not aggress every time they have a conflict or feel provoked (Finkel et al., 2012). The present study involved descriptive exploration and thematic analysis of episodes of physical aggressive conflict as a way of understanding how and why IPA is used and understood by community men and women. Although previous research has documented the antecedents of aggression within samples of battered women, less is known about the situational contributors to physical aggression within community samples. We also considered the subjective meaning of IPA episodes for participants. A qualitative approach can improve understanding of the context and subjective meanings associated with aggressive events that cannot be fully captured or understood with quantitative ratings or acts-based measures such as the Conflict Tactics Scales (see Murphy & O'Leary, 1994; Testa, Livingston, & VanZile-Tamsen, 2011).

Understanding Situational Influences on IPV Episodes

IPA among community samples has been conceptualized as representing an occasional, situationally-provoked response to a conflict, with emotions and anger escalating to physical aggression (Johnson, 2008; 2011). Situational violence, by far the most common represented within general population samples, is about equally likely to involve males and females as perpetrators (Archer, 2000; Johnson, 2011). Finkel et al. (2012) describes these occasional episodes of partner aggression as reflecting a “perfect storm”, in which an individual with low self-control or a tendency toward aggression feels provoked by his or her partner and is unable to inhibit aggressive urges in the situation. Understanding of IPA can be enhanced by examining the motives or situational triggers for aggression within relationship conflicts. Several studies have attempted to do this by asking participants about their reasons for perpetrating physical aggression, using a variety of methods (e.g., open-ended versus choosing or rating a varying number of listed motivations, specific to each study, see Bair-Merritt et al., 2010; Flynn & Graham, 2010; Langhinrichsen-Rohling, McCullars, & Misra, 2012a for reviews). Frequently identified motives for physical aggression include anger, stress, jealousy, self-defense, getting a partner's attention, retaliation for emotional hurt, and expression/communication when verbal means fail (e.g., Leisring, 2013; Neal, Dixon, Edwards, & Gidycz, 2015). A limitation of this approach is that perpetrators may not always recognize their motivations and some motives may be more socially acceptable to report than others (Langhinrichsen-Rohling, et al., 2012a). As described below, qualitative methods may provide deeper insight into how and why physical aggression develops within conflict situations.

Qualitative Studies of Partner Aggression

Much of what we know about partner physical aggression is derived from survey studies which have used acts-based measures such as the CTS-2 (Straus, Hamby, Boney-McCoy, & Sugarman, 1996) to characterize an individual's use of violent acts over some period of time. Acts-based measures have been criticized for their failure to consider the context in which aggression occurs (e.g., Dobash, Dobash, Wilson, & Daly, 1992; Langhinrichsen-Rohling, 2010; Straus, 2012). For example, a shove may differ in severity and meaning depending on the gender of the perpetrator, the motive for the aggression, and the cultural context in which it occurs. Moreover, participants may view the aggressive act as relatively more memorable, distressing, or meaningful depending on contextual factors such as the content of the conflict in which it occurred (e.g., discussing divorce) or prior history of violence. Qualitative studies permit holistic examination of the situational factors and contexts as well as subjective meanings of violence that cannot be captured with quantitative methods (Murphy & O'Leary, 1994; Testa et al., 2011). For example, the participant's own words in describing a physically aggressive event can provide insight into how he or she views that experience.

Studies using this holistic, data-driven approach have provided important insights into the types and functions of IPA episodes. Foshee, Bauman, Linder, Rice, and Wilcher (2007) conducted follow-up interviews with 116 adolescents who had indicated on an acts-based questionnaire that they had perpetrated physical aggression in their dating relationships. Many of the events involved defensive aggression, in response to their partner's aggression. Girls' accounts also included as reasons for physical aggression: to express anger, in response to their partner's on-going abuse and controlling behavior, and to let the partner know that he had done something wrong. Studies of situational violence within adult community samples also point toward the importance of frustration, anger, and loss of control as participants' perceived motives for using physical aggression during a conflict with their partner (e.g., “exploding”, “losing it”, Cascardi & Vivian, 1995; Stanley, Bartholomew, Taylor, Oram, & Landolt, 2006; Stith et al., 2011). For example, male perpetrators described their physical aggression as a buildup of rage that finally exceeded a threshold while also blaming their own aggression on their partner (Whiting, Parker, & Houghtaling, 2014). Other motives that emerged included aggression as a means of getting the partner to do something (e.g., get away, stop yelling) and self-defense, particularly by women (Cascardi & Vivian, 1995; Orengo-Aguayo & Lawrence, 2014). Communication patterns also appear to play a role in aggressive incidents, for example, physical aggression was sometimes used to get a withdrawing partner's attention (Stanley et al., 2006). Not surprisingly, and consistent with conceptions of Situational Violence (Johnson, 2011), the vast majority of violent events in these studies emerged from escalation of an ongoing argument (Cascardi & Vivian, 1995), with physical aggression emanating from poor verbal or conflict resolution strategies (Stanley et al., 2006; Stith et al., 2011).

Beyond examining situational influences on use of physical aggression within intimate relationships, we were particularly interested in how participants described these experiences and the meaning or significance they found in them. Although many community couples engage in occasional physical aggression, partner violence is socially proscribed. This cultural inhibition against partner violence may influence the way participants understand and describe their experiences. For example, in a qualitative study of Situational Violence episodes, male and female participants minimized the physical aggression they experienced and failed to label these interactions as violent (Stith et al., 2011). When asked to describe violent acts reported on a questionnaire, 17% of adolescent participants “recanted” and claimed these were not actually acts of dating aggression (Foshee et al., 2007). Because male-to-female violence is viewed as less socially acceptable than the reverse (Felson & Feld, 2009; Hammock, Richardson, Lamm, Taylor, & Verlaque, 2016), we might expect gender differences in the way IPA is expressed, described, or understood.

The Present Study

The present study was designed to provide in-depth examination of physically aggressive relationship conflicts as experienced by adult men and women from the community. We were particularly interested in understanding how and why physical aggression is used within these conflict episodes, from the perspective of the participants. The focus of our analysis was not on what led to the conflict (e.g., Cascardi & Vivian, 1995; Stephenson, Martslof, & Draucker, 2011) but rather on what led to the use of physical aggression within the conflict and its function and consequences within the conflict event. We examined both male and female descriptions of IPA incidents, allowing us to examine whether there were gender-specific uses of aggression or perspectives on the use of aggression. We expected that nearly all aggression would reflect Situational Violence, that is, physical aggression arising within the context of a specific argument (Johnson, 2011). The sample included a large proportion of heavy episodic drinkers, and thus was uniquely well-suited to explicating the role of alcohol in partner aggression (e.g., Testa & Derrick, 2014).

Most prior studies have identified aggressive episodes by administering the CTS or other acts-based measures of IPA and then asking those with a positive response to describe a physically aggressive episode (e.g., Cascardi & Vivian, 1995; Foshee et al., 2007; Olson, 2002). However, this method appears to result in defensiveness, denial, or downplaying the aggression as “playful” (Foshee et al., 2007; Lehrner & Allen, 2014). Moreover, the most recent or physically severe episode may not be the most memorable or meaningful to the participant. For example, verbal aggression events frequently included more emotionally meaningful or distressing conflicts (e.g. discussion of divorce) and were better recalled and more vividly described than physical aggression events (Testa, Derrick, & Leonard, 2010). To avoid these concerns, we instead asked participants to describe the relationship conflict that they considered most severe or upsetting and then limited our analyses to the subset of conflicts that included some physical aggression.

We approached narratives holistically and without a priori theories or categories, seeking to identify patterns in the data using an inductive or data-driven approach (Hsieh & Shannon, 2005). We focus on the participants' words and perspectives to understand these incidents; however, as IPA researchers, our background and experiences shape our approach to the narratives and the meaning we derive from them. Although we were particularly focused on the situational or immediate influences on physical aggression, our perspective is that episodes of IPV reflect multiple levels of influence, from societal (e.g. cultural acceptance of violence), to individual and dyadic (e.g., personality characteristics, relationship dynamics), to situational (see Bell & Naugle, 2008; Finkel & Eckhardt, 2013). These broader contexts are also likely to shape how participants view their experiences.

Method

Participants

Participants consisted of 27 male and 29 females who indicated that their most severe relationship conflict episode in the past year – subjectively defined - had included some physical aggression. All had been recruited 2 years earlier to participate in a larger 3-wave prospective study designed to examine heavy episodic drinking (HED) and intimate partner aggression among married and cohabiting heterosexual couples. Couples (N = 280) in which both partners were between 21 – 45 years old were recruited from Erie County NY using address-based sampling. Because of the aims of the parent study, couples in which one or both partners was a heavy episodic drinker (HED, 4/5 drinks per occasion, at least weekly) were oversampled (see Testa et al., 2012 for details on sampling and recruitment). As a result, 200/280 (71.4%) couples in the original sample included at least one partner who engaged in weekly HED.

The 27 male and 29 female participants who provided a description of a physically aggressive conflict represented 51 different couples. There were only five couples in which both partners described an aggressive event and only one couple in which partners described the same conflict event. Thus, we considered the narratives of males and females to represent independent rather than dyadic data. Compared with women, men were significantly less likely to be White and married and more likely to report past year partner aggression on the CTS-2 and to represent a heavy drinking couple (see Table 1).

Table 1. Descriptive Information about Male and Female Participants and Narratives.

Participant Characteristics	Male (N = 27)	Female (N=29)	Z/t
Age	39.08 (5.83)	37.60 (6.06)	-.93
White	20 (74.0%)	28 (96.6%)	2.41*
Married and living with partner	14 (51.9%)	24 (82.7%)	2.47*
College Graduate	10 (37.0%)	17 (58.6%)	1.62
Heavy drinking couple	21 (77.8%)	16 (55.2%)	-1.79+
Past year CTS aggression, By either partner	21 (80.7%)1	16 (55.2%)	-2.01*
Narrative Characteristics
Low severity violence described1	12 (44.4%)	20 (69.0%)	1.85+
Female perpetrator2	14 (73.7%)	17 (68.0%)	-.41
Male drank at the time	14 (51.9%)	8 (26.6%)	-1.94+
Female drank at the time	14 (53.8%)3	3 (10.0%)	-3.52**
Violence most severe aspect	14 (51.9%)	15 (46.4%)	-.41
How upsetting at the time (1-7)	5.93 (1.24)	5.86 (1.43)	-.18
How upsetting now (1-7)	3.59 (1.85)	4.07 (2.30)	.85

¹Low severity events consist of throwing or hitting an object or pushing, by a single perpetrator (Category 1 in Table 2)

²Based on 19 male, 25 female events. Category 3 and 5 from Table 2 are omitted since perpetrator not easily determined.

³Missing data for one man; percent based on 26 cases

⁴Missing data for one woman; percent based on 28 cases

⁺p < .10,

^*p<.05,

^**p<.01

Procedure and Assessment

Narratives were derived from Wave 3 interview assessments that were conducted in private rooms by one of four White, female interviewers. Although recruited as couples at Wave 1, individuals were allowed to participate in subsequent assessments if their relationship ended or their partner declined to participate. Of the original 280 couples, 240 men and 243 women (representing 252 couples) participated. They completed several computerized questionnaires including the physical aggression subscale of the CTS-2 (Straus et al.,1996), referring to the past 12 months. After completing these measures, participants were asked to describe the incident of relationship conflict in the past year that they considered the most severe for whatever reason. Conflict interviews were completed by 173 women and 168 men. The primary reason for not completing an interview was inability to recall a specific conflict to discuss. In contrast to previous studies which have administered the CTS-2 to screen for violence and then asked those with positive responses to describe a physically aggressive event (e.g., Foshee et al., 2007; Lehrner & Allen, 2014), we conducted the conflict interview independent of CTS-2 responses. The interviewer did not have access to CTS-2 data. As a result, individuals who reported past year aggression on the CTS-2 but identified a non-physically aggressive conflict as the most severe (18 men, 16 women) are not included in the present analysis which includes only those conflicts that involved some physical aggression.

The Wave 3 interview began with the interviewer asking the participant to recall the worst or most severe argument they had had with their partner over the past year and to describe what happened in their own words: “Start with where you were, how and why the conflict got started, how it proceeded, how you were feeling and so on.” Interviewers were instructed to allow the respondent to describe the incident without interruption, asking for clarification only when necessary or to redirect back to the original event. After the participant had described the incident, the interviewer asked several standardized questions including whether each had used alcohol or drugs and what made the incident the most severe. Participants were also asked to rate, on a 7-point scale, how upsetting the event was at the time and now. If physical aggression was used, the interviewer asked why the person had used aggression. At the end of the interview, the interviewer asked whether each partner had engaged in yelling, insulting, throwing things, grabbing or restraining, pushing or shoving, or any other physical contact. Although the aggression was usually apparent in the narrative, in a few cases these questions permitted clarification of each partner's use of physical aggression. Throwing or aggression toward objects as well as grabbing, pushing, shoving and other physical acts constituted physical aggression. Verbal aggression (yelling, swearing) was common; however, the present analysis focused on use of physical aggression.

The average length of the entire interview, including questions not relevant to the current analysis, was 16.45 (SD=5.63) minutes (range 6.83 – 34.42 minutes). Digitally recorded interviews were transcribed verbatim by a research assistant and these transcripts were used in subsequent qualitative analysis. We used the entire transcript in deriving codes and themes, including the description of the conflict as well as responses to interviewer questions regarding why aggression was used, what made the event the most severe conflict, and whether and how alcohol influenced the event. Responses to these closed-ended questions were subject to statistical testing. There were 29 male and 31 female narratives that included a positive response to one of the physical aggression items; however, upon initial reading of the transcripts, we agreed that 4 incidents did not involve partner aggression and were not included, yielding a sample of 27 male and 29 female narratives for qualitative analyses.

Data Analysis

Our primary method of analysis was thematic analysis, a multistage, iterative process (see Braun & Clarke, 2000; Green et al., 2007), also described as conventional content analysis (Hsieh & Shannon, 2005). We chose thematic analysis for its flexibility and ability to generate unanticipated insights within this understudied area (Goldberg & Allen, 2015; Testa, Livingston, & VanZile-Tamsen, 2011). We used an inductive, data-driven approach (Braun & Clarke, 2006; Hsieh & Shannon, 2005), seeking not to confirm pre-existing hypotheses but instead to identify themes and patterns expressed by the participants, in their own words. We adopt a critical or subtle realist paradigm (Guba & Lincoln, 1994; Hammersley, 1992). That is, we assume that narratives describe an objective reality but that our ability to understand this reality is imperfect, influenced both by the subjective perspectives of the narrators as well as by our own experiences, values, and perspectives.

Thematic analysis allowed us to approach the narratives holistically, considering acts of aggression within the context in which they occurred and as they were understood by participants. In thematic analysis, themes need not be mutually exclusive and narratives may provide evidence of more than one theme. Although it was not possible to ignore the gender of the narrator, we read all female and male narratives together as we derived themes. We began by considering the reasons for and function of aggression within the conflicts, however, we also identified patterns and themes that cut across all data.

Thematic analysis began with the first and third authors reading all transcribed male and female narratives, approaching them without an a priori framework. Although we believed that males and females might have different perspectives on the emergence of aggression (Foshee et al, 2007; Johnson & Ferraro, 2000), our primary goal was to identify patterns across all narratives, considering gender differences secondarily. We followed the phases of thematic analysis described by Braun and Clarke (2006). First, we familiarized ourselves with the data, reading the transcripts multiple times. Coders independently recorded significant statements and then compared them. This led to a preliminary set of ideas and themes (Testa, Crane, & Leonard, 2014). The first and second authors then continued the iterative process of coding data by identifying, discussing, fitting, and refining themes (see Allison et al., 2008). Recognizing the potentially different perspectives of the coders, we used strategies such as analyst triangulation and negative or deviant case analysis (Patton, 2001) to incorporate these perspectives and produce more credible conclusions. For example, in a few cases in which there were disagreements, we recognized that there was both a primary and a secondary function of physical aggression in the scenario. The patterns that we identified initially and presented in 2014 were maintained throughout analysis, although some of the initial themes were combined into larger and broader themes. Finally, all transcripts were read independently, multiple times, by the first two authors, to verify and refine themes into the final coding scheme described below. The small number of disagreements between coders were resolved with discussion.

Results

Overview of Results

Before commencing thematic analysis, we first characterized the objective characteristics of the physical aggression described. Next, we sought to identify themes and patterns in the data that revealed the reasons for the use of physical aggression. These revealed three primary motives or functions of physical aggression: Expressive, Instrumental, and Punishment. Our multiple reading of the narratives also suggested a primary theme that cut across the narratives: that partner physical aggression is Meaningful and Unacceptable.

Characteristics of Aggressive Events

Although events were identified independent of the CTS, we used the minor and severe subscales as a guide to characterizing their objective severity. We also sought to identify the gender of the aggressor(s) and to characterize the aggression as involving a single perpetrator versus bidirectional aggression. Descriptive findings are summarized in Table 2.

Table 2. Characteristics of Aggressive Events.

	Narratives
	Female(n = 29)	Male (n = 27)
1. Single perpetrator, low severity aggression
1a. Hitting or throwing object, not at partner 4 female, 10 male perpetrators	9	5
1b. Throw object at partner 4 female, 2 male perpetrators	4	2
1c. Single push/grab/shove 12 female, 0 male perpetrators	7	5
2. Mutual low severity aggression (pushing, fighting over object) difficult to identify a primary perpetrator	3	7
3. Single perpetrator, moderate to severe aggression, not reciprocated (slap, poke, punch) 5 female, 2 male perpetrators	4	3
4. Single primary perpetrator, moderate to severe aggression, with restraint or defense 5 female aggression / male restraint	1	4
5. Prevent or stop partner aggression toward someone else 3 female aggression / male restraint of female	1	2

We identified 32 incidents that included a single act of unreciprocated, minor physical aggression and agreed that these represented low severity physical aggression (Category 1). This category included incidents in which one person threw an object (e.g., phone, TV remote, suitcase) or hit an object (e.g., punch wall, slam door) without contact with the partner. Somewhat more severe were incidents in which it appeared that the object was thrown at the partner, the most serious being a man throwing a rock at a woman's car windshield. None resulted in injury. We also included in this category incidents that involved a single act of mild aggression involving physical contact with the partner but no reciprocation (e.g., a single push, attempt to grab an object from the partner). There were gender differences in the perpetrators of the low severity events: throwing or hitting objects (category 1a) involved mostly male perpetrators whereas unreciprocated single acts (category 1c) all involved female perpetrators.

We considered the remaining aggressive events to be more severe because they involved mutual violence and/or more severe acts that could cause pain or injury. Eight events involved mutual minor aggression (e.g., push, throw), with acts that were equal or equivalent. It was difficult in these cases to identify the initial perpetrator, as they described fighting over an object or pushing each other. Seven events involved more severe (e.g., slapping, punching), unreciprocated acts and five events involved severe aggression that was met with restraint or self-defense by the partner. Finally, there were three events in which a male partner physically restrained his partner when she was aggressive toward someone else. Technically, the man was the perpetrator of IPA; however, it was the woman's aggression toward someone else that was the genesis of the event. Nearly all of the severe violence events involved a female as the primary aggressor with the male either disengaging, restraining the woman, or defending himself. The two most severe male-perpetrated events involved poking the woman in the forehead and putting a hand on her throat (but not choking); female-perpetrated events tended to involve more severe acts such as punching.

To examine whether there were systematic differences in male versus female narratives, we compared them on characteristics derived from the descriptions (severity, gender of perpetrator) and on closed-ended interview items. As shown in Table 1, male narratives tended to describe more severe events involving alcohol. Male and female narratives were equally likely to indicate that the female was the perpetrator. Although men's events tended to be more severe than women's, ratings of how upset they were by the conflict did not differ for men versus women nor did the percentage who indicated that the violence was the most severe aspect of the event.

Although the characteristics and severity of physically aggressive events displayed in Tables 1 and 2 provide a helpful overview, objective characteristics cannot capture the context or significance of the event for the narrator. For example, one event was categorized as low severity (category 1a) because it involved throwing an object and no other violence. However, this act occurred within an event which suggested Intimate Terrorism rather than Situational Violence: the man threw his wife's suitcase because he was enraged by her plans to leave. He then disabled her car to prevent her from leaving their rural home and she called the police for assistance. She stated that even though he was not physically preventing her from leaving, she felt intimidated and not free to leave. This vivid example illustrates the potential limitations of documenting acts apart from the context in which they occur.

Reasons for Aggression

Thematic analysis was used to identify the most commonly expressed reasons for or functions of physical aggression within conflicts. The first and second authors were able to classify 55/56 narratives as reflecting one of three primary functions of aggression: Expressive, Instrumental, or Punishment. Initial inter-rater agreement was excellent (Kappa = .85) with the few discrepancies resolved through discussion¹.

The most common function of physical aggression was Expressive (n = 37), which we observed in the majority of male and female narratives. Participants described physical aggression as allowing them to express their feelings to the other person when verbal expression was inadequate. Examples, both from females, include: “I wanted him to see how angry I was and I threw the phone at the wall” (108192-F) and “It was probably just to get each other's attention, because we weren't listening to each other” (108421-F). In some cases, the display of aggression is used to make a strong and final statement to break the deadlock of an ongoing argument. For example, “I just threw all her presents at her and I was like, “Here, here, here.” I was like, “We're done.” (300098-M).

Expressive narratives seemed to best reflect Situational Violence as commonly understood, that is, inability to handle conflict verbally or control aggressive impulses when provoked. Expressive narratives frequently included words like “boiling” and “blew up” to describe the building and then explosion of anger into physical aggression as they “couldn't take it anymore”. Examples include: “It escalated to the point where he got so incredibly angry with me he threw a beer can” (112506-F) and “I couldn't take this yelling and swearing at me anymore” (100396-M). These examples imply physical aggression as a loss of self-control but also as a rather inevitable consequence of the extreme anger and frustration (“it's like shaking a pop bottle,” 101024-M) or, in some cases of the other person's actions. Within the Expressive motive were examples of the partner as the “cause” of the aggressive response, such as: “He can push my buttons and get me to that point where I lose control and slap him” (117905-F) and “I egg him on. I infuriated him and he doesn't know how to control his anger at all” (107822-F).

In 11 narratives (7 by men, 4 by women), we identified the primary motive or function as Instrumental in that a physical act was used to accomplish a specific, practical purpose. For example, aggression was used to push the partner away: “He'll be standing in the way trying to stop me so we can talk about this or just argue about it and I just push him out of my way” (110339-F). Alternatively, a physical act was used to prevent the partner from walking away from an argument or leaving, for example refusing to relinquish the car keys or throwing a suitcase as the partner tried to pack. Thus, aggression was used both to push a partner away, thereby ending the conflict, or alternatively, to get the partner to stay and continue the conflict. Although some Instrumental narratives included elements of Expressive, these narratives lacked the emotional volatility and lack of control as well as the escalation or explosion of conflict into a physical expression of anger that was prominent in the Expressive incidents.

In some cases, physical aggression was used instrumentally to stop or prevent one partner from aggressing toward another person (see Category 5 in Table 2). In these events, the aggression did not represent an escalation of a conflict between partners (as in the Expressive scenarios). Rather, as described in the following excerpt, the use of aggression triggered or exacerbated conflict, as the recipient of the aggression reacted angrily. For example:

When you're drunk you don't want to listen… She didn't like me telling her what to do and I felt like I had to…I pulled a drink out of her hand and threw it. She reacted angrily.get up and push by me and go get another one and I would knock it out of her hand again. My sister and brother got involved and it escalated. I had to restrain her to the car so we could get out of there. (115932-M)

We identified the motive of Punishment in 7 narratives (6 from females) that were strikingly similar to each other. In fact, this was the first pattern that we identified in the data because they stood out so clearly from the rest. In these narratives, upon learning of her partner's serious transgression (e.g., infidelity, drug use), the perpetrator - a woman in every case - immediately uses physical aggression to hurt and punish her partner. Although these narratives contain some elements of expression, they do not arise from an escalating conflict as the Expressive scenarios did. Moreover, although they could be viewed as a subset of Instrumental, in that the aggression was used for the purpose of punishment, they stand apart qualitatively from the Instrumental narratives. For example, an important distinguishing feature of these scenarios is that the male partner is described as feeling guilty and deserving of punishment. He does not retaliate or respond to the aggression but rather apologizes, cries, or does nothing, as described in this example:

I slapped him when I first found out. I found out from a text on his cell phone … and then when I asked him he lied about it and then I proved it to him and he lied again and then I think I slapped him. (What did he do?) He just took it. He knew he deserved it (110457-F).

The only event in which male and female partner described the same incident was a Punishment incident. The husband's account, although less emotional, corroborated the wife's account of her discovery of her husband's infidelity.

Although the sample consisted primarily of heavy drinking couples, it is noteworthy that participants did not spontaneously invoke alcohol as a reason for the physical aggression. In a few cases, drunkenness was mentioned or it appeared that it contributed to the conflict (e.g., wife was angry about husband's drunken behavior the night before, leading to an argument). When explicitly asked about the role of alcohol at the time of the event, only a few participants responded positively with a comment such as, “Your inhibitions go down when you drink… I think it was the catalyst” (116371-M). However, it was somewhat surprising given the heavier than average drinking levels of the sample that alcohol was rarely viewed as a significant reason for the aggression and did not emerge as a theme or even sub-theme.

Primary Theme: Violence is Unacceptable and Significant

Thematic analysis permits identification of meaning beyond the surface or semantic level. Although we began by identifying motives and reasons for using physical aggression, through our immersion in the narratives and the patterns within them, we identified a robust latent theme. That is, physical aggression was viewed by participants as unusual, socially undesirable, and unacceptable, a response of last resort. As a result, when partner aggression occurs it stands out as significant and meaningful, both at the time of the event and in the recall and interpretation of the incident.

Recall that our desire to minimize social desirability bias in discussing aggression led us to avoid the common method of reminding participants of their reports of physical aggression on the CTS and then asking them to describe one of those violent events. Instead, we asked participants to identify conflict events that they considered severe for whatever reason. To avoid arousing defensiveness, interviewers did not ask who the aggressor was or label the event as violent. Nonetheless, we identified a clear social desirability theme throughout the narratives, suggesting to us that participants internalized the belief that partner violence is not socially acceptable, making it significant and meaningful when it does occur.

At the simplest level, several respondents noted that at the time of the event he or she was concerned with others overhearing, disapproving, or being upset by the aggression. Examples include: “He said ‘this is bad', we shouldn't do this in front of (son)” (118902-F) and “It's a good thing we live way out in the country or the neighbors would think we were ghetto” (101850-F). These examples suggest that concerns about social desirability were encoded as part of the incident, and expressed when it was recalled. This woman's verbal stops and starts suggest her efforts in describing and making sense of the incident.

He went to grab me away from (child) and then my immediate reaction was just,I turned, and I just pushed him away which then led him to - he wasn't - this is so out of character for us - which made him kind of just secure me against the wall, which all that did is just infuriate me even more ‘cause now he's thinking I'm gonna hurt my kids and I would never, ever do that and it escalated just because of that, and it was lack of sleep… This is the most awful thing because this isn't us (101921-F)

Although interviewers strove to be neutral and non-judgmental as they elicited descriptions of conflict events, some participants seemed concerned with how their stories and actions might be interpreted. This resulted in downplaying of the seriousness of aggression, either to convince the interviewer or to justify to oneself that the aggression was not that serious. For example, “You know about the physical fights that you see in the movies, that's not how it is with us” (101730-F) and “It's just like a jokingly thing, kind of not serious” (101024-M). The tendency to blame the partner for causing the aggression, for example “Some of the things she was saying initiated it” (104300-M), is also consistent with this theme, as participants deflected responsibility for their own socially unacceptable behavior. Participant narratives included effortful attempts to downplay the aggression and convince themselves, or the interviewer, that their aggressive behaviors were not really violence, as shown in these two examples, both by men:

If I come towards her she'll put her arm up or something but it's like, we don't really, like, there's no punches, like, no-one's punching anybody or choking anybody like some of the questions (laughs). Like, it don't get to that point. It gets to, like, there's pushing and shoving but it's not like (trails off, 100396-M).

I think there is a strong difference between playfully smacking someone in the butt and physically assaulting. I don't think it ever got to that point. I think that she just reacted to hit back. I don't think it was, there's never situations of violence in our home at all, it was just, you know, a playfulness that turned into something beyond that (107082-M).

The occurrence of physical aggression made the incident stand out in recollection of the event and its significance at the time. Even in cases in which the aggression was quite minor, the occurrence of physical aggression marked the crossing of a boundary or an indication that “this is getting out of control.” For example:

I got really frustrated and I grabbed a pillow or something and I threw it across the room. And my wife had told me before that that really upsets her when I throw things and I haven't, I don't do it a lot, but when I do it, you know, a couple of times a year, it's really upsetting to her. .she was really quiet and just sort of went away and that was almost worse because there was no reaction. .That was a very clear line for her, throwing things or being violent in any way that to her represented violence (108142-M).

Some participants expressed dismay at their involvement in physical aggression: “I remember for the past 2 years saying oh no, we don't fight like that, and now it's like, yeah, I guess we kind of did” (113125-M). When asked what the most severe aspect of the conflict was, this man responded: “Probably because I touched her. I'm not a physical guy like that, that I got so outraged which, that's probably what scared me the most” (105322-M).

Gendered Violence

Several aspects of the data indicate that there are differences not only in men's and women's aggressive behaviors, but in the meaning of male and female aggression. Thus, our conclusion that IPA is viewed as unusual, unacceptable, and meaningful is qualified by gender: male-to-female violence is more unusual, less acceptable, and more serious than female-to- male violence. We base this conclusion on the patterns and types of violence perpetrated by men and women as well as on the way men's and women's aggressive acts are discussed. For example, the Social Desirability expressions presented in the above section came disproportionately from male narratives, suggesting that men felt more defensive or concerned about their aggression and how it would be viewed than did women. By objective standards, women in this sample were more physically aggressive than men. Most perpetrators were female, particularly for acts of unreciprocated aggression and severely violent acts such as punching (see Table 2). There were no corresponding events involving severe male-to-female aggression. None of the women's narratives described using violence in self-defense. All of the Punishment narratives were strikingly gender-specific: women described using physical aggression to punish a man for a serious transgression. Despite the one-sided nature of these events, women did not go through obvious efforts to downplay their aggression, suggesting that they believe that physical punishment of a male partner may be justified when his transgression is serious (e.g., infidelity). This apparent inconsistency of these Punishment narratives with the Social Desirability theme led us, through Deviant Case analysis, to recognize the importance of gender in the meaning of partner aggression. Men seemed to accept the physical punishment as well, expressing that they deserved it both in these Punishment narratives as well as in others: “She didn't mean it, she just dug in (with her nails). I deserved it for sure, you know” (111626-M). We did not identify comparable statements by female victims.

In contrast to women's initiation of severe aggression, men did not reciprocate but described restraining women when they were behaving aggressively.

I was more or less grabbing her, holding her arms, her hands, trying to keep her from picking up stuff and throwing it. She was trying to push me off of her so she could grab something to throw or hit me (300193-M).

Men's ability to restrain a woman to prevent her from inflicting injury presumably reflects their typically larger size and strength. There were no comparable events involving a woman restraining a man. Men also described consciously not retaliating when experiencing female aggression, which we believe reflects internalization of the norm that aggressing against women is socially unacceptable, as evidenced in the quote below. We did not find evidence of this selfcontrol in female narratives.

It makes me feel aggressive. It makes me feel like I want to retaliate back. I never do, I never have but that's what upset me the most that when that happens it makes me feel like I have to retaliate (107082-M).

The fact that men more often were the perpetrators of aggression toward objects (e.g., punching a hole in a door, smashing remote against the wall) may also be interpreted within the framework that men displace their aggression as an alternative to aggressing against women.

Yet, despite the fact that women were more frequent aggressors, and perpetrated more severe violence than did men, we found evidence in both male and female narratives that women were afraid and intimidated by the potential for male violence, even when little or none was actually displayed. For example, a man was shocked that his wife punched him twice in the face yet he walked away and did not retaliate although his wife was afraid that he would:

I was smiling like I am now and extremely - she was - she got scared after that. We actually talked about that after the fight. I said “why did you back down like that after I stood up? Did you think I was going to hit you?” And she goes, “Yes, I did. You should have seen the look in your face”. “Have I ever hit you before?” and she was like “No”. I was angry but I don't think - honestly, I don't think I would ever do that (109766-M).

In contrast, we found no evidence of a comparable male fear of female aggression.

Discussion

Partner physical aggression is meaningful and consequential to community men and women, both at the time of the event, and in their recollection and telling of the episode. The significance of IPA reflects the belief that partner violence is unacceptable, a theme expressed across the narratives in different ways including downplaying of aggression, embarrassment over others observing the aggression, and distress that one has perpetrated or been involved in a physically aggressive conflict. This theme was qualified by the conclusion that male-to-female IPA is viewed as less acceptable than the reverse, consistent with previous quantitative research (e.g., Archer, Fernandez-Fuertes, & Thanzami, 2010; Felson & Feld, 2009).

Findings also provide insight into some of the critiques and controversies surrounding the CTS and other acts-based measures of IPA: that these measures cannot capture the context or significance of these acts (see Langhinrichsen-Rohling, 2010; Straus, 2012). Narratives describing subjectively severe conflict events included more examples of female- to- male violence than the reverse, including more acts of severe violence and more female-initiated aggression. Women perpetrate more IPA than men (Archer, 2000), are more likely to initiate violence (Capaldi & Crosby, 1997), and are more likely to be the only perpetrator within the couple (Langhinrichsen-Rohling, Misra, Selwyn, & Rohling, 2012b). Yet, thematic analysis revealed that both women and men viewed men's physical aggression as more serious than women's. Men, but not women, described containing and controlling their aggressive urges, perhaps recognizing the greater potential for male aggression to result in injury (Archer, 2000). Quantitative studies reveal that women who experience IPA victimization express more fear than do victimized men (Houry et al., 2008; Kar & O'Leary, 2010; Sillito, 2012). Our results extend these findings by suggesting that women fear the potential for male violence. Thus, even implicit threats of male violence (e.g., “that look on your face”) gave men some control over women's behavior. Although women may perpetrate more IPA than their partners by objective standards, this may not be the best gauge of the severity and consequences of these acts.

Qualitative examination of physically aggressive events provided insight into the dynamics, contexts, and functions of aggression. Expression was the most common function of violence within these incidents, as physical aggression served as an outlet for emotions that could not be expressed verbally and made a statement. The expressive function of partner violence has been identified in previous quantitative (Langhinrichsen-Rohling et al., 2012a) and qualitative studies (Cascardi & Vivian, 1995; Foshee et al., 2007; Orengo-Aguayo & Lawrence, 2014), and is consistent with the conception of Situational Violence as aggression that erupts due to poor conflict resolution and self-control (Johnson, 1995). The prominence of this theme also suggests that deficits in communication and self-control may contribute to partner aggression, offering a potential target for prevention intervention efforts within community samples (see Babcock, Graham, Canady, & Ross, 2011; Denson, DeWall, & Finkel, 2012).

The punishment function, though less common, was also identified in Foshee et al.'s (2007) study of adolescents (labeled “Ethic reinforcement”), and as here, was specific to female-to- male aggression. We believe that this dynamic fits within the overarching theme that violence is unacceptable, but that female-to-male violence is not quite so unacceptable and may even be justified under extreme circumstances (e.g., partner's infidelity). We also found evidence that partner aggression was used instrumentally: to obtain an object, to make the partner stay away or prevent him or her from leaving, or to stop or prevent the partner from aggressing toward someone else. To our knowledge, this function of aggression has not previously been identified and may have emerged in part because our events were not limited to those identified by the CTS. Its prevalence may also reflect the fact that our sample contained a relatively high proportion of heavy drinkers, resulting in some incidents in which physical means were used to prevent or mitigate the partner's drinking or drunken aggression toward another person. Many of these events did not readily fit the schema of a marital conflict escalating to verbal and then to physical aggression. Rather, consistent with the theme that IPA is unusual and unacceptable, in some cases physical acts (e.g., restraining) precipitated conflict between partners.

Implications for Measuring and Understanding IPA

Although not intended as a way of assessing the validity of the CTS, study findings have implications for measurement of partner violence, long an area of concern (see Follingstad & Rogers, 2013). We identified incidents of aggression by asking participants about their most serious episode of partner conflict, recognizing that an incident may be recalled as severe due to its emotional content, regardless of the presence or amount of physical aggression. A substantial number of respondents who described an aggressive event, including nearly half of the women in our sample, did not report past year aggression on the CTS-2 that preceded the interview. In part, the mismatch reflects our broader definition of aggression; we included events in which objects were thrown or hit, whereas the CTS-2 specifies throwing something at a partner that could hurt. However, participants described throwing a pillow or punching a door as violence that “crossed the line,” suggesting that including these less severe acts are considered aggression and their assessment is important to capturing subjective experiences of IPA. Previous research has in fact found that the CTS does a poorer job capturing less severe violence (Jose, Olino, & O'Leary, 2012). The poor match between CTS responses and interview data in general has been noted in previous research (e.g., Lehrner & Allen, 2014) and may reflect the very different cognitive tasks involved in estimating frequency of past behaviors over a year or more, without context, versus recalling a subjectively meaningful and upsetting event. Recent studies suggest that it is the way that IPA experiences are recalled that is particularly important in terms of their meaning for the individual and impact on relationship functioning, even though partners do not recall the same events (Burrus & Cobb, 2011; Derrick, Testa & Leonard, 2014). The fact that there was almost no overlap in the events identified by male and female partners suggests that there may be gender differences in how men and women recall or interpret conflict incidents, a potentially fruitful direction for future research.

In addition to suggesting that even minor acts of aggression are viewed as unacceptable and meaningful, our findings also suggest that perpetrating aggression can be as distressing as being victimized. Consistent with the theme that IPA is viewed as unacceptable, participants expressed distress that they were aggressive, that they wished to aggress, or that there was aggression in their relationships. These findings are consistent with a recent daily diary study showing that self-reported episodes of partner physical aggression, whether involving victimization or perpetration, had negative effects on subsequent mood and relationship functioning (Derrick et al., 2014).

Limitations and Directions for Future Research

Our findings are based on a fairly small number of incidents, derived from a community sample of heterosexual and primarily White adult couples in which heavy drinking was deliberately oversampled. Although generalization beyond the present investigation is not typically an intent of qualitative analysis (Creswell, 2009), different themes and patterns may have emerged in a more diverse or less heavy drinking sample. Our narrow focus on understanding the development and function of physically aggressive episodes, rather than on the broader development of conflict, verbal, or psychological aggression, is also a potential limitation. Conclusions are potentially limited by the characteristics and skill of the interviewers and the quality of the interviews; different questions may have elicited different descriptions. Nonetheless, we believe that the themes we identified are true reflections of participants' understanding of situational violence and not simply a function of the researcher's questions.

Qualitative analysis of IPA events has considerable potential for understanding how and why physical aggression is used within intimate relationships, including the degree to which these recalled events correspond with quantitative measures of IPA. However, we believe the true potential for this approach - and its implications for future research - lies in the potential for understanding the subjective meaning that aggressive events have for individuals and for relationships, apart from - or as a supplement to - quantitative assessment of acts perpetrated or received. Researcher- or clinician-focused assessment of IPA cannot capture these subjective perspectives, yet they may be critically important for understanding its consequences as well as for developing effective means of preventing IPA.