Abstract
The purpose of this study was to understand sexual assault perpetrators’ emotional responses to perpetration to facilitate treatment development and to better understand processes that may give rise to repeat perpetration. Sixty-one firsthand narratives of sexual assault perpetration, posted on Reddit.com, were analyzed using qualitative text analysis. The analysis revealed four primary emotional responses to perpetrating sexual assault: shame, guilt, depression, and anger. Each emotional response was associated with different contextual features that appeared in the narratives. Shame co-occurred with perpetrator alcohol use and consent confusion, guilt co-occurred with perpetrators’ stated self-growth, anger co-occurred with denial of responsibility and hostility toward women, and depressed affect co-occurred with social isolation following perpetration. The findings indicate certain emotional responses may be more adaptive than others for protecting against repeat perpetration. This research has important implications for the treatment of perpetrators and supports the idea that self-image and perceived social context may be important treatment targets.
Women experience sexual assault, defined here as any non-consensual sexual activity, at alarming rates. A nationally representative survey found that 43.9% of women in the United States have experienced some form of sexual assault in their lifetimes (Breiding et al., 2014). Moreover, researchers estimate that between one quarter (Swartout et al., 2015) to three quarters (Lisak & Miller, 2002) of perpetrators commit multiple sexual assaults. Thus, intervening among men who have perpetrated sexual assault against women in the past could significantly reduce rates of sexual assault. Yet, to intervene with this group, we must first understand the emotional responses to perpetration, as this knowledge may help inform the motivation to perpetrate again. The purpose of the current study is to examine perpetrators’ responses to committing sexual assault against women to aid in the development and refinement of treatment for perpetrators.
Although research has explored the outcomes of sexual assault for the victim, little is known about the psychological outcomes for perpetrators. Studies that have examined perpetrators’ affective or interpersonal functioning have been limited to risk factors such as antisociality (Abbey & Jacques-Tiura, 2011) or hostile attitudes toward women (Malamuth, Sockloskie, Koss, & Tanaka, 1991) that presumably existed prior to offending. Certain theoretical models of repeat sexual offending, primarily focusing on incarcerated offenders and child sex offenders, suggest that shame following perpetration may contribute to future sexual assault perpetration (Marshall, Marshall, Serran, & O’Brien, 2009; Wolf, 1989). However, perpetrators’ emotional reactions have not been studied empirically, and theoretical accounts rarely make reference to unidentified perpetrators of sexual assault against women. The current study addresses this gap with a qualitative analysis that examines emotional responses (e.g., shame, guilt, or sadness) in firsthand accounts of sexual assault perpetration. We specifically analyze men’s narratives of sexual perpetration against women, as the vast majority of sexual assaults are cases of male-to-female violence (Breiding et al., 2014).
Individual-Level Models of Sexual Assault Perpetration
Despite the dearth of empirical evidence on emotional responses to sexual assault perpetration, some theoretical models of sexual violence make predictions regarding men’s likely responses to perpetration (Marshall et al., 2009; Ward, Hudson, & Keenan, 1998; Wolf, 1989). Although we focus solely on sexual assault against women and define sexual assault broadly, ranging from unwanted touching to rape, the models that we review here focus on convicted offenders. Therefore, in these models, “sexual assault” typically refers to severe behaviors, namely, rape and child sexual abuse. Furthermore, child sex offenders differ from other convicted sex offenders in several ways, including less reliance on physical force (L. J. Cohen, Frenda, Mojtabai, Katsavdakis, & Galynker, 2007) and greater social skills deficits (Geer, Estupinan, & Manguno-Mire, 2000).
In Wolf’s (1989) model of sexual offense, which was based on clinical work with convicted sex offenders, cognitive and affective reactions to an assault can motivate a perpetrator to re-offend. According to this model, the initial offense is motivated by the perpetrator’s negative self-image, which involves an undue focus on his own needs and goals. He compensates for his negative self-image with fantasies of control. Guilt surrounding these fantasies and the offense is reduced with cognitive distortions, such as justification or minimization of harm inflicted on the victim. For example, research has shown that perpetrators may justify their behavior by attributing sexual assault to uncontrollable biological processes or by blaming the victim for not being more forceful in her resistance (Hipp et al., 2015). In Wolf’s model, any remaining guilt following the assault may further lower the perpetrator’s self-esteem and perpetuate the cycle of sexual offending. In addition, the cognitive distortions intended to protect the perpetrator’s self-image may also contribute to re-perpetration, as denial of responsibility prevents him from changing his behaviors. Wolf defined guilt in a way that coincides with our definition of shame—a negative evaluation of one’s own character (Tangney, Wagner, Fletcher, & Gramzow, 1992).
Other theories have proposed that cognitive and affective reactions to perpetrating sexual assault can include shame, cognitive distortions, and lack of empathy toward the victim, which may all be functionally related. Marshall et al. (2009) argued that convicted sex offenders’ experience of shame, which is characterized in part by excessive focus on the self, can result in a lack of empathy for the victim. Because empathy has been found to protect against sexual assault (Abbey, Parkhill, Clinton-Sherrod, & Zawacki, 2007; Wheeler, George, & Dahl, 2002), reduced empathy may increase the likelihood of re-offending. In addition, victim blame and other cognitive distortions, intended to protect against negative self-image, may reinforce attitudes such as hostility toward women that predict sexual assault (Malamuth et al., 1991). Furthermore, past research has found that shame indirectly predicts aggression through externalizing blame (Stuewig, Tangney, Heigel, Harty, & McCloskey, 2010).
The model proposed by Marshall et al. (2009) was informed by general research on guilt, shame, and empathy, not specific to incarcerated sex offenders. Nevertheless, the authors postulated the process of conviction and incarceration may increase shame, and, therefore, their model may be most relevant for convicted offenders. Likewise, Wolf’s (1989) model was intended to apply to convicted sex offenders. Due to the lack of research on emotional responses in undetected perpetrators, it is unclear whether these affective reactions specified in the models of re-offending reviewed above could be generalized to apply to sexual assault perpetrators who may not have been reported or convicted.
Interpersonal Models of Sexual Assault Perpetration
In addition to individual level risk factors, sexually coercive behaviors are influenced by associating with sexually aggressive peers (Jacques-Tiura et al., 2015; Swartout, 2013), and culturally bound gender role norms (Malamuth et al., 1991). Men who perceive pressure from peers to engage in sexual activity “by any means,” and whose peers talk about women in objectifying terms, are more likely to perpetrate sexual assault (Jacques-Tiura et al., 2015), as are men who have a fewer distinct social domains or communities in which they interact (Kaczkowski, Brennan, & Swartout, 2015). Furthermore, aggressive male peers provide informational support in the aftermath of an act of violence against a partner (DeKeseredy, 1990). Informational support provides a new interpretive frame for the assault, such as placing the blame on the victim, to protect the aggressor from stress.
The interpersonal models underscore the importance of peer attitudes and peer network structure in predicting individual attitudes that give rise to sexual assault. Taken one step further, one can argue that peer perceptions also influence how the perpetrator makes meaning out of his actions. The current study may help inform whether the social context, such as the perpetrator’s relationship to the victim, or peer approval or disapproval of sexual assault, plays a role in the self-reported emotional responses to perpetration.
The current study is a qualitative analysis examining the self-reported emotional reactions of men who perpetrate sexual assault against women in firsthand narratives obtained from an anonymous online community (Reddit.com). This analysis addresses whether the perpetrators’ narratives include reactions to the assault that align with the models of sexual re-perpetration outlined above, or whether a new or refined model could better explain emotional responses to perpetration. Although qualitative analyses cannot determine which theory most accurately predicts psychological responses to sexual assault perpetration, data from firsthand accounts could lay the groundwork for developing hypotheses to test different models. Moreover, just as many have argued for the existence of multiple etiological pathways toward sexual offending (e.g., Malamuth et al., 1991; Ward et al., 1998), one might expect multiple types of responses to sexual assault.
Because the current study takes an inductive approach to analyzing the data (Creswell, 2007), specific hypotheses are not outlined at this time. However, the following two, general research questions were developed a priori:
Based on theoretical models of repeat sexual assault, we expect that several narratives will include reports of negative emotional responses to perpetrating sexual assault. Furthermore, the evidence that perceived peer attitudes are important in predicting individual rape-supportive attitudes suggests, in the aftermath of a sexual assault, the perpetrator’s perception of his peers’ reactions will influence his own reaction.
Method
Data Collection
Reddit.com is an anonymous online news networking forum. A nationally representative survey found that 6% of adult Internet users in the United States reported visiting Reddit, and men between the ages of 18 and 29 constitute the largest demographic of Reddit users (Duggan & Smith, 2013). Users can post videos and articles and start discussions with other users—all anonymously. The content seen by users is influenced by the user voting system, whereby more popular posts that receive more user “up-votes” will appear before less popular content. On July 26, 2012, an anonymous Reddit user posted the following question: “Reddit’s had a few threads about sexual assault victims, but are there any redditors from the other side of the story? What were your motivations? Do you regret it?” The user who posted the question was in no way connected to the researchers involved in this study. Upon learning of this thread, the researchers involved in this study obtained approval from the local Institutional Review Board to retrieve posts made in response to this question. The data collected were publicly available and posts were not linked with any identifiable information.
Data reduction
For feasibility, the data gathered by the research team were first limited to the 1,128 posts that appeared on the first page of the Reddit thread and that were written in response to the original question, rather than a comment responding to another writer’s response. Next, the data were further limited to the 779 comments posted in the first 2 days. Narratives that did not describe perpetration were then eliminated, leaving 113 posts. The responses that were eliminated at this stage were stories of sexual assault victimization, secondhand accounts of sexual assault perpetration, or miscellaneous comments that did not include information on a sexual assault. The responses were further narrowed down to 75 narratives of male-to-female sexual assault, excluding instances of child sexual abuse and reports of false accusation without information on emotional responses. Narratives in which respondents claimed to have been falsely accused by the victim were included in cases where the respondents also wrote about their emotional reaction to the accusation, as we believed these narratives could contribute to our understanding of the emotional outcomes to perpetration. Furthermore, it was impossible to know whether these respondents were actually falsely accused or were resistant to accepting responsibility. Finally, as we were interested specifically in emotional responses to perpetration, we eliminated 14 of the 75 narratives that did not make reference to an emotional reaction following perpetration. Sixty-one narratives remained in the final dataset.
Respondent Characteristics
Due to the nature of Reddit, which is an anonymous online community, we were not able to assess the demographics of the respondents. However, Reddit users are primarily men between the ages of 18 and 39 (Duggan & Smith, 2013).
As with all self-report data, there is a risk of inaccurate reporting. However, the anonymity of an online forum may have positively influenced reporting accuracy (Joinson, 1999), compared with a setting where respondents would have some level of contact with the researcher and may be motivated to appear in a favorable light.
Qualitative Methodology
The current study is a qualitative text analysis (Kuckartz, 2014) with elements of phenomenological design (Creswell, 2007; Willig, 2012) and discursive analysis (Willig, 1999, 2012). Qualitative text analysis does not assume an interaction between the researchers and the respondents. This is an appropriate approach for the current study, given the data were gathered from an anonymous online public forum, and respondents were not influenced by the researchers.
Phenomenological analysis is a method of qualitative inquiry used to understand a particular experience. The guiding research question of the current study, examining the experiences of perpetrators in the aftermath of the assault, fits within the framework of a phenomenological design. Discursive analysis rests on the premise that language is performative (Emerson & Frosh, 2004; Willig, 1999). Whereas a phenomenological design focuses on people’s experiences, discursive analysis focuses on how these experiences are communicated (Emerson & Frosh, 2004; Willig, 2012). This entails a critical examination of the choice of language to convey an idea that could be described in a different way, or the decision to include certain details peripheral to the event. The data analyzed for the current study came from people who freely elected to post intimate, stigmatized, and incriminating details of their personal lives in the public domain. One must ask what the respondents were trying to accomplish through communicating their stories, and therefore the construction of the narratives warrants attention.
Data Analysis
Kuckartz (2014) outlined several stages for qualitative text analysis that guided our analysis. The qualitative analytic stages include the (a) development phase, (b) test phase, and (c) coding phase. Analyses were conducted using ATLAS.ti Version 7 qualitative data analysis software. Due to the brevity of each response, we analyzed each post as a single text segment and applied the inductively developed codes to the entire text segment for each post. The stages of analysis for the current study are outlined below.
Development phase
The development phase is the initial identification of themes and development of codes. Although the research questions were developed prior to analysis, the analysis itself was content-driven (Creswell, 2007); themes and codes were not determined a priori, but rather were derived from the text itself. The analytic lens chosen for the current study, a combination of phenomenological and discursive analysis, focused the attention to specific elements of the narratives. To reduce the potential subjective bias in the identification of salient themes, a team of two researchers independently read through a portion of the narratives (25%) and developed a list of themes to be used as codes. Potential codes and definitions of the codes were determined once consensus had been reached between the researchers (Miles, Huberman, & Saldaña, 2014).
The data were first read through for general impressions, followed by a closer reading of the meanings in each statement—the descriptive stage (Willig, 2012). At each stage of analysis, codes became successively more abstract to eventually identify commonalities across narratives (Creswell, 2007). Salient ideas that emerged in a single narrative were then phrased in more general terms to accommodate similar ideas that occurred in a separate narrative, thereby identifying common patterns across narratives. For instance, if one respondent wrote about binge drinking and another wrote about consuming illegal drugs, these ideas were abstracted and combined under the code of “substance abuse.” This process is also referred to as “factoring” in the guidelines for analyzing qualitative data laid out by Miles et al. (2014).
In line with a phenomenological research design, the focus in coding the data was to provide a textural and structural description of psychological responses to perpetrating sexual assault (Creswell, 2007). The textural description paints a picture of what the respondents experienced. To this end, we categorized data specifically related to the emotional experience of a self-identified sexual assault perpetrator following the assault. The structural description provides information on contextual details that are relevant to the phenomenon of interest. In the current study, these were factors described in the narratives that may have influenced the respondents’ experiences in the aftermath of sexual assault perpetration, with particular attention to the social context. The contextual themes coded in the current analysis, including the social context, were not intended to relay information on some objective reality, but rather indicated the respondents’ subjective experiences and the way in which their narratives were constructed. With discursive analysis, the elusive task of obtaining a “true” analysis of the respondents’ inner experiences does not invalidate the interpretation.
Testing and coding phase
The validity of qualitative research lies within the iterative nature of the analytic process (Morse, Barrett, Mayan, Olson, & Spiers, 2002). The researchers repeatedly checked to ensure the codes fit the data and were applied predictably throughout the process of coding the narratives. To ensure consensus and reliability, inter-rater reliability was assessed early in the analytic process, once 25% of the data had been coded by both researchers using the agreed upon set of codes. This process helped determine any areas of confusion in the codes and definitions. Below satisfactory agreement (κ < .60) was resolved through adapting the code conceptualization to ensure a better fit with what was explicit in the data. Once consensus had been reached with satisfactory inter-rater reliability, the primary researcher coded the remaining 75% of the data.
Results
The purpose of this study was to examine the primary emotional responses to perpetrating sexual assault and the contextual and psychological features that co-occurred with these primary reactions. The analysis revealed four themes relating to the emotional outcomes of sexual assault perpetration: shame, guilt, anger, and depression. These four outcomes are negative emotional reactions directed at the perpetrator himself (shame), his behavior (guilt), the victim or women in general (anger), or generalized (depression). The themes were largely independent of one another in the narratives, and all themes were recurrent in the data. The four themes also co-occurred with different contextual themes. The classification of primary versus contextual themes was based on the research question: Themes that described emotional outcomes were primary themes, while other salient themes that emerged were considered contextual or secondary themes. The perpetrator narratives have been quoted exactly without changing language or spelling, although some narratives have been shortened in the interest of space. Additional narratives that represent each of the four emotional themes can be found in the appendix.
Shame
Shame, characterized by a negative self-concept following sexual assault perpetration, was the most prominent of the four primary emotional themes. Although shame and guilt share overlapping conceptual features, shame is self-directed. The event is seen as evidence of a flawed self. For example, the respondent quoted below attributed his behavior to a fundamental flaw in his character: I carried that mark on my conscience for years . . . I don’t condone rape or sexual assault. It’s a terrible thing to impose your will upon another person. But I think to commit these sort of things shows a significant and somewhat demented flaw in your character. I always (and still) have had a lot of trouble connecting with other people. (Respondent 126)
Several other respondents demonstrated shame by speaking of themselves in disparaging terms, for example, “That moment in time made me feel like scum” (Respondent 341). Another respondent wrote about his struggles to accept himself following multiple acts of sexual assault perpetration, which resulted in his expulsion from college. His narrative begins as follows: Five years ago or so I was in the middle of a really bad breakup with someone who was manipulating me. Granted I could simply have cut her off but I was more naive at the time and didn’t know how to say no to all the things she asked me for. In order to cope with my negative emotions I would pop Adderall during the day, smoke weed all the time, and drink until I passed out every night. Around this time I hooked up with a girl who lived down the hall from me. We were both really drunk, but she was near blackout and I was not nearly as gone. She invited me into her room, puked in the trash can, took her pants off and pulled me into bed. So I took this as her wanting to hook up and so we did . . . After one particularly bad blackout I drag myself out of bed and go to class. Upon my return I find that I have an email telling me to go straight to Student Affairs. I do so and am promptly arrested and taken to the local municipal jail. Apparently during my blackout I went into a girl’s room who I had a class with and volunteered with, and I woke her up by rubbing her thigh . . . She told her roommates who told her to go to the campus security . . . I kept going to class, until a week later I was called in and found out that three girls had come forward. One was the girl I drunkenly hooked up with (6 months ago at this point) who accused me of sexual assault . . . Now it is five years later, and though I understand I crossed some lines and was a creep, I refuse to believe I was a rapist. I am still dealing with some self-shame issues that keep me from fully accepting myself but I hope with work I can become the most complete human I can be. (Respondent 311)
This respondent expressed shame when he indicated that he has had difficulty accepting himself. However, although this respondent related a story that meets the definition for rape, he rejected the label “rapist.” He further denied his role as the aggressor by introducing his narrative in a way that positioned him as a victim of manipulation: “five years ago or so I was in the middle of a really bad breakup with someone who was manipulating me.”
Shame, perpetrator alcohol use, and consent confusion
Respondents who expressed shame also appeared to minimize their responsibility through highlighting consent confusion and perpetrator alcohol use. For example, Respondent 311 quoted above reported being heavily intoxicated at the time of the assaults. Although there is no reason to doubt the veracity of this claim, the choice to include this detail in the narrative may represent an effort to show that this behavior was atypical for the respondent, and the respondent is externalizing blame onto situational factors (i.e., alcohol). Likewise, reports of consent confusion, which was coded when the perpetrator stated that he believed that the sexual assault was consensual when it happened, might indicate that the perpetrator was trying to place himself in a more favorable light.
Guilt
For this analysis, guilt was defined as remorse or regret directed at the specific sexual assault behavior without necessarily an indication that the respondent viewed himself negatively. Research supports differentiating between shame and guilt (Cohen, Wolf, Panter, & Insko, 2011; Tangney et al., 1992). Similarly, we found guilt and shame appeared independently and were easily distinguished from one another, as shown by the strong inter-rater reliability. The following quotation is from a respondent who expressed guilt about sexually assaulting an acquaintance.
However badly I feel about it, and I do feel fucking revolted about it, I can only imagine she feels worse. I would literally sell my soul to be able to take that night back. If there’s any silver lining, and I know that any silver lining to such a thing is FUCKING TINY by comparison to how much I hurt her, it’s that what happened made me a better man and drove me to be more active in feminism. I now would never ever EVER do something like that again, and in all my relationships since have always been respectful of my girlfriends when they dont want to. But that’s it. It’s an important lesson to learn but I really fucking wish I could have learned it without hurting someone like I did. (Respondent 41)
Respondent 41 described immense regret (“I would literally sell my soul to take that night back”), yet his negative appraisal is directed at a specific behavior occurring within a discrete time frame.
Guilt and self-growth
The theme of self-growth co-occurred more strongly with guilt than with the other three primary emotional themes. Self-growth was coded when the respondent indicated that following the perpetration, he was actively working toward being a better person, or actively working to prevent another act of perpetration. For example, Respondent 41 expressed motivation toward self-improvement because of the assault. Likewise, Respondent 143’s narrative demonstrates a similar case of guilt and commitment to self-improvement: MUCH later I realized that I had basically assaulted her . . . I still think about it sometimes, and I feel terrible . . . I’ve never done anything like that ever again, and never will. I just wish I hadn’t learned that lesson in that way, and that she had to suffer for it.
Anger Toward the Victim
Some respondents expressed negative reactions directed toward the victim—or the alleged victim. We present this theme with a caveat: Several of these respondents denied perpetrating sexual assault. The theme of denial of responsibility strongly co-occurred with anger toward the victim. As demonstrated in the following example, these respondents indicated that the sexual interaction was consensual and expressed anger at being falsely accused.
Am I a rapist? Well fuck no. Did I feel like one anyway? Oddly, yes. Some of the stories in this thread don’t involve actual rape, just situations where someone with a highly biased opinion would try to make the guy in the situation feel like he had done something wrong. This happens way. too. fucking. much. My kneejerk reaction to a girl saying she’s been raped has become “uh huh yeah whatever bitch” because I’ve heard women cry false rape way too many times on my buddies, as a way to get out of things or as a way to get sympathy . . . (Respondent 557)
Although this respondent adamantly denied committing rape, he admitted to feeling like a rapist. This suggests either that he doubts his innocence, or that being seen as a rapist by others is enough to make one feel like a rapist.
In the following example, Respondent 529 explained that whenever he thought about a past sexual assault that he perpetrated, his first reaction was to blame the victim: I know making out doesn’t mean yes to sex, but I really thought that she wanted it. I don’t remember her saying no, but I believe her that she said it. I feel godawful and think about it every day. My thoughts go from “she didn’t say no she just thinks I raped her because she’s an attention grabber” to “holy shit how can I think like that. That’s just me trying to feel less guilty about the situation, you ARE a RAPIST, YOU RAPED HER” to “ok it’s in the past, just don’t do it again, you are a good person.” She texted me a few months ago too. She has a few good friends that she’s told and she’s in therapy so hopefully she can live with it, even if it’s going to be hard.
This narrative exemplifies the relationship between anger and shame. Although the respondent at times blamed the victim, he then chastised himself: “you ARE a RAPIST.” This narrative depicts a pattern whereby negative self-concept that follows perpetration gives way to cognitive distortion: accusing the victim of lying to seek attention.
Anger and hostility toward women
Narratives that expressed anger frequently included language that was hostile toward women or described common rape myths, most notably, that women generally lie about rape, as in the case of Respondent 557’s narrative quoted above. This pattern could be seen even in cases where there was no denial of responsibility. In the following example, Respondent 579, clearly admitted to rape, and he still expressed hostility toward the victim through denigrating language: I fucked her there and walked away, disgust in my eyes as she told me how pathetic I was. She railed something about her innocence and how I was a monster. I didn’t care. I had been pushed over an edge and my primal instinct was there to defend itself . . . I called her a whore as I left.
Although this respondent admitted to rape, he also blamed the victim for pushing him “over an edge.” He further evaded responsibility by attributing his behavior to a “primal instinct.” Given the prevalence of hostility toward women in narratives that expressed anger, the perpetrator’s emotional response and interpretation of events may have been influenced by pre-existing negative attitudes about women and sexual assault.
Depression
Depression following sexual assault perpetration emerged as a salient theme in several narratives. In this analysis, depression was coded when respondents explicitly wrote about depressed affect or suicidality following perpetration. In some cases, they admitted wrongdoing, while in others, they claimed to be wrongfully accused. The following example demonstrates the emergence of suicidality after the respondent was accused of perpetration and subsequently rejected by his peers: My friends from college, the few who found out, immediately treated me like I was Ted Bundy or some other piece of human trash and cut ties . . . There isn’t a day that passes, there has not been since this happened more than a decade ago, that the thought of suicide has not crossed my mind. That word “Rape” will follow me to my grave. I could be the best person on Earth and it will still follow me. (Respondent 200)
Depression and social isolation
Nearly all instances of suicidality among perpetrators were accompanied by feelings of social ostracism following the assault. Respondent 200, quoted above, conveyed daily suicidal thoughts tied specifically to being socially outcast and bearing the “rapist” label. The following example conveys a similar pattern of suicidality and social isolation following an act of sexual assault perpetration: I have never had many friends. She told at least some of my friends and my roommate. He kicked me out, and I lost all my friends. I still do not know how I live after this . . . I wish I knew how to apologize for this. I wish my friends could have the courage to help me. Sometimes I wish I had never promised myself that I would never commit suicide. (Respondent 217)
The respondent in the next example related a period of depression after finding out that a sexual encounter he had thought was consensual was perceived by the victim as rape. In this case, there was no explicit indication that he was socially ostracized as in the previous examples.
I found out she felt like I had raped her, I was so blown away by this I had no idea what to do. I was confused and hurt and just felt like shit . . . I have never in my life felt as shitty and depressed as when she told me that she felt what happened was rape. The depression made me have to drop out of school and go live back home . . . I’m over my depression now but I never, and will never, feel as low as I did because of that night. (Respondent 149)
Although this respondent was not rejected by either the victim or his peer group, his depression began when the label “rape” was applied to his experience. As such, this narrative resembles Respondent 200’s narrative when he explained, “That word ‘Rape’ will follow me to my grave.” Rape is socially reprehensible; therefore, the label “rape” or “rapist” may represent actual or feared ostracism.
Integration of Themes
Figure 1 depicts the relationship between primary themes and contextual themes as they relate to theories on repeat perpetration and the empirical literature on aggression. The arrows represent only hypothesized causality based on the extant literature, as our findings cannot indicate casual relationships. First, we found a strong co-occurrence between shame and reports of consent confusion and perpetrator externalization of blame, which aligns with theories that suggested a relationship between negative self-concept and cognitive distortions intended to reduce the perpetrator’s culpability (Marshall et al., 2009; Wolf, 1989). Although the themes of shame and anger did not consistently overlap in the current study, they shared important similarities: externalization of blame and rejection of the “rape” label. Anger and shame may contribute to repeat perpetration, as blame externalization has been found to mediate the relationship between shame and aggression (Stuewig et al., 2010), and because the perpetrator who denies responsibility may not be motivated to correct his behavior (Wolf, 1989). In contrast, respondents quoted in the section on guilt acknowledged wrongdoing (e.g., “I realized that I had basically assaulted her”), which may be an important step in preventing a future sexual assault (Marshall et al., 2009; Pollard, 1994).
Model of emotional responses and repeat offending.
Discussion
Qualitative analyses of sexual assault perpetrators’ responses to their behavior are rare. We were able to gather naturally occurring data posted on Reddit to investigate whether emotional responses to perpetration aligned with theories of offending and repeat perpetration. The purpose of this study was to examine the emotional responses to perpetration, an area of research that has previously been neglected. The analysis revealed four salient emotional outcomes to perpetrating sexual assault: shame, guilt, anger, and depression. Each of these responses co-occurred with different contextual or psychological themes (e.g., consent confusion, social isolation, and self-growth). The co-occurrence of the primary emotional themes with second-level themes sheds light on the factors that contribute to these emotional reactions and the way the perpetrator interprets his behavior. First, the results are in many ways in line with the theories of repeat perpetration outlined in the introduction. In addition, perpetrators highlighted the influence of social factors on their emotional responses to perpetration.
The model in Figure 1 indicates the potential relationship between negative self-concept and cognitive distortions (e.g., victim blame or minimization of the harm inflicted) among perpetrators (Marshall et al., 2009; Wolf, 1989). In the present study, this pattern appeared in the strong co-occurrence between shame and reports of consent confusion and alcohol use, which may have served to externalize blame. Likewise, respondents who expressed anger denied responsibility and blamed external factors, which could be evidence of an attempt to reduce shame. Cognitive distortions, such as victim blame, may prevent the perpetrator from taking steps to correct his behavior, or may reinforce attitudes that are hostile toward women, thereby contributing to repeat perpetration. In contrast, we saw that guilt was related to reports of self-growth and may therefore be a more adaptive response to perpetration.
In addition, several respondents suggested that their emotional reaction to perpetrating sexual assault was connected to the social context. In several narratives, peer rejection following sexual assault perpetration preceded depression and suicidal ideation. This finding indicates that peer attitudes not only contribute to sexual assault behavior (Franklin, Bouffard, & Pratt, 2012; Jacques-Tiura et al., 2015; Swartout, 2013) but are also important in the aftermath of sexual assault and can shape how the perpetrator reacts to the event. DeKeseredy (1990) argued that a perpetrator’s peers provide a favorable interpretive lens in the aftermath of sexual assault to protect him from stress. The results from the current study indirectly support DeKeseredy’s theory by demonstrating that the absence of peer support is linked to a negative emotional response to perpetrating sexual assault.
The significance of social contexts also appeared in the construction of these narratives. The use of a discursive analytic lens allowed for a critical examination of the choices made in framing narratives. Several respondents expressed frustration that after being accused of perpetrating sexual assault, they did not get to share their side of the story. For example, Respondent 557 proclaimed that men do not have a voice to defend themselves against false accusations: “our society prevents men from retaliating or defending themselves.” As such, social dynamics do not only play out within the narratives, but also when considering the purpose of these narratives. The fact that many narratives appear to be framed to reduce culpability (e.g., by blaming alcohol or placing the perpetrators in the role of the victim) and publicly shared tells us that the perpetrators may seek some type of social acceptance.
Limitations
Because the respondents answered an open-ended question and could include any details they thought relevant, each narrative included some themes but not others, making it difficult to assess the existence of, and relationships between, themes across all narratives. For example, it is not clear whether respondents who expressed guilt were also concerned with the label “rape,” simply because the label “rape” was infrequently mentioned in narratives containing guilt. In addition, it is impossible to know the experience of the perpetrators beyond what they presented in their narratives. Although several narratives contained elements that were in line with theoretical models of repeat perpetration, we do not have evidence on whether these respondents go on to perpetrate again in the future.
By gathering data from Reddit, we were able to gain in-depth knowledge of the experiences of self-identified perpetrators. These respondents may have been more reluctant to share these experiences in the presence of a researcher or without the total anonymity allowed by Reddit. However, there were limitations to using this source of data. First, the respondents were not selected at random by researchers, but rather freely chose to share their stories online. Therefore, we cannot say with any certainty that their experiences are representative of sexual assault perpetrators. In addition, due to the anonymity of Reddit, we were not able to report ages or other demographic information. Finally, as with all self-report data, there is the possibility of bias in the data due to inaccurate responding.
Research Implications
Findings from qualitative studies are not intended to be generalizable, but can be used to inform future quantitative research. Future research could investigate whether certain psychological responses to perpetrating sexual assault predict recidivism with a longitudinal study. Anonymous online recruitment of perpetrators could be used to assess emotional responses and the frequency of sexual assault over time. By using validated measures, this line of research could also quantify the relationship between constructs such as shame, remorse, and externalization of blame, and could assess the presence of these reactions in different groups of offenders (e.g., undetected perpetrators, convicted offenders, and child sex offenders).
Clinical and Policy Implications
Although qualitative results cannot be generalized, the richness of this data provides important insight into potential intervention targets for sexual assault perpetrators. Existing programs to prevent sexual violence, primarily on college campuses, involve an educational component on sexual assault and consent (DeGue et al., 2014). The prevalence of reports of consent confusion and victim blame among respondents in the current study highlights the need for continued efforts to increase education about sexual assault using empirically-supported prevention programs.
Moreover, the results from the current study indicate the presence of emotional reactions that could give rise to self-harm or future sexual assault perpetration. Some respondents reported suicidal ideation, while other narratives included elements such as externalization of blame that are predictive of aggression (Stuewig et al., 2010). Future quantitative research should continue to investigate emotional consequences for perpetrators, as this line of research can be used to improve therapeutic interventions for perpetrators. Prior research has indicated elevated risk for self-harm among adjudicated offenders and child sex offenders (Jeglic, Spada, & Mercado, 2013; Stinson & Gonsalves, 2014). The present study indicates that suicide may also be a risk for non-adjudicated perpetrators, and future research should evaluate the extent of self-harm risk in this population. Furthermore, our findings indicate that suicidality may be tied to perceived or actual social ostracism. This finding, if supported by further research, could indicate that treatment for some perpetrators may benefit from addressing the perpetrator’s perception of social isolation.
In addition, as displayed in Figure 1, our findings, in conjunction with prior theoretical models, suggest that treating shame and anger may be important for preventing repeat perpetration. However, the causality laid out in this model cannot be determined from the data analyzed in the current study, and further research is needed to test this model and examine the relationship between shame, anger, and repeat perpetration. Dynamic assessments used to determine recidivism risk for sex offenders, including the Stable-2007 and the Acute-2007 (Hanson, Harris, Scott, & Helmus, 2007), assess for negative emotionality and loss of social supports as potential risk factors. Our findings indicate that these factors may also be important for undetected offenders.
Finally, treatment for these unidentified perpetrators of sexual assault against women may benefit from a similar approach used for unidentified pedophiles and hebephiles in the Dunkelfeld project (e.g., Beier et al., 2015). In this program, self-identified pedophiles who did not go through the criminal justice system were able to receive therapy anonymously, which led to reductions in offense-supportive cognitions and increased victim empathy (Beier et al., 2015). Although there are important differences between child sex offenders and perpetrators who offend against adults (L. J. Cohen et al., 2007; Geer et al., 2000), further research could evaluate whether a similar program could be used for undetected perpetrators who offend against adults.
Conclusion
The purpose of this study was to understand the emotional responses to perpetrating sexual assault to better understand processes that give rise to repeat perpetration. We analyzed 61 firsthand narratives of sexual assault perpetration posted on Reddit.com, using qualitative text analysis (Kuckartz, 2014). The results revealed four primary emotional responses to perpetrating sexual assault: shame, guilt, anger, and depression. Each emotional response was associated with different contextual features that appeared in the narratives. The findings of this study are in several ways in line with theories of repeat offending and suggest that guilt may be a more adaptive response than shame. In addition, the results indicate the importance of the social context when considering the four primary emotional outcomes identified in this study. This research has important implications for the treatment of perpetrators and supports the idea that self-image and perceived social context may be important treatment targets.
Footnotes
Appendix
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.