Online Abuse of Feminists as An Emerging form of Violence Against Women and Girls

Abstract

Abuse directed at visible and audible women demonstrates that cyberspace, once heralded as a new, democratic, public sphere, suffers similar gender inequalities as the offline world. This paper reports findings from a national UK study about experiences of online abuse among women who debate feminist politics. It argues that online abuse is most usefully conceived as a form of abuse or violence against women and girls, rather than as a form of communication. It examines the experiences of those receiving online abuse, thereby making a valuable contribution to existing research which tends to focus on analysis of the communications themselves.

Introduction

Online abuse aimed at women ‘celebrities’ is well covered in mainstream media (Bracchi 2013; McNally 2015) and that directed at women engaged in feminist debate has also received coverage. Early examples include the campaigner Caroline Criado-Perez being subjected to extensive, extreme abuse after calling for women’s representation on banknotes. Professor Mary Beard (2013) blogged about her experiences of online abuse after a TV appearance and reflected on the classical and contemporary ‘cultural awkwardness’ about women’s public voice. Anita Sarkeesian’s experiences of abuse after she criticized the representation of women in video games, and Jessica Ennis-Hill’s after commenting on the case of footballer and (at that time) convicted rapist Ched Evans, are further examples of abuse towards high-profile women. Growing public discussion of the impact of abuse on women’s participation in civic life was exemplified in the Parliamentary launch in 2016 of a cross-party campaign—‘Reclaim the Internet’ (www.reclaimtheinternet.com)—which recalls earlier responses—‘Reclaim the Night’ in the United Kingdom; ‘Reclaim the Streets’ in the United States—to street sexual harassment and violence. Indeed, various organizations (e.g. End Violence Against Women Coalition 2013) have framed concern about online abuse of women in terms of violence against women. Similarly, a working group on Broadband and Gender, chaired by UNDP and UN Women, states that cyber violence against women and girls (VAWG)¹ can have ‘adverse impact on the exercise of and advocacy for free speech and other human rights’ (2015: 2).

However, much scholarship about online abuse has assumed it is unique to the cyber environment and early scholarship focused on defining and categorizing it as a form of communication, as we discuss below. By contrast, our approach is to explore online abuse of this kind as an extension of offline gender relations which are marked by abuse and VAWG. In this paper, we first explore the recent development of perspectives on online abuse. Three gaps in the extant literature are identified: first, a failure to develop a robust gendered analysis; second, a lack of comparative analysis of online and offline VAWG; and third, a lack of victimological examination of online abuse experienced by women and girls. Following that, we outline our methodology and then present findings relating to the nature and impact of online VAWG and responses to it.

Perspectives on Online Abuse

Since the recognition of online hate and abuse, scholarship has sought to define, explain and understand this growing phenomenon. Problematically, extant research has tended to treat online abuse as separate from ‘real-world’ experiences. Jane (2015) defines the ‘three paradigmatic waves of flaming-related research’ comprised of, firstly, technological determinism (the idea that ‘flaming is the result of the medium’; Jane 2015: 67), secondly, attempts to define the variously named phenomenon of ‘aggressive or hostile communication occurring via computer-mediated channels’ (O’Sullivan and Flanagin 2003) and, thirdly, a tendency to minimize or overlook the phenomenon. Each of these waves has generated debate, as would be expected in an emerging body of scholarship. They also reflect a broader distinction in the cybercrime literature that separates out offending that could only be committed in an online environment (e.g. many banking frauds or identity theft) from that which might be exacerbated by technology but could be committed offline (Wall 2008). However, reviewing this work to illuminate online abuse directed at women and, particularly, those engaged in feminist debate online reveals three important and related gaps in the research. First, there is a glaring lack of gendered analysis of a phenomenon that is frequently gendered. Second, the focus on online abuse as a form of communication overlooks commonalities with other forms of VAWG. Third, accounts from recipients of online abuse which would reveal the experience and impacts of it are absent. We discuss each of these shortcomings below.

Firstly, research has rarely foregrounded a gendered analysis of these aggressive communications. While some refer to the sexualized or sexist nature of the content, many scholars have not acknowledged—let alone prioritized—this aspect. In mitigation, the sexualized and sexist nature of abusive communications need not always be the focus. Indeed, given the diversity of terms for the phenomenon being investigated—‘flaming’ (Lea et al. 1992), ‘trolling’ (Donath 1999), ‘provocation’ (McCosker 2013), ‘invective’ (Vrooman 2002), ‘cyber hate’ and ‘hate speech’ (Citron and Norton 2011), ‘ebile’ and ‘gendered vitriol’ (Jane 2014b)—there is a need to set parameters around the phenomenon. However, for an investigation of gendered online abuse against women and girls, it is important to move beyond a focus on definitions and a tendency to group together diverse forms of hostile communication, without acknowledging specific features of misogynistic communication that require distinctive classification and explanation. O’Sullivan and Flanagin (2003: 80) begin to move away from this monolithic approach. They offer the ‘interactional norm cube’ to identify ‘the possible combinations of message interpretations, as a function of various message sender/receiver and third-party viewpoints, and as informed by normative cues’. This highlights that flaming is varied, contextual and relational, distinguishes it from other forms of harassment and hate speech, and provides a platform for understanding sexualized and misogynistic abuse.

Definitional work is not alone in failing to incorporate gendered analysis; Lee’s (2005) analysis of responses to flaming neglects a gender awareness, even when presenting data which demonstrate the gendered nature of the flaming. He presents extracts from a discussion forum which include the terms ‘bitch’, ‘dumb broad’, ‘ignorant little slut’ but defines these simply as ‘name-calling’ without recognizing inherently gendered and sexist features. Claims that ‘provocation… extends the public that forms’ around news events (McCosker 2013: 215) and that flaming can be seen as productive form of conflict which ‘sometimes plays an important role in demarcating group boundaries’ (Lee 2005: 392) seems of limited value when analysing explicitly sexist, sexualized, threatening and violent behaviour directed at women online which may in fact restrict women’s civic engagement (see, e.g., Citron and Norton 2011). An exception to this lacuna is William’s (2006) analysis of hate speech and sexual abuse within the online community Second Life. While not directly comparable to the online abuse analysed in this paper (because, e.g., the cyberworld he analyses is an environment in which users adopt holistic virtual identities), Williams usefully critiques the Internet as a site of social and cultural reproduction that reflects real-world patterns. Equally valuable, in relation to the discussion further below, is Williams’ analysis of how such environments provide routes of resistance to abuse as well as a conduit for offensive behaviour.

Some scholarship pays attention to gendered dimensions of online abuse. Herring was perhaps the first to recognize the gendered nature of much online abuse and examines gendered differences in styles of communication (2004), as well as feminist responses to ‘trolling’ (2002). Jane (2014a) focuses explicitly on ‘ebile’ directed at women online, arguing that this receives considerable media coverage but little scholarly attention. Citron and Norton (2011) argue that the gendered nature of online abuse compromises women’s ‘digital citizenship’ and is a civil rights violation (Citron 2009). This echoes work noting that racist hate crime attempts territorial exclusivity and the delegitimization of minority communities in some geographical areas (Bowling 1998). Halder and Jaishankar (2009: 2011) argue that part of the logic of victimization of women on social networking sites is to exclude certain voices from cyberspace. Megarry (2014) argues that online abuse polices women’s voices, thereby limiting their use of online fora for feminist activism. Focusing on the ‘performance’ of online abuse, Vrooman (2002: 64) examines communications on alt.flame as ‘resolutely masculinist displays of the prowess and skill of a chosen identity, an aspect of masculine display’. As we argue below in relation to our data, recipients of such communication often identify it as an exclusionary attempt to delegitimize their online presence although it is frequently counterproductive since their resolve to political engagement can be strengthened as a result. This indicates how our survey of ‘victim’ impact and responses reshapes understanding of general online abuse.

The second problem we observe in much scholarship is its conceptualization of online abuse as distinct from real-world contexts. The technologically deterministic approach exemplifies this, seeing aggressive communications as resulting from the technology used, allowing as it does anonymity and unaccountability in a disinhibiting environment. Such work has been the subject of thorough empirical and theoretical critique, including its failure to recognize the wider social context of gendered norms and sexist behaviours (Vrooman 2002). Others have noted this failure to contextualize online abuse as an extension of offline behaviour. For example, with reference to the #mencallmethings hashtag, Megarry (2014: 47) argues online harassment ‘should be conceptualised as online sexual harassment and a form of excluding women’s voices from the digital public sphere’. However, if online abuse is to be seen as an extension of real-world behaviour, then it might follow that those who are not misogynistic offline are unlikely to become so online simply because they are in an uninhibited environment, as a technologically determinist position might imply. Or ‘[i]n other words, people are jerks not only when they are in anonymous Internet spaces, but also when they are in spaces where they can get away with being jerks’ (Shaw 2014: 274). Claims that the Internet is inherently criminogenic are critiqued by Wall (2008: 49) who rejects assumptions that cyberspace ‘corrupts normally law-abiding individuals who go on a moral holiday when on the Internet’. Moreover, the focus on online abuse as a form of communication minimizes its significance as ‘just’ words. The ‘real’ and the virtual are not separate experiential realms; activities that take place in the virtual world are still experienced as reality, with material consequences. Proper understanding of these experiences requires that we move beyond analysis of texts to engagement with those who receive them, an approach we adopt in the analysis below.

Although recent studies usefully examine the text used in online abuse, the third shortcoming we note is the failure to consider experiences of receiving online abuse. Jane (2014a) insists that we do not censor the ‘unspeakable’ ebile to protect the sensibilities of readers; excising harsh or offensive language, she argues, contributes to a ‘tyranny of silence’ which benefits perpetrators (2014b: 533). We support Jane’s insistence (2014a: 81) that ebile ‘must be spoken in its unexpurgated entirety’ to document and understand the phenomenon. However, analysis needs to go further and consider the experience of receiving such online abuse. Without analysing recipients’ experiences, claims that, e.g., online abuse ‘silences women’ remain unsupported by empirical evidence. Textual analysis is valuable but limited since without examination of recipients’ experiences we lose their interpretations, relying instead on researchers’ own analysis. Thus, we cannot explore how the experience of abuse intersects with other aspects of life and identity, and we do not learn how such experiences are incorporated—or not—into daily activities, and political engagement, online and offline. Considering the perspectives of recipients asserts their agency and capacity to respond to abuse and challenges not only perpetrators but also the conceptual and ideological context that underpins offensive behaviour. In common with wider victimological perspectives, we emphasize that recipients of online abuse retain the capacity to respond in a range of ways that cannot simply be ‘read’ from the content of the abuse. The diversity of responses to online abuse is further elaborated below.

As outlined above, given that it has now been established that there are forms of abusive, threatening and violent online communication towards women, we focus specifically on that phenomenon directed at women who engage in feminist debate online. Anecdotal evidence indicates that online feminism attracts both shocking levels of threats and violence but also more routine, even mundane levels of sexism, prejudice and misogyny. Rather than treating it as a form of communication, we locate it in wider forms of behaviours which constitute VAWG. This approach enables us to learn from examination of and theorizing about VAWG, to better understand the social role of online abuse of feminists and its impacts. In doing this, we contribute not only to knowledge about patterns and impacts of victimization but also to development of criminological analysis of offending and harm experienced online. While there is clearly a growing body of work exploring the extent and techniques of online crime, there remains relatively little empirical or theoretical insight into the nature and impact of such offending, including its impact on engagement in political movements.

The next section of the paper outlines our methodology. Then, analysis is presented of the nature of gendered abuse online, its impacts on recipients and the various responses developed. In the conclusion, we consider the extent to which existing knowledge of VAWG in real-world contexts can be applied to this emerging field.

Methodology

There are several reasons for this study’s focus on women who engage in feminist debate. Feminist civic engagement is flourishing and of growing academic interest (Dean and Aune 2015; Lewis and Marine 2015). Online activity has been significant in the resurgence of feminist communities, debates and theories. Anecdotal evidence suggests that women espousing feminist views are particularly targeted for abuse online. This cohort therefore provides a starting point for further research about online abuse, gendered and otherwise. We do not claim that their perspectives or experiences can be extrapolated more widely. However, given their interpretations and perspectives, those who engage in feminism online may offer useful insight into their experiences, given the centrality of VAWG in feminist politics. Finally, a particular aim of the research was to explore how those abused online respond to the abuse and whether this constitutes a form of activism.

To explore these matters, two data collection strategies were used: a survey and in-depth interviews. An online questionnaire (conducted June–October 2015) contained multiple-choice and open questions about: the use of social media for feminist debate; the nature, frequency, duration and volume of abuse; forms of social media used to communicate abuse; the topics being discussed when abuse began; what made the communications feel abusive; whether any aspects of identity (such as gender, sexuality, ethnicity, disability) were targeted; how many perpetrators were involved and whether they were known to the respondent; whether the abuse was linked to offline experiences; the emotional and offline impacts; responses to the abuse; reporting behaviour and satisfaction with responses from others. Data were gathered about a range of online abuse—harassment and sexual harassment, threats of physical and sexual violence, flaming and trolling, stalking, electronic sabotage, impersonation and defamation—and definitions, drawn from relevant contemporary research, were provided for each. Asking about ‘general’ and specific (‘the last incident’) experiences captured the range and specificity of abuse without focusing disproportionately on experiences which might skew the data towards the ‘worst’ incidents. Responses indicate that abuse can be experienced over extended periods, so an individual ‘incident’ can consist of a single communication or of many, over weeks or months. The open questions generated fulsome responses, creating an extensive qualitative data set.

A sample was created for the survey by inviting personal/professional contacts to complete the questionnaire and promote it among relevant networks. Initial contacts included about 60 women’s organizations, approximately 30 individual feminists, including journalists, activists and academics, and five organizers of feminist events concurrent with the research. This approach enabled snowball sampling, reducing the impact of initial selection bias and reaching a greater number and range of participants. To further reduce bias, we paid attention to the type of politics and topics supported by the individuals and organizations contacted; e.g., we made sure not to invite only those supporting radical feminism, or only those focusing on violence against women, but also local, regional, and national networks of Black women, religious women, service providers and activists. For several reasons, the research was not explicitly promoted to high-profile feminists, although, due to anonymity of respondents, we do not know whether such respondents did participate. We were aware that some high-profile feminists are subjected to extreme levels of abuse which may not be typical. Many of these high-profile feminists are regularly trolled, and so we risked the questionnaire being sabotaged through trolling or cyberattacks. These concerns about security were central to designing the methodology; the questionnaire was hosted on SurveyMonkey which was deemed to provide sufficient data security and some protection against sabotage by preventing more than one response per IP address. These strategies proved effective; only 14 responses were deemed to be inauthentic (so identified because they included irrelevant, extensive text and/or sexualized responses).

In total, 226 valid responses were received. For this analysis, respondents who self-identified as men (n = 9) were screened out of the data set, so only those who identified as women were included.² It was not possible to verify the credentials reported and, in common with other online surveys, we have to rely on the integrity of respondents’ reports. The characteristics of the final sample are outlined in Table 1.

Table 1

Survey respondents’ demographic characteristics (n = 226)

	n	%
Race/ethnicity
White	182	88
Asian (Indian, Bangladeshi, Chinese)	7	3
Black (African, Caribbean)	6	3
Mixed	6	3
Other	7	3
Age (years)
18–25	40	18
26–35	64	28
36–45	64	28
46–55	41	18
56+	17	8
Location
Northern England	53	23
Central England	55	24
Southern England	89	39
Scotland	15	7
Wales	9	4
Northern Ireland	5	2
Sexual orientation
Heterosexual	111	54
Bisexual	50	24
Lesbian	34	17
Other	10	5

The second data collection strategy was a set of 17 in-depth interviews exploring emergent themes from the survey data, particularly responses to and impacts of abuse. Interview volunteers were recruited through the survey and further snowballing. Their demographic characteristics are presented in Table 2. Interviews were conducted via Skype, telephone or in person, typically lasted about an hour, and were recorded and transcribed.

Table 2

Interview respondents’ demographic characteristics (n = 17^a)

	n	%
Race/ethnicity
White	12	75
Asian (Indian, Bangladeshi, Chinese)	1	6
Black (African, Caribbean)	2	13
Other	1	6
Location
Northern England	4	23
Central England	1	6
Southern England	11	65
Scotland	1	6
Sexual orientation
Heterosexual	7	54
Bisexual	3	23
Lesbian	2	15
Other	1	8

^aSome respondents declined to provide some demographic characteristics.

Qualitative survey and interview data were analysed thematically, through collaborative processes of reading and rereading the data, discussing emerging themes and then coding the data. The study has benefitted from the exceptional richness of data provided by respondents. Unedited data are presented below in line with Jane’s (2014a) call for unexpurgated documentation, to break the tyranny of silence around cyber violence against women.

The quantitative analysis distinguishes between levels of online activity as a proxy measure for levels of engagement in online feminism. Use of social media to discuss feminism ranged from less than one hour a day (85 respondents)—the group we call ‘low users’; 1–2 hours per day (73 respondents)—‘moderate users’; to 3 or more hours a day (68)—‘high users’, which included 50 respondents who were intensely engaged, discussing feminism online for 6 or more hours a day. As this was not a random sample, inferential statistics such as chi-square could not be used to generalize to the wider population, so bivariate relationships between variables are examined only to establish patterns within this sample.

Findings

The following three sections present qualitative and quantitative data about the nature of the abuse, its impacts, and social and legal responses to it. This leads to a conclusion that considers the findings in terms of what is known about VAWG more generally. We argue that online abuse of feminists is best understood, analysed and theorized as a form of VAWG.

The nature of the online abuse

In an effort to understand the context of online abuse, respondents were asked about their online activity. The most commonly used social media for feminist debate was Twitter (80 per cent of the sample), followed by Facebook (74 per cent of the sample) and blogs (35 per cent). Other forms (e.g. email, news sites, private/invite-only fora) were less commonly used. Respondents experienced most abuse on Twitter; some respondents reporting that abuse started when they began to use Twitter. Eighty-eight per cent of those who use Twitter regularly for feminist debate had been abused on it, compared with 60 per cent of Facebook regular users, 46 per cent of blogs regular users and 29 per cent of news sites users. The greater frequency of abuse on Twitter might be due to the open access of this social media relative to others. Given the popularity of Twitter, the specific aspects of this platform as a site for abusive communications are worthy of further consideration.

The data show that there is no single pattern of experiences of online abuse. Rather, there is a continuum of online abuse ranging from concentrated, frequent, highly threatening and hateful to, at the other end of the spectrum, comparatively sporadic and less inflammatory, unpleasant, non-threatening messages. To some extent, this reflects wider experiences of victimization, including VAWG (Kelly 1987) and some forms of hate crime. These include extreme incidents but also routine low-level offending, which might have a significant impact for the very reason that it becomes normalized and persistent (Bowling 1998; Chakraborti and Garland 2009).

As might be expected, ‘high users’ are exposed to more abuse with a third (35 per cent) of this group reported ‘constant’ abuse. However, this level of abuse was also a feature for others, with 24 per cent of moderate users and 16 per cent of low users receiving ‘constant’ abuse. Only 7 per cent of the sample reported that they experienced it less than once a year. Respondents were asked about ten types of abuse, using terms widely used online.³ Experience of multiple types was common; only 17 per cent had experienced a single form and a quarter of high users had experienced eight or more types (compared with 7 per cent of moderate and 6 per cent of low users). Figure 1 indicates that high users experience greater levels of abuse across all ten types. However, the difference in their experience was more marked in relation to some types (physical threats, sexual harassment, incitement to abuse, sexual threats, stalking, electronic sabotage and impersonation) than others (flaming and trolling, harassment, and defamation). The current data set does not enable explanation of these differences in experiences of type and incidence of abuse, but they are worthy of further study.

Fig. 1

Types of abuse ‘ever’ experienced.

At one end of the continuum, threats to rape and to kill were commonly reported. Three responses to the question inviting open text details of the most recent incident illustrate the more extreme forms of abuse:

I was told I deserved to die a painful death. (Respondent 74⁴)

I sent out a tweet saying ‘Guns are not the issue, racism is’ and one male sent me a message of abuse (can’t remember exactly because I reported and blocked) saying along the lines: ‘You fucking spaz feminazi would shoot you in the head if I could then you wouldn’t give a fuck about gunlaws’. (Respondent 54)

Threat to kill me and my son. (Respondent 196)

Not all experiences of abuse were so threatening. Examples of the other end of the continuum include:

Hostile violent comments (not threats towards me). (Respondent 189)

Told to shut up, sworn at etc. (Respondent 15)

Abusive language in private mails and on topic threads. (Respondent 92)

Although these comments suggest some experiences are more mundane, this does not imply that they were experienced as less impactful or harmful. The experience of abuse is extremely subjective, making it difficult to create reliable scales of severity. Instead of developing hierarchical scales, we rely instead on the idea of a continuum and on respondents’ reports of their reactions to and the impacts of abuse, as we discuss in the following section.

Media coverage of online abuse has highlighted the sexualized nature of much abuse of women. Our data show that 40 per cent of the sample experienced sexual harassment and 37 per cent experienced threats of sexual violence; high users were more likely to have experienced these (see Figure 1). These included rape threats, as further open text survey responses illustrate:

I was sent messages on a daily basis, sometimes several times a day, on a number of platforms telling me that I was a slut and a whore, that I’m not a real lesbian because I’ve ‘had sex’ with men, despite the fact that my only experience with and around men is as a trafficking survivor. I was called a ‘cum-whore’, a ‘bi-slut’; I was told I deserved my rapes, I was told it was ‘regret not rape’. I was told that I ‘enjoyed it’, I was told that a must have just been a horny kid (I was trafficked from the age of 5), I was told that dykes don’t like dick so I can’t be a lesbian. I was told to kill myself, I was threatened with rape, I was told I like cock, I was told I loved the taste of semen. (Respondent 198)

I was Tweeting about #EverydaySexism and received emails from several men detailing how they were going to sexually abuse me to remind me who was in control in society. (Respondent 103)

I was abused for discussing breast feeding in public! Told that I should never breed, that he should be able to wank off next to my kids and have sex next to me and my kids on a bus! Called disgusting and a disgrace to women. (Respondent 127)

Some received images as well as written abuse; high users were more likely to receive these (33 per cent, compared to 24 per cent of moderate and 22 per cent of low users). Many of these were sexualized, including respondent’s own image incorporated into pornographic content:

My image was photo-shopped on to various other images and posted to everyone in my uni class. (Respondent 90)

Following my tweet about a feminist event, I received a tweet the next day, of three photographs from an unknown sender. The photographs were of a white, older, long-haired unknown naked male, bending over and stretching his hugely gaping anus open to the camera (and so, to me as the viewer), with a really horrible distorted/angry expression on his face. (Respondent 165)

Media coverage of this topic tends to focus on cases where recipients receive huge volumes of online abuse. While most of our respondents had not experienced mass abuse, a minority reported very high volumes from a large number of perpetrators (6 per cent reported there were 50 or perpetrators in their last incident):

I took a picture of a pink office supply item advertised as ‘for women’ and made a sarcastic comment about how now women can work too and tagged #everydaysexism in an attempt to point out even these little things are still a representation of sexism. This was immediately shared by GamerGate all over Twitter, Reddit, and various other sites. Within a few hours it had over 25,000 views and 650 abusive comments on Reddit not including the comments on Twitter. My picture, name, twitter handle, location, profession, were all shared. I feared for my online security as Gamergate is known to hack people’s accounts. It took days before I could get moderators to remove my personal information that was shared across sites. I was threatened with rape, abuse, etc. (Respondent 126)

I said something about women in science (I am a chemist). I got a barrage of abuse targeting both me and my daughter (not my sons, whose photos are also on my feed - they were never mentioned) - it was mostly variations of ‘fuck off back to the kitchen’ It went on for months and every time it started up again men would encourage others to join in. (Respondent 31)

One person ‘set’ their 10k followers on me for talking about radical feminism. I was told to ‘get raped’ ‘Die in a fire’ & that I needed ‘excorcised’ to name but a few. (Respondent 130)

Although these high-volume attacks were often relatively short-lived, they could be sustained in a manner that makes them akin to harassment. Examples include:

It was defamatory and aimed at getting me fired. It was also relentless from a person I have blocked. Felt invasive and intrusive - they are monitoring me even though I’ve blocked them. (Respondent 67)

The most recent is the most ongoing. He sets up accounts to start discussions under pseudonyms so he can abuse me, incites other people to abuse me, emails to tell me he is watching me (he largely isnt its only online activity he sees now). It has decreased in frequency and is only occasional now. (Respondent 116)

At the other end of the continuum, for 47 per cent of respondents the last incident was comprised of fewer than ten abusive communications. Low users were most likely to receive a single communication (51 per cent compared with 44 per cent of moderate and 31 per cent of high users):

I had retweeted some stories about the street harassment of women. A stranger tweeted at me a couple of times saying I was only concerned about this issue because it would never happen to me as I was ugly and obviously frigid etc (w specific reference to my twitter profile photo). (Respondent 176)

I linked to an old Spectator article in which Boris Johnson wrote that the problem with the UK was that British men could not ‘control’ their women (the context was pregnancy/reproduction). I tweeted it out drawing attention to its repugnant message, and how it was at variant with Johnson’s image as a good-natured clown. Most of the responses were positive/discursive but one was extremely personal, telling me to ‘fuck off you crazy feminist c*nt’ or something along those lines. (Respondent 6)

In response to a comment I made about male violence I received a tweet from a man who made sexual insults and suggested that my position on male violence was because, as a feminist, I wasn’t getting enough sexual attention from men (phrased in an abusive and sexually graphic way). (Respondent 57)

These extracts illustrate considerable diversity in the duration and volume of abuse. The same can be said about the perceived characteristics of perpetrators. As far as respondents could ascertain, most commonly there were one or two perpetrators (49 per cent) and the incident lasted for one day (44 per cent). In contrasting intensity, 31 per cent received 10–50 communications in the last incident, for 35 per cent the incident lasted about a week and for 33 per cent there were 3–10 perpetrators. Respondents were not asked about the gender of abusers because it is not possible to reliably ascertain this online. Many intimated that abusers were male but it is possible that some may have adopted male online personas; indeed, one of the first people in the United Kingdom convicted of sending abusive tweets was a woman who included references to rape as well as threats to kills (see R v John Raymond Nimmo and Isabella Kate Sorley 2014). This indicates that online abusers may adopt the discourse of misogyny regardless of their gender. This is not ‘male violence’ so much as ‘masculinized violence’; i.e., violence that is generally perpetrated by men against women and girls, but may be perpetrated by women, and which draws on and generates misogynistic discourses. The finding of recent research by Demos (2016), that half of those sending abusive tweets containing the words ‘slut’ and ‘whore’ were women, was emphasized in many press reports of the research, indicating ignorance that misogynistic terms have been incorporated into public discourse.

For 61 per cent of respondents, the perpetrators were thought to be strangers, while for a third, the perpetrators were members of their online community or a known persona. A fifth reported that the abuse was sent by a ‘well-known’ troll.⁵ In 77 per cent of cases there was no link to offline abuse; however, 23 per cent reported that their experience of online abuse was somehow linked to ‘real-world’ abuse. Respondents reported those experiences as particularly impactful:

he named the train station local to me in an oblique way. Later on the same forum he had a conversation with himself about making a special visit to a particular person (me) & named the station he’d be catching the train to. This man is a known rapist...He specified his visit would take place over the w/e. The police advised me not to stay alone at my home - or, if I did, to phone them if I heard any odd noises. I live alone so of course it unnerved me. I consider myself to be strong & independent, but he managed to intimidate and frighten me. (Respondent 85)

I started a girl only group at a secondary school to discuss issues affecting young women. The fact that it was girl only meant a group of five boys took exception and began to Tweet about me and post defamatory messages on Facebook. This was all amongst students (I’m a teacher) and I did not see them until students showed me them on their phones. Ultimately, the campaign led to me leaving teaching. (Respondent 132)

For some of the respondents, the sexism or misogyny in the online abuse intersected with other forms of oppression, such as racism and homophobia:

We are talking about the conscious and the unconscious here. It was not precisely obvious that I am a working class Jewish woman, but bullies find it very easy to sniff out people who might be vulnerable to attack - or else they felt threatened by me, so they had to take me out. (Respondent 163)

I had used a hashtag when discussing a recent news event and started to receive hostile or derailing tweets from racist and anti-feminist users who appeared to be monitoring the hashtag in order to prevent feminists having an uninterrupted discussion with each other. (Respondent 115)

I was quoted in a press article speaking out about violence against women.the Facebook newspaper page included (not anon individuals) comments like ‘she needs a good kicking in the cunt’..she’s a man hating lesbian and needs a good fucking to sort her out ‘..’someone should shut her up by sticking a cock in her mouth’..’why doesn’t someone kick the shit out of that ugly bitch’...I could go on and on giving examples of the abuse posted. (Respondent 94)

These data show the nature, duration, volume and frequency of online abuse. We now turn to the under-researched issue of the impact on recipients.

Impacts of online abuse

For the majority, receiving abusive messages was significant. In relation to the last incident, only 7 per cent reported they were ‘not bothered’. Again, there was a continuum of responses from ‘it was just one of those things, I shrugged it off’ (39 per cent of the sample), ‘I was upset and it had a significant impact but I’ll get over it’ (36 per cent), to ‘it was really traumatic and I keep thinking about it even thought I don’t want to’ (26 per cent). For some, the sense that it was ‘just one of those things’ points to the normalization of online abuse. It has become a part of everyday online life and some respond by working to ‘manage’ their emotional reactions to minimize the harm done by it. The following demonstrates the kind of ‘emotion work’ (Hochschild 1979) involved in being able to ‘shrug it off’:

It’s something I experience quite often, and just for being a feminist. On an almost daily basis I have to deal with messages from men, many of which contain pictures or content that’s sexual and unwanted. It upsets me greatly but I’ve gotten used to it and I can’t afford to let it upset me. (Respondent 9)

Another strategy to manage impact is to compare one’s own experience to others’. Commonly respondents minimized their own experiences in light of others, sometimes blaming themselves for using social media incautiously. Perhaps significantly, there were no instances of respondents emphasizing the magnitude of their own experiences in light of others. For example:

It happens to all women so it’s almost not worth mentioning as it’s so unremarkable. (Respondent 201)

It was a much more minor incident than the sustained harassment (in some cases from users with offline positions of power) that friends of mine have received - relatively speaking it did not matter that much. (Respondent 115)

This ‘normalization’ occurred even in response to death and rape threats, which might reasonably be judged as very serious. Given that some had received voluminous abusive messages, detailing threats of physical and sexual violence, a ‘simple’ abusive message may, in comparison, be experienced as relatively mild. ‘Normalization’ can be an effective strategy for dealing with online abuse but raises significant concerns about the longer-term, insidious harm of considering death and rape threats as ‘normal’. The accumulative effect of routine, everyday abusive encounters can be highly significant. Indeed, respondents who experienced it more often (‘most weeks or constantly’) were more likely to experience stronger reactions, as shown in Table 3. The majority (64 per cent) of those who found it ‘really traumatic’, received abuse ‘constantly’ or ‘most weeks’, while the majority (71 per cent) of those who ‘shrugged it off’ received it once a month or less often. While some women interviewed stated that the frequency of online abuse lessened its impact, the survey data indicate that, in general, more frequent abuse increased impact. This indicates that far from becoming diluted by its frequency, the effects of the ‘wallpaper of sexism’ (Lewis et al. 2015) are cumulative and exacerbated. Again, there is a parallel with the normalization thesis whereby offline VAWG leads survivors to downplay their experiences because they are part of the ordinary and mundane routine of everyday life.

Table 3

Frequency of abuse and reaction to it

	Reaction to last incident of abuse
	‘I shrugged it off’		‘Upset but I’ll get over it’		‘Really traumatic, I keep thinking about it’
	n	%	n	%	n	%
Frequency of abuse
‘Constantly’ or most weeks	21	30	55	26	30	64
Once a month	17	24	10	15	5	11
Less than once a month	33	47	20	30	12	26

As indicated above, some respondents reported the abuse had serious impacts. Some referred to ‘triggering’ (Lewis et al. 2015) whereby reactions to other abusive experiences are relived. Triggering occurred when women had previous offline experiences of abuse, for example:

Exacerbation of pre-existing mental health conditions. Particularly PTSD [post-traumatic stress disorder] related to past sexual violence. (Respondent 79)

Depression and anxiety, triggering of past experiences of real-life abuse, increased mistrust of people. (Respondent 69)

More routine impacts of receiving abusive messages included emotional and physical effects. Forty-two per cent of respondents reported they were ‘worried’ after the last incident and ‘stress’ and ‘anxiety’ were frequently reported. For example, one respondent reported she felt ‘stress, fear, anxiety, depression, hopelessness, anger’. For some, the impacts manifested physically, but mental health consequences were also identified, as the following comments illustrate:

I ended up being prescribed beta blockers in the short term as I would wake up in the night with palpitations. (Respondent 85)

I have severe ME. My energy is very limited, and simply talking about the weather online is tiring. Talking about feminist issues is something I have to ration or the energy output makes me iller. A mass of abuse for it made me substantially more sick, and it took around 2–3 months before I was back to a useful level of ‘health’ again. (Respondent 83)

While harmful consequences should not be minimized, many respondents felt in some way ‘galvanized’ by their experience. Fifty-four per cent agreed it made them ‘more determined in your political views’. A third (33 per cent) agreed it made them feel ‘motivated to continue to engage in debate’ and ‘motivated to do something’ (34 per cent). Analysis of impacts shows that, while emotions such as anger, worry, vulnerability, fear and sadness reduce over time, feelings of being galvanized to act increase over time. This complicates the claim that online abuse ‘silences’ women; while it undoubtedly has that impact for some women at some times, abuse also galvanizes participation in this form of civic life. We discuss this further in the following section.

Personal, social and legal responses to online abuse

Just as the nature and impact of online abuse varied, so too women who receive abuse respond in a variety of ways. This section explores practical and political responses to online abuse.

In contrast to offline VAWG, our survey suggested relatively high levels of reporting, detailed in Table 4. While 30 per cent said they did not report the last incident to anyone, 70 per cent talked to informal contacts (friends, family, online contacts and colleagues) and nearly half (43 per cent) to formal contacts (police, Internet service provider (ISP), Twitter, Facebook). There may be several reasons for this. The relative ease of reporting online and the availability of documentary evidence reduce two of the key obstacles to reporting some forms of offline VAWG. Rates might also be higher among this group because the sample was comprised of respondents attuned to feminist politics and so perhaps more inclined to report.

Table 4

Reporting of ‘last incident’

	Contacted for support		Satisfied with response
	n	%	n	%
Informal contacts
Friends	131	58	104	79
Online contacts	91	40	68	75
Family	86	38	46	53
Colleagues	50	22	23	46
Formal contacts
Facebooka	41	62	4	10
Twittera	84	54	14	29
Police	44	19	7	16
ISP	37	16	1	3

^aPercentages for Facebook and Twitter are calculated from the population of those abused on this form of social media.

Levels of satisfaction with the responses of those consulted vary markedly, with much more satisfaction with informal than formal contacts. The survey found particularly low rates of satisfaction with ISPs (3 per cent of those who’d reported) and Facebook (10 per cent). Satisfaction rates with the police were also low at 16 per cent which compares badly to general levels of victim satisfaction with the police. Data compiled by Her Majesty’s Inspectorate of Constabulary (HMIC undated) show overall satisfaction with police among crime victims was 84 per cent in England and Wales in 2014. Reasons for dissatisfaction with the police response included perceived indifference to online abuse, responses that echo criticisms of responses to VAWG offline. Some concerns mirrored wider challenges in terms of policing cyberspace. The interviewees quoted below noted officers were uncertain about the legal and organizational capacity to respond:

My local police, it’s not their fault, it’s genuinely not their fault, some of them are lovely people, but they don’t have any idea how to deal with it, so they say things like ‘have you upset anyone recently?’ and I’m like, ‘yeah, well it’s my job’… And then they say things like, ‘have you thought about you know changing your Twitter handle?’ or, you know, like ‘can you not just block these people?’ or things like this. (Interview 10)

It’s really difficult to do anything because they always say, it’s, you know, not really a threat. I don’t know what the actual law is at the moment because I know they’re having quite a few changes but I think the police tend to think until something actually happens they can’t do anything so they will monitor it but they will actually wait for him to actually go and harm someone before they do anything. (Interview 11)

One explanation for the low rates of reporting offline abuse is a sense of shame or stigma, and the belief that such victimization is a private matter to be dealt with alone rather than a concern for social or legal agencies. By contrast, women experiencing online abuse seem less likely to invoke notions of shame and stigma; 14 per cent reported embarrassment or shame about the last incident and 9 per cent said this preventing them disclosing it. However, a third did not report because they did not believe anyone could do anything about it or would take it seriously.

As well as reporting to others, participants had various ways of responding to the perpetrator/s of the online abuse. A third (34 per cent) confronted the perpetrator/s online; high, moderate and low users had comparable rates of confronting but this was more likely to be an ongoing rather than one-off exchange for high (65 per cent) than for moderate (45 per cent) and low (48 per cent) users. This suggests that shame and stigma were not strong responses to online victimization. A third (35 per cent) ‘used security measures to protect myself’ (e.g. blocking contacts or changing security settings); more high users (49 per cent) than moderate (34 per cent) and low (26 per cent) did this. A quarter said they ‘discussed the problem online to draw support’; this was more common for high users (34 per cent) than moderate (25 per cent) or low (13 per cent) users, perhaps indicating a stronger activist identity and online support network. A quarter of the sample (26 per cent) reported that they were ‘more cautious about how I took part in discussions and/or which topics I discussed’, suggesting that, as noted in the literature on cybersecurity and offline crime prevention, recipients felt ‘responsibilized’ to take measures to prevent recurrence. In circumstances where individuals become responsible for their own security, the role of public agencies becomes reduced, which may result in a deteriorating cycle such that reporting to external agencies comes to be seen as less worthwhile. The focus then shifts towards self-protection rather than the collective need to tackle perpetrators or to create safe spaces for civic engagement (Lee 2007). This ‘responsibilization’ might be attributed to a neo-liberal influence in discourses around crime. However, it has a longer history in terms of VAWG whereby victims are blamed and held responsible for their own safety.

Discussion

This research has shown that, far from being a form of behaviour unique to the cyber environment, online abuse of feminists shares several features of offline abuse of women. As with real-world VAWG, forms are multiple and varied. Most women experienced multiple types of abuse and almost half experienced it as a routine part of their online lives. In this way, it is experienced as a course of behaviour rather than a set of individual acts. Indeed, women reported their frustration with police responses which treated each individual communication as a discrete act, rather than grasping the harm caused by the accumulation of abuse. This reflects broader concerns that the criminal justice systems fails to conceptualize the cumulative impact of apparently low-level offending, a concern that informed legislation to respond more effectively both to antisocial behaviour and hate crime (Burney 2009; Chakraborti and Garland 2009). Similarly, intimate partner violence has high rates of repeat victimization (Walby et al. 2015), although it is often framed and treated, problematically, as discrete acts (Kelly and Westmarland 2016).

As with some forms of VAWG, some online abuse seeks to sexually degrade women. Significant features of online abuse are sexual harassment and threats of sexual violence, experienced as degrading violations, and violent pornographic depictions are also used.

There are similarities too in terms of the perpetrators. VAWG is committed most often by perpetrators ‘known to’ victims, demonstrating that risks occur not just in public, but also private, familiar and familial spaces. Even in the relatively anonymous online environment, a third of our sample reported that perpetrators were members of their online community. VAWG online and offline is often committed by lone men but harassment in public places, in particular, can have a public, performative aspect; Phipps and Young (2015) see ‘laddish’ harassment as a form of homosocial bonding. Online abuse is experienced in an ambivalent space that is simultaneously private and public. Social media may offer forms of private space whereby interaction is performed only in front of those ‘followed’ or ‘befriended’ rather than to a wider public. However, it is easily recirculated and might be considered public in the sense that it is shared, even if only among invited contacts. In terms of the content, many respondents suggested perpetrators intended their messages to have communicative action that served to exclude targets from online spaces often conceived as ‘creative commons’, a place in which cultural, social and political exchanges occur. There is a ‘performative’ aspect to online abuse; the motivation and impact may be not only to demean or exclude the individual victim, but thereby to build up the identity and status of the communicator. The experience of receiving abuse may be individual, private and solitary, even while the communication of abuse is public, social and performative.

Online abuse, like offline VAWG, has significant impacts. While both are ‘normalized’ in wider discourses in ways which support dominant heteronormativity and gender relations (Hlavka 2014), this does not undo the harm caused. At the most extreme, both can lead to symptoms of post-traumatic stress disorder (see Pain’s (2015) comparison of intimate partner violence with warfare). The frequency of abuse is a feature of its impact and our analysis suggests that, far from diluting its effects, frequency exacerbates significance. However, it is apparent that, far from being ‘silenced’, some are galvanized by experiencing online abuse and motivated to continue political engagement. This follows in the tradition of VAWG emerging as a social problem largely through the activism of survivors of violence (Emerson Dobash and Dobash 1992).

As well as similarities, this comparison highlights key differences between online and offline abuse. While offline abuse is characterized by low levels of reporting, our respondents, who may not be typical of online populations, were frequent reporters. In reporting to informal and formal contacts, they challenge patterns of shame, stigma and self-blame and instead involve others in confronting it. While shame is a ‘self-regulating practice … of male power’ (Baker 2013: 145), this group of feminists appear to disavow these norms in challenging online abuse.

There may be other distinctive features to the experience of online abuse. Williams (2006: 103) argues that online, where identities are less secure because of their reliance on text, ‘the permanency and visibility of violent narratives online gives a certain longevity’ to the abusive text. Traces of abuse remain, occupying cyberspace and infiltrating online identities and reputations; the tentacles of abuse can be enduring in ways unique to the online environment. There may also be significant differences between the experience of online abuse and the motivation behind it. Perpetrators may be motivated by a transitory sense of entertainment, boredom or ‘humour’ and be unaware that abuse may be experienced as intensely threatening and frightening, with enduring impacts; further research about perpetration of online abuse would reveal more about the motivations and intended effects.

Other future research could usefully focus on strategies to mitigate impacts on individuals and on online social and political activism. To address the question about whether feminists, or women, are at particular risk of abuse, further examination of online abuse in other social movements, in wider civic debate and in popular culture, would be valuable. In light of concerns about apparently deteriorating behaviours in wider political debate, scholarship could also examine whether/how abusive engagements come to (re)shape social movements and political dialogue. Further research could build on the methodological limitations of this study—the relatively small, self-selected sample group which is appropriate for a purposive study of feminists who use social media, could be complemented by a wider-scale study of gendered and abusive online engagements, and a comparison of women and men’s experiences. As the analysis presented indicates, a difficulty of Internet surveys is that online identities are malleable and it is impossible to verify the credentials of participants. Moreover, as with other victim survey research, the various forms of abuse discussed here are inherently subjective and interpretive; experiences of abuse will be filtered by the wider social and personal characteristics of each individual.

Once heralded as a haven for free speech and democracy, the Internet is also revealed as an extension of offline gendered realities, where violence and abuse is the ‘wallpaper’ (Lewis et al. 2015) of everyday life for women and girls. As women and girls challenge patriarchy offline and online, and seek to occupy these spaces on equal terms, we have witnessed a ‘backlash’ against demands for voice and space in civic engagement. Thornton (1995: 318, cited in Lee 2007: 129) argues women are rendered ‘less fit for public sphere responsibilities according to both historical and prevailing democratic norms’; exclusionary intent in online space may be an extension of attempts to exclude women from public spaces, town halls and common spaces of contemporary and historical democracy.

For criminologists, online abuse demands urgent attention. It highlights ways in which abusive, harmful behaviour, some of which is criminal, is part of the process through which gender discourses, and feminist politics, in particular, are contested, negotiated and developed. It demands attention to questions of victimization and regulation. To fully grasp the significance of this phenomenon, we argue it is vital to recognize that online abuse towards feminists, and maybe towards women more generally, bears striking similarities to offline VAWG, notwithstanding some important differences, and so should be considered first and foremost a gendered phenomenon which extends the reach of patriarchal oppression. We also argue that attempts to understand and theorize it are aided by attention to experiences of receiving abuse, as well as to textual analysis of communications. As in other spheres, the impact of offensive speech, abuse and threats are experienced differently by recipients dependent on their wider social circumstances and personal biography. A more thoroughgoing victimology is required to better understand the impact of online abuse and the agency expressed or denied by those subjected to it.

Funding

This work was supported by a BA/Leverhulme Small Research Grant.

Acknowledgements

We wish to thank Karen Ross, Jamie Harding and Paul Stretesky for their insightful, helpful comments on earlier drafts of this paper.

1

We use the term ‘violence against women and girls’ (VAWG), rather than ‘gender-based violence’, because it reveals the gendered nature and direction of this violence and because of its contemporary currency in United Kingdom and international policy contexts. We use the term ‘violence’ to refer to interpersonal physical violence (including sexual), and ‘abuse’ to encompass both this violence and other forms of behaviour which are part of the continuum (Kelly 1987) of behaviours (including sexual harassment, sexual violence, coercive control, intimidation, humiliation and threats) reflecting patriarchal oppression which intersects with other oppressions such as racism, disablism, classism and heteronormativity.

2

The low number of men who responded to this survey prevents comparison of women and men’s experiences of online abuse; this would be a valuable topic for further research. .

3

Harassment: repeated unsolicited communications and/or violations of privacy

Sexual harassment: repeated unsolicited communications of a sexual nature, including unwanted sexual images

Threats of physical violence

Threats of sexual violence

Stalking: someone sought and compiled information about you and used it to harass, threaten and/or intimidate you

Flaming & Trolling: posting deliberately inflammatory or off-topic material to humiliate and/or provoke a response or emotional reaction from you

Electronic sabotage: e.g. spamming or viruses sent by someone

Impersonation: your identity was stolen

Defamation: hostile misinformation and false messages were posted about you

Inciting others to abuse or threaten you

4

Survey data are identified as ‘Respondent n’; interview data are identified as ‘Interviewee n’.

5

These options do not total 100 per cent because respondents were asked to select up to three categories which best described the identity/ies of the perpetrator/s.