The Gendered Representational Costs of Violence against Politicians

Women face more harassment and intimidation as politicians than men, but little is known about how this affects representation. I develop a theoretical framework for studying the gendered costs of political violence for descriptive, substantive, and symbolic representation. Testing the framework using interview and survey data on Swedish women and men politicians, I uncover the costs of violence against politicians in all three dimensions empirically. Violence makes women more frequently than men consider leaving politics and enforces male-coded characteristics for political candidates, affecting prospects for gender-diverse descriptive representation. Substantive representation is harmed by violence silencing feminist debates and obstructing women politicians’ policy debate activities. Women’s symbolic representation is threatened by violence coercing women to decrease their visibility.

The Gendered Representational Costs of Violence against Politicians

Sandra Håkansson
Women face more harassment and intimidation as politicians than men, but little is known about how this affects representation.I develop a theoretical framework for studying the gendered costs of political violence for descriptive, substantive, and symbolic representation.Testing the framework using interview and survey data on Swedish women and men politicians, I uncover the costs of violence against politicians in all three dimensions empirically.Violence makes women more frequently than men consider leaving politics and enforces male-coded characteristics for political candidates, affecting prospects for gender-diverse descriptive representation.Substantive representation is harmed by violence silencing feminist debates and obstructing women politicians' policy debate activities.Women's symbolic representation is threatened by violence coercing women to decrease their visibility.

H
arassment of politicians has become a normalized part of politics.Several studies find that women are targeted more often (Collignon and Rüdig 2020;Håkansson 2021;Mechkova and Wilson 2021;Rheault, Rayment, and Musulan 2019;Thomas et al. 2019) and more viciously than men (Erikson, Håkansson, and Josefsson 2021;Ward and McLoughlin 2020).Violencewhether physical or psychological-against politicians has costs for its targets: it affects their mental health, it jeopardizes their political ambition, and protecting against threats requires money and time.Attacking a politician also has immediate implications for democracy, since it can hinder representatives in the process of standing and acting for the represented (see Pitkin 1967).
I focus on the gendered costs that violence against politicians has for women's political representation.How do such costs manifest?Connecting attacks on politicians to the impacts on political representation is imperative because of politicians' unique role in the system of representative democracy.Previous scholarship on gendered political violence has investigated its scope and character, and has emphasised the importance of studying gendered impacts (Bardall, Bjarnegård, and Piscopo 2020;Krook 2020).This is the first study that explicitly and comprehensively theorizes gendered consequences for representation.I develop and test a new theoretical framework for studying the costs for women's descriptive, substantive, and symbolic representation.It specifies which costs may be expected, the mechanisms generating them, and their implications for democracy.
I test this framework on Swedish data, primarily forty-six interviews with women and men politicians, with additional quantitative illustrations based on survey data in three waves (N=22,000).Sweden-generally considered peaceful and without significant cleavages along ethnic, religious, or political lines-is a case where violence would not be expected to inflict significant harms to democratic representation.Costs to representative democracy may be even graver in contexts with higher political instability.Moreover, Sweden features among the highest share of women in politics globally and women have held substantial political leadership positions for several decades.Gender patterns in representational costs found in Sweden are hence likely not exaggerated compared to other contexts.
Using the representation framework reveals a broad range of gendered costs of violence against politicians.While previous studies have focused on violence making women leave politics (Bjarnegård, Håkansson, and Zetterberg 2022;Erikson, Håkansson, and Josefsson 2021;Herrick and Franklin 2019), this is the first to add politician ideals as a mechanism that can deter women's descriptive representation.I find that violence against politicians enforces masculine coded ideals such as A list of permanent links to Supplemental Materials provided by the author precedes the References section.

Sandra Håkansson is a Doctoral Candidate in Political
Science in the Department of Government at Uppsala University (Sandra.Hakansson@statsvet.uu.se,Sweden).Her research focuses on gender aspects of violence against politicians, representation, and political leadership.toughness, which increases women's perceived unsuitability for politics.Previous research on how violence affects substantive politics highlights that women spearheading feminist agendas suffer backlash (Biroli 2018;Sanín 2020).I find that violence in politics hinders both women's and men's substantive representation of women's rights, and that women's substantive policy influence across policy domains is obstructed by violence disproportionately consuming their time and energy.Last, whereas others have highlighted that sexist harassment of women politicians might increase women citizens' distancing from politics (Krook and Sanín 2016), this study focuses on women representatives' visibility as a key mechanism behind symbolic representation.I find that violence makes women representatives decrease their public visibility, which reinforces the image of politics as a male arena.

Why Should Gendered Costs of Violence against Politicians Be Expected?
Following the World Health Organization, I conceptualize violence as acts intended to cause physical or psychological harm.In contexts where physical political violence is endemic, most politicians experience violence of various forms (e.g., Bjarnegård, Håkansson, and Zetterberg 2022;Piscopo 2016).However, psychological violence, such as threats, denigrating tropes, and harassment, is also widespread in peaceful contexts.For example, political harassment has targeted 83% of U.S. mayors (Herrick and Franklin 2019), 87% of New Zealand MPs (Every-Palmer, Barry-Walsh, and Pathé 2015) and 49% of British political candidates (Collignon and Rüdig 2021).
Research on violence against women in politics (VAWIP) outlines how political violence targeting women as women, often using gendered tropes, can be particularly damaging (Bardall 2013;Krook 2020;Krook andSanín 2016, 2020).VAWIP aims to reinforce pre-existing notions about women's unsuitability for politics and hence affects the whole of society by discouraging women's political participation.Even political violence that does not have a gendered motive (such as excluding women from politics) or a gendered form (such as sexual threats), can have a gendered impact (Bardall, Bjarnegård, and Piscopo 2020).
Violence against politicians can be expected to have gendered consequences and affect women's representation disproportionately for several reasons.First, there are gender patterns in violence against politicians.Several studies find that women, particularly powerful and visible women, experience more violence than men (Collignon andRüdig 2020, 2021;Håkansson 2021;Herrick et al. 2019;Mechkova and Wilson 2021;Rheault, Rayment, and Musulan 2019). 1 Furthermore, while men are attacked for their roles as policymakers, attacks on women are more often personal, sexual, and hate-based (Bjarnegård 2021;Bjarnegård, Håkansson, and Zetterberg 2022;Erikson, Håkansson, and Josefsson 2021;Ward and McLoughlin 2020).The scale and kind of attacks women experience is likely detrimental to their political representation.
Second, the pervasiveness of male privilege in political institutions may shape how women and men representatives are affected by violence.Men have privileged access to resources such as political allies, credibility, and support (see, e.g., Bjarnegård 2018;Puwar 2004), which can cushion the impact of violence against them.Being seen as a more legitimate politician can limit the impact of denigrating attacks and supporters and allies can intervene against for example online abuse.Furthermore, gendered morality standards can make character assassinations of a sexual nature, such as infidelity rumours, more damaging for women than men (Bardall 2013;Bjarnegård 2021).The same type of violence can hence have a differential impact on women and men.
Based on pioneering interview research with women politicians, Krook (2020) conceptualizes VAWIP's implications for democracy, human rights, and gender equality.Other studies have uncovered costs for specific outcomes, such as political ambition and campaigning (Bjarnegård, Håkansson, and Zetterberg 2022;Collignon and Rüdig 2021;Erikson, Håkansson, and Josefsson 2021;Herrick and Franklin 2019) or politicians' mental health (Herrick and Franklin 2019;James et al. 2016).The present study broadens the focus compared to this work, covering all classic dimensions of representation (Phillips 1995;Pitkin 1967). 2

Specifying the Gendered Representational Costs of Violence against Politicians
From a societal and democracy perspective, how this violence harms political representation is arguably the main concern related to violence against politicians.Modern democracy relies on representatives linking citizens to governance, and illegitimately intervening in that process undermines the system of representative democracy.By centring the different dimensions of representation, my framework connects some of the costs of political violence identified in previous research to wider implications.For example, affecting politicians' mental health not only constitutes a workplace problem, but jeopardizes political representatives' capacity to fulfil their policymaking role.Using the framework presented here to analyse the costs of violence against politicians illuminates the social ramifications of violence against political representatives as representatives.
Representation research demonstrates that descriptive, substantive and symbolic representation are all required for political inclusion (Lombardo and Meier 2018;Phillips 1995).Furthermore, it demonstrates that the different components of representation do not necessarily coincide or follow from each other in predictable ways (e.g., Htun 2016;Weldon 2002).Different aspects of gendered representation may be affected by violence by different means.A comparatively higher violence exposure among women may be important for certain outcomes, such as depressing women's political ambition.Women's or feminist substantive representation can instead be harmed by attacks on any proponent, male or female, of women's or feminist perspectives.I argue that costs to descriptive, substantive, and symbolic representation should be studied in their own right and without assuming that violence will affect each dimension by the same means.
Table 1 summarizes my framework for analysing gendered costs to descriptive, substantive, and symbolic representation.The first two columns outline costs and mechanisms experienced by political representatives, which is the empirical focus of this study.The last column outlines implications for democracy, which is what ultimately motivates studying gendered representational costs of violence against politicians.The three representation dimensions often overlap empirically.However, keeping them analytically separate helps specify the representational costs with more precision, and disentangles the processes whereby violence affects each dimension respectively.The framework can be applied for this purpose in any context.While studying each dimension alone provides important insights, we can arrive at a more comprehensive understanding for the magnitude of damage inflicted by violence against politicians by studying all three simultaneously.

Descriptive Representation
Presence in political institutions is a fundamental question of democratic equality: representatives should mirror the represented in relevant demographic aspects (Phillips 1995;Pitkin 1967;Young 2000).Violence against politicians can affect legislatures' descriptive composition by decreasing women's willingness to run for and remain in office, termed the supply side of political recruitment (Norris and Lovenduski 1995).Popular debates often warn that violence will cause massive dropouts of women from politics (e.g., Perraudin and Murphy 2019).The few studies that so far investigate gendered impacts on political ambition empirically suggest that women's ambition is equally, or less, negatively affected by violence as men's (Bjarnegård, Håkansson, and Zetterberg 2022;Erikson, Håkansson, and Josefsson 2021;Herrick et al. 2019).Perceived threats can sometimes even increase women's political ambition by heightening the sense of urgency and importance of affecting political processes (Dittmar 2020).There is hence inconclusive evidence in existing scholarship as to whether political violence harms descriptive representation by decreasing women's political ambition.
Importantly, descriptive representation is not solely a result of political ambition.Men and women are evaluated differently, according to gendered standards, in the recruitment process for political offices.Gendered recruitment criteria, and the cognitive association between men and ideal politician characteristics, explain skewed descriptive representation to a high extent (Bjarnegård 2013;Bjarnegård and Zetterberg 2019;Carroll and Sanbonmatsu 2013;Kenny 2013;Niven 1998).Although stereotypes are changing, men continue to be perceived as more rational, strong, and tough, and women as more compassionate, emotional, and weak (Alexander and Andersen 1993;Eagly et al. 2020;Eagly and Karau 2002;Hentschel, Heilman, and Peus 2019;Huddy and Terkildsen 1993).Whereas men are assumed to possess qualities seen as essential for politicians (Schneider and Bos 2014), women face a "burden of doubt" and are required to prove their possession of such qualities (Murray 2014;Puwar 2004).
Violence can amplify this male privilege.If recruiters anticipate a violent political climate, characteristics such as strength and toughness become even more desirable for political candidates.Since women are assumed to be the opposite, violence can increase the burden on women to prove their suitability for politics.For example, external threat contexts trigger a strong preference for male leaders (Holman, Merolla, and Zechmeister 2011).Violence can hence affect what Norris and Lovenduski (1995) call the demand side of political recruitment and can harm representatives' descriptive gender diversity by decreasing the demand for women and female-coded characteristics.Furthermore, violence can distort representatives' horizontal distribution across policy domains.By shaping the characteristics associated with certain policy areas, violence can affect how assignments are distributed (see.e.g.Barnes and O'Brien 2018).Individuals' self-selection into specific policy domains or roles can be similarly affected, such as when these positions come to be associated with higher risks for women than men.

Substantive Representation
Substantive representation relates to how representatives "act for" and are responsive to their constituents (Pitkin 1967).Mansbridge (1999) contends that representing the substantive interests of the represented is the primary function of representative democracy.Violence can compel representatives to consider other things in place of the people they represent and the (party) platforms on which they were elected when they decide whether and how to promote a policy.It can silence specific debates, or representatives of a perspective that perpetrators dislike.Backlash can be expected against actors who challenge hegemonic male policy interests (Sanbonmatsu 2008;Sanín 2020).For example, political violence aiming to silence feminist agendas has been found in Brazil (Biroli 2018).Silencing any individual or group acting in the interest of women's rights or advocating feminist agendas, is thus a gendered infringement on substantive representation (Childs and Krook 2009).
Furthermore, silencing women politicians as a group, regardless of the political agenda, can be conceptualized as a cost to women's substantive representation.Gendered online abuse has been found to circumscribe women MPs' debate activity on social media (Erikson, Håkansson, and Josefsson 2021).Since opinion formation is a vital part of policymaking, silencing women in public debates constitutes a systematic infringement on women's capacity to shape policy on equal terms with men.
Moreover, policymaking requires adequate working conditions.In a report by the Inter-Parliamentary Union (2018), half of the women parliamentarians who had experienced violence said it affected their ability to work normally.Similarly, substantial portions of violence exposed politicians across contexts experienced psychological harm such as fearfulness, sleep difficulty, and difficulty concentrating (Every-Palmer, Barry-Walsh, and Pathé 2015;Herrick and Franklin 2019;James et al. 2016).This hinders policymaking.Having to devote time to handling and devising strategies to avoid violence crowds out the time and energy available for the core of representatives' roles.Stress, anxiety, and time-consuming activities related to violence can hence be conceptualized as not only costs to individuals or to democracy in general, but specifically as costs to substantive representation.

Symbolic Representation
Symbolic representation refers to whether the represented feel that they are being fairly and effectively represented (Pitkin 1967), and is particularly important for politically alienated groups (Mansbridge 1999).Women representatives can function as role-models and increase political participation among women citizens (Wolbrecht and Campbell 2007), enhance their trust in government (Schwindt-Bayer and Mishler 2005), and foster more gender-egalitarian attitudes in society at large (Beaman et al. 2012).Many women politicians worry that violence against them will discourage other women's political participation (Krook 2020, 255).Moreover, in order for the empowering impacts of women's political presence to be realized, women politicians must function as visible cues of the fact that politics is not only for men; in other words, women politicians must be visible (Hinojosa and Kittilson 2020).A key harm to women's symbolic representation that violence can inflict is to coerce women representatives into decreasing their visibility.Power and visibility are associated with significantly more violence for women than men politicians (Håkansson 2021;Rheault, Rayment, and Musulan 2019).A plausible implication is that women representatives might select out of visibility and avoid the most visible roles in politics, as well as avoid media and social media visibility.Ultimately, this risks affecting gendered attitudes to politics negatively.Without visibility, women politicians' presence in politics is less likely to enhance women constituents' affective connection, trust, interest, and participation in politics (Hinojosa and Kittilson 2020).
Another way that constituents' connection with politics can be enhanced is through constituency work, with politicians listening to constituents and being visible as elected representatives who care about their constituents' circumstances (Eulau and Karps 1977).British women politicians report decreasing their availability to constituents by no longer having open surgeries (drop-in meetings with local constituents) in order to protect themselves from violence (Krook 2020, 250).In this way, violence can harm symbolic representation by making activities such as interacting with constituents riskier for women.

Operationalizing the Framework and Specifying Expectations
The previous section describes how costs, mechanisms and implications for democracy are linked theoretically.I next describe empirical tools for assessing whether the costs are gendered.As this study focuses on consequences of violence experienced by political representatives, I describe how each dimension's mechanisms manifest in representatives' experiences and specify theoretical expectations.
Descriptive.The question of whether women's ambition (in general and toward specific sectors) is more affected by violence than men's is analysed by comparing women's and men's experiences.Based on previous knowledge on women politicians' higher exposure to violence and to particularly damaging forms of violence, I expect violence to depress women's political ambition more than men's.
Moreover, I expect politician ideals to become more male coded as a result of violence in politics.Analysing whether violence increases the perceived importance of masculine-coded characteristics (rationality, strength, toughness) and the perceived devaluing of female-coded characteristics (compassion, emotionality, weakness) does not primarily imply comparing women's and men's experiences.The proliferation of such ideals, including perceiving sanctions and rewards connected to fulfilling them, will likely be experienced by women and men representatives alike.It is their differential implications for women's and men's descriptive representation that make them gendered, since the ideals risk reducing the desirability of women as political candidates to party selectors and/or voters.The central analysis consists of establishing whether violence has made such ideals widespread among politicians at large.

Substantive.
Based on previous research on backlash against policy that violates the gender hierarchy, I expect feminist agendas to be targeted with violence.Politicians (regardless of their gender) refraining from proposing or debating feminist policy because of violence indicates that violence harms gendered substantive representation.Hence, it is the topic as such rather than whether women or men representatives are affected in this aspect that makes it a gendered representational cost.
I also expect women's substantive influence at large to be reduced more than men's, based on previous research on women's particular exposure to political violence.This can manifest as women decreasing their participation in policy debates more than men, or violence claiming more of women's time and energy.Comparing women's and men's experiences provides information about reduced debate activities or depleted time and energy as possible mechanisms behind gender-unequal policy influence.
Symbolic.Likewise, the question of whether women representatives, more than men, select out of visibility due to violence in politics is analysed by comparing women's experiences to men's.The same applies to the analysis of whether women representatives, more than men, select out of engagement with constituents on account of violence.While avoiding debates is conceptualized as a cost to substantive representation, withdrawing from public visibility counts as a symbolic cost.Withdrawing from a debate does not necessarily entail that the politician becomes less visible overall as they may maintain their visibility by.e.g., continuing to post non-sensitive content on social media.If they explicitly select out of visibility, e.g., by opting out of social media entirely or declining visible roles or activities, not for the sensitivity of issues but for the simple fact that they will be seen by a broad public, this amounts to a symbolic cost.
Based on previous research, including the increased risk of violence connected with visibility for women politicians, I expect that women will be particularly prone to withdraw from visibility as a result of violence.Regarding distancing from constituents, women representatives seem to prioritize constituency contacts more than men (Herrick 2010;Thomas 1992).This might indicate that they are less prone to decrease their constituent interactions due to violence.

Methods
To investigate these costs, I rely primarily on interviews with politicians.The scarcity of previous research makes qualitative data suitable for exploring how gendered representational costs of violence manifest empirically.Politicians can shed light on the mechanisms outlined in table 1, and semi-structured interviews are suited to capturing how representatives experience a range of impacts.In the Swedish proportional representation system, MPs are both part of the pool of recruits and recruiters.Many MPs are part of selection committees and are familiar with desirable candidate characteristics.They are thus well positioned to provide insights on potential impacts on both the ambition and recruitment side of descriptive representation.Representatives are also well positioned to explain how they may be hindered in their policymaking work.Even though this study does not encompass citizens' political behaviour or attitudes, elected politicians can explain how violence affects their visibility in public arenas and hence generate knowledge on a crucial aspect of symbolic representation.
I interviewed 46 politicians (23 women and 23 men) from all political parties between November 2017 and June 2020, with the purpose of understanding how politicians experience being affected by violence.Interviews are randomly denoted with numbers to ensure anonymity.In recruiting respondents, I aimed to include a variety of perspectives from for example national and local politics, different political parties, geographical regions, and age groups (refer to the interview demographics in table A1 in the online appendix).Most respondents are MPs and have held both national and local office.The respondents' party composition largely corresponds to the parties' parliamentary representation, with the exception of the Sweden Democrats who are under-represented in this sample.To be able to discern in what ways perspectives and experiences are shared by women and men, and in what ways they differ, I interviewed comparable women and men in terms of party, local/national experience, age, and degree of media visibility.This can highlight whether certain consequences of violence are more common for women than men.Including men as well as women respondents can furthermore triangulate the question of gendered impacts.The existence of norms that grant men privilege can be validated by such norms being mentioned by politicians across gender identities.
The sample encompasses politicians with and without personal experiences of violence.To fully grasp the implications for democracy and representation, it is insufficient to study impacts on the direct targets of violence in politics.Attacks on politicians, not least when targeting politically marginalized groups, are often designed and interpreted as message crimes communicating that the social group of the target, such as women or an ethnic minority, do not belong in politics (Krook 2020).Hence, knowledge is needed on how this violence affects larger groups of politicians, and on its influence on norms that may affect politics on a systemic level.While some respondents had gone public about experiencing violence, in the majority of cases I did not know beforehand whether or not the respondent had faced violence.Many examples of how politicians are impacted by violence in politics emerged from respondents who had not themselves been targeted, but whose representation had been affected nevertheless.
Some analyses of the interview material consisted of comparing women's and men's accounts on a certain theme, while other analyses consisted of gaining an understanding of how respondents perceive a theme that has gendered implications in and of itself.Further descriptions of qualitative methods, including interview themes, are found in online appendix 1.
As a complement to enhancing understanding for how violence impacts political representation, I use three waves of the Politicians' Safety Survey, PTU, to investigate the magnitude of some impacts. 3This survey is collected by the Swedish National Council for Crime Prevention and targets all elected politicians.With a response rate consistently over 60% (8,000-9,000 observations per wave and no systematic attrition, refer to table A2) it is the most extensive data material in the world on violence against politicians and its consequences.I use this data to investigate the frequency and size of gender gaps of some consequences among Swedish politicians at large.I hence triangulate the research question by capturing the same phenomena as the qualitative material using another type of data source that, for example, grants respondents complete anonymity and eliminates interviewer effects.The survey data also provide additional information about the extent to which the manifestations of costs described in interviews can be generalized to the general population of Swedish politicians.Combining semi-structured interviews with survey data thus provides in-depth understanding for the processes generating costs to representation as well as their propagation and distribution.
Previous studies analyse correlations between violence exposure and the willingness to remain in politics.The PTU survey instead contains a direct question on whether politicians have been affected by violence in several areas relating to both descriptive and substantive representation (refer to table A3).Hence the PTU captures more outcomes of relevance for political representation than previous research.Moreover, as the question on consequences of violence is posed to all respondents, it is possible to investigate impacts beyond direct targets and to estimate the difference in consequences experienced by politicians with and without personal experiences of violence.Despite their advantages, these data have not been used to analyse consequences of violence against politicians before.
The statistical analyses mainly consist of OLS regression analyses that compare the frequency of various consequences of violence for women and men politicians.Online appendix 2 presents other specifications including logit regressions.Logit estimations consistently demonstrate the same results as OLS regressions in terms of direction and statistical significance of relationships. 4 Analyses are carried out separately for municipal politicians and MPs to account for any possible differences between them in the forms of violence and types of perpetrators.The intensity of violence, as well as the types of perpetrators also vary across parties and municipalities (Collignon and Rüdig 2021;Håkansson 2021).Analyses on municipal politicians cluster standard errors at the municipal level, and analyses on MPs at party and year.

Empirical Manifestations
The empirical analysis focuses on evaluating manifestations of the gendered costs to women's descriptive, substantive, and symbolic representation outlined in the mechanism column of table 1.

Descriptive Representation
Reduced political ambition.From time to time the thought of whether it is worth enduring threats and harassment to remain in politics hits many politicians.Their reasoning suggests that there can be a tipping point when the price becomes too high: That contributed, I can say, in fact, to that I eventually sort of gave up.Because I was entirely spent by then … .The price is getting too high when it starts to creep in and they start threatening … to kill my children.Then I felt that; no.The survey asks whether respondents have left politics as a consequence of exposure to or worrying about violence, and only 1% of respondents answer "yes" (3% of politicians directly exposed to violence, refer to table A3). 5 It is more common to consider leaving politics.Overall, 6% report having considered leaving politics as a consequence of violence and 17% for politicians who had been personally exposed to violence.This demonstrates that even though personal violence exposure matters, there are also residual impacts.A group of politicians who are not personally exposed consider leaving politics due to worrying about violence.
Table 2 shows that violence leads women to be more likely to consider leaving politics due to violence than men.The difference is one percentage point among municipal politicians (Model 1), and five percentage points among MPs (Model 3).These differences are not driven by newcomer status, age, immigrant background, party identity, or differences across municipalities or years.Gender differences are most pronounced among the politicians with personal experiences of violence (refer to table A5).However, there are no gender differences in actually leaving (refer to table A4).This suggests that while violence makes more women than men consider leaving politics, violence may not substantially harm descriptive gender diversity in Sweden through depressing women's ambition.
Similarly, more women than men interviewees talked about seriously considering leaving politics as a consequence of violence.Nevertheless, they had all decided to remain in (or return to) politics for the foreseeable future despite having felt significantly emotionally impacted by violence during certain periods.Some mentioned support and encouragement from party colleagues and allies as important for this decision, while others manage to keep up their motivation without it.MPs comprise a selected group of individuals, highly motivated and invested in politics, as illustrated in the following quote: I have made my life choice.I don't see this as a career.Of course, if you think that this occupation entails hard stuff, then I'll change to another occupation.I don't quite see it like that.What is it that I would leave, then?Shall I leave myself?Others affirmed that being a politician is highly rewarding despite the negative aspects.The findings suggest that among elected Swedish politicians, violence does not lead to depressed ambition on a large scale.Similar findings have been obtained for U.S. mayors (Herrick et al. 2019).It is possible that the ambition of prospective candidates earlier in the pipeline is more severely affected by violence in politics, and that prospective women candidates are particularly affected (see Anlar 2022).
Both the qualitative and quantitative data suggest that violence impacts sectorial self-selection, but not in gendered ways.No interview respondents indicated that violence shapes gendered norms related to specific offices.The most common role that both women and men politicians actively avoid or would turn down is as a spokesperson for migration, due to the violent activists involved.The survey data indicates that very few leave specific roles, but each year, 7% consider leaving specific roles due to violence, and with slightly more among women than men in municipalities (refer to tables A3, A5, and A6).This indicates that while violence intervenes unduly in representatives' sectorial selection, the impact does not appear to be gendered.
Male-coded ideals for politicians.Even more striking than impacts on political ambition, violence imposes obstacles for women's presence in politics by enforcing male-coded norms as desirable politician qualities.Malecoded characteristics such as strength, toughness, and nonemotionality are seen as essential for politicians in relation to violence in politics.A recurrent theme in the interviews is that politicians need to present as strong and if targeted by violence, as unaffected by it.One woman said that the image of what a politician should be like is a person who is strong, stable, and doesn't let anything get to them (Woman-29).Others said politicians get used to "a pretty tough façade" (Woman-43), that you cannot be an MP "if you scare too easily" (Woman-20), and that "it's pretty widespread that we politicians are supposed to be able to take a bit more.You're an MP after all, you're a politician after all, you will get some crap" (Woman-26).Similarly, a man explained that when he had experienced violence he worried that discussing it would give him a victim label: It hinders you from acting and speaking if you're seen as a victim all the time.It becomes an identity: "That's [NN], he's hated".Okay, is that all he is?That's a risk if that becomes your profile.This theme emerged clearly in response to questions on whether respondents had told their colleagues about abuse targeting them in the course of their political work.Most often, they had not.One of the men said the only reason he had talked to his close colleagues about the threats against him was because they were made publicly.If the threats had not been public, he would have been "the man's man that deals with it myself" (Man-10).He said that if he were to be "darn honest", he might have told his colleagues "in a joking manner", if at all.Some attested that there can be a stigma attached to talking about threats because "there's few colleagues that you want to show yourself scared or weak in front of" (Man-37), and "vulnerabilities" are not something politicians "want to Notes: Survey item Model 1-4: "Have you at any point during the previous year, due to exposure and/or worrying considered leaving all political assignments?"Model 5-8: "Have you at any point during the previous year, due to exposure and/or worrying, avoided engagement in or making statements about a certain issue?"The coefficient for Female reports the difference between women and men.The Constant in model 1, 3, 5 and 7 reports the average share of men who experienced each consequence.In these models, the coefficient for Female can be interpreted as reporting the difference between women and men in percentage points.The practices of punishing politicians by portraying them as whiny and not strong enough for politics if they are emotionally affected by violence reproduces informal ideals of heroic, tough, and rational rather than emotional candidates.This ideal echoes traditionally masculine stereotypes that dominate notions of desirable politician characteristics (Alexander and Andersen 1993;Murray 2014).Quotes from two men illustrate how masculinity and toughness were bundled together in relation to not being scared of violence.When asked whether those who threatened him had been at his house, one of the men replied that "I'm almost two meters tall and fairly strong so I guess they haven't dared!" (Man-10).Another man talked about the advantage of having worked as a security guard and practiced martial arts with policemen as it means that he is not easily scared by online abuse (Man-6).Both men and women recognize toughness and indifference to violence as ideal characteristics for politicians.Adhering to traditionally male stereotypes, this discourse reproduces an informal barrier for women's political presence.Women not being believed to embody stereotypically male characteristics at face value enforces the "burden of proof" to demonstrate their suitability for politics (Murray 2014;Puwar 2004).
Not everyone agreed that there is a stigma .One (highprofile) politician did not feel that there is a stigma, but that people can feel embarrassed if they "actually get affected" instead of "just shaking it off" (Woman-22).Furthermore, there are even instances when being targeted with violence can confer status.In some cases, it is seen as a confirmation of one's prominence and importance .Status conferral does not apply across the board, however.It is contingent on remaining tough and indifferent, as well as for what reason and in what way one has been targeted.I asked one man who had been threatened by Nazis what kind of support he received.His reply exemplifies the expectation of being unaffected as well as the status derived from being targeted by political opponents: Well, I didn't really get that much support.I told [my party] but it was more something that you laugh a little about like "ha ha, such nutheads, and they think they can scare us".Like that.So I got some kind of support, but perhaps not questions about how I was doing but rather like a pat on the shoulder for standing up against it somehow.(Man-32) Being patted on the shoulder for standing up against political enemies illustrates that status conferral depends on whether one is targeted for a reason that is considered politically important.Likewise, it is contingent on standing up for issues in line with the party's stance, as explained by a spokesperson for a high-profile issue for which she has been extensively attacked by opponents: I don't experience any stigma but rather the opposite, that people think you're a bit more important … ."Oh but it means that you've done something that matters, that people care about what you're saying" … .If you challenge the party leadership and get beaten up, then nobody will say such things.(Woman-21) Furthermore, status conferral seems to be contingent on the type of violence one has experienced: For me it's pretty easy to talk.But I don't think it is for everyone.

It also depends on what kind of threat you've experienced … .
There was someone who got nude pictures of themselves spread.There it might be harder to, like, say "oy, it wasn't me!" (Woman-26) Gendered forms of attacks are mentioned repeatedly as stigmatizing for targets, as well as difficult to talk to colleagues about due to how they interplay with social gender hierarchies: I think it connects to these personal stories that one might not want to tell.I think precisely about such things that to me connect to abuse and things that have happened to me as a woman.And that they are so closely related to each other makes it maybe not something that I'd want to talk to everyone about.Sexual violence constitutes a particular form of violence in terms of its association to stigma, and is overwhelmingly waged on women politicians (see, e.g., Bjarnegård 2021;Håkansson, and Zetterberg 2022).It is unsurprising that this type of violence does not seem to function as political capital that can confer status to its targets.In this respect, gendered forms of violence lead to gender inequalities in the consequences of violence.The findings related to status conferral shed new light on the conclusion in previous research that men are more often targeted for their political views, whereas attacks on women are more personal, sexual, and related to their gender (Bjarnegård 2021;Erikson, Håkansson, and Josefsson 2021;Ward and March 2024 | Vol.22/No. 1 89 McLoughlin 2020).Men are more often targeted in ways that can grant them legitimacy and admiration, whereas more of the violence women experience is seen as politically unimportant and embarrassing for the target.
Related to the ideal of toughness, several of the men, but none of the women described that they think the problem of harassment and threats against politicians is exaggerated.They claim there are people that speak about their experiences of violence to get attention (Man-2; Man-9, .According to one man, some speak out publicly in order to get media attention and sympathy, even if "they might have received a lot less threats than I have" (Man-41).Another man stated that most people who write violent comments to politicians "would never commit a deed" and because of that he does not see it as "all that serious", but others blow their exposure out of proportion (Man-46).The following quotes are illustrative: My view is that it's stupid to go out publicly and snivel and "Oh, I've been threatened" and such things.There was [a politician] that did so now, and I think it's inappropriate.Because if you really have a threat, if it really is something serious, then you just risk it escalating if you do that.(Man-9) In general, I've always been a bit annoyed at politicians who parade that they've received hatred (laughs).I can be like what the heck, you've chosen this profession, you're a public person, you should be able to handle it … .There's no taboo or anything.On the contrary, I think people (laughs) talk too much about it.Or I think people talk, I think people exaggerate things.(Man-2) The tough and unemotional candidate ideal is not in itself related to a gendered pattern of violence against politicians.Regardless of whether women face more violence, or whether the violence they face is particularly nasty in its form, the mere occurrence of violence against politicians spurs such ideals and as a consequence puts women at a disadvantage.Male-coded characteristics are enforced as ideal for candidates through several mechanisms.Social rewards consist of applauded bravery when being indifferent to violence, and punishments consist of being considered a person who whines and exaggerates to get attention when raising the adverse consequences of violence against politicians.MPs are in constant need of being perceived as desirable to recruiters, and often act as recruiters themselves as members of nomination committees and part of parties' internal primaries for ballot positions.Violence hence increases informal barriers to the recruitment of women by amplifying the value of stereotypically male characteristics which women are not automatically assumed to possess.

Substantive Representation
Feminist policy.Virtually no respondents report changing policy (e.g., voting differently) in any form as a consequence of violence in either the survey or interview data, either on feminist issues or other issues (tables A3 and A8).The most commonly reported consequence of violence, however, is to avoid debating certain topics (table A3).Of politicians in general, 13% report being silenced by violence; 26% of politicians with personal experiences of violence have been silenced.Across political parties and hierarchies, the majority of both women and men interviewees mentioned limitations to taking part in public debate, and they largely mentioned the same topics as the ones they actively avoid.After migration, feminist issues such as men's violence against women or gender equality were the most recurrent topics mentioned as attracting extensive violence when brought up in public debates.Such issues were mentioned by as many men as women.One woman described that the issues that have generated threats against her are violence against women, child abuse, and rape (Woman-31).Women's as well as men's substantive representation of women and feminism is hence delimited by violence.Policy areas that are important to women risk being forgotten as politicians refrain from debating them.As a consequence, these policy areas might receive lower priority, less demand for action from the public, and less development.Intended beneficiaries of a policy such as support to battered women might receive less information about the policy's existence; which threatens the policy's success.
Politicians silenced in debates.Several interviewees described drawing back from policy debates, such as the following quote: My own experience from getting so many threats and serious death threats in relation to some debates naturally means that I'm partly reluctant to write or propose things in certain areas.Another woman described avoiding speaking publicly about an issue she works on in her parliamentary committee because of violent activists involved in the issue (Woman-18).Yet another woman described "subduing" herself for a while after having been severely attacked in relation to a specific debate: I didn't participate in public debates for example, where I would be present physically.But it also meant that I didn't take the debate on my own Facebook page.I posted something and there was a lot of hate and I actually didn't reply.In this way, representatives are delimited in voicing perspectives their voters have elected them to promote.
The survey data demonstrate that women silence themselves substantially more than men.Among MPs, women are 12 percentage points more likely than men to be silenced by violence (refer to table 2 Model 7), and women municipal politicians are 3 percentage-points more likely to suffer this consequence than men counterparts (Model 5).These gender differences are even larger when focusing specifically on politicians with personal experiences of violence, but personal violence exposure does not account entirely for this pattern (table A9).The process of being deterred from debate participation by witnessing attacks on others was described by interviewees: It's very clear that many primarily female colleagues get very battered.And then I think that I can't do that.So then I keep silent … .In particular it's enormously much more common that they have to tell women that they're ugly and such, which is extremely irrelevant in political debate.This shows how violence can silence women politicians even without specifically targeting those who otherwise might have participated in debates.By targeting a few, a larger pool of political voices is silenced.As this quote illustrates, both the gendered frequency and form of attacks silence other women's voices.Even though the topics they refrain from debating may not in themselves be gendered, this implies that women representatives enjoy less policy influence than men.That the relative silencing of women politicians compared to men is the most pronounced at the MP-level is deeply troublesome.Participating in debates-including in media outlets, political events, and on social media-is particularly central to the roles of MPs.
The issue mentioned most often by both women and men as attracting violence is migration.Similar to feminist issues, migration policy relates to equality and to challenging hegemonic male substantive interests.Even politicians who have not personally been exposed to violence avoid discussing contentious issues such as migration publicly for fear of hostility (e.g.Man-8).When asked if there are particular topics that violence makes them refrain from debating, more women than men mention migration, which implies that the topic risks becoming an increasingly male-dominated policy area.Respondents maintain that they stand firm in their policy stances, but avoid debating these in public: When it comes to immigration issues or xenophobia I'm a bit more watchful with the statements I make than I used to … .It can completely drain you of energy … .When you've been exposed to it a number of times you feel that it's not worth it.Unfortunately.At the same time it tells you that you should do that even more because you really need to be heard on those issues.But it takes such an incredible amount of energy.So I don't back down, I don't change what I stand for or anything like that, but maybe I don't stand as much on the barricades in those issues all the time as I would otherwise.Particularly on the migration issue, I ask myself "do I have the energy to write this?" … .I'll write, and then what happens?Well then I have five hours discussion afterwards.I can't take it today.So the debate climate makes me self-censor.In this way violence unjustifiably restricts policy discourse and particularly silences women on this topic.
Time and energy for policymaking.Several respondentswomen in particular-talked about how violence against them takes up time and energy that they would rather devote to the core of their roles as representatives.Without being prompted, almost all of the women and a few of the men described being affected in this way.For example, one woman described how much energy it took having to repeatedly block a person who would turn up under new account names, incessantly commenting in very rude words on everything she posted online (Woman-12).Another woman explained that in her work on the education committee, after a while she did not get to focus on the core issues such as improving education: "Instead it became all about handling aggressive people and disappointed people and so on, which took a terrible amount of time and energy from you" (Woman-43).One of the men recounted a situation when people wrote hundreds of online messages about attacking him and taking his child.In addition to this being a highly distressing experience, he also commented that I spent two days, in the middle of the election campaign, on sanitizing the effects of this.When I was one of the most visible national politicians I had to devote my time to this crap.In the middle of the election campaign.That was very unpleasant.Other respondents described difficulties concentrating and exhaustion in relation to threats.One woman kept going over scenarios in her head that she had discussed with the security services, regarding a person who was stalking her: It takes energy to think about that all of a sudden one day a person can be standing there, and then I should do the following.It takes your energy from other things.So that's a little exhausting.In a similar vein, one woman noted that "every sound I hear when I'm at home I'm thinking 'This is them coming now', and that's been really tough in periods" (Woman-12).A man who received threatening phone calls explained how he felt very worried and "it was a little difficult sleeping and such because you get worked up" (Man-32).Sleeping difficulties and not feeling safe at home were mentioned by several politicians, some of whom have had aggressors come to their homes and others who have not.The following quote demonstrates that political violence can have a tangible impact on politicians without personal violence exposure: That winter when I became active they set fire to a home for refugees.In principle people could have been inside and died.It was in the middle of the night.It touched me and my friend very deeply.We had just entered politics and we felt that gosh, there are so many lunatics!And we're on the other side from them that we consider lunatics.Well, it's mainly the xenophobic ones at that side that we see using violence and threats and setting fires and we are the explicit enemies.And then it felt risky to put oneself in that line with your name in public.….At the time after the fire I used to lay awake at night a lot and listen and had a really hard time falling asleep."Is someone coming now and how do I get the kids out, and do we have our ladders and fire alarms in order?" (Woman-44).
March 2024 | Vol. 22/No. 1 91 Without being prompted, a few men, but far more women, mentioned expecting violence and devising strategies as part of preparing for how and when to raise issues in public debate.The following quote illustrates how routinized harassment is for some politicians: I know that if I do this, it will show in the emails, and then one can feel a knot in the stomach.I remember once in the Chamber, I whispered to [NN] who was sitting next to me "Okay, I'm going up and I know what this entails" and it was like that, we both knew it … .I do it anyway, but I know, and I can also feel it, there's an apprehension about what will happen.Necessitating politicians to devote their time and energy to handling and worrying about violence indicates that they thus have less time and energy for policymaking.One woman remembered a period when she felt scared every time the phone beeped or there was a noise outside.It was tough to be in a situation where you are "frozen, scared for your life", and at the same time "you're supposed to manage your duties without anything showing" (Woman-29).This constitutes a tangible infringement on women politicians' substantive policy work, which seems to be driven by the fact that more women than men experience violence.As a result, women are compelled to re-direct their time, energy, and mental availability from policymaking tasks to violence mitigation, which ultimately threatens their performance as representatives.

Symbolic Representation
Visibility.It is beyond the scope of this study to investigate how citizens' attitudes to gender and politics are affected by violence against politicians.Among the consequences of violence experienced by politicians, however, the main harm to women's symbolic representation is that violence coerces them to decrease their visibility to the citizenry.Without being prompted, almost half of the women, but no men, mentioned decreasing or selecting out of visibility in general due to violence.One woman said that she would decline a role on the municipal assembly, because it is broadcasted online and that would increase her visibility to potential perpetrators (Woman-44).A couple of women spontaneously mentioned that they have opted out of social media entirely, or drastically decreased their social media presence.One woman described that she has left Twitter because "it takes too much mentally" (Woman-22).Another commented that others have told her what people have written about her on social media and she prefers not knowing (Woman-11).At the same time, many respondents talked about how important social media is in their work to present themselves to voters, connect with constituents, become known to a large audience, be able to provide lengthy and nuanced responses to critiques against them, and get their side of a story out.Violence hence delimits women's access to tools that are highly valued by other politicians.Some women, but no men, spontaneously mentioned pressures to adjust their campaign activities to risks of violence.One woman (Woman-39) described a campaigning situation where a few men waited around until there were no men in her party's canvassing booth and then approached the two women there very aggressively.She said that it was the worst thing she ever experienced, and they had to call on people from other parties' canvassing booths to come and help get rid of the aggressors.In the aftermath, her party adopted a policy that women should not be "alone without men" in the canvassing booth, as they perceived it as evident that the aggressors had targeted the women.One man recounted intervening in a similar situation when an older man was harassing a female canvasser, commenting that these things happen more often to younger and female canvassers (Man-36).Another woman described that at one point during an election campaign with repeated threats against her, "finally, the security department thought that I should stay at home instead" (Woman-23).The specific targeting of women, online and offline, seems to be an important factor that disrupts women representatives' opportunities for visibility.
Women's visibility as women is furthermore prevented by gendered forms of violence.One woman perceived gendered and sexualizing comments about her as worse than the extensive threats she had received: That's what wears me down the most, and that's what's changed how I dress.I never wear dresses if I'm going to be in the Chamber or in situations where I think that it increases, it triggers their reactions, the female body.This quote exemplifies how women in political institutions can be seen as "bodies out of place" (Puwar 2004).Violence reduces the possibility of women representatives functioning as symbols of women belonging in politics.Notions about women's unsuitability for politics are reinforced by women understating their visible femaleness in order to avoid violence.
Distancing from constituents.Another cost to symbolic representation is that some representatives decrease their availability to and opportunities for direct interactions with constituents.Women's accounts of such adaptations to violence were largely similar to men's, suggesting that this cost is not gendered.Several respondents mentioned that previous harassment has made them stop answering their phones and actively limiting their availability to constituents (Man-2, Man-37, Woman-1, Woman-4, Woman-12, Woman-34).Two mentioned feeling uneasy about crowds of people such as constituency events.Hence, violence restricts opportunities for constituents to interact with their representatives, which can be important for fostering affinity with politics among marginalized groups.Nevertheless, such adaptations to violence did not appear to be widespread and no clear gender patterns emerged.

Summarizing the Gendered Costs of Violence against Politicians
Violence against Swedish politicians demonstrably has gendered costs for all aspects of representation.Descriptive gender diversity in politics seems to be affected to some degree by depressing women's political ambition, since more women than men consider leaving politics.The impact of political violence on the demand side of political recruitment is potentially even more wide-raging.Violence foments male-coded characteristics as ideal for candidates, and strengthens the cognitive association between maleness and ideal politicians.These ideals increase the male privilege of being viewed as a natural inhabitant of political institutions, and the higher efforts required from women to prove their suitability for political office.Violence does not seem to drive women out of specific policy domains, however.
For substantive representation, I find that violence imposes barriers to women's policy influence.For women politicians more than men, handling, worrying about, and taking measures against violence is a significant portion of their itineraries.Rather than giving any kind of pay-offs, this task claims energy and time from representatives' policymaking tasks.Women more than men also actively select out of policy debates in order to avoid violence.By hindering women representatives' debate activities more than men's, violence systematically excludes women from opinion-formation processes.Both women and men, furthermore, experience specific risks related to raising feminist perspectives.Violence hence constitutes an obstacle to women's policy influence at large, as well as to both women's and men's substantive representation of women's rights and feminism.
A few women and men have decreased their direct interactions with constituents, but there does not appear to be a significant damage to women's symbolic representation in this regard.Rather, women's symbolic representation is hindered substantially by violence making visibility costlier for women than men, and women, as a consequence, selecting out of public visibility.Gendered attacks lead women politicians to decrease their presence on social media, to decline visible roles and activities, to adapt their campaigning activities, and to tone down their visible femaleness.Women representatives' possibility to function as visible cues of women's political inclusion is hence diminished.

Conclusion
Previous research has analysed whether women face more and specific forms of violence compared to men in politics, but has paid marginal attention to the consequences for political representation.This article expands knowledge on the gendered nature of violence in politics by specifically directing attention to the gendered costs of violence for representation.Presenting a theoretical framework for costs to descriptive, substantive, and symbolic representation marks the article's first contribution.With minor adaptations, (e.g., relating to the electoral system) the framework may be applied in any context.Second, the article provides the first empirical investigation of gendered costs for the three dimensions of representation.Using unique interview and survey material on Swedish politicians, it demonstrates with precision what these costs consist of in a case characterized by a comparatively high level of gender equality and a comparatively low degree of social cleavages.Investigating various costs to the three dimensions of representation in the same study provides a comprehensive understanding for how violence against politicians threatens the representative democratic system.Applying the framework empirically also illuminates several previously overlooked consequences of violence against politicians.
One main gendered representational cost of violence in the Swedish case is the enforcement of male-coded politician ideals.This increases the dissonance between stereotypes about women and ideal political candidates, implying that women risk being seen as less suitable for politics than men.Future research should investigate how the selectorate perceive violence-exposed candidates, and how these perceptions apply to women and men candidates.In addition, violence may pose yet another barrier to women's descriptive representation since a strengthened association between politics and conflict has been found to depress women's political ambition (Schneider et al. 2016).There are also other important policy implications of stereotypically male traits such as risk acceptance emerging as highly valorized against the backdrop of violence against politicians.Risk acceptance is not unequivocally a desirable characteristic of policymakers, as it might imply accepting high risks against the citizenry.
I furthermore find that violence limits both women's and men's capacity to substantively represent feminist perspectives.This sheds new light on the question of under what conditions men substantively represent women (Bergqvist, Bjarnegård, and Zetterberg 2018;Höhmann and Nugent 2022).To make it possible to challenge hegemonic male-substantive dominance, political parties may need to provide specific support to representatives who promote issues such as gender equality and migration.Furthermore, policy attention is needed to women's withdrawal from issues such as migration, which risks its democratic legitimacy.Gender-sensitive support might also possibly alleviate the fact that violence makes women withdraw from visibility.The finding in previous research that women in European politics receive less media attention than men (Van der Pas and Aaldering 2020) underscores the critical ramifications of violence 1 Although a few find that they experience similar levels (Bjarnegård, Håkansson, and Zetterberg 2022;Erikson, Håkansson, and Josefsson 2021;Ward and McLoughlin 2020). 2 In addition to descriptive, substantive, and symbolic representation, Pitkin's traditional conceptualization also includes formalistic representation, i.e., rules and procedures for selecting political leaders.As few countries continue to formally exclude women from politics by, e.g., reserving the right to hold elected office for men, this dimension tends to be excluded in contemporary studies of gender and political representation.3 Refer to online appendix 1 regarding data access.4 The outcomes investigated are binary (Yes/No responses to survey questions on whether violence has affected respondents in a certain way).Linear regressions risk incorrect standard errors when applied to binary outcomes.On the other hand, logit estimations, while more compatible with non-continuous dependent variables, present results that are less easily interpretable.5 The survey targets all politicians who were in office at the beginning of the year of the survey (a couple of months prior to survey dissemination).The fact that 84-94 respondents each year report that they have left politics due to violence demonstrates that the survey manages to target respondents who have dropped out.

Table 1
Gendered costs of violence against politicians for women's representation Young is defined as below 35 years, Newcomer as serving one's first term, and Immigrant background as being foreign-born or having foreign-born parents.
You're supposed to be like a statue … .You cannot be fragile as a politician.It's really filthy to be fragile.….If you let it get to you, if you talk about it and don't just throw it in the bin, there's a stigma, absolutely.