This thesis was submitted in fulfillment for the Research Master Social Sciences
Part of the job?
The effect of exposure to the online intimidation of politicians on political ambition
Jade Vrielink, 11268891 University of Amsterdam Graduate School of Social Sciences
Thesis supervisor: Dr. Daphne van der Pas Second reader: Dr. Eelco Harteveld
Word count: 9274
Date: 17th of June 2022, Amsterdam
It is increasingly common for politicians to experience online harassment. Operating in the public sphere is especially trying for women politicians who often pay a higher price than men for this. This pre- registered project examines whether exposure to the online intimidation of politicians weakens the interest of the general public, women especially, in pursuing political office (pre-registration link https://osf.io/8yx9v). It is set in the UK where the prevalence of online abuse is well-documented relative to other countries. The participants in the survey experiment were randomly assigned to receive one of three manipulations, two treatments versus one control. The experimental material consists of a fictional non-partisan politician Twitter profile, a Twitter post and five comments. The comments were varied across conditions to contain neutral, abusive or sexist language addressed to the politician.
Drawing from the existing literature on political ambition, violence against women in politics and gender stereotypes it is hypothesized that seeing intimidating Twitter comments addressed to politicians leads to less political ambition. It is further expected that participants are affected more if their gender identity aligns with the gender of the targeted politician. Moreover, women are hypothesized to be affected more by the sexist comments than by the non-gendered abusive comments. This research is instrumental in determining to what extent online intimidation influences the desirability of political office in general.
Additionally, it contributes to efforts to gain more knowledge on the effect of violence against women in politics on equal representation. This project also potentially contributes to knowledge on the mechanisms through which the current well-established gender gap in political ambition is reproduced.
Key words: Political ambition, Gender, Online Intimidation, Twitter, Violence against Women in Politics, VAWIP, Survey Experiment, United Kingdom
There is a growing concern among political parties and scholars that the rampant online intimidation against women politicians will scare off potential candidates who do not relish the prospect of having to put up with hate and aggression online (Saris & Van de Ven 2021; Harmer & Southern 2021). The proposed research will study to what extent online abuse against politicians frustrates the political ambition of bystanders. The consequences of social media as political communication tools are paradoxical for women politicians. On the one hand social media, such as Twitter, hold the promise to benefit women politicians. Twitter creates the opportunity to get a message across without the interference and possible bias of traditional media. An example of this bias is that women politicians generally receive more coverage based on their appearance and personal life than their counterparts (Van der Pas & Aaldering 2020). The flipside of the coin is that women politicians are also disproportionally affected by the unfiltered character of social media. In the United Kingdom, the case study of this project, women MPs are increasingly more often exposed to online abuse (Collignon et al.
2022) . The urgency of the problem is apparent in recent attempts at mapping the scale of the issue. In 2019 45 per cent of UK candidates expressed having suffered abuse on social media versus 29 per cent in 2017 (ibid). Data that is not based on self-reporting show even higher numbers of abuse, namely 62%
(Ward & McLoughlin 2020). Women reported higher levels of abuse than men (Collignon et al. 2022).
While many politicians are harassed online, women politicians are often targeted based on their gender identity (Harmer and Southern 2021; Ward & McLoughlin 2020).
Little is known thus far about the effect of this phenomenon on the general population. Will less people have the desire to enter the political arena when they know that online abuse is part of the job? Are women disproportionally affected by the intimidation of women politicians? For example, sexist media coverage has been shown to negatively affect the amount of women who are willing to enter politics (Haraldsson & Wägnerud 2019). It is important to gain more knowledge on this topic, because of the possible consequences for (equal) representation if it frustrates people’s political ambition. This project also contributes to knowledge on the mechanisms through which the current well-documented gender gap in political ambition is reproduced.
The prevalence of online abuse is well-documented in the UK relative to other countries and the phenomenon has been on the political and societal agenda for some time. MPs from both sides of the political spectrum have spoken out quite strongly against the harassment they receive while fulfilling office and the effect this has on themselves and their loved ones (BBC 2021). This is a critical case since politicians in the UK political system MPs are particularly vulnerable compared to other countries due to their close contact with constituents during surgeries and work visits in combination with a polarized two-party system. Moreover, 90 per cent of UK MPs have a twitter account and thus sustaining an online presence is part of the job (Ward & McLoughlin 2020). Therefore, UK bystanders are the most likely to be deterred compared to other countries.
To determine whether witnessing online abuse has any impact on the political ambition of UK citizens a survey experiment is employed. Participants in the treatment group are exposed to fictional politicians receiving abusive or sexist comments from seemingly random Twitter users. After which, participants are asked about their interest in ever pursuing political office in the future. The research design, hypotheses, sampling plan and analysis plan, were pre-registered before the data collection took place (https://osf.io/8yx9v).
The direct and indirect consequences of the online intimidation of politicians
A growing amount of literature is concerned with mapping the extent of online abuse (Collignon &
Rüdig 2020; Collignon et al. 2022; Ward & McLoughlin 2020). Considerably less empirical literature focuses on the direct consequences of the problem on politicians and no studies thus far have attempted to uncover the indirect effect on the general population. For example, online harassment is feared to restrict MPs ability to speak and act freely, with due consequences for the quality of representation.
Interviews with EU politicians validate this concern. A substantial portion of interviewees who have experienced online harassment report curtailing their (online) behavior in order not to provoke (IPU 2018). For example, by self-censorship or by closing accounts on some platforms (idem: 10-11). This is worrying, because politicians who are less visible online might have trouble campaigning as a candidate and might eventually even be less likely to be elected. Twitter is mainly used as a broadcasting platform by political candidates, but it has also been found that Twitter use in general and the use of a more interactive style in particular is connected to more electoral support (Baxter et al. 2016; Graham et al 2013; Kruikemeier 2014). Apart from electoral costs there are also personal costs. Many UK politicians have stated that it can be quite taxing for themselves and those around them (BBC 2021). In some cases online harassment plays a role in the decision to quit politics all together. Four out of nineteen departing UK MPs in 2019 noted the intimidation they received online as one of the reasons they decided to leave office (Collignon 2022 et al.: 2).
This paper concentrates on a wide scope. Rather than focusing on politicians, political candidates or potential candidates, the focus of this research is the general UK population. The rationale behind this broad population is that there is little knowledge available on the barriers that ordinary people perceive with regards to fulfilling office, while at the same time it is found that parliament should be a better reflection of society (Hakhverdian & Schakel 2017). If online harassment is increasingly becoming “part of the job” fulfilling political office might become considerably less desirable for anyone.
Political ambition: static or dynamic?
Political ambition is defined as the interest an individual has in pursuing a political career. This is sometimes referred to as nascent political ambition since it is purely a general desire and it does not concern concrete steps towards this goal (Lawless and Fox 2010a). The level of reported political ambition of the general population is well-documented to be affected by relatively static factors. A study of the British population has found that there is a gender gap, a social class gap, an education gap, a north–south divide, and a personality gap (Allen & Cutts 2018). While this might give the impression that not much can boost or depress political ambition, as it is dependent on societal inequalities, a longitudinal study shows that political ambition is a dynamic concept which increases and decreases as political and personal circumstances change (Lawless and Fox 2011). Several scholars have identified volatile factors that can affect political ambition. For example, receiving encouragement from others or seeing more women of the same party run for office can increase the desire to seek office (Crowder- Meyer 2020; Costa & Wallace 2021). Perceiving high personal costs in running for office or stereotype threat on the other hand can diminish political ambition (Fulton et al. 2006; Pruysers & Blais 2017).
Defining (gendered) online abuse
Comments that are defined as abusive comments are messages directed at a specific person with the intent to cause harm or distress. This can include casual use of slurs of derogatory stereotypes and profane language, so long as it is specific and it is personal attack at ‘you’ (Bartlett 2014: 24). Examples of abusive comments used in the experimental material are “You’re literally presiding over the erosion of the UK with your thumb jammed up your arse, you utter fuckwit ” and “You are a disingenuous tw@t, you do not give a sh1t. Piss off and let true patriots run the country again”. A distinction is made between general abusive comments and sexist comments in order to identify whether there is a difference in impact for bystanders. Gendered harassment includes comments that criticize, attack, marginalize, stereotype, or threaten a person based on attributes of gender or sexuality (Edström, 2016). Tweets that contain sexism use women’s gender identity to position them as poor representatives. This can include both hostile and benevolent sexism. Both forms of sexism seek to justify male power and traditional gender roles. Hostile sexism is characterized by the use of overtly misogynistic and derogatory language (“Bitch” and “Milf”,) while benevolent sexism consists of more subtle use of gender stereotypical attitudes (“Who is looking after your poor child?!???” and “She cares more about her eyebrows than people in need” and) (Glick and Fiske, 1997).
Study 1: The role of gender and feeling at risk
It should be made clear that gender is not the only determining factor in the odds of being targeted. For example, levels of abuse are found to differ between parties, across elections and between levels of name recognition (Collignon et al. 2022; Ward & McLoughlin 2020). The political context has an impact on
online harassment. Violence against politicians in a more general sense is often used as a tool to influence policy or to hinder politicians from doing their job effectively (Daniele and Dipoppa 2017;
Dal Bó et al. 2006). Additionally, men and women should not be thought of as homogenous groups, or binary groups for that matter. Intersectionality plays a large role in the likelihood of receiving abuse.
For example, UK MPs who identify as being Black, Asian or from an ethnic minority are targeted more often (63% of BAME women and 38% of BAME men) than their white counterparts (45% of white women and 34% of white men) (Collignon et al 2022). Those MPs who speak out against discrimination or inequality in society are generally more likely to receive online harassment (Sobieraj 2020). So while in this paper the focus is on gendered impact and gendered forms of harassment it should be noted that this does not paint a full picture of the effects of the phenomenon.
The role of gender has been central in the literature on this issue. Women are found to be at higher risk for receiving online abuse as the prevalence of violence against women in society in general translates to women in the public eye (Krook and Restrepo Sanin 2020). 54% of women MPs in the UK and 40%
of men report receiving abuse on social media (Collignon et al. 2022). Observational studies reach less clear conclusions on the role of gender on the extent of abuse (Southern & Harmer 2021: 266; Ward and McLoughlin 2020). However, it is clear that women UK MPs are more likely to receive hate speech than men (Ward and McLoughlin 2020). This has the potential as being experienced as more threatening since it is directed at the MP specifically. Women and men are generally found to experience abuse differently. 34% of women candidates in the UK report feeling very fearful due to online harassment while campaigning, compared to 14% of men candidates (Collignon et al. 2022: 6). Qualitative research indicates that women MPs experience the environment in which they have to do their job as more hostile (Wagner 2020).
One way through which online harassment, targeted at women, can reproduce the gender gap in political ambition is when it becomes an extra barrier to pursue office. A political career requires women to be public figures, online and offline, while the online environment can be experienced as particularly unsafe. This theory is substantiated by research into women’s experiences of safety in the digital sphere in general. The online disinhibition effect posits that harassment is so rampant online, because people feel more anonymous (Fox et al. 2015). Women curtail their online behavior in the same way that they adapt their behavior when they are harassed on the street, by silencing themselves to avoid escalating the violence and abuse (Megarry 2014: 52). It was found that 42% of women who have received hateful comments directed toward group characteristics say they will become more cautious online (Nadim &
Fladmoe 2021). Preemptively constraining behavior and the fear of becoming a target is therefore not limited to politicians. The vulnerability associated with being a public person might lead bystanders to preemptively express less political ambition (Bardall et al. 2020).
This study attempts to mimic the real-life scenario of bystanders being confronted with the online abuse of politicians on social media. While exposure to women politicians is often thought to boost political ambition among women, this is conditional on how the potential career is framed (Schneider et al. 2016;
Campbell and Wolbrecht 2006). Politicians can also represent a tale of caution instead of success. For example, when women are exposed to politicians who share realistic experiences such as difficulty combining a political career with caring for small children they do not report a boost in political ambition (Foos and Gilardi 2019). The association of politics with conflict and competition might also have an adverse effect on women’s political ambition (Schneider at al. 2016). It is found that women report higher levels of conflict avoidance than men (Wolak 2022). This originates in gendered expectations for men and women. Traditional gender roles encourage women to avoid conflict and incongruity with these roles is likely to be met with prejudice (Schneider et al. 2022; Everitt et al. 2016). In contrast men tend to be drawn to politics precisely because of the conflict and disagreement (Wolak 2022).
So far it has been established that male bystanders are less frequent targets (because of their gender) and less fearful of online abuse compared to their female counterparts. How does seeing the online harassment that politicians receive change male bystanders perception of political careers? There are several reasons to expect that the negative effect of exposure to online harassment will also be found in men bystanders even though they might not be the ones most at risk. First, political ambition is generally higher in men (Lawless and Fox 2014). A negative effect is more likely to be found when political ambition is high then when it is already quite low (Lawless and Fox 2014). Second, exposing participants to negativity in politics is found to decrease political ambition regardless of which group the information is targeted at (Pruysers & Blais 2017: 242). Simply the negative association of incivility and harassment with politics might deter bystanders from wanting to pursue a political career.
This results in the following hypotheses:
H1a Seeing abusive Twitter comments addressed to politicians leads to less political ambition
H1b Seeing abusive Twitter comments addressed to politicians leads to less political ambition among women bystanders
H1c Seeing abusive Twitter comments addressed to politicians leads to less political ambition among men bystander
It is expected that people who feel more at risk to receive online harassment are also more likely to report less political ambition. The fear of being targeted then functions as a mediating variable between the exposure to abusive comments and political ambition.
H2 (partial mediation)The negative effect of exposure to online abusive comments on political ambition is partially mediated by the fear of being targeted
In addition, it is hypothesized that bystanders will be affected more when they are exposed to abusive comments directed to politicians from their own identity group since they are more likely to identify with the targeted politician.
H3a The negative effect of seeing abusive comments is greater when bystanders share an identity with
the subject of the abuse.
H3b The negative effect of seeing abusive comments is greater for women bystanders if the subject of the abuse is a women rather than a man.
H3c The negative effect of seeing abusive comments is greater for men bystanders if the subject of the abuse is a man rather than a woman
Study 2: Sexism and internal political efficacy
Gender roles and misogynous language are often visible in the content of uncivil tweets that UK MPs receive (Bardall et al. 2020: 2005-2008; Ward and McLoughlin 2020). In sexist tweets women MPs are implied to be unsuited for office or unwelcome in public life. For example, by referring to the domestic duties of women or their sensitive nature. Tweets that MPs receive also often refer to their appearance and private life instead of the content of their office (Harmer & Southern 2021). Moreover, these tweets tend to dismiss the existence of discrimination against women or include victim-blaming (Harmer &
Southern 2021: 2008; Sobieraj 2020).
The literature suggests that exposure to sexist comments could diminish political ambition among women bystanders. Priming participants with the message that women are less capable and less likely to be successful in politics is found to decrease women’s political ambition (Pruysers & Blais 2017).
Women in the treatment condition were half as likely to place a career in politics in their top-five preferred careers as their counterparts in the control group. They were also less likely to say that running for office had crossed their mind (ibid). Additionally it has been found that hostile sexism negatively affects the performance based self-esteem of women bystanders (Bradley-Geist et al. 2015: 39).
Participants with lower performance based self-esteem also expressed lower career aspirations (ibid). It is hypothesized that sexist comments function as information cues reinforcing negative conceptions about women’s competence in politics.
H4a Seeing sexist Twitter comments addressed to women politicians leads to less political ambition H4b Seeing sexist Twitter comments addressed to women politicians leads to less political ambition among women bystanders
While gendered harassment is something that affects mainly women politicians, the effect of sexist comments on political ambition is likely not limited to women bystanders. It was found that observing hostility towards women in the workplace affects all employees similarly regardless of gender (Miner- Rubino & Cortina 2007). One line of reasoning for this is that while men are not targeted themselves it could lead them to perceive the organization in general as more unjust and unfair (ibid). Political office is quite an open and public work environment which means that hostility can be observed by the general public. I argue that cues indicating a hostile work environment, exposure to sexist comments in this study, inform bystanders and consequently lower their desire to pursue a career in that field.
H4c Seeing sexist Twitter comments addressed to women politicians leads to less political ambition among men bystanders
However, compared to men the impact of online abuse is likely greater for women since they are found to be more responsive to the evaluations of others in their self-evaluation than men (Roberts & Nolen- Hoeksema 1994). Additionally, while the negatively and unjust nature of the comments is non- discriminatory, the content of the negative evaluation in sexist comments are directed towards women’s performance specifically.
H5 The negative effect of seeing sexist comments addressed to a women politician is larger for women bystanders than for men bystanders
It is further hypothesized that women in the treatment condition will express less internal political efficacy and subsequently less political ambition than women in the control condition. Women bystanders thus lose confidence in their own abilities and perceive political office as unattainable as a response to sexist comments consequently lowering their political ambition. There is a positive relationship between internal efficacy and political ambition. A longitudinal study into the fluctuation of political ambition in potential candidates has found that internal efficacy explains much of the variation in political motivation (Lawless and Fox 2005). Related to the political ambition gap there is also a gap in internal efficacy. Women potential candidates in the US were twice as likely to rate themselves not at all qualified to run for office compared to their men counterparts (Lawless and Fox 2010b). Their self-perceived qualifications are also more likely to influence future considerations of running for office in women than in men (idem: 121). People who believe that they can participate effectively in politics are more likely to do so. Efficacy can be primed through positive feedback leading to higher levels of self-reported political interest (Preece 2016). In this study participants are exposed to negative priming with sexist comments which will effectively function as negative feedback leading to lower levels of efficacy.
H6 (partial mediation):The negative effect of exposure to online sexist comments on political ambition is partially mediated by internal efficacy in women bystanders
Figure 1 shows a visualization of all the mechanisms discussed in the theoretical framework so far.
Figure 1 Visualization of the hypotheses. The dashed line indicates that the effect only applies to women.
The challenge effect
A competing theory is that exposure to gender discrimination, personally or through the media, actually has the effect of mobilizing women. Rather than diminished ambition this would have the effect a positive effect on women’s reported political ambition (Moore 2005; Haraldsson 2021). The negative effect on political ambition is then only seen in men respondents. This is supported by research that shows that women are positively challenged to perform well rather than disengaging due to stereotype threat (Derks et al. 2011). However, since online abuse represents a greater personal threat for (potential) candidates this effect is not expected. Moreover, Haraldsson (2021) only finds the challenge effect for expressive political ambition whereas there is no gender effect on nascent political ambition, which is the focus of this study.
Comparing study 1 and study 2: the effects of sexist versus abusive comments
To determine whether women’s political ambition is affected more by exposure to sexist comments than by exposure to abusive comments the two stimuli are compared. The reasoning behind this is that sexist comments have a dampening effect on political efficacy as well as increasing women’s perception of being at risk while abusive comments do not play into women’s (lack of) competence specifically. For men bystanders however witnessing sexist comments does not affect their political efficacy or their sense of being at risk, but both abusive as sexist comments affects their perception of political careers negatively.
H7a Women bystanders express less political ambition when seeing sexist comments towards a woman politician than when seeing abusive comments towards a woman politician
H7b Men bystanders express the same amount of political ambition when seeing sexist comments towards a woman politician as when seeing abusive comments towards a woman politician
Research design and implementation
This study employs a between-subjects experiment to test whether online harassment of politicians has an effect on people’s political ambition. An experimental design is suited for this research question since it has the advantage that all other factors except for the treatment can be held constant. This lowers the possibility that observed loss of political ambition is caused by other factors than the assigned condition.
Additionally, to understand whether the type of abuse plays a role in the effect, by creating two treatment conditions. One is categorized as abusive, but it does not contain tropes relating to the politician’s gender. The other is categorizes as sexism as it contains negative tropes relating to women. Study 1 has a factorial design of 2 (abusive / non-abusive tweet) x 2 (gender targeted politician) conditions. Note that the gender of the fictional politician is also randomized for the control condition (non-abusive tweet). Study 2 consists of a basic randomized design with two conditions (sexist / non-sexist tweet).
Politicians’ gender is not randomized in this instance since this might come across as unrealistic to participants. Men politicians are not typically a frequent target on the basis of their gender. So participants will be shown either sexist comments received by a women politician or neutral comments received by a women politician. The female control condition is the same as in study 1. Participants are randomly assigned to either of five treatment conditions (abusive tweet m/v, control tweet m/v or sexism tweet).
The survey has been built in Qualtrics and administered online through Lucid among UK respondents.
Quota sampling was used to ensure that the sample was similar to the general UK population with regards to age and gender. Respondents below the age of 18 were redirected out of the survey, since they are not eligible to stand for office. The online survey experiment was conducted between 3 May and 26 May 2022. Lucid is a marketplace made up of providers who direct respondents to Lucid.
Participants are compensated per completed survey through the third-party provider, usually in cash, gift cards or reward points (Coppock & McClellan 2019). While Lucid respondents are not a probability sample, studies show that demographic and experimental effect size estimates are similar to those of probability samples (Coppock and McClellan 2019). To determine the sample size a power analysis was performed using the software program G*Power. The conclusion of the analysis was to collect a total sample of 3220 respondents distributed unevenly across treatment groups as some groups require more respondents to gain enough power (see appendix A).
The hypotheses, survey design, sampling plan, measures and analysis plan were pre-registered before the start of the data collection, on the 28th of April 2022. The pre-registration link is: https://osf.io/8yx9v.
Pre-registration increases the credibility and replicability of this research by functioning as a safeguard against the selective reporting of results (Logg and Dorison 2021). It is still possible to explore the data, but there is a clear distinction between confirmatory and exploratory analysis. This benefits the general cumulation of knowledge on this topic and researchers building on this work.
The survey experiment: treatment and control conditions
Participants in the survey are first asked some basic demographic question regarding, age, gender, ethnicity followed by questions about their political interest and attitudes. They are then randomly assigned to one of five conditions in which they are prompted to look at the Twitter profile of a politician and subsequently to read five (violent or neutral) comments addressed to a post of this politician. All experimental material is presented in Twitter layout to mimic real world media consumption. Please see appendix B for an overview of the material used in this study. The Twitter profile, the Twitter post and two out of five comments are constant across treatments. Three of the comments that participants are shown are varied across treatments. In the control condition respondents are shown five neutral comments. In the abusive treatment condition respondents are shown two neutral comments and three abusive comments. In the sexist treatment condition respondents are shown two neutral comments and three sexist comments. The comments were shown in randomized order. Several measures were taken to prevent partisan bias. First, copyright free pictures of German politicians are used which the respondents are unlikely to recognize. No additional information was given about their party affiliation (see figure 2).
Figure 2 The fictional politicians "Christine Jones" and “Paul Smith”
Second, the Twitter post entails a message about a work visit to a local school and does not contain references to partisan activities (see figure 3). The name/Twitter handle are anonymized so as to not relate to an existing UK MP. The fictional UK politicians are named “Christine Jones” and “Paul Smith”.
Figure 3 The Twitter post which remains constant in all experimental material
The treatment conditions were pre-tested among a small group of 40 respondents ahead of administering the survey to Lucid respondents. These respondents were gathered among personal contacts. Special
attention was paid to whether respondents found the experimental material realistic and whether respondents did not find the comments too offensive or disturbing. During the pre-test respondents were also asked to rate the comments on agreeableness (1-10 scale) as a manipulation check. Respondents found the treatment conditions considerably more unpleasant (between 1.4 and 3.3) than the control conditions (between 7 and 7.3).
To increase external validity the comments that were used for the treatment conditions are real hate comments received by UK politicians on Twitter in February and March 2022. The Twitter handles and photos of the commenters were anonymized to prevent bias. See figure 4 for an example of a neutral and an abusive comment. See appendix C for the full survey and all experimental material.
Figure 4 A neutral comment (above) and an abusive comment (below)
It is important to note that participants might have already been exposed to online abuse pre-treatment (Gaines et al. 2007: 12-17). This could lead to an underestimation of the effects as is shown in previous studies where more common or subtle forms of sexism show no effect (Bradley-Geist et al. 2015: 39;
Pruysers et al. 2020: 44-46). Therefore, the treatment should be strong enough that participants will pick up on it, while remaining realistic. There are ethical implications to this which I will discuss further on.
I have also included a pre-treatment check to be able to differentiate between respondents who are likely pre-treated and those who are less likely to be pre-treated: ”Do you follow one or more politicians on social media (for example on Twitter, Facebook or Instagram)?” (“Yes, I follow one politician”, “Yes, I follow more than one politician”, “No”, “I don’t use social media”).
Several measures were put in place to increase the quality of the results. First, the survey contains two attention checks regarding the gender and profession of the fictional politician (“What is the profession of the person's Twitter profile you just saw?” (multiple choice: “MP”, “Teacher”, “Journalist”, “I don’t know”) and “Was the person you just saw a man or a woman?”). The survey was terminated for 394 respondents who failed these simple attention checks. Second, to prevent speeding and to ensure that respondents read the stimulus attentively the click through button was delayed, for 6 and 17 seconds respectively, to ensure that respondents remain on the page long enough to have read the material. The
average time that respondents took to fill in the survey is 6,3 minutes. 65 respondent filled in the survey in less than 100 seconds and therefore could not reasonably have read the questions. These respondents were not used in the analyses. There was some additional attrition. There were 405 respondents who did not complete the survey. These respondents did not differ from the rest of the sample with regards to age or ethnic group, but they were generally lower educated, female and less likely to follow a politician on social media.
Importantly there are ethical concerns of showing abusive comments. This is mitigated by a) only showing abusive messages that occur in reality on social media and as such are not extreme enough to be removed by the platforms; b) informing participants beforehand through a consent form that they can be confronted with comments containing abusive language, and exclusively proceeding with their consent (see appendix D for an informed consent form); c) debriefing respondents with a text aiming to reaffirm their political efficacy; d) debriefing respondents with a text to provide information regarding the deception of fictional politicians that occurred in the experiment (see appendix E for the debriefing form);. This study has also been given ethical clearance from the Examinations Board of the Research Master Social Sciences at the University of Amsterdam based on these measures. The ethical clearance was gained before survey distribution took place. The full survey including treatment material, the consent form and debriefing form were all viewed by the Examinations Board and thesis supervisor beforehand.
The main outcome variable is nascent political ambition. This will be measured by asking participants
“In general, how interested are you in ever running as a candidate in an election for public office? (for example as local councillor or as a Member of Parliament)” (on a scale of 0 “extremely uninterested” to 100 “extremely interested”). The skewness of the variable was found to be .95 indicating that the distribution was right-skewed (see appendix F, figure 13). In other words, most people report having little political ambition and a small number of respondents report high levels of political ambition. This was expected and included in the pre-registration. As a solution the variable was recoded into a dummy variable (0 = 0-10 little to no political ambition, 1 = 11-100 some political ambition). This coding was used in all of the analyses, apart from the robustness check for H1.
A more common way to measure political ambition is through the question “if you have never run for office, have you ever thought about running for office?” (Lawless and Fox 2010b). This question is not suited for this study since it provides information on the past political ambition of the participant while the focus of this study is the present. To increase the reliability of the measure and to ensure continuity
with other studies an additional question was added, namely a rank order question: “Please rank the following jobs from 1 to 8 (where 1 represents the job you are most interested in having and 8 represents the job you are least interested in having)” (adapted from Pruysers and Blais 2017). The job ranking question was recoded to categorical variables where respondents who had Member of Parliament or Local Councillor in their top 3 were coded as 1 or 2 and the rest of the respondent as 0.
The two mediating variables are internal political efficacy and perceived risk of harassment. First, internal political efficacy is measured using the four-item index by Niemi et al. (1991). This is found to be a valid and reliable measure of internal efficacy (Morrell 2003). 1. “I consider myself well-qualified to participate in politics”, 2. “I feel that I have a pretty good understanding of the important political issues facing our country”, 3. “I feel that I could do as good a job in public office as most other people”
and 4. “I think that I am better informed about politics and government than most people”. This is measured on a five-point scale from “disagree strongly” to “agree strongly”. The responses on the four items were recoded into an index with a 4-20 scale (Morrell 2005). Second, perceived risk of harassment is measured by one item “To which extent is the possibility of receiving hateful or abusive comments a barrier for you to pursue public office, where 0 means no barrier and 10 means a large barrier”.
In addition, respondents answered some demographic questions to determine the composition of the sample, namely age, gender, education and ethnicity (BAME). As well as factors which could influence the effect of the stimuli, namely political ambition, political interest and left-right placement. Since random assignment is used in this study, which ensures that the exposure to the treatment material is not correlated with any omitted variables, these variables are not used as control variable in the analysis.
The sample consisted of 1652 women and 1559 men. More than 90% of the sample self-identified as white and 8% of the sample identified as black, Asian or from an ethnic minority.
The mean age was 47,9. Half of the sample was between the ages 18 and 47. In terms of ideology, 46%
of the respondents identified as centrist on a left-right placement scale, 28% identified as more right- wing and 26% as more left-wing. 39% of respondents indicated they would vote Labour if elections were held today and 28% of respondents would vote Conservative. To determine how representative the sample is of the UK population a couple of similarly measured demographics were compared with data from the European Social Survey (ESS 2021). The sample used in this study is generally lower educated (ESS mean = 5.52, sample mean = 4.23), but comparable on left-right placement, gender and age. This could affect the result of this study since lower-educated individuals are generally less politically ambitious and would respond less strongly to the stimuli (Allen & Cutts 2018). There was some attrition.
405 respondents did not complete the survey. These respondents did not differ from the rest of the sample with regards to age or ethnic group, but they were generally lower educated, female and less
likely to follow a politician on social media. Further basic descriptive statistics are enclosed in appendix F. The number of subjects in each treatment condition by gender is reported in table 1. The random assignment was performed in Qualtrics. Since there is no reason to assume the random assignment was not successful, no balance test or randomization check is performed. Please find a note on the assumption of unit homogeneity and the need for balance tests in appendix G.
Table 1 Number of respondents in each experimental condition
Control Men Politician
Total N 1651 1559
Figure 3 provides a descriptive overview of the mean reported politician ambition by gender (0 = women and 1 = men). The expected gender gap in political ambition is clearly visible. Male respondents’
mean political ambition is 10 points higher on a 0-100 scale than women’s political ambition. Similarly 27% of women listed one or two political jobs in their top
3 preferred jobs compared to 37%
of men respondents. The similar pattern between the two measures of political ambition increases the validity of the scale measurement.
Contrary to our expectations political ambition is not consistently lower in treatment groups as opposed to control groups for neither men nor women.
Figure 5 Mean of political ambition between groups (0 = women, 1 = men)
The first hypothesis argues that seeing abusive Twitter comments directed to a politician leads to less political ambition. To test this hypothesis a logistic regression with political ambition as binary dependent variable and receiving the treatment as an independent variable is performed. It is found that the effect of receiving the treatment is negative.
Participants in the treatment group had 8% less probability of reporting political ambition above 10%
on a 100% scale. See figure 6 for a visualization of this effect. However, the effect is not significant (ß= -.085 p=.296, OR = .92).
Thus no convincing evidence is found to support this first hypothesis. As a robustness check different coding options for the political ambition dummy are inspected (see appendix H). The option to recode political ambition into a dummy was included in the pre-registration, but the threshold indicating the
absence or presence of political ambition was not specified. The robustness check shows that the cut-off point indicating the presence of absence of political ambition does not determine the outcome of the analysis (lower cut-off point ß=
-.087 p=.291 OR=.92, higher cut-off point (ß= - .082 p=.391 OR=.92). An additional question to measure political ambition in which participants are asked to rank political and non-political jobs was also included in the survey. Performing the
same regression analysis with the job ranking measure yields similar results, which substantiates the reliability of this result (ß= -.115 p=.178 OR=.89). Combining the two-items into a single measure of political ambition variable also does not yield sufficient evidence to support the first hypothesis (ß= - .11 p=.14 OR=.90).
A potential explanation for the lack of significant findings of exposure to abusive comments could be contamination from real-world experience. Violence against politicians could be so normalized and often-discussed that a one-shot treatment will not affect the amount of political ambition that bystanders express. To gain some insight into the extent of pre-treatment a question was added to the survey where respondents can indicate whether they follow politicians on social media. It was found that 26,2% of the sample follows at least one politician on social media. Additionally, to determine whether respondents who follow politicians are also more knowledgeable about the online harassment of politicians an OLS
Figure 7 The effect of following a politician on social media on the perception of what percentage of politicians receive abuse 0 "I don't use social media" 1 "No" 2 "I follow one politician" 3"I follow more than one politician"
Figure 6 Treatment effects of abusive comments on political ambition (predicted)
regression is performed with the estimated percentage of politicians that receive abuse as the dependent variable and following a politician as an independent variable. This analysis is exploratory and was not included in the pre-registration. Following a politician has a significant positive effect on the percentage of politicians that participants estimate receive abuse, although the effect is very modest (ß= 1.49 p=.002). Adding whether participations follow politicians on social media to our initial logistic regression analysis provides evidence for the pre-treatment theory. Participants who do not use social media are negatively affected by the treatment condition in terms of political ambition. They are about 50% less likely to report political ambition in the treatment group (ß= -.72 p=.076 OR = .49) while participants who follow more than one politician are not affected by the treatment (ß= .179 p=.449 OR=1.20). Since real life examples were used for the stimulus it could be that participants did not pick up on the treatment, because of previous exposure through social media use. Lastly, there is a reciprocity problem which cannot be solved by randomization completely. Politically ambitious people are probably more likely to follow politicians which increases the likelihood of previous exposure to the online harassment of politicians. This then decreases the odds of finding a treatment effect.
It is further hypothesized that abusive Twitter comments lead to less political ambition for both female (H1b) and male bystanders (H1c). A split sample logistic regression does not support the hypothesis that female bystanders experience less political ambition after viewing the stimulus (ß= .01 p=.92 OR=1.01).
Contrary to our expectation there only seems to be a negative treatment effect for male bystanders, although still above the threshold of .05 significance (ß= -.19 p=.122 OR=.84). An explanation for this result is that men generally report having more ambition to pursue office while women generally report very low levels of ambition. Therefore, it is more likely that a substantive negative effect is found among male respondents. In other words, when respondents already have near zero desire to pursue office at the beginning of the experiment then no amount of stimulus can bring on a detrimental effect.
The second hypothesis proposes that the treatment effect is partially mediated by the extent to which someone fears becoming a target themselves. First, an OLS regression is performed to determine whether the stimulus affects the perceived risk of being targeted. It was found that participants in the treatment condition were significantly more likely to report that being targeted is a barrier for them in pursuing office (ß= .303 p=.015). Performing a split sample analysis it was found that this effect only holds for women bystanders (ß= .424 p=.012). The split sample analysis was not included in the pre- registration. Figure 8 shows the treatment effects for women. On a scale of 1-10 women in the treatment condition reported experiencing a 0.42 scale points larger barrier. Male bystanders were about as likely to report experiencing a barrier in the treatment condition as in the control condition (ß= .177 p=.327).
Following exposure to online harassment, women are more likely to feel at risk of receiving online harassment than men do.
This is in line with our expectation that women are more likely to be deterred by conflict. However, men were also hypothesized to be deterred by the negative representation of the political job. This does not seem to be the case.
Adding the perceived risk of being targeted to our original logistic regression provided no evidence for the partial mediation hypothesized in H2. There is no significant effect of the reported fear of being targeted on political ambition (ß= -.084 p=.302 OR=.92). Performing an additional (non-preregistered) split sample analysis among female bystanders also does not provide evidence for mediation. It seems that, contrary to what was previously hypothesized, the fear of being targeted does not translate to general loss of political ambition for women (ß= -.009 p=.621 OR=.99). This is an optimistic conclusion since it seems that online harassment does not directly affect the amount of women that is willing to enter politics. Although women indicate that they see online harassment as a barrier after exposure to abusive comments, this barrier does not deter them from expressing interest in fulfilling political office.
It was further hypothesized that when participants are exposed to fictional politicians of their own gender a larger decrease in political ambition would be observed (H3). In the survey experiment bystanders were randomly exposed to either men or women fictional politicians. To test this two logistic regressions were performed with an interaction between the gender of the fictional politician and receiving the treatment in split samples containing only men and only female respondents. It was not found that the gender of the fictional politician in the stimulus had a significant effect on the amount of political ambition reported by bystanders so there is no evidence to support H3 (women ß= -.068 p=.663 OR=.93, men ß= -.031 p=.896 OR=.97) .
So far the effect of the abusive comments stimulus was discussed. Study 2 concerns the sexist comments stimulus. It was hypothesized that these comments lead to less political ambition (H4), and that this is especially salient among female bystanders (H5). A logistic regression with political ambition as the dependent variable and receiving the sexist treatment material as the independent variable shows that viewing sexist comments has a negative effect on political ambition. Participants in the treatment
Figure 8 Treatment effects of abusive comments on the reported barrier of online harassment on pursuing political office (predicted)
condition have a 14% lower probability of reporting political ambition. However, this effect is not significant (ß= -.155 p=.164 OR=.86). Performing the same analysis in a split sample to look at differential results between male and female respondents (H4b and H4c) yielded similar results (women ß= -.073 p=.635 OR=.92, men ß= -.198 p=.227 OR=.82). Figure 9 visualizes these results.
The treatment effect for male respondents approaches significance. It is striking that men’s reported political ambition is more negatively affected by sexist comments than women’s political ambition. Men in the sexist treatment condition have a 18% lower probability of expressing political ambition than men in the control condition, although again this is not statistically significant.
To test H5 a logistic regression is performed with an interaction between the gender of the respondent and the treatment condition. No evidence is found that the effects of sexism on political ambition is dependent on the gender of the respondent (ß= -.198 p=.402).
To determine whether the effect of sexist comments is partially mediated by internal political efficacy (H6) it is first determined if sexist comments have an effect on internal political efficacy for women. An OLS regression is performed with internal political efficacy as independent variable and exposure to sexist comments as an independent variable in a subsample of women respondents. There is no significant effect of sexist comments on internal political efficacy (ß= -.059 p=.836). Including internal political efficacy in our logistic regression with political ambition as the outcome variable shows no evidence for mediation. Although political efficacy has a positive effect on political ambition as expected (ß= .271 p=.000 OR=1.31), there is no treatment effect (ß= -.073 p=.666 OR=.93).
It should be noted that although no evidence is found for a dampening effect of exposure to sexist comments to political ambition, there is also no evidence for the challenge effect which states that respondents would experience a boost in political ambition when confronted with gender discrimination (Moore 2005; Haraldsson 2021).
Study 1 versus study 2
Lastly it was hypothesized that in comparison women would respond more to the sexist comments than the abusive comments (H7a) while for men there would be no difference between the stimuli (H7b). It
Figure 9 Treatment effects of sexist comments on political ambition (predicted)
is found that there is no significant difference between the stimuli for women (ß= -.003 p=.980 OR=.99) nor for men (ß= .055 p=.678 OR=1.06).
The contribution to the literature of this study is twofold. First, it is the first experiment that examines the possible consequences of online harassment on bystanders as opposed to politicians. Second, it is the first large scale quantitative study in which the difference in impact between gendered and non- gendered abuse is explored. A limitation of this experimental set-up is that it was not possible to use an intersectional approach due to the required power when many subgroups are involved. In this case it was not financially feasible to oversample for BAME respondents to such as extent that reliable inferences could be made. Future research should look into the impact of race as it is found to play an important role in the risk of being targeted. Black and Asian women receive 30% more abuse than their white counterparts (Amnesty International UK 2017). It is important to gain knowledge on the effect of racist comments on bystanders for instance or the response of different groups of bystanders to witnessing online harassment of a BAME politician. Another limitation of this study is that it is a one shot treatment and possibly does not represent people’s experiences in real life (Gaines et al. 2007).
Perhaps a better method with regards to external validity is to measure the effect of online harassment on political ambition by exposing participants to several stimuli over longer periods of time. This does bring up ethical implications which need to be considered. Another concern that has already been addressed in the results section is the possibility of contamination from real-world experience. It is not clear to what extent the participants have previous knowledge about the occurrence of online abuse or have been pre-treated through social media use before partaking in the study. Lastly, a disadvantage of using fictional politicians for the experimental material is that respondents are unlikely to identify with or sympathize with them as they do with real-world politicians they know. Therefore, the effect could have been dampened by using fictional politicians. A possible avenue for future research would be to use real-life politicians and to explore whether partisan identification influences the treatment effect.
Figure 10 Comparison of treatment effect of exposure to abusive comments versus sexist comments
Although this study has not found evidence for the hypothesized effects there are some possible avenues for future research. Online harassment does not seem to deter bystanders from political office, but the finding that women respondents experience a barrier due to exposure to abusive comments suggests that women might curtail their political participation in other ways. It is interesting to explore for example whether there are differences between the online campaign strategies of candidates in this regards.
The majority of politicians in the UK have to deal with online harassment. Research has shown that this phenomenon is increasingly more prevalent and that women are often the targeted based on their gender (Collignon et al. 2022). The goal of this study was to understand whether, and how, the exposure to online harassment of politicians leads to less interest in pursuing political office for bystanders. It was further speculated that gendered harassment would have a larger impact on women than men especially through lowering their perception of being able to participate effectively in politics.
Contrary to expectations, this survey experiment has not found convincing evidence that exposure to online harassment in the form of Twitter comments leads to less political ambition. Due to the large sample size which ensures enough power to find a small treatment effect it is likely that this is a reliable finding. Interestingly, there was a significant treatment effect of exposure to abusive comments on experiencing online abuse as a barrier to pursuing office. This treatment effect was only found in a subsample of female respondents. While women in the treatment group feel more at risk than women in the control group, this does not mean that they are also less interested in pursuing office in the future as was expected. Does this mean that concerns about women being deterred from office by online abuse are unfounded? For the general UK population this would seem so. However, it is important to note that the outcome variable in this research is the hypothetical future interest in pursuing office or the general desirability of the job. It is unclear to what extent this finding is generalizable to expressive political ambition. The UK was a most-likely case which means this finding might not be generalizable to countries without the first-past-the-post system. In countries with more proportional systems such as the Netherlands less visibility is generally required from political candidates. Political office might then be perceived as less high risk.
An additional finding is that men actually seem to be affected more by the stimuli due to their generally higher levels of political ambition. As opposed to what was expected the phenomenon might have a very modest role in closing the gender gap, because of the dampening effect on men’s political ambition.
What are the implications for the literature on violence against women in politics? So far the focus has been on politicians and (potential) political candidates. In this study it is determined that the negative implications experienced by politicians are not applicable to the general population. Since the general population likely does not have very strong feelings about pursuing political office they are generally
unaffected by online abuse on this aspect. It would be interesting to explore whether there is an effect on expressive political ambition as opposed to nascent political ambition. Moreover, in line with previous research it is determined that stereotype threat does not lead to a challenge effect with regards to nascent political ambition (Haraldsson 2021).
The main takeaway from this study is quite positive, namely that people generally do not seem to be less interested in a political career due to exposure to online harassment. However, policymakers and party recruiters should be attentive that women might feel more at risk of receiving online abuse than men. This means that they might also need different kinds of encouragement and support while weighing the possibility to pursue office.
Bibliography Primary sources
Allen, P. and Cutts, D. (2018). “An Analysis of Political Ambition in Britain”, The Political Quarterly, 89: 73-81.
Bardall, G., Bjarnegård. E. and Piscopo, J.M. (2020). “How is Political Violence Gendered?
Disentangling Motives, Forms, and Impacts”, Political Studies, 68(4): 916-935.
Bradley-Geist, J.C., Rivera, I. and Gerinegr, S.D. (2015). “The Collateral Damage of Ambient Sexism:
Observing Sexism Impacts Bystander Self-Esteem and Career Aspirations”, Sex Roles, 73: 29- 42.
Collignon, S., Campbell, R. and Rüdig, W. (2022). “The Gendered Harassment of Parliamentary Candidates in the UK”, The Political Quarterly. Available online, through
Haraldsson, A. & Wängnerud, L. (2019). “The effect of media sexism on women’s political ambition:
evidence from a worldwide study”, Feminist Media Studies, 19(4): 525-541.
Harmer, E. and Southern, R. (2021). “Digital microaggressions and everyday othering: an analysis of tweets sent to women members of Parliament in the UK”, Information, Communication &
Society, 24(14): 1998-2015.
Krook, M. & Restrepo Sanín, J. (2020).”The Cost of Doing Politics? Analyzing Violence and Harassment against Female Politicians”, Perspectives on Politics, 18(3): 740-755.
Pruysers, S., & Blais, J. (2017). Why Won't Lola Run? An Experiment Examining Stereotype Threat and Political Ambition. Politics & Gender, 13(2), 232-252.
Schneider, M.C., Bos, A.L. and DiFilippo, M. (2022) “Gender Role Violations and Voter Prejudice:
The Agentic Penalty Faced by Women Politicians”, Journal of Women, Politics & Policy, 43(2): 117-133,
Schneider, M.C., Holman, M.R., Diekman, A.B. and McAndrew, T. (2016) “Power, Conflict, and Community: How Gendered Views of Political Power Influence Women's Political Ambition”, Political Psychology, 37: 515-531
Ward, S. and McLoughlin, L. (2020). “Turds, traitors and tossers: the abuse of UK MPs via Twitter”, The Journal of Legislative Studies, 26(1): 47-73.
Amnesty International. (2017).” Black and Asian Women MPs Abused More Online”,
https://www.amnesty.org.uk/online-violence-women-mps. Accessed March 26, 2021.
Baxter, G., Marcella, R. and O’Shea, M. (2016). “Members of the Scottish Parliament on Twitter:
good constituency men (and women)?”, Aslib Journal of Information Management, 68(4):
BBC (2021). “Our families fear for our personal safety, say MPs”,
https://www.bbc.com/news/ukpolitics-58913530. Assessed on June 7 2022.
Campbell, D.E. and Wolbrecht, C. (2006). “See Jane Run: Women Politicians as Role Models for Adolescents”, The Journal of Politics, 68(2): 233-247.
Collignon, S. and Rüdig, W. (2020). “Harassment and Intimidation of Parliamentary Candidates in the United Kingdom”, The Political Quarterly, 91: 422-429.
Coppock, A., & McClellan, O. A. (2019). Validating the demographic, political, psychological, and experimental results obtained from a new source of online survey respondents. Research &
Politics, 6(1), 2053168018822174.
Costa, M. and Wallace I. (2021). “More Women Candidates: The Effects of Increased Women’s Presence on Political Ambition, Efficacy, and Vote Choice”, American Politics Research, 49(4): 368-380.
Crowder-Meyer, M. (2020). “Baker, Bus Driver, Babysitter, Candidate? Revealing the Gendered Development of Political Ambition Among Ordinary Americans”, Political Behavior, 42:
Dal Bó, E., Dal Bó, P. and Di Tella, R. (2006). “‘Plata o Plomo?’: Bribe and Punishment in a Theory of Political Influence”, American Political Science Review, 100(1): 41–53.
Daniele, G. and Dipoppa, G. (2017). “Mafia, Elections, and Violence against Politicians”, Journal of Public Economics, 154: 10–33.
Derks, B., Ellemers, N., Laar, C. van and Groot, K. de (2011). “Do sexist organizational cultures create the Queen Bee?”, British Journal of Social Psychology, 50: 519-535.
Dong, N. and Maynard, R. (2013). “PowerUp! A tool for calculating minimum detectable effect sizes and minimum required sample sizes for experimental and quasi-experimental design studies”, Journal of Research on Educational Effectiveness, 6: 24–67.
Edström, M (2016) “The trolls disappear in the light: Swedish experience of sexualized hate speech in
the aftermath of Behring Breivik”. International Journal for Crime, Justice, and Social Democracy, 5(2): 96–106.
European Social Survey European Research Infrastructure (ESS ERIC). (2021). ESS9 - integrated file, edition 3.1 [Data set]. Sikt - Norwegian Agency for Shared Services in Education and
Everitt J, Best L.A. and Gaudet D. (2016) “Candidate Gender, Behavioral Style, and Willingness to Vote: Support for Female Candidates Depends on Conformity to Gender Norms”, American Behavioral Scientist, 60(14):1737-1755.
Foos, F. and Gilardi, F. (2019). “Does exposure to gender role models increase women’s
political ambition? A field experiment with politicians”, Journal of Experimental Political Science, 7 (3):157-166.
Fox J, Cruz, C. and Lee, J.Y. (2015) “Perpetuating online sexism offline: Anonymity, interactivity, and the effects of sexist hashtags on social media”, Computers in Human Behavior, 52: 436–
Fox, R.L. and Lawless, J.L. (2005). “To Run or Not to Run for Office: Explaining Nascent Political Ambition”, American Journal of Political Science, 49: 642-659.
Fox, R.L. and Lawless, J.L. (2011). “Gaining and Losing Interest in Running for Public Office: The Concept of Dynamic Political Ambition”, The Journal of Politics, 73(2): 443-462
Fox, J., Cruz, C., & Lee, J. Y. (2015). “Perpetuating online sexism offline: Anonymity, interactivity, and the effects of sexist hashtags on social media”, Computers in Human Behavior, 52: 436–
Fulton F.A., Maestas C.D., Maisel, L.S. and Stone W.J. (2006). “The Sense of a Woman: Gender, Ambition, and the Decision to Run for Congress”, Political Research Quarterly, 59(2): 235- 248.
Gaines, B.J., Kuklinski, J.H., and Quirk, P.J. (2007) ‘The Logic of the Survey Experiment Reexamined’, Political Analysis, 15(1): 1-20.
Glick, P. and Fiske, S.T. (1997) “Hostile and Benevolent Sexism”, Psychology of Women Quarterly, 21: 119-135.
Graham, T., Broersma, M., Hazelhoff, K. & Haar, G. Van ‘t (2013) “Between Broadcasting Political Messages and Interacting with Voters”, Information, Communication & Society, 16(5): 692- 716.
Hakhverdian, A. and Schakel, W. (2017). Nepparlement? Een pleidooi voor politiek hokjesdenken.
Amsterdam University Press: Amsterdam.
Haraldsson, A. (2021). “Media Discrimination and Gender Differences in Political Ambition in a
Laboratory Experiment”, Political Research Quarterly. Available online, through https://doi.org/10.1177/10659129211046890.
Inter-Parliamentary Union (IPU) (2018). “Sexism, Harassment and Violence Against Women Parliamentarians”, https://www.ipu.org/resources/publications/issue-briefs/2018-10/sexism- harassment-and-violence-against-women-in-parliaments-in-europe. Accessed 17 January 2021.
Krook, M. & Restrepo Sanín, J. (2016).”Gender and political violence in latin America. Concepts, debates and solutions”, Politica Y Gobierno, 23(1): 125-157.
Kruikemeier, S. (2014). “How political candidates use Twitter and the impact on votes”, Computers in Human Behavior, 34: 131-139.
Lawless, R.L. and Fox, J.L. (2010a). “If Only They’d Ask: Gender, Recruitment, and Political Ambition”, The Journal of Politics, 72(2): 310-326.
Lawless, R.L. and Fox, J.L. (2010b). It Still Takes A Candidate: Why Women Don’t Run for Office.
Cambridge University Press: New York
Lawless, R.L. and Fox, J.L (2014). “Uncovering the Origins of the Gender Gap in Political Ambition”, American Political Science Association, 108(3): 499-519.
Logg, J.M. and Dorison, C. A. (2021). “Pre-registration: Weighing costs and benefits for researchers, Organizational Behavior and Human Decision Processes”, 167: 18-27.
Megarry,J. (2014). “Online incivility or sexual harassment? Conceptualising women's experiences in the digital age”, Women's Studies International Forum, 47(A):46-55.
Miner-Rubino, K., & Cortina, L. M. (2007). Beyond targets: Consequences of vicarious exposure to misogyny at work. Journal of Applied Psychology, 92(5), 1254–1269.
Moore, R.G. (2005). “Religion, Race, and Gender Differences in Political Ambition”. Politics &
Gender, 1(4): 577-596.
Morrell, M. E. (2003). “Survey and Experimental Evidence for a Reliable and Valid Measure of Internal Political Efficacy”, Public Opinion Quarterly, 67(4): 589-602.