Misogyny, politics and social media – Determinants of hostile engagement of female Parliamentarians on Twitter

Social media has become a major arena for political discourse and competition, but also an increasingly hostile environment. For women in particular, social media offers direct communication with citizens without the intermediary of traditional media. However, they are also more exposed to uncivil responses from their followers. Is this a particular phenomenon for female politicians, and does the form of interaction differ to other MPs? We investigate whether female MPs in Germany and Switzerland are more likely to use uncivil language in their online campaigns and whether they are more likely to be criticized, attacked, or subjected to online incivility by other users. When the media covers female leaders, they often make use of sexist stereotypes: Traditional notions of leadership include exclusively masculine qualities such as competence, assertiveness, and strength, while female leaders are portrayed as empathetic, caring and consensus oriented. Because politics is an area largely dominated by men, male MPs are automatically considered qualified for public office. On the contrary, women are often assigned domestic roles such as wife or mother by the media, which puts their ability to hold public office into question. Female politicians are therefore confronted with a double-bind: they are either portrayed by the media as feminine, likeable, and sympathetic - or as competent. This double-bind also depends on party affiliation: Female stereotypes are often associated with liberal parties, which leads to women being perceived to be more liberal than they actually are. The immediacy of online communication gives female politicians the opportunity to combat such gendered stereotypes by communicating in a more confrontational and aggressive manner. We test the hypothesis that female politicians are more likely to attack their opponents on a personal level using uncivil language. Because stereotypes are also associated with party affiliation, we assume that female politicians of conservative parties are more likely to use uncivil language in their tweets. However, female MPs might also be judged more harshly by voters when using such tactics. Therefore, we test the hypothesis that if female politicians use uncivil language on their Twitter account, they are also more likely to get uncivil replies from their followers, compared to their male colleagues. We developed a codebook to code 12,000 German tweets for personal and uncivil attacks in order to train a BERT transformer model to classify all responses to social media posts of German and Swiss politicians. We find that compared to male politicians, women are less likely to attack their opponents using uncivil language, and that they are also less likely to be attacked by other users but that they are over proportionally targeted if they use uncivil language themselves.