Gender politics online? Political women and social media at election time in the United Kingdom, the United States and New Zealand

Digital optimists have claimed that Internet technology, and especially social media, would revolutionise politics and empower previously marginalised groups. The reality is somewhat different: online like offline politics is the preserve of narrow elite of mostly men, while women are still less likely to discuss politics online. This article offers a comparative study of Theresa May, Hillary Clinton and Jacinda Ardern’s political communications. It investigates whether they used digital technologies during elections to feminise politics and evaluates the extent to which women politicians adopt leadership and communicative styles that challenge masculine norms of political behaviour, whether they prioritise policy areas that are likely to make a difference in women’s lives and if they speak on behalf of other women.

and democracy to thrive and endure' (AmericanRhetoric, 2015). In 2005, Theresa May, the then Member of Parliament (MP) for Maidenhead (Berkshire), was only one of 17 female MPs in the Conservative Party (9%). In order to redress this gender imbalance, May founded Women2Win, a pressure group aiming at making the Conservative Party more representative of modern Britain. Writing for the feminist blog Villainesse in 2015, the then Labour List candidate for Auckland Central, Jacinda Ardern, recalled the dismay she caused at a business meeting when she claimed to be a feminist because she believed that women and men should be paid the same for doing the same jobs (Ardern, 2016).
This article focuses on these three politicians to evaluate whether there is a relationship between their commitment to women's rights, their gender and their political communication. In particular, the article evaluates whether there are lessons to be learnt by studying women representing women in different ideological, political and legislative contexts. It claims that by analysing the ways in which May, Clinton and Ardern use modern forms of political communication to articulate their message, we can begin to uncover the role that gender plays in political communication research.
The question of political representation is a complex one. Women make up a little over 50% of the population in the United Kingdom, the United States and New Zealand (World Bank, 2017). Yet, globally, only 15 Heads of State 1 out of 193 UN Member States (10%) are women (Geiger and Kent, 2017). The picture at national level is equally disheartening: women 2 hold only just over 20% of seats (on average, worldwide; Inter-Parliamentary Union, 2017). This under-representation of women in political institutions has consequences for how democratic a society is seen to be. This is not just a matter of figures and how they limit the ability of women to descriptively represent their constituents. It is also about substantive representation: how politicians' identity influences their political interests and priorities (Paxton and Hughes, 2015).
In policy term, women are seen to have more of a stake (and therefore, be more interested) in so-called women's issues (childcare, reproductive politics and domestic violence), which are catered for through 'soft' policy areas. The naming is not accidental: they are 'soft' because they are considered less important (compared to policing, the economy and foreign policy) and not having the same overall impact on the 'majority' of the population. It is then traditionally left to women to engage with, speak about (Bird, 2005) and vote in favour of these issues (Campbell, 2004), in order to act for women (Childs, 2005;Lovenduski, 2005). Women politicians, who substantively represent other women, are seen as feminising (Lovenduski, 2005) politics: they bring different values, priorities and styles to political discourse that challenge masculine norms of what politics is and how it is to be conducted (Cramer Walsh, 2002). By feminising politics, women politicians redefine what counts as political concern and make (female) voters feel represented.
The perceived lesser importance of women and their interests is reflected in the way they are treated in the media, where men are the experts whereas women are asked how they juggle family and career and are judged on their physical appearance and their age (Ross, 2016). So, women politicians are constrained in their ability to represent women and to speak on their behalf both within the very democratic institutions that are designed to guarantee equal representation and in mainstream media. This article evaluates whether women seeking office turn to digital technologies to bypass traditional gatekeepers and redress the representation gap between what counts as politically meaningful and what is considered a 'minority' concern. It does so by asking two questions, one of the content and one of the tools of communication: do open supporters of women's rights, talk about women and their concerns in their election campaigns and, in so doing, do they make a claim to represent other women? And, are social media platforms, such as Twitter and Facebook, central in defining the conversation, gendering the debate and bypassing mainstream media gatekeepers?

The United Kingdom, New Zealand and the United States: A comparative analysis
This article focuses on three political campaigns: Hillary Clinton's bid for President of the United States (2016), Theresa May's Conservative Party leadership campaign (2016) and Jacinda Ardern's General Election campaign (2017). There were several differences between these politicians and the campaigns they led: from their ideological stance, to their career stage, their age and political experience, and the political, electoral and media systems within which they operated. These differences combined to create complex scenarios. For instance, political affiliation, ideology and election priorities impacted on their understanding of gender equality: May was a Liberal Conservative (Bryson and Heppell, 2010), who believed that her party should be modernised and made more representative of British society (Childs and Webb, 2012). In this sense, she was a committed supporter of the rights of women. Nevertheless, after the referendum, in order to maintain its hold on Westminster, the Conservative Party had to be seen to deliver Brexit. As a consequence, May's political priority had to be readjusted to this goal (Cowley and Kavanagh, 2018).
As members of progressive parties, Ardern and Clinton, were ideologically inherent supporters of gender equality, but voters' priorities and political climate complicated this assumption in their cases too. In 2016, the United States appeared to have reached a post-feminist phase, where economic and racial inequalities were seen as more pressing priorities (Bordo, 2017). In New Zealand, Ardern, was catapulted into the Labour Leadership in the latter stages of the Election Cycle and had no apparent chance of setting a gender (let alone feminist) agenda. Voters' priorities were 'skyrocketing house prices, clogged roads, under-resourced schools and hospitals [and] degraded rivers' (Ardern, 2018: 34). Labour, under the previous administration, had been portrayed as uncapable of delivering on those priorities. Ardern's task was to convince voters that Labour would listen to their concerns.
The second difference between these politicians was the electoral context. As a Labour List MP (selected by her own party to guarantee a more representative system, rather than elected by the public), Ardern was asked to lead the Party to the next election and became New Zealand's Prime Minister in a Mixed-Member Proportional Representation (MMP) system. In the United States' Presidential system, 3 Clinton won the popular vote (by over 2%; nearly three million votes) but ultimately lost Electoral College votes to Donald Trump. In the United Kingdom, May ran a Conservative Election Leadership campaign: she was elected by fellow Conservative MPs, who whittled down the contestants from the initial five. 4 The difference in electoral systems was indicative of both the type of constituents these politicians appealed to and the style of communication they had to adopt. Notably, May spoke to Conservative grassroot and MPs; Ardern to a country that had voted on the other side of the ideological spectrum for 10 years 5 ; Clinton to a country swept by anti-establishment and anti-government mood and fragmented along identity politics lines (Denton, 2017). 6 Yet, despite the points of departure, it would be wrong to conclude that these politicians and their campaigns should not be compared. In fact, they shared several conceptual attributes (Rose and Mackenzie, 1991) that travelled across geographical, ideological, socioeconomic and structural boundaries (Mazur and Hoard, 2014). First, they were advocates for women's rights. For instance, May had been Minister for Women andEqualities (2010-2012) and Shadow Secretary of State Education (2011). During this time, she introduced a married couple's tax allowance and child-care package. Without wanting to deny that these policies were at least in part designed to win women's votes, they nonetheless showed a commitment to women and their concerns. Ardern had not held a Cabinet post before she became Prime Minister, but her portfolio as Labour spokesperson included traditionally 'soft' policy areas such as Social Development (2011) and Arts, Culture and Heritage and Children (2014). As First Lady, Clinton actively rejected the traditional role as host as well as that of sponsor for non-governmental matters (Parry-Giles and Blair, 2002). Instead, she rebelled against the status quo by, for instance, refusing to take her husband's surname and politicising healthcare reform. Later on, as Secretary of State, she championed gender and sexual equality at home and abroad.
The second trait that made these politicians comparable was the treatment they received in the news media along the gendered lines identified by Norris (1997) and others (e.g. Sreberny-Mohammadi and Ross, 1996) more than two decades ago. For instance, around the time under investigation, journalists reported on May's passion for shoes (Hawkes and Cole, 2016), her legs (Vine, 2017) and her reputation for being a 'bloody difficult woman' (Wooding, 2016) rather than her experience and qualifications. Clinton's vilification in the news media is well documented (Center for American Women and Politics, 2016). Of note here were criticism of her smile (@DavidFrum, 2016;@PGourevitch, 2016), physical appearance (@MashableNews, 2016; @mmfa, 2016) and her fitness to run for President (Rafferty, 2016;Seitz-Wald et al., 2016;Thrush et al., 2016). Unsurprisingly, Ardern's age and experience were prominent in press coverage of her candidacy (Hurley, 2017;Newshub, 2017b), but also her appearance (Newshub, 2017a;Robinson, 2017) and her intentions to have a family (Bracewell-Worrall, 2017;Jones, 2017) were covered. By stereotyping Ardern, May and Clinton in this way, the news media denied their achievements, experience and competence.
The third feature that allows for a comparison was their use of social media as tools for political campaigning. Despite the differences in polity (in parliamentary systems, candidates have less incentives to use social media because the party is responsible for setting out specific policies; in candidate-centred systems, social media are used to reach the electorate and build a particular image), social media have become essential tools for political campaigning in all three countries.
The rest of this article evaluates whether Ardern, May and Clinton were able to use social media to overcome the barriers they faced and to represent the interests and priorities of other women in ways that allowed them to make a difference in women's lives. In order to answer these questions, the analysis that follows focuses on whether May, Clinton and Ardern's online communication challenged masculine norms of how politics is to be conducted, what issues were framed as politically meaningful and to what extent their online communication gendered political debate by appealing directly to a female electorate.

Method
The data for this study were collected through the Web Data Research Assistant (Carr, 2017) and hand coded. It included original tweets and Facebook posts sent from @ HillaryClinton between 26 July 2016 (date of Clinton's formal nomination at the Democratic National Convention) and 8 November 2016 (the day of the Presidential Election); @the-resa_may and @TheresaMayOfficial between 10 June 2016 (when May announced her candidacy for the leadership of the Conservative Party) and 13 July 2016 (when she became UK Prime Minister); @jacindaardern, @NZLabour 7 and NZLabourParty, between 1 August 2017 (when Ardern became leader of the Labour Party) and 19 October 2017 (when she was sworn in as Prime Minister). These data excluded retweets and replies, because evidence suggests that politicians use social media to counter-balance existing power structures, rather than as tools for interacting with voters (Grant et al., 2010;Jackson and Lilleker, 2011;Stromer-Galley, 2014).
The choice of multiple social media platforms 8 was driven by a number of considerations. First, the need to capture a richer data set that would allow for a more systematic comparison: whereas Clinton was a prolific social media user (as dictated by American political campaigning tradition for over a decade), the UK and NZ politicians have been slower adopters of digital technologies (Cardo, 2018;Fountaine et al., 2019;Margetts, 2017). In addition, different platforms captured distinctive communicative styles aimed at different constituencies (Morozov, 2013): across all three countries, Twitter was the preferred tool for male millennials (YouGov, 2019b), whereas Facebook was more popular with older multi-ethnic female users (Pew Research Center, 2018;YouGov, 2019a).
It was often impossible to distinguish between posts written by the politician and those written by her campaigning team. This should not be seen as a limitation: this study aimed at evaluating communication that purported to come from these politicians and the focus was on their message, regardless of whom the source of this communication may be. This makes sense because, both in candidate-and party-centred systems, campaigning teams 'stay on message' to convey the core party/candidate strategy.

Analysis
Consistently with other studies of this type (Evans and Clark, 2016;Jungherr, 2016;Meeks, 2016), the collected tweets and posts were coded for two variables: leadership and communication style and language and policy concerns. These variables were designed to measure the extent to which these politicians feminise political debate, challenge masculine ideas of what is a legitimate political concern and substantively represent their constituents.

Leadership and communication style
This category grouped the different styles and language that Ardern, Clinton and May used in their online communication, in order to find out how they presented themselves and whether they used masculinist, adversarial discursive norms (Walsh, 2001) or a more consensual leadership style. Historically, women have favoured the latter (Sreberny-Mohammadi and Ross, 1996) but were forced to adopt the former. This was not due to their biological makeup (indeed Thatcher embraced adversarial vernacular 9 ; Fairclough, 2015), but because of the performative expectations of politics as a male endeavour and the double bind that came with them.
The analysis of the data revealed a variety of communicative styles that complicated the relationship between gender and leadership, consensual and adversarial style. All three politicians showcased both male and female styles. Clinton's election campaign slogan 'stronger together' exemplified this well by bringing together the value of collaboration (typical of a consensual style politics) and the typically male attribute of strength.
Ardern's tweets provide several alternative examples. For instance, 'we'll build a better, fairer Aotearoa' 10 (@nzlabour, 2017c) and 'I want to build a country where every child grows up free from poverty' (@nzlabour, 2017b) combined a soft policy area to a typically male activity: building. A leader, for Ardern, would erect a structure strong enough to support a generation. By using the pronouns 'I/we' (rather than Labour), she conveyed the message that, as leader, she would be the one to lead in this endeavour.
Consistent with existing scholarship (for instance, Evans and Clark, 2016), Twitter proved to be the platform that most encouraged an adversarial tone. For instance, when Conservative MP Ken Clarke called May a 'bloody difficult woman', she capitalised on the remark by renewing her no-compromise stance on the EU Referendum: 'There must be no attempts to remain inside the EU, no attempts to rejoin it by the back door and no 2nd referendum' (@theresa_may, 2016a). In this way, May turned a potentially damaging allegation 11 into political capital: her 'difficult' character became an asset for strong leadership rather than a hindrance to the country.
Clinton also displayed a confrontational style by attacking directly her opponent. Although there is no consensus on whether women politicians are more likely than men to 'go negative' (Evans and Clark, 2016) or whether this style is overall a useful campaign strategy, Clinton used this confrontational style to both challenge Trump's fitness to represent the country and to claim a place as a woman in a masculine world. Her Twitter feed combined negative messages framed as statements of fact (e.g. 'Donald Trump [. . .] demeans women, and mocks people with disabilities' (@HillaryClinton, 2016b); 'oppose reproductive rights, support conversion therapy and proposed a Muslim ban' (@ HillaryClinton, 2016j)) with the words of her supporters, for instance, Obama's claim that 'Hillary Clinton is that woman in the arena' (@HillaryClinton, 2016d). In this statement, Clinton was framed as a fighter in a man's world (the arena) and 'there has never been a man or a woman more qualified [. . .] to serve as president' (@HillaryClinton, 2016h).
The adversarial tone of Clinton's attacks on Trump, then, were framed as necessary for her to demonstrate her ability to stand in a men's world. The calibre of her backers and their words merely confirmed that she should be trusted regardless of her gender. As she put it, 'it is ridiculous that there has never been a female president before' (@HillaryClinton, 2016e): the (traditionally male) system has failed, not women.
A second area of contention, when it comes to gendered political communication, is personalisation of politics. By allowing individual voices to be heard, social media are well placed as tools for a personalised style of communication (Enli and Skogerbø, 2013;Meeks, 2016), especially Facebook. Personalisation encourages politicians to reveal details about their private lives, such as their relationships and family status and, in so doing, to appear approachable, down-to-earth and 'like us' (Langer, 2012). It is an asset for politicians seeking to draw attention to themselves in an overcrowded multimedia environment (Lilleker and Jackson, 2010). Women politicians, who personalise their campaigns and reveal personal details about being a woman, a wife and a mother, however, not only become victims of vitriolic online abuse (Southern and Harmer, 2019). They are also caught in a double bind that calls into question their competence and ability to do their job. Research finds that, as a consequence, women politicians actively avoid self-representing in this way online (Lee, 2013;Meeks, 2016).
Discussions of motherhood provided a good example: 'Among our intake of 17 new MPs, we've got two who've just become mums [. . .] Here's what they had to say about juggling motherhood and the big job [. . .]!' (New Zealand Labour Party, 2017a) and 'Watch Jacinda speak to young mums this morning about the challenges and opportunities that women face, and how to embrace them' (New Zealand Labour Party, 2017b). In both these cases, Ardern used motherhood to expose the reality of women's lives and the way they are disrupted by motherhood. She mainstreamed motherhood by presenting it as a phenomenon that affects a large portion of the population and that it is still mostly the responsibility of women. Crucially, she spelt out that, once they become mothers, women occupy multiple spaces and are able to multitask. This includes their ability to both be mothers and politicians.
In one case, when motherhood became a personalised issue, it was rejected outright on the basis of relevance, at least online. 12 When Andrea Leadsom, May's opponent in the Conservative Leadership campaign, suggested that mothers make better leaders, May depersonalised the attack by bringing the attention back to 'what is important': 'Yesterday, I launched my clean campaign pledge, and invite @AndreaLeadsom to join me in signing it' (@theresa_may, 2016c). This rejection conveyed a sense that the personal need not be political. On the contrary, in this tweet at least, the two are diametrically opposed. This separation between the personal and the political was also apparent in Clinton's communication style. When comedian Louis C. K. came out in support of Clinton's campaign as the 'tough bitch mother, who does the job', Clinton replied, 'Thanks, Louis C.K.-but I prefer "Madam" Tough Mother' (Hillary Clinton, 2016b). By using the formal title for high political position, Clinton's reply was designed to bring back the conversation from women as mothers to women as leaders.
The data analysed here complicated this relationship between communication, gender and political representation. It showed that, Ardern, Clinton and May adopted both female and male communicative styles to create a 'complete' politician: one that would at the same time stand up to her opponents, compromise and care. Personal life for this politician was only relevant because it spoke of the life of women, a traditionally underrepresented section of society. As such, discussing personal experience in the public sphere meant bringing marginalised perspectives to the forefront of political discourse; it meant feminising politics.

Policy concerns
This category grouped online discussions of policy concerns in order to establish what constituencies Ardern, Clinton and May sought to represent. The data revealed that all three politicians addressed a combination of soft and hard policy areas in ways that complicated the essentialising gender/policy binary. Family was a good example: 'parents just want to give their kids a better life. That's what growing up in small business families meant [. . .]. We're going to keep fighting so every family has that opportunity' (Hillary Clinton, 2016a) and 'your family's economic security is too important to risk. Who do you trust to get the best Brexit deal for your family and the UK?' (TheresaMayOfficial, 2017). In both cases, families had economic value; they mattered for the country because they contributed to its economic success. To care for one's family meant guaranteeing economic prosperity for the country.
Discussions of education made a similar point: 'School Leavers' Toolkit -another way we're helping young people prepare for the future' (@nzlabour, 2017d) 'with the practical knowledge and skills to help them succeed in the world of work and in our communities' (New Zealand Labour Party, 2017c). Both posts spoke of the school system as a tool to predict the economic development of a nation: good schools prepare future generations for the economic success of the country; good governments provide good schools because, ultimately, this benefits society as a whole. In the case of discussions about both family and education, soft policy areas were framed as politically meaningful for the 'majority', rather than a woman's concern.
Hard policy areas 13 constituted the core of Ardern, Clinton and May's online communication. These areas were also reframed in complex ways, as the economy examples that follow demonstrate: 'a successful economy is one that serves its people -not the other way around' (@nzlabour, 2017a); '[. . .] We don't just believe in markets, but in communities' (@theresa_may, 2016b). This framing of the economy as a compassionate endeavour suggested that economic prosperity should be seen as an investment for all; not just as profit, but as a public service. The focus on the ways in which economic prosperity can benefit the people were also evident in this tweet: 'helping small businesses succeed is about more than policy -it's personal ' (@HillaryClinton, 2016c). Here, the focus shifted away from the market on to the people. This femininisation of the economy was even more evident when it was framed around (gender) equality. For instance, 'women deserve equal pay across our nation ' (@HillaryClinton, 2016a) and 'in 2017 there should be no such thing as a gender pay gap in New Zealand' (NZ Labour Party, 2017). In both statements, the suggestion was that economic inequality is anachronistic and should not occur in rich democracies.
Foreign policy and national security were also popular online topics for these politicians, especially Clinton and May, given their prior expertise. 14 Competence in these masculine policy areas is associated with leadership skills (Cramer Walsh, 2002). It should be no surprise then that these topics were often discussed online. In the case of Clinton, the discussion alternated endorsements by powerful men (e.g. former acting director of the CIA: 'she will deliver on the most important duty of a president -keeping our nation safe' (@HillaryClinton, 2016g); then President of the United States: 'She [. . .] forcefully argued in favor of the mission that took out Osama bin Laden' (@HillaryClinton, 2016f)) and emotional appeals to 'stand up against terror and come together to prevent these attacks. -H' (@HillaryClinton, 2016i). By using her initial, Clinton suggested that national security was important because it was personal: when the nation is under attack, 'we' all are. This personalisation of a hard policy area feminised it too.
This analysis suggests that, despite their marginalisation in mainstream media, these politicians did not see social media as alternative platforms to discuss women's concerns; at least not in a straightforward fashion. Instead, they framed policy areas in two interesting ways: first, they demonstrated the relevance of soft policy areas for the majority. Second, they argued that women can be trusted with hard policy areas, because these areas have a powerful impact on women too. In this sense, women possess the expertise they need to tackle them. So, online discussions of policy concerns gave Ardern, Clinton and May the chance not only to substantively represent women but also to make a broader representative claim on the 'majority' of the population.

Substantive representation?
This comparative study between New Zealand, the United Kingdom and the United States aimed at finding out whether women political leaders used social media to feminise politics at election time. In particular, it evaluated the extent to which they adopted leadership and communicative styles that challenged masculine norms of political behaviour, whether they prioritised policy areas that were more likely to make a difference in women's lives and if they spoke on behalf of women. Similarly unequal political systems, their treatment in mainstream media and a shared interest in women's issues and concerns, facilitated this comparison.
Consistent with the finding that party allegiance is a more accurate predictor of politicians' values than gender is (Norris, 1996(Norris, , 2000Norris and Lovendunski, 1995), this analysis showed that ideological values partly determine the politician's feminisation of her political communication. May's failure to challenge Leadsom's statement that mothers make better leaders, for instance, could be seen as adherence to conservative regulatory gender frameworks that celebrate motherhood and prescribe specific roles to women in society. It also suggested that she may have avoided confrontation because women are more likely to adopt consensual styles of communication. However, May reversed these ideological expectations when she declared that the next PM would be a woman. This was a way to disrupt politics as usual and feminise debate, by challenging traditional masculine norms of politics.
Similar complexities emerged around the content of communication. By and large, Ardern, May and Clinton stayed on message online like they did offline. This however did not limit their ability to speak with their own distinctive expert woman's voice. The fact that they did not use social media platforms to directly address women's issues may tell us more about the priorities and specificities of the elections they fought than their commitment to speak on behalf of other women. Social media played a role in constraining the content and style of these politician's online communication, since the content of online communication is often used by journalists as news sources. This means that women politicians are hampered, online as well as offline, by regulatory frameworks and masculine rules of politics.
Consistent with other research (Childs and Webb, 2012), this analysis showed that descriptive representation did not necessarily correspond to substantive representation, especially when it came to gendering the debate and when feminist (rather than women's) concerns were at stake. However, it also revealed that Ardern, May and Clinton found a space on social media to combine complex communication styles to showcase some of the expertise that is traditionally denied to them in mainstream media. In this sense, this study complicated the relationship between women's concerns, styles of communication and leadership. (Bordo, 2017). May followed a slightly different political path, elected Conservative MP for Maidenhead in May 1997, when she was 41 years, but having been involved in local politics since 1986 (Theresa May MP, n.d.). 7. Analysing multiple accounts was necessary in the case of Ardern, because, soon after becoming leader of the Labour Party, she moved all her communication to the official Labour Party account. 8. Of note here, Theresa May made very little use of Facebook. This is reflected in the lack of Facebook data in the analysis of her online behaviour. Instagram would have added a richer analytical layer to this study, but the legal requirements of this particular platform made inclusion of these data impossible. 9. Thatcher's statement, 'I fight on. I fight to win', made after the inconclusive first ballot in the Tory leadership election in 1990, was one such example. 10. Original Māori name for New Zealand. 11. In fact, however clumsily, Clarke had not meant the remark to damage May. On the contrary, he had argued that, despite being a 'bloody difficult woman', May would make a better Prime Minister than the alternative. The news media did not report this in full and the negativity of the statement stuck. 12. Ardern had a similar reaction when asked by a journalist whether she wanted children. She openly berated the journalist (who had also suggested that employers should have the right to know women's family plans before giving them a job) but made no reference of the incident on social media. 13. Discussions of hard policy areas mostly took place on Twitter, whereas both Facebook and Twitter provided platforms for discussions of soft policy areas. 14. May was in the Home Office between 2010 and 2016 and Clinton served as Secretary of State from 2009 to 2013.