“Don’t Put Color in Your Hair, Don’t Do This, Don’t Do That”: Canadian Mayors’ Mixed Gender Performance on Social Media

Abstract Although mayors can have important impacts on citizens’ daily lives, local politics remains understudied, especially compared with national and regional politics. This study focuses on Canadian mayors’ digital political gender performance—or self-presentation—on Facebook, Twitter, and Instagram and the context in which this gendered performance arises. Overall, results confirm that mayors’ gendered performances are on a continuum rather than binary. Results from a visual content analysis of nine Canadian mayors’ social media accounts show that, broadly speaking, women mayors gravitate toward congruent, mixed gendered performances and avoidance strategies, whereas men mayors also display mixed performance of their gender, while more freely exploring congruent and incongruent approaches to gendered stereotypes. Additionally, semistructured interviews with these mayors show that women mayors still work under added constraints because of their gender, which translates into comments on their appearance, attitude, and lifestyle choices; increased aggression and lack of respect; and a generally greater mental load.

). In addition, the visual communication literature shows the importance of politicians' digital image production in maintaining and reinforcing positive political attitudes and influencing political decision-making (Strachan and Kendall 2004). Although some scholars have studied well-known politicians' image production strategies, such as Justin Trudeau's Instagram posts or Angela Merkel's appearance, little work has been done on mayors (Flicker 2013;Lalancette and Raynauld 2017).
This study thus examines Canadian mayors' gender performance on their professional Facebook pages, Twitter accounts, and Instagram accounts through a visual content analysis of their posts, and the context in which the gender performance arises through semistructured interviews. The following research questions are explored: How do Canadian mayors perform gender through their professional Facebook pages, Twitter accounts, and Instagram accounts? In what context does this gendered performance arise?
Results from a visual content analysis of nine Canadian mayors' social media accounts show that, broadly speaking, mayors demonstrate mixed gender performance, regardless of their gender, although women mayors also tend to adopt avoidance strategies when they are not favoring a congruent approach. Additionally, interviews highlight a context in which women mayors face comments on their appearance, attitude, and lifestyle choices. This can-and in certain cases, does-impact their appearance, whereas men mayors do not report such comments, and results from content analyses suggest that men have greater freedom to explore congruent and incongruent approaches to gender stereotypes. Results also confirm that mayors' gendered performances are on a continuum rather than binary.
This study shies away from mayors of big cities to focus on the realities of local government outside the limelight. Its methodology relies on a collection of gender performance indicators on a continuum ranging from feminine to masculine and encompassing many facets of gendered visual cues, from their facial expressions to the events they attended. Hence, it offers an intimate glimpse into these mayors' daily lives through semistructured interviews and an in-depth analysis of their digital political image, thus contributing to limited research on Canadian mayors, the role of gender in local political actors' digital communications, and the mental load it entails.

Social Media and Municipal Politics in Canada
Canadian mayors' work has important impacts on citizens' daily lives, and yet we know very little about their responsibilities, as shown in Graham's (2018) research on urban Canadian mayors. Despite this, scholars often forget local politics, preferring federal and provincial politics, which attract a greater deal of attention. Larsson and Svensson (2014), who examined the literature on the use of digital tools by politicians, noted the importance of studying local politics to balance the number of studies on the national level requires greater scientific attention in political communication.
Although the literature on Canadian municipal politics has mostly focused on electoral behavior, digital political communication scholars are beginning to take an interest in local politics, particularly because "many local governments rival their provincial and federal counterparts in terms of social media deployment and usage" (Gruzd and Roy 2016, 80; see also Riarh and Roy, 2014;Cutler and Matthews 2005). There are now studies on the role of social networks in the formation of local electoral dynamics during the municipal election in Ottawa in 2010 (Raynauld and Greenberg 2014) and social media's potential as a tool in the 2010 Niagara elections (Hagar 2014).
However, little attention has been devoted to social media platforms outside of these extraordinary situations. Indeed, it has been established that politicians participate in a permanent campaign, and yet a great deal of research continues to focus on electoral campaigns that do not reflect daily political digital practices (Cutler and Matthews 2005;Dumitrica 2014;Hagar 2014;Marland, Giasson, and Esselment 2017;Raynauld and Greenberg 2014 As the position of mayor varies greatly across the country, going from a fulltime to a part-time position, the level of professionalization may vary greatly. However, as local media are increasingly disappearing and social media platforms gain in popularity to share information, mayors may be motivated to use these free tools (Enli 2017). Additionally, it is possible that mayors' gender performance may matter more than that of national and provincial political actors, especially in municipalities where local politics is nonpartisan. Indeed, such constituencies may allow local political actors to be "closest to the people," thereby focusing on their needs and concerns (Oliver 2012). Lucas, McGregor, and Tuxhorn (2022, 189) find that in large nonpartisan cities, the personal vote-or a personal connection-is very strong.
Considering the importance of a mayor's role in a community, as well as the normalization of social media as a local information broadcasting tool, these political actors are an interesting population case to examine digital political gender performances.

A Theoretical Framework on Digital Gender Performance
Gender is viewed as a continuum of attributes ranging from femininity to masculinity, rather than a binary social construct (Goodyear-Grant 2019). It is instead constitutive and performative and created by personal behaviors, such as grooming and body movements, publicized interactions, and staged photographs (Butler 1990;Wagner and Everitt 2019). The performative nature of gender comes from the repetition of words and actions in an endogenous cycle, where it is unclear whether one acts a certain way because of one's gender, or whether said act contributes to reinforcing gender (Butler 1990). According to Butler (1988, 526), "gender is an act which has been rehearsed, much as a script survives the particular actors who make use of it, but which requires individual actors to be actualized and reproduced as reality once again." This political gendered self-presentation can be influenced by many factors, including incumbency, public opinion, and the media. The last often prioritizes a male gaze when covering politics-for example, by using sports and war metaphors-in line with hegemonic masculinity, which ritualizes, naturalizes, and reproduces dominant forms of masculinity (Sabin and Kirkup 2019, 47;Trimble and Sampert 2004). Hence, gender stereotypes-the traits individuals are believed to possess by virtue of their specific gender-still guide what society views as acceptable behavior (Wagner and Everitt 2019). According to the gendered mediation thesis, news media uphold the gender binary by reinforcing the notion that women are an anomaly in politics and belong to the private spheres of life (Ross and Comrie 2012).
By explicitly signifying the gender of politicians, news media are practicing "gender marking," which politicizes gender by underlining the uniqueness of women in politics, reinforcing the understanding that politicians should be men, constructing a woman's ability to lead as being limited by her body, and asserting that women politicians may use the "gender card" for political gain (Falk 2010;Falk 2013, 196;Meeks 2012;Nicholson 1994, 81;Trimble 2017). Hence, gender marking can harm women politicians' political viability, as their appearance is used to evaluate whether they are performing their gender correctly and make them seem less competent (Falk 2010;Mandziuk 2008).
This gender marking can be particularly prevalent given the rise of personalization by the media and by politicians themselves (Strömbäck 2008;Van Zoonen 2005). This intimization, or the publicizing of information and imagery of the politician as a person, their family life, as well as their personal spaces, may affect politicians differently according to their gender (Stanyer 2013).

Gender Stereotypes
Schneider's (2014a) typology of the uses of gender stereotypes in politicians' political communication focuses on four strategies: reinforcing gender stereotypes (congruent), overturning stereotypes (incongruent), mixing stereotypes (mixed), and avoiding gender stereotypes (avoidance). Goodyear-Grant (2019) explored how Schneider's typology fits with current and future research on selfpresentation of gendered identities and noted that "overturning" is a hefty order. Hence, the extent to which gendered self-presentation is congruent with stereotypes, incongruent, mixed, or completely avoids any reference to gender will be examined. It is also important to note that research on gendered mediation tends to reinforce gender dualism, which, in turn, reinforces hegemonic gender schemas (Lünenborg and Maier 2015, 183). This generally happens when quantifying news stories referring to politicians according to feminine and masculine characteristics, behaviors, and policy issues, as there needs to be room for performances that go beyond the heteronormative binary (Trimble 2017). Thus, Schneider's (2014a) typology was selected to take a more comprehensive and flexible approach to gender performance.
Reinforcing Gender Stereotypes (Congruent) Schneider (2014b) analyzed communication strategies based on gender stereotypes and hypothesized that a politician may either reinforce or bend a gender stereotype. This is consistent with strategic stereotype theory, according to which politicians capitalize on gender stereotypes that benefit their political aims while attempting to counteract potentially damaging gender stereotypes (Fridkin and Kenney 2014).
For example, gendered frames in political news coverage may refer to women politicians by reinforcing gender stereotypes as "girly moms" and to men by reinforcing their gender as "alpha males" (Lemarier-Saulnier and Giasson 2019). Such framing reinforces preexisting stereotypes-for example, that women politicians are better at handling issues like education, health care, elder care, and "women's interests," whereas men politicians are more competent when dealing with issues such as defense, business, and crime (Lawless 2004;Schneider 2014a).
Chen, Park, and Joo (2020) examined the role of visual self-presentation on social media in gender stereotypes during the 2018 U.S. general election. Results show that highly feminine candidates displayed a friendly, attractive, and maternal personality, often smiling and appearing with family. Highly masculine candidates were found to share photos of strong and authoritative activities, such as visiting construction sites, shaking hands, and sharing work-related achievements. Prediction experiments also found that these masculine traits were predictive factors for winning elections (Chen, Park, and Joo 2020). Hence, some issues are traditionally seen as feminine, such as education, arts and culture, health care, seniors, addiction, the environment, and women's rights, whereas the economy, law enforcement, business, and criminal justice are seen as agentic (Brands, Kruikemeier, and Trilling 2021;Dolan 2005).
During the 2015 Canadian federal election, Stephen Harper fit the "alpha male" criteria, having interpreted a traditional form of masculinity by embodying a stoic, desexualized, and managerial persona, or a "typical 1950s suburban dad" (Sabin and Kirkup 2019). Canadian men politicians favor suits in formal settings, and some, like Prime Minister Justin Trudeau, prefer a dress shirt with rolled-up sleeves and a tie in casual settings (Lalancette and Raynauld 2017;Mattan and Small 2021).
Additionally, women are expected to make others feel comfortable through emotional displays, such as smiling, and do so more than men (Fischer and LaFrance 2015). However, smiling can be a social act, especially when interacting with others (LaFrance 2011; LaFrance and Hecht 2011). This has been clear on the campaign trail, as Hillary Clinton performed far more intense smile displays than Donald Trump during the 2016 presidential debates, and Representative Madeleine Dean reported having an aide hold a sign during campaign events to remind her to smile (Astor 2019;Senior, Ridout, and Stewart 2019).

Overturning Stereotypes (Incongruent)
According to Goodyear-Grant (2019), while men politicians tend to prefer a reinforcement strategy, women politicians often attempt to cue masculine stereotypes. By doing so, some women politicians are described as "iron ladies" and seen as rigorous, ambitious, competent, and independent. An iron lady often focuses on public work, business, and law and displays agentic leadership while valuing a successful career (Lemarier-Saulnier and Giasson 2019). Margaret Thatcher gained the epithet after defying expectations of how women should behave (Pullen and Taksa 2016). Cueing masculine stereotypes can improve a woman politician's evaluation, but it can also lead to critiques of toughness, failed femininity, or even masculine mimicry (Bauer 2017). Indeed, women are judged on their ability to "perform" their gender by being self-effacing, compliant, submissive, and cooperative, which is communicated by their appearance and clothing choices (Eagly and Karay 2002;Mandziuk 2008).
It is important to note that gendered performances can be difficult for both men and women, as the former must navigate between hegemonic and subordinate masculinity (Sabin and Kirkup 2019). Masculine nonconformity can be punishing for men politicians and is commonly experienced by gay men, racialized men, and men who publicly express emotions or behaviors associated with femininity (Cooper 2009;Everitt, Best, and Gaudet 2016;Golebiowska 2002). This overturning of the masculine stereotype is associated with traits such as indecision, passiveness, weakness, and emotiveness (Conroy 2015). After becoming Liberal Party leader in 2013, Justin Trudeau was attacked by his opponents with implicit and explicit connections between his masculinity and his fitness for government. This was further shown by Trudeau's attempts to recuperate his masculinity by accentuating his physicality-which could be seen as a mixed strategy (Sabin and Kirkup 2019).

Mixing Stereotypes (Mixed)
Women politicians are now expected to display agentic behavior and penalized if they do not (Bongiorno, Bain, and David 2014). Likewise, men politicians are expected to possess more traditionally communal-or feminine-qualities (Everitt, Best, and Gaudet 2019). Hence, some men politicians emphasize traits such as empathy, loyalty, and dependability by appearing with family members, children, and babies (Goodyear-Grant 2019). Thomas Mulcair, then leader of the New Democratic Party, attempted to reframe his gendered performance by recasting his masculinity ("angry Tom") in friendlier terms before the 2015 federal elections but failed from lack of authenticity (Sabin and Kirkup 2019). Racialized men candidates may also choose a mixed strategy to avoid triggering negative stereotypes. Barack Obama faced a double bind during his campaign, and a careful balance had to be struck as masculinity could easily lead to the "bad Black man" image and femininity would not be seen as presidential (Cooper 2009, 637).
However, a double bind may still exist, according to role congruity theory (Eagly and Karau 2002). Butler and Geis (1990) show that competent, assertive women speakers elicit more negative facial expressions than do equally competent, assertive men. According to Flicker (2013), women in political leadership positions must walk a fine line. When they perform and dress femininely, they may be perceived as deficient, but when they refuse typical feminine looks in favor of male dress codes, they are perceived as conspicuous, further "othering" women in politics.
Similarly, motherhood can accentuate the conflict between women's public and private roles, leading to doubts about their capacity to balance politics with family. Indeed, mothers can be viewed as selfish for pursuing their professional ambitions (Van Zoonen 2005). The "intensive mothering" ideology portrays women as self-sacrificing and child-absorbed primary caregivers for their children (Hays 1996). A great example of a mixed strategy is Sarah Palin's "Mama Grizzly" frame, which combined motherhood with aggressive masculinity (Goodyear-Grant 2019). Indeed, during her nomination as the Republican vice presidential candidate, Palin's political career was framed as a mother's quest to ensure her children's well-being, rather than to further her career.
Avoiding Stereotypes (Avoidance) Finally, practical examples of gender avoidance are less common, but Flicker's (2013) analysis of Angela Merkel's fashion choices shows that the German chancellor avoided any reference to her femininity, preferring a generic haircut, a pantsuit with blazer, and minimal makeup and jewelry. Merkel also refused to have her handbag included in any photograph (Flicker 2013).
Additionally, studies on the inclusion of politicians' children on U.S. congressional websites show that mothers tend to de-emphasize their children online, whereas fathers tend to include their family (Meeks 2016;Stalsburg and Kleinberg 2016). The presence of children in women politicians' pictures may negatively accentuate communal traits and highlight their worklife balance. For men politicians, including children in photos is generally a reinforcement strategy and is received positively: it softens their "agentic" characteristics.

Image Production
According to Schneider's (2014a) typology of the uses of gender stereotypes, politicians alter their appearance, the values they champion, and the people they surround themselves with to better perform their gendered political communication strategy. Although visual communication remains an emerging field, scholars are increasingly showing interest in the importance of visuals in the construction of political images, as they represent the dominant mode of learning (Barry 2005;Filiminov, Russman, and Svensson 2016;Schill 2012). Indeed, a single image can impact citizens' assessment of politicians' competence, agreeability, leadership, and integrity (Rosenberg et al. 1986). In addition, when verbal and visual messages conflict, individuals tend to retain visual information (Lang 1995;Shea and Burton 2001). More importantly, visual communication is a social process that considers the target audience (Kenney 2009). Therefore, in this study, in line with Hall (1973) and Barthes (1964), images are viewed as containing signs-or codes-to be deciphered by receivers.
This image production, which stems from the professionalization of political communication, can be achieved by creating a simple message, saturating communication channels to increase its salience, leveraging the support of well-known personalities, and convincingly delivering the message (Bennett 2014;Schill 2012). This can help reach the audience, both intellectually and emotionally, to maintain and reinforce positive perceptions in order to influence personal political attitudes as well as political decision-making (Lalancette and Raynauld 2017;Lalancette and Tourigny-Koné 2017;Strachan and Kendall 2004).
Hence, as both men and women politicians are increasingly expected to perform their roles by adopting some traits from the opposing gender while maintaining normative gender stereotypes, it is expected that mayors will tend to adopt a mixed approach to gender stereotypes (Bongiorno, Bain, and David 2014).
H 1 : Both women mayors and men mayors will tend to adopt a mixed approach to gender stereotypes in their visual social media publications.
However, as some politicians are punished for transgressing stereotypical gender roles, it is also expected that women mayors who do not adopt a mixed strategy will avoid stereotypes altogether, whereas men mayors who do not adopt a mixed strategy will reinforce gender stereotypes.
H 2 : Women mayors who do not adopt a mixed strategy will tend to avoid stereotypes in their visual social media publications. H 3 : Men mayors who do not adopt a mixed strategy will tend to reinforce gender stereotypes in their visual social media publications.
The COVID-19 pandemic shed much light on the invisible work that women do within and outside the home. This "mental load" refers to the cognitive workload associated with maintaining every aspect of daily life, which includes doing household chores, having a career, and performing gender roles (Robertson 2017: Robertson et al. 2019. Occhiuto (2021, 3) defines it as "the mental effort and processes involved in the optimal daily and generational reproduction of the household, family and the self." According to Weeks (2011), individuals also perform their gender in the workplace. This is in line with Dean, Churchill, and Ruppanner (2022, 14), who argue that the mental load is both cognitive and emotional labor and operates at home and in society according to three characteristics: it is invisible, boundaryless, and enduring. Similarly, a political actor's gender performance may be experienced as an additional mental load, especially if, and when, it requires reflection and planning, and when it is compounded with news coverage practicing gender marking.

Sample Selection
The sample for this study was selected from a database (Sullivan 2021a) of 3,525 Canadian mayors that was created using Statistics Canada's 2016 census, electoral results, as well as data from Facebook pages and Twitter and Instagram accounts. This original method of data collection began in mid-November 2018 and ended in mid-January 2019.
Mayors who used all three social media platforms between December 1, 2018, and January 12, 2019, were identified, and more than half of them were invited by email to participate in a 30-to 45-minute semistructured interview. 1 Out of the 21 mayors contacted twice, 9 accepted the invitation. The small size of the sample is attributable to the few mayors who fit the criteria and had the availability or desire to participate in the study. These 9 mayors, however, published a great quantity of social media posts, generating a large digital data set.
As indicated in Table 1, the mayors in the sample came from small-to medium-sized municipalities in the provinces of Ontario, British Columbia, Quebec, Northwest Territories, New Brunswick, Prince Edward Island, and Alberta and, for the most part, fulfilled their duties on a part-time basis. This limited the level of professionalization of their digital communications, suggesting a more authentic digital communication strategy.
Finally, to respect the mayors' anonymity, women mayors are referred to as Woman Mayor "x" (or WMx) and men mayors as Man Mayor "x" (or MMx).

Content Analysis
Mayors' posts from December 1, 2018, until May 31, 2019, were collected manually. This six-month period followed municipal elections in Ontario, Yukon, Northwest Territories, British Columbia, New Brunswick, Manitoba, and Prince Edward Island, offering a glimpse into some of these mayors' first months in office. An overview of the mayors' social media activity can be found in Table 2.
It is important to note that this study focuses on Facebook pages, rather than Facebook profiles, for ethical reasons. Some mayors had hybrid profiles, mixing their professional and personal lives. It was thus difficult to pinpoint what is meant for public consumption and what is meant for family and friends.

Interviews
The mayors were invited to discuss their digital communication practices and their experiences in local politics during 30-to 45-minute interviews. 2 Once the invitation was accepted, the content analysis of their social media posts was conducted. The goal was to complete preliminary content analyses before speaking with the mayors to avoid subsequent bias. Interviews took place on Skype, FaceTime, and Facebook Messenger as the mayors in the sample were spread across the country, and this strategy offered greater accessibility and flexibility.

Content Analysis
Schneider's modified typology (2014a) was operationalized to examine mayors' visual gender performance. Only posts containing at least one image were analyzed, and images containing mostly text were excluded. Posts were then manually coded according to their nature; the mayor's presence, facial expression, and physical appearance; the presence of other individuals; and indicators of intimization. To narrow down pertinent posts, most of the proportions presented in the tables were calculated based on the number of visual publications (image or video), or the number in which the mayor was present, as presented in Table 2. The framework is composed of five dimensions. The first examines mayors' facial expressions according to three categories: (1) smiling, (2) focused, and (3) other. The second dimension focuses on mayors' physical appearance, divided into three components: (1) feminine, (2) neutral and (3) masculine 3 . The feminine subgroup includes indicators such as makeup, hairstyle, the presence of a handbag and jewelry, as well as clothing, such as a skirt or dress (see Table A1 in the Supplementary Material). The neutral subgroup contains a selection of casual attire, outerwear, and holiday accessories (Table A2). Finally, the masculine subgroup contains indicators such as sportswear, (pant)suits, and hockey jerseys ( Table A3).
The third dimension is the presence of other people in visual publications. This dimension is divided into two components: (1) feminine and (2) masculine. The feminine subgroup includes indicators such as artists, children, seniors, health professionals, and educators. The masculine subgroup includes indicators such as businesspeople, law enforcement, athletes, and blue-collar workers.
The fourth dimension focuses on events, which are divided into two components: (1) feminine and (2) masculine. The feminine subgroup includes indicators of learning settings, community events, and arts and culture, whereas the masculine subgroup is composed of indicators such as public relations, meetings, and sports.
Finally, the level of intimization, the fifth dimension, was examined by identifying the proportion of posts containing an image in which a partner, friend, and/or children were easily identifiable, as well as activities at home or out of a mayor's public function.
The protocol was tested on three mayors' social media accounts before it was applied to all mayors. Also, the coding categories are not mutually exclusive. For example, an image of an event can include both politicians and teachers.

Interviews
Mayors were asked to answer a dozen open-ended questions about their experience in local politics, their digital political image, and the context in which they perform their mayoral duties. Interviews were recorded, transcribed, and then analyzed to identify general themes, such as their preferred appearance, external pressures to conform, gender biases, and social media strategies.

Results
This section presents results from the visual content analysis and interviews. It first offers a glimpse of mayors' positions on the gendered performance axis, then focuses on aggregate results of mayors' facial expressions, physical appearance, people who appear in their visual publications, general trends in events and the level of intimization, followed by mayors' experiences navigating gendered performances as mayors.

Gendered Performance Overview
To begin, every mayor in the sample was placed on the gendered performance axis (Figure 1) to offer a visual representation of results. Mayors on the left of the figure displayed the most incongruent gendered performance stereotypes, mayors in the bottom center presented an image of themselves that mixes gender stereotypes, mayors in the upper center avoided any reference to gender, and mayors on the right shared a political digital image in line with their gender's stereotypes. Mayors appear in Figure 1 according to their individual results, from both the content analysis and interviews. This overview has no pretention to be a formal gradation, but rather acts as a visual aid. Hence, Figure 1 offers a simplified placement of mayors, according to their gendered performance on Facebook, Twitter, and Instagram, as well as their responses during interviews, on an axis based on Schneider's (2014a) typology. For example, WM5's gendered performance was somewhat congruent with her gender but also showed a lot of avoidance.

Facial Expression
Results in Table 3 demonstrate that most mayors, regardless of gender, tend to smile in most of their visual publications that include their facial expressions, particularly on Instagram. WM3 was the woman mayor who published the most pictures in which she looked focused, both on Facebook (29%) and Twitter (46%). Men mayors tend to appear more focused than women mayors, except for WM3. Table 4 presents results regarding mayors' gendered appearance. Results for each platform represent an average number of gendered indicators by visual publication, focusing solely on publications in which mayors are present. Results confirm that mayors' gendered performances are on a continuum rather than binary. For example, WM4 did not display many gendered traits online. On Instagram, her gendered performance was slightly more neutral (0.90) than feminine (0.71) or masculine (0.81).

Physical Appearance
Also, all mayors displayed a mix of feminine, masculine, and neutral traits to different extents. Results vary greatly among mayors of the same gender and across social media platforms.
The highest numbers of gendered traits correspond with the feminine indicators for women mayors and masculine indicators for men mayors. The more pronounced congruent gendered performances can be found among women mayors, especially WM2, WM3, and WM5 across all three platforms, and MM1 on Twitter. Overall, despite low averages of neutral indicators, women mayors appear to avoid references to gender in their appearance more than men mayors. Indeed, whereas most men mayors seem to prefer a mixed or congruent approach, results suggest that when women mayors prefer to avoid references to their gender, they will adopt both neutral and masculine traits to a greater extent.

People in the Frame
Furthermore, other individuals making appearances in mayors' visual publications were identified to shed some light on issues and target groups.
Results from Table 5 demonstrate, overall, low averages of gendered indicators by publication when considering other individuals in the frame.
Among women mayors, there were three published images in which they find themselves in the company of stereotypically masculine people, such as  businesspeople, athletes, and law enforcement. Indeed, WM3 (0.53), WM4 (0.62) and WM5 (0.33) seem to have favored Twitter to share such images. Among men mayors, MM2 also displayed an incongruent approach, especially on Instagram (0.50), where he showcased individuals who signal a more feminine gendered performance, with images of health care professionals, educators, children, artists, and seniors. In addition, whereas WM3, WM4, and WM5 used Twitter for an incongruent approach, WM1 and WM2 preferred Facebook, and men mayors, Instagram.

Types of Events
The types of events featured in mayors' visual publications were also analyzed. Results are presented in Table 6.
Results suggest that, broadly speaking, mayors did not show highly stereotypical displays of gender in their choice of events. As expected, both men and women mayors showed a mixed strategy. However, on average and across all three platforms, women mayors displayed the most congruent approach to their visual gender performance.
In addition, Instagram appears to represent an outlet for more feminine gender performances among men mayors, similarly to results in Table 5. Twitter may represent women mayors' digital space for an incongruent approach, although results are less pronounced.

Intimization
Results in Table 7 show that very few mayors included their partner in their posts, especially among women mayors. Indeed, they made appearances in 1% of WM2's Facebook posts and 4% of WM5's Instagram posts. Aside from MM3, all men mayors shared images of their partner to some extent, especially on Instagram. This platform appears to offer a more intimate look into mayors' lives, as most mayors shared a good number of images of their life space, either at home or outside of their work. However, images of children remain scarce, as only WM1, WM3, MM2, and MM4 shared one or two visual publications of their little ones.

Interviews
This section offers a glimpse into Canadian mayors' experiences in local politics by focusing particularly on the context in which their gendered performance arises.
To begin, when discussing gender in politics, especially regarding the news coverage, all women mayors perceived a difference in the coverage of women and men politicians, citing examples such as Christy Clark. One shared that, during the electoral campaign, journalists had inquired about her marital status. Alternatively, responses were mixed coming from men mayors. While MM1 had not noticed a gendered difference in the news coverage, he did note that women do not speak to him the same way a man would, the latter being much more aggressive and impolite.
When asked about the image they aimed to project online, it became clear that women mayors had already put some thought into their digital persona. Nearly half spoke of the importance of authenticity online, while the other half shared that they tended to avoid posting images of themselves. For WM5, this was   Politics & Gender motivated by the impression that it is a feminine practice. This is not to say that men mayors did not speak of authenticity, but they shared less on their personal image and more about representing their community, a motivation shared by WM3.
Similarly, on the topic of their physical appearance, whereas nearly all women mayors stated that they either had received unsolicited comments on their appearance or had been asked to modify how they dress, none of the men mayors reported anything similar. In MM3's case, he had received suggestions on the content he shares online related to his work as mayor, but nothing about his appearance.
There also appears to be a consensus among women mayors that there is pressure to conform. For example, WM2 was told by her team, during the municipal election, how to present herself-or rather, how not to present herself-by staying away from her usual forms of self-expression. This is a topic that recurred during discussions with other women mayors, as both WM3 and WM5 chose to change certain aspects of their physical appearance to limit unpleasant comments and for self-preservation. Even things such as their smile or their emotions had been the topic of discussion from citizens and colleagues alike: "I've had 'don't be so emotional'" (WM2).
This sparked discussions on gendered double standards, as these comments, according to the women mayors, are rarely directed toward men politicians. "I never hear that with men. I don't hear 'that your chest hair is showing, or I think those shoes are a little too flashy for your job'" (WM2). MM2 agreed that there exists a double standard and that nobody would comment on a man's appearance.
In addition to the pressure to appear a certain way, or rather to not appear too feminine, women mayors brought up other difficulties, such as a lack of respect on behalf of men colleagues, either at conferences or even the council chamber. WM1 did rationalize this difficult work environment by pointing out that, as a woman mayor, she is now in a field created by men for men, making it easier for them to navigate. Hence, WM1 recounted that at first, her different approach "definitely got under people's skin." This lack of respect can border on inappropriate behavior at work, as WM4 had faced aggression from colleagues during professional disagreements and condescending comments, which echoes WM5's experience with men colleagues acting aggressively toward her and her staff members. This can certainly affect a woman politician's experience, as WM2 received inappropriate comments on her appearance from a man colleague during a professional event. Unfortunately, as some women mayors pointed out, sexism in the workplace can be subtle but just as harmful. Citizens also contribute to creating a hostile working environment for some women mayors. Indeed, both WM3 and WM5 reported having members of the community openly criticize their appearance according to their preferences or commenting on their choice to pursue a career in local politics.
Finally, regarding sharing details of their private life online, mayors did not all speak on the topic, but those who did, including WM1, WM5, MM3, and MM4, shared that it was a conscious decision. For some, such as WM1 and MM3, they lived and worked in small communities where they were well known and chose to keep their private life separate online. In some cases, as for WM5, this involved asking loved ones to avoid engaging with trolls online, even if it could be difficult to watch. As for MM4, sharing some aspects of his private life was a way of humanizing himself by showing that he is a husband and a father.

Discussion and Conclusion
This study examined Canadian mayors' gender performance on their professional Facebook pages, Twitter accounts, and Instagram accounts through a visual content analysis of their posts, as well as the context in which gender performance arises through semistructured interviews. I sought to answer the following research questions: How do Canadian mayors perform gender through their professional Facebook pages, Twitter accounts, and Instagram accounts? In what context does this gendered performance arise?
Results show that, overall, women mayors continue to perform a stereotypical congruent version of their gender, while men mayors show a wider range of approaches. The restrictiveness of women mayors' gender performance is demonstrated in both the content analyses and the semistructured interviews.
For example, although most mayors smile in their social media posts, men mayors appear more focused compared with women mayors. This is consistent with research showing that women continue to be expected to make others feel comfortable through emotional displays, such as smiling (Fischer and LaFrance 2015). Similarly, women mayors revealed during interviews that they feel a definite pressure to conform and, in some cases, modify their appearance, and they expressed frustration about the existence of a gendered double standard in politics. This pressure was not expressed by men mayors, as shown in their greater use of incongruent gender performances in regard to their physical appearance. Indeed, while both genders showed an inclination toward mixed gender performance, women mayors showed a greater congruent approach on all levels, including their appearance, as the people in their published images and the events they attended.
Results also illustrate a gendered use of social media platforms for visual selfpresentation. Whereas women mayors appear to prefer Twitter to embrace an incongruent gender performance, men mayors preferred Instagram. This platform appears to represent a more intimate digital space, as mayors who chose to share details of their private life online, mostly used Instagram to offer a glimpse into their world outside of work.
It is also important to note that the broad collection of gender performance indicators used in this study allowed to examine mayors' gender performance in detail by going beyond the usual binary operationalizations. This made it possible to capture nuances better examined on a continuum. Indeed, mayors' physical appearance results reveal that rather than being binary, mayors' gendered performances are instead on a continuum. Beyond these empirical results, this study makes theoretical and conceptual contributions-namely, by exploring the gendered differences in digital self-presentation practices among political actors who often operate in nonpartisan environments, while creating strong bonds with constituents.
One aspect of this study that would warrant further research is the mental load experienced by women mayors, as well as mayors who fall outside the current political norm. The term "mental load" was never explicitly used in interviews, but a majority of women mayors detailed the thought process behind their current appearance, the changes they had made to appease the angry comments, and, in some cases, their self-censorship online to avoid cueing their femininity. These additional cognitive gymnastics were not mentioned by men mayors.
In addition, semistructured interviews revealed that women and men mayors describe different working environments and constraints. Discussions about gender biases in politics, as much in the council chamber as in the media coverage, highlighted women mayors' experiences of exclusion, disrespect, and othering. All women mayors-except one-had received unsolicited comments on their appearance, and many had modified how they dress and present themselves. None of the men mayors had received unsolicited feedback on their appearance. Instead, comments tended to be about the content they shared online. Women mayors also reported facing aggression and subtle sexism in their workplace. Altogether, this suggests that invisible gendered barriers still exist in local politics.
Thus, this study shows the importance of analyzing politicians in context, rather than simply focusing on their appearance, as it can be misleading when other aspects, such as the type of events they attend or the people they showcase, can either emphasize a gendered performance or subdue it. Hence, a woman mayor can wear dresses and a lot of makeup but still present herself as a more stereotypically masculine politician.
This study did not control for mayors' age, the size of their municipality, or the number of years spent in politics. The aim was not to generalize these results to all Canadian mayors, but to offer a glimpse of gendered performances of local politicians who actively used Facebook, Twitter, and Instagram to share information with their constituents. Its operationalization was also fraught with difficulties, as coding had to be done by relying on the description (or lack thereof) of images by mayors, tagged users, and recognizing recurring individuals and spaces. As images only offer a partial view of an event, certain elements of a mayor's appearance, such as their shoes or even a skirt, were difficult to identify when images were cut at the waist.
Finally, such a study has never been done on local politicians' digital gender performance. Those that do focus on political gender performance tend to examine the national level and rely on limited and binary gender indicators. Federal politicians, however, usually work under different circumstances, such as greater budgets and communication teams, as well as greater media scrutiny. The mayors in this study are much more relatable, often having been part of their community their whole lives, meaning that both their professional and private lives are intertwined.
Hence, this study aimed to examine how citizens with ambitions for their community and varying levels of resources present themselves and to widen the lens with which visual gender performance may be examined. A future study could gain tremendously by including an ethnographic element, thereby avoiding crafted answers during interviews.
Supplementary Materials. To view supplementary material for this article, please visit http:// doi.org/10.1017/S1743923X23000041.