Running Head: THE ODD ONE OUT
University of Amsterdam – Graduate School of Communication Research Master’s Program Communication Science
Master Thesis
Supervisor: dr. Penny Sheets Thibaut Date of Completion: 28-06-2018 Word count: 9448
The Odd One Out –
An Investigation into the Association between Gender Identity, Party Affiliation and Competence and Likeability in Female Politicians
Lea Kremer 11386304
THE ODD ONE OUT
Abstract1
Despite much progress in recent decades, many female politicians are still
underrepresented and alienated from the political system. Most research on the portrayal of female politicians in the media focuses on prominent individual politicians in a US context. This study combines a mixed-method approach to analyse the construction of a female identity in politicians. Study 1 extends the boundary conditions of previous literature by systematically analysing the portrayal of male and female gender identity in politicians in the UK during the election period 2017. It finds that female politicians are not portrayed according to expected gender norms, but are depicted with significantly more stereotypical masculine traits compared to male politicians. It proposes that party affiliation affects the perception of female politicians, by simultaneously creating expectations regarding their behaviour. Study 2 experimentally tests the assumption that party- and gender-conforming behaviour predict likeability and competence in female politicians. It finds that female politicians that behave contrary to gender stereotypes are perceived as less likeable, but finds no predictor of competence. Hence, female politicians are not feminine enough to gain on female gender stereotypes, but also loose likeability on male
stereotypes, while not gaining on competence. This indicates that the role of women in politics is vague and expectations of it are undefined and it points towards a need for redefining the identity of female politicians.
1 Due to the combination of two methodologies in this study, this thesis exceeds the proposed word limit of 7500 words.
THE ODD ONE OUT
“Ah! Belva, with all your womanly wisdom you are getting the cart before the horse.”2 ("The State Republic," 1884)
While Belva Lockwood, the first female candidate appearing on an official ballot in the USA in 1884 was a novelty; women’s participation in politics is continuously on the rise. In 2016, 22.8% of national parliamentarians worldwide were female (Women in National Parliaments, 2016).
Nevertheless, women still face adversity in the political arena. Many press offices working for female members of congress describe the struggle of creating an adequate
representation that combines their role as woman and as politician (Niven & Zilber, 2001). While women have successfully entered the political arena, femininity is still an outlier in the
androcentric political system (Lünenborg & Maier, 2015; Rosenwasser & Dean, 1989). Thus, female politicians have to be careful how they portray themselves, to not be stuck in what researchers call a “double-bind” (Dabbous & Ladley, 2010, p. 183; Hall & Donaghue, 2013, p. 6). That is, to be perceived as competent female politicians have to exhibit stereotypical masculine traits, such as assertiveness and confidence (Vos, 2013). However, a rejection of stereotypical female traits, such as sensitivity and nurturance can cost them likeability (Hall & Donaghue, 2013; Hehman, Carpinella, Johnson, Leitner, & Freeman, 2014). This paradoxical demand on female politicians hinders many women from succeeding in the politics.
While many researchers focus on the way voters use stereotype heuristics to assess the viability of female political candidates (Brians, 2005; Norton, 1999; Schneider & Bos, 2014), they tend to neglect the role the media plays in shaping these heuristics (Adcock, 2010). As media reproduces dominant gender stereotypes in society (Adoni & Mane, 1984), a systematic analysis of the portrayal of female politicians in the media can reveal society’s perception of
THE ODD ONE OUT
female politicians. Whereas many studies focus on the media coverage of individual prominent politicians (Dabbous & Ladley, 2010, Meeks, 2013a, Lünenborg & Maier, 2015), only few studies assess media coverage of female politicians in general (Bystrom, Robertson & Banwart, 2001; Meeks, 2012). However, a focus on prominent female politicians neglects the realities of female gender in every day politics, thus prohibiting the study of gender identity at a macro-level. As research (i.e. Heflick & Goldenberg, 2009; McGinley, 2008) continues to focus solely on how gender is constructed in female politicians, it continues to alienate women from the political system (Stern, 1993). Only by assessing both the female and the male gender identity in politicians, can we understand where those identities align, and where they differ.
RQ1: How is the female identity constructed in media coverage and how does it differ from the
coverage of male politicians?
As media coverage is shaped by dominant gender stereotypes, it in turn influences our perception of how female politicians are to behave. Particularly, as female politicians are often still perceived as less competent than male politicians (Hehmann et al., 2014, Lawless, 2004 Okimoto & Brescoll, 2010) the media can have a strong influence on how female politicians are perceived by voters. Thus, we have to equally consider how the media portrayal of a female gender identity in politicians shapes perceptions about them.
RQ2: How does gendered media coverage affect attitudes and perceptions about female
politicians?
While female participation has increased in politics, women are still under represented in nearly every national legislature (Golder et al., 2017) and bias against women in politics still affects voting behaviour (Karpowitz, Monson, & Preece, 2017; Streb, Burrell, Frederick & Genovese, 2007). An increased representation of female politicians does not only increase political involvement in adolescent girls (Campbell & Wolbrecht, 2006), it also empowers
THE ODD ONE OUT
women’s behaviour and leads to a more positive self-evaluation (Latu, Mast, Jammers & Bombari, 2013). Research on women in politics is thus not only essential to uncover biases and limitations that prevent equal participation, it also is important for women’s empowerment.
This study utilizes a mixed-method approach to evaluate the construction and reception of a female gender identity in politicians through the media. By combining a content analysis of news coverage during the 2017 UK election period, and an experiment examining the effects of coverage on attitudes towards hypothetical male and female candidates this study is able to systematically assess the female identity in politics.
Theoretical Background The Female Gender Identity
For a majority of communication science gender is a demographic variable that predicts an outcome rather than explaining (Reeder, 1996; Aelst, Sehata, & Dalen, 2010; Vos, 2014). Here, the sociological construct of gender is equated with the biological sex. However, differential outcomes are not created by the sex. Very often, the alienation of the female gender in society is rather the result of a process of socialization (Reeder, 1996; Barge, Schlueter and Pritchard, 1989). By defining gender solely as a dichotomy between male and female, we neglect the more nuanced expression of gender as a collective and individual identity (Lünenborg & Maier, 2015). Simultaneously with the struggle for gender equality in society, many feminist scholars have focused on the experience and creation of female gender identity as a major step in the transformation of the traditional gender binary (see here Holmes & Meyerhoff, 2008). Particularly, West and Zimmermann’s (1987) interactionist perspective on gender, as well as Butler’s (1990, 1993, 2004) constructionist analysis of gender have shaped research of gender identity and expression. While both perspectives differ in their analytical approach, they agree that gender is a performative act that arises from the interaction with society rather than the
THE ODD ONE OUT
biological sex. Building on Goffman’s theory of performative identity (1978), every aspect of gender identity is expressed in a performance that is evoked and shaped by societal structures. Success or failure of such a performance is largely based on adherence to social norms. Consequently, the expression of gender is not so much based on having a female body, but on historical conditions, expectations and limitations placed on the female identity. Being Woman is not an inherent and consistent attribute found in all women, but rather an active performance that is shaped by the surrounding power relations (Butler, 2003). A failure to “do gender”
appropriately through adherence to societal expectations can lead to a rejection by society (West and Zimmermann, 1987). Throughout western history the dialectic between the male and female has led to a subordination of the female in an androcentric society (Marlow, 2002; Mead, 1963; Oakley, 1974). The female identity is thus defined in its divergence from the male. By focusing on the difference the female displays from the androcentric, we do not only assume a shared female identity, we also place femininity at the very centre of every female identity. Whereas man can be an individual, woman is defined foremost by being woman (Alcoff, 1988).
Accordingly, the concept of female identity is as much collective as it is individual (Gardiner, 1981).
Revolutionary changes in the last century have led to women entering previously uniquely masculine areas, forcing a re-invention of the societal expectations and acceptance of the female gender identity (England, 2010). One such highly masculinized domain is the area of politics (Meeks, 2013a). Whereas some women in western countries only achieved suffrage less than 50 years ago3, other women have made large advancements in the political world, currently serving as heads of governments in 10 countries (Women in National Parliaments, 2016). While many
3 Switzerland in 1971 and Liechtenstein in 1984 are the last Western countries to have passed women’s suffrage (Paxton, 2000; Paxton et al., 2003). Nevertheless, many women around are still restricted in their voting rights today (Dewey, 2013).
THE ODD ONE OUT
female politicians today do not face the same adversity from the political system, femininity is still very much an outlier in many areas of the androcentric political system (Lünenborg & Maier, 2015; Rosenwasser & Dean, 1989). As women enter politics a new female identity is created in the dialectic with the androcentric political arena. The advancement of female politicians has allowed researchers to study how female politicians portray themselves and are in turn portrayed by society.
Study One: Women in Politics and the Media
Particularly, the media plays an important role in shaping the discourse on women in politics (Dabbous & Ladley, 2010; Kahn, 1994; Uscinski & Goren, 2010). Mediated discourse “(…) (re-)produces the dominant discourse about the role women should play in society”
(Garcia-Blanco & Wahl-Jorgensen, 2012, p. 422) and reinforces established gender roles (Adoni & Mane, 1984). The media mediates our perception of social reality by depicting these
established gender identities and framing them as relevant and valid constructs. Media is created by journalists, whose perception is shaped by a hierarchy of influences ranging from professional norms, organizational and ideological context to the society surrounding them (Reese, 2001). Such professional norms of objectivity within the profession and organization prohibit journalist from challenging the perceived norm of gender roles (Schudson, 2001; Tuchmann, 1978). Moreover, research suggests that even among female journalists, masculine values, experiences and socialization are considered the norm, reducing the influence of a journalist’s gender on news coverage (De Bruin, 2000; Everbach, 2006; Meeks, 2013b). Consequently, journalists are more likely to reproduce dominant gender norms in society than challenge them.
As society accepts the mediated depiction of gender, it is then reproduced by the media (Takeshita, 1997). This cycle, which traps female politicians in self-reproducing stereotypes, continuously alienates them from the political arena. Van Zoonen (1998) finds that people only
THE ODD ONE OUT
think of politicians in a limited number of categories. Thereby, individuality in politicians is rejected in favour of clear-cut identities. Consequently, it is not only relevant how the media displays dominant gender roles in political coverage, but also how these in return affect perceptions and attitudes about politicians, particularly female ones.
After what Tuchmann (1978) called the “symbolic annihilation” of women, through their complete disregard by the media, the anomaly of women entering the political arena and vying for office forced the media to take notice. Women in male spaces are a deviance and deviances are always newsworthy (Meeks, 2013a). Thus, the media primarily focused on the outsider role of women in politics, through an increased focus on the viability of female political candidates (Kahn, 1992, 1993, 1994). While the increase of female candidates has led to an overall reduction of the novelty aspect of female politicians (Smith, 1997; Banwart, Bystrom & Robertson, 2003), gender in current political media coverage is still being singularized and appraised, albeit in a more subtle fashion. Understanding the constructions of gender identities by the media requires an awareness of how gender identities are created and perceived by society. Many previous researchers have focused on the analysis of individual female politicians such as Hillary Clinton (Meeks, 2013a; Uscinski & Goren, 2010), Nancy Pelosi (Dabbous & Ladley, 2010), Julia Gillard (Hall & Donaghue, 2013) or Angela Merkel (Lünenborg & Maier, 2015). While a singular focus analysis is important to understand the constraints placed on the individual female identity, we thereby neglect to explore the shared collective female identity that is expected of female
politicians. Only few studies (Bligh, Schlehofer, Casad & Gaffney, 2012; Bystrom, Robertson & Banwart, 2001; Meeks, 2012) attempt to do so by expanding their focus to a larger number of female politicians. The focus of a majority of these studies is a North-American context. Of the little research on women in politics an even smaller number focuses on a diverging contexts (i.e. Baird, 2004 (Australia); Garcia-Blanco & Wahl-Jorgensen, 2012 (EU); O’Neill & Savigny, 2014
THE ODD ONE OUT
(UK); van Zoonen & Harmer, 2011 (Germany)). A systematic analysis of media representation of female politicians is necessary to understand how the female identity of women politicians is shaped in diverging contexts. Such analysis has to go beyond the novelty aspect of female politicians and focus on more subtle divergences of the female identity in the androcentric arena of politics by drawing on a broad base of gender research.
Previous research has identified four essential areas in which the media coverage of female politicians can diverge from that of male politicians; gender-traits, body presentation, private-public divide and gendered language. Based on this research hypothesis for the analysis of media coverage of female politicians are identified.
Gender-trait stereotypes describe the dichotomy between agentic, or stereotypical masculine, and communal, or stereotypical female traits (Huddy & Terkildsen, 1993). Earlier studies find that while agentic traits, such as assertiveness or ambition, are considered crucial qualities of political competence, a display of communal traits, such as nurturance or pleasantness is essential for a female politician’s likeability (Dabbous & Ladley, 2010; Gidengil & Everitt, 2003). As stereotyping is an important component of maintaining the social order male and female politicians will be more likely associated with the according gender-conforming traits (Talbot, 2008).
H1a: Male politicians are more often connected to agentic traits than female politicians.
H1b: Female politicians are more often connected to communal traits than male politicians. As competence is connected to masculinity, so are power and ambition still inherently constructed as male. Accordingly, the explicit expression of power or of related emotions, such as anger or passion, is considered natural in men, but unnatural in women (Lakoff, 2008). While male politicians display power and engage in conflict, female politicians are expected to restrict such ambition and display more harmonious and cooperative behaviour.
THE ODD ONE OUT
H2: Male politicians are more often referred to in relation to controversy/ conflict than female politicians.
Likewise, Vos (2013) finds further evidence of the agentic-communal dichotomy in the area of policy issues. Thereby, issues such as health or education, that are closely related to communal or caring aspects, are often perceived as more feminine “compassion issues” (Bligh et al., 2012, p. 561), while security or the military are seen as more masculine issues (Huddy and Terkildsen, 1993; Dayhoff, 1983; Sanbonmatsu, 2003).
H3a: Male politicians will be referred to more often in relation to masculine-coded policy areas, such as defence and national security, international security and terrorism, and military in general than female politicians.
H3b: Female politicians will be referred to more often in relation to feminine-coded policy areas, such as social welfare, education, health care or culture and arts.
According to Foucault (1976) the body is more than a sole physical shell. It has an inherent (political) power that is infused through the repudiation and simultaneous desire with which society regards it. Female politicians, as women, do not escape the politicization of their bodies. The (self-) sexualisation of the female body that is prevalent in celebrity culture however does not translate to the political arena, as it is considered unsuited for political office (van Zoonen & Harmer, 2011). Nevertheless, female politicians are often judged for diverging from the ideal body, while at the same time regarded as less competent if fulfilling it. Thus, research has found an increased focus on appearance in the coverage of female politicians (Devitt, 1999; Heldman, Caroll &Olsen, 2000, 2005). A male politician is solely considered in his function as politician, whereas a female politician is woman and politician at the same time and cannot escape the societal scrutiny that is inherently attached to the female body (Sreberny-Mohammadi & Ross, 1996, Gidengil & Everitt, 2003).
THE ODD ONE OUT
H4: The appearance of female politicians will be more often referred to than for male politicians. Furthermore, there is an increased focus on the private life of female politicians (Heldman et al., 2000). Although, there is an overall trend towards “celebrity politics” (Van Zoonen, 2006 p. 288) or personalization in politics (Langer, 2007), private life for male and female politicians is constructed differently by the media. Whereas, male politicians show their dependability and their reliability through their private life, for female politicians, their private life represents a return to the deep-rooted women-private, men-public dichotomy (Muir, 2005). The depiction of private life in the media is more frequent for female politicians, to show the double strain of the family and political life and hence return her to an outsider position in the political arena (Van Zoonen, 1998).
H5: Female politicians are more often referred to in relation to family members than male politicians.
Gender in political coverage is not only made evident through a diverging focus regarding male and female politicians, but also through the use of gendered language. In strongly gendered languages, such as German or Italian, this becomes evident through the use of the generic
masculine and dissymmetric feminine, which stresses the asymmetry of the genders (Sensales, Areni & Dal Secco 2016). This “asymmetric language treatment” (Pauwels, 2003, p. 553) is equally prevalent in languages with a less dominant grammatical gender, either through the use of the generic masculine as an all-encompassing form or through the accentuation of the female as diverging from the denominal masculine. In political coverage this can be expressed through the explicit use of the first name for female politicians, compared to the last name for male
politicians (Uscinski & Goren, 2010) or through other forms of hierarchical naming. While overt linguistic sexism has been significantly reduced over the years, also due to the inclusive use of
THE ODD ONE OUT
new language (Sensales et al., 2016), the asymmetric treatment is still prevalent in political coverage.
H6: Female politicians will be more often referred to with linguistic sexism (hierarchical naming and gender marking) than male politicians.
A majority of the research has resulted from a US context. To better understand the role of female politicians as portrayed by the media we have to broaden this focus. This study has chosen to focus on media portrayals of female politicians in the UK. Not only does the UK share the English language with the US, making it easier to apply research on (linguistic) sexism, but also 30% of politicians in the UK are female (“Election 2017”, 2017) compared with only 20% in the US (Women in Elective Office, n.d.). Moreover, the ascent of Theresa May to Prime Minister in 2016 positions a female politician at the centre of the political arena. The UK media system presents “ a mid-position between U.S. professionalism and continental European partisan tendencies” (Umbricht & Esser, 2013, p.5. after Schudson, 2001) rendering it an ideal bridge between US focused research and a new European context. The more partisan nature of the UK media also offers a stronger focus on individual candidates compared to other European media landscapes (Holtz-Bacha, Langer & Merkle, 2014).
Study Two: Media Coverage and the Perception of Female Politicians
Not only does media reproduce dominant paradigms in society, it also shapes attitudes and perceptions. This significance of news coverage in political campaigns has been well
established in previous research, that showed the influence of news media on voters’ perceptions of political candidates as either positive or negative, as well as their overall evaluations of candidates (e.g., Bennett, 2000; Krosnick & Kinder, 1990). Female politicians are often still perceived as less competent than male politicians (Hehmann et al., 2014, Lawless, 2004) or when seeking office (Okimoto & Brescoll, 2010). Thus, the effect media coverage has on that
THE ODD ONE OUT
perception becomes even more relevant. Nevertheless, while numerous studies have researched the effect of portrayals of female politicians on attitudes about those same, all did depend on the results of previous research (Bligh et al., 2012, Dabbous & Ladley, 2010, Gidengil & Everitt, 2003, Hall & Donaghue, 2013; Hehman et al., 2014, Schneider, Tinsley, Cheldelin &
Amanatullah, 2010). To fully understand the portrayal of female politicians through the media and their reception by the public we have to combine both. Only by structurally analysing the media coverage of female politicians and its effect on the audience, can we begin to deconstruct established gender roles.
Accordingly, a large number of studies have traced the “double-bind” (Dabbous & Ladley, 2010, p. 183; Hall & Donaghue, 2013, p. 6) women face in male dominated areas. An overt display of agentic qualities is required to show (political) competence, but a rejection of communal values is seen as a deviation from the expected female identity expression coming at cost to a female politician’s likeability (Hall & Donaghue, 2013; Hehman et al., 2014, Bligh et al., 2012). Here, female politicians are required to “do gender” in an opposing manner. To be perceived as competent, women have to adapt to the male identity of politician. However, a rejection of their femininity comes at cost to their likeability. The audience, who expects gender conformity, will reward women who exhibit a female gender performance with likeability but deny their competence. Contrary, female politicians who reject a female identity in favour of the androcentric role of politicians will be rewarded by the media audience with competence but rejected as unlikeable. I thus propose the following hypothesis.
H7: Female politicians who are depicted as gender conforming will be perceived as more likeable than female politicians who are depicted as non-gender conforming.
H8: Female politicians who are depicted as gender non-conforming will be perceived as more competent than female politicians who are depicted as gender conforming.
THE ODD ONE OUT
Rudy, Popova and Linz (2010) define the three possible goals of a gender-oriented content analysis as the support of feminist claims, the comparison of media with reality and the prediction of effects of media on audience. Thus, we do not only have to understand how media portrayals of gender roles reflect our social reality but also how they in turn shape audience perception through suggested gender ideal. This research will combine the assessment of media portrayals by means of a content analysis of UK newspapers and an experimental study that will test the effects of the former on attitudes towards female politicians, enabling us to better
understand how gender identities shape and restrict the existence of female politicians.
Study One
First, a quantitative content analysis using articles from three UK newspapers was conducted. The Guardian and the Daily Telegraph are both broadsheet newspapers aligned with the opposite ends of the political spectrum with the Guardian being a left-leaning newspaper often aligned with the Labour party and the Daily Telegraph being a conservative newspaper more aligned with the Conservative party (Johnson, Culpeper & Suhr, 2003). Additionally, the
Sun is included as right-leaning tabloid paper appealing to popular audience (Vincent, 2004). On
the 18th of April 2017, Theresa May announced that a general election was to take place on the 7th of June 2017 (Boyle & Maidment, 2017). Previous research finds that personalization, the focus on individual politicians vis-à-vis their party, increases during election times (Garcia, 2014). Consequently, we assume an equally increased focus on the portrayal of individual politicians during the proposed time period that makes the analysis of media portrayals of male and female politician more apprehensible.
A constructed week sample (Riffe, Aust & Lacy, 1993) of 150 articles was selected from the LexisNexis database using the search terms: election AND (Brit* OR UK OR Engl* OR
THE ODD ONE OUT
Scot* OR Wales OR Welsh). Overall, 45 articles 4from the Daily Telegraph, 45 articles from the Sun and 50 articles from the Guardian were coded. An additional coder coded 15 articles or 10% of the original dataset to test the reliability of the codebook. Reliability was assessed using Krippendorff’s alpha 5(Krippendorff, 1970). Actors appearing in the article were coded
individually. 794 actors were coded, of which 576 actors were political actors. These actors were divided between 57.8% national politicians, 0.8% local politicians, 12.5% International
politicians and 3.1% political administrative actors. The majority of actors were male 6 (66.4%). Variables and Operationalization
Agentic and communal traits were chosen based on the stereotyping measures of Hoffmann and Hurst (1990). The presence or absence of six agentic (e.g. assertiveness, confidence or competitiveness) and five communal traits (e.g. gentleness, kindness or emotionality) was coded (on a 0-1 binary scale) for each actor mentioned in the article. The scores of the relevant items were added to create a positive and negative agentic trait scale, as well as a positive and negative communal trait scale. Table 1 presents the average scores for all trait scales, as well as the scores for the harmony and conflict scale.
4As LexisNexis included different regional versions (e.g. The Sun –Scotland) of the newspapers, some articles
appeared twice in the sample. However, these articles were only coded once reducing the sample size slightly.
5 All reliability results not presented in the text are presented in Appendix B, table 1. Levels of Krippendorff’s alpha were used to account for the use of binary variables. Overall results showed good reliability.
6 Both male and female actors were included in the content analysis to avoid further supporting the idea that male actors are the status quo while female actors are a divergence, thereby allowing the researcher to assess both portrayals in comparison with another (Stern, 1993).
THE ODD ONE OUT
Table 1. Distribution of Dependent Variables
Source Krippendorff’s α M (SD) Agentic Traits Positive .66 .3 (.71) Negative .74 .12 (.16) Communal Traits Positive .74 .03 (.71) Negative .88 .06 (.27) Conflict .89 1.29 (.89) Harmony .73 .31 (.55) N 576
Conflict and harmony scales were adapted from the conflict frame developed by Semetko and Valkenburg (2000). Three items were coded on a 0-1 binary scale for conflict, focusing on disagreement between actors or one actor criticizing another. Likewise, three items were coded for the harmony scale focusing on actors working together or one actor praising another. The score of these items were combined to create a scale ranging from 0 to 3. Articles scored on average M = .99 (SD = .89) on the conflict scale and M = .22 (SD = .47) on the harmony scale.
To assess the presentation of the body in the coverage of politicians, any relation to appearance was coded. Only for 1.2% (female: 2.6%, male: 0.5%) of all actors was there a
mention regarding their appearance. Likewise, any mention of family was coded for each actor. A reference to family was made for 2.4% of all political actors (female: 2.6%, male: 2.3%).
Using the research of Pauwels (2003) linguistic sexism was assessed through the use of gender markings (e.g. the use of specific female role descriptors) and hierarchical naming (the
THE ODD ONE OUT
deprecating use of, e.g. girl or boy for an adult person). Explicitly female gender markings and hierarchical naming were only present for 0.2% of all actors. Lastly, policy topics were coded on a 0-1 binary scale from a list of twenty topics. One reference to a topic was sufficient for it to be coded. There was no limitation to the number of topics coded per article. On average articles mentioned 3 topics (M = 3.31, SD = 1.94). EU issues (19.3%), national politics (15.7%) and health care (15%) were the most common policy topics appearing.
Results
Of the 576 political actors coded in the sample, 193 were female and 383 male. Table 2 depicts the mean scores across the dependent variables by gender.
Table 2. Distribution of dependent variables by gender of political actor
Source Male Actors
M (SD) Female Actors M (SD) Positive Agentic .22 (.54) .45 (.95) Negative Agentic .09 (.39) .19 (.56) Positive Communal .02 (.14) .04 (.19) Negative Communal .04 (.2) .1 (.36) Conflict 1.32 (.87) 1.23 (.91) Harmony .32 (.55) .3 (.55) N 383 193
H1 expected that in accordance with gender stereotyping, agentic traits would be mentioned
THE ODD ONE OUT
for female political actors (H1b). An independent samples t-test7 was conducted for all trait scales
in male and female political actors.
There was a significant difference in positive agentic traits (t(256.54)=-3.19, p = .02) and in negative agentic traits (t(288.64)=-2.24, p = .03), with women scoring higher on both scales (positive: M = .45, SD = .95; negative: M = .19, SD = .56) than men (positive: M = .21, SD = .53; negative: M = .09, SD = .39). While there was a significant difference, there is no increased connection between male politicians and agentic trait, but rather between female politicians and agentic traits—thus the opposite direction than expected. Here, women were more often
described with positive agentic qualities like ambitiousness (χ2 = 9.22, p = .01), assertiveness (χ2 = 17.68, p < .001), competitiveness (χ2 = 10.65, p = .005), independence (χ2 = 17.45, p < .001) or outspokenness/confidence (χ2 = 10.71, p = .005). Therefore, H1a has to be rejected.
Moreover, there was a significant difference in negative communal traits (t(254.56)= -2.6, p = .01) with women again scoring higher (M = .1, SD = .36) than men (M = .04, SD = .2). Thus, women were more often described with a lack of communal traits, such as affection (χ2 = 6.78, p = .034) and gentleness (χ2 = 10.94, p = .007). Contrary, there was no significant difference in positive communal trait mentions (t(308.14)=-1, p = .32). Although there is only a significant difference in negative communal traits, this can nevertheless point to an increased focus on stereotypical female traits, or the lack thereof in female politicians. Consequently, H1b has to be
partially accepted.
A consequent independent samples t-test was conducted to test for an increased mention of conflict for male politicians (H2). Neither a significant difference in conflict mentions between
7 As the Levene-Test for Equality of Variance was significant for each trait scale, all results are reported assuming no equality of variance.
THE ODD ONE OUT
male and female politicians (t(574)=1.09, p = .28), nor a significant difference in harmony mentions (t(574)= .32, p = .75) could be found. Thus, the hypothesis has to be rejected.
A Chi-square test for association8 was conducted to assess whether different policy topics were more often mentioned in relation with male or female politicians (H3). A significant
difference was found for the topics National Politics (χ2 = 7.17, p = .007), Agriculture and Fishing (χ2 = 4.57, p = .03) and Scottish Referendum (χ2 = 9.34, p = .002), with female political
actors scoring higher on all three topics. Due to the lack of clear gender association with either topic, we cannot assume that the increased mention for female politicians is due to a gender bias, but rather we have to assume another influencing factor such as diverging expertise or party focus9. As neither an increased focus on male-coded policy topics for male politicians, nor an increased focus on female-coded policy topics for female politician could be found, H3a and H3b
have to be rejected. Likewise, H4 and H5 have to be rejected as no significant association between
gender and mention of family or appearance was found.
H6 predicted an increased use of linguistic sexism for female politicians. However, no
significant difference on hierarchical naming and use of female gendered titles could be
established. While, there was a significant association between male/neutral gendered titles and gender (χ2 = 4.63, p = .03) these titles were more often used for women (54.9%) than men
(45.5%). The use of male gendered titles also differed significantly across newspapers (χ2 = 8.29, p = .02) with the Daily Telegraph using more male gendered titles (57.5%) than the Sun (44.4%) or the Guardian (44.7%). However, only in the Guardian did the use of male gendered titles for
8 All results for the Chi-square test not mentioned in the text are reported in the Appendix A, Table 1-5.
9 E.g., the majority of political actors from the Scottish National Party (SNP) in the sample are female (67.7%) thus possibly leading to a higher mention of the topic Scottish referendum for female political actors compared to male political actors.
THE ODD ONE OUT
women (54.4%, χ2 = 3.03, p = .08) approach significance. As no overall increased linguistic sexism in the coverage of female political actors could be established, H6 has to be rejected.
Largely, the data of this study was not able to replicate previous findings. Particularly, the significant association between female politicians and positive agentic traits was contrary to the expectations of a focus on gender-conforming traits in politicians. During the analysis, another novel finding emerged, that could additionally influence the portrayal of female politicians through the news media and explain the increased focus on agentic traits in women: party ideology. While female politicians of conservative parties have been characterized using
androcentric traits, the opposite appears to be the case for female Labour politicians. Thus, as the
Daily Telegraph praises “Theresa May's strong and stable style” (Leaper, 2017, p.20f.) or the Guardian points out that May has “adopted a hard line that has effectively put Ukip out of
business” (O’Hagan, 2017), Labour politicians are characterized as explicitly less agentic, with the Sun proclaiming Labour shadow secretary Diane Abbott as “guess[ing] at the annual cost” (“Labour is a Slapstick act”, 2017) or the Daily Telegraph accusing her of “floundering on details”. (Hughes, 2017) or the Guardian pointing to the “emollient tone” of Shadow Secretary Rebecca Long-Bailey (Boffey & Stewart, 2017). There seems to be a different perception of conservative female politicians that is not influenced by newspaper ideology.
Likewise, this focus on agentic traits in female conservative politicians also seemed to emerge in the coverage of foreign conservative politicians, such as the German chancellor Angela Merkel. Merkel, for example, was described as “a mistress of the put-down”, and was explicitly compared to May in that she is also “known to carefully deliberate and weigh up the facts before she pounces” by the Daily Telegraph (Qvortrup, 2017). The similarity in their media portrayal becomes most clear in the frequent comparison of both politicians to the Iron Lady (Qvortrup,
THE ODD ONE OUT
2017, Rayner & Swinford, 2017, Hyde, 2017), both as successors of Margaret Thatcher and in a more metaphorical manner.
To assess this perception, party affiliation for all political actors was added to the data. Overall, there were 119 female conservative and 35 female labour political actor mentions. An independent sample t-test was conducted to assess the difference for agentic and communal traits in female conservative and labour politicians.10 Overall, female conservative politicians scored higher on positive agentic traits (M = .52, SD = 1.07) than female labour politicians (M = .29, SD = .57; t(107.42) = 1.7, p = .0911). Contrarily, female labour politicians scored higher on negative agentic traits (M = .31, SD = .76) than female conservative politicians (M = .2, SD = .56; t(152) = -.96, p = .34). While the differences in traits were not significant, this can most likely be
attributed to the small number of female labour politicians (n = 35).
Previous researchers (Dolan, 2014; Hayes, 2011; King & Matland; 2003, Meeks & Domke, 2016; Schneider & Bos, 2014) have established a significant interaction effect between party affiliation and gender on the evaluation of political candidates in the US, whereby party-affiliation heuristics eschew gender stereotypes. Here, Winter (2010) finds that the conservative Republicans are more associated with masculine traits, while the more liberal Democrats are more likely associated with feminine traits.
Following, politicians are not only subjected to expectations that arise from the gender stereotypes, but are also influenced by stereotypes and expectations that arise from their party affiliation. Both gender and party create requirements for the female politician that have to be fulfilled through role-conforming behaviour or will be punished with a negative perception
10 All means and standard deviations are reported in Appendix 1, table 6.
11 Due to the significant Levene test of variance equality (p =.02), results for positive agentic traits are reported assuming no equality of variance.
THE ODD ONE OUT
through the public. Thus, we have to question how party and gender expectations interact or diverge to create role expectations.
This emergent finding fed directly into the design of Study 2, which was designed to pursue this party-gender interaction, and to focus on its possible effects on respondents’ attitudes towards the politicians in question. In particular, the results of the content analysis, combined with previous research, led to the following expectations:
H9a: Female labour politicians who are depicted as not conforming to gender stereotypes will be perceived as less likeable than female labour politicians who are depicted as such.
H9b: Female conservative politicians who are depicted as not conforming to gender stereotypes will be perceived as equally likeable as female conservative politicians who are depicted as such.
H10: Female conservative politicians who are depicted as not conforming to gender stereotypes will be perceived as more likeable than female labour politicians who are depicted as such.
Study Two
To assess the effect of gender and party-conforming behaviour of female politicians in news coverage on the perception of likeability and competence, an online experiment using a 2 (between-subjects: gender conforming vs. gender non-conforming) x 2 (between subjects: labour vs. conservative) design was conducted through an online survey on the platform Qualtrics. Data was collected between May 24th and June 15th 2018. Participants were recruited from the student pool of the University of Amsterdam, through social media and using the online survey platforms
Survey Circle and Poll-Pool. Overall, 223 individuals participated. 20 participants were excluded
from the analysis due to excessive missing values resulting in N = 203. Table 3 depicts the experimental groups and the number of participants for each group.
THE ODD ONE OUT
Table 3 Experimental Design and Number of Participants in Brackets
Experimental Factor: Gender-conforming Experimental Factor: Party-Affiliation non-conforming conforming Conservative 1 (47) 3 (51) Labour 2 (51) 4 (54)
Variables and Operationalization
The independent variables applied to the depiction of female politicians were gender-conforming behaviour and party affiliation. Both variables were operationalized within a newspaper article that focused on a debate between a male and female politician over land-clearing laws in Australia.12
Based on the findings in study one, the focus of gender (non)-conforming behaviour was placed on agentic traits—those traits related to assertiveness, confidence and independence. Accordingly, female politicians in the non-gender conforming condition were presented as being very assertive and aggressive by focusing on combative behaviour. On the contrary, female politicians in the gender-conforming condition were depicted as more reserved and measured in their behaviour and as looking for compromise. For example, in the gender non-conforming condition the female politician is described as “aggressively attacking perceived policy
loopholes” and as “combat-ready” towards further policy change, while the female politician in the gender conforming condition merely “points out some perceived policy loopholes” and as “ready to re-evaluate her position” when facing further policy change. Party affiliation was
THE ODD ONE OUT
operationalized by presenting the female politician as either belonging to the conservative or the labour party. The male politician in the article was presented as being from the opposing party.
The dependent variables were competence and likeability; these were measured for both the female and the male politician. To assess likeability, four items13 were chosen from Reysen’s (2005) likeability scale and presented in a random order. Participants were asked to indicated on a 7-point Likert scale from “Strongly disagree” (1) to “Strongly agree” (7) if each politician was friendly/ likeable/ warm/ approachable. The scores on all 4 items were averaged to create a likeability scale, which showed excellent reliability (Cronbach’s ⍺ = .83). On average
participants rated the political actors M = 3.9 (SD = .62), with the male politician scoring higher (M = 4.07, SD = .89), than the female politician (M = 3.73, SD = 1.11).
Competence was assessed through four items chosen from a list of items proposed by Blais and Thompson (200914). Again, participants were asked to indicate on a 7-point Likert scale from “Strongly disagree” (1) to “Strongly agree” (7) if each politician was competent/ knows what they are doing/ is good at solving problems/ is highly skilled. On average participants rated the political actors M = 4.3 (SD = .61), with the male politician scoring higher (M = 4.35, SD = .82), than the female politician (M = 4.2, SD = 1.06). Again, the scale showed good reliability (Cronbach’s ⍺ = .79).
In order to verify that participants have perceived the stimuli in the intended way, the following manipulation check was employed. Participants were asked to indicate the presence of four items measuring agentic traits on a scale from 0 (not at all) to 10 (completely) (Cronbach’s ⍺ = .78). Participants in the gender non-conforming conditions scored the female politicians higher
13 Reysen’s (2005) scale consists of 11 items measuring likeability of which the four most suitable items were chosen. As established by Bakker & Leikes (2016) such a short scale is equally well equipped to measure the construct as a longer scale of all items.
14 Blais and Thompson (2009) present a scale of 21 trust in leadership items of which seven items are focused on competence. Of these items again the four most applicable were chosen to form a short scale, which can be considered as reliable to measure the construct as a longer scale (Bakker & Leikes, 2016).
THE ODD ONE OUT
(M = 7.13, SD = 1.8), than those in the gender conforming conditions (M = 5.3, SD = 2.07). This difference was significant (t(201)=6.65, p < .001). Additionally, participants were asked to recall the political party of the female politician out of a list of options. 60% of participants15 correctly indicated the political party in their experimental group.
To control for possible confounding effects, participants were further asked to indicate their interest in politics (M = 3.33, SD = 1.16) and their trust in politicians (M = 2.24, SD = .87) on 5-point scales. Additionally, political cynicism was measured using three items on a 5-point scale (Cronbach’s ⍺ = .57). Participants showed an average level of political cynicism (M = 2.5, SD = .72). Participants were asked to indicate their political orientation on two 10-point scales from left (0) to right (10) and conservative (0) to progressive (10). Both items were significantly correlated (r = -.59, p < .001). After inverting the conservative-progressive scale, both items were combined to form a political orientation measure on which participants scored on average M = 3.24 (SD =1.75, Cronbach’s ⍺ = .72) indicating a more left and progressive leaning.
Lastly, participants were asked to indicate their gender, country of residence, age and education level. 69.8% of participants were female (n = 141) with an average age of 26 (SD = 8.33). The majority of participants were college graduates (37.4%) or post-graduates (31%). 1.5% of participants had not completed high school. 83.5% of participants were from Europe of which the majority of participants resided in the Netherlands (41%), Germany (19.5%) or the UK (13.5%). Of the non-EU participants 5.5% were from the US, followed by Australia (3.5%)16 and Canada (2%).
15 To ensure the reliability of the study, all analyses have also been conducted excluding participants who failed the recall. These results will be presented in the Appendix A, table 12 – 16. No difference in results was found.
16 As the experimental stimulus was focused on politics in Australia and independent t-test was conducted to assess if respondents from Australia perceived the male and female politicians significantly different across conditions. There was a significant difference for perceived likeability (t(49) = -3.3, p = .002) and competence (t(49)= -2.1, p = .041) for the female politician in the experimental condition 2 (gender non-conforming/labour). Nevertheless, due to the
THE ODD ONE OUT
Participants were told that the purpose of the survey was to explore how they read political news. After giving their informed consent, participants were randomly assigned to one of four articles, corresponding with the four experimental groups and instructed to read the article carefully. Once participants proceeded, they were asked to indicate the likeability and
competence for each politician. Subsequently, the manipulation checks were presented.
Eventually, participants were asked about their demographics, followed by questions regarding their political interest, trust and cynicism. On the final page, participants were debriefed and thanked for their cooperation.
Results
Table 4 displays the means and standard deviations of each experimental group for the dependent variables likeability and competence for the male and female politician.
Table 4 Means and Standard Deviations by gender of politician
Source Con./Non (1) Labour/Non. (2) Con./Con. (3) Labour/Con. (4)
M (SD) M(SD) M(SD) M(SD) Female Likeability 3.34 (1.08) 3.52 (1.07) 3.93 (1.06) 4.1 (1.09) Competence 4.12 (.86) 4.02 (.92) 4.21 (.84) 3.95 (.97) Male Likeability 4.15 (.92) 4.36 (1.15) 4.13 (1.09) 4.18 (1.08) Competence 4.32 (.73) 4.2 (.84) 4.47 (.85) 4.42 (.86)
small sample (n=2) and the lack of systematic difference across conditions, participants from Australia were included in the analysis.
THE ODD ONE OUT
Randomization checks 17were unable to detect significant differences across the groups regarding the control variables age, gender and education. Thus, we can conclude that
participants were randomly divided.
H7 proposed that female politicians who are depicted as conforming to gender stereotypes will be perceived as more likeable, while H8 assumed that female politicians who are depicted as non-conforming to gender stereotypes will be perceived as more competent. To assess the difference in likeability and competence participants of all conditions were combined into gender-conforming/ non-conforming groups irrelevant of party affiliation. Table 5 records the Means and Standard Deviations for each group
Table 5 Means and Standard Deviations for Gender Conforming/ Non-Conforming Groups
Source Gender Conforming Gender Non-Conforming
M (SD) M (SD) Female Likeability 4.01 (1.07) 3.4 (1.07) Competence 4.07 (.86) 4.26 (1.04) Male Likeability 4.07 (.91) 4.06 (.89) Competence 4.44 (.84) 4.26 (.79) Female Labour Likeability 4.09 (1.09) 3.5 (1.07) Competence 4.18 (1.08) 4.36 (1.14) Female Conservative Likeability 3.93 (1.06) 3.34 (1.08) Competence 4.13 (1.09) 4.15 (.92)
A independent samples t-test found, that there was a significant difference in perceived likeability between the two conditions (t(201) = -3.87, p < .001) with female politician in the
THE ODD ONE OUT
gender-conforming condition being perceived as more likeable (M = 4.01, SD = 1.07), than those in the gender non-conforming condition (M = 3.4, SD =1.07). Consequently, we can accept H7. Contrarily, there was no significant difference in perceived competence (t(201) = .7, p = .48), leading to a rejection of H8.
The results of study one led to the assumption that female labour politicians who are depicted as gender non-conforming will be perceived as less likeable than female labour
politician who are depicted as gender conforming (H9a), while there would be no such difference for female conservative politicians (H9b). An independent samples t-test found a significant difference in likeability (t(103) = -2.73, p = .008) between gender-conforming female labour politicians (M = 4.09, SD =1.09) and those that were not gender conforming (M = 3.5, SD =1.07) leading us to accept H9a. However, the difference in likeability between gender-conforming female conservative politicians (M = 3.93, SD =1.06) and those that were not gender conforming (M = 3.34, SD =1.08) was also significant (t(96) = -2.75, p = .007) leading to a rejection of H8.
H10 proposed that female conservative politicians who are not conforming to gender stereotypes would be perceived as more likeable than female labour politicians who are not gender conforming. However, no significant difference (t(96) = -.85, p = .4) could be established, Following, we have to equally reject H10.
Additionally, an Analysis of Variance 18(ANOVA) was conducted across all four experimental conditions for the perceived likeability and competence of the female politician. There was a significant difference in likeability (F(3, 202) = 5.43, p = .001) between the groups but not on perceived competence (F(3, 202) = .59, p = .67).
Using Tukey’s honest significant difference test, differences between the groups were established. Figure 1 represents the differences in mean likeability by gender conformity and
THE ODD ONE OUT
party affiliation. Female politicians who are conservative and gender non-conforming were perceived as less likeable (M = 3.34, SD = 1.08, MDiff = -.59, 95% CI [-1.16; -.03], p = .03) than
female politicians who are conservative and gender conforming (M = 3.93, SD =1.06). Likewise female politicians who are conservative and gender non-conforming were perceived as less likeable (M = 3.34, SD = 1.08, MDiff = -.76, 95% CI [-1.32; -.2], p = .003) than female labour
politicians who are gender conforming (M = 4.1, SD =1.09). Female labour politicians who are gender non-conforming were perceived as less likeable (M = 3.52, SD = 1.07, MDiff = -.58, 95%
CI [-1.12; -.03], p = .033) than female labour politicians who are gender conforming (M = 4.1, SD =1.09). Results point towards a significant effect of gender conformance but not of party conformance on likeability.
Figure 1 Bar Graph for Mean Likeability for the Female Politician across by Gender conformity and Party affiliation To further test for the interaction between gender conformity and party conformity on likeability, experimental groups were additionally combined based on party affiliation. Age,
4,01 4,09 3,93 3,4 3,5 3,34 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5
Female Labour Cons. Female Labour Cons.
Conforming Non-Conforming Me an ( L ike abi li ty)
THE ODD ONE OUT
gender, education, political cynicism, political trust and political orientation were included as control variables. All results of the two-way ANOVA are reported in table 6.
Table 6 Two-way ANOVA results for Party and Gender Conformance on Likeability in Female Politicians with control variables
Source Sum of Sqaures
(Type III) df Mean Square F Party 1.67 1 1.67 1.49 Gender 14.54 1 14.54 12.96* Party x Gender .03 1 .03 .02 Age 33.06 32 1.03 .92 Sex19 7.09 1 3.55 3.16* Education 0 1 0 0 Political Orientation 4.5 1 4.5 4* Political Cynicism .99 1 .99 .89 Political Trust .004 1 .004 .004 Political Interest .58 1 .58 .52 Error 178.31 159 1.12
Note: R2 = .28, adj. R2 = .09; All significance tests were conducted at an alpha-level of .05. p* < .05
The analysis found a significant effect of gender conformity (F(1, 202) = 14.54, p < .001) on the perceived likeability of female politicians, even when controlling for socio-economic variables and political attitudes. Political orientation (F(1, 202) = 4.5, p = .047) and gender of
THE ODD ONE OUT
participants (F(2, 202) = 3.16, p = .045) were also a significant predictor of likeability with men perceiving the female politician as significantly more likeable (M = 4, SD = 1.03, t(197) = 2.42, p = .017) than women (M = 3.59, SD = 1.1). Again, no effect of party conformity on likeability could be established. A second two-way ANOVA20 was conducted to assess the interaction of gender and party conformity on competence of female politicians. No significant predictor of competence in female politicians could be established.
Concluding, study two was not able to establish an interaction of gender and party conformity as proposed by study one but found that gender stereotypical behaviour in female politicians is a significant predictor of likeability irrelevant of party.
Discussion
This study examined how the media portrays female politicians, particularly how the female identity is constructed and how this portrayal affects likeability and competence. While female politicians have become a fixture of the political arena, it is still predominantly male (Meeks, 2013a; Lünenborg & Maier, 2015; Rosenwasser & Dean, 1989). Such male dominance was also seen in this study: despite a sitting female UK prime minister, only 33.6% of political actors coded in UK news coverage were female, further underlining the existing disparity between the genders. Consequently, the analysis of female political representation is even more essential to reveal biases that prohibit an equal representation of women in politics.
Two studies were conducted to explore these dynamics—a content analysis of news coverage during the 2017 UK election period, and an experiment examining the effects of coverage on attitudes towards hypothetical male and female politicians. While the analysis of political news media coverage during the 2017 UK election period was not able to replicate previous research on gender stereotypical coverage in female politicians, it did offer new insights
THE ODD ONE OUT
into how the female identity is constructed by the media. Contrarily to previous expectations, female politicians were significantly more associated with both agentic traits - e.g. assertiveness and confidence, and the lack thereof. Likewise, there was a significant association between female politician and the lack of communal traits—e.g., affection and gentleness. The female politicians in this study were portrayed as explicitly not fulfilling feminine gender stereotypical traits, thereby distancing them from their femininity, while also not completely fulfilling the masculine ideal of their role as politician. That is, women seem to fit into neither the role of woman, nor the role of politician. Consistent with earlier research (Githens & Prestage, 1977, Schneider & Bos, 2014), female politicians seem to constitute their own subtype of women— neither feminine enough nor too powerful that they are considered masculine. Whereas news coverage about male politicians uses stereotypes similar to news coverage of non-politician men, media coverage of female politicians does not share the same stereotypes as that of non-politician women. This indicates that there still seems to exist a dissonance between the female identity and positions of power that can only be rectified by explicitly rejecting femininity in female
politicians. Likewise, study one was not able to find attributes, such as a focus on family, appearance or harmonic or communal behaviour that is expected of non-politician women (Devitt, 1999; Muir, 2005).
In line with previous research on the evaluation of US candidates (Dolan, 2014; Meeks & Domke, 2016; Schneider & Bos, 2014) party affiliation also seemed to influence the portrayal of female politicians in the UK, leading to an increased focus of agentic traits in conservative female politicians. Thus, the assumption was made that conservative female politicians will be afforded more leniency when engaging in agentic behaviour as it is expected from conservative politicians. Following, conservative female politicians were expected to be perceived as likeable despite not depicting dominant gender stereotypes. Study two was designed to assess this
THE ODD ONE OUT
association between gender conforming and party conforming behaviour on the perceived likeability and competence of female politicians.
Study two could not establish an association between gender and party conformity, nor an association of either with perceived competence in female politicians. However, the female politician was continuously perceived as less likeable when not confirming to female gender stereotypes with gender conformity being a significant predictor of likeability in female politicians. It was women, rather than men, that perceived the female politician as overall less likeable, indicating dominant gender stereotypes are not restricted to one gender, but accepted across the population. Additionally, participants continuously indicated that the female politician was more aggressive 21both, when conforming and not conforming to female gender stereotypes. Although female politicians do not share stereotypes with women, they are still perceived as less likeable when displaying non-gender conforming behaviour through agentic traits. Female politicians are not only losing on male stereotypes, but also do not gain any advantage from female stereotypes (Schneider & Bos, 2014).
The double-bind (Dabbous & Ladley, 2010; Hall & Donaghue, 2013) between
competence and likeability for female politicians might rather be a single bind by which female politicians are unable to fulfil the vague and unclear expectations society places on them. Thus, while they loose likeability when not conforming to gender stereotypes, they do not gain any competence. Here, female politicians can neither be too female nor too male. The continued devaluation of individual female politicians for both their femininity, e.g. Nancy Pelosi (Dabbous & Ladley, 2010) or Sarah Palin (Heflick & Goldenberg, 2009) and their lack thereof, e.g. Hillary
21 When asked to directly compare the aggressiveness exhibited by the female politician and the male politician on a scale from 0 (female politician) to 10 (male politician), participants rated the gender-conforming politician (M = 4.67, SD = 2.36) and the non gender conforming politician (M = 2.7, SD = 2.28) as more aggressive than the male politician.
THE ODD ONE OUT
Clinton (Meeks, 2013; Uscinski & Goren, 2010) or Angela Merkel (Lünenborg & Maier, 2015) underlines the difficulty for women of defining themselves in the political arena.
Limitations and Future Research
As voters use stereotype heuristics (McDermott, 1997; Van Zoonen, 1998) to assess the viability of political actors, we have to define what those heuristics are for female politicians. If female politicians are neither considered according to female stereotypes, no according to male, then what defines their identity vis-à-vis society? As Schneider and Bos (2014) have made a first step in re-assessing stereotyping for female politicians and defining their role as a new subgroup that exists independent of stereotypes applied to women, we have to broaden our research on gender stereotypes. Consequently, we cannot assume that there is one female gender identity. As Butler (2011) rejects the idea of a universal femininity, we have to consider other factors that shape and enact the female identity. Thus, a focus on interaction of sex, gender and race could give bigger insights on how different female politicians are portrayed and perceived in turn (Livingston, Rosette& Washington, 2012; McGinley, 2008). This however can only be achieved by regarding the portrayal of female politicians outside of WEIRD – western, educated,
industrialized, rich and democratic - populations (Henrich, Heine & Norenzayan 2010). Focusing on gender identity outside of a western context could give clearer input on the struggle and strive of female politicians in Muslim or non-Western countries that use a different value system to define the role as woman and politicians, thereby further adding to the growing research on intersectional gender identities.
Likewise, this study has applied one conception of stereotypical female and male traits. As Schneider and Bos (2014) show this conception can however vary significantly between studies. We thus, have to consider the results within the boundaries of this trait conception. Many studies on gender identities in female politicians neglect to analyse underlying concepts and the
THE ODD ONE OUT
explanation of dominant gender paradigms that shape identities (Lünenborg & Maier, 2015). While this study has tried to assess gender as a multi-faceted construct in a larger sample in a systematic way, it has in no way achieved the depth of a qualitative content analysis. Hence, as issues of gender and asymmetric power relations become subtler and intertwined, we have to deliberate the social context that has shaped and restricted identity construction (Lazar, 2007). Qualitative content analysis has to uncover those nuances of gender that are not captured with statistical analysis but rather become evident only through a detailed interpretation of the text at hand to reconsider the role of female politicians independent of stereotypes applied to non-politician women.
This study has proposed that party affiliation might be an equally strong heuristic used to assess female politicians. Nevertheless, only 60% of participants correctly recalled the party cue provided in study two. While results remained constant22, this could indicate that participants either were unable to perceive the party cue correctly, or due to the diversity of the participants from 25 different countries, not all respondents understood the parties represented in the same way. While most labour/socialist and conservative parties across Europe share a similar ideology (Bakker et al., 2015), this cannot be said for labour/socialist and conservative parties
internationally. Particularly, the US does not share the same ideological perceptions of the political parties. Consequently, while this study did not find a significant interaction of party conformity, we cannot dismiss it as a predictor of likeability and competence of female
politicians. Future studies should focus on parties in a more homogenous context such as within one country to adequately assess if party heuristics play the same role in, a e.g. European context as they do in the US and what expectations and restrictions they place on female politicians.
22 As indicated by the analysis conducted using only participants that correctly answered the party cue question. The results are reported in Appendix A and do not differ from the results oft he analysis with all participants.
THE ODD ONE OUT
In communication science we often focus on how individuals or groups are portrayed and received. This however can, diminish the role female politicians play in shaping the discourse surrounding them (Foucault, 1982). While we have to consider how female politicians are perceived and how their identity is created in different media systems, only by additionally analysing how female politicians themselves define their role as women and politicians, can we positively contribute to the definition of identity for female politicians. As many female
politicians still struggle to reconcile their femininity with their role in the political system, the media can play an important role in either including or alienating women from the political arena.
This study has contributed to the growing body of research on female politician, by applying a mixed method approach for a systematic analysis of the portrayal of female politicians outside of the dominant US context. By revealing, that female politician face an unclear role conception, arising from the persistent dissonance between femininity and power, it has shown the continued alienation of women from the political system. Future research has to set in to define this unclear identity and to further eliminate bias towards women in politics, paving the way for equal representation and facilitating women’s empowerment. While female politicians do not face the same hostility as Belva Lockwood, they still have to shape their space in politics.
THE ODD ONE OUT
References
Adcock, C. (2010). The Politician, the Wife, the Citizen, and her Newspaper: Rethinking Women, Democracy, and Media (Ted) Representation. Feminist Media Studies, 10(2), 135-159.
Adoni, H., & Mane, S. (1984). Media and the Social Construction of Reality Toward an Integration of Theory and Research. Communication Research, 11(3), 323-340.
Aelst, P. V., Sehata, A., & Dalen, A. V. (2010). Members Of Parliament: Equal Competitors for Media Attention? An Analysis of Personal Contacts Between Mps and Political
Journalists in Five European Countries. Political communication, 27(3), 310-325. Alcoff, L. (1988). Cultural Feminism versus Post-Structuralism: The Identity Crisis in Feminist
Theory. Signs, 13(3), 405-436.
Baird, J. (2004). Media Tarts: How the Australian Press Frames Female Politicians. Melbourne, Australia: Scribe Publications Pty Limited.
Bakker, B. N., & Lelkes, Y. (2016). Selling Ourselves Dhort? How Abbreviated Measures of Personality Change the Way We Think About Personality and Politics. Unpublished manuscript), University of Amsterdam.
Bakker, R., De Vries, C., Edwards, E., Hooghe, L., Jolly, S., Marks, G. & Vachudova, M. A. (2015). Measuring Party Positions in Europe: The Chapel Hill Expert Survey Trend File, 1999–2010. Party Politics, 21(1), 143-152.
Banwart, M. C., Bystrom, D. G., & Robertson, T. (2003). From the Primary to the General Election: A Comparative Analysis of Candidate Media Coverage in Mixed-Gender 2000 Races for Governor and US Senate. American Behavioral Scientist, 46(5), 658-676.
