Abstract
This study discusses variations in the media coverage of the most prominent male and female EP candidates from Romania in the four weeks leading up to Election Day (May 8 to June 7, 2009). The verbal- and visual-framing analyses conducted focus on the visibility of these candidates, their viability (horse-race frame), and the balance between issue-related coverage (issue frame) and gendered coverage (trivialization frame). Our sample encompasses 249 news stories from the web sites of the most influential broadsheets and tabloids in Romania, namely, Cancan, Evenimentul Zilei, Gândul, and Libertatea. The results point toward a gender bias on the part of media. Whereas women dominated the tabloid outlets, men were featured prominently in the broadsheets. The trivialization and the issue frames appeared more often for female candidates, whereas the results for the horse-race frame were mixed. While factors other than gender (such as experience) might have influenced the coverage of these particular candidates, it is still too early to proclaim equal treatment in the media, especially since women were considered unable to deal with the issues they were linked to.
A vast body of research suggests that the media coverage of female candidates in electoral races differs from the coverage of their male counterparts with regard to the quality and quantity of the coverage (Kahn and Goldenberg 1991). As the literature review will show, this mostly involves their visibility, the viability assessments related to them, and the balance between issue-related and traits coverage. Framing theory is one very prominent theoretical lens through which such variations in coverage have been analyzed. Framing is an umbrella term under which various communication scholars have scrutinized the practice of selection, emphasis, and presentation to an audience of only a fraction of the available information (Entman 1993).
Most interaction between candidates and citizens nowadays “occurs in living rooms, watching television, reading the newspaper or surfing the Internet” (Heith 2001: 335). Since voters’ opinions of those running for office and the voting behavior of both laymen and pundits can be affected by election news coverage (Bartels 1993; Bystrom et al. 2001; Carlin and Winfrey 2009; Gidengil and Everitt 2003; Kahn 1994a), including visuals (Barrett and Barrington 2005b; Gilliam and Iyengar 2000; Graber 1987; Palvio 1991; Rosenberg et al. 1986), knowledge about the potential biases in this coverage is crucial. In the context of this debate, the research question we want to pursue is whether there is a gender bias in the news coverage of Romanian candidates vying for the European Parliament.
To answer this research question, we conduct a verbal and visual-framing analysis of the media coverage of the four most prominent male and female candidates in the four weeks leading up to Election Day (June 7, 2009). According to ActiveWatch, the Romanian Media Monitoring Agency, these are Elena Băsescu, George Becali, Adrian Severin, and Monica Macovei. Adrian Severin (born 1954) of the Social Democratic Party (center-left) and Monica Macovei (born 1959) of the Democratic Liberal Party (center-right) are experienced politicians. Elena Băsescu (born 1980) of the Democratic Liberal Party (center-right) and George Becali (born 1958), who made his bid on the list of the Greater Romania Party (radical right-wing, ultra-nationalist), have turned to politics after pursuing careers in modeling and business, respectively.
Our scrutiny of gender bias involves analysis in terms of visibility, viability, and the balance between issue-related coverage and traits coverage. We consider this to be a valuable endeavor for four reasons. First, because while the literature on European politics in general and European elections in particular has grown vastly over the past decade, the issue of unequal treatment of male and female candidates in the media remains rather unexplored in this context. 1 Even when similar research questions are posed, those studies typically assess the written or spoken words of news reports, completely ignoring visual content. Thus, the second strength of this article is its focus on both verbal and visual elements in news coverage. Third, by analyzing not only broadsheets but also tabloids, we gain access to the “media diet” of a wide audience. Last, by addressing this question in a country not extensively analyzed in the literature, we contribute to increasing knowledge in the field of mediated political communication in emerging democracies.
The discussion of gender, press, and politics is particularly interesting in the Romanian context. This is because of the ways in which expectations surrounding womanhood have shifted in the past century, making a full circle from homemakers, through wage earners, to homemakers again (Băban 2000; Woodcock 2007). After adding the patriarchal value system in Romania to the equation, one would not expect politics to be a female-friendly environment. In fact, less than 10 percent of the members of the present-day Romanian Parliament are women (Romanian Ministry of Labour, n.d.). On the other hand, 36 percent of the Romanian Members of the European Parliament (MEPs) currently in office are women.
Theoretical Background: Framing Theory
Framing can be understood as the process of drawing a cropping frame around a picture. While the elements within the frame are focused on, competing, distracting, or contradictory elements located outside the border are marked off. In the case of research on election coverage, framing refers to the practice of emphasizing a certain standpoint on a given person, issue, or event, and downplaying others. Thus, a frame “simplifies and condenses the ‘world out there’” (Snow and Benford 1988: 137). Frames are often classified into “generic” and “issue-specific”: While generic frames apply to a wide array of topics, issue-specific frames only apply to specific issues (Semetko and Valkenburg 2000). When the issue at hand is of interest, scholars analyze the elements of such frames as problem definition, causal interpretation, moral evaluation, and treatment recommendation (Entman 1993). When examining wider matters, applicable to an array of issues—this is arguably the case for gender studies—scholars often decide to focus on generic frames. News frames become noticeable in the choice of issues covered or the lack thereof, the employed language, and the accompanying visuals (Entman 1993; Gamson and Lasch 1983; Gamson and Modigliani 1989; Pan and Kosicki 1993; Tankard 2001; Van Gorp 2007).
After a frame has been presented repeatedly to an audience (and has thus become salient), some elements—for example, particular figures of speech or visuals—suffice for the activation of an entire known frame or one particular interpretation of the received information (Gamson and Lasch 1983; Gamson and Modigliani 1989).
With few exceptions (Dan and Ihlen 2011; Reynolds and Barnett 2003; Semetko and Boomgaarden 2007), most scholars analyze only verbal messages when conducting framing analyses. Yet visuals were held important even by key framing scholars like Goffman (1979); we too consider them to be essential with regard to the discussion of gender-balance in news coverage. This is due to the fact that visuals laced with gender stereotypes may occur in the media, but this type of stereotyping has a harder time passing newsroom scrutiny when communicated in words (Coleman 2010). Bearing in mind that visuals are more likely to be processed unconsciously, through people’s preexisting schemas 2 (Devine and Monteith 1999), the potential impact of visual stereotypes in the media is vast. Actually, visual framing makes stereotyping less obvious than verbal framing (Messaris and Abraham 2001) and is thus able to overcome the cognitive barriers that diminish the potential effects of verbal framing (see Callaghan 2005; Messaris and Abraham 2001; Petty and Cacioppo 1986 for details). Moreover, if the information conveyed visually is incongruent with the information conveyed verbally, viewers tend to believe and remember the visual message over the verbal one (Coleman 2010). Hence, an article showing a gendered picture, while the accompanying words make no reference to gender, is arguably more damaging than misogynistic words and a neutral photograph.
Despite the fact that the above account emphasizes the importance of imagery, focusing only on news visuals seems to be just as incomplete as focusing only on the verbal channel, for most mass media provide news consumers with written or spoken words and still or moving images. Thus, this research pays attention to both channels of communication in candidate framing.
The following section maps the research that has been conducted on the media coverage of politicians vying for office, allowing us to show the relation between this research agenda and framing research. What then follows is a detailed description of our methodology, before we move on to presenting and discussing our findings.
Media Coverage of Male and Female Candidates
Previous research shows important differences in the media coverage of male and female candidates. Tuchman’s (1978) seminal work on “symbolic annihilation,” in particular, has fueled research on the exclusion and trivialization of women in news.
Numerous empirical studies have arrived at the conclusion that female candidates often receive less (issue) coverage than their male counterparts (Aday and Devitt 2001; Banwart et al. 2003; Bystrom 2006; Devitt 2002; Fox 1997; Heldman et al. 2005; Kahn and Goldenberg 1991; Ross and Comrie 2012; for a different perspective, see Bystrom et al. 2001). Such variations in the quantity of male and female candidate coverage are important since voters will rarely support a candidate unknown to them (Goldenberg and Traugott 1984).
Moreover, previous investigations showed several gender-related variations in the quality of the media coverage. Their authors focused on one or more of the following three trends in campaign coverage: horse race, issues, and trivialization. These trends overlap with the concept of generic frames. Journalists’ decisions to cover elections in terms of issues or horse race are consequential: People attending issue frames tend to describe the election in terms of issues, while those attending strategy frames lean toward describing the election in strategic terms (Gamson 1992). We now briefly describe each of these three constructs.
Horse race or strategy frames refer to the positions of candidates in public opinion polls, their campaign resources, and their viability (Cappella and Jamieson 1997; Jamieson 1992; Kerbel 1997; Patterson 1994). Such frames are characterized by the language of competition, as known from coverage of wars and sports. Most studies agree on the fact that media often portray the candidacies of women as less viable than those of men (Aday and Devitt 2001; Bystrom 2006; Gidengil and Everett 2000; Heldman et al. 2005; Iyengar and Kinder 1987; Kahn 1994a; Kahn and Goldenberg 1991; see Bystrom et al. 2001 and Smith 1997 for claims to the contrary). This is conveyed not only through words, but also through visuals. For instance, Semetko and Boomgaarden (2007) have analyzed the electoral coverage in Germany in 2005 and found that Angela Merkel was portrayed more negatively than the male incumbent Gerhard Schröder, who was depicted as a “winner.” This is important because several studies—like the one from Barrett and Barrington (2005b)—demonstrated that subjects reading the same article viewed the candidate with a favorable photograph more positively than the candidate with an unfavorable photograph.
Frames referring to the candidates’ stands on public policy matters are often described as issue frames (Cappella and Jamieson 1997; Patterson 1994). Earlier studies have arrived at the conclusion that women often receive less issue coverage than men (Braden 1996; Bystrom 1999; Devitt 2002; Valenzuela and Correa 2009). Moreover, when journalists do address women’s issue positions, they tend to report on “female” issues like education more readily than on “male” issues such as the economy or gun control (Bystrom et al. 2001; Huddy and Terkildsen 1993; Smith 1997). This appears to be the case especially with female reporters (Kahn and Goldenberg 1991). Such coverage is perhaps why voters believe that a candidate’s gender is responsible for their dealing more effectively with certain issues (Bystrom et al. 2001; Kahn 1994a; Kahn and Goldenberg 1991; Major and Coleman 2008; Rosenwasser et al. 1987; Sapiro 1982; Semetko and Boomgaarden 2007).
A focus on style over substance can trivialize people and their positions (Lind and Salo 2002; Stevens 2007). In the context of elections, references to individual attributes—for example, personal appearance, clothes, size, personality, gender roles—or visual depictions thereof, point toward the trivialization frame (see also Davis 1982; Jamieson 1995; Jolliffe 1989; Lind and Salo 2002; Major and Coleman 2008; Trimble and Treiberg 2010; Turk 1987). Similar constructs are sometimes referred to in the literature as “sex-specific narrative frames” (Gidengil and Everitt 2003), personal frames (Valenzuela and Correa 2009), or gender frames (Boomgaarden and Semetko 2007). Several studies report more such coverage of women candidates than of their male counterparts (Aday and Devitt 2001; Bystrom 1999; Devitt 2002; Heldman et al. 2005; Kahn 1994b; Ross 2002; Valenzuela and Correa 2009). In addition, just as with issue frames, traits can be grouped into “male” or “female,” meaning they are consistently associated with men or women, respectively. Relying on research on sex stereotypes (Ashmore and del Boca 1979), Kahn and Goldenberg (1991) consider traits like independence and eagerness to be among the “male” traits, while dependency and compassion are among the “female” traits. Again, just as in the case of issue frames, research shows that the discussion of “female” traits appears prominently in coverage of female candidates, especially with female reporters (Braden 1996; Kahn and Goldenberg 1991).
Method
As previously mentioned, this study sets out to discuss whether there is a gender bias in the news coverage of Romanian candidates vying for the European Parliament. To this end, we employ a verbal- and visual-framing analysis of the media coverage of the four most prominent male and female candidates during the four weeks leading up to Election Day (June 7, 2009).
Our analysis is based on articles published on the web sites of the most influential broadsheets and tabloids in Romania: Cancan, Evenimentul Zilei (EvZ), Gândul, and Libertatea. We selected these outlets because they have the largest numbers of unique visitors in Romania 3 —as identified by SATI, the local bureau monitoring web traffic—and are targeted at different audiences. Evenimentul Zilei (EvZ) and Gândul are broadsheet newspapers, while Cancan and Libertatea are tabloids. Moreover, EvZ, Libertatea, and Cancan are considered to be center-right, whereas Gândul is perceived as center-left. We searched these web sites for occurrences of the names of either the selected candidates or their acronyms for the period May 8 to June 7, 2009. After excluding duplicates and irrelevant hits, our sample included 249 news accounts, of which 122 were illustrated. The unit of analysis was the individual article, composed of words and images. We coded the articles in which two candidates were mentioned (n = 31) twice, once for each of the portrayed candidates. This led us to 280 data entries in our sample.
This study employs a deductive approach by seeking to find frames that have been identified before in the literature. Yet, despite the richness of this literature, we decided against establishing the presence of a frame in a news report by means of single dichotomous categories (frame present/frame not present). Rather, we preferred to rely on earlier studies to develop several dichotomous questions for each frame, as shown in Table 1.
Frames and Framing Questions.
When analyzing the horse-race frame, we investigated each article with respect to the employed language and the viability assessments made, as well as the attention to campaign resources and standings in public opinion pools (Table 1). In addition to the dichotomous variables measuring the presence or absence of this type of information, we included a range of variables we considered relevant, as described in the next paragraphs.
Viability assessments can be made verbally and visually. Verbally, this referred to a candidate’s suitability to represent Romania in the European Parliament. To be precise, a candidate is considered to be unsuitable when she or he appears to be unprepared for future tasks, for instance due to lack of experience or understanding of politics. Visually, this refers to the favorableness of the accompanying photographs. We assessed the favorableness of a picture by basing individual coding decisions on quick impressions of each photograph, as we wanted to replicate the average reader’s reaction to pictures in the news. Here, we coded the photographs on a three-point scale, ranging from −1 (unfavorable) to 1 (favorable). In doing so, we implicitly took into account camera angles, facial expressions, and backgrounds. Favorable photographs showed the candidate smiling or looking confident, sometimes against a theatrical background. In one photograph judged to be favorable, a well-dressed George Becali is shown smiling from a low angle sitting in a chair with gold frame. Correspondingly, unfavorable photographs showed the candidate with a bewildered or angry look in their face, caught in an awkward physical position or frowning. For instance, a close-up of an absent-minded Elena Băsescu with a wooden smile and a bruise under her eye was coded as unfavorable. Finally, a typical neutral photograph in our sample was a simple head shot of a candidate with a blank expression on his or her face against a minimal background. While our way of assessing favorability might be criticized as vague, we produced high intercoder reliability. Bearing in mind that the size of photographs in newspapers can have an influence on readers’ perceptions (Wanta 1988), we also examined the pictures in terms of size, on a scale from 1 (very small) to 4 (large). Relying on Barrett and Barrington (2005a), we argue that two favorable small head shots of a candidate could be countered by one large and unfavorable photograph of the same candidate. Therefore, we generated a new variable by multiplying the favorableness scale by photograph size. The potential range of values for this variable was −4 (for a large unfavorable photograph) to 4 (for a large favorable one).
Besides simply noting the presence or absence of sports and/or war metaphors, we were also interested in analyzing which metaphors were used—for example, “defending oneself” or “attacking the opponent.” For this purpose, we included a string variable where we wrote down the metaphors used. Moreover, if campaign resources were mentioned in the article, we further wanted to know if they were described as low, medium, or high.
Within the section issue frame, we coded up to five issues per article, according to the level of emphasis on the issue in the story. Our codebook contained definitions of each issue included. For instance, “education” was operationalized as mention of any topic related to study migration, the Bologna agreement, quality of teaching, the need to create a well-educated workforce, and the need for more funding for higher education. When coding, we looked for each of these sub-issues and other related ones. For example, an article mentioning Elena Băsescu’s intention to support the legalization of marijuana if elected to the European Parliament was coded as “European legislation.” For the analysis, we distinguished between “male” and “female” issues. 4 According to the literature described above (Huddy and Terkildsen 1993; Kahn and Goldenberg 1991), the following issues were considered to be “male issues”: European legislation, economy, military/defense, justice, and energy policy. On the other hand, education, environment, health care, and social programs/issues were considered “female issues.” Furthermore, to account for issues we did not expect, we included a string variable. Yet, given the low prevalence of issues in our sample and the fact that we developed the codebook inductively, we barely ever made use of this variable.
At the same time, we were interested to find out if the candidate was described as able to deal with the issue at hand or not. Whenever positive statements that indicated confidence in the candidate’s ability to handle the issue were made by the reporter, the candidates themselves, or someone else (including the other candidates), we assigned a positive code. Sometimes, no evaluations regarding the ability of a candidate to handle the issue (or just as many positive and negative evaluations) were found in relation to the issues mentioned. This was coded as neutral/balanced. Finally, negative codes were assigned when statements offered an unfavorable impression of the candidate’s ability to handle the issue.
To determine whether the analyzed media have conveyed a trivialization frame, we looked for remarks on candidates’ appearance, character traits, economic status, gender roles, and the relationship between candidates’ gender and competencies. Moreover, we were interested to see whether it was suggested that politics was a male domain and whether the images portrayed the candidates in a trivial way. In the next paragraphs, we briefly describe how we operationalized each of these constituting constructs.
We considered that articles mentioned the candidate’s appearance when they discussed facial features, age, attire, hairstyle, hair color, weight, height, or size. As far as character traits are concerned, we included and defined the ones encountered in previous studies discussing the media coverage of politicians. 5 In addition, we included a string variable to account for character traits we did not anticipate. The included traits were rational, aggressive, (pro) active, ambitious, independent, dishonest, objective, sensitive, passive, noncompetitive, dependent, honest, and subjective (Braden 1996; Kahn and Goldenberg 1991; Maier and Tenscher 2006; Siune 1983). For the analysis, and in accordance with this literature, we distinguished between “male” and “female” traits. We treated sensitive, passive, noncompetitive, dependent, honest, and subjective as female traits and the rest as male traits.
We considered that the economic status of a candidate was mentioned when they were portrayed as poor, well off, or rich, either explicitly or implicitly (e.g., by describing the things they can afford). To find out if the article was presenting the candidate in a way that emphasizes their gender or gender role, we looked for references to their gender (male/female), gender expectations (say, provider), marital status (wife/husband), and family relationships (daughter/son; mother/father). To offer an example, we considered that an article presenting Elena Băsescu as a “daddy’s girl” was—among other things—emphasizing her gender role. Moreover, we assessed gender role visually. For instance, visual assessments of the gender role would portray Elena Băsescu together with her father.
When accounting for the relationship between the candidate’s gender and the competencies of the candidates, we anticipated situations such as someone questioning a woman’s availability for last-minute engagements due to her being a mother. The references to politics as a male domain had to be explicit, say by suggesting that women cannot cope with the stress. For instance, George Becali stated that politics is a male domain par excellence and that he would never allow his daughters to enter politics. 6
Finally, we examined the images to find out whether they presented the candidate in another role than the political–professional one. The favorableness of the picture (favorable, unfavorable, neutral) described earlier does not offer any information to assess triviality. The reason is that, while a picture might be aesthetically flattering, it can serve to undermine somebody’s political aspirations. For instance, we coded glamorous photos of Elena Băsescu from her past as a fashion model as both favorable and trivial.
While the first author developed the codebook inductively, the second author conducted the pretest, which allowed us to make further adjustments before the second author coded the material. Apart from some string variables, we mostly employed binary coding in our data collection. Thus, we used simple yes–no categories to measure the occurrence of frame-elements (and thereby frames) in the news. We chose this method due to its potential for high intercoder reliability. We are, however, aware of the risk related to measurement errors and the fact that correlations among such variables are lower than between ordinal or interval variables.
To test for intercoder reliability, and thus whether both authors obtain similar results using the same codebook, the first author coded a randomly selected subsample (10 percent). Due to the straightforward nature of the codebook and the fact that we developed it inductively and conducted a pretest, we obtained very good reliability coefficients. Holsti’s formula yielded an average intercoder reliability of .97. The test returned values of 1.00 for the formal variables and for the variables included in the section “issue frame.” For the rest of the variables—the ones conveying the horse-race and trivialization frames, and the photographs—we obtained coefficients between .99 and .85.
We used IBM SPSS Statistics 19.0 (Statistical Package for the Social Sciences) to analyze the data. The decision for dichotomous variables had implications for the method employed for the identification of frames. More precisely, this scaling did not allow for the factor analysis employed by others to this end. In addition, we had to dismiss a cluster analysis as counterproductive, since many articles in our sample contained more than one frame. Yet, according to the cluster approach, precisely these articles—otherwise of interest for the answer of our research question—were not assigned to any cluster. Thus, we decided to create an index from the variables assumed by previous research to belong to the same frame. We built two indices: one for the horse-race frame and one for the trivialization frame. For each of them, we relied on the described research to assign different weights to the constituting variables. 7 The higher the predicted importance for the latent constructs (aka the frame), the higher the score in our index. Since the issue frame was very straightforward (Does the story mention any issues? If yes, which ones?), we considered a positive answer to be equivalent with the presence of the issue issue frame.
For analysis of differences in news coverage, we used descriptive statistics, recording frequencies and cross-tabulations where appropriate. To test whether female candidates are more or less likely to be framed in a given fashion, we use paired-sample t -tests for comparing the two samples. 8
In the next section, we present the findings of our study. We start by offering an overview of our sample, and then move on to analyzing variations in quantity and quality of the coverage of male and female politicians.
Findings
As Table 2 shows, we analyzed mainly articles originating from highbrow outlets: EvZ and Gândul represent together more than 70 percent of the sample (n = 205). Among the tabloid outlets, it was especially Libertatea which ran stories on the candidates (21.8 percent, n = 61). Even though the amount of articles is fairly low, this is far from surprising. In fact, other scholars had noticed that EP election stories amount to only about 28 percent of the average number of national election stories (Wilke and Reinemann 2007).
Distribution by Media Outlet.
When assessing variations in the quantity of coverage, we looked at the number of articles as a whole, as well as the number of photographs and direct quotes. Overall, women were shown slightly more often, but were described and “heard” (through direct quotes) less often than men. Splitting the data according to the media outlets tells a different story, however. As shown in Figure 1, female candidates visually and verbally dominated the coverage in tabloid outlets, whereas their male counterparts were the focus of photographs and articles in the broadsheets. This is an unsettling result, though in line with previous research (Braden 1996; Bystrom 1999; Kahn 1994b; Kahn and Goldenberg 1991; Semetko and Boomgaarden 2007; Witt et al. 1994). The worst-case scenario would be a tacit agreement between the analyzed media outlets, as in women are to be covered by tabloids, whereas broadsheets deal with men. However, purely in quantitative terms, this concern becomes marginally less acute considering that due to stark variations in unique visitors’ numbers, news reports appearing on the tabloid web sites included here are likely to be read by more people than the broadsheet articles.
Variations in quantity (distribution by candidate gender, channel of communication, and media format).
In about 40 percent of the articles in our sample (n = 113), the candidates had the chance to explain their opinions with their own words. The number of direct quotes was slightly lower for female candidates; the male-issued direct quotes accounted for 55.7 percent (n = 63) of the articles featuring direct quotes. While the trend observed here contradicts the results of others (Falk 2008), it is worth noting that the tabloid press offered more direct quotes for Elena Băsescu than for George Becali.
As far as the quality of coverage is concerned, we now turn to variations in the use of horse-race, issue, and trivialization frames for the four candidates included in our sample.
As Table 3 shows, female candidates altogether received considerably more horse-race coverage than their male counterparts. The lion’s share of this type of coverage was found in the tabloid outlets for all four candidates. Specifically, a vast majority (81.2 percent) of all news stories featuring Monica Macovei and just over two-thirds of all news reports mentioning Elena Băsescu (70.6 percent) focused on the horse-race frame. On the other hand, the values for male politicians are mixed; this type of coverage was found in 73.6 percent of the articles featuring Adrian Severin, but only in slightly over a quarter of the coverage of George Becali (27.4 percent).
Variations in Quality (Distribution by Candidate).
Note: Values are percentages of all news stories per candidate.
One way to account for these variations is the rarity of women in high-ranking national politics. Thus, journalists working for highbrow outlets might think that comparing minority types (female politicians) to majority types (their male counterparts) would help the electorate gain a better impression of the candidates, which in turn could help them make an informed electoral decision. After all, horse-race coverage can make the viability of candidates more salient to voters (Iyengar and Kinder 1987). Another possible explanation could be that the prevalence of horse-race frames in broadsheets mirrors the high competition among media outlets: Given that this type of framing is likely to attract readers and the fact that broadsheets are losing ground to tabloids, highbrow outlets might make a conscious decision to cover elections in this manner.
The expected language or metaphors from the world of sports or war were hardly ever found (only in 10.7 percent of the sample, n = 30) within the horse-race frame. In these articles, the journalists discussed “defending” or “attacking” a person or their political views, and talked about “fighting” (and even “wrestling”), “being at war,” “protesting,” “winning” an argument, and “opponents” and “supporters.” This type of language was noticed in particular for Elena Băsescu (n = 12, 40 percent) and Monica Macovei (n = 11, 36.7 percent), the two female candidates. Public opinion polls were mentioned in 29 articles, which represent 10.35 percent of our sample. Most of these articles referred to Elena Băsescu (n = 17), followed by George Becali (n = 8), Monica Macovei (n = 3), and Adrian Severin (n = 1). Campaign resources were only mentioned in the coverage of female candidates, with an emphasis on the limited amount of those resources (Elena Băsescu, 4.7 percent, n = 5, and Monica Macovei, 3.1 percent, n = 1).
As shown in Table 4, all analyzed outlets were reticent in making viability assessments; the tabloids in particular. Yet, when the analyzed articles contained such evaluations, the candidates were more often described as unsuitable than as suitable. Adrian Severin was considered to be unsuitable in more than half of the articles in which he was mentioned (52.6 percent), followed by Elena Băsescu and Monica Macovei with 39 percent and 34.3 percent, respectively; for George Becali, we recorded 16.9 percent. The candidate with the most favorable suitability assessments in their sample was Monica Macovei with 21.9 percent, followed by Adrian Severin and Elena Băsescu with 15.8 percent and 11.4 percent, respectively. Only 4.8 percent of the articles featuring George Becali described him as suitable.
Viability Assessment (Distribution by Candidate and Media Genre).
Note: Values are percentages from the total number of news stories per candidate. Suitable and somewhat suitable were combined for the analysis.
In addition to these verbal assessments, the media can also make assessments of the candidate visually. While we did not include this variable in the horse-race frame, we can compare the way the candidates were described with the way they were portrayed in terms of favorability.
As Figure 2 shows, overall, visuals were more favorable for female candidates. Almost half (43.8 percent) of the articles featuring Elena Băsescu were accompanied by a favorable picture, while only 12.4 percent of those articles contained unfavorable pictures. Elena Băsescu was the only candidate for whom we could notice a difference between the tabloid and the broadsheet outlets: Tabloids portrayed her mostly in an unfavorable fashion. Monica Macovei, the other female candidate, was also mostly portrayed favorably (9.4 percent). On the other hand, 33 percent of the articles mentioning George Becali contained a favorable or neutral picture, while Adrian Severin was portrayed exclusively unfavorably (21.1 percent). This can probably be best explained by the fact that journalists often used staged stock pictures of Elena Băsescu from her past career as a fashion model. Moreover, the sample of pictures of Monica Macovei was very small.
Valence of pictures and picture size (distribution per candidate).
Issue frames were virtually nonexistent in the sample for male candidates. 9 For female candidates, however, the issue frame represented a fairly important part of the coverage, both in tabloid and broadsheet outlets (23.8 percent for Elena Băsescu, 68.7 percent for Monica Macovei). This is a surprising result, given that it challenges much of the literature on gender and election coverage (Kahn 1994b). A possible explanation for the higher distribution of issue frames among female candidates might be the fact that they are more likely to make issues “a cornerstone of their campaign” (Kahn and Gordon 1997: 74). Yet, since we have not analyzed the campaign materials of the candidates, this is only an assumption.
As mentioned above, we made a distinction between “male” and “female” issues. Table 5 above shows that all candidates received more coverage on male rather than female issues. In fact, only Elena Băsescu received some coverage on social problems (4.8 percent, n = 5), an issue typically associated with women. Most of the male issue-related coverage comes from broadsheets: 16.2 percent (n = 17) for Elena Băsescu and 65.6 percent (n = 21) for Monica Macovei. Most of the issue-related coverage featuring Elena Băsescu related to European legislation, 6.1 percent (n = 17), followed by foreign policy, 2.1 percent (n = 6), energy, 1.8 percent (n = 5), and justice, 0.4 percent (n = 1). The focal points in the coverage of Monica Macovei were justice, 6.1 percent (n = 17), economy, 3.2 percent (n = 9), and European legislation, 0.7 percent (n = 2).
Distribution of Male/Female Issues and Media Genre.
Note: Values are percentages from the total number of news stories per candidate.
As to whether the candidates were considered capable of dealing with the issues mentioned, the data show a bleak result for the female politicians. When such assessments are made, the female candidates are mostly considered unqualified. This was the case 24 times (22.8 percent) for Elena Băsescu and 15 times (46.8 percent) for Monica Macovei. To give an example, Elena Băsescu was deemed unqualified to deal with the issue of European legislation by one journalist due to her intention to support the legalization of marijuana in the EU. After presenting a presumably truncated version of her line of reasoning (that marijuana is legal elsewhere), the male journalist argues: “Of the many foolish things the youngest daughter of the President said . . . this one beats them all.” 10 He continues to ridicule the suggestion by comparing it with the legalization of harems and incest. His conclusion is that such politicians should not represent Romania in Brussels.
Thus, as welcome as the fact may be that women received more issue coverage than men, it is equally regrettable to see that women were mostly deemed unsuitable to deal with such issues. On the other hand, the reason why men did not receive this type of coverage might also be rooted in the aspects they chose to emphasize during the campaign. Also, in a patriarchal society, journalists might be assuming that men know what they are doing and that there is no value in discussing political matters. This does not seem reasonable, especially in relation to inexperienced politicians like George Becali.
As already mentioned, we operationalized the trivialization frame as an index consisting of remarks on a range of variables surrounding candidates’ appearance, personality, roles, and nonpolitical pictures. We encountered this construct more prominently for female candidates. We recorded 39.3 percent for Elena Băsescu and 16.1 percent for George Becali; most of these values hail from tabloid articles. Nonetheless, there was no sign of this frame in the articles featuring Monica Macovei and Adrian Severin. Since the dividing line between the two groups is their experience in the realm of politics, it seems plausible to assume that unconventional newcomers received more coverage through the trivialization frame than old stagers. We now briefly describe our results for all the variables encompassed in this frame.
The media reports included a limited amount of remarks about the candidates’ appearance; in fact, such remarks only referred to the two most prominent candidates. While five articles featuring George Becali (4 percent) contained such remarks, the percentage from the sample of Elena Băsescu was much higher (n = 30, 28.5 percent). Some of these remarks on Elena Băsescu’s appearance do not stem from journalists, but rather from George Becali. On one occasion, he said, “That’s a pretty girl, a beautiful girl . . .; that’s a girl who would look good in the European Parliament.” 11 This is a backhanded compliment: Although George Becali seems to support Elena Băsescu’s candidacy, this is based solely on her appearance. He justifies his stance with the claim that other MEPs lack both qualification and good looks. By implying that she lacks qualifications, George Becali deprecates the role of female politicians as accessories in a male-dominated environment. On a different occasion, an article mentioned Elena Băsescu being dubbed as the “Page Five Girl” of this election, since her name appeared on the fifth page of the ballot. This is a condescending allusion to the section of the tabloid Libertatea featuring photographs of topless female glamor models. 12
While Monica Macovei and Adrian Severin were depicted in a strictly work-related fashion at all times, both Elena Băsescu and George Becali were portrayed in trivial ways (thus in contexts other than politics). This was the case in 22.5 percent (n = 14) of the pictures of Elena Băsescu and in almost one-third (29.8 percent, n = 17) of the pictures of George Becali.
Regarding the coverage of character traits, the analysis showed that—of all the candidates—George Becali acquired the most traits coverage (77.4 percent), followed by Monica Macovei (62.5 percent), Elena Băsescu (45.7 percent), and Adrian Severin (36.8 percent). For all candidates, the traits coverage was mostly encountered in broadsheets, with just a few exceptions in the tabloids. Overall, we also noticed more coverage of male traits (37.1 percent, n = 104) than of female traits (19.6 percent, n = 55). In fact, Adrian Severin was exclusively covered in relation to male traits (n = 7, 36.8 percent). Both for George Becali and Monica Macovei, the gap between male and female traits was about 25 percent in favor of male traits, as shown in Table 6. The smallest gap and thus the most balanced traits coverage was registered for Elena Băsescu: 25.7 percent (n = 27) on male traits and 20 percent (n = 21) on female traits.
Traits Coverage (Distribution Per Candidate).
Note: Values are percentages from the total number of news stories per candidate.
Interestingly enough, it was not the female candidates who received the most coverage on female traits, but rather a male candidate. Monica Macovei acquired 18.7 percent coverage on female traits and Elena Băsescu 20 percent, but George Becali got 26.6 percent. A closer examination of the respective articles reveal that George Becali was portrayed as being sensitive over two very specific issues he was dealing with at the moment: troubles with the football team he sponsors and his short-term imprisonment. Both were rather humbling experiences, which resulted in George Becali taking offense over statements and actions and/or caused an emotional outburst on his part. Given how unusual this situation is, we would not expect this particular observation to be replicated in other contexts—at least not more for male than for female politicians.
Economic status received media attention only in the case of Elena Băsescu and George Becali; this involved thirteen articles for Elena Băsescu (12.3 percent) and fifty-five for George Becali (44.3 percent). Most of these portrayed both of them as privileged: Elena Băsescu (n = 12) and George Becali (n = 48).
Our inquiry into the coverage of gender role in the analyzed articles involved both a verbal and a visual element. Verbally, the data showed that the news reports emphasized candidate gender and gender role for the female candidates: While 72.3 percent (n = 76) of the articles featuring Elena Băsescu focused on her gender role, only six (n = 4.8) did the same for George Becali. Visually, we noticed that all candidates were mostly portrayed alone; in the cases of Adrian Severin and Monica Macovei, this portrayal was exclusive. Elena Băsescu, on the other hand, was portrayed six times with family members (9.6 percent).
Furthermore, we noticed that the media coverage related the gender of the candidate to their political prospects only seven times, all for Elena Băsescu (6.6 percent). This connection was negative. Also in relation to Elena Băsescu, the articles mentioned six times that politics is a male domain (5.7 percent). Yet no story mentions that the election outcome or voters’ behavior might (partly) depend on a candidate’s gender.
Following this description and discussion of our results, the next section is aimed at providing the reader with a final answer to our research question and discussing the limitations of this study.
Conclusion and Discussion
Our study showed several variations in the quantity and quality of coverage, most of them to the detriment of female politicians. While women dominated the tabloid outlets (both verbally and visually), men were featured prominently in the broadsheets. The trivialization and the issue frames appeared more often for female candidates, whereas the results for the horse-race frame were mixed.
If we accept the already presented argument that when images contradict the words in a news account, viewers tend to believe and remember the visual message over the verbal message (see Coleman 2010), then a potentially worrying bias becomes noticeable. In the case of Elena Băsescu, 20 percent of the accompanying pictures portrayed her as a fashion model, whereas her previous career was hardly ever addressed through words. If a candidate is perceived as just a pretty face in one-fifth of the cases, this is likely to stand in the way of her political ambitions. We acknowledge, however, that this might be rooted in journalists’ reliance on stock photos, and that the pile of political pictures of Elena Băsescu is dwarfed by the pile of fashion pictures.
In the light of these results, we have to answer our research question in the affirmative. There was a gender bias in the news coverage of Romanian candidates vying for the European Parliament. Although women may no longer be symbolically annihilated in the media (Tuchman 1978), it is still too early to proclaim equal treatment in the media. From this perspective, the media coverage of politicians is still on the beaten path: Female candidates are portrayed in a way that undermines their authority (Cowley and Childs 2003; Wilson 2004).
In this study, we laid out our case for an integrative framing analysis and explained the advantages. While we consider this to be a valuable approach, we might be criticized for using newspaper articles instead of video material. A similar analysis of video material would have been more complex, yet possibly richer in content. Our choice to analyze still photographs was motivated by the fact that the limited amount of images used alongside online newspaper articles can increase the importance of any single image, by allowing the viewer to attend to them more closely (Barrett and Barrington 2005a). Linked to this criticism, the method we used to measure the favorableness of the photographs can be disapproved of for being too subjective. Yet we produced a high intercoder reliability, which indicates that our quick impressions of the photographs were consistent.
In more general terms, the fairly low proportion of issue frames in the coverage is disquieting. From the perspective of participatory democracy, the lack of issue information in the press might hinder the informed evaluation of policy suggestions made by the candidates running for office (Kahn and Goldenberg 1991). Granted, this particular finding is partially the outcome of our candidate-centered sampling criteria. We included all articles on the four candidates analyzed, including articles presenting Elena Băsescu as a fashion model and George Becali as a football manager. While this was a deliberate decision, since we wanted to include all the available information, we acknowledge that had we selected the articles by using the search term “European elections” instead, our sample would have been considerably smaller, but the percentage of issue frames within would have been higher. Regardless, the absolute number of issue frames remains low.
Moreover, we noticed a prevalence of male issues and male traits in the coverage, which can be interpreted in several ways. Some might find this disquieting and a hint toward woman annihilation in a male-dominated domain. Others might consider these results able to prompt a reevaluation of these male–female classifications of issues and traits. Still others might argue that female politicians themselves are emphasizing male issues and traits in their campaign materials to succeed in a male-dominated field. But regardless of the origins of such coverage or one’s opinion about it, there is one important implication: Voters may regard male issues and male traits to be most important and use them to evaluate the candidates (Iyengar and Kinder 1987; Kahn and Goldenberg 1991; Markus 1982). Thus, a deliberate linkage between female politicians and male traits and issues could lead voters to perceive those traits as “human” rather than “male.” 13 Relying on Newton and Williams (2011), we suggest that masculine and feminine stereotypes impose artificial boundaries for behavior, personality, and political competencies. Thus, these exclusive constructs should be replaced by inclusive archetypes of politicians, regardless of their gender. On the other hand, given the patriarchal value system in Romania, an emphasis on feminine qualities might be quite convenient. Future audience research needs to address this type of questions.
The above described limitations of our study suggest directions for future research. In terms of research methodology, the refinement of the measurement scale for favorable pictures and trivial pictures seems particularly important. Furthermore, future research asking questions similar to the one posed here should incorporate video material in their analysis. The case of European elections would be particularly interesting here, especially if national coverage were to be compared with coverage in European channels like Euronews. Also, the analysis of candidate framing in still visuals should be expanded to include political cartoons and info graphics (Coleman 2010).
From the perspective of (visual) framing theory, future studies should contemplate the analysis of the whole process of mass communication, both verbally and visually. With regard to the media coverage of female and male politicians, the analysis should start with assessing the stereotypes in the respective culture and continue with the assessment of campaign materials, journalists’ perceptions, media coverage, and effects on the members of the audience. In the present study, we proposed a way to analyze the media coverage; this can serve as a starting point for such an analysis. Yet, it has to be expanded in such a way to allow for the analysis not only of generic frames but also of issue-specific frames.
In the case presented here, we are confident that the gender of the candidates was the main reason for the variations in coverage reported. However, the fact that other conditions (like the political standing of the candidates) were sometimes able to elucidate differences in media coverage points toward the need for a more complex explanatory model. It seems to us that scholars embarking on this exciting task should distinguish between elements linked to (1) the candidate and (2) the context.
With regard to the (1) candidate, elements such as age and experience, political standing and perceived respectability, incumbency, electoral prospects, and relationship with the media might bring along additional explanatory power. A case in point is the study by Ibroscheva and Raicheva-Stover (2009); the authors were able to show that a female university professor running for office was not subject to the same gendered coverage as the rest of the female candidates. Ross and Comrie (2012), to offer another example, showed how gender can work together with age in disadvantaging candidates. Finally, Edwards and McDonald (2010) attributed variations in the portrayal of Hillary Clinton and Sarah Palin in political cartoons not to their gender, but to their personalities.
As far as (2) the context is concerned, we argue for a systematic consideration of the underlying culture, the political party, the targeted political position, and the media outlet. The present study exposed important differences between the media coverage of the candidates in popular and highbrow outlets. Future investigations should contrast media outlets not only based on their mass-appeal but also on their scope. To be precise, a comparison of the media frames on the same candidates in national and European outlets points to rewarding results. With regard to the targeted political position, it seems plausible to assume that European races—just like mayoral ones (Atkeson and Krebs 2008)—are more open toward female politicians, at least when compared with patriarchal national settings. Thus, widening the focus of research to include more types of elections appears worthwhile.
Footnotes
Acknowledgements
The authors would like to thank Verena Pflieger for her valuable advice on the data analysis. We are also grateful to the editor and three anonymous reviewers for their insightful comments and suggestions. The second author is beneficiary of a scholarship with the program Doctoral Scholarship for a Sustainable Society, a project cofinanced by the European Union through the European Social Fund, Sectorial Operational Programme Human Resources and Development 2007-2013.
Declaration of Conflicting Interests
The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.