18 Million Cracks, but No Cigar: News Media and the Campaigns of Clinton, Palin, and Bachmann

Abstract

Decades of research within political science, political communication, and mass media found pervasive gender biased media coverage of female political candidates. However, recent research suggests that gender stereotypes do not have a consistent effect in all campaign environments and when gender stereotypes are not activated, female candidates are not disadvantaged. As a result, if we see a reduction in reliance on gender stereotypes in the media, female candidates should enjoy a more level playing field. In this analysis, we focus on mass media’s coverage of female candidates in elite executive political races. This study conducts a content analysis of media coverage of three recent women candidates for the United States’ highest executive offices: Senator Hillary Clinton, Congresswoman Michelle Bachmann, and Governor Sarah Palin. Our analysis of newspapers and television news coverage confirms the media do not discuss female and male candidates in neutral terms, but instead fall back onto traditional gender stereotypes and emphasize female candidates’ physical appearances and family roles far more frequently than they do for male candidates. This may, in turn, prime gender stereotypes in voters, impair candidates’ fundraising ability, and limit the electoral ambition of future generations of female candidates.

1. Introduction

The United States is experiencing a banner year in efforts to break the glass ceiling in American politics. At the time of manuscript preparation, former First Lady, New York Senator, and Secretary of State Hillary Clinton has just become the first woman to be the presidential nominee of a major political party. Women are far more common in political office today than during any previous era, but their penetration is uneven across governmental levels. Current achievements include securing 24.5 percent of seats in state legislatures and 104 federal congressional seats [1]. However, despite only making up around 20 percent of the body, women in Congress have received praise in recent years for being able to work across party lines and cooperate with broader coalitions than their male counterparts—“Women are the Only Adults Left in Washington,” “Senate Women Lead in Effort to Find Accord,” “Men Got Us Into the Shutdown, Women Got Us Out” [2,3,4]. Unfortunately, once women make it into Congress, they face an atmosphere that can be unwelcoming—20 of the 21 House chair positions were given to men for the 114th Congress [5] and overt sexism still exists in the halls of Congress [6,7].

Decades of research within political science, political communication, and mass media have found a widespread and enduring pattern of gender biased media coverage of female political candidates. Female candidates have historically been subject to coverage focusing on their physical attributes, emphasizing their familial ties, questioning their emotional fitness for office, and coverage that focuses significantly less on issues than their male counterparts [8,9,10,11].

However, research published in recent years demands that we reexamine our previous understanding of media coverage of female candidates. This new wave of research makes a strong argument that gender stereotypes do not have a consistent effect in all campaign environments and thus may not be the election barrier asserted by previous research. In her analysis of the 2010 U.S. House races that involved both male and female candidates, Dolan [12] uncovers no evidence that voters hold traditional gender stereotypes and factor them into their voting decisions. Hayes [13] draws a similar conclusion in a study of the 2006 U.S. Senate contests. Brooks [14] finds—based on a series of experiments—that women who act tough, get angry, or even cry on the campaign trail are not viewed any differently than are men who do the same thing. Hayes, Lawless and Baitinger [15] use an experiment to find that negative media coverage of candidates’ appearance does affect voter evaluations of professionalism, but the penalty is the same for both male and female candidates. Hayes and Lawless [16] find no systemic gender bias in local newspaper coverage of congressional candidates during the 2010 midterm elections and conclude that traditional indicators of vote choice trump any gender stereotypes held by voters. Baur’s [17] findings suggest that when gender stereotypes are not activated, female candidates are not disadvantaged relative to their male counterparts. As a result, if we were able to see a reduction in reliance on gender stereotypes in the media, female candidates would be enjoying a more level playing field.

Collectively, this emerging line of research suggests that gender stereotypes in media coverage either never existed, no longer exist, or do not a practical impact. In light of these recent findings, we must determine if gender bias in media coverage is still a ubiquitous feature of campaign coverage, and under what circumstances gender stereotypes and gendered media bias have an impact. In this analysis, we focus narrowly on mass media’s coverage of female candidates in elite executive political races as the glass ceiling remains intact at this electoral level. We also question the assumption made by several experiments that a balanced media environment currently exists. Gender stereotypes may affect male and female candidates in a similar way, but if there is an imbalance in who receives gendered coverage, the consequences are not distributed equally. We seek to determine if media has created a campaign environment free of abundant gender priming and examine the mass media’s emphasis of candidates’ physical appearance and familial roles, the “hair, husband and hemline” problem (also referred to as H3) ([11], p. 37; [18]). As noted above, research suggests that female candidates are subject more often to discussions of their physical appearance—clothes, hairstyles, body shape, attractiveness, etc.—and their domestic roles—wife, mother, grandmother, etc.—than their male counterparts. This is traditionally believed to have a detrimental effect on female candidacies as the time and space devoted to these characteristics could have been spent on policy positions and experience, and it continues to reinforce American culture’s objectification of even our most successful women. However, recent studies question the existence and impact of this barrier.

This study conducts a content analysis of media coverage of three recent women candidates for the United States’ highest executive offices: Senator Hillary Clinton, Congresswoman Michelle Bachmann, and Governor Sarah Palin. It seeks to identify differential media treatment of the women candidates concerning emphasis on matters irrelevant to their ability to govern across both newspaper and television coverage. We argue that unequal time and space spent discussing outfit choices, attractiveness, hair styles, and familial obligations is wasted and continues to undermine and objectify female candidates relative to their male competitors.

2. Why Women? And Why Not?

In this study we chose to focus on female candidates for elite executive office because while legislative positions are critical for setting policy agendas and crafting legislation, executive positions are unique in the singularity of the power they afford their holder. Rather than working primarily in committees, executive actions are perceived as individual actions. While women are receiving positive coverage for their congressional work, executive offices are perceived as more independently powerful, and here women are underrepresented drastically. Moreover, comparing the news coverage of Clinton, Palin, and Bachmann to their primary opponents allows us to remove partisanship from the equation—a notable advantage over general election studies. The number of women in political office within the United States continues to grow, but women have made far bolder advances in legislatures than executive positions. Hillary Clinton succinctly summarized the situation in her June 2008 concession speech, “Although we weren’t able to shatter that highest, hardest glass ceiling this time, thanks to you it’s got about 18 million cracks in it.” As of 2015, only 37 women have ever served as governor and we currently have just 6 female governors serving in the United States [1]. Positions at the federal executive level remain limited to 7 cabinet seats and women only hold 24.5 percent of state-level executive positions, far from gender parity [1]. The United States is also behind many of its peer nations when it comes to gender parity in political offices [19].

Research indicates that female legislators often concentrate on issue areas of importance to women voters [20,21,22,23,24,25,26]. The presence of descriptive representation for women changes the political system in terms of legislative output and encourages more women to run for office and desensitizes voters to the novelty of a female candidate [14,27,28]. Pew confirmed these ideas in a recent survey where it found that 78 percent of women feel that having women in top leadership positions would have least some positive impact on all women’s lives [29].

Scholars have identified a substantial and pervasive political ambition gap between women and men—as adolescents, college students, and professionals alike [12,28]—which can be attributed to socialization, situational factors, and structural variables [30,31,32,33]. In order to significantly increase gender parity in political representation, women must be socialized to recognize when they are qualified to run for political office [31,33]. While women and men use the same criteria to assess their potential as candidates, women underestimate the degree to which they meet the criteria [34]. The traditional masculine framing of politics and our traditional gender role expectations interact to reduce women’s sense of efficacy [34,35,36]. In addition, situational factors and gender norms must accommodate women’s entrance into political life. Despite women’s increasing role in the workforce, women still absorb a larger share of household responsibilities [37,38,39]. Without considerable support from their spouses, family, or hired staff, women simply may not feel they have the time to enter political life until they are older, sacrificing prime years to develop leadership and political experience.

In addition to the effects of socialization and situational factors that disproportionately discourage women from entering politics, structural variables also inhibit women’s candidacies. Party gatekeepers and fundraising heavyweights must learn to recognize the value and viability of female candidates [40]. While studies show that quality female candidates are viable and win at comparable rates to male candidates, this understanding is not shared by local and state party gatekeepers and women are significantly less likely to be recruited for public office than men with similar backgrounds and experience [34,40]. Not only are women less likely to be recruited as candidates, they are less likely to be recruited intensely or encouraged by multiple people [36]. Without support from low level party organizations, women’s representation in the political pipeline is reduced [40,41].

Americans are far less rigid regarding gender roles today, and most Americans no longer view women as incapable or ill-suited to politics. Emerging research finds little evidence that female candidates for office face widespread, overt bias that inhibits victory. Scholars argue that traditional shortcuts such as party identification, incumbency, and ideology trump gender effects in determining vote choice [12,13,14,15,16]. The lack of overt bias is important, but it is only part of the story. Recent research into the implicit biases of voters reveals that the average person still associates the words “president” and “executive” with male names more easily than female names [42]. Female candidates also face more primary opponents than their male counterparts [43,44] and, while women may win at the same rates as men, this parity occurs when the female candidates are of higher quality than their male opponents [14,16,44,45,46]. Media emphasis on H3 characteristics can complicate the opportunity for female candidates to establish their higher qualifications with voters. In her study on implicit biases, Mo [42] found that voters who implicitly prefer males for leadership positions, but also explicitly claim to be open to female leaders, will use a dual process account of reasoning to ultimately select the most qualified candidate regardless of their sex. But that result rests on the voter being presented with the female candidate’s qualifications more often than her outfit choices.

3. Contemporary Media Priming

The above explanations place much of the weight for our current gender disparities on individual choice and decision making, whether that of the potential female candidate, party leadership, or the voter. However, this assumes a balanced media environment that treats candidates equally regardless of their gender. We align with other recent studies that argue that when media coverage emphasizes familial connections, personal traits, gender, and candidates’ physical appearances, gender stereotypes are primed and female candidates are often disadvantaged [16]. When media primes gender stereotypes, the electorate can lose focus on substantive policy positions the candidates are advocating and inaccurately attribute characteristics or competencies to candidates. Media coverage that employs gender stereotypes or emphasizes the novelty of a female candidate’s bid for office may negatively impact perceptions of her viability and by extension her fund raising capability [14]. Applying a sexual frame to politics is especially damaging to women because cultural stereotypes about the powerful, sexual man reinforces perceptions of his power and strength, while the application of the sexual frame puts women in a double bind, where women must “combat the double bind between femininity and competence, media frames that cast them as sex objects undermine their credibility as leaders in ways that the same frames do not undercut male candidates” ([47], p. 336). We follow this line of theory and argue that imbalanced H3 references reassert gender as a primary heuristic through which voters make their electoral decisions and thus can have negative consequences for female candidates, especially at the executive level.

Female candidates often receive media coverage that differs substantively and significantly from that of their male counterparts by emphasizing candidates’ physical appearance and familial roles. This pattern is known as the “hair, husband, and hemline” (H3) problem ([11], p. 37; [18]). Falk [48] presents evidence of a continued bias in media coverage stretching from Victoria Woodhull’s 1872 presidential campaign through Sen. Clinton’s 2008 presidential bid. Female presidential candidates receive more emotional descriptions, are referenced more often in relation to their families or spouses, and are referred to by their first name more often than the men they campaign against [48,49]. They are also more often described in terms of their appearance with special attention given to their clothes and physique [48] and often receive less news coverage overall [50].

Elizabeth Dole’s 1996 bid for the Republican presidential nomination provided one of the first opportunities to extensively examine media coverage of female presidential candidates. Analysis of the coverage showed Dole was frequently depicted in domestic and sexual situations and often in relation to her spouse [51]. In addition, coverage of Dole frequently focused on her personal traits and appearance, rather than her qualifications or issue positions [8,46,51]. Similar patterns have been found in a variety of congressional and gubernatorial races [52,53,54,55]. In an analysis of political cartoons from the 2008 Democratic primary, Conners [56] found Sen. Clinton’s coverage frequently focused on her emotions and physical appearance, and often showed her requiring the assistance of her husband. Miller and Peake’s [10] analysis of newspaper articles from Palin’s 2008 vice-presidential run found that gender, physical appearance, and her family were mentioned far more often in relation to her than her male counterparts. Time spent covering a candidate’s physical attributes, fashion choices, and familial roles is time that could be devoted to their stance on substantive issues relevant to their potential governance. Following these previous findings, our first hypothesis is that we expect that there will be significantly more H3 references related to the female candidates than the male candidates in all three races under review.

In order to accurately capture the prevalence of H3 problem, this analysis includes both television and newspaper sources. Content analyses of news coverage were traditionally relegated to newspapers and, too often, studies have limited their newspaper analyses to just one or two newspapers. Examination of newspaper coverage continues to be necessary for temporal comparison with past research, but television news sources have secured a top spot in the bid for news-seekers. As of 2013, 55 percent of those surveyed reported turning to television as their main source of news compared to 9 percent turning to newspapers and 21 percent to the internet [57]. The internet and social media are fascinating lines of research, but the difficultly in controlling for the original source of the information puts them beyond our current research scope. Newspapers have limited space to fill and the average reader has a limited attention span. Television news sources, particularly cable networks, have virtually unlimited time to fill, present information with visuals, and can address the limited attention spans of viewers by switching between stories frequently. Moreover, the blurred line between news and opinion on cable networks provides an atmosphere where the content of the story is less controlled. These factors lead to our second hypothesis that television news broadcasts will include significantly more H3 mentions.

4. Data and Methods

4.1. Sample

Employing conventional procedures for analyzing news coverage, we conducted a content analysis of both print media and television newscasts. Five newspapers were carefully selected for the study based upon rank in the top newspapers according to circulation, geographic spread, and full access inclusion in the Newsbank Database or Lexis-Nexis Database. USA Today was selected because it has the second-highest circulation rate in the United States and unlike the top circulating paper (Wall Street Journal), it is geared towards the general public. The New York Times was also included in the sample as a national news source that increases the comparability of this study with the existing literature. Following the findings of Fox and Smith [58], the Chicago Sun-Times, Dallas Morning News, and San Jose Mercury News were selected to avoid a potential East Coast bias. While the LA Times has a larger circulation than the San Jose Mercury News, it was not available in either of the databases utilized in this study. San Jose Mercury News was identified as the highest circulation West Coast newspaper available in Newsbank and Lexis-Nexis at the time of this study. Fifty articles were randomly pulled from each newspaper source for each of the three races under study. Newspaper articles that were editorials or letters to the editor were excluded from the analysis by the coders after the sample was pulled to isolate news coverage from opinion. The number listed in

Table 1. Sample Overview.

Source	Clinton	Palin	Bachmann	Total
Television Sample Overview
CNN	34	33	47	114
MSNBC	49	50	50	149
FOX	49	50	49	148
Total Valid Transcripts	132	133	146	411
Total # TV Transcripts	150	150	150	450
Newspaper Sample Overview
USA Today	37	47	45	129
New York Times	34	31	37	102
Chicago Sun	25	36	13	67
Dallas Morning News	17	35	39	91
San Jose Mercury	36	32	40	108
Total Valid Newspaper Articles	149	181	174	497
Total # Newspaper Articles	250	250	250	750

for each newspaper represents the number of valid articles remaining after editorials and articles about the wrong person were removed.

The television news sample includes three cable networks to represent a broad ideological range of coverage (FOX News, MSNBC, and CNN). For each of the 3 races in this study, 50 random transcripts were pulled from each of the 3 cable news television sources. The inclusion of 50 transcripts provides reasonable insight into the television content, given the amount of content in each transcript. In total, our results reflect the content of 411 television news transcripts and 497 newspaper articles. The distribution of sources utilized in this study can be seen in Table 1.

The search terms included the surname of the female candidate and her chief rival identified as the person who won the primary nomination or her counterpart in the other party (“Clinton OR Obama OR McCain”; “Bachmann OR Romney”; and “Palin OR Biden”). While Sen. Obama was Sen. Clinton’s chief rival, Sen. McCain was included in the Clinton race search terms for two reasons: the extraordinary length of the Democratic primary made McCain a relevant competitor for both Sen. Clinton and Sen. Obama and Sen. Obama himself is beyond the parameters of the “traditional” American political candidate. The date range for each race begins the day the female candidate announced their candidacy until they formally suspended their campaign. The Bachmann race includes articles and transcripts from 13 June 2011 to 4 January 2012. The Palin race is confined between 29 August 2008 and 4 November 2008. The length of 2008 Democratic primary returned too many results for either of our databases to process thus we spilt the Clinton race into an early primary period (20 January 2007–31 December 2007) and a late primary period (1 January 2008–7 June 2008). For this race, twenty-five articles or transcripts were randomly selected from each primary period for each source.

4.2. Coding Methods

The articles and transcripts were coded to identify descriptors that referenced non-election relevant information about the candidate. The unit of analysis for the study was each descriptor, or descriptive phrase, that referenced information provided specifically about a candidate but not traditionally identified as important to voter decision making. The above sections expanded on the H3 problem ([11], p. 37; [18]). This category has a coding procedure that requires the researcher to assess the contribution of the information to the political campaign. Items typical of H3 mentions include: “Mitt Romney… must be squirming in his unbuttoned collar and pressed blue jeans” [59] or “So there was Bachmann at the Iowa State Fair in Des Moines the other day, downing a corn dog with gusto and inspiring a thousand Google Images and countless easy jokes on Twitter, Facebook and on the comedy shows” [60]. Not all H3 mentions are as easy to dismiss: “Hillary Clinton has been derided on some right wing talk radio stations as being unattractive...Calling her the Ice Queen isn’t a personal attack. That is your interpretation of her demeanor. Saying that she has a fat butt...” [61]. Each mention was counted. For example, one count for Romney’s jeans, one count for Bachmann’s corn dog, one count for calling Clinton unattractive, one count for demeanor, and one count for referencing “a fat butt.” In order to capture coverage emphasis on candidates’ familial connections, or familial responsibilities, we also separately code references to spouses, children, or family. For example, the statement “Her son, Track, will deploy to Iraq in the next few weeks and she, Governor Palin is waking up in her home town in Wasilla with her family for the first time today since she became John McCain’s running mate” [62] contains two mentions of Gov. Palin’s family. For the female candidates, we also code direct references to them as a woman or the novel female candidate. Many of the debates surrounding the consequences of gender stereotyped media coverage rest on the argument that they are only relevant when gender is primed. We intended to see how often media outlets directly pointed to the candidate’s gender or sex.

We chose to use each H3 reference as the unit of analysis because we sought a precise measure of the frequency of H3 mentions. Television transcripts, in particular, are often quite lengthy and we argue that there is a substantive difference between a transcript including two H3 references, versus a transcript containing 35 references. If we only coded articles that contained at least one reference, we would have just a binary, rather than a continuous measure. Moreover, given our primary interest in how coverage of the female candidates compares to their male counterparts, a more precise measure using H3 mentions better fit our needs.1 This unit of analysis is a common approach within the literature and content analysis research (For example, [13,46,63]).

Manual coding was utilized to capture subtler characterizations than can be identified than with the use of content analysis computer programs. Specifically, the descriptions under investigation in this study required manual coding in order to make conclusions about the relevancy of the statement to a candidate’s campaign. Human content analysis research can suffer from analysis by only a single coder, which raises significant reliability concerns [64]. This research utilized two coders, the authors. To guard against reliability issues, the authors developed specific coding procedures and each article and transcript was reviewed by at least one author, and a random selection was examined by both coders to ensure reliability in the results. The two coders produced a value of 0.92 using Cronbach’s alpha, suggesting that coding was highly consistent between coders.

5. Results

The prevalence and consequences of the “hair, husband, and hemline” problem has been well documented ([11], p. 37; [18,44,46,47,48,50,65]) and our findings corroborate earlier studies. Female candidates for president are far more likely to have their marital and parental status discussed in newspaper and television news coverage than their male opponents. As revealed in

Table 2. Summary Overview—Female vs. Male Candidates.

Category	Female Candidates	Male Candidates	Total
# H3 Mentions	65% (386)	35%(205)	100%(591)
# Spouse, Parent	88% (607)	12%(80)	100%(687)
Chi squared 97.35 (1) p < 0.0001

, nearly 90% of all familial mentions were in reference to female candidates (88% to 12%). Even with the Bachmann-Romney race, where Rep. Bachmann received less overall news coverage as one of many primary competitors, her spousal and parental connections were mentioned twice as often as her male counterpart (See

Table 3. Summary Overview.

Category	Clinton	Obama	McCain	Palin	Biden	Bachmann	Romney	Total
# Articles/Transcripts	400			400		400		1200
# Valid (not excluded)	281			314		320		915
# H3 Mentions	30% (53)	62% (110)	8% (15)	88% (297)	12% (39)	47% (36)	53% (41)	591
# Novelty	100% (93)	0% (0)	0% (0)	100% (172)	0% (0)	100% (50)	0% (0)	315
# Spouse, Parent	89% (278)	9% (29)	2% (7)	93% (278)	7% (21)	69% (51)	31% (23)	687

). Gov. Palin had small children during her candidacy and she mentioned them often. However, there is still a large gap between Palin’s 278 references to her family compared with Sen. Biden’s 21 references (Table 3). For example, CNN featured a discussion between panelists on The Situation Room where it was pointed out that “There is actually a debate raging whether Palin, as a mother of five, can lead the country” [66]. Similar to Gov. Palin, Sen. Obama had small children during the campaign while his competitor’s child was an adult. However, Sen. Clinton still managed to receive 278 familial mentions compared to Sen. Obama’s 29 familial mentions (Table 3). We should caution that references to Sen. Clinton’s husband, former President William J. Clinton, were included in this count when directly linked to Sen. Clinton. However, we do not believe this diminishes the finding because many of Clintons’ familial references featured language like “if she needs Bill Clinton, a man to defend herself at all times, how can she portray herself as commander in chief” [67]. This quote highlights the problems with unequal familial mentions. Because of the gender social norms in the United States, references to husbands and children may serve to remind some voters that female candidates have responsibilities at home that may compete with their commitments to the nation. This same assumption in not generally applied to men, as even today, the domestic workload of “husbands” and “fathers” is far less than that of “wives” and “mothers.” Women still bear the brunt of childcare and household chores, and women are often asked how they will balance these domestic duties with professional ambitions, in a way that men never are [65,68,69].

Moreover, Table 2 reveals that our female presidential and vice presidential candidates were 30 percent more likely to have the media discuss their appearance, clothing, and other trivial distractions than their male opponents (65% of H3 mentions are regarding female candidates). This aligns with our expectations given the literature. However, a closer analysis raises interesting questions. Table 3 displays the raw counts and percentages for individual candidates within each race and shows that the degree of fluff coverage varied greatly between the races. Sen. Obama actually received twice as many H3 mentions as Sen. Clinton. Congresswoman Bachmann’s fluff references were relatively close to Gov. Romney’s although she received less news coverage than the other women in the sample. Palin’s H3 mentions, however, were off the charts- Palin received more than seven times as many fluff mentions as Biden (Table 3).

Given the small number of candidates, it is impossible to confirm why these differences exist, but we would attribute it to either a partisan or introductory effect. It is possible that Republican female candidates cultivate a campaign to fit within expected gender stereotypes. They may intentionally highlight their families and fashion choices because those are expected, and accepted, expertise areas of women. Given the high occurrences of H3 for both Obama and Palin, it is also possible that candidates without an established national profile receive more H3 coverage as the media introduce them to the public. Both these questions are beyond the scope of this paper but provide interesting lines for future research.

Comparatively, references to our three female candidates as the novel woman candidate varied greatly. Gov. Palin’s sex was mentioned most often with 172 references with Sen. Clinton receiving 93 direct references to her sex. Representative Bachmann was the least novel of our female candidates receiving only 50 references to her sex (Table 3). If media coverage asserts and reasserts female candidates as something unique, novel, and different simply because of their sex, we can assume that gender is primed as a heuristic for the voter and gender stereotypes will be in play. At the aggregate level, we are left with dramatic inequalities between female and male candidates regarding marital status, parenthood, and physical appearance that are difficult to dismiss.

As mentioned earlier, coverage of Sen. Clinton’s campaign was sampled from the early and the late primary season. This can provide initial insight into any difference in H3 coverage between early and late seasons. However, as Gov. Palin was not announced until the closure of the primary season and Rep. Bachmann’s campaign during the early primary season, we can only compare across one candidate, Sen. Clinton. We would expect H3 coverage to be more prevalent during the early primary as media outlets are introducing candidates to the public. We would expect that by the late primary season, items of more substance and importance would be covered. However, we find that 84 percent of all H3 references during the Clinton race occurred during the late primary season. This also holds at the candidate level: 79 percent of the H3 references used to describe Sen. Clinton occurred during the late primary season. While purely descriptive in nature, this particular result raises concerns about the media’s attention span and raises the possibility that female candidate may receive more H3 coverage when they are not the leading candidate in the polls.

As discussed earlier, this paper offers a comparative analysis of television and newspaper coverage and gender stereotypes. Figure 1 compares television and newspaper coverage across the H3 problem (note: t = 8.77, (df = 2), p < 0.01). While television coverage made more total H3 and familial mentions overall, both modes are guilty of gendered coverage. In both television and newspaper coverage, the female candidates received 89 percent of familial references. There is a small difference between television and newspaper concerning H3 descriptions, but minimal compared to past research [70]. In both modes, female candidates received over 60 percent of the H3 references, with television being the slightly larger perpetrator (68 percent compared to 63 percent; See Figure 1). In a space- and time-limited environment, H3 and familial coverage means less issue coverage. This is damaging to female and male candidates, but the results in Figure 1 confirm that female candidates are disproportionately covered in this manner and thus face an additional hurdle to election.

Figure 2 provides a more nuanced examination of the H3 references across sources within each medium. In all sources under review expect for The New York Times, female candidates received a majority of H3 references. Fox News and MSNBC were relatively similar in their gendered use of H3 references with the female candidates receiving respectively 68 and 62 percent of references within their coverage (Figure 2). The Dallas Morning News was actually the worst culprit concerning imbalanced use of H3 references, with 92 percent of the references in their coverage going to the female candidate. The San Jose Mercury News distributed 72 percent of their references to the female candidate, while the Chicago Sun, New York Times, and USA Today provided the most gender balanced use of H3 references in the entire sample. Concerning the “hair, husband, and hemline” problem, cable television sources are consistently imbalanced in their distribution of H3 references between male and female candidates, while newspaper sources show a greater diversity in gendered H3 coverage (Figure 2).

Finally, we also examine familial references across sources to gain insight into how that consistent 89 percent in Figure 1 plays out among sources. The gender disparity is much higher along this variable with 5 of our 8 sources directing over 90 percent of familial references to the female candidate (Figure 3). All three television news sources performed poorly with female candidates receiving 92 percent of familial references at Fox News, 87 percent at MSNBC, and 91 percent at CNN. San Jose Mercury News is the biggest culprit across this measure with 97 percent of familial references being directed towards female candidates, while USA Today is the least incriminated with only 68 percent of familial references linked to the female candidate. Concerning references to spouses, children and familial obligations, both news media provide astoundingly imbalanced gendered coverage. It is no wonder that 67 percent of respondents in the 2008 Pew Survey supported the argument that “Women’s responsibilities to family don’t leave time for politics” [71].

6. Discussion

Building on a literature that is evolving quickly, we seized the opportunity to examine the relative treatment of candidates in campaign coverage of Senator Hillary Clinton, Congresswoman Michelle Bachmann, and Governor Sarah Palin. We find clear evidence that the contemporary media environment covers female and male candidates in disparate manners. Female candidates are far more likely to be the subject of discussions of physical appearance and familial roles than their male counterparts. Because female candidates often receive less overall news coverage than their opponents, and their coverage is disproportionately focused on H3, the public is receiving far less policy and leadership information on female candidates. This pattern makes it more difficult for female candidates to demonstrate their ability to lead and convince voters that their policy initiatives are best suited for the Oval Office. This may, in turn, prime gender stereotypes in voters, impair candidates’ fundraising ability, and limit the electoral ambition of future generations of female candidates.

Emerging research suggests that in a neutral environment, Americans do not rely heavily on gender stereotypes to evaluate candidates for political office. When gender stereotypes are not primed and candidates are treated equally, the importance of gender on candidate evaluations and vote choice is limited [15]. Unfortunately, the American news media is not a neutral environment. Our analysis of newspapers and television news coverage confirms the media do not discuss female and male candidates in neutral terms, but instead fall back onto traditional gender stereotypes and emphasize female candidates’ physical appearances and family roles far more frequently than they do for male candidates. Research suggests that when media coverage emphasizes personal attributes, these personal qualities become more important to voters and stereotypes are activated. When voters have low information levels and rely on news sources driven by the commercial bias, and in races where partisanship and incumbency do not distinguish between candidates (open races, primary races, etc.), reliance on gender stereotypes becomes more likely and more influential.

One major change in the media environment that requires further analysis is treatment of gender in internet news coverage. More and more voters are using the internet and phone applications as their primary news sources. This is also the news content people are most likely to share via social media, which further extends its reach. Online news sites are operated by a variety of news organizations and the results of an analysis of television, newspaper, and online news content is needed to better reflect what Americans are being exposed to (particularly younger generations). We have reasons to believe that younger generations are more open to dismissing traditional gender roles [72], but a more thorough investigation of internet news coverage is needed reveal whether their news sources are also reinforcing traditional gender roles. In addition, including network newscasts in future analyses is desirable. Although network news viewership has declined in recent years, it remains an important component of the media landscape.

7. Conclusions

While women are making strides into Congress and state level politics, patterns of media gender discrimination continue. Although Mitt Romney had 18 grandchildren at the time of his 2012 presidential run, no one questioned if a grandfather could lead the nation and keep his grandfatherly emotions and responsibilities in check in the White House. This role was rarely mentioned, outside of confirming he was a legitimate “family man.” Yet just hours after the news broke that Hillary Clinton’s daughter was expecting, “Grandmother in Chief?” became a major headline [73]. The recent 2016 presidential election cycle featured two women making a bid for their parties’ nominations. While the 2015 Pew survey on Women and Leadership [29] reveals that 38% of surveyed adults hope that the US elects a female president in their lifetime, our research confirms that they must overcome additional hurdles created by a gendered media environment compared to their male counterparts. Despite the optimism of much of the recent research on gender bias in electoral politics, gender remains pervasive in media coverage and media approaches to female and male candidates remain decidedly uneven.