Cabinet ministers hold key posts at the top of the executive branch, developing and implementing policy and navigating executive-legislative relations.Footnote 1 The media’s portrayal of cabinet ministers is how the public learns about these executive branch leaders: their credentials for their post, and how they manage challenges that arise in their department’s policy domain. As more women are appointed to cabinet posts, it is important to examine how the media presents women cabinet ministers compared to the portrayal of men, and whether the media’s coverage is gendered, as has often been found for media coverage of female politicians running for elected office. If the media provides less coverage of women than men in the cabinet, or focuses on trivial topics such as how they look rather than on their work leading their department, it may impede the credibility of women who have achieved these top political offices to do their job in the minds of the public and the legislature. Media treatment of women cabinet ministers also matters because if the media presents women in a more negative fashion than their male colleagues, then media treatment of women ministers would negatively affect public perceptions of the government and the president.
Yet, to our knowledge, only one study examines how media treatment of women who hold cabinet posts compares to coverage of their male colleagues: Fernández-García (Reference Fernández-García2016) about cabinet ministers in Spain.Footnote 2 We expand research about media treatment of women compared to men who hold appointed posts by examining news coverage of cabinet ministers in three presidential systems with differing levels of women’s incorporation into politics: Costa Rica, Uruguay, and the United States. This comparative research design allows us to examine whether media treatment of female and male appointees who run executive branch departments differs depending on the degree of incorporation of women into the cabinet and with the type of portfolio.
Most of the literature about media treatment of women politicians has focused on women candidates, primarily for legislatures, but also for the head of government post. But posts in presidential cabinets differ from elected posts in ways that may impact whether media coverage is gendered. First, the president appoints cabinet ministers. In contrast, to become a legislator or head of government an aspirant must win an election, and during the campaign, the candidate (who they are, their background, traits, policy stances) and their elections chances are the news story. Second, a minister’s work focuses on policy areas defined by the department’s enabling legislation and the president’s policy initiatives. The minister’s policy decisions, and how they handle problems in their policy domain are news. Legislators, in contrast, can choose the policy topics they work on, and heads of government (from presidents to mayors) are responsible for all policy areas assigned to their level of government. Third, many ambitious legislators want a cabinet post because they control a budget and staff (Chasquetti et al. Reference Chasquetti, Buquet and Cardarello2013, 27). This means that as women receive more cabinet appointments, they are limiting men’s access to powerful posts.
Cabinet departments oversee specific policy areas, so the minister can have credentials that demonstrate experience with these topics. Ministers must be effective at negotiating with the legislature to get executive policies passed into law (Neto Reference Neto2006) and able to handle questioning by the legislature. The nature of the job of cabinet minister prompts our question of whether the media gives parallel coverage of men and women cabinet ministers. Is coverage focused on their policies and implementation, or as is often found in media treatment of women candidates for elected office, is media coverage of women mostly about non-germane topics such as looks or family?
Women are being appointed to cabinets in greater numbers and hold more diverse and prestigious portfolios in many countries, even though some posts (e.g., finance, defense) remain bastions of male power (Annesley et al. Reference Annesley, Beckwith and Franceschet2019; Armstrong et al. Reference Armstrong, Barnes, O’Brien and Taylor-Robinson2022; Barnes and O’Brien Reference Barnes and O’Brien2018; Escobar-Lemmon and Taylor-Robinson Reference Escobar-Lemmon and Taylor-Robinson2016). We examine whether quantity and quality of coverage varies systematically with the sex of the person who holds the post, the type of policy covered by the post, and the extent of women’s incorporation into government.
While appointments to cabinets are a gendered process (e.g., Annesley et al. Reference Annesley, Beckwith and Franceschet2019; Barnes and O’Brien Reference Barnes and O’Brien2018; Bauer and Darkwah Reference Bauer and Darkwah2022; Borrelli Reference Borrelli2010), we examine coverage after the appointment. This is a similar approach to studying coverage of legislators during the “routine period,” when a legislator is in office, rather than studying election coverage (Aaldering and Van der Pas Reference Aaldering and Joanna Van Der Pas2020; Van der Pas and Aaldering Reference Van der Pas and Aaldering2020). By focusing on the period post-appointment (in Costa Rica and Uruguay) or after confirmation (in the United States) and the initial few weeks of the new administration, we evaluate media treatment during a “routine” period. Further, the early honeymoon period of a new administration allows a level playing field for all ministers, as presidents enjoy a grace period after winning election. Yet, even newly installed administrations can face problems that place a new minister in the “hot seat.” Examples from our period of study include the economic crisis faced by the US Treasury Secretary from the date he took office in January 2009, the national teachers’ strike that Costa Rica’s education minister faced that began the day she was sworn into office in May 2014, and multiple ministers in the March 2015 Uruguay administration were immediately faced with a potable water crisis.
Research shows that favorable press coverage influences presidents’ popularity and their ability to achieve their legislative agenda (Calvo Reference Calvo2014; Chasquetti Reference Chasquetti, Alemán and Tsebelis2016; Edwards III Reference Edwards and George2009). The same may be true for the ability of cabinet ministers to pursue the administration’s agenda. Thus, how the media portrays female, compared to male members of a president’s cabinet can matter for the capacity of women to do their job once they hold top executive branch posts.
We use an original data set from three presidential systems to conduct large-N analysis. Results indicate that media treatment of women cabinet ministers is more equivalent to treatment of men where there is greater incorporation of women into government. The media does not punish women holding posts outside stereotypically feminine policy areas. The media more often covers credentials of women than men when there is limited incorporation of women into the cabinet, even when controlling for objective experience/credentials.
Expectations for Minsters in Presidents’ Cabinets and Gendered Media Treatment
Politics has been a male domain, so women need to establish their qualification for office. Part of the challenge for women politicians to do so can come from journalists, who may give men and women unequal coverage (measured here as word count, % of the article, # of quotes, tone of coverage), and not present women’s qualifications, instead focusing on appearance or traits (Aday and Devitt Reference Aday and Devitt2001; Álvarez Monsiváis Reference Álvarez Monsiváis2021; Bystrom Reference Bystrom and Murray2010; Hayes and Lawless Reference Hayes and Lawless2016; Hayes et al. Reference Hayes, Lawless and Baitinger2014; Kahn Reference Kahn1994; Murray Reference Murray, Murray and Santa Barbara2010). Research, however, has primarily studied election coverage with little work about officials in office (see Van der Pas and Aaldering [Reference Van der Pas and Aaldering2020] for meta-analysis, but see Font and Ponce [Reference Font and Ponce2019] about senators in Uruguay; Cárdenas Arias [Reference Cárdenas Arias2023] about capital city mayors in Colombia and Spain). Generalizability of findings is also uncertain because most studies are of Anglophone nations.
Some scholars conclude that as women become more common as candidates the media focuses less on the candidate’s sex, looks, or personality, and there is more balance across men and women candidates in coverage of issues (Atkeson and Krebs Reference Atkeson and Krebs2008; Kittilson and Fridkin Reference Kittilson and Fridkin2008; Hayes and Lawless Reference Hayes and Lawless2015). Yet, equal treatment is not the conclusion of all studies (Fernández-García Reference Fernández-García2016; García Beaudoux et al. Reference García Beaudoux, D’Adamo, Berrocal Gonzalo and Gavensky2020; Gidengil and Everitt Reference Gidengil and Everitt2000; Thomas et al. Reference Thomas, Allison Harell and Rijkhoff2021). For nominees to national high courts, Escobar-Lemmon et al. (Reference Escobar-Lemmon, Valerie Hoekstra and Caul Kittilson2016) find that much media attention focuses on the woman’s status as the first, and highlights professionally irrelevant factors such as family life instead of professional qualifications. Murray’s (Reference Murray, Murray and Santa Barbara2010) cross-national study of female candidates for executive office also found the “first woman” frame to be common.
Media coverage of women ministers in presidents’ cabinets may differ from coverage of women running for office. As outlined above, cabinet ministers implement policy in a defined area, and handle problems that arise in their policy domain. Thus policy, more than the minister’s sex or traits, should attract media coverage. Building on literature that finds women candidates are treated more equally to men candidates when women lose their novelty as candidates, we expect a similar pattern as women become more common in the cabinet. Literature about mental templates indicates that people update their image of who “fits” a job based on who they regularly see doing the job. Thus, when women are regular figures in politics, which can occur across a range of offices and types of posts, women should fit people’s, including journalists’, image of a leader (Koenig et al. Reference Koenig, Eagly, Mitchell and Ristikari2011; Taylor-Robinson and Geva Reference Taylor-Robinson and Geva2023). We thus hypothesize:
H1 As the incorporation of women in the cabinet increases, there will be fewer differences in media treatment of men and women cabinet ministers, in quantity and quality of coverage.
Feminine stereotype congruence is argued to impact how the media treats women candidates (Álvarez Monsiváis Reference Álvarez Monsiváis2021; Font and Ponce Reference Font and Ponce2019; Fowler and Lawless Reference Fowler and Lawless2009; Murray Reference Murray, Murray and Santa Barbara2010). Women may be advantaged when running for state governor because states manage stereotypically feminine policy domains like education and health. National politicians are expected to handle masculine stereotyped policy domains such as economics and security (Carlin et al. Reference Carlin, Carreras and Love2020; Falk and Kenski Reference Falk and Kenski2006). Extending this logic to unitary systems, women should have an advantage for some cabinet posts, as the national government administers feminine stereotyped policies including education and health. Surveys that ask “what is the most important problem facing the country” indicate those policy areas can be high visibility (Escobar-Lemmon and Taylor-Robinson Reference Escobar-Lemmon and Taylor-Robinson2016, 71–74). Our second hypothesis builds upon predictions that there are negative reactions to women who work outside of stereotyped-feminine fields where women have long been present.
H2 Tone of media coverage of women will be more critical when women hold posts that clash with gender role congruity expectations.
We also consider the role of gender in media emphasis of experience and qualifications. Women in the US Senate are often not given credit for their experience and abilities (Dolan Reference Dolan2014). Women high court justices may be painted by the media as less experienced or lacking the necessary qualification for their duties (Escobar-Lemmon et al. Reference Escobar-Lemmon, Valerie Hoekstra and Caul Kittilson2016), while the media highlights the qualifications of male politicians (Van der Pas and Aaldering Reference Van der Pas and Aaldering2020). Since cabinet ministers oversee a specified policy area, it is possible to define experience and expertise that is relevant for the post, and the media could provide information about the minister’s resume. Gender politics research has shown that women are often better qualified than men before they will run for office (Bauer Reference Bauer2020; Fulton Reference Fulton2012), but having credentials for a job does not guarantee that the media covers those credentials for women. We thus hypothesize that:
H3 The media will less often portray women than men as experienced and having the necessary qualifications for their post.
Research Design
Case selection
We examine media coverage of ministers appointed to presidential cabinets in Costa Rica, the United States, and Uruguay. These presidential countries provide variance in incorporation of women into government, which is our primary explanatory variable for whether the media will provide equal coverage of women and men ministers. Costa Rica has extensively incorporated women in the executive and legislature for several decades. In Uruguay, women’s incorporation into cabinets and the legislature has been limited, even with a gender quota for legislative ballots, though the number of women increased markedly in the initial Vázquez cabinet in 2015. The United States has been slow to reach the world average for women in legislatures, but women’s presence in the cabinet has increased, including several women secretaries of state. The cases also vary regarding women’s appointment to stereotypically masculine posts, which is the second factor we hypothesize will affect media treatment of women ministers. (Appendix A provides detailed information about women in government for each country.)
Costa Rica and Uruguay are the new countries for analysis since the United States has been extensively studied about gendered media coverage of candidates. They vary in incorporation of women into government but are similar in some aspects of institutional design and government policy agenda. Both countries are unitary with little devolution of policy or financial power to local government (Bentancur and Busquets Reference Bentancur and Miguel Busquets2017; Muinelo-Gallo and Rodríguez Miranda Reference Muinelo-Gallo and Rodríguez Miranda2017). In both countries, the national government is responsible for social welfare programs, and the national government’s agenda has long emphasized social welfare policy. Cabinet posts that are stereotyped as feminine are often high profile with large budgets. Those posts match gender stereotypes about women’s expertise, but they are desired by male politicians. Neither country has the military power to emphasize defense (Costa Rica abolished its military in 1948), though crime is an important topic of concern in both countries (AmericasBarometer 2010, 2014). In the United States, with its federal system, social welfare policies are largely handled by the states, and the national government plays a lead role in world politics, often acting as a global policeman (Scruggs and Hayes Reference Scruggs and Hayes2017). In all three countries, AmericasBarometer surveys in 2008 and 2012 show large majorities of the public (ranging from 70% to 84%) disagree or strongly disagree with the statement that men make better political leaders than women.
The countries’ party systems are different. The United States has a two-party system, Uruguay since the return of democracy has had three major parties, and Costa Rica’s two-party system had fragmented by the period we cover. The party system will influence the president’s decision-making about cabinet appointments to aid relations with the legislature, which may influence appointment of women (see Dos Santos and Jalalzai Reference Dos Santos and Jalalzai2021). Whether it would influence media treatment of women and men ministers is unknown in the present literature.
In all three countries, presidents can face challenges from the legislature. Costa Rica’s president is constitutionally weak and is often frustrated by lack of cooperation from the Legislative Assembly (Vargas Cullel Reference Vargas Cullel2007). Uruguay’s parties are factionalized, so the president must bargain within their party to establish enough legislative support for their agenda, even when the party has the majority (Chasquetti et al. Reference Chasquetti, Buquet and Cardarello2013). US presidents face constraints under divided government. (Appendix B provides details about executive-legislative relations in each country and its impact on cabinets.)
All three countries are consolidated democracies as reflected in their Polity and Freedom House scores.Footnote 3 In consolidated democracies, we can assume rules of the game for executive-legislative relations are known and presidents expect to serve their full term. Thus, presidents can select cabinet ministers and pursue their policy agenda without the expectation of being removed from office prematurely. However, this research design decision limits generalizability of our findings beyond consolidated presidential democracies.
In sum, structural and historical differences provide different challenges across these three countries for all ministers. A cabinet minister’s ability to navigate these challenges could be affected by how the news media presents the minister, which makes assessing whether media treatment is gendered a question of interest. To conduct that assessment we collected data about media coverage of full cabinet rank ministers from the initial cabinets of two presidents in each country from a similar time frame: Costa Rica’s presidents Laura Chinchilla (2010, National Liberation Party) and Luis Guillermo Solís (2014, Citizens Action Party), US presidents George W. Bush (2001, Republican Party) and Barack Obama (2009, Democratic Party), and Uruguay’s Presidents José Mujica (2010, Broad Front) and Tabaré Vázquez (2015, Broad Front). We examine President Bush’s and President Obama’s first terms despite the temporal discrepancy with the proximal Costa Rican and Uruguayan cases to maintain the “honeymoon” period in all cases. Bush’s 2004 and Obama’s 2012 presidency would not have had honeymoon periods because consecutive terms are generally considered a carryover from the first term. The administration between Vázquez’s first (2005) and second terms allows for two separate honeymoon periods.
Data
We examine media coverage of all cabinet ministers appointed to the initial cabinet during approximately the first month in office of six presidents for 97 total minsters, 31 of whom are women (32%), in 1,864 total newspaper articles. During this early period of the administration all ministers should benefit from the honeymoon effect. Finding bias in the honeymoon period would be a particularly strong indication of the media treating women cabinet ministers more harshly than men. Additionally, focus on the president’s initial appointees allows us to overcome how to model media treatment of ministers who are appointed as replacements, possibly in response to crisis in the policy area of the ministry, or declining popularity of the president. While these advantages are useful for examining gendered differences, focusing on that initial period does limit the temporal generalizability of our findings, as we discuss in the conclusion.
Even in the honeymoon period, cabinet posts differ. Ministers oversee different policy domains – some viewed as stereotypically feminine (though still often run by men), some masculine, and some neutral. We exploit the stereotype variance across posts to test when post may interact with minister sex to determine differences in media coverage. We code the “gender” of the portfolio based on Krook and O’Brien (Reference Krook and O’Brien2012), with 20 ministers coded in stereotypically feminine posts, 50 in stereotypically masculine posts, and 27 ministers in neutral posts.
We use Krook and O’Brien’s (Reference Krook and O’Brien2012) Gender Power Score (GPS) as a systematic way to measure incorporation of women into cabinets. The GPS includes the percentage of cabinet seats held by women, weighting posts to account for the percentage of masculine/neutral/feminine portfolios held by women, and the percentage of high/medium/low prestige posts held by women. The largest possible GPS score is 12 which would mean all ministers are women. A score of 3 indicates a cabinet is 50% women, and that women hold half of all types of post. Table 1 presents the GPS scores for the six initial cabinets we study (bolded) and additional cabinets to contextualize women’s cabinet incorporation in each country.
Sources: (1) GPS for additional administrations calculated based on the Nyrup and Bramwell (Reference Nyrup and Bramwell2020) data (WhoGov within v2.0), (2) % women in the legislature for Costa Rica and Uruguay from the Inter-Parliamentary Union (ipu.org accessed August 26, 2023), the United States from Congressional Research Service. October 13, 2022. “Women in Congress: Statistics and Brief Overview” https://crsreports.congress.gov R43244 Appendix A Table A1.
Given our interest in administration level (GPS) and minister-level (sex, post stereotype) explanatory variables, we use a variety of multilevel models to test our expectations of gendered media coverage, allowing random effects for administration. Specific model specification is determined by the dependent variable of interest. Methodologically, the use of multilevel models allows us to include administration-level controls of theoretical interest, while not over emphasizing their explanatory power. Specifically, by including the GPS, we can meaningfully situate the six cases on a single scale for comparison as all ministers in a cabinet are clustered under one GPS, but each cabinet is situated between those that are most similar regarding women’s incorporation. Beyond the inclusion of the GPS as an important theoretical predictor, the multilevel statistical approach allows for inclusion of other contextualizing control variables. Given the role of ideology and the economy in how governments are evaluated, as robustness checks, we include a dummy variable for left government, the percent unemployment (measured January the year the president is inaugurated with inaugurations in May in Costa Rica, January in the United States, and March in Uruguay), and their interaction. All results presented here are robust to the inclusion of ideology and economic predictors as shown in the appendix.
By studying appointed ministers nested within individual administrations’ initial cabinets, we hold party constant (i.e., the president’s party), though appointees can come from different factions within a party. Only three ministers, all in the United States and all men, are not from the president’s party.Footnote 4 In studies of candidates in US elections, the Democratic Party is associated with issues concerning social welfare, policies stereotyped as feminine, while the Republican Party is associated with issues related to defense and economics. Additionally, the issue priorities of the Republican Party are associated with masculinity, and the Democratic Party with femininity (McDermott Reference McDermott2016). Party issue ownership, which covaries with the sex of candidates, makes it difficult to disentangle the effect of party vs. candidate sex on media coverage. Here, because 97% of all ministers, and all women ministers, are from the president’s party, we avoid this problem.
Media and gender politics research often relies on print media coverage, generally in the major national newspapers, or the major newspapers for a candidate’s state, and most studies cover a one-to-two-month window preceding the election (Dunaway et al. Reference Dunaway, Lawrence, Rose and Weber2013; García Beaudoux et al. Reference García Beaudoux, D’Adamo, Berrocal Gonzalo and Gavensky2020; Hayes and Lawless Reference Hayes and Lawless2015; Meeks Reference Meeks2012).Footnote 5 Articles are found based on keyword searches for politician names. Some researchers exclude letters to the editor and op-eds from their analysis (see Aday and Devitt Reference Aday and Devitt2001; Escobar-Lemmon et al. Reference Escobar-Lemmon, Valerie Hoekstra and Caul Kittilson2016; Meeks Reference Meeks2012), while others include them (see Kahn Reference Kahn1994; Hayes and Lawless Reference Hayes and Lawless2015). We follow the research design choice made by Escobar-Lemmon et al. (Reference Escobar-Lemmon, Valerie Hoekstra and Caul Kittilson2016) in their cross-national study of high court appointments to exclude editorials because, as they explain, “policies of the different newspapers differ regarding editorials” (p. 260).
Commonly used measures in the literature are (1) volume of coverage (measured here as word count, % of article), (2) tone of the article overall, (3) tone of coverage of the candidate (minister in our analysis), (4) number of quotes, and (5) professional backgrounds and credentials (Aday and Devitt Reference Aday and Devitt2001; Álvarez Monsiváis Reference Álvarez Monsiváis2021; Bystrom Reference Bystrom and Murray2010; Escobar-Lemmon et al. Reference Escobar-Lemmon, Valerie Hoekstra and Caul Kittilson2016; Fernández-García Reference Fernández-García2016; García Beaudoux et al. Reference García Beaudoux, D’Adamo, Berrocal Gonzalo and Gavensky2020; Hayes and Lawless Reference Hayes and Lawless2015; Kahn Reference Kahn1994; Kittilson and Fridkin Reference Kittilson and Fridkin2008; Serignese Woodall et al. Reference Serignese Woodall, Fridkin, Carle, Murray and Santa Barbara2010; Van der Pas and Aaldering Reference Van der Pas and Aaldering2020). We add (6) the number of ministers covered in the article, which measures whether a minister received focused attention, or was primarily covered in group particles that describe the cabinet as a whole. Candidate studies often measure coverage of appearance, family, and traits. They also measure mentions of novelty labels such as “first,” which can have both positive and negative frames (in addition to the above sources see Atkeson and Krebs Reference Atkeson and Krebs2008; Cárdenas Arias Reference Cárdenas Arias2023; Meeks Reference Meeks2012; Verge and Pastor Reference Verge and Pastor2018). We collected data about these topics for all articles about ministers, but do not analyze them here because they were so rarely mentioned. Appendix D2, Tables D1–D4 and D8 provide detailed information about the very limited occurrence of mentions of being the first, appearance, family, and other traits (by country, presidential administration, and sex of minister). Appendix D1 also explains journalistic norms in Latin American countries and how these norms affect reporting style and shape what is accepted as news coverage, providing context on how the cross-national nature of this project limits our ability to engage with techniques employed by gender mediation scholars.
We use a major national newspaper in each country that was available for all necessary dates in the same news database, which is important as some databases use different algorithms (additional details on selection can be found in Appendix C). Thus, we used La Nación in Costa Rica, The New York Times in the United States, and El País in Uruguay. All are in the top two most consumed print medias for the country. Costa Rica’s La Nación is considered centrist. Uruguay’s El País is considered right-leaning. The New York Times is considered left-leaning in the United States. The lean is only potentially problematic in the United States because party is constant (center-left) across the Uruguayan administrations. However, partisan treatment of each cabinet in the United States should not differ between the men and women in the same cabinet.
Temporal windows varied across countries to tailor to nuances in cabinet minister selection and electoral realities. Costa Rican and Uruguayan presidents independently appoint their ministers, but US appointees face confirmation by the Senate. In Uruguay, the coverage is four weeks before inaugurationFootnote 6 and two weeks after (February 1–March 15) in both cabinets. In Costa Rica, the coverage is five weeks before and four-and-one-half weeks after (April 1–June 10).Footnote 7 Given confirmation delays for some nominees in the United States, the general weeks of coverage around each secretary’s confirmation is three weeks before and after, but the dates may not be the same for all cabinet secretaries. All articles within the designated windows that returned the minister’s name were analyzed if the article was determined to be about the person of interest.
Coders recorded the following for each article: article length (word count, number of paragraphs),Footnote 8 number of quotes from the minister,Footnote 9 overall tone of the article, tone of the article when discussing the minister, number of ministers covered in the article, mentions of the minister’s experience/credentials. The dataset includes 1864 articles. However, articles often include more than one minister, so we have 2028 minister-article observations. Descriptive statistics about cabinets, ministers, variables, and information about missing data are presented in Appendix E. Article tone was coded 1 for negative tone, 2 for balanced/neutral tone, and 3 for a positive tone. We code overall article tone, and tone of coverage of the minister because many articles included both positive and negative coverage, especially if more than one minister was included in the article. A second coder coded portions of the dataset for each country, which is of particular importance for article/minister tone, which are the most subjective measures. Kappa’s Cohen’s test indicated strong intercoder reliability (please see Appendix F for greater detail).
Examination of Media Treatment
Hypothesis 1 predicts fewer differences in media treatment of women and men cabinet ministers where women’s incorporation into government has been greater. We use the cabinet GPS score to operationalize women’s incorporation, with higher GPS scores indicating greater incorporation. We interact the minister’s sex with their administration’s GPS score to proxy whether differences in men’s and women’s media coverage diminish as women become more mainstream in executive branch politics.
To test for quantity of coverage, we examine overall word count, percent of paragraphs (sentences for Uruguay) pertaining to the minister, number of quotes from the minister, and number of ministers covered in the article (Table 2). When appropriate, hierarchical linear models were used for continuous models, and mixed-effect negative binomial models were used for count models.
Note: *p<0.1; **p<0.05; ***p<0.01. Models 1 and 2 present Hierarchical Linear Regression results. Models 3–6 are Ordinary Least Squares Regression results, as statistical tests indicate that there are no model fit increases when allowing random intercepts at the administration level. Discrepancies across the number of observations are explained in Appendix E1. Models 3 and 4 only include Costa Rica and the United States. Models 5 and 6 only include Uruguay. Models 7–12 present Mixed-Effects Negative Binomial (MENB) Regression results. Positive alpha parameters (dispersion parameters) indicate that a MENB fits better than a mixed-effect Poisson model. Models with word count controls (9, 12) are robust to the use of standardized rather than raw word count measures.
Notable regarding quantity of coverage is that minister sex and GPS are often suggestively significant predictors of amount of coverage, but their interaction is significant only for the number of ministers mentioned in an article. Given our theoretical interest in whether minister sex is less impactful as the GPS increases, we focus on the significant interaction, but briefly contextualize the other significant results.
For the United States and Costa Rica % paragraphs captures coverage, and shows us that in those countries, women receive approximately 2.4% more coverage than men, holding GPS constant (model 4). However, given the near significance of the interaction (p=0.119), it appears that the lower levels of GPS in the United States are primarily driving these results with women in the Bush cabinet getting 5.5% and women in the Obama cabinet getting 3.8% more paragraphs than men in those cabinets. For % sentences (model 6), members of the Mujica cabinet receive more coverage, on average, than members of the Vázquez cabinet (approximately 2.7% more), but the interaction is not significant. Regarding number of quotes (model 9), on average, a one unit increase in GPS results in 0.508 more quotes on average per article. When considering predicted probabilities, in the Mujica administration (GPS = 0.326) a woman averages 0.672 quotes, but a woman minister in the Solís administration (GPS = 1.358) 1.171. Generally, more inclusive administrations are associated with more quotes by both men and women, but less space dedicated to ministers overall.
When we turn our attention to number of ministers mentioned per article (Table 2, Models 10–12), we see a more nuanced pattern. Women ministers are consistently “sharing” articles with more ministers, which is an indication of unequal treatment of these women politicians. However, examination of the role of gender over levels of the GPS, shows that the number of ministers/articles is most divergent between men and women at the lowest levels of the GPS and closes as the GPS increases. Women at a GPS of 0.326 (equivalent to the Mujica administration) share articles with 2.39 ministers on average, while men share with only 1.85 ministers on average. When the GPS is equal to 1.526, this gap closes to 2.42 and 2.31 ministers respectively. Overall, when concerning whether inclusion of women reduces gaps in quantities of coverage for women compared to men, our tests support Hypothesis 1 when we measure quantity as number of ministers in article and (suggestively) % of paragraphs covering the minister.
We also expect the quality of coverage of women ministers to be more like that of men when there is greater incorporation of women into government. Here we examine tone of the article and tone of coverage of the minister within the article (Table 3, Models 1 and 2 respectively). Models of tone present multilevel Ordered Logistic Regression Coefficients with tone coded from 1 (negative), 2 (neutral), to 3 (positive), and Figure 1 presents the predicted percentage share of each tone level by minister sex over levels of GPS.
Multilevel Ordered Logistic Regression Coefficients. Standard errors in parentheses.
*** p<0.01, ** p<0.05, * p<0.1. Models present the fixed effect coefficients for gender, GPS, and their interaction, the cut points (representing the log odds of being in a higher category when all predictors are zero), and the random effect (intercept) for administration-level variation that is important in explaining variation in tone, beyond what is captured in the GPS (measured at the administration level).
This analysis shows that gendered media treatment of ministers is contingent on women’s incorporation. For both overall tone and minister tone, the interaction gender*GPS is significant. The impact on a single article’s tone is very small, but when considering the percentage of each tone of coverage culminating over many articles, there appear to be meaningful differences. For example, when we compare tone of overall coverage for women at the lowest and highest levels of simulated GPS (0.326 and 1.826), women are predicted to move from 14.49% negative, 69.89% neutral, and 15.62% positive coverage to 9.28% negative, 67.40% neutral, and 23.31% positive. By comparison, for men there is little change in overall coverage tone, as men are predicted to move from 12.97% negative, 69.95% neutral, and 17.38% positive to 12.63% negative, 69.55% neutral, and 17.81% positive. This suggests that women’s incorporation may not change how men are covered but may improve how women are covered. Turning to predicted probabilities for minister tone, women are predicted to move from 8.46% negative, 77.53% neutral, and 14.05% positive to 3.81% negative, 69.53% neutral, and 26.65% positive when GPS moves from lowest to highest simulated values, and for men they move from 8.47% negative, 77.54% neutral, and 13.99% positive to 6.43% negative, 75.74% neutral, and 17.83% positive. Increases in women’s inclusion in the cabinet are associated with particularly notable increases in positive tone coverage of women ministers. Generally, these results show us that as GPS increases, women and men see increases in positive coverage, but more so for women than for men, providing suggestive evidence for more equal coverage as women’s incorporation increases.
Our final test for Hypothesis 1, and part of our test of Hypothesis 3, is about mentions of minister qualifications. We code a binary variable for whether the article mentions the minister’s experience or qualifications and use mixed-effects logistic regression with minister gender and GPS as IVs (Table 4 and Figure 2). The main effect of gender is significant, but counter to our expectations in H3, the media mentions qualifications of women ministers more than for men at lower levels of GPS and less at higher levels of GPS. At the lowest simulated GPS level, women are predicted to have their qualifications mentioned 26.28% of the time, and for men that rate is 18.61%. When we move to the highest level of simulated GPS, women will receive an experience mention just 16.55% of the time and men 20.80%, an almost 10% drop in likelihood for women.
Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Mixed-Effects Logistic Regression Coefficients (interpretable as log odds).
Importantly, these overall patterns hold when we control for the actual level of a minister’s experience (Appendix D2, Tables 5–7 provide details by case of the minister experience/credentials mentioned by the media, showing that most mentions are positive). Most ministers in these cabinets had extensive experience related to the policy area of their portfolio. We code “extensive experience” following Escobar-Lemmon and Taylor-Robinson (Reference Escobar-Lemmon and Taylor-Robinson2016, 80–81), with 28 of 31 women, and 50 of 65 men in our dataset having such a background, which means that these ministers had credentials that could have been covered in the newspaper. Actual experience is coded for low experience or high experience related to the portfolio. For example, Vázquez’s education minister was a medical doctor, so she was coded as low experience, whereas Vázquez’s health minister was also a medical doctor, and given the portfolio of health, he was coded as high experience.Footnote 10 Appendix H3 provides tests for differences in minister experience by gender and country and finds no significant differences. Even when controlling for actual experience, there is more experience coverage for women than men when there is less incorporation of women into the cabinet, but as women’s incorporation increases, men get more experience mentions. This suggests that the media may be treating women differently when they are relatively novel, just not in the traditional way of highlighting their novelty, but rather in highlighting their qualifications.
In sum, for some measures of quantity or quality of coverage we find support for H1: coverage of women is more equivalent to coverage of men with greater incorporation of women into the cabinet. However, regarding coverage of women’s experience for their posts, women received more coverage of their credentials when women were less common in the cabinet.
For H2, our interest is in whether women’s coverage is negative if they hold stereotyped-masculine posts (e.g., economics, agriculture, transportation, labor, foreign policy, defense/security), compared to posts that comport with gender stereotypes (e.g., education, health, social welfare, women’s issues, culture). To test this, we interact minister gender and the gender stereotype associated with the post. First, we note that masculine posts get more media attention: 68% of the articles are associated with masculine posts, 14% with neutral posts, and 18% with feminine posts (Table E1 in Appendix E).
Since tone of coverage was different for men and women ministers, and differences diminished in contexts in which women are more commonly seen in government, here we examine overall tone and minister tone (Table 5 and Figure 3). Post stereotype is a categorical variable, and we use Feminine Stereotype posts as the reference category.
Multilevel Ordered Logistic Regression Coefficients. The reference category is Feminine Stereotype. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Models 1 and 2 present the fixed effect coefficients for gender, post stereotype, and their interaction, the cut points (representing the log odds of being in a higher category when all predictors are zero), and the random effect (intercept) for administration-level variation that is important in explaining variation in tone.
For both Overall and Minister Tone, the interaction gender*masculine stereotyped post is significant, with the coefficient for gender being significant for minister tone (and approaching significance for overall tone, p = 0.147), suggesting that men and women also receive different coverage when occupying feminine stereotyped posts. Counter to our expectations for H2, it appears that incongruency with the post’s stereotype improves the tone of coverage for individual ministers. Women who hold masculine stereotyped posts receive less critical coverage than men in masculine posts, while women receive more critical coverage than men in feminine posts.
When we consider the predicted probabilities of negative, neutral, and positive minister-specific coverage over these posts for men and women, we get a better picture of these unexpected gender effects (Figure 3b). A man in a masculine post is predicted to get 7.88% negative, 78.21% neutral, and 13.91% positive coverage while a woman is expected to get 4.76% negative, 64.71% neutral, and 32.70% positive coverage. In feminine posts, in contrast, men are predicted to get 4.67% negative, 73.74% neutral, and 21.59% positive coverage while women are predicted to get 8.00% negative, 78.28% neutral, and 13.71% positive coverage. We see a similar pattern for overall tone. So, in both posts, the more critical coverage results are not driven by more negative than positive coverage, but a relative increase in neutral and negative coverage and relative decrease in positive coverage when the minister’s gender is congruent with post stereotype.
While these findings contradict our H2, they add to the overall picture of findings indicating that women in office are treated differently than women who are running for office. They suggest that even when women hold posts that clash with gender stereotypes the media presents them as able “to do the job.”
Finally, we examine whether media mentions of minister qualifications vary with the gender stereotype of their post. We use a binary variable for whether the article mentions the minister’s qualifications, and mixed-effects logistic regression with minister gender and the gender stereotype of the post as predictors (Table 6). Overall, women are less likely to get experience mentions, which support H3, but this only holds for feminine and neutral posts, running counter to our expectations for H2. As visualized in Figure 4, in feminine posts, men will get an experience mention in 36.43% of articles, while women will only get a mention in 14.60% of articles. In masculine and neutral posts, 16.70% and 26.85% of articles will mention men’s experience, respectively, but women will receive experience mentions in 21.20% and 24.33% of articles. When the minister’s sex is incongruent with a post’s stereotype the media may help establish a minister’s expertise, enhancing their credibility for the job.
Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Mixed-Effects Logistic Regression Coefficients (interpretable as log odds). Feminine Stereotype is the reference group. DV: Binary indicator if the article mentions the minister’s experience.
Conclusion
This study of six initial cabinets in three presidential democracies finds more similar media treatment of women and men cabinet ministers when there is greater incorporation of women into government. Women ministers also receive positive coverage when they hold masculine stereotyped policy portfolios. These are positive findings about how major newspapers treat women cabinet ministers, and they contrast with the treatment that is often found for women who are running for office, and with the gendered coverage of heads of government.
Differences in treatment of women, compared to men in these presidential cabinets are small regarding quantity of media coverage, regardless of extent of women’s incorporation into government or type of portfolio. However, for quality of coverage we observe more equal treatment of women and men where the Gender Power Score is higher, which supports arguments by Hayes and Lawless (Reference Hayes and Lawless2016) and Wagner et al. (Reference Wagner, Trimble, Curtin, Auer and Woodman2022) that how the media treats women will be more similar to men when women are no longer novel in politics. We cannot, however, claim that greater incorporation of women into the halls of power causes the more equal treatment. Presidents may appoint more women after they observe the media treating women as competent political players, also possibly observing positive coverage in other areas, such as business.
This initial cross-national study of cabinet ministers adds to the literature on gendered media. Gendered media treatment of cabinet ministers appears less common than gendered treatment of candidates for elected office. These findings regarding women ministers are especially interesting given the specific cabinets studied. This positive coverage of women ministers is interesting in Costa Rica considering the negative press received by President Laura Chinchilla (PLN 2010–14) by the end of her term when her public approval was very low. Negative reviewers of a woman president did not carry over to the women appointed to President Solís’s (PAC) initial cabinet in 2014. Positive coverage of women ministers is interesting in Uruguay given the historically low level of incorporation of women into Uruguayan politics. The expansion in cabinet seats filled by women in President Vázquez’s initial cabinet in 2015 did not produce backlash in the form of a negative tone in media coverage of women ministers overall, though one woman, Carolina Cosse, Minister of Industry, Energy, and Mines, had a negative relationship with the press. Positive tone of coverage of women is interesting in the United States, particularly in President Obama’s initial cabinet where women held the powerful secretary of state post, the masculine stereotyped homeland security post, as well as health and human services tasked with health care reform that was an important part of Obama’s agenda. The lack of backlash effects is interesting, especially given findings of overtly gendered coverage of women candidates for the executive in Latin America (Dos Santos and Jalalzai Reference Dos Santos and Jalalzai2021).
Given the unique challenges and institutional settings of our cases, we cannot be certain that findings about women in the cabinets included here would extend to all contexts or time periods. Study in more countries, presidential and parliamentary, is needed to assess if the findings are generalizable. Future study should examine media treatment of women defense ministers, which is particularly important given the still low representation of women in that highly masculine post, and a topic we could not examine due to a lack of women in defense/security portfolios during these administrations.Footnote 11 In addition, these findings are based on the early honeymoon period of these administrations. Sexism could come later when the administration loses popularity as controversial policy reforms are implemented, or if a president who highlighted women in their initial cabinet to signal gender equality, stops caring about signaling their commitment.Footnote 12
Future work should examine whether women ministers are criticized more harshly by the media than men when there is a policy problem associated with their department, or if the media presents women ministers as capable of handling the problem. We explored whether women are treated differently than men when they hold politically hot posts, by comparing women and men ministers in high visibility and low visibility posts at the beginning of these administrations (full results in Appendix G3). That analysis confirms that women are not limited to low visibility posts or to masculine stereotyped posts during good times. At least during the honeymoon period of the administration, women in high visibility posts receive more positive coverage than men.
Future work also is needed to explore if media treatment of women ministers becomes more negative over time, for example, if the president pledged to appoint a parity cabinet, or a cabinet that represents the race/ethnic and generational diversity in the country (such as President Boric’s initial cabinet in Chile in 2022). Positive coverage generated by a cabinet that looks new may turn to harsh criticism of women and other new types of ministers when the challenges of governing become apparent. Alternatively, the media could ignore women ministers once the novelty of a more gender balanced cabinet is no longer being highlighted by the president.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/lap.2023.42
Competing interests
The authors have no conflict of interest to declare.