Beyond Neutrality: A Longitudinal Analysis of Gender,Party,and Media Visibility in the Tone of US Political Coverage

Abstract

Research shows that female politicians often receive more negative news coverage than their male counterparts, which can undermine voter support and contribute to their underrepresentation in politics. Despite the increase in politically active women and heightened attention to gender issues in political discourse in the United States, the past decade has also witnessed a rise in divisive rhetoric and political polarization. These changes not only transformed the political environment but also may have impacted media coverage of political figures. Despite the evolving sociopolitical landscape, existing research on news sentiment lacks a longitudinal perspective. This study uses transformer-based classifiers for sentiment detection and state space models for time series analysis to provide new longitudinal evidence on the evolution of positive and negative coverage of 1,095 US politicians in almost 900,000 news articles during the 2010s. The results indicate that while some female politicians experienced more negative coverage than men, the pattern varied depending on the politicians’ party affiliation and media visibility. In both election and routine periods, the media portrayed female Democrats, who received moderate levels of media coverage, less negatively than male Republicans, but more critically than female Republicans and male Democrats. Notably, Democratic and Republican women received slightly more positive coverage in most election years of the 2010s. Highly visible politicians, especially women, received more negative attention overall. The study and its findings advance our understanding of how news sentiment varies across different groups of politicians and how it fluctuates between election and nonelection periods.

The tone of news reporting has a profound impact on candidate popularity, with positive coverage potentially boosting voter support and negative portrayals decreasing it (Aaldering et al. 2018). Coverage sentiment may be significantly influenced by gender stereotypes that associate female politicians with women in general, rather than with strong leaders. This incongruence has led to the hypothesis that female politicians receive more negative coverage than their male counterparts; however, the evidence remains inconclusive (Van der Pas and Aaldering 2020). In addition to gender, other characteristics of politicians, such as party affiliation and media visibility, may also influence the tone of reporting. High-profile politicians may be subjected to more negative coverage due to heightened media scrutiny, while Republican female politicians may face more criticism due to a perceived mismatch between the masculine image of their party and traditional feminine roles (Hayes 2011). The intersection of these factors has rarely been studied; however, if differences in media portrayals exist, these narratives may potentially create barriers to the political representation and advancement of women from different parties and levels of prominence.

In terms of women’s participation in politics, the past decade has witnessed a surge in politically active women and an increased focus on gender issues in political discourse (Pew Research Center 2021). Research suggests that these developments have led to changes in gender stereotypes, affecting both voter perceptions (Van der Pas et al. 2023) and media portrayals of politicians (Andrich et al. 2023). However, research on gendered media coverage frequently lacks a longitudinal perspective, with many studies relying on data collected more than a decade ago or focusing on high-profile politicians during narrow pre-election periods.

This exploratory study examines media coverage trends from 2010 to 2020, using a large dataset with sentiment classification models fine-tuned with custom-labeled data and time series analysis to track coverage sentiment over time. The research examines differences in positive and negative media coverage of female and male politicians, accounting for their media visibility and party affiliation (RQ1), and how these differences have evolved (RQ2). Findings show that some female politicians received more negative coverage than men; however, this varied by party affiliation and visibility. Female Democrats with moderate visibility received less negative coverage than male Republicans, though more critical coverage than female Republicans and male Democrats. Additionally, election years saw slightly more positive coverage for both Democratic and Republican women. However, highly visible politicians, especially women, received more negative attention overall. Finally, while female Republicans did not receive more negative coverage overall, by the decade’s end, they were the only group whose negative coverage increased while positive coverage decreased.

This article contributes in several ways. Theoretically, it challenges traditional gender stereotypes, showing that female politicians do not consistently receive more negative coverage than males and that less stereotypical portrayals have not resulted in more positive reporting. Empirically, it analyzes how gender, party affiliation, and media visibility shape media coverage tone. Methodologically, it uses a large, diverse sample of politicians and news items, incorporating advanced sentiment analysis and time series techniques.

Literature Review

Gender Differences in the Tone of Political Reporting

The differential treatment of female and male politicians in the media, including in terms of sentiment, can be attributed to the pervasiveness of gender stereotypes (Van der Pas and Aaldering 2020). These stereotypes are deeply rooted in societal observations of traditional roles, namely the assumption that women are primarily responsible for caregiving and men for financial security (Eagly 1987). Gender stereotypes are harmful in that they both describe and prescribe how social groups should behave (Burgess and Borgida 1999). For instance, they imply that women lack certain characteristics deemed necessary for effective political leadership. This is largely because leadership has historically been associated with masculine and agentic traits, such as dominance and aggression (Koenig et al. 2011), which are also favored by voters (Rosenwasser and Dean 1989). Thus, journalists may be inclined to portray female politicians less positively due to (1) the incongruence between the perceived characteristics of women and the expected attributes for political leadership (Eagly and Karau 2002) and (2) the perceived failure to project a balanced image of a political leader who is neither overly feminine nor overly masculine (Gershon 2012). This may lead to coverage that questions female candidates’ viability, leadership abilities, communication skills, and campaigning strategies (Kahn 1994; Kahn and Goldenberg 1991; Wagner et al. 2019), overemphasizes their combative behaviors, and uses more aggressive language (Gidengil and Everitt 2003; Van der Pas and Aaldering 2020).

Given this context, it is not surprising that research on gender differences in news coverage often expects more negative sentiment in the portrayal of female politicians compared to their male counterparts (Van der Pas and Aaldering 2020). However, despite a general trend toward neutral coverage for both genders (Bystrom et al. 2001; Heldman et al. 2005), no consensus exists on whether male or female politicians are portrayed more positively or negatively in the media, or whether any significant differences exist at all (Kittilson and Fridkin 2008; Van der Pas and Aaldering 2020; the analysis of members of parliament by Semetko and Boomgaarden 2007). For example, some studies show that female politicians garner more critical media coverage (Kahn 1994; Kahn and Goldenberg 1991), while others indicate more favorable reporting for female candidates in US elections (Banwart et al. 2003; Bystrom et al. 2001; Dunaway 2012; Robertson et al. 2002) and elections in Europe, including Belgium, Norway, and the Netherlands (D’Heer et al. 2021; Lühiste and Banducci 2016; Midtbø 2011; Van der Groot et al. 2021). In contrast, some evidence suggests that male politicians receive more negative reporting, both in the United States (Banwart et al. 2003; Bystrom et al. 2001; Smith 1997) and European electoral contexts (Lühiste and Banducci 2016).

The tone of news coverage critically influences political careers by shaping public perception, with positive portrayals potentially increasing voter support and negative coverage possibly reducing voter trust and turnout (Aaldering et al. 2018; Ansolabehere et al. 1994). However, negative framing particularly harms female politicians, reinforcing perceptions of them as outsiders and hindering their representation in politics (Van der Pas and Aaldering 2020). Nevertheless, the past decade has seen more women, including prominent figures such as Nancy Pelosi, Hillary Clinton, and Kamala Harris, running for political office, increasing their media visibility and gradually transforming public stereotypes about female politicians (Van der Pas et al. 2023). Indeed, research shows that as social roles change, stereotypes evolve accordingly. For example, as more women obtain an education and enter the workforce, they are increasingly perceived as more agentic and competent (Diekman and Eagly 2000; Diekman et al. 2004; Eagly et al. 2020). Both stereotypes about women leaders and women politicians are also shifting, blending positive agentic and communal characteristics (Beaman et al. 2009; Eagly and Sczesny 2009; Koenig et al. 2011; Van der Pas et al. 2023). Moreover, studies on the representation of women in text indicate that they are increasingly associated with intelligence (Garg et al. 2018), while associations with feminine traits have diminished over time (Bhatia and Bhatia 2021). Importantly, recent research in political communication shows that media portrayals are similarly evolving to link women more with leadership qualities (Andrich et al. 2023). These changes suggest that media coverage of female politicians may have become less critical over time, reflecting an adaptation in stereotypes and societal expectations.

Other Determinants of Coverage Sentiment

In addition to the gender of politicians, other factors that reflect structural aspects of contemporary US politics, including increased male visibility in the news and partisan stereotypes, can significantly influence the tone of political reporting. For instance, although increased media visibility can enhance voter recognition of candidates (Gattermann and De Vreese 2017) and improve electoral success (Hopmann et al. 2010), it often comes with heightened scrutiny, as visible politicians are more likely to be held accountable for the actions of their party or administration. Election frontrunners, for example, frequently attract more negative coverage than trailing candidates, who are often portrayed more positively (Graber 2006; Miller et al. 2010). Findings on gender and media visibility are mixed, with some studies suggesting that female politicians receive less coverage than male politicians (Johnstonbaugh 2018; Lühiste and Banducci 2016; Shor et al. 2019) and others reporting no gender differences (Atkeson and Krebs 2008; Gershon 2012; Hayes and Lawless 2015). Thus, if male politicians receive more media attention, they may also garner more negative coverage than their female counterparts. Conversely, female politicians may still be regarded as a novelty and thus may receive more, and more unfavorable, media attention than their male colleagues (Kahn and Goldenberg 1991; Lühiste and Banducci 2016). In addition, the coverage of highly visible female politicians holding or seeking influential political positions may be more critical due to the increasing incongruity between leader stereotypes and the feminine gender stereotype, as higher-status leadership roles are more strongly linked to masculinity (Koenig et al. 2011; Midtbø 2011). Studies indicate that prominent female politicians, including presidential candidates (Lawrence and Rose 2010), party leaders (Gidengil and Everitt 2003), ministers (Fernandez-Garcia 2016), and gubernatorial candidates (Rausch et al. 1999), often receive more critical coverage than their male counterparts. Yet, some research finds that coverage of female leaders is more neutral (Smith 1997), more positive (Bystrom and Dimitrova 2014; Semetko and Boomgaarden 2007; Zulli 2019), or similar in tone to that of male leaders (Heldman et al. 2005). Despite its importance, the intersection of gender and media visibility remains underexplored, as most studies focus only on high-profile figures.

Gender stereotypes are further complicated by their overlap with partisan stereotypes. The Democratic Party is often associated with traditionally feminine traits (e.g., compassion and empathy), while the Republican Party is linked to masculine qualities (e.g., tough and strong leadership; Hayes 2011). This intersection of gender and partisan stereotypes can shape perceptions of female candidates and influence their media portrayals. For female Republicans in particular, navigating both the masculinity associated with their party and societal expectations of femininity presents unique challenges. This incongruence may lead to more negative evaluations by voters (Bauer 2017), as female Republicans may be perceived as incompatible with both the Republican Party’s image and conventional gender roles, thus appearing to conflict with the identities of a Republican, a politician, and a woman (Bauer 2017; Dolan 2014). Nevertheless, research on the intersection of gender and party in coverage sentiment remains limited, with most studies focusing on influential figures. For example, Elizabeth Dole and Michele Bachmann did not receive more negative coverage than male Republicans in their presidential nomination bids in 1999 and 2012, respectively (Bystrom and Dimitrova 2014; Heldman et al. 2005), while Hillary Clinton experienced more negative attention than male Democrats during her 2008 presidential campaign (Lawrence and Rose 2010; Miller et al. 2010). Additionally, studies generally show that Democrats receive more favorable media coverage than Republicans, as seen in analyses of US Representatives (Gershon 2012) and vice-presidential candidates (Conroy et al. 2015).

Present Study

The reviewed literature underscores how traditional gender roles and leadership expectations can lead to less favorable media portrayals of female politicians, an issue potentially complicated by factors like media visibility and political affiliation. High-profile politicians face greater scrutiny, often resulting in negative coverage (Miller et al. 2010), and female Republicans may receive critical coverage due to the mismatch between their party’s masculine image and traditional feminine roles. However, the interaction of these factors with gender remains under-researched.

Over time, perceptions of women, including female politicians, have become more accepting of their leadership abilities (Van der Pas et al. 2023). While media portrayals of female politicians in the United States have reflected these changes over the past decade (Andrich et al. 2023), it remains unclear whether there has been a comparable shift in the tone of political coverage over time. Despite progress, the United States still trails many nations in female political representation, often lacking the supportive policies like quotas found elsewhere (Brechenmacher 2018; Inter-Parliamentary Union 2018). This highlights the need to examine the unique challenges of gender equality in the United States within its influential political landscape.

Much existing research relies on data collected more than a decade ago (Smith 1997) or focuses on high-profile politicians in brief pre-election periods (Heldman et al. 2005). This study fills these gaps by analyzing a large media dataset from 2010 to 2020, using advanced BERT-based (Bidirectional Encoder Representations from Transformers; Devlin et al. 2019) sentiment analysis and time series methods. It provides new longitudinal insights into how gender, visibility, and party affiliation intersect in political reporting throughout the 2010s. This article addresses the following research questions (RQs):

• RQ1. How does the share of positive and negative coverage vary for female and male politicians with different levels of media visibility and from different parties?

• RQ2. How has this difference changed over time?

Methods

Sample

Politicians

A dataset of politicians was compiled, encompassing their names, unique identifiers, gender, party affiliation, and political positions held between 2009 and 2020 (n = 1,095). Detailed information on members of the 111th to 116th US Congress was obtained using the ProPublica Congress application-programming interface (API). ¹ Additionally, information on politicians serving in the Cabinet of the United States from 2009 to 2021 was gathered by scraping relevant Wikipedia pages. ² The final sample included members of the US Congress, Cabinet officials, and presidents who served during the analyzed period. Excluding politicians without party affiliation and those who received no coverage during the analyzed period (15 males and 6 females), the final dataset consisted of 1,074 politicians. The last decade saw 160 female Democrats and 57 female Republicans. Federal-level politicians were selected because they are likely to attract significant media attention and engage in national discourse. This selection also aligns with the choice of national media sources, which extensively cover federal politics.

Media Reports

LexisNexis APIs were used to collect the entire corpus of digital copies of news stories produced by 21 major US newspapers, magazines, and online outlets. ³ Depending on the outlet, unique readerships ranged from 6.5 to 125 million per month (see Supplemental Appendix A for more details). The sample was constructed to reflect a diverse media landscape, primarily featuring center-leaning outlets. The New York Times leaned liberal, while National Review and The Weekly Standard leaned conservative, according to the classification by Weeks et al. (2023; see Supplemental Appendix A for more details). Subsequent supplementary analyses found no significant differences in coverage tone by outlet leaning (see Supplemental Appendix A). Although the sample includes one of the largest national media outlets, it admittedly does not represent the full spectrum of media in the United States. The final dataset included 891,116 news stories published between January 2010 and December 2020, with identical stories eliminated.

Text Preprocessing

Punctuations, numbers, and special characters were removed. As politicians can be mentioned in news articles in various ways (e.g., Donald Trump, Donald J. Trump, or President Trump), each mention of every politician from the sample was replaced with the person’s unique identifier (e.g., donald_trump; detailed information can be found in Supplemental Appendix A).

Sentiment Classification

Sentences mentioning politicians were extracted from the media dataset, resulting in 7,262,946 sentences. From these, 8,192 random sentences were selected for training, validation, and testing. Three student assistants classified the sentences as positive, negative, or neutral based on Van Dalen et al.’s (2012) definitions. Specifically, a sentence was labeled positive if the politician mentioned in the sentence was presented in a positive light, for example, if her or his qualities, abilities, and merits were highlighted. A sentence was labeled as negative if the politician mentioned in the sentence was described in a negative light, for example, if the politician’s failures and inabilities were highlighted, or if a clear negative tone and critical evaluation of the politician could be distinguished.

All sentences were independently labeled by three student assistants, with the final labels determined by majority agreement among them. A random sample of sentences was annotated by the expert (the author of the paper), and these ratings were then compared to the majority labels to evaluate the inter-rater reliability between the expert and the majority (Cohen’s kappa of 0.81). Once the student assistants had completed labeling the data, the author acted as a fourth rater to resolve any cases where a majority decision could not be reached. The human-defined labels served as a gold standard for fine-tuning a state-of-the-art, pretrained RoBERTa model, an optimized variant of the BERT model (Devlin et al. 2019) by Liu et al. (2019). These models, trained on vast data with an efficient architecture, are highly accurate in understanding context and detecting linguistic properties, including sentiment. Specifically, the RoBERTa model was initially configured as a three-class sentiment classifier by Barbieri et al. (2020) on more than 124 million English texts. In this study, the model was further fine-tuned for the specific context of political news using the abovementioned human-labeled training data. The resulting fine-tuned model performed exceptionally well at identifying neutral sentiment (F₁ score = 0.93) but showed moderate performance in the detection of negative (F₁ score = 0.6) and positive sentiment (F₁ score = 0.59). Therefore, it was only used to detect neutral sentences (n = 6,296,539). To classify the remaining sentences as positive or negative, a binary BERT-based classifier called SiEBERT (Hartmann et al. 2023), which was pretrained on various sentiment-related tasks, was fine-tuned using the training set that only contained positive and negative sentences. Random oversampling was used to address the overrepresentation of the negative class. The classifier performed well on the held-out test set for both positive (F₁ score = 0.69) and negative classes (F₁ score = 0.97). Ultimately, 60,690 sentences were classified as positive, and 905,717 sentences were negative. Details on data annotation, inter-rater reliability measures, and fine-tuning can be found in Supplemental Appendix B.

Measures

Sentences mentioning the same politician were aggregated within each article to calculate an average sentiment score indicating whether the coverage was positive, negative, or neutral. For time series analysis, these scores were further aggregated by month, calculating the proportions of positive, negative, and neutral articles for each group of politicians (i.e., female/male and Democrat/Republican). A monthly time unit was selected due to limited coverage for most politicians, which led to many missing values at shorter intervals (see Supplemental Appendix E for robustness checks with other time units).

As is common with such datasets, a small number of politicians received most of the media coverage. To ensure diverse representation, politicians were grouped by media visibility: those with over 1 percent of monthly coverage were classified as highly covered, while the remaining politicians were categorized as moderately covered. Media attention shifts with events, crises, and legislative actions, at times giving less influential politicians coverage levels comparable to high-profile figures. Thus, the sample of highly covered politicians was adjusted monthly to capture fluctuations in media focus. Female Republicans received high coverage only sporadically; thus, their results are included in the Supplemental Appendix (Figures C4 and C5). A sensitivity analysis excluding politicians with more than 10 percent coverage showed minimal changes in trends, except for male Democrats (Figures E7 and E8, Supplemental Appendix E).

Statistical Modeling

The study aimed to describe changes in positive and negative media coverage for highly and moderately covered female and male Democrats and Republicans (RQ1) and to model these trends over time (RQ2). To capture temporal variations, state space models (Kalman 1960) were used to incorporate monthly proportions of positive and negative coverage, allowing for smooth transitions without assuming data stationarity. These models account for temporal autocorrelation, separating genuine process variations from observational errors (Durbin and Koopman 2012), making them useful in political science for tasks like analyzing election poll data (Petropoulos et al. 2022). Unlike traditional time series, state space models conceptualize changes via unobserved state variables, using Bayesian inference to treat parameters as random and time-varying (Gelman et al. 2013). Smoothing algorithms optimize state predictions and uncertainties, integrating all available data for more accurate trend analysis (Durbin and Koopman 2012).

In the article, trends from state space model smoothing algorithms are termed smoothed trends, using Bayesian estimation techniques (Gelman et al. 2013). Uncertainty is expressed through 95 percent credible intervals (95% CIs are reported in Results in square brackets) based on 10,000 posterior draws from the posterior distribution, with a maximal Monte Carlo standard error of 0.008. The Bayesian 95 percent credible interval directly reflects a 95 percent probability that the true estimate lies within the interval, based on the observed data (Hespanhol et al. 2019). The median of these intervals is provided as the centrality estimate. Similar to previous studies (Bystrom et al. 2001; Dunaway 2012), positive and negative sentiments were modeled separately for both moderately and highly covered politicians using four distinct models. The prior distribution is induced by the state space model with hyperparameters, such as variances of the disturbances and measurement noise, fitted by maximum likelihood estimation. The modeling employed the KFAS package in R (Helske 2017). For mathematical details of the models, refer to Supplemental Appendix D.

Results

Negative Sentiment Toward Moderately Covered Politicians

The estimated monthly parameters of negative coverage for moderately covered politicians from January 2010 to December 2020 are shown in Figure 1. Generally, the proportion of negative coverage was low, raw values ranging from 0.4 percent to 26.8 percent. However, over time, negative coverage increased for most groups, with female Republicans seeing the most significant rise—a median increase of 2.4 percent, 95% CI [0.7, 4.1] by 2020. Conversely, negative coverage for Democratic men decreased by 1.6 percent [−2.5, −0.7] over the same period. Across all groups, negative coverage peaked every 2 years during the summer and fall of US federal election cycles. For instance, male Republicans experienced a median increase of 2.1 percent [1.1, 3.1] in 2014 and 1.9 percent [0.9, 3.0] in 2016 than in the nonelection years of 2013 and 2015, respectively. Conversely, in 2019, an atypical nonelection year, negative coverage was unusually high, but by 2020, it decreased for most groups—female Democrats, male Democrats, and male Republicans by 2.7 percent [−3.7, −1.7], 2.4 percent [−3.1, −1.7], and 2.7 percent [−3.9, −1.7], respectively. In contrast, female Republicans saw a slight increase of 0.2 percent [−1.3, 1.6] from 2019 to 2020.

Figure 1.

The evolution of negative news reporting on politicians with moderate coverage.

Overall, Republican men consistently received the highest proportion of negative coverage compared to other groups over the last decade. For instance, the gap between them and male Democrats widened from about 1 percent [−0.2, 2.1] in 2010 to 3.3 percent [2.2, 4.4] in 2020. They also garnered more negative stories in both election and nonelection years than female Democrats, notably peaking in 2016 with a median difference of 2.5 percent [1.5, 3.5]. Similarly, Republican men saw greater negative coverage than Republican women, except in 2020, when female Republicans experienced a slight increase in negative stories compared to their male counterparts by 1.4 percent [−0.2, 2.8].

Female Democrats generally faced more negative reporting compared to male Democratic and female Republicans, with differences becoming more pronounced over time. By 2020, the gap in negative coverage between Democratic women and men widened to 2.1 percent [1.1, 3.1]. Throughout the decade, female Democrats also consistently received more negative attention than female Republicans, peaking in 2012. However, in 2020, this trend reversed with Republican women receiving 2.6 percent [1.1, 4] more negative coverage than Democratic women.

Female Republicans and male Democrats generally received the least negative coverage, though female Republicans faced more negative reporting than male Democrats every election year, with the difference growing to 4.6 percent [3.3, 6] by 2020. Initially, Democratic men experienced slightly more negative coverage than Republican women during nonelection years, but this trend reversed by 2019, with the latter receiving up to 2 percent [0.9, 3.2] more negative stories.

To summarize, overall, in the analyzed dataset, negative reporting for all moderately covered politicians increased toward the end of the 2010s, peaking around federal elections. Male Republicans consistently faced the most negative coverage throughout the decade, while female Democrats received more negative attention than female Republicans and male Democrats. By the decade’s end, male Democrats had the least negative coverage. Although female Republicans generally experienced less negative coverage, by 2020, they faced the most unfavorable reporting of all groups.

Negative Sentiment Toward Highly Covered Politicians

The estimated parameters from the model for negative coverage of highly covered politicians for each month from January 2010 to December 2020 are shown in Figure 2. Overall, highly covered politicians received a greater range of negative reporting, from 3.4 to 38.7 percent, than moderately covered politicians. Over the decade, negative coverage increased for all groups, especially for men. Specifically, negative reporting for male Republicans and male Democrats rose substantially by 8.6 percent [5.4, 11.9] and 7 percent [4.2, 9.6], respectively, at the end of the last decade compared to its beginning. Highly covered politicians also showed seasonal patterns of coverage that align with election cycles. Female Democrats and male Republicans saw their peaks in unfavorable coverage during the 2016 election, while male Democrats saw their biggest spike during the 2020 presidential election.

Figure 2.

The development of negative reporting on highly covered politicians.

Throughout election cycles and in between, male Republicans often received more critical coverage than Democrats. For instance, the gap between them and Democratic men peaked at 6 percent [3.4, 8.4] in 2016 but decreasing to 1 percent [−2, 4] by 2020. Republican men also faced more negative coverage compared to Democratic women, with disparities reaching 6.5 percent [3.4, 9.6] in 2020. However, in 2016, female Democrats experienced the highest negative coverage at 7.5 percent [4.7, 10.3] more than male Republicans.

In the early 2010s, Democratic men garnered more negative coverage than Democratic women, with a 2.7 percent higher rate in 2012. By the mid-2010s, female Democrats saw a significant increase in negative coverage, peaking at 13.4 percent [11, 16] in 2016. However, in 2020, male Democrats again faced more unfavorable coverage, with a median difference of 5.6 percent [2.7, 8.4].

In sum, in the media outlets examined in this study, highly covered politicians experienced more negative attention overall than their moderately covered counterparts. Negative coverage generally increased over time and peaked during election cycles. Like their less prominent colleagues, highly covered male Republicans received the most unfavorable reporting, while negative coverage for female and male Democrats alternated throughout the 2010s.

Positive Sentiment Toward Moderately Covered Politicians

The estimated parameters from the model for monthly positive coverage of moderately covered politicians are illustrated in Figure 3. Overall, positive coverage was minimal, ranging from none to 5.1 percent. Only among female Democrats did it increase, by just 0.2 percent [−0.2, 0.6] from 2010 to 2020. Seasonal effects were less pronounced compared to trends in negative coverage, with Democrats seeing their highest peaks in positive stories in 2012 and Republicans in 2014.

Figure 3.

The trend of positive sentiment among politicians with moderate coverage.

In both election and routine years, Democratic and Republican women often garnered slightly more positive coverage than men politicians. For instance, during the 2018 midterms, Republican women had 0.6 percent [0.2, 1] and 0.5 percent [0.1, 1] more positive coverage compared to Republican and Democratic men, respectively. In the 2020 elections, female Democrats gained 0.6 percent [0.3, 1] and 1 percent [0.5, 1.2] more favorable reporting than male Democrats and Republicans, respectively.

Throughout the 2010s, the positive coverage of Republican and Democratic women alternated. For example, in 2013, female Democrats received 0.3 percent [−0.1, 0.7] more positive stories, while in 2015, female Republicans had 0.2 percent [−0.2, 0.6] more. Democratic men consistently outperformed Republican men in positive coverage, with the largest gap in 2013 at 0.4 percent [0.2, 0.6].

In summary, in this dataset, in contrast to negative coverage, all politicians with moderate media visibility received minimal positive reporting, with women experiencing more favorable coverage than men during election and routine periods. While coverage of female Democrats and Republicans alternated over time, male Democrats received more favorable coverage than male Republicans.

Positive Sentiment Toward Highly Covered Politicians

The estimated parameters from the model for monthly positive coverage of highly reported politicians are displayed in Figure 4. Positive coverage for highly visible politicians was low, ranging from 0.2 to 4.5 percent. Over the decade, positive coverage increased, particularly for male Democrats, who saw a median increase of 0.6 percent [0.3, 1] by 2020 compared to 2010. Seasonal peaks occurred mostly during election years, with significant increases for both Democratic and Republican men during the 2012, 2016, and 2020 presidential elections, while Democratic women saw major increases during the last two presidential elections of the decade.

Figure 4.

The evolution of positive reporting on politicians with high media visibility.

During all election and routine periods, female Democrats consistently received more positive coverage than male politicians. In 2016, they received 0.4 percent [−0.02, 0.7] more positive coverage than male Democrats and 0.5 percent [0.2, 0.9] more than Republican men. By 2020, the gap in positive coverage between Democratic women and Republican men expanded to 0.8 percent [0.4, 1.2], while the difference between female and male Democrats almost disappeared.

Generally, highly covered male Republicans received less positive coverage than other groups. For example, Democratic men consistently received more favorable media portrayal, peaking in 2020 with a difference of 0.7 percent [0.4, 1] compared to their Republican counterparts.

To summarize, highly covered politicians, compared to those with moderate visibility, experienced an increase in positive reporting over time. Like their less visible counterparts, female Democrats received more favorable coverage than male politicians, particularly during the last two elections of the 2010s. Among male politicians, Republicans consistently had the lowest share of positive coverage.

Discussion

Contrary to the common expectation of increased negative media coverage for female politicians (Kahn 1994; Kahn and Goldenberg 1991), this study presents a more complex picture. The analysis of a sample of national US media outlets revealed that, among moderately covered politicians, male Democrats received the least negative attention throughout the decade, followed by female Republicans. In contrast, female Democrats and particularly male Republicans faced heightened media scrutiny. This trend persisted for highly covered Republican men. On the Democratic side, highly reported men encountered more negative portrayals in the early part of the decade, while highly covered women faced more negativity in the second half of the 2010s. This is in line with previous research (Lawrence and Rose 2010; Miller et al. 2010), potentially reflecting a response to the growing participation of female Democrats at the highest levels of politics.

Regarding positive coverage, the findings align with existing evidence (D’Heer et al. 2021; Dunaway 2012; Midtbø 2011), demonstrating that moderately covered women, both Democratic and Republican, have consistently received more positive portrayal than their male counterparts across election and nonelection periods in the analyzed media dataset. A similar pattern emerged for highly covered female Democrats. Among men, Democrats received more positive reporting than Republicans; a trend observed for both moderately and highly reported politicians. However, the fraction of positive coverage was very small, as were the differences in favorable coverage between politicians. While the media have maintained a predominantly neutral tone over the past decade, this study finds an overall increase in negativity, consistent with earlier research (Bystrom et al. 2001), with the spike in negative coverage coinciding with election periods.

The findings suggest that coverage was generally more negative for men, especially those with high media visibility. While consistent with some prior studies (Banwart et al. 2003; Bystrom et al. 2001; Lühiste and Banducci 2016; Smith 1997), these patterns are not readily comparable to previous research. The impact of party affiliation and media visibility introduces a degree of complexity to these findings and prevents from interpreting them as a straightforward “black and white” outcome. Nonetheless, the following discussion explores the impact of nonneutral coverage based on existing studies and contextualizes the results within political and social changes that took place in the United States during the 2010s.

The relatively negative coverage of Democratic women in the present dataset may reflect recent shifts in female representation in US political institutions. It is possible that as more women entered politics, particularly within the Democratic Party (Pew Research Center 2021), they faced increased media scrutiny and criticism. This rising visibility of Democratic women may have led to more negative coverage, contradicting broader, positive changes in the acceptance of women in leadership (Andrich et al. 2023; Van der Pas et al. 2023). The effect of gender novelty has been measured in several contexts (Trimble et al. 2021), but not in terms of news sentiment. Future studies can examine how the increasing presence of women in political positions affects the tone of media coverage. Moreover, as races that feature female candidates often receive more game-frame reporting, which focuses on the competitive aspects of politics, highlighting who is winning or losing across various events like elections and legislative actions, and is typically negative (Kahn 1996), the coverage may have been influenced by the mere presence of women in elections. Future research can offer a more comprehensive examination of the differences in tone between various media frames. In addition, as the number of women in US government has grown, so has the number of minority women. The Democratic Party’s increased diversity in the last decade may have led to more negative media attention, as previous research shows that women of color are subject to more negative coverage than their counterparts (Gershon 2012; Ward 2016). However, more research is needed to confirm this speculation. Additionally, future studies should investigate whether nonbinary political figures and people of color face comparable levels of media negativity. This study’s findings, therefore, suggest that women in politics may still face persistent challenges, even within a party known for inclusivity. These obstacles can be detrimental to their careers, as negative media attention can lead to voters developing unfavorable impressions of female candidates during campaigns and perceiving them as lacking political expertise during routine times (Aaldering et al. 2018). The current analysis shows that female Republicans did not receive more negative coverage overall but were the only group to see an increase in negative and a decrease in positive coverage by decade’s end. This suggests a possible link between party and gender stereotypes, though further research involving potential voters is needed to confirm whether citizens still associate the Republican Party with masculinity. In addition, while the number of Republican women in the US Congress has increased over the past decade, they remain significantly underrepresented (Pew Research Center 2021). Therefore, future longitudinal studies are needed to determine whether the coverage sentiment of Republican women will be comparable to that of Democratic women.

The observed increase in negative coverage over the past decade across the analyzed outlets is emblematic of the heightened political divisiveness in the United States during the 2010s, shaped by evolving media dynamics in response to broader sociopolitical shifts. Early in the decade, emerging partisanship and populism were fueled by the slow recovery from the 2008 financial crisis, which led to growing economic disparities and public frustration with the government. These sentiments galvanized the rise of both right- and left-wing political and social movements, including the Tea Party movement, Occupy Wall Street, and Black Lives Matter. The country’s deepening economic, social, and cultural divisions, dramatic increase in racial resentment, and disappointment with Barack Obama’s presidency (Abramowitz and Webster 2018) may explain the higher proportion of negative reporting about highly covered Democratic men, particularly in relation to highly reported female Democrats, at the beginning of the decade.

The analysis of this study’s media sample highlights a consistent negative portrayal of both moderately and highly covered Republican men throughout the 2010s. While Republicans maintained control of the US Congress for much of the decade, research suggests that their primary focus was often on thwarting the popular Democratic president rather than addressing public grievances (Bradberry and Jacobson 2015). Their predominantly negative coverage can also be best explained by the party’s own post-2012 election review, which acknowledged public perceptions that the Republican Party seemed uncaring, “scary,” “narrow-minded,” and “out of touch,” with its members often characterized as “stuffy old men” (RNC 2013: 6). Additionally, the Republican Party’s shift to the far right, facilitated by the Tea Party and Donald Trump, has led to criticism from conservative media ⁴ and many Republican voters (Abramowitz and Webster 2018). Thus, the party’s emphasis on divisive rhetoric rather than substantive policy discussions, coupled with its proximity to Trump’s controversial persona, ⁵ as well as the debated effectiveness of the Trump administration’s response to the COVID-19 outbreak (Parker and Stern 2022), likely contributed to the increased negative coverage found in this research. This study’s results corroborate Gershon’s (2012) findings, indicating that Democrats tend to receive more favorable and less negative media coverage compared to Republicans. Understanding the impact of political parties on coverage sentiment may be linked to their strategies and public perceptions during the 2010s. To confirm this, future studies can analyze the sentiment of policy issues coverage as well as the tone of campaign communications over the past decade.

In the 2010s, significant changes occurred not only in politics but also in the media landscape. The decade witnessed the emergence of partisan outlets and the transformation of social media into a divisive and toxic space (Barberá 2020). Despite the supplementary analysis of this study indicating no significant differences between the selected outlets and their political leaning with regard to coverage sentiment, these shifts may have influenced coverage sentiment of the past decade, with media outlets adapting to the polarized climate by framing politicians based on partisanship and responses to divisive issues. Future research can investigate the differences in tone of coverage between polarized media and how coverage sentiment contributes to the deepening political divide. Partisan media have a substantial effect on election outcomes, with studies showing that Fox News can significantly increase Republican vote shares (Martin and Yurukoglu 2017). Thus, the increase in negativity observed in this study raises concerns about its potential impact on the electorate. Further research is needed to understand how nonneutral coverage shaped voter perceptions and influenced recent elections. Finally, over the past 50 years, the United States has seen a notable increase in negative partisanship, in which voters align themselves against one party rather than identifying with another (Abramowitz and Webster 2018). This phenomenon has significant implications for political representation and can reshape the legislative process (Abramowitz and Webster 2018). Future research should examine the potential contribution of negative news coverage to the rise of negative partisanship among American voters.

Limitations

This study highlights the importance of analyzing sentiment in news coverage by considering politicians’ gender, party affiliation, and media visibility. However, the grouping of politicians may oversimplify the diversity within these categories, potentially overlooking important differences within the same political groups. Due to the unavailability of consistent data across large datasets and the challenges associated with accurately and sensitively categorizing certain factors over extended periods, such variables were not included here. Future research should aim to incorporate intersectional variables to more accurately capture how different identities affect coverage sentiment.

Additionally, the study covers a decade marked by significant shifts in political, social, and cultural landscapes. Although some of the findings in this work are contextualized within historical events, the discussion of their role in the results is speculative, since the current analysis does not fully account for the evolving nature of these contexts, which affects the generalizability of the findings.

Furthermore, the present study does not entirely explore the political bias of media outlets. Despite its overall importance, however, research suggests minimal variance in sentiment between conservative and liberal media, with some studies showing sympathetic portrayals toward ideologically aligned politicians (Aday 2010; Smith and Searles 2014) and others reporting weak or no correlation (Shor 2019; Soroka 2012). More importantly, research reveals no significant effect of the political orientation of media outlets on the tone of coverage of women in the United States (Shor 2019), indicating that partisan preferences may not always result in biased reporting. Nevertheless, future studies can examine whether the political leanings of media outlets contribute to differences in coverage sentiment of politicians from opposing parties, especially from a longitudinal perspective due to the increasing polarization of American politics.

In addition, determining the sentiment of media coverage is inherently challenging, even with a rigorous manual labeling process employed in this study. Despite using advanced automated sentiment classifiers with excellent performance, there remains a possibility of misinterpretations in evaluating the sentiment of news articles. Finally, other factors may contribute to coverage sentiment, including the share of women in politics and media ownership structures.

Conclusion

This study explores the tone of political news coverage in the United States during the 2010s, analyzing nearly 900,000 stories about more than 1,000 US politicians. Employing state space models and advanced sentiment classifiers, the research challenges the expectation that female politicians consistently receive more negative coverage. Instead, coverage varied by party affiliation and media visibility, with female Democrats often facing less negativity than male Republicans but more than female Republicans and male Democrats. This underscores the significant influence of gender and party dynamics on media portrayals. Moreover, the differences in coverage sentiment between moderately and highly covered politicians emphasize the importance of media visibility for coverage sentiment, suggesting that greater media visibility may lead to increased media scrutiny.

The longitudinal analysis revealed trends in news sentiment across election and nonelection periods, showing an increase in both negative and positive coverage during elections and reflecting the decade’s growing political polarization. By providing new longitudinal insights into gendered media coverage, this study enriches the discourse on media representation and political dynamics, highlights the need for further research in diverse political systems, and emphasizes the importance of longitudinal approaches to understanding the evolving relationship between media, gender, and politics.

Supplemental Material

sj-pdf-1-hij-10.1177_19401612251318187 – Supplemental material for Beyond Neutrality: A Longitudinal Analysis of Gender, Party, and Media Visibility in the Tone of US Political Coverage

Supplemental material, sj-pdf-1-hij-10.1177_19401612251318187 for Beyond Neutrality: A Longitudinal Analysis of Gender, Party, and Media Visibility in the Tone of US Political Coverage by Aliya Andrich in The International Journal of Press/Politics