Over the last decade, gender has been particularly salient in US elections. For example, in 2016, Hillary Clinton was the first woman to win a major party’s nomination for the presidency. On the flip side, Donald Trump displayed misogyny during the election, from his treatment of female opponents on the campaign trail to the surprise October release of the Access Hollywood tape, in which he was caught on a live mic bragging about sexually harassing women. A record number of women ran for Congress in 2018Footnote 1 and for the Democratic nomination for the presidency in 2020. In the 2024 election, Donald Trump highlighted hypermasculinity on the campaign trail: think of his appeals to men enamored with mixed martial arts and of Hulk Hogan speaking at the Republican National Convention. The surprise this time around was Biden’s withdrawal from the race in July following a disastrous debate performance, which led to the nomination of Kamala Harris as the party’s nominee, the first woman of color nominated for president by a major political party. How does the public react to these distinct features of the political environment, in which gender is salient, albeit in different ways?
We define a gendered political context (or GPC) as one that relates to gender and is politicized. That is, the context should have relevance for the status of women and/or girls in society and the association to politics should be clear. Some of the gendered political contexts we highlighted above are positive, and signal hope and progress with respect to the upward mobility of women. However, other GPCs, such as Trump’s treatment of women, are negative, and threaten the status of women in US society. While these are quite different signals, the media and political elites often linked both to increased political engagement. For example, after Donald Trump called Clinton ‘such a nasty woman’Footnote 2 during the final presidential debate, Elizabeth Warren (D-MA) quipped at a campaign rally: ‘Get this Donald: Nasty women are tough. Nasty women are smart. And nasty women vote!’Footnote 3 In 2024, the following headline appeared on BBC: ‘“A lot of women are rising up”: Harris hopes hinge on female vote’.Footnote 4
Scholars have examined linkages between both types of GPCs and political participation. Abundant scholarship has considered how the presence of women running for office, especially high-profile offices, influences political engagement, particularly on the part of women (for example, Alexander and Jalalzai Reference Alexander and Jalalzai2020; Atkeson, Reference Atkeson2003; Barnes and Taylor-Robinson Reference Barnes, Taylor-Robinson, Alexander, Bolzendahl and Jalalzai2018; Campbell and Wolbrecht Reference Campbell and Wolbrecht2006, Reference Campbell and Wolbrecht2025; Fridkin and Kenney Reference Fridkin and Kenny2014; Ladam et al. Reference Ladam, Harden and Windett2018), and the findings are generally mixed. Comparatively less work has considered threats to women and their status in society and political engagement (Archer and Kam Reference Archer and Kam2020; Campbell and Wolbrecht Reference Campbell and Wolbrecht2020; Cassese et al. Reference Cassese, Ondercin and Randall2025; Clayton et al. Reference Clayton, O’Brien and Piscopo2023; Lawless and Fox Reference Lawless and Fox2018). This literature more consistently finds a mobilizing effect when women face gendered political contexts that pose a threat to them. For both types of GPCs, it is not always clear what mechanisms contribute to these relationships.
In this article, we explore a mechanism that may help us to understand why GPCs lead to increased political engagement: emotional reactions. We ask three key questions. First, how does the public react emotionally to GPCs that signal progress for women or pose a threat to women? Second, are the emotional reactions generated felt more strongly for some groups? Do all women react similarly, or do reactions vary by party? Do some groups of men also have an emotional reaction to GPCs? Finally, what are the implications of these emotional reactions? Are these effects fleeting or do they motivate individuals to become more engaged in politics?
Drawing from theories of emotion, we contend that individuals will feel distinct emotions in reaction to GPCs that signal hope or threat and these reactions will be more intense for some groups. In encountering threats to women, we theorize that the dominant emotional reaction will be feelings of anger, though individuals may also experience anxiety and dampened enthusiasm. Given threats to women over the last decade, we argue that these feelings should be felt more strongly among Democrats, women, Democratic women, and highly educated women. Conversely, when GPCs signal hope and empowerment for women, it should generate feelings of enthusiasm, and more intensely among these same groups. Finally, elevated anger and enthusiasm should generate greater political engagement.
To test these arguments, we turn to the context of the 2016 US presidential election, which was unique in that gender was relevant in ways not seen in any prior presidential election. We first fielded an online experiment in February of 2019 with Qualtrics panelists drawn to be representative of the US population. Respondents were randomly assigned to read a news vignette that discussed either Trump’s treatment of women, Clinton’s historic run for office, or a non-political article. The timing of our study makes this a hard test of our argument in that it takes place three years after the 2016 election. We find the public reacts with greater anger (and lower enthusiasm) to the Trump vignette, and this effect is particularly pronounced for Democratic women and men, but also holds for Republican women. These elevated feelings of anger increase intentions to engage in politics. Only highly educated women react with greater enthusiasm to the Clinton vignette, and these feelings increase their likelihood of political engagement. We supplement these main findings with two additional experiments, the same vignette experiment with a convenience sample of highly educated white women, and one fielded during the 2020 presidential election, focusing on Trump’s treatment of women. The findings from these two studies reinforce the findings from Study 1.
This manuscript makes several theoretical contributions of relevance to work on emotions and politics, gender and politics, and political participation. First, while a number of scholars have explored the behavioral consequences of elevated emotions in a number of domains, such as race, terrorism, infectious diseases, and the economy (for example, Albertson and Gadarian Reference Albertson and Gadarian2015; Banks Reference Banks2014; Kam and Estes Reference Kam and Estes2016; Merolla and Zechmeister Reference Merolla and Zechmeister2009; Phoenix Reference Phoenix2019), little work has explored the precursors to emotional reactions, how gendered political contexts affect emotional reactions, or for whom these reactions are more deeply felt (exceptions include Cassese and Holman Reference Cassese and Holman2019; Clayton et al. Reference Clayton, O’Brien and Piscopo2023 and Valentino et al. Reference Valentino, Wayne and Oceno2018). Our work theorizes which emotions individuals are likely to experience when they encounter GPCs that signal a threat to women or signal progress. Second, while a great deal of work has looked at the impact of high-profile women candidates on women’s political empowerment, less scholarship has looked at GPCs that signal threat, and it is not always clear what mechanisms drive either relationship. Our contribution is to highlight specific mechanisms that might help us to understand why these GPCs lead to greater political engagement. Some have theorized about emotional reactions to isolated GPCs (Bankert Reference Bankert2020; Chira Reference Chira2020; Kam and Archer Reference Kam and Archer2021). We develop a broad theory of how different GPCs might impact a range of emotional reactions and how those in turn shape engagement. Third, we develop a theory for how the intersection of identities impact the intensity of emotional reactions to GPCs. Few who study emotional reactions to the political environment have considered how the interplay of identities influences reactions (exceptions include Cassese et al. Reference Cassese, Ondercin and Randall2025; Gutierrez et al. Reference Gutierrez, Ocampo, Barreto and Segura2019; Phoenix Reference Phoenix2019). Fourth, our study contributes to the political participation literature by highlighting how GPCs can shape political engagement through emotional reactions. Canonical theories of participation often focus on more stable facets of individuals (see Willeck and Mendelberg Reference Willeck and Mendelberg2022 for a review), which are slow to change over time, while emotional reactions can help us understand short-term shifts in political engagement. Our theoretical arguments are meant to be general and thus can be applied to other GPCs in the United States, such as reactions to the #MeToo movement, the Dobbs decision, or Kamala Harris’ run for the presidency in 2020 and 2024, as well as GPCs in other country settings, such as the pro-choice protests in Poland (Drążkowski et al. Reference Drążkowski, Trepanowski and Mikołajczak2024), the presence of women political leaders in other countries (for example, Alexander and Jalalzai Reference Alexander and Jalalzai2020; Wolbrecht and Campbell Reference Wolbrecht and Campbell2007), and reactions against misogynistic leaders (Ash Reference Ash2019; Kaul Reference Kaul2021; Xavier Reference Xavier2015).
We also make several empirical contributions. First, while scholars have studied the emotions-to-engagement link, a common approach is to look at the impact of emotions separately. We evaluate what happens when a stimulus from the environment impacts more than one emotion, and how this impacts the likelihood of engagement. For example, we show that Trump’s treatment of women leads to a mobilizing effect through elevated anger, but also a demobilizing effect through diminished enthusiasm. When combined, these countervailing effects lead to greater engagement only among women and particularly Democratic women. Second, given our experimental approach, we are better able to assess whether people’s emotional reactions and increased engagement are due to Trump’s treatment of women and Clinton’s historic run for office, rather than reactions to the candidates more generally (Lawless and Fox Reference Lawless and Fox2018; for a study that uses this approach to look at political ambition, see DeMora et al. Reference DeMora, Lindke, Long, Merolla and Osorio2023). Third, given the timing of some of our studies, our results also show that reminders of past GPCs can still generate strong reactions. This has implications for political campaigns, which may have incentives to remind the public about past actions to spur engagement. Finally, our findings provide insight into why political engagement has been increasing over the last decade, particularly among Democratic women (White Reference White2018; also see Appendix Figures C1–C4).
GPCs and Civic Engagement
The first element of our theory of emotional engagement is gendered political contexts, or GPCs. The political environment is broad in scope and encompasses the many ways in which an individual observes and interacts with the political world. Individuals might be exposed to the political environment as they read or listen to news, talk to people in their community, or as they go about their daily lives. We focus on features of the political environment that are relevant to women in some way, whether it is a negative stimulus or a positive stimulus, and that are politicized. That is, in order to qualify as a GPC, a stimulus must be both relevant to women and/or girls and the information environment should link that stimulus to politics. This could occur when a woman runs for a high-profile political office, but it could also occur in a race with two men when an issue of relevance to women is highly salient or when a leader uses sexist rhetoric on the campaign trail.
What do we know about GPCs and political engagement from existing research? We first consider GPCs that signal progress for women and their status in society: women running for high-profile political office. Some studies have found associations between women’s presence on the ballot or in elected office and higher levels of political knowledge, greater interest in politics, more political efficacy, and increased participation in politics (for example, Atkeson Reference Atkeson2003; Campbell and Wolbrecht Reference Campbell and Wolbrecht2006; Fridkin and Kenney Reference Fridkin and Kenny2014; Hansen Reference Hansen1997; Hinojosa and Kittilson Reference Hinojosa and Kittilson2020; Jones Reference Jones2014; Stauffer Reference Stauffer2021; Verba et al. Reference Verba, Burns and Schlozman1997). However, other research has found largely null effects (for example, Broockman Reference Broockman2014; Dolan Reference Dolan2006; Lawless Reference Lawless2004; Wolak Reference Wolak2015, Reference Wolak2020). A growing body of work finds the effect of women’s presence in politics on political engagement is conditional. Women are mobilized by women leaders when they share the leader’s party affiliation or ideology (for example, Mariani et al. Reference Mariani, Marshall and Mathews-Schultz2015; Reingold and Harrell Reference Reingold and Harrell2010). Young women are more activated by women leaders than older women since their political attitudes are less crystallized (for example, Campbell and Wolbrecht Reference Campbell and Wolbrecht2006, Reference Campbell and Wolbrecht2025). Furthermore, women only become more engaged in politics when high-profile women leaders are running, since citizens are more likely aware of them (Alexander and Jalalzai Reference Alexander and Jalalzai2020; Barnes and Taylor-Robinson Reference Barnes, Taylor-Robinson, Alexander, Bolzendahl and Jalalzai2018; Fridkin and Kenney Reference Fridkin and Kenny2014; Ladam et al. Reference Ladam, Harden and Windett2018).
Clinton’s nomination for president in 2016 fits into the category of a high-profile example of progress for women and their status in government, as she was the first woman to receive a major party nomination for the office in US history. There is some evidence that connects Clinton’s historic run to engagement-related outcomes. Bonneau and Kanthak (Reference Bonneau and Kanthak2018) found that exposure to a political advertisement about Clinton increased political ambition among women favorably disposed towards her, and Scott and Collins (Reference Scott and Collins2020) found that Black women were more likely to express desire to run if they had favorable feelings towards Clinton. DeMora et al. (Reference DeMora, Lindke, Long, Merolla and Osorio2023) showed that reading about Clinton’s historic run increased political ambition among a sample of highly educated white women, especially Democrats. While these findings are suggestive, they all only look at a desire to run for political office, which is quite rare among the public. Furthermore, only one (DeMora et al. Reference DeMora, Lindke, Long, Merolla and Osorio2023) considers whether emotional reactions link this GPC to greater political engagement. We build on existing work by considering a mechanism that has not received much attention in this literature, emotional reactions, and by looking at how these reactions are felt more intensely depending on the intersection of gender, education, and party.
As noted earlier, there has been more limited scholarly exploration of the impact of GPCs that signal threats to the status of women in US society. Some research has demonstrated that women who experience policy threat and political exclusion react with greater political ambition (Clayton et al. Reference Clayton, O’Brien and Piscopo2023). The #MeToo movement was associated with greater political engagement in the 2018 congressional elections, particularly among Democrats (Castle et al. Reference Castle, Jenkins, Ortbals, Poloni-Staudinger and Strachan2020) low in sexism (Kam and Archer Reference Kam and Archer2021; see also Bankert Reference Bankert2020 for personal experiences with sexism), those who experienced sexual harassment (Castle et al. Reference Castle, Jenkins, Ortbals, Poloni-Staudinger and Strachan2020), and those high in linked fate with women (Jenkins et al. Reference Jenkins, Poloni-Staudinger and Strachan2023). The Dobbs decision was also correlated with greater political engagement among women (Paris and Cohn Reference Paris and Cohn2022; Sommer et al. Reference Sommer, Rappel-Kroyzer, Adamczyk, Lerner and Weiner2023), younger voters (Amos and Middlewood Reference Amos and Middlewood2024), and those with more intense preferences on abortion (Cassese et al. Reference Cassese, Ondercin and Randall2025). For the case we look at, the 2016 election, Campbell and Wolbrecht (Reference Campbell and Wolbrecht2020) found stronger reactions among Democratic girls, who, though disillusioned after the 2016 election, became more interested in protesting. In a survey of highly educated women, Lawless and Fox (Reference Lawless and Fox2018) found that Democratic women were more likely to report feeling appalled and depressed in reaction to Trump’s victory in 2016, and these negative feelings were associated with greater political engagement.
What are some mechanisms linking GPCs that signal threat to political engagement? While Clayton et al. (Reference Clayton, O’Brien and Piscopo2023) focus on feelings of efficacy and not emotional reactions, the focus group responses in their research show that the combination of threat and exclusion led to higher degrees of anger and higher mobilizing potential. Reacting to the perceived threat presented by Donald Trump specifically, a focus group participant stated, ‘It was really the aftermath of the election of Donald Trump; as it began to settle in… I felt myself saying, “I don’t have a choice, I have to run.” It’s like a fire in my belly that’s not going to go away if I don’t stand up as a woman and someone who has thought about it before. If not now, when? And if not me, who?’ (Clayton et al. Reference Clayton, O’Brien and Piscopo2023, 1468). Kam and Archer (Reference Kam and Archer2021) suggest that emotional reactions likely undergird the mobilizing effects they observed in response to the #MeToo movement, particularly anger, but they were not able to test this mechanism directly given data limitations (see also Bankert Reference Bankert2020 for emotional reactions to personal experiences with sexism and consequences for participation). Using data from the American National Election Study, Cassese et al. (Reference Cassese, Ondercin and Randall2025) show how Democrats with more intense preferences on abortion were more likely to feel angry and anxious about the Dobbs decision, while Republicans with more intense preferences were more likely to feel hope, but they do not empirically connect these emotional reactions to greater engagement. Lawless and Fox (Reference Lawless and Fox2018) look at associations between feeling appalled and depressed about Trump to increased engagement, but they cannot isolate whether Trump’s treatment of women is driving this effect, they do not look at anger or related emotions like anxiety, and they only study highly educated women.
In the next section, we develop a theory of emotional engagement that helps us understand why GPCs that signal threat or hope for women might lead to greater political engagement.
A Theory of GPCs and Emotional Engagement
Gendered political contexts (GPCs) have been salient in US politics over the last decade, with some signaling progress for women and their status in society, and others signaling threats. How does the public react emotionally to these different types of GPCs? Who reacts most strongly and with what consequences for political engagement? In what follows, we contend that GPCs that signal progress for women, like Clinton’s historic run for office, should generate enthusiasm, particularly among some groups, and this elevated enthusiasm should lead to a greater inclination to engage in politics. In contrast, GPCs that signal threats to women are likely to instead generate anger, and to a lesser extent anxiety, particularly among some groups, and this should also translate to greater political participation.
We draw from cognitive appraisal theories of emotion to develop expectations for which emotions individuals are likely to experience when they interact with GPCs. According to this theory, individuals first evaluate whether a particular stimulus from the environment is pleasant or unpleasant and whether it is blocking progress towards their goals, which can lead to positive or negative emotions. Individuals then consider secondary appraisals of the environment, which lead them to experience distinct emotions. In total, individuals appraise their environment on six different cognitive dimensions: ‘certainty, pleasantness, attentional activity, control, anticipated effort, and responsibility’ (Lerner and Tiedens Reference Lerner and Tiedens2006, 117; also see Lazarus Reference Lazarus1991). How they appraise their environment across these dimensions influences the particular emotions that they experience. We focus on emotions particularly relevant to politics: anger, enthusiasm, and fear.
Anger and fear are similar on the dimension of unpleasantness, whereby individuals feel they are being blocked from making progress towards a goal. However, they differ on the dimensions of certainty, control, and responsibility (see work by Carver Reference Carver2004; Harmon-Jones et al. Reference Harmon-Jones, Harmon-Jones, Abramson and Peterson2009; Lazarus Reference Lazarus1991; Lerner and Keltner Reference Lerner and Keltner2000; Lerner and Tiedens Reference Lerner and Tiedens2006). Individuals are more likely to experience fear when an unpleasant context is higher in uncertainty, where they feel little control, and where they are not necessarily sure where to place blame for the situation. Individuals are more likely to experience anger when they have greater certainty, can attribute blame for the unpleasant situation, and where they feel greater control (Carver Reference Carver2004; Harmon-Jones et al. Reference Harmon-Jones, Harmon-Jones, Abramson and Peterson2009; Lazarus Reference Lazarus1991; Lerner and Keltner Reference Lerner and Keltner2000, Reference Lerner and Tiedens2006). Enthusiasm is different from fear and anger in that it is more likely to result from a situation where the initial expectancy is favorable (Carver and Scheier Reference Carver and Scheier1990) and individuals are receiving positive feedback on attaining goals from the environment.
How does this help us understand the emotional reactions individuals are likely to have towards GPCs? We contend GPCs that signal threats to women should lead to greater anger or anxiety, as they block progress towards goals. Which emotion is most prevalent depends on secondary appraisals. For example, if individuals have a sense of who to blame for the negative situation, are certain of the threat, and feel like they have a sense of control, they are more likely to experience anger than anxiety. GPCs that signal hope provide positive feedback of progress towards goals and are likely to lead to greater enthusiasm.
If we turn to the context of the 2016 election, being reminded of Trump’s treatment of women and Clinton’s historic run for office should lead to differing emotional reactions. Trump’s mistreatment of women was salient during the election, not only in allegations of sexual harassment and assault, but also in his treatment of other women running for president, such as Carly Fiorina, whose appearance he criticized during a debate, and Hillary Clinton, who he called a ‘Nasty Woman’ in another debate. Much of this came to a head with the release of the Access Hollywood tape in which he was caught on a live mic bragging about sexual assault. Cognitive appraisal theory suggests that exposure to this type of stimuli should lead to higher levels of anger: it is unpleasant to read about, it is certain that Trump is the culprit, and it is clear who to blame. Thus, our first hypothesis is that exposure to Trump’s treatment of women should increase anger (H1). It is also likely that individuals will feel some anxiety given that Trump was in office at the time of our study, and individuals may feel uncertain about the status of women. They should also feel lower levels of enthusiasm since progress towards goals is blocked.
In stark contrast to Trump, reading about Clinton putting cracks in the glass ceiling should generate more pleasant appraisals, as it highlights advances made by women, who have historically been absent from the highest rungs of government. Since such content does not highlight party or issue differences, it should not signal threats to any particular group. We therefore hypothesize that individuals should feel higher levels of enthusiasm when reflecting on Clinton’s historic candidacy (H2).
But, will all individuals experience the same intensity of emotions to these GPCs? We expect reactions will be stronger among Democrats than Republicans. For one, in today’s polarized environment, individuals are likely to process information through a partisan lens (for example, Berinsky Reference Berinsky2023; Druckman et al. Reference Druckman, Peterson and Slothuus2013; Groenendyk and Krupnikov Reference Groenendyk and Krupnikov2021; Klar and McCoy Reference Klar and McCoy2021; Lavine et al. Reference Lavine, Johnston and Steenbergen2012; Nicholson Reference Nicholson2012; Zaller Reference Zaller1992). They should be more motivated by a candidate who shares their partisanship, and less critical of an in-party candidate. Second, the Democratic party has developed a stronger reputation on issues related to gender equality, and is perceived as being better able to handle sexual misconduct (Holman and Kalmoe Reference Holman and Kalmoe2024; Masuoka et al. Reference Masuoka, Grose and Junn2023; Wolbrecht Reference Wolbrecht2000), while the Republican party, in contrast, has staked out a position that upholds traditional gender norms and values. These elite positions are also reflected among the mass public: individuals who identify as Democrats are more favorable towards feminism and feminists, are starkly lower in sexism, and hold more progressive views on gender compared to Republicans (for example, Archer and Kam Reference Archer and Kam2020; Banda and Cassese Reference Banda and Cassese2022; Cassese and Holman Reference Cassese and Holman2019; Castle et al. Reference Castle, Jenkins, Ortbals, Poloni-Staudinger and Strachan2020; Elder et al. Reference Elder, Greene and Lizotte2021; McCabe Reference McCabe2005; Oceno and Morell Reference Oceno and Morell2024). Applying this logic to the cases we examine, a Democrat reading about Trump’s mistreatment of women should be inclined to believe the information and react with greater anger, while a Republican may discount the information, argue against it, and react with reduced or no emotions (H1a). Conversely, a Democrat reading about Clinton’s historic run should feel greater enthusiasm than a Republican (H2a), because of shared partisanship and greater value placed on women’s political leadership.
Second, we expect emotional reactions to GPCs will be more intensely felt among women. Research in social psychology and issue publics show that individuals are more likely to search for and process information when it is relevant to them, such as when it is related to an identity group, (for example, Iyengar et al. Reference Iyengar, Hahn, Krosnick and Walker2008; Jerit Reference Jerit2008; Johnson and Eagly Reference Johnson and Eagly1989; Krosnick Reference Krosnick1990; Petty et al. Reference Petty, Cacioppo and Goldman1981), and this has been shown to hold for women (Bolsen and Leeper Reference Bolsen and Leeper2013). Other research has found women candidates are uniquely able to prime gender identity for women (Holman et al. Reference Holman, Schneider and Pondel2015). Since GPCs are related to women specifically, we expect women will be more attuned to Trump’s negative treatment of women and Clinton’s historic run for office and, therefore, react with stronger emotional reactions (H1b and H2b, respectively).
Third, the intersection of party and gender identity should influence the intensity of emotional reactions (H1c and H2c). The public is sometimes confronted with cross-pressure due to conflicting identities (Berelson et al. Reference Berelson, Lazarsfeld and McPhee1954; DeMora et al. Reference DeMora, Merolla, Newman and Zechmeister2021, Reference DeMora, Lindke, Long, Merolla and Osorio2023; Klar Reference Klar2013). Women hold a variety of political views and some may be more salient than gender identity, and pull them in a different direction (see also Barnes and Cassese Reference Barnes and Cassese2017; Cassese and Barnes Reference Cassese and Barnes2019; Cassese and Holman Reference Cassese and Holman2019; Junn Reference Junn2017; Crowder-Meyer Reference Crowder-Meyer2022; Junn and Masuoka Reference Junn and Masuoka2020; Wolbrecht and Corder Reference Wolbrecht and Corder2020). In other cases, the intersection of identities might work in tandem and lead to even stronger reactions (Mariani et al. Reference Mariani, Marshall and Mathews-Schultz2015; Mason Reference Mason2016; Masuoka, et al Reference Masuoka, Grose and Junn2023; Merolla and Zechmeister Reference Merolla and Zechmeister2013; Reingold and Harrell Reference Reingold and Harrell2010). For the cases we explore, we expect Democratic women to experience the most intense anger in reaction to Trump’s treatment of women and the most intense enthusiasm in reaction to Clinton’s historic run, since both are salient to their partisan and gender identities. Conversely, Republican men should experience the least intense emotions given that neither GPC is particularly relevant to their gender group, they have incentives to discount critiques of their leader, and they may feel threatened by a Democratic woman leader (Campbell and Wolbrecht Reference Campbell and Wolbrecht2025; Valentino et al. Reference Valentino, Wayne and Oceno2018). Finally, Democratic men and Republican women may fall in between these two poles since they have one identity that might incline them towards greater anger in reaction to Trump or enthusiasm in reaction to Clinton, partisanship in the case of Democratic men and gender in the case of Republican women.Footnote 5
We also consider highly educated women (H1d and H2d), since this demographic in particular may be more attuned to and feel attacked by Trump’s mistreatment of womenFootnote 6 and more drawn to and inspired by Clinton, given her background, the criticisms she has faced for bucking traditional gender roles, and related shared experiences. Highly educated women tend to be more acutely aware of the extent of gender inequality in the public sphere (Kane Reference Kane1995), and have demonstrated strong negative emotional reactions to Donald Trump (Lawless and Fox Reference Lawless and Fox2018). We therefore expect to see highly educated women react to Trump’s treatment of women with greater anger and to a lesser extent anxiety (and lower enthusiasm). Highly educated women were also supportive of Clinton because she was actively attempting to heighten their sense of group consciousness (Campbell and Wolbrecht Reference Campbell and Wolbrecht2020; Cassese and Holman Reference Cassese and Holman2016). They may therefore feel greater enthusiasm in thinking about her historic run. DeMora et al. (Reference DeMora, Lindke, Long, Merolla and Osorio2023) found that being reminded of Clinton’s historic candidacy generated enthusiasm among highly educated white women.
Finally, what implications do these elevated emotions have for political participation? According to research on emotions, enthusiasm is an action-oriented emotion. As a positive emotion, enthusiasm can lead to participation in activities merely for the pleasure of the act regardless of the likelihood that participation will result in the achievement of a particular goal (Fredrickson Reference Fredrickson2001; Groenendyk and Banks Reference Groenendyk and Banks2014). Studies have documented higher political engagement among individuals experiencing positive emotions such as enthusiasm and hope about political figures and campaigns (Brader Reference Brader2006; Groenendyk and Banks Reference Groenendyk and Banks2014; Marcus et al. Reference Marcus, Neuman and MacKuen2000; Nichols and Valdéz Reference Nichols and Valdéz2020; Phoenix Reference Phoenix2019,Reference Phoenix2020; Valentino et al. Reference Valentino, Brader, Groenendyk, Gregorowicz and Hutchings2011).
The higher levels of certainty, greater sense of control, and higher tolerance of risk associated with anger also leads to higher action-tendencies (Lerner and Tiedens Reference Lerner and Tiedens2006). Anger has been linked to greater political engagement among those induced to feel angry about an election (Valentino et al. Reference Valentino, Brader, Groenendyk, Gregorowicz and Hutchings2011), and among communities targeted by voter identification laws (Valentino and Neuner Reference Valentino and Neuner2016). It was also a motivating factor for Latino voters in the 2016 presidential election, who were angered by Trump’s divisive rhetoric (Gutierrez et al. Reference Gutierrez, Ocampo, Barreto and Segura2019). Those high in sexism induced to feel anger were more likely to say they would turn out in the general election and support Trump (Valentino et al. Reference Valentino, Wayne and Oceno2018). Individuals high in sexism became more likely to participate when exposed to Trump’s attack on Clinton for playing the woman card (Cassese and Holman Reference Cassese and Holman2019).Footnote 7 Given this body of work, we hypothesize that elevated anger (H3a) and elevated enthusiasm (H3b) in reaction to GPCs should lead to increased political engagement. For the GPCs we explore, individuals who feel greater anger in reaction to Trump’s treatment of women should be more likely to participate in a range of political activities, as should individuals who experience enthusiasm about Clinton’s historic candidacy.
Experimental Design
It is not possible to completely isolate the effects of Clinton’s historic run from the effects of Trump’s mistreatment of women, since these things were occurring simultaneously and are likely linked in people’s minds. We turn to an experimental approach with our main goal being to make one of these relatively more salient to respondents (Zaller Reference Zaller1992). We recognize it may be hard to fully disentangle each GPC, as exposure to one may bring to mind the other. However, this should dampen any effect we find, if, for example, reading about Clinton’s historic run brings to mind Trump’s win and reduces the hypothesized enthusiasm associated with Clinton. We randomly assigned participants to read short vignettes for these treatments,Footnote 8 and also included a control vignette that did not contain any political content. We therefore use a between-subjects design where we compare responses between conditions.
The first study was conducted from 4 to 26 February 2019 online with Qualtrics panelists (n = 586), and was targeted to be nationally representative. The sample reflects the national population on gender, race/ethnicity, and education (compared to the 2010 US census) but is slightly more Democratic than the population (compared to Gallup data; see S.I., Table A1). Respondents were evenly balanced on these characteristics across experimental conditions (see S.I. Table A2). The second study was conducted online with NORC panelists in June–July 2019 (n = 605).Footnote 9 We were given the opportunity to place our experiment on a study being done with a targeted population, white women with a Bachelors degree or higher. This enables us to have a large sample of a population we expected to be particularly affected by the Clinton treatment, highly educated women for whom Clinton is a descriptive representative with respect to gender and race. We can also assess if the main findings replicate in a sample similar to the one used in Lawless and Fox (Reference Lawless and Fox2018), but using a different methodological approach. Study 2 is not representative of the US population since it only targeted highly educated white women (see S.I. Table A3 and Table A4 for balance checks).Footnote 10
In Study 1, respondents were asked a series of demographic and political predisposition questions (Study 2 has this information from the panel). Respondents were then randomly assigned to one of three experimental conditions. After exposure, they filled out a post-treatment survey that asked our main dependent variables on emotional reactions and intended future political engagement.Footnote 11 Those in the control condition read a neutral non-political article, which was a review of Amazon’s Echo. The treatment vignettes were drawn from several news articles, edited together by the investigators, and followed a similar format. They first introduced the main theme, highlighted quotes from the political figure, referenced reaction by citizens, and closed with a statement about the implications for gender equality (See S.I. for treatment text). Prior experimental work on emotions and politics has used a vignette approach (for example, Gadarian Reference Gadarian2010; Merolla and Zechmeister Reference Merolla and Zechmeister2009).
The Trump condition sought to highlight his mistreatment of women. It opened by noting Trump was the first presidential candidate to be accused of sexual harassment by at least ten women, included an excerpt from the Access Hollywood tape, and discussed setbacks for women’s equality. To refresh, we expect the Trump treatment to increase levels of anger (Hypothesis 1) relative to the control condition, and this should be more pronounced among Democrats (H1a), women (H1b), Democratic women (H1c), and highly educated women (H1d). We also explore whether the treatment leads to increases in anxiety and decreases in enthusiasm.
The Clinton condition highlighted her accomplishment as the first woman to have secured the presidential nomination from a major political party, making her a role model for young girls, as well as her prolific list of accomplishments as a civil servant. The treatment included an excerpt from Clinton’s announcement speech that daughters can be anything they want, even ‘President of the United States’, and closed by noting progress for women’s equality. We expect to find that the condition leads to higher levels of enthusiasm (Hypothesis 2) relative to the control condition, and this should be more pronounced among Democrats (H2a), women (H2b), Democratic women (H2c), and highly educated women (H2d).
As noted, the goal of the treatments was to bring to the top of respondents’ minds distinct features of the political environment related to gender, the appraisals of which should be more likely to generate anger in the case of Trump’s treatment of women and enthusiasm in the case of Clinton’s historic run. While this approach is commonly used, one limitation is that we cannot isolate which particular pieces of information in the treatments are driving emotional reactions. We opt for this approach for a few reasons. First, we wanted to mirror the way in which individuals experience news. Second, any one of the examples in the treatments could generate anger in the case of Trump or enthusiasm in the case of Clinton. We are less concerned with pinpointing exactly which pieces of information generate these effects, and are more concerned with how they might lead to an effect together. In this sense, our treatments are molar, or multi-part, treatments, meant to increase external validity (Shadish et al. Reference Shadish, Cook and Campbell2002), but we recognize the internal validity of the design is less crisp.
Another thing to note about our studies is the timing: since they took place in 2019, individuals were likely already familiar with the information in the vignettes. Therefore, the treatments bring historical events to the top of their head, which may serve as a more conservative test, as emotional reactions may be weaker several years removed from the election. Further, since the studies happened after the 2018 congressional elections, in which women were more successful, propensities to participate among women may have been more elevated in general, which may lead to a more conservative test. Since Trump was still in office at the time of our studies, the treatment may generate stronger effects compared to the Clinton treatment. The Clinton treatment effect might also be muted if reminding folks of the 2016 election triggers sadness as well as enthusiasm.
Our first set of dependent variables assess emotional reactions after exposure to a vignette. In Study 1, using a five-point Likert scale, respondents were asked to indicate to what extent they feel a particular emotion (fear, anxiety, worry, enthusiasm, hopefulness, pride, hate, contempt, bitterness, and resentfulness) right now, with the end points being ‘very slightly or not at all’ to ‘extremely’. These are based on the PANAS battery (Watson et al. Reference Watson, Clark and Tellegen1988), and are recommended by scholars to measure the constructs of anger, anxiety, and enthusiasm (Marcus et al. Reference Marcus, MacKuen, Wolak, Keele and Redlawsk2006). We create three additive scales for anger (α = 0.89), fear/anxiety (α = 0.87), and enthusiasm/hope (α = 0.86),Footnote 12 which were rescaled to run from 0 to 1 with higher values being greater expressions of the given emotion.
Questions about political engagement were asked towards the end of the survey and therefore at some remove from the emotion battery. Respondents were asked to indicate on a four-point scale how likely it would be that they would engage in the following activities in the next twelve months: contact a government official to express your policy views; donate to a campaign or cause; attend a meeting of a town or city government or school board; and join in a protest march, rally, or demonstration. We created an additive scale from all of the participation questions and rescaled the measure to run from 0 to 1 (Study 1 α = 0.85, Study 2 α = 0.59). One limitation is that we ask about future intended behavior rather than actual behavior, so people may be more likely to say they will engage in these activities due to social desirability bias. However, mean values on the engagement scale are not particularly high, 0.21 for Study 1 and 0.30 for Study 2. If anything, this should make finding an effect more challenging. Since we are interested in the effects of the treatments through emotional reactions on political engagement, we run mediation analyses to test whether elevated anger in reaction to the Trump treatment (H3a) and elevated enthusiasm in reaction to the Clinton treatment (H3b) increase intended political engagement.
The Effects of the Political Environment on Anger
We begin by showing mean anger by condition for Study 1 in Figure 1. As expected (H1), respondents in the Trump condition have significantly higher levels of anger (mean = 0.28) compared to respondents in the control condition (mean = 0.13, p = 0.00) and the Clinton condition (mean = 0.14, p = 0.00). Also as expected, we do not see a significant increase in anger for individuals in the Clinton treatment (p = 0.48). These effects hold controlling for variables that were unevenly distributed across conditions (see Table A5, column 2).
Figure 1. Mean level of anger by experimental condition, Study 1.
Note: average levels of anger by experimental condition. All bands represent 90 percent confidence intervals.
We next examine whether the treatment effects are moderated by party, gender, and then the intersection of the two. We dichotomize party and gender, where 1 equals Democrat and woman.Footnote 13 Each model includes dummy variables for each treatment, the moderating variable/s, and interactions between the two (see S.I. Table A5). We find a significant moderating effect for the Trump condition by party (p = 0.00). If we calculate the marginal effects, Republicans are not angered by the treatment (p = 0.24), while Democrats are about a quarter of a unit higher in anger (b = 0.26; p = 0.00) compared to their counterparts in the control group, a large effect for a scale that runs from 0 to 1, and consistent with H1a. The interaction between the Trump treatment and respondent sex is also significant (p = 0.00). As seen in the marginal effects in Figure 2, men and women in the Trump condition have significantly higher levels of anger than their counterparts in the control condition, but the difference is higher for women (H1b; b = 0.18; p = 0.00) compared to men (b = 0.12; p = 0.00).
Figure 2. Change in anger moving from the control group to the Trump condition, by gender and party, Qualtrics panel, Study 1.
Note: average marginal effects of the Trump condition compared to the control group by gender and partisanship. All bands represent 90 percent confidence intervals.
When we turn to the interaction of treatment, party, and gender, we find a significant moderating effect between party and the Trump condition among women (p = 0.06). As illustrated in Figure 2, Republican women, Democratic men, and Democratic women have significantly higher anger in the Trump treatment compared to their counterparts in the control group, while there is no impact among Republican men, as expected (H1c). The effect is more than double (b = 0.26; p = 0.00) for Democratic women compared to Republican women (b = 0.11; p = 0.06), though there is no meaningful difference between Democratic women and men.
Finally, when we look at the interaction between gender and education, we find that women, regardless of educational attainment, react with greater anger towards the Trump condition compared to the control (p < 0.01). Men also react with greater anger, although this effect is only marginal for less educated men (p = 0.050). Thus, we do not find support for H1d (see Table A5, column 7).
Next, we explore whether the findings replicate in Study 2 (the sample of highly educated women). In Figure 3, we show mean anger levels by experimental condition. We find similar effects for the Trump (mean = 0.38) and Clinton (mean = 0.02) conditions as we found in Study 1, where the former has significantly higher levels of anger compared to the control group, and the latter is not significantly different from the control group (mean = 0.02; p = 0.00 and 0.97, respectively). When we look at interactions with partisanship in Figure 4, we again observe that the effect of the Trump condition on anger is more pronounced (b = 0.52; p = 0.00) for Democratic women compared to Republican women (b = 0.10; p = 0.00); for ordinary least squares (OLS) results see S.I. Table A6. The size of the effect for Democratic women is particularly striking, as it constitutes half of the anger scale. These findings are consistent with Lawless and Fox (Reference Lawless and Fox2018), whose study also looked at highly educated women.
Figure 3. Mean level of anger by experimental condition, NORC panel, Study 2.
Note: average level of anger by experimental condition. All bands represent the 90 percent confidence interval.
Figure 4. Change in anger moving from the control group to the Trump and Clinton conditions, by party, NORC panel, Study 2.
Note: average change in anger by experimental condition and by partisanship. All bands represent 90 percent confidence intervals.
As a last step, we test if elevated feelings of anger translate into greater political engagement. To assess whether anger mediates the relationship between the Trump condition and future political engagement, we calculate the indirect effect of the treatments on engagement through anger using a structural equation model.Footnote 14 In both studies, as expected (H3a), individuals who experience higher levels of anger in reaction to the Trump treatment are significantly more likely to report that they will engage in politics in the next twelve months (see Table 1).
Table 1. Mediation analysis through anger on political engagement scale
The Effects of the Political Environment on Enthusiasm
We next plot mean enthusiasm by experimental condition in Figure 5 for both studies. In Study 1, as expected, mean enthusiasm is higher in the Clinton condition (H2; mean = 0.45; p = 0.26) compared to the control group (mean = 0.42), but the difference is not statistically significant. Meanwhile, mean enthusiasm is significantly lower in the Trump condition (mean = 0.31; p = 0.00) compared to the control group. Turning to Study 2, we find stronger effects, likely because the sample only consists of highly educated women. Individuals in the Trump treatment have significantly lower levels of enthusiasm compared to those in the control group (mean = 0.01, p = 0.00), while those in the Clinton condition (mean = 0.48; p = 0.00) have significantly higher levels of enthusiasm.Footnote 15
Figure 5. Mean enthusiasm by experimental condition.
Note: average level of enthusiasm by experimental condition. Panel A shows the Qualtrics study, and panel B shows the Fall 2019 NORC study. All bands represent the 90 percent confidence interval.
We next examine whether the treatment effects for Study 1 are moderated by party, gender, and education (see Table A7). We do not find that Democrats are more likely to experience enthusiasm from reading about Clinton’s historic run (H2a), nor are women (H2b). When looking at the impact of party and gender, we do not observe a moderating effect for the Clinton treatment (H2c). Finally, the Clinton treatment does not increase enthusiasm among less-educated women in our sample (p = 0.95), but it has a marginally significant effect on increasing enthusiasm among highly educated women (b = 0.099, p = 0.082 one-tailed). While marginal, this finding pairs well with what we observe from Study 2’s sample of highly educated women, and provides some support for H2d.
Since Study 2 comprises highly educated white women, we limit our examination of moderating effects to partisanship (H2c), and find a significant moderating effect for the Clinton treatment (b = 0.32, p = 0.00). As shown in the marginal effects in Figure 6, the treatment increases enthusiasm among Republicans (0.19, p = 0.00) and Democrats (0.51, p = 0.00), but the effects are much stronger among Democrats.Footnote 16
Figure 6. Change in enthusiasm moving from the control group to the Trump and Clinton conditions, by party, NORC panel, Study 2.
Note: average change in anger by experimental condition and by partisanship. All bands represent 90 percent confidence intervals.
To test if the treatments impact engagement through enthusiasm, we again estimate the indirect effects with a structural equation model (Table 2).Footnote 17 We find a positive and significant mediation result through enthusiasm from the Clinton condition for Study 2, lending support to H3b. While enthusiasm increases engagement, the Trump treatment decreases enthusiasm leading to an overall negative effect through this emotional pathway in both studies. In sum, both treatments impact enthusiasm, but in different directions, and with different consequences for future engagement.
Table 2. Mediation analysis through enthusiasm on political engagement scale
Up until this point, we have considered mediation paths through anger and enthusiasm separately. What happens if we consider the impact of the treatments on engagement through emotions simultaneously? This is important in the case of the Trump condition since it has countervailing effects, increasing engagement through anger but reducing engagement through enthusiasm. If we model multiple mediators, the combined indirect effect of the Trump treatment through anger, enthusiasm, and fear is not statistically significant in Study 1 (0.010, p = 0.370), but is significant in Study 2 (0.080, p = 0.000). The combined indirect effect of the Clinton treatment is significant in Study 1 (0.015, p = 0.082) and in Study 2 (0.096, p = 0.000). If we disaggregate the data, the effect is strongest among Democratic women: the Clinton and Trump treatment yield a positive combined indirect effect (0.066, p = 0.028; 0.170, p < 0.001, respectively) in Study 2.Footnote 18
Robustness Checks and Alternative Explanations
A reader might wonder if the effects are similar for fear as they are for anger. We do find elevated fear in reaction to the Trump treatment, but the effects are more muted than what we find for anger (see S.I. Figure A3; for other effects of fear see S.I. Tables A11 through A15).
To test the robustness of the main results, we first ran all of the analyses with a series of controls for age, income, and ideology, and the findings remain consistent (see S.I. Tables A16–A21). Furthermore, we investigate whether our vignettes did more than just echo existing candidate attitudes, by including pre-treatment vote choice as a covariate in Study 1 (see third column in S.I. Tables A19–A21). The effects reported in our manuscript remain robust to these pre-treatment inclusions. In fact, the effect of the Clinton condition on enthusiasm becomes statistically significant once baseline attachments are controlled for, underscoring that it is the GPCs themselves (not just pre-existing feelings) driving these responses.
A skeptical reader might wonder if it is really Trump’s treatment of women that is driving the findings. Although the Trump treatment explicitly mentions instances of Trump’s sexism, perhaps participants are thinking of his treatment of other groups, such as minorities, and are reacting to that instead. We think this is unlikely given that Democratic women in particular reacted most strongly to the treatment, and we also find effects among Republican women. One might also wonder if any negative story about Trump would generate the same effects. We are able to probe some of these concerns with a follow-up experiment we fielded on a module of the 2020 CES, in which we included an emotion induction survey experiment where respondents were randomly assigned to answer an open-ended question of what makes them angry about Trump; his treatment of women, his treatment of minorities, or a relaxation control treatment.Footnote 19 If there is something distinct about Trump’s treatment of women compared to Trump in general or his treatment of minorities, we should observe differences in engagement with the experimental stimuli between women and men co-partisans. We find some evidence supportive of this expectation. Republican women were more likely to write about their anger towards Trump’s treatment of women compared to Republican men, while we do not observe any significant differences in how much Republican women wrote compared to Republican men in the other conditions (see SI. Figures B4 and B5). While Democratic men and women have similar compliance rates in the Trump’s treatment of women condition (and both are higher than Republicans), Democratic women wrote more in response to the prompt than men in this condition but there are no meaningful differences between the two groups in the other experimental conditions. Finally, we looked at the text of the open-ended responses more carefully to ensure that respondents were reflecting more on Trump’s treatment of women. Respondents in the Anger-Trump women condition specifically mentioned the Access Hollywood tape and accusations of sexual assault made against him, his poor treatment of female opponents, and his wife, and did not mention his treatment of other groups. These responses did not appear as frequently in other conditions.
Another concern may be that our vignette studies do not directly manipulate the mediator. In this follow-up study, where we do directly manipulate anger towards Trump’s treatment of women, we find that respondents who complied with this treatment report a higher likelihood of engaging in political activities compared to the control (see Table 3).Footnote 20
Table 3. Treatment effects among compliers, CCES 2020
***p < 0.001; ** p < 0.01; * p < 0.05.
A reader might also wonder whether highly educated women are reacting with enthusiasm to Clinton because of her policy stances or candidacy in general, rather than the historic nature of her candidacy. We cannot cleanly distinguish between these possibilities with the current design; however, since the effects we observe for Study 2 are observed among not only Democratic women, but also Republican women, who are less likely to share policy stances with Clinton, or be enthusiastic about her as a candidate, we think it is more likely that they are reacting to the historic nature of her candidacy. This would also square with recent research which finds that novelty, like historic firsts, drives engagement more than the mere presence of women candidates (Campbell and Wolbrecht Reference Campbell and Wolbrecht2025). Likewise, readers might wonder if the saliency of women running for Congress in 2018 or sadness regarding Clinton’s loss affected their reaction to the treatment. We have no clear way of testing this, but by using an experimental design, we are increasing the salience of specific aspects of the environment in respondents’ minds.
It is also possible that these GPCs affect political engagement through additional pathways. Study 1 and Study 2 included a question on gender-linked fate post-treatment. When we look at the effect of the treatments on gender-linked fate among both samples, we do not find any meaningful effects.
Conclusion
Since the 2016 election, gender has remained salient in American political life. Our study takes a novel look at how gendered political contexts shape emotional reactions, and how that varies across individuals, building on scholarship on emotions and politics (for example, Albertson and Gadarian Reference Albertson and Gadarian2015; Cassese and Holman Reference Cassese and Holman2019; Groenendyk and Banks Reference Groenendyk and Banks2014; Phoenix Reference Phoenix2019). We show how Trump’s treatment of women increases anger and reduces enthusiasm, particularly among Democratic women, and these elevated feelings increase political engagement. We also show how a very different GPC, Clinton’s historic run for office, increases enthusiasm among highly educated women (especially Democrats), which in turn increases inclinations to engage in politics.
Our findings on the mediating effects of emotions on political engagement expand on existing scholarship. Like recent work, we observe strong effects for anger in increasing the likelihood of future intended political participation in activities beyond voting (Groenendyk and Banks Reference Groenendyk and Banks2014; Valentino et al. Reference Valentino, Brader, Groenendyk, Gregorowicz and Hutchings2011). We also find mediating effects of enthusiasm on intended future engagement (Groenendyk and Banks Reference Groenendyk and Banks2014; Marcus et al. Reference Marcus, Neuman and MacKuen2000; Valentino et al. Reference Valentino, Brader, Groenendyk, Gregorowicz and Hutchings2011). What is novel is that we also consider the combined effects of emotional reactions to GPCs on engagement.
In terms of substantive effects, the GPCs we tested only had small effects on engagement directly, but they do move the needle in meaningful ways when it comes to emotional reactions. The Trump vignette had a medium-sized effect on anger (Cohen’s d = 0.58) in Study 1, and a large effect size in Study 2 (d = 1.37). These effects were even larger among Democratic women (Study 1 d = 0.94; Study 2 d = 2.21). The Clinton vignette had a large effect on enthusiasm in Study 2 (d = 1.4) and an even larger effect among Democratic women in the sample (d = 1.91). These stark changes in emotions contribute to increased political engagement.
Our work also builds on scholarship on gender and engagement in the aftermath of the 2016 election (Cassese and Holman Reference Cassese and Holman2019; Campbell and Wolbrecht Reference Campbell and Wolbrecht2020; Lawless and Fox Reference Lawless and Fox2018; Valentino et al. Reference Valentino, Wayne and Oceno2018),Footnote 21 by exploring different GPCs, exploring a range of emotional reactions as potential mechanisms to increased engagement, examining both a general population sample and a sample of highly educated women, and using experimental designs. Our findings suggest that the increased political engagement that we have observed since 2016, particularly among Democratic women, is at least somewhat related to the more intense emotional reactions to the political environment around issues of gender, including several years removed from 2016. Additionally, we provide evidence that the political environment may affect Republican women in similar ways, but at a smaller magnitude. This is particularly important in today’s polarized partisan environment because it shows that party does not lead to the wholesale rejection of information.
An important question for future scholarship is how reactions vary by gender, party, and race. For example, women of color may have reacted even more strongly to Trump’s treatment of women, given their experience with multiple types of marginalization (Brown and Gershon Reference Brown and Gershon2016), greater sense of gendered consciousness (Harnois Reference Harnois2015), and reactions to his attacks on other marginalized groups.
While our main goal was to tease apart Trump’s treatment of women and Clinton’s historic run, since we expected them to affect different emotions, another follow-up question is how these worked in tandem to affect emotional reactions and political engagement. To maximize experimental power, we looked at each separately, but future research could consider the combined effects of the political environment. For example, might the combination of elevated anger from Trump’s treatment of women and greater enthusiasm from Clinton’s historic run have led to an even greater likelihood of political engagement?
While our empirics focused on the context of the 2016 election, the theory is meant to be broad and can be applied to other GPCs, such as the Women’s March, the #MeToo movement, the Dobbs v. Jackson Women’s Health Organization decision, and the record number of women who ran for the Democratic nomination in 2020. Some research on the impact of the Dobbs decision is consistent with our framework. For example, using data from the American National Election Study, Cassese et al. (Reference Cassese, Ondercin and Randall2025) show how Democrats with more intense preferences on abortion were more likely to feel angry and anxious about the Dobbs decision, while Republicans with more intense preferences were more likely to feel hope. Other work has shown that registration rates by women increased after the decision (Paris and Cohn Reference Paris and Cohn2022; Sommer et al. Reference Sommer, Rappel-Kroyzer, Adamczyk, Lerner and Weiner2023).
Another extension is to consider the example we raised in the introduction, the 2024 election. Trump was running for office again, and during the election was involved in several legal battles related to women, including the Stormy Daniels hush money trial, where he was convicted of a felony. The salience of his treatment of women was more subdued than in 2016, especially given concerns about inflation and immigration, but we would still anticipate that raising the salience of Trump’s treatment of women should generate anger and in turn increase engagement, particularly among Democratic women. Furthermore, the patterns we observe for Trump could hold in other cases that are highly salient and reflect explicit examples of harassment towards women. One thing to tease out more in future work is the impact of explicitly sexist rhetoric towards female rivals and sexual harassment of women (both of which are combined in our treatment). On the flip side, we had another example of a trail-blazing nominee in Kamala Harris. We expect that Harris’ nomination generated enthusiasm, particularly among Democratic women, and in turn should have increased political engagement. The one addition we would make to these expectations is that we also expect race to be particularly relevant, where Democratic women of color, particularly Black women, should have experienced enthusiasm. Emerging scholarship provides preliminary support for these expectations.Footnote 22
This theory can also be applied to contexts outside of the United States. For example, with respect to GPCs that signal progress for the status of women, some scholars have found that the presence of high-profile women leaders (Jalalzai Reference Jalalzai2016) and collective representation of women in legislative bodies leads to greater political engagement among women in contexts outside of the United States (Barnes and Burchard Reference Barnes and Burchard2013; Wolbrecht and Campbell Reference Wolbrecht and Campbell2007; but see Schwindt-Bayer and Reyes-Housholder Reference Schwindt-Bayer and Reyes-Housholder2017). It could be that enthusiasm towards these leaders, particularly among highly educated women, helps to explain the association between women’s representation and political engagement, as we find in the case of Clinton. For GPCs that signal threats to the status of women, there are examples of leaders in other countries employing sexist and misogynistic rhetoric, most notably Jair Bolsonaro in Brazil (Setzler 2021) and Rodrigo Duterte in the Philippines (Banquiles Reference Banquiles2025; Presto and Curato Reference Presto, Curato, Facal, Micheaux and Noren-Nilsson2024). Emotional reactions to this rhetoric may help to explain mobilization by women against these leaders, with for example the #elenão (#nothim) movement in Brazil (O’Doherty Reference O’Doherty2018). As another example, reactions against abortion bans predicted greater participation in pro-choice protests in Poland among both men and women (Drążkowski et al. Reference Drążkowski, Trepanowski and Mikołajczak2024). Anxiety or anger at the bans may undergird this increased political engagement, as scholars suggest occurred in reaction to the Dobbs decision in the United States (Cassese et al. Reference Cassese, Ondercin and Randall2025).
Supplementary material
Supplementary material for this article can be found at https://doi.org/10.1017/S0007123425101166.
Data availability statement
Replication data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/DVN/LEFTNE
Competing interests
The authors declare none.
Open access
