When Identities Go Viral: Themes and Aversiveness in TikTok Discourse on Kamala Harris’s Intersectional Identities During the 2024 U.S. Election

 

Gyo Hyun Koo[1]

Dasia Clemente

Howard University, USA

 

 

This study examines TikTok posts from the 2024 U.S. election to explore how users discussed Kamala Harris, with particular attention to references to her racial and gender identities. Using social representation theory and the polarized intersectionality framework, this study analyzed 28,435 TikTok posts to identify patterns in how Harris was portrayed as a woman of color in politics and to assess the quality of the discourse using aversiveness measures. A qualitative thematic analysis reveals the following themes: (1) identity-based attacks questioning her legitimacy, (2) counterarguments defending her, (3) voter enthusiasm, (4) denial of identity salience, and (5) post-election collective fate perceptions. A computational analysis of aversiveness shows that references to intersecting identities heightened overall aversiveness. Among these posts, those containing more Trump-related hashtags show higher aversiveness scores than those containing more Harris-related hashtags. These findings reflect ongoing tensions in the representation of women of color in politics and point to polarized intersectionality among TikTok users.

 

Keywords: social representation, intersectionality, polarized intersectionality, aversive online content, TikTok, Kamala Harris, deliberative democracy

 

Gyo Hyun Koo: [email protected]

Dasia Clemente: [email protected]

Date submitted: 2025-09-09

 

 

TikTok has become a major hub for political conversations (Amado, 2024; Zeng & Abidin, 2021). Its short-form videos, interactive features, and active user base (Widjaya et al., 2024) position the platform as a powerful tool for political visibility and voter mobilization during elections (Gorman et al., 2024; Gorman & Goldenberg, 2024). These videos are distributed through an algorithmic system that allows users, such as political figures, to amplify their reach beyond their initial following (Literat & Kligler-Vilenchik, 2023; Zeng et al., 2021). In the lead-up to the 2024 U.S. presidential election, both Republican and Democratic candidates launched TikTok accounts, underscoring the platform’s growing influence on political discourse. TikTok is a conducive environment for political meaning making, in part because its user- and influencer-driven messaging is not controlled by elites (Lee & Abidin, 2023), whereas discourse on platforms such as X or Facebook is more often shaped by network structures and elite control (Bossetta, 2018).

 

This study investigates how TikTok users discussed Kamala Harris during the 2024 election campaign, the first Black and South Asian woman to run for the U.S. presidency on a major party ticket. Taking intersectionality into account (Crenshaw, 1989, 1991), this study explores how race and gender together shape representations of women of color in politics. It has two main objectives: (a) analyzing representations of Harris as a woman of color candidate and (b) assessing the quality of discourse. Drawing on Moscovici’s (1988) theory of social representations, which explains how individuals construct shared knowledge to navigate the material and social world, this study uses qualitative thematic analysis to examine how Harris’s intersecting identities were represented. We then computationally examine the aversiveness of these posts, defined as expressions of dislike such as insults, profanity, or toxicity, to assess discourse quality. Finally, we apply the concept of polarized intersectionality (Pavan & Martella, 2021), which highlights how the politicization of marginalized identities intensifies polarization, and test whether posts with more Trump-related hashtags versus Harris-related hashtags show greater aversiveness, as well as whether references to her intersecting identities are associated with higher aversiveness.

 

The data set consists of 28,435 TikTok posts published between June 20 and November 8, 2024, a period spanning from one month before President Biden announced Harris’s nomination through the days immediately following the election outcome. The findings reveal five central themes in the discourse: (1) identity-based attacks questioning Harris’s legitimacy, (2) counterarguments defending her, (3) expressions of voter enthusiasm, (4) denial that her identities influenced the election outcome, and (5) post-election perceptions of collective fate. Aversiveness varied across posts: Posts with a higher prevalence of Trump-related hashtags exhibited greater aversiveness than those with a higher prevalence of Harris-related hashtags, and posts referencing intersectionality-related terms showed higher aversiveness than those without such terms. Taken together, these results reflect the tension between longstanding ways of marginalizing women of color in political life and emerging counternarratives that challenge such exclusion.

 

This study makes three key contributions. First, it advances scholarship on identity and representation by illustrating how social identities function simultaneously as resources for political mobilization and as mechanisms of structural exclusion (Mollenkopf, 2013). Second, extending the literature on the roles of media and institutions in identity construction (Kreiss et al., 2020), this study positions public discourse as a dynamic arena in which political identities are challenged and reshaped. This focus is especially important given TikTok’s growing role as a political arena for younger generations (McClain, 2024), highlighting how users on the platform shaped discourse around a woman of color candidate and her intersecting identities during the 2024 U.S. election. Third, this study offers practical contributions by showing how discourse quality, proxied by aversiveness, varies across specific keywords and hashtags. Higher levels of aversiveness may limit participation and productive exchange in online deliberation, particularly in discussions involving marginalized identities. Because posts referencing Harris’s identities were more aversive than those that did not, these findings raise questions about the safety of online deliberative environments and suggest the need for more adaptive platform policies to address identity-based attacks and support constructive online conversations.

Literature Review

 

This literature review begins by situating identity at the center of American politics, highlighting how politicians from minoritized groups, particularly women of color, have been marginalized in political discourse. It then draws on theories of social representation and discourse quality, along with deliberative democratic frameworks, to examine how these dynamics extend into digital environments. In doing so, it identifies a key gap in understanding how intersectional identities are represented and contested in online political discourse, informing this study’s focus on discussions of Kamala Harris during the 2024 U.S. election.

 

Identity and Representation

 

Identities shape how people perceive themselves, relate to others, and communicate (Howarth, 2011). In politics, identity plays a central role in shaping people’s behavior and collective actions (Bernstein, 2005). A “strong, internalized subjective identity” can enhance ingroup cohesion while simultaneously fostering antagonism toward outgroups (Huddy, 2001, p. 130). This dual dynamic reinforces unity among those with shared identities and drives political mobilization (Egan, 2012). Consequently, identities function both as sources of political empowerment and as mechanisms of social and political exclusion (Mollenkopf, 2013).

 

Representation, understood as the process of creating and conveying meaning within a culture through language, signs, and images, connects mental concepts to both real and imagined worlds and enables people to interpret individuals, objects, events, and ideas (Hall, 1997). It shapes perceptions of what is considered normal and who is regarded as belonging, thereby reinforcing systems of inclusion and exclusion (Hall, 1997). In doing so, representation actively contributes to the construction of social norms, collective values, and dominant discourses (Howarth, 2011). Hall (1997) emphasized its inherently political nature, particularly underscoring the media’s ideological power and the role of public institutions in shaping and circulating dominant meanings. From this perspective, the representation of identities can function as a strategic mechanism for either reinforcing or contesting dominant ideologies. For instance, political actors draw on a candidate’s identities, such as race, ethnicity, gender, or class, to foster connection and mobilize constituencies (Sides et al., 2018). Identities are thus dynamic and deliberately constructed through political communication, with media acting as a central platform for their articulation and expression (Kreiss et al., 2020), yet meaning making is not limited to institutions or media; we argue that the public also plays an active role in constructing and contesting identity representations. Accordingly, this study conceptualizes public discourse as a site where the meanings of social identities are continuously negotiated and redefined.

 

Social Representation Theory

 

Moscovici’s (1988) theory of social representations provides a useful analytic framework for examining how Harris’s intersecting identities are communicated and interpreted. Moscovici (1988) underscored the central role of representations in communication, particularly in shaping how knowledge is shared and identities are constructed. He distinguished among three types of representations: hegemonic, emancipatory, and polemic. Hegemonic representations reflect dominant narratives that often, though not always, reinforce existing power structures. Emancipatory representations emerge when individuals, particularly historically marginalized groups, challenge societal norms by creating alternative narratives that challenge and subvert existing power structures. Polemic representations arise from social conflicts and debates, revealing competing narratives and ideological tensions (Höijer, 2011; Moscovici, 1988). Taken together, social representations illuminate how societies collectively construct and interpret individuals’ identities, roles, and legitimacy within the political sphere. These representations are shaped by intersecting narratives of race, gender, and power, which frequently shift between visibility and marginalization (Runderkamp et al., 2022). This study adopts these three forms of representation as its analytical framework.

 

Representation and Polarized Intersectionality

 

The literature on the representation of women of color in politics highlights persistent issues of misrepresentation and divisiveness (e.g., Pavan & Martella, 2021; Trimble et al., 2015). Research shows that news media tend to place greater emphasis on the identities of politicians who are not white men, which can reinforce perceptions of their lack of political belonging and further amplify gendered and racialized portrayals (Burge et al., 2019; Runderkamp et al., 2022; Trimble et al., 2015). This occurs despite evidence that congresswomen frequently outperform their male counterparts in areas such as securing funding and passing legislation (Anzia & Berry, 2011). However, media narratives, public perceptions, and the broader political climate often remain unfavorable, as female politicians are penalized for displaying ambition or seeking power, traits that are perceived as conflicting with gendered expectations of selflessness and principle (Okimoto & Brescoll, 2010; Runderkamp et al., 2022).

 

Social representations of women of color in politics have significant consequences. While some scholars note that women from minority ethnic backgrounds can at times draw political strength from their intersecting identities, serving as bridges across diverse groups rather than facing only compounded barriers (Celis & Erzeel, 2015), this potential is offset by persistent challenges. Intersecting forms of discrimination based on race, gender, class, age, or religion often contribute to their misrepresentation, which is frequently weaponized to undermine both their identities and political agendas (Martella & Pavan, 2024; Pavan & Martella, 2021). In polarized contexts, partisan narratives exploit these identities to construct forms of political otherness, thereby producing what Pavan and Martella (2021) describe as polarized intersectionality and amplifying social and cultural disparities.

 

The 2024 U.S. presidential election provides a critical case. Kamala Harris became the first woman of color to be nominated as a major-party presidential candidate (NPR Washington Desk, 2024), making her campaign a key site for examining how candidates from historically underrepresented racial, ethnic, and gender groups are discussed. This study examines how Kamala Harris’s intersecting racial, ethnic, and gender identities are constructed in online political conversations, focusing on thematic patterns in TikTok content to assess how hegemonic, emancipatory, and polemic social representations (Moscovici, 1988) emerge within the digital public sphere:

 

RQ1: Which themes emerge in online discourse regarding Kamala Harris’s intersecting identities?

 

Discourse Quality

 

Building on the flexibility of thematic analysis in identifying patterns and meanings (Clarke & Braun, 2016), this study further examines the quality of online conversations through levels of aversiveness. Online spaces create new opportunities for political expression and exposure to diverse viewpoints. At the same time, concerns persist about the quality of discourse (Mutz, 2006; Papacharissi, 2004), including the rise of aversive content targeting minoritized groups (Corbu et al., 2024). The anonymity and lack of face-to-face interaction on these platforms create conditions under which communication may depart from social norms, fostering uncivil and aversive political discourse despite expectations that informational heterogeneity would produce more constructive engagement (Hmielowski et al., 2014; Ophir et al., 2023; Rossini, 2020; Wells et al., 2017).

 

Studying discourse quality is important because deliberative democracy depends on both the quality and quantity of social interaction (Mutz, 2006), requiring individuals to approach political decision making with open-mindedness, interest in public affairs, and a commitment to respect and tolerance (Katz, 2014). Everyday political talk is central to this process, as it facilitates the exchange of diverse perspectives, fosters mutual understanding, and encourages individuals to shift from personal concerns to collective issues (Jacobs et al., 2009). Scholars have conceptualized and measured discourse quality in various ways, including the use of connective language (Lukito et al., 2024), levels of intolerance and incivility (Rossini, 2020), and politeness (Yeomans et al., 2018). This study examines aversiveness as a measure of deliberative quality in online representations of intersectional identities. While no universally agreed-on definition exists, this study conceptualizes aversiveness as a form of incivility that violates norms of respectful communication.

 

Our justification for using aversiveness rests on a conceptualization of discourse quality that holds that participants in a dialogue ought to treat one another with respect (Steenbergen et al., 2003). Respect entails recognizing the rights of social groups, considering opposing demands as potentially legitimate, and thoughtfully engaging with counterarguments, practices that foster attentive listening and the fair evaluation of diverse viewpoints. (Steenbergen et al., 2003). Aversiveness in online conversations can affect user engagement (Koo et al., 2025), underscoring its importance in analyzing social media discussions about candidates whose identities have been historically marginalized. To evaluate aversiveness, we employ Google’s Perspective API, developed by Google Jigsaw, which assesses text across attributes such as toxicity, severe toxicity, insult, profanity, threat, and identity attack (Perspective API, n.d.). This computational approach provides consistent, scalable scores, enabling systematic assessment across the large data set.

 

This study examines whether aversiveness varies across posts that reference Kamala Harris’s intersecting identities as a Black and Asian woman and those that do not, as well as between posts with greater Trump-related hashtag use and those with greater Harris-related hashtag use. If posts referencing her identities exhibit higher aversiveness than those that do not, this would suggest identity-focused content corresponds to shifts in discourse quality, as reflected in aversive expressions. Similarly, differences in aversiveness between posts characterized by greater Harris- or Trump-related hashtag use would indicate broader disparities in the tone of candidate-centered political discourse. Accordingly, we ask:

 

RQ2: How does the level of aversiveness in online discourse about Kamala Harris differ by (a) the use of Trump-related versus Harris-related hashtags and (b) the presence of references to her intersecting identities?

 

Methods

 

Data Collection

 

This study used Junkipedia, a tool developed by the Algorithmic Transparency Institute (National Conference on Citizenship, n.d.), to retrieve TikTok posts. The data were collected through Junkipedia’s monitoring feature, which automatically scrapes TikTok for videos from selected channels, hashtags, or search terms. Using “election” as the keyword, we collected posts created over a five-month period (June 20–November 8, 2024). This timeframe spans from one month before Biden’s announcement of Harris as the Democratic presidential candidate (July 21, 2024) to the period when election results became clear following the November 6 presidential election (N = 52,261). For analysis, we created a subset by filtering posts that explicitly mentioned “Harris,” resulting in a final data set of 28,435 posts. This data set comprises both the audio content (converted to text using Google transcription) and the text embedded within the posts. As part of text preprocessing, all text was converted to lowercase.

 

Data Analysis

 

To address RQ1, which explores emerging themes surrounding Harris’s intersecting identities, we employed a grounded theory approach to conduct a qualitative thematic analysis. Two researchers independently familiarized themselves with the data and coded each post to identify emerging themes, which they then compared and discussed. Given the large sample size, the coding was divided temporally, with one researcher starting from the earliest date and the other from the latest date. This iterative process continued until theoretical saturation was reached (Guest et al., 2014). The resulting themes were subsequently reviewed, analyzed, and integrated into the theoretical framework of Moscovici’s three types of representation. Following an initial reading of the data, no substantial differences were found between thematic patterns in the text and audio content; therefore, the results are presented without distinction between the two modalities.

 

RQ2 asks how the level of aversiveness in online discourse about Kamala Harris varies based on (a) the use of Trump-related versus Harris-related hashtags and (b) the presence of references to her intersecting identities (i.e., race and gender). To examine this question, we conducted a log-transformed linear regression with aversiveness as the dependent variable. The aversiveness scores were highly right-skewed, with most values concentrated near zero. To address this and better meet the assumptions of linear regression, we applied a log transformation (West, 2022). Predictors included the use of Trump-related versus Harris-related hashtags, the presence of intersectionality-related terms, and post length (detailed in the following section).

 

Measures

 

Predictors

 

The Use of Trump-Related Versus Harris-Related Hashtags

 

We used a dictionary-based approach and identified keywords commonly associated with each candidate. First, we extracted the top 100 hashtags from the data set (see Online Supplemental Appendix A^[2]) and identified two hashtag groups: Trump-related hashtags (#maga, #trump2024, #makeamericagreatagain) and Harris-related hashtags (#voteblue, #harris2024, #voteblue2024). To improve labeling accuracy, we adopted a conservative approach to minimize noise and enhance the reliability of group distinctions. Posts containing both Trump-related and Harris-related hashtags or neither were excluded to avoid misclassifying mixed or ambiguous cues. We excluded general party and candidate names (e.g., #harris, #trump, #republican, #democrat) from the hashtag lists, as such terms are frequently used in both supportive and critical contexts.

 

Based on this classification, we created a new column in the data frame with two factor levels: posts with more Harris-related hashtags (n = 4,874) and posts with more Trump-related hashtags (n = 3,959). To avoid misclassification or overinterpretation, we do not classify users or posts as Harris or Trump supporters.

 

Referencing Intersecting Identities

 

We identified posts referencing Harris’s intersecting identities by creating an “identity” variable using a keyword-based approach. A value of 1 was assigned to posts containing at least one racial identity keyword (e.g., “Black,” “Asian,” “African”) and one gender identity keyword (e.g., “woman,” “women,” “female”) in either the text or transcribed audio. For example, this includes phrases such as “Black women,” “Black woman,” “Black female,” “Asian women,” “Asian woman,” and “Asian female,” among others.^[3] Posts that did not meet the criteria were coded as 0. The total number of unique posts referencing identity terms in either text or audio format was 1,301.

 

Outcome Variable

 

Aversiveness

 

We used the Perspective API to detect six aversive traits: toxicity, severe toxicity, identity attack, insult, profanity, and threat. The API was applied separately to audio and text content, generating six aversiveness scores for each modality. Each trait is scored on a scale from 0 to 1 and summed, yielding a total score of 0–6 per modality. The text and audio aversiveness scores were then summed to generate a single overall aversiveness variable (M = .80, SD = .81, range = .06–7.09).

 

Covariate

 

Post Length

 

We controlled for the length of each post. Text and audio transcripts were combined to represent the full content of each post, and the total word count was used (M = 226.39, SD = 296.73).

 

Results

 

Qualitative Thematic Analysis

 

RQ1 asks about themes around Harris’s intersecting identities, and the thematic analysis results in identifying five themes: (1) questioning legitimacy through identity-based and gendered attacks criticizing Harris’s leadership qualifications as a “DEI hire,” (2) counterarguments against attacks on Harris, (3) voter enthusiasm, (4) denial of the influence her identities might have had on the election outcome, and (5) post-election collective fate perception.

 

The first theme captures how users challenged Kamala Harris’s suitability for the presidency by framing her success as unearned or strategically manufactured. Some attributed her rise to identity politics, labeling her a diversity, equity, and inclusion (DEI) hire, which frames her advancement as being attributed to her identities rather than to her qualifications or suggesting her candidacy was merely a strategy to access Biden’s campaign funds. These narratives diminish her professional experience and leadership abilities and resonate with broader patterns of gendered disinformation reflecting resentment toward women in power (Stabile et al., 2019). Particularly, disinformation, such as an alleged affair with former San Francisco mayor Willie Brown, served to delegitimize Harris’s qualifications while reinforcing misogynistic tropes aimed at undermining her credibility and place in political leadership.

 

The second theme centers on users’ counterarguments to attacks on Harris. For instance, people emphasized her extensive experience as a prosecutor, attorney general, and U.S. senator, framing her as an ideal Democratic candidate because of her accomplishments and progressive stance, which contrasts with President Biden’s. This is accompanied by people pointing to systemic barriers that limit Black representation in leadership, arguing that the lack of Black leaders reflects structural inequalities rather than a lack of ability.

 

The third theme centers on voter enthusiasm following Biden’s announcement to withdraw from the race and endorse Harris as the nominee. Posts reflecting this theme expressed excitement about her candidacy and emphasized the generational change she represents. This enthusiasm, especially among young voters, highlighted her intersectionality as a strength and underscored the historical significance of her position as the first woman of color to run for president.

 

The fourth theme shows people denying the impact of Harris’s identities on the election results. Instead, they attributed the outcome to Biden or the Democratic Party’s failures in areas like economic stability and foreign policy, including the war in Gaza, rather than Harris’s identity as a Black and Asian woman.

 

The final theme highlights disappointment over the election results, emphasizing the collective fate of racial, gender, and socioeconomic groups. Harris’s identities were framed as symbolic of broader issues, such as women’s rights or liberal values being overshadowed by money, with her defeat seen as a broader setback for democracy. We present exemplary posts for each theme in Online Supplemental Appendix B.

 

We interpret these five themes through Moscovici’s (1988) three types of social representations. First, hegemonic representation, indicative of dominant narratives (Moscovici, 1988), was shown in user discourse that either engaged in gendered and racialized critiques of Harris (theme 1) or downplayed the relevance of social identities in American politics (theme 4). For instance, some themes emphasized Harris’s identity by framing her candidacy as a product of identity politics, including characterizations of her as a DEI hire, thereby reducing her qualifications to race and gender rather than political experience. These narratives undermine her achievements and sustain the distinct challenges faced by women of color in political leadership. At the same time, some users rejected the idea that Harris’s identities influenced the election outcome, despite existing evidence that social identities are deeply intertwined with electoral behavior and experiences (e.g., Carey & Lizotte, 2019; Cassese & Barnes, 2019). Taken together, these themes suggest that increasing female representation in traditionally masculine roles does not automatically improve voter perceptions (Okimoto & Brescoll, 2010), but instead reflects the persistence of hegemonic representations that continue to shape how candidates are framed and assessed in U.S. elections.

 

Second, emancipatory representation, with subgroups constructing alternative representations and providing different versions of the same phenomenon (Höijer, 2011; Moscovici, 1988), highlighted expressions of hope and inclusivity in American politics, grounded in Harris’s intersecting identities. In this category, users celebrated her identity as a historic milestone (theme 3), portraying her as a symbol of progress and representation for underserved communities. As the first Black and South Asian woman elected to the U.S. vice presidency and a presidential candidate from a major party, her rise was viewed by some as a significant shift in American history. However, following the election, users also expressed disappointment and frustration (theme 5), viewing it as a discouraging moment for those hoping for collective change. This reaction reflects emancipatory representations that articulate unmet aspirations for greater inclusion in political leadership.

 

Third, polemic representation, which arises from ideological tensions and public debate (Moscovici, 1988), captures the conflicts among competing political perspectives. For example, counternarratives (theme 2) directly addressed and challenged false allegations and gendered disinformation presented in theme 1, including claims about Harris’s personal life and professional qualifications. Polemic tensions were also evident between theme 4, in which users denied that Harris’s identities influenced the election outcome and attributed the results to other factors, and theme 5, in which users centered on her intersecting identities and interpreted the outcome as a broader setback for collective progress. These themes captured contradictory viewpoints, showing different interpretations of Harris’s candidacy and its implications.

 

Computational Analysis

 

RQ2 asks how the level of aversiveness in online discourse about Kamala Harris varies by (a) the use of Trump-related and Harris-related hashtags and (b) by whether her intersecting identities are referenced. A linear regression was conducted on log-transformed aversiveness scores to examine the effects of hashtag use, intersectionality mention, and post length. The overall model was statistically significant [F(3, 7196) = 50.74, p < .001].^[4] Posts with more Trump-related hashtags (M = .95, SD = .89, n = 3,206) were significantly more aversive than those with more Harris-related hashtags (M = .90, SD = .94, n = 3,994, β = .12, SE = .02, p < .001). Posts that referenced an intersecting identity (M = 1.32, SD = 1.00, n = 343) were also significantly more aversive than those that did not (M = .09, SD = .91, n = 6,857, β = .52, SE = .05, p < .001). Additionally, post length was a positive predictor of aversiveness (β = .0001, SE = .00005, p = .02). Table 1 shows mean aversiveness scores for Trump-related and Harris-related hashtag groups by the presence of intersectionality references.

 

Table 1. Aversiveness Scores by Trump- and Harris-Related Hashtag Use and Intersectionality Reference.

Intersectionality Reference	Trump-Related Hashtags	Harris-Related Hashtags	Trump-Harris Mean Difference
Present	M = 1.33, SD = 1.02 (n = 140)	M = 1.31, SD = .98 (n = 203)	+.02
Absent	M = .93, SD = .88 (n = 3,066)	M = .88, SD = .93 (n = 3,791)	+.05

Note. n = 7,200. Positive values in the Trump-Harris mean difference column indicate greater aversiveness in posts with more Trump-related hashtags than in posts with more Harris-related hashtags.

 

Discussion

 

This study examined TikTok discourse surrounding the intersecting identities of Kamala Harris, the first woman of color to be nominated as a major-party presidential candidate in the 2024 U.S. election. Using a mixed-methods approach, this study first identified five themes that illustrate how different forms of discourse emerge in the meaning-making process surrounding intersecting identities, either by reinforcing dominant narratives that uphold existing power structures or by challenging prevailing social norms (Höijer, 2011). We also examined these themes within Moscovici’s (1988) three social representation categories. Importantly, themes do not map neatly onto a single representational category, but may reflect multiple forms simultaneously (as illustrated by theme 5). In addition, we assessed discourse quality in terms of aversiveness. We first discuss the findings from the aversiveness analysis and then contextualize them within the thematic analysis.

 

In analyzing discourse quality using measures of aversiveness, we found that posts with greater Trump-related hashtag use exhibited higher aversiveness than those with greater Harris-related hashtag use. This pattern suggests that posts with Trump-related hashtags are associated with a more confrontational or negative tone in the surrounding discourse. Because identifying the specific targets of aversiveness falls outside the scope of this study, we cannot determine whether these posts reflect ingroup affirmation or outgroup derogation. Still, differences in aversiveness observed across the two groups suggest that, because hashtags often signal support, disagreement, liking, or disapproval (Johnson et al., 2019), variation in aversiveness by hashtag usage may reflect broader patterns of affective polarization in the associated discourse (Wojcieszak et al., 2021).

 

Posts referencing intersectionality-related keywords were generally more aversive than those that did not. This pattern suggests that references to intersecting identities may activate entrenched, divided narratives and shift political debate into identity-based conflict, providing support for the concept of polarized intersectionality. Interestingly, although both identity references and the use of Trump- or Harris-related hashtags are associated with aversiveness, references to intersecting identities show a stronger relationship with aversiveness. Across both Trump- and Harris-related hashtag groups, posts referencing intersectionality also exhibited higher raw aversiveness scores. Although we remain cautious about using hashtags as proxies for candidate support, these findings suggest that the politics of identity, when refracted through polarized social media environments, can contribute to more aversive online discourse.

 

These findings of hashtag-group differences in aversiveness can be linked to the thematic analysis of polemic representations, as they reflect intergroup conflict within the discourse. In the thematic analysis, we observed polemic representations characterized by conflicts between users, with some promoting false allegations and gendered disinformation and others responding with counternarratives that contested these portrayals.

 

Moreover, we identified themes that represent Harris’s candidacy in ways that minimized her racial and gender identities, treated her qualifications as unrelated to identity, or rejected the relevance of identity to electoral outcomes. Although some users introduced counternarratives and forms of emancipatory representation, our computational analysis showed that discourse on intersectionality was consistently associated with greater aversiveness. This highlights how intersecting identities are more frequently invoked in negative reactions, including critiques and narratives of marginalization, reinforcing their role as contested sites of polarization.

 

Finally, longer posts were associated with higher aversiveness. It aligns with previous research indicating that longer conversations tend to be more toxic, both across multiple platforms (Avalle et al., 2024) and within single platforms such as Twitter (Salehabadi et al., 2022). Although further research is needed to examine TikTok’s various affordances, particularly the combined use of audio and text and how message length relates to aversiveness, longer text or audio appears to provide users with additional space to express stronger aversive sentiments. Together, these findings suggest that references to intersectionality, as well as platform-level affordances such as word or time limits, may be meaningfully linked to the quality of online political conversation.

 

Limitations and Future Research

 

We acknowledge several limitations. First, aversiveness alone is not a definitive marker of deliberative quality; misleading or harmful content can still be conveyed in seemingly civil ways. Moreover, strict adherence to discourse norms may suppress expression, particularly among marginalized voices (Papacharissi, 2004). We therefore remain cautious about uncritically promoting civility or suppressing aversive discourse, recognizing that such expressions can, at times, serve as powerful tools for social change (Chen et al., 2019). Second, we compared posts with more Trump-related hashtags and those with more Harris-related hashtags, which only capture aggregated differences in discourse tone rather than individual user intent. Lastly, while this study focused primarily on content, future research could examine how such messages are received and disseminated by analyzing patterns of user interaction and network. As researchers have noted, certain voices receive disproportionately greater visibility on TikTok (Krutrök & Åkerlund, 2023), and future research can investigate whether and how social media algorithms contribute to this imbalance by amplifying certain themes or forms of discourse over others.

 

Despite its limitations, this study advances an identity-based perspective on political communication (Kreiss et al., 2020) by conceptualizing public discourse as a dynamic arena where political identities are constructed, negotiated, and contested. It also contributes to research on social representations by showing how intersecting identities can simultaneously enable political mobilization and reproduce structural exclusion through processes of representation (Mollenkopf, 2013). Adopting an intersectional perspective, the study examines the overlapping barriers faced by women of color political candidates (Brown et al., 2022). Our findings indicate that these intersecting identities function not merely as descriptive categories, but as contested sites through which candidates receive both support and enthusiasm as well as delegitimization and attack, underscoring intersectionality as an ongoing terrain of struggle in political representation.

 

Findings on aversiveness further extend scholarship on identity and digital platforms by suggesting that intersectional identities can become focal points of polarization and shape the quality of online discourse. TikTok’s user- and influencer-driven messaging creates a conducive environment for political meaning making (Lee & Abidin, 2023). At the same time, discussions involving intersecting identities can become sites of heightened hostility in online political discourse. Because intolerant discourse is more likely to emerge in discussions of historically marginalized groups (Rossini, 2020), these dynamics underscore the need for more responsive moderation, at both the platform and user levels, to foster more constructive online discourse.

 

References

 

Amado, A. (2024). Pop politics beyond populism: Popular culture as political communication. In A. Casero-Ripollés & P. C. López-López (Eds.), The Routledge handbook of political communication in Ibero-America (pp. 91–104). London, UK: Routledge. https://doi.org/10.4324/9781003388937

Anzia, S. F., & Berry, C. R. (2011). The Jackie (and Jill) Robinson effect: Why do congresswomen outperform congressmen? American Journal of Political Science, 55(3), 478–493. https://doi.org/10.1111/j.1540-5907.2011.00512.x

 

Avalle, M., Di Marco, N., Etta, G., Sangiorgio, E., Alipour, S., Bonetti, A., Alvisi, L., Scala, A., Baronchelli, A., Cinelli, M., & Quattrociocchi, W. (2024). Persistent interaction patterns across social media platforms and over time. Nature, 628, 582–589. https://doi.org/10.1038/s41586-024-07229-y

 

Bernstein, M. (2005). Identity politics. Annual Review of Sociology, 31(1), 47–74. https://doi.org/10.1146/annurev.soc.29.010202.100054

 

Bossetta, M. (2018). The digital architectures of social media: Comparing political campaigning on Facebook, Twitter, Instagram, and Snapchat in the 2016 U.S. election. Journalism & Mass Communication Quarterly, 95(2), 471–496. https://doi.org/10.1177/1077699018763307

 

Brown, N. E., Clark, C. J., & Mahoney, A. (2022). Women of color political elites in the US: An introduction, personal reflections, and a call for scholarly engagement. Journal of Women, Politics & Policy, 43(1), 1–7. https://doi.org/10.1080/1554477X.2022.2004844

 

Burge, C. D., Hodges, M. J., & Rinaldi, R. (2019). Family matters? Exploring media coverage of presidential candidates’ families by gender and race. Politics, Groups, and Identities, 8(5), 1022–1042. https://doi.org/10.1080/21565503.2019.1584748

 

Carey, T. E., Jr., & Lizotte, M.-K. (2019). Political experience and the intersection between race and gender. Politics, Groups, and Identities, 7(2), 243–266. https://doi.org/10.1080/21565503.2017.1354036

 

Cassese, E. C., & Barnes, T. D. (2019). Reconciling sexism and women’s support for Republican candidates: A look at gender, class, and whiteness in the 2012 and 2016 presidential races. Political Behavior, 41(3), 677–700. https://doi.org/10.1007/s11109-018-9468-2

 

Celis, K., & Erzeel, S. (2015). The complementarity advantage: Parties, representativeness and newcomers’ access to power. Parliamentary Affairs, 70(1), 43–61. https://doi.org/10.1093/pa/gsv043

 

Chen, G. M, Muddiman, A., Wilner, T., Pariser, E., & Stroud, N. J. (2019). We should not get rid of incivility online. Social Media + Society, 5(3), 1–5. https://doi.org/10.1177/2056305119862641

 

Clarke, V., & Braun, V. (2016). Thematic analysis. The Journal of Positive Psychology, 12(3), 297–298. https://doi.org/10.1080/17439760.2016.1262613

 

Corbu, N., Buturoiu, R., Ștefăniță, O., & Dumitrache, A. (2024). Exposure to online hateful content and users’ engagement: A silencing effect. International Journal of Communication, 18, 1351–1371. https://ijoc.org/index.php/ijoc/article/view/22449

 

Crenshaw, K. (1989). Demarginalizing the intersection of race and sex: A Black feminist critique of antidiscrimination doctrine, feminist theory, and antiracist politics. University of Chicago Legal Forum, 1989(1). http://chicagounbound.uchicago.edu/uclf/vol1989/iss1/8

 

Crenshaw, K. (1991). Mapping the margins: Intersectionality, identity politics, and violence against women of color. Stanford Law Review, 43(6), 1241–1299. https://doi.org/10.2307/1229039

 

Egan, P. J. (2012). Group cohesion without group mobilization: The case of lesbians, gays, and bisexuals. British Journal of Political Science, 42(3), 597–616. https://doi.org/10.1017/S0007123411000500

 

Gorman, L., & Goldenberg, C. (2024, September 19). TikTok in the 2024 US presidential race: Trending campaign strategy or passing trend? The German Marshall Fund of the United States. https://www.gmfus.org/news/tiktok-2024-us-presidential-race-trending-campaign-strategy-or-passing-trend

 

Gorman, L., Goldenberg, C., & Nieminen, I. (2024, October 30). TikTok tactics: 2024 US candidates dance around security risks. The German Marshall Fund of the United States. https://www.gmfus.org/news/tiktok-tactics-2024-us-candidates-dance-around-security-risks

 

Guest, G., MacQueen, K. M., & Namey, E. E. (2014). Introduction to applied thematic analysis. In G. Guest, K. M. MacQueen, & E. E. Namey (Eds.), Applied thematic analysis (pp. 3–20). Thousand Oaks, CA: SAGE Publications. https://doi.org/10.4135/9781483384436

 

Hall, S. (1997). The work of representation. In S. Hall (Ed.), Representation: Cultural representations and signifying practices (pp. 13–74). Thousand Oaks, CA: SAGE Publications.

 

Hmielowski, J. D., Hutchens, M. J., & Cicchirillo, V. J. (2014). Living in an age of online incivility: Examining the conditional indirect effects of online discussion on political flaming. Information, Communication & Society, 17(10), 1196–1211. https://doi.org/10.1080/1369118X.2014.899609

 

Höijer, B. (2011). Social representations theory: A new theory for media research. Nordicom Review, 32(2), 3–16. https://doi.org/10.1515/nor-2017-0109

 

Howarth, C. (2011). Representations, identity and resistance in communication. In D. Hook, B. Franks, & M. W. Bauer (Eds.), The social psychology of communication. London, UK: Palgrave Macmillan.

 

Huddy, L. (2001). From social to political identity: A critical examination of social identity theory. Political Psychology, 22(1), 127–156. https://doi.org/10.1111/0162-895X.00230

Jacobs, L. R., Cook, F. L., & Delli Carpini, M. X. (2009). The discursive turn: Citizens talking together. In L. R. Jacobs, F. L. Cook, & M. X. Delli Carpini (Eds.), Talking together: Public deliberation and political participation in America (pp. 1–27). Chicago, IL: University of Chicago Press. https://doi.org/10.7208/chicago/9780226389899.001.0001

 

Johnson, O., Hall-Phillips, A., Chung, T.-L., & Cho, H. (2019). Are you connected through consumption? The role of hashtags in political consumption. Social Media + Society, 5(4), 1–14. https://doi.org/10.1177/2056305119883427

 

Katz, E. (2014). Nowhere to go: Some dilemmas of deliberative democracy. In K. Kenski & K. H. Jamieson (Eds.), The Oxford handbook of political communication (pp. 85–94). Oxford, UK: Oxford University Press.

 

Koo, G. H., Lukito, J., Masullo, G. M., Overgaard, C. S. B., & Orr, B. (2025). The consequences of “the bird is free”: A computational analysis of aversive LGBTQIA+ tweets and engagement trends before and after Elon Musk dismantled the platform’s moderation system. New Media & Society, 1–22. https://doi.org/10.1177/14614448251356240

 

Kreiss, D., Lawrence, R. G., & McGregor, S. C. (2020). Political identity ownership: Symbolic contests to represent members of the public. Social Media + Society, 6(2), 1–5. https://doi.org/10.1177/2056305120926495

 

Krutrök, M. E., & Åkerlund, M. (2023). Through a white lens: Black victimhood, visibility, and whiteness in the Black Lives Matter movement on TikTok. Information, Communication & Society, 26(10), 1996–2014. https://doi.org/10.1080/1369118X.2022.2065211

 

Lee, J., & Abidin, C. (2023). Introduction to the special issue of “TikTok and Social Movements.” Social Media + Society, 9(1), 1–8. https://doi.org/10.1177/20563051231157452

 

Literat, I., & Kligler-Vilenchik, N. (2023). TikTok as a key platform for youth political expression: Reflecting on the opportunities and stakes involved. Social Media + Society, 9(1), 1–3. https://doi.org/10.1177/20563051231157595

 

Lukito, J., Chen, B., Masullo, G. M., & Stroud, N. J. (2024). Comparing a BERT classifier and a GPT classifier for detecting connective language across multiple social media. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing (pp. 19140–19153). Miami, FL: Association for Computational Linguistics.

 

Martella, A., & Pavan, E. (2024). “We hate her . . . and you too”: Polarized intersectionality in Italy throughout changing political scenarios. New Media & Society, 26(12), 6975–6994. https://doi.org/10.1177/14614448231160706

 

McClain, C. (2024, August 20). About half of TikTok users under 30 say they use it to keep up with politics, news. Pew Research Center. https://www.pewresearch.org/short-reads/2024/08/20/about-half-of-tiktok-users-under-30-say-they-use-it-to-keep-up-with-politics-news/

 

Mollenkopf, J. (2013). Dimensions of immigrant political incorporation. In J. Hochschild, J. Chattopadhyay, C. Gay, & M. Jones-Correa (Eds.), Outsiders no more? Models of immigrant political incorporation (pp. 107–118). Oxford, UK: Oxford University Press.

 

Moscovici, S. (1988). Notes towards a description of social representations. European Journal of Social Psychology, 18(3), 211–250. https://doi.org/10.1002/ejsp.2420180303

 

Mutz, D. C. (2006). Hearing the other side: Deliberative versus participatory democracy. Cambridge, UK: Cambridge University Press.

 

National Conference on Citizenship. (n.d.). The algorithmic transparency institute. https://www.junkipedia.org

 

NPR Washington Desk. (2024, August 6). It’s official: Kamala Harris becomes Democrats’ 2024 presidential nominee. NPR. https://www.npr.org/2024/08/01/nx-s1-5060456/harris-democratic-nominee-roll-call-delegates

 

Okimoto, T. G., & Brescoll, V. L. (2010). The price of power: Power seeking and backlash against female politicians. Personality and Social Psychology Bulletin, 36(7), 923–936. https://doi.org/10.1177/0146167210371949

 

Ophir, Y., Pruden, M. L., Walter, D., Lokmanoglu, A. D., Tebaldi, C., & Wang, R. (2023). Weaponizing reproductive rights: A mixed-method analysis of white nationalists’ discussion of abortions online. Information, Communication & Society, 26(11), 2186–2211. https://doi.org/10.1080/1369118X.2022.2077654

 

Papacharissi, Z. (2004). Democracy online: Civility, politeness, and the democratic potential of online political discussion groups. New Media & Society, 6, 259–283. https://doi.org/10.1177/1461444804041444

 

Pavan, E., & Martella, A. (2021). The release of #SilviaRomano: A case of mediated polarized intersectionality. Problemi dell’informazione, 3, 377–411. https://doi.org/10.1445/102513

 

Perspective API. (n.d.). About the API—Key concepts. https://developers.perspectiveapi.com/s/about-the-api-key-concepts?language=en_US

 

Rossini, P. (2020). Beyond incivility: Understanding patterns of uncivil and intolerant discourse in online political talk. Communication Research, 49(3), 399–425. https://doi.org/10.1177/0093650220921314

 

Runderkamp, Z., van der Pas, D., Schotel, A. L., & Mügge, L. (2022). Space invaders and norm-politicians: How the media represent the intersectional identities of Members of Parliament. European Political Science Review, 14(4), 566–582. https://doi.org/10.1017/S1755773922000339

 

Salehabadi, N., Groggel, A., Singhal, M., Roy, S. S., & Nilizadeh, S. (2022). User engagement and the toxicity of tweets. https://arxiv.org/abs/2211.03856

 

Sides, J., Tesler, M., & Vavreck, L. (2018). Identity crisis: The 2016 presidential campaign and the battle for the meaning of America. Princeton, NJ: Princeton University Press.

 

Stabile, B., Grant, A., Purohit, H., & Harris, K. (2019). Sex, lies, and stereotypes: Gendered implications of fake news for women in politics. Public Integrity, 21(5), 491–502. https://doi.org/10.1080/10999922.2019.1626695

 

Steenbergen, M. R., Bächtiger, A., Spörndli, M., & Steiner, J. (2003). Measuring political deliberation: A discourse quality index. Comparative European Politics, 1(1), 21–48. https://doi.org/10.1057/palgrave.cep.6110002

 

Trimble, L., Raphael, D., Sampert, S., Wagner, A., & Gerrits, B. (2015). Politicizing bodies: Hegemonic masculinity, heteronormativity, and racism in news representations of Canadian political party leadership candidates. Women’s Studies in Communication, 38(3), 314–330. https://doi.org/10.1080/07491409.2015.1062836

 

Wells, C., Cramer, K. J., Wagner, M. W., Alvarez, G., Friedland, L. A., Shah, D. V., Bode, L., Edgerly, S., Gabay, I., & Franklin, C. (2017). When we stop talking politics: The maintenance and closing of conversation in contentious times. Journal of Communication, 67(1), 131–157. https://doi.org/10.1111/jcom.12280

 

West, R. M. (2022). Best practice in statistics: The use of log transformation. Annals of Clinical Biochemistry, 59(3), 162–165. https://doi.org/10.1177/00045632211050531

 

Widjaya, R., Bestvater, S., & Smith, A. (2024). Who U.S. adults follow on TikTok. Pew Research Center. https://www.pewresearch.org/data-labs/2024/10/08/who-u-s-adults-follow-on-tiktok/

 

Wojcieszak, M., Sobkowicz, P., Yu, X., & Bulat, B. (2021). What information drives political polarization? Comparing the effects of in-group praise, out-group derogation, and evidence-based communications on polarization. The International Journal of Press/Politics, 27(2), 325–352. https://doi.org/10.1177/19401612211004418

Yeomans, M., Kantor, A. & Tingley, D. (2018). Detecting politeness in natural language. The R Journal, 10(2), 489–502.

 

Zeng, J., & Abidin, C. (2021). “#OkBoomer, time to meet the Zoomers”: Studying the memefication of intergenerational politics on TikTok. Information, Communication & Society, 24(16), 2459–2481. https://doi.org/10.1080/1369118X.2021.1961007

 

Zeng, J., Abidin, C., & Schäfer, M. S. (2021). Research perspectives on TikTok and its legacy apps. International Journal of Communication, 15, 3161–3172. https://ijoc.org/index.php/ijoc/article/view/14539

 

Copyright © 2026 (Gyo Hyun Koo and Dasia Clemente). Licensed under the Creative Commons Attribution Non-commercial No Derivatives (by-nc-nd). Available at https://ijoc.org.

https://doi.org/10.65476/187qx715

 

[1] The authors declare no conflicts of interest.

[2] The online supplemental appendix can be accessed at https://osf.io/jqa2b/?view_only=e659c3b6917846c2a962fbb785549276.

[3] While other social identities, such as socioeconomic status, age, and religion, are also part of Harris’s identity, this study focuses on her racial and gender identities, as these were the most prominently spotlighted in public discourse. Future research could expand this work by examining a broader range of her intersecting social identities.

[4] The analytic sample is smaller than the full data set because the log transformation analyses were restricted to posts with nonzero toxicity scores (n = 4,642) and clear classification of candidate-related hashtag use and intersectionality reference.