Introduction
Political competition increasingly revolves around social identities as well as group-based sympathies and animosities (Achen and Bartels Reference Achen and Bartels2016). Parties not only seek to promote policies or ideologies but also to define the social groups they represent and, crucially, who their opponents do not. Group appeals thus serve not only as mobilization tools but also as instruments of conflict and differentiation, which can be used to bolster parties’ own connections to key constituencies but also to undermine rivals’ claims to represent particular groups.
While previous research has shown how group appeals can shape party reputations and mobilize support (Dassonneville, Stubager, and Thau Reference Dassonneville, Stubager and Thau2026; Dausgaard and Hjorth Reference Dausgaard and Hjorth2026; Robison, Stubager, Thau et al. Reference Robison, Stubager, Thau and Tilley2021), we know far less about how they are used in inter-party conflict. Understanding their association with negative campaigning offers insights into the broader dynamics of representational competition, a central but understudied feature of identity-driven politics. Accordingly, this paper asks: when and why do parties use social group appeals to discredit their rivals?
Previous research has established various determinants of negative campaigning, including issue competition, performance in office, or leadership traits, and its effects on voters (Haselmayer Reference Haselmayer2019; Lau and Brown Rovner Reference Lau and Brown Rovner2009). Separately, a growing literature has shown that social group appeals are central to how parties connect with constituencies (Huber Reference Huber2022; Stuckelberger and Tresch Reference Stuckelberger and Tresch2024; Thau Reference Thau2021, Reference Thau2024). However, despite the potential to shape voter perceptions and electoral coalitions, the intersection of these strategies remains largely unexplored. In contrast to previous studies of negative campaigning, we examine attacks that challenge a party’s claim to represent specific social groups, showing how negative campaigning can operate in the domain of group representation. This allows parties not only to criticize their opponents in targeted and resonant ways but also to weaken the targeted group’s attachment to the rival party. Examining this practice is essential for understanding how parties compete over group representation.
We develop a framework for group-based negative campaigning grounded in social identity theory and reference group theory. Specifically, we hypothesize that group-based attacks are more likely when the attacking party has a strong reputation or when the targeted party has a weak reputation for representing a particular group, and when the asymmetry between the two parties is particularly strong. To test our expectations, we draw on a novel dataset of more than 200,000 social media posts on Facebook and Twitter from political parties in the UK between 2016 and 2024, combined with survey data from the British Election Study Internet Panel (BESIP). This approach allows us to systematically link voter perceptions of group representation with what parties say in public and how they compete with their rivals. Our findings suggest that group-based attacks follow a strategic logic of representational competition: parties target opponents who are weakly associated with a group, especially when the attacker holds a reputational advantage.
Literature review
Political parties frequently communicate using appeals to social groups, including references to social classes, racial or ethnic communities, or demographic cohorts. A growing body of research has documented how such appeals shape voter attitudes towards candidates (Dassonneville, Stubager, and Thau Reference Dassonneville, Stubager and Thau2026; Robison, Stubager, Thau et al. Reference Robison, Stubager, Thau and Tilley2021) and policies (Huber, Meyer, and Wagner Reference Huber, Meyer and Wagner2024) and help parties manage group-based reputations (Dausgaard and Hjorth Reference Dausgaard and Hjorth2026). Through these appeals, parties signal alignment with specific parts of the electorate (Converse Reference Converse2006; Miller, Wlezien, and Hildreth Reference Miller, Wlezien and Hildreth1991) and can modify their group images over time (Thau Reference Thau2021, Reference Thau2024).
At the same time, negative campaigning has attracted considerable attention in the literature on electoral campaigns (for an overview, see Haselmayer Reference Haselmayer2019; Lau and Brown Rovner Reference Lau and Brown Rovner2009). Generally, this strategy is defined as attacking another party’s policy proposals, its past record, or personnel, thereby contrasting with self-promotional ‘positive’ campaigning (Lau and Brown Rovner Reference Lau and Brown Rovner2009: 286). A core insight from this literature is that negative messages follow a strategic logic shaped by issue ownership, ideological distance, or electoral competitiveness (Haselmayer Reference Haselmayer2019; Lau, Sigelman, and Brown Rovner Reference Lau, Sigelman and Brown Rovner2007; Lau and Brown Rovner Reference Lau and Brown Rovner2009; Nai and Walter Reference Nai and Walter2015). However, this research has not examined how parties use negative campaigning in the context of social group representation.
Despite the insights offered by each strand of literature, no study to date has explored how group appeals and negative campaigning intersect. The fact that these two literatures have remained largely disconnected is notable given that group appeals can enhance the potency of attacks by connecting them to voters’ social identities and group attitudes. Recent work has begun to theorize this dynamic. Huber and Dolinsky (Reference Huber and Dolinsky2023) argue that both direct and indirect group-based appeals are central to representational competition, including by challenging their rivals’ connection to certain social groups. But to date, there are no systematic empirical studies of this connection.
This paper contributes primarily to the literature on group appeals by showing that these messages serve not only to mobilize support but also to attack opponents and challenge their representational standing. At the same time, it extends research on negative campaigning by identifying a form of attack that targets parties’ reputational linkages with social groups rather than their policy positions or leadership traits. In doing so, our findings also speak to theories of identity targeting and issue ownership, suggesting that parties may use group-based appeals to exploit vulnerabilities in their opponents’ representational standing.
Theoretical framework
Political parties compete not only over policies or ideologies but also over who speaks for whom in society. Group appeals play a crucial role in this competition as tools for mobilizing identity and reinforcing or altering partisan attachments (Thau Reference Thau2024). We argue that they are equally central to negative campaigning. Our approach is grounded in the idea that group-based negative campaigning reflects a strategic effort to reshape the social boundaries of political competition. Rather than treating group appeals and attacks as separate, we consider them as interconnected strategies that parties use to pursue two goals: first, to delegitimize rivals by framing them as unfit representatives of specific groups, and second, to reaffirm or improve their own credibility with those groups (or with other groups of voters that care about them). In other words, attacking an opponent’s relationship to a group is also a way of making claims about one’s own (Huber and Dolinsky Reference Huber and Dolinsky2023).
Drawing on social identity theory (Tajfel and Turner Reference Tajfel, Turner, Worchel and Austin1986) and reference group theory (Miller, Wlezien, and Hildreth Reference Miller, Wlezien and Hildreth1991), we understand these attacks as appealing not only to voters who identify with the group in question but also to those who feel sympathy towards the group and care about how it is treated (Conover Reference Conover1988). Voters respond to messages that resonate with their in-group identities but also to signals about other (out-) groups based on their attitudes towards the group. When parties frame an opponent as betraying, ignoring, or failing a group, they activate not just political disagreement but affect and identity-based responses that can shape how voters judge both the attacker and the target.
Group-based attacks serve multiple strategic functions because they speak to different types of audiences in different ways. First, they can reinforce a party’s standing with its core support base by signalling continued loyalty to a shared group identity. For instance, when the Labour Party criticizes the Conservatives for neglecting working-class interests, it may not necessarily be trying to convert new voters but rather to reaffirm its own commitment to its traditional base. In this sense, the attack acts as a signal of in-group protection and a reaffirmation of ‘we are with you’ directed at existing supporters. It helps to sustain partisan identity, prevent defection, and increase mobilization, particularly among voters who are highly attuned to group-based cues.
Second, group-based attacks can serve as persuasive tools aimed at voters who are politically or emotionally aligned with the targeted group, even if they are not core supporters of the attacking party. These are voters who may not see the attacking party as their primary representative but who may be open to shifting their support if they come to view a rival party as untrustworthy or harmful to a group they care about. For example, if the Conservative Party attacks Labour for failing pensioners, it may not be addressing pensioners who support Labour but instead appealing to undecided or swing voters in that demographic. The goal in such cases is to weaken rival-party attachments among voters who are not yet firmly committed to the targeted party.
Third, group-based attacks may speak to broader symbolic concerns among voters who do not belong to the group in question but care about how certain social groups are treated in public life. These voters may be motivated by sympathy towards the targeted group or other values such as fairness, justice, or inclusion (Conover Reference Conover1988) and respond positively to messages that highlight representational failures or injustices, even when they are not personally affected. For instance, an attack that frames a rival party as neglecting women may resonate with voters who view the group positively, even if they do not belong to the group themselves. Here, the appeal operates less on the basis of group identity and more through affect or solidarity, suggesting that group-based attacks can shape broader political perceptions beyond narrowly targeted constituencies.
Across these functions, what unites group-based attacks is that they allow parties to position themselves as attentive or trustworthy while portraying opponents as inattentive or indifferent. They are not merely policy critiques but attempts to redraw the symbolic boundaries between social groups and political parties. Importantly, these strategies do not come without trade-offs or potential costs. Highlighting an opponent’s lack of support for a group could backfire by prompting sympathy for the rival or alienating others who see such appeals as divisive or opportunistic. These potential side effects suggest that parties will not employ group-based negative campaigning indiscriminately. Rather, they are likely to do so under specific conditions that make such attacks more effective and less risky.
We propose that reputational asymmetries between parties are key to understanding when group-based negative campaigning occurs. Issue ownership refers to the long-term reputational association between a party and a policy domain, leading voters to perceive the party as most competent for handling a particular issue (Petrocik Reference Petrocik1996). Parties that own an issue are seen as more credible and effective on that domain, which both motivates and legitimizes attacks on rivals who are perceived as weak on the same issue (Damore Reference Damore2002; Elmelund-Præstekær Reference Elmelund-Præstekær2011; Poljak and Seeberg Reference Poljak and Seeberg2024). We argue that the same logic applies in the group domain. Just as parties develop ownership over policy issues, they develop reputational associations with specific social groups, associations that are similarly built over time through sustained appeals, historical alignments, and perceived group loyalty. Group-based negative campaigning can thus be understood as an extension of the ownership logic: parties challenge rivals precisely where those rivals are perceived as weak representatives of a group, mirroring the dynamic whereby parties attack opponents on issues they do not own.
At the same time, parties may also be more inclined to attack rivals who are perceived as weak representatives of a group, seeking to exploit these vulnerabilities to discredit their opponents or disrupt existing group-party linkages. In this way, group-based negative campaigning becomes a tool for managing reputational dynamics in the party system, used to shape perceptions of who truly represents the interests of which groups. Just as parties may use negative campaigning to challenge an opponent’s issue competence when it is weak (Elmelund-Præstekær Reference Elmelund-Præstekær2011), they can target weak group reputations as well. By drawing attention to a rival’s shortcomings, parties can more credibly damage that rival’s overall reputation, even if their own group credibility is not dominant.
Based on this framework, we formulate our first hypothesis, comprising two sub-hypotheses concerning the reputational standing of both the sender and the target party:
Hypothesis 1a (Sender): Parties are more likely to engage in group-based negative campaigning when they have a strong reputation for representing a specific social group.
Hypothesis 1b (Target): Parties are more likely to engage in group-based negative campaigning when their opponent has a weak reputation for representing a specific social group.
While we conceptualize sender and target group reputation as distinct drivers of group-based attacks, their effects are inherently relational. Parties assess their own standing not in isolation, but relative to that of their competitors. A party that enjoys a clear reputational advantage over its rival can credibly claim to ‘defend’ a group against misrepresentation, whereas a party facing a similarly competent or more credible rival may find such attacks less persuasive or even risky. Thus, group-based negative campaigning should be most likely when reputational asymmetries exist between sender and target, that is, when the sender’s perceived group reputation exceeds that of its opponent.
Hypothesis 2 (Asymmetry): Parties are more likely to engage in group-based negative campaigning when their perceived reputation for representing a social group exceeds that of the target party.
Data and methods
To investigate how political parties use group appeals in negative campaigning, we combine (1) social media posts from UK parties between 2016 and 2024 and (2) survey-based perceptions of party-group linkages from the British Election Study Internet Panel (BESIP).
The UK represents an interesting case for studying group-based negative campaigning due to its historically strong party-group associations. Specifically, the class cleavage has traditionally shaped party support and voter alignments, while Brexit has introduced a new social division. The UK’s party system, coupled with a highly competitive media environment, provides fertile ground for parties to engage in both group appeals and strategic attacks. However, its single-member district electoral system also generates incentives for personalized, candidate-centred campaigning. This means that group-based messaging in the UK context may be driven at least in part by individual politicians rather than parties as collective actors. We address this by focusing exclusively on official party accounts, which holds the unit of analysis constant and allows us to study the party-level strategic logic independently of individual personality effects, even though this design also means our findings likely underestimate the full volume of group-based negative campaigning in the UK.
We collected all publicly available Facebook and Twitter posts published by the official accounts of eight major UK partiesFootnote 1 : the Conservatives, Labour, Liberal Democrats, SNP, Greens, Reform UK, UKIP, and Plaid Cymru. The time frame (2016–2024) covers multiple election campaigns, as well as periods of regular political activity. During this period, social media has become a key medium for political communication.
Facebook and Twitter were chosen as the largest and most influential social media platforms. Even though their audiences are not representative of the general population, these platforms are particularly relevant for studying negative campaigning because they enable rapid, unfiltered communication and provide immediate visibility for attacks. In this context, attacks involving group appeals should be especially likely to surface, making social media an appropriate arena for capturing such strategies. The two platforms also differ in their audience composition: Facebook skews towards an older demographic and is widely used by the general public. Twitter (now X), by contrast, functions more as an elite network frequented by journalists, political elites, and politically engaged citizens, making it a venue for signalling to opinion leaders rather than mass mobilization (Jungherr Reference Jungherr2015). Our data capture organic party posts, which are publicly visible and distributed freely to followers. These differ from paid, targeted advertising, which allows parties to micro-target specific social groups based on demographic and behavioural data. Since organic posts function as broadcast messages to all followers, our findings speak to the public broadcast dimension of group-based negative campaigning rather than its targeted persuasion dimension.
To examine the use of group appeals in negative campaigning by UK political parties, we proceed in three steps. First, we identify posts referencing political opponents using a dictionary-based approach (see Table A.1 in the online Appendix). Specifically, the dictionary includes names of political parties, abbreviations and common informal references, and names of party leaders during the period of study. Overall, we find that of the approximately 212,000 posts in our dataset, about 12% mention another party. We tested the performance of our dictionary on a hand-coded sample of 200 postings stratified by network and party. The performance score for recall yields highly satisfactory results (0.98).
Second, to identify social group mentions, we rely on a token classification model developed by Horne, Dolinsky, and Huber (Reference Horne, Dolinsky and Huber2024). Again, the recall value of 0.97 for a random sample of 200 postings stratified by network and party indicates a very good performance of the token classifier. Postings containing both a party mention and a group mention were then manually coded by a student assistant and one of the authors to check whether the target party and social group had been identified correctly and to determine the higher-level meaningful group category that the individual group token can be allocated to.Footnote 2 Crucially, we did not treat all inter-party references as negative campaigning. We specifically code postings for the presence of an actual attack or criticism directed at a rival party, capturing the tonal dimension of negative campaigning as well as its directional aspect. This leaves us with about 8,600 posts that combine a genuine attack of a rival party with a social group appeal (examples are provided in Table A.3).
Third, we link these supply-side data to public opinion data on voters’ perceptions of party-group linkages provided by the BESIP (Fieldhouse, Green, Evans et al. Reference Fieldhouse, Green, Evans, Mellon, Prosser, Bailey, De Geus, Schmitt and van der Eijk2024). Several BESIP waves contain questions on how closely respondents believe parties look after the interests of specific social groups.Footnote 3 These items primarily tap into a broader evaluative judgment that goes beyond performance capacity alone, encompassing perceived group alignment, loyalty, and trustworthiness. This means we are unable to empirically distinguish between the competence and integrity dimensions of group reputation (cf. Bjarnøe, Adams, and Boydstun Reference Bjarnøe, Adams and Boydstun2023) and therefore treat group reputation as a unified concept.
Table A.2 in the online Appendix lists the social groups and parties covered in each wave. As the BESIP covers only a limited set of social groups, our analysis is restricted to group references that can be matched to these survey-based measures, constraining the observable set of party-group combinations. We use mean imputation for missing values across waves and parties to maximize coverage and normalize the sum of all responses across parties to 100% per group and wave.
Figure 1 shows the mean perceived group representation scores for each party across the different social groups covered in the BESIP, indicating which parties are viewed as more or less representative of each group. It highlights notable variation in group-party linkages. For instance, Labour is seen as strongly representing groups like the working class and unemployed people, while the Conservatives score higher for groups like the middle class and Christians.
Perceptions of Group Representation in the BESIP.
Figure 1. Long description
The image contains multiple dot plots, each representing the perceived political party representation across different social groups. Each plot has political parties listed on the vertical axis and a scale from 0 to 8 on the horizontal axis, indicating the level of representation. Panel A: Black and Asian. Panel B: Middle Class. Panel C: Working Class. Panel D: Unemployed. Panel E: Muslims. Panel F: Jews. Panel G: Christians. Panel H: Atheists. Panel I: Local people. Panel J: White British. Panel K: Londoners. Panel L: Young people. Panel M: Pensioners. Panel N: Highly educated. Panel O: Low educated. Panel P: Private employees. Panel Q: Public employees. Panel R: Women. Panel S: Men. Each dot plot shows the perceived representation levels of different political parties (UKIP, Reform, Con, Plaid, SNP, Green, Lib, Lab) within each social group. The positions of the dots indicate the level of representation, with higher positions indicating stronger perceived representation.
Finally, we merge the BESIP data with the social media data based on the field time of the respective wave. To preserve the temporal ordering between perceptions and messaging, we use party-group linkage scores from the most recent survey wave conducted before the respective social media post in question. This ensures that voter perceptions temporally precede the observed communication. While party rhetoric may shape perceptions (e.g., Dausgaard and Hjorth Reference Dausgaard and Hjorth2026), the temporal structure of the data reduces the risk that the estimates capture immediate feedback from messaging to perceived reputations. We then arrange the data in a sender party × target party × group category × wave format (n = 6,303) and calculate the shares of group-based attacks for each combination. Afterwards, we add the perceived group representation values for both the sender and the target party.
The dependent variable is the proportion of the sender party’s total attacks directed at a specific target party on a particular social group within a given wave. This variable is bounded between 0 and 1 and reflects the intensity of group-specific negative campaigning. Key independent variables capture the perceived group reputation of both sender and target parties, as well as their difference. Specifically, we subtract the reputation scores of the target party from those of the sender party, which means that positive values indicate a better group reputation of the sender party and vice versa. To account for group-specific salience and temporal variation in campaign dynamics, we include fixed effects for group categories and survey waves.
The data structure includes repeated observations of sender–target party pairs across groups and time. To account for this, we employ a mixed-effects fractional logit regression with a random intercept for dyads and robust standard errors. This modelling strategy is appropriate for bounded proportional outcomes (Papke and Wooldridge Reference Papke and Wooldridge1996), while the dyad-level random intercept accounts for unobserved heterogeneity in the underlying relationship between party pairs over time. Due to missing values for the dependent variable or main independent variables, the number of observations in the main regression model drops to ∼2,500.
To test the robustness of our results, we estimate a series of alternative specifications. These include fixed-effects fractional logit models, models that control for traditional predictors of attack behaviour (party size, government status, and ideological extremity), and models with sender and target fixed effects to account for unobserved time-invariant party characteristics. We further estimate models excluding observations with imputed group reputation values to ensure that our results are not driven by the imputation procedure, as well as models excluding outliers based on extreme group reputation values. Finally, we assess model consistency across different subsets of the data, focusing on the most frequent senders and targets of group-based attacks and the most commonly referenced social groups. Across all specifications, results are consistent with the main model, reinforcing the robustness of our findings (see Appendix D).
Analysis
We begin by examining the effect of sender group reputation on the likelihood of group-based negative campaigning. The results of the main regression model are presented in Table C.1 in the online Appendix, while Figure 2 displays the predicted probabilities. The coefficient for sender group reputation is close to zero and statistically insignificant, indicating that a party’s own reputation for representing a social group does not systematically influence its likelihood of using group-based attacks. The left panel of Figure 2 accordingly shows a flat line across all levels of sender reputation. This finding offers no support for Hypothesis 1a and remains consistent across all robustness checks (Table D.1), suggesting that strong group associations do not motivate parties to engage in negative campaigning.
Predicted Share of Group-Based Negative Campaigning at Different Levels of Perceived Group Representation.
By contrast, results strongly support Hypothesis 1b. The coefficient for target reputation is negative and statistically significant at the 1% level. To assess the substantive size of this effect, we calculate predicted probabilities across the observed range of target group reputation scores. As shown in the centre panel of Figure 2, parties with high group representation scores face a much lower likelihood of being targeted with group-based attacks (about 4%), while parties with low reputation scores are attacked at more than double that rate (just over 10%). These results indicate that parties are more likely to direct group-based attacks at opponents who are perceived as weak representatives of specific social groups.
Turning to H2, we find that parties strategically exploit reputational asymmetries to their advantage. As shown in Figure 2 (right panel), the predicted share of attacks on another party concerning a given group is about 6% when the sender suffers a substantial reputational deficit (−63 percentage points). This share more than doubles to roughly 13% when the reputational balance reverses in favour of the attacking party (+63 percentage points). Together, these results provide strong empirical support for the idea that group-based negative campaigning is selectively applied where it is most likely to resonate.
Several group categories also exhibit systematic differences in how likely they are to be referenced in attacks. Groups such as the working class, young people, pensioners, and public employees are significantly more likely to be referenced in group-based attacks, indicating their high political salience (see Table C.1 in the online Appendix). The prominence of groups like pensioners and public employees suggests these attacks function as symbolic signals about parties’ broader representational priorities, directed at the general electorate rather than at the targeted group itself.
Party-level differences in attack behaviour are also substantively important. As illustrated in Figure B.3 in the online Appendix, the Conservative Party is by far the most targeted party, receiving attacks from across the party system while itself engaging in relatively few group-based attacks. Labour, by contrast, is the most active sender. These patterns are consistent with government-opposition dynamics (the Conservatives were in government for nearly the entire study period), but they persist even when controlling for government status, suggesting that party-specific factors beyond incumbency also shape who attacks whom and on what groups.
Conclusion
This study examines how parties strategically use group-based rhetoric in negative campaign communication on social media. Drawing on Facebook and Twitter data from eight UK parties (2016–2024) and survey data from the BESIP, we find that parties target opponents with group-based attacks when those opponents have a weak reputation for representing the group in question, and especially when the attacker enjoys a reputational advantage. This suggests that decisions about when to attack are shaped not only by issue ownership or electoral competitiveness, as emphasized in existing research, but also by reputational asymmetries in social group representation. More specifically, our findings point to a logic of targeted credibility destruction: parties invest in group-based attacks when they can most plausibly frame a rival as an inadequate representative, and especially when they themselves hold a reputational advantage that lends credibility to the attack. This functions less as direct persuasion towards the attacker and more as dissuasion from the target, convincing undecided and weakly attached voters that the targeted party’s poor representational record is real and disqualifying.
Importantly, our findings point to a distinct logic underlying negative group-based attacks compared to positive group-based appeals. While positive appeals typically aim to strengthen a party’s own association with a group (Thau Reference Thau2021, Reference Thau2024), negative attacks seek to delegitimize a rival’s claim to represent a group, weakening voters’ trust in that rival and thereby reducing their choice set. This mirrors the issue ownership logic: just as parties attack rivals on issues where the rival is perceived as weak (Elmelund-Præstekær Reference Elmelund-Præstekær2011), they likewise exploit reputational vulnerabilities in the group domain. The result is not primarily about converting existing supporters of the attacked party but about shaping undecided voters’ perceptions of the targeted party’s representational failures. Whether such attacks actually produce downstream shifts in group-party alignments among voters remains an important open question that future work should address, for instance, by examining their effects on group identification, party sympathy, and vote intention.
These findings highlight a more group-based logic of negative campaigning that could be particularly potent in an era where social identities and group attitudes have become central to political conflict. From a normative perspective, this raises important questions about the consequences of group-based negative campaigning. While group appeals can strengthen political engagement among specific communities, their use in negative campaigns risks polarizing public discourse along social lines and reinforcing intergroup divisions. This tactic may sharpen political distinctions, but at the cost of fostering greater antagonism in political debates.
While our study is the first to examine the role of group appeals in negative campaigning, it also comes with several limitations. First, since we focus on the UK context, the generalizability of our findings to other multiparty democracies remains to be explored. Different party systems, media environments, and political cultures may shape how group-based negative campaigning operates. Second, while we rely on a robust measure of party reputations using voter perceptions from BESIP, these perceptions may not fully capture how parties themselves interpret or react to group reputational dynamics. Third, prior research shows that the prevalence and tone of negative campaigning can vary across communication channels, as parties adapt their strategies to different media environments (Walter and Vliegenthart Reference Walter and Vliegenthart2010). Social media may encourage more direct and conflictual messaging because parties can bypass journalistic mediation and compete for attention in highly visible online environments. While the strategic logic identified here should also apply in other forms of party communication, the frequency and intensity of group-based attacks may differ across channels such as press releases or mediated coverage in newspapers. Future research should examine whether similar patterns emerge in other communication contexts.
A further limitation concerns our focus on official party accounts. In the UK’s personalized political system, individual politicians may be more likely to engage in negative campaigning than parties as collective actors. By focusing on party accounts, our study likely underestimates the overall volume of group-based negative campaigning and cannot speak to the role of individual politicians in this process. At the same time, this design choice allows us to study the strategic communication of parties as organizations independently of individual personality effects and personal career incentives. Future research should systematically compare group-based negative campaigning between parties and individual politicians.
In sum, our findings underscore that group-based negative campaigning is an important tool for party communication. Understanding when and why parties choose to wield it sheds light on broader patterns of political competition and identity mobilization in contemporary democracies.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1475676526101844.
Data availability statement
The British Election Study Internet Panel data are available for download from https://www.britishelectionstudy.com.
Acknowledgements
Previous versions of the paper were presented at the Department of Political Science at the Paris Lodron University of Salzburg in November 2025, at the 2025 EPSA conference in Madrid, and at the Party Politics and Government Coalitions Workshop at Aarhus University in October 2024. We thank all participants as well as the three anonymous reviewers for their helpful comments and suggestions.
Funding statement
The authors received no external funding for this research.
Competing interests
The authors are not aware of any conflicting interests.
Use of AI
Generative AI tools were used for language editing; all analytical decisions, interpretations, and final text are the responsibility of the authors.
Ethical standards
The analyses are based on publicly available posts from official party accounts and anonymized secondary data, so no ethics approval was required.
Open access
