This is not America

The literature on campaigns has been dominated by the US experience, with many of the best accounts written by academics and practitioners focused on what works and why in the context of American elections (e.g. Brader Reference Brader2006; Green and Gerber Reference Green and Gerber2015; Hillygus and Shields Reference Hillygus and Shields2008; Plouffe Reference Plouffe2010; Vavreck Reference Vavreck2009). Although there were some accounts of campaigning in other parts of the globe (e.g. Bowler and Farrell Reference Bowler and Farrell1996) and some descriptive accounts of campaigns within studies of individual elections, the USA was the primary focus in most of the literature, with much of the rest of the coverage focused on Western Europe, particularly the UK. A large slice of the literature derived from the US or UK experience seemed to indicate that – despite the noise, energy and excitement around campaigns – ultimately they did not have much of an impact on election outcomes, which were driven rather by the fundamentals in what was labelled the ‘minimal effects’ model (Finkel Reference Finkel1993). Nonetheless, it is striking that scholars focused on the local level of campaigning in the UK did find effects (e.g. Fisher et al. Reference Fisher, Cutts and Fieldhouse2011) linking to points about place which I return to below.

Much of the discussion around the adoption of new modes of campaigning in regions such as East Asia is explicitly or implicitly linked to what is dubbed an ‘Americanization’ of campaigning (Kiyohara et al. Reference Kiyohara, Maeshima and Owen2018), often as a synonym of ‘modernization’, highlighting how the USA is often taken as the starting point for analysis. The US bias in the study of campaigning has manifested itself in two ways: as a source of inspiration for practitioners and as a benchmark for scholars. In terms of inspiration, some USA-based campaigns acquire almost mythical status. Barack Obama and Donald Trump’s triumphs in 2008 and 2016, for instance, became sources of inspiration to friends and foe alike. Campaigning is often linked to fads and fashions, with a herd-like mentality, copying and borrowing approaches that have worked elsewhere. Such borrowing, especially from the US case, is fuelled by the well-developed industry of political consultants and operatives who have been at the forefront of devising strategies and utilizing technological advances. Indeed, there are good reasons to expect an adoption of American strategies given both the army of campaign professionals from the United States and widespread knowledge of the English language. But there are also reasons at the very least for expecting a less than wholesale adoption of the American model and rather some form of hybrid between American style and the indigenous ways (Kiyohara et al. Reference Kiyohara, Maeshima and Owen2018: 5). Moreover, the spread of tools and strategies that work, what Taylor Boas (Reference Boas2016: 18) labels ‘success contagion’, may be stronger within countries and proximate cases than from developments in a geographical and culturally distant United States.

We would be wise to avoid seeing the United States as some kind of benchmark against which we assess campaigning practices. First, the electoral and party systems in the USA make it an unusual case to use as a standard. Given a two-party system and deep partisan links, campaigning in the USA has historically been mostly about mobilization, convincing voters inclined towards one party to turn out on election day (Green and Gerber Reference Green and Gerber2015). Although the nature of partisanship is changing in the US case – now often driven more by animosity towards the opponent rather than love for one’s own side – campaigning remains largely about mobilization. In contrast, elsewhere across the globe campaigns are conducted in fluid multiparty systems with weak ties between the subjects and objects of voting and where, by extension, campaigns focus as much on conversion (i.e. seeking to persuade voters to switch parties) than on mobilizing those voters to turn out on election day. Second, the sheer volume of money swashing around in US politics stands in stark contrast to many other countries, even those that are labelled as advanced democracies. Indeed, comparative studies of data-driven campaigning in countries such as the UK, Australia, Germany and the Netherlands show that the United States is the outlier given the deeper pockets of American campaigns (Kefford et al. Reference Kefford2023).

The past decade, however, has seen a geographical rebalancing, with much more scholarly attention paid to campaigning beyond the USA and UK (e.g. Bleck and van der Walle Reference Bleck and van der Walle2018; Horowitz Reference Horowitz2022; Kefford Reference Kefford2021; Kiyohara et al. Reference Kiyohara, Maeshima and Owen2018; Muñoz Reference Muñoz2019; West Reference West2020). These books, alongside recent scholarship focused on specific modes of campaigning in the USA and Western Europe (e.g. Dommett et al. Reference Dommett, Kefford and Kruschinski2024; Gibson Reference Gibson2020), offer insights into campaigning in different parts of the globe, but also link campaigning to wider themes of politics. These accounts highlight that the United States is not an ideal benchmark for many other countries given the electoral system, strong partisan ties, the stability of the party system and the significantly larger amounts of money involved (Johnson Reference Johnson2016; Oklobdzija Reference Oklobdzija2024). Moreover, much of the emergent literature also underscores variations not just between different regions of the globe, but within the same systems, with newer and challenger parties often more likely to embrace new technologies.

As I discuss below, some of the most significant contributions to the literature from outside the Anglo-American world focus on more traditional forms of campaigning. The research of Paula Muñoz (Reference Muñoz2019) and Jeremy Horowitz (Reference Horowitz2022) on rallies in Peru and Kenya in the 2010s was conducted at a time and in places where we might have expected these modes to matter. But will we see a convergence on a Western or American model? Technological advances in recent years are clearly reshaping campaigning, but their impacts are also shaped by interaction with wider social trends, such as the growing size of the youth vote and processes of urbanization in Africa that are changing the calculus of campaigning. But even here we would be wise to assume the movement is always towards what some might dub more advanced countries. Technological advances rarely occur in the same way in different places. Moreover, innovation is sometimes present in countries often seen as less developed. It is worth remembering that the ruling Jubilee coalition in Kenya worked with Cambridge Analytica in the 2013 election where some of the techniques used in 2016 by Trump and the Brexit referendum were first employed (Nyabola Reference Nyabola2018: 161).

New technology and tools

Political campaigning is changing. Not only are technological advances altering the way citizens acquire information about events and the manner in which politicians and parties convey their messages to voters, they are impacting how political actors organize their campaigns and can facilitate a much quicker sharing of recipes for success. Indeed, Holli Semetko and Hubert Tworzecki (Reference Semetko, Tworzecki and Fisher2018: 293) argue that we have entered a new era of campaigning thanks to the massive increase in the volume of electronically stored information, an era characterized by a similarity of strategies and techniques used across the globe, aided and abetted by the ‘transnational circulation of money, personnel and know-how’. Indeed, much of the focus on campaigning tends to stress how changes in campaigning are driven by technological advances and the way in which communication can be directed to an increasingly specialized and targeted audience through the steps from broadcasting through narrowcasting and microcasting to nanocasting (West Reference West2018). Although we might question the simple linearity in some of these accounts (the reality is often much messier) (Paget Reference Paget2019), there is a clear general direction of travel.

It is not just the direction of travel that matters, but also the speed. The tempo of technological change and academic publishing are very different. Many of the findings from research into campaigning already feel distinctly dated by the time they are published and read. Indeed, Glenn Kefford (Reference Kefford2021: x) openly acknowledges in the preface to his book that thanks to the ‘pace of change in the campaign environment’, by the time his account is published it ‘will already be dated’. The velocity of change becomes particularly striking when we consider the use of different social media platforms. Although Facebook, for instance, was central to Rodrigo Duterte’s electoral success in 2016, six years later the preferred social media platform in elections in the Philippines was TikTok, and it is a worthwhile bet that in another six years a different platform may come to dominate. Nonetheless, although messages are forged to resonate with contemporary audiences, they are frequently variations on well-worn themes linked to performance, capability and the risks and rewards of change.

At the heart of books by Shoko Kiyohara and colleagues (Reference Kiyohara, Maeshima and Owen2018) and Rachel Gibson (Reference Gibson2020) is the view that the internet and social media have been game-changers in political campaigning. Drawing on the experience of four countries (the UK, Australia, France and the US), Gibson provides a clear and well-marshalled account of developments. Few would disagree with a central claim of When the Nerds Go Marching In that ‘over the past two decades we have seen digital technology move from the margins to the mainstream of political campaigning within Western democracies’ (Gibson Reference Gibson2020: 210). The great strength of Gibson’s book is the way she charts the development of digital campaigning through four main phases: experimentation; standardization and professionalization; community-building and activist mobilization; and direct/individual voter mobilization. It is a story from the early experimental hit-and-miss days to highly sophisticated targeting. Digital technology, as Gibson (Reference Gibson2020: 2) argues, has helped change the ‘“art” of campaigning into something more of a science’.

Although Gibson offers excellent coverage of her four main cases, as for all studies she has to draw a chronological line in the sand for her data analysis (in her case 2015), meaning she does not offer insights into some of the most striking and consequential electoral events in those countries’ recent history: the Brexit referendum in the UK and the elections of Emmanuel Macron in France and Donald Trump in the US. While studies of the 2022 French presidential election have cast doubt on the power of social media given its limited use (Neihouser et al. Reference Neihouser2022), the embrace of digital technology clearly played a role in Trump’s 2016 campaign, and the over-reliance, almost slavish belief in the power of data and her team’s algorithm led Hillary Clinton’s rival campaign to plough the wrong furrows. Moreover, received wisdom of the Brexit campaign has tended to emphasize the importance of social media and effective targeting in the final two weeks of the campaign. Nonetheless, in the cases of both Brexit and Trump, messages mattered. Not only did close observers of Clinton’s campaign suggest the real weakness was the lack of a simple clear message (Allen and Parnes Reference Allen and Parnes2017), but the impact of social media targeted campaigning in the Brexit referendum underlined the power of the alluring slogan ‘Take Back Control’ and the infamous pledge on the side of a bus that promised to give the National Health Service rather than Brussels £350 million a week.

The four phases of digital development, however, have not played out in the same way across the globe. Rather than following the same linear path in a process of replication, as with common patterns of technological development, we see bypassing and leapfrogging as those behind the curve seek not just to catch up, but to jump to the most recent innovation (Lee Reference Lee and Lee2021). Having not gone through all the stages, previous laggards are less encumbered. Indeed, a desire to be at the head of the game provokes political operatives to go straight for the new, shiny methods. But the adoption of certain tools is also linked to wider technological changes and the state of communications. Given, for instance, the dominance of mobile-based platforms like m-Pesa in Kenya for the sending and receiving of money (thanks to the reluctance of physical banks to provide services to large swathes of the population), it is no surprise that campaigners quickly realized the advantages of communicating campaign messages in ways also consumable on such devices (Nyabola Reference Nyabola2018: 64).

Data-driven campaigning

Whilst the work of Gibson (Reference Gibson2020) and others highlights the overarching trends and possibilities of data-driven campaigning, in a series of pioneering publications, Katharine Dommett, Glenn Kefford and colleagues examine the reality (Dommett et al. Reference Dommett, Kefford and Kruschinski2024; Kefford Reference Kefford2021; Kefford et al. Reference Kefford2023). One of the great strengths of their scholarship is the casting of some doubt on the claims about the prevalence and power of data-driven campaigning (DDC). Some of what is written ‘is the work of public relations teams looking to drum up business for their clients. Some is campaigns exaggerating their capacities to scare their opponents’ and ‘the rest are fantasies about how a new software program swung voters one way or the other, or how a new dataset completely revolutionalised how we understand women aged between 35 and 45 who live in regional areas’ (Kefford Reference Kefford2021: 2). This is a useful corrective. It is easy to get dazzled and bewitched by technological advances. Indeed, as Kefford argues, much media coverage ‘conflates the perceived sophistication of data and analytics operations’ with the ‘claimed efficacy of such campaigns’ (Kefford Reference Kefford2021: 4, emphasis in the original).

Kefford and Dommett and colleagues’ research underscores that there is often a large gap between the rhetoric and the reality. Indeed, the ‘vast majority of parties’ data-driven practices are mundane, predictable, and removed from the hyperbolic accounts that dominate popular commentary’ (Dommett et al. Reference Dommett, Kefford and Kruschinski2024: 5). Drawing on the cases of Australia, Canada, Germany, the UK and the US, their work offers two main contributions to our understanding of campaigning, one related to the raw material – data – and one to the deployment of that data. At the heart of DDC is data and these data need to be acquired. Such data can be publicly available, disclosed (given freely by citizens), generated by monitoring and inferential. But there are important variations in the supply and ability to use those data in different countries given the variety of rules about what data are publicly available, what citizens are willing to freely disclose and what government regulations exist. Although much can be gleaned from data publicly acquired or through sophisticated monitoring and profiling of the digital space, there is a still a vital role for grassroots actors to collect information to feed into the calculations, underlining not just the benefits of activists on the ground, but also the advantages of larger parties who have the capital and labour resources to collect and acquire data, and the reinforcement of existing hierarchies in party systems (Dommett et al. Reference Dommett, Kefford and Kruschinski2024: 55; Kefford et al. Reference Kefford2023: 452).

Dommett and colleagues highlight a gap between the potential, accorded by the acquisition of data and the statistical and algorithmic tools that can be deployed, and the reality. This gap is driven by resources, time and beliefs in its efficaciousness. In most systems – the USA here is an exception – limited funds mean that many parties do not engage in the complex segmentation and modelling necessary to identify and target highly granular personality profiles, preferring to focus on broader groups. Moreover, in the heat of the campaign there may not be the time to deploy data-driven campaigning to its full extent. The lack of resources or time could be overcome by a change in the funding environment, and the development of more sophisticated tools that require less human input and can be deployed more cheaply. Nevertheless, whilst there may be developments that change elements in the cost–benefit calculations, there are lingering doubts whether such a focus on fine-grained analysis is the best use of resources. Sometimes that scepticism may be well founded given the perceived flimsy foundations of data on which the models are built. But even if the data can identify the ‘right’ people, such data may be far less fruitful in identifying which messages have persuasive power (Tappin et al. Reference Tappin2024: 7).

More broadly, scholarship focused on the reality of campaigning underscores that the impact of technological advances is not about supplanting traditional forms of campaigning nor that there is a battle between digital and pre-digital forms of campaigning. It is both clicks and mortar: a campaign’s social media posts of a field campaign can then be amplified by sharing on social media in a process of ‘digital circulation’ of campaign content (Baldwin-Philippi Reference Baldwin-Philippi2015: 68). Indeed, data can help make a ground game more efficacious by maximizing a candidate’s time with voters who might respond well to a knock on the door and having their flesh pressed. Moreover, studies of new movement parties that trumpeted their internet credentials, like the Five Star Movement in Italy, highlighted the continued relevance of traditional partisan grassroots organization and bottom-up mobilization of potential supporters. Although social networks played a role in ‘amplifying the effectiveness of a campaign’ (Bischof and Kurer Reference Bischof and Kurer2023: 29), there was a particular potency of place-based political mobilization.

While Dommett, Kefford and colleagues offer much to understand the use of data in campaigning, scholarship from other parts of the globe illuminates how that data and the new tools for spreading that data work in practice. Not only has research shown, for instance, that Nigerian parties were far better at harnessing platforms like WhatsApp than those in the United Kingdom in the 2010s (Cheeseman et al. Reference Cheeseman2020), but it also underscores the broader ‘social media ecosystem’ involving supporters, influencers, trolls and intermediaries (Sinpeng et al. Reference Sinpeng, Gueorguiev and Arugay2020). These actors are not just objects of influence, but rather are also subjects shaping the message. Individuals at the centre of campaigns may have significant power, but they can unleash forces over which they have little control. Indeed, research on elections in the Philippines, Indonesia and Malaysia suggests we might be better off conceiving of two campaigns: one formal, mainstream, media-driven discourse; and on social media platforms, a different campaign that is subversive, underground and based around identity politics and disinformation (Tapsell Reference Tapsell2021). Moreover, research into campaigning in Africa highlights well that the power of social media messages lies as much in their role in stimulating the spreading of those messages off-line through ‘pavement radio’ and the equivalent of watercooler moments as in the direct impact of social media.

The online and off-line role of social media underlines a deeper interconnected media ecosystem, blurring the distinction between the ‘connected’ and the ‘disconnected’ (Cheeseman et al. Reference Cheeseman2020; Gadjanova et al. Reference Gadjanova, Lynch and Saibu2022). These studies of East Asia and Africa highlight that message control and message discipline can be both enhanced and undermined by digital tools. WhatsApp, for example, can re-route voters’ attention towards a central campaign message, but it can just as easily be taken off on a very different path – whether by accident or design. Furthermore, certain social media platforms, such as Instagram, for instance, accord influencers a potentially significant impact. There can be a tendency to lump all forms of digital campaigning into one basket, but just as large-scale rallies, town hall meetings and door-knocking play contrasting roles and have varied effects in field campaigns, so different digital platforms and digital tools communicate in different ways and have the ability to reach different audiences, whether different generations or those more or less engaged in politics, often requiring variations in tone, language and imagery (e.g. Albertazzi and Bonansinga Reference Albertazzi and Bonansinga2023; Williams et al. Reference Williams2022).

Parties, party systems and power

Technological advances do not just provide new modes of campaigning, they also have impacts on both power structures within parties and power dynamics between parties (Bennett and Segerberg Reference Bennett and Segerberg2013; Kiyohara et al. Reference Kiyohara, Maeshima and Owen2018). On the one hand, the expertise needed to run DDC sees a rise of the nerds and at least a partial changing of the guard at the top. But it can be empowering for grassroots activists. Moreover, although DDC may invoke images of data geeks hunched over their laptops, it is worth stressing that DDC requires a basket of expertise: not just algorithmic programming and statistical modelling, but also experts in advertising, digital communication and so on (Dommett et al. Reference Dommett, Kefford and Kruschinski2024).

Indeed, the choice of technology impacts on the power dynamics within a campaign. Hierarchical structures can be challenged by citizens’ autonomy in cyberspace. Free from the control of a managed campaign, ordinary citizens can reproduce campaigning messages in different and interactive ways which can be highly effective (Meikle Reference Meikle2016). Aim Sinpeng and colleagues' (Reference Sinpeng, Gueorguiev and Arugay2020: 370) study of the 2016 elections in the Philippines highlighted how Duterte’s Facebook presence was ‘underwhelming, unengaging and generally unprofessional’, yet his Facebook page quickly emerged as a centre of attention thanks in no small part to his vocal and ardent digital supporters (and, it must be added, some bots and trolls). Nonetheless, research on campaigning in Africa challenges the idea that digital has changed power structures in political parties. Examining the study of the use of WhatsApp in two countries with very different levels of party institutionalization, Nigeria and Ghana, Jonathan Fisher and colleagues (Reference Fisher, Gadjanova and Hitchen2024: 935) found that although space had been formed for younger digitally savvy entrepreneurs, ‘ultimately’ that group’s ‘wider political reach continues to be filtered through existing patrimonial structures’, leading to an intensification of ‘pre-existing patterns of party politics’.

In terms of party systems, technological change has offered opportunities to outsiders. Given limited access to traditional mainstream media for insurgents, necessity is sometimes the mother of invention. But new technology can not only convey the energy and excitement around those new entrants on the political scene (Stromer-Galley Reference Stromer-Galley2019) and be a ‘valence characteristic’ (Bleck and van der Walle Reference Bleck and van der Walle2018: 174), it can also reinforce their newness, challenger status and modernity, as we see in a variety of cases like the Pirate parties, Nayib Bukele in El Salvador and a host of new parties in Central and Eastern Europe (Haughton and Deegan-Krause Reference Haughton and Deegan-Krause2020; Lupu et al. Reference Lupu, Ramírez Bustamante and Zechmeister2020). Nevertheless, although new parties have some positional advantages, they are more likely to suffer from a resource disadvantage. Significant resources are required to develop and run complex voter-management systems. New data-driven modes of digital campaigning, therefore, confront smaller parties with an even greater challenge of competing with their bigger rivals. In short, whilst technological advances seemingly help forge a more open and competitive space, ‘as embedded political structures begin to see the benefits’ of the new tools, ‘pre-existing power relationships re-emerge’ (Gibson Reference Gibson2020: 12). But in the ruthless world of politics such positions of power can be precarious. Political parties need both resources and specialized skills, as well as ‘organizational flexibility’ and a willingness to use new tools to keep pace with ‘ongoing changes in political communication environments and the wavering political loyalties of the electorate’ (Mykkänen et al. Reference Mykkänen, Nord and Moring2022: 1).

Rallying the troops

While the spotlight on campaigning in the USA and Western Europe has been largely focused on the deployment of data-driven tools, some of the most insightful research on campaigning in Africa and Latin America has focused on campaign events, especially rallies. The emerging literature underscores three key aspects of campaigning: the informational role of campaign events, the means to lure attendees to such events and the importance of timing.

Just as using modern technology carries a meaning of modernity, so using rallies as a campaigning tool carries meaning. The key informational role is to signify strength, popularity and momentum, but rallies can also send messages related to place and to specific groups if using certain languages or dancing in particular styles (Kramon Reference Kramon2017; Paget et al. Reference Paget, Beardsworth and Lynch2023: 235). Drawing on the experience of Peru, Muñoz (Reference Muñoz2019) suggests rallies offer a message to those present, and also to a broader audience about the popularity of a candidate or party. ‘In uncertain and volatile electoral settings’, she argues, ‘high turnout at rallies affects the dynamics of the race by establishing name recognition, maintaining electoral reputation, narrowing the field of viable contenders, and attracting strategic voters in the final rush’ (Muñoz Reference Muñoz2019: 15). Projecting a message of popularity can be important both for activists and supporters who may need to be fired up and reminded they are not alone, as well as shifting voters’ preferences. Muñoz (Reference Muñoz2019: 125) cites an El Instituto de Opinión Pública survey from 2012 which suggests that nearly 60% of voters were convinced to support or not a candidate from campaign events.

Nonetheless, these tools can fail and backfire. Poorly organized and sparely attended rallies can harm a campaign. The visual effect of the sheer volume of attendees can be powerful, as indeed can the type of people who come, both in terms of connecting kindred spirits and reinforcing the perception that a campaign is grassroots-driven. But rallies have informative functions in other ways, not just to existing and potential patrons and donors that the candidate or party is a horse worth backing (Szwarcberg Reference Szwarcberg2015): they can be a means of harvesting contact details for party-building purposes (Kumar Reference Kumar2022). Moreover, campaign rallies, particularly the ‘heavily attended, boisterous and sometimes controversial’ ones, can feed the media’s appetite for newsworthy items, a fact not lost on the organizers of Donald Trump’s rallies (Stromer-Galley Reference Stromer-Galley2019: 182). Furthermore, akin to other forms of campaign visits (Cutts and Middleton Reference Cutts and Middleton2025), rallies can send out a strong message about the importance of the place and its inhabitants. Physical presence is important, but it is worth recalling this can be manufactured. In addition to attending 437 rallies in person across India in 2014 during his campaign, Narendra Modi also attended several rallies as a 3D hologram (Jethwaney and Kapur Reference Jethwaney and Kapur2019: 169, 174–5). A lack of physical presence is also not a weakness when we think of perhaps the greatest informational power of rallies in modern politics. The sights and sounds of the rally can be – and are – shared on social media, meaning it is not only the messages of the rally that can be conveyed, but also the spectacle.

The informational function of rallies is central to Gabrielle Lynch’s (Reference Lynch2023) idea of ‘rally-centric’ campaigning. Her study of campaigns in Kenya suggests that rallies constitute a hybrid form of political communication that simultaneously and deliberately targets face-to-face and mediatized audiences with tailored messages. Indeed, the modern media environment provides both opportunities and constraints for more traditional modes of campaigning. Previously, messages from rallies could be directed clearly to specific audiences and almost hermetically sealed off, but in an era of social media anything you do at a rally can now be clipped and broadcast. The focus of Dan Paget’s (Reference Paget2019) and Lynch’s research is on countries such as Tanzania and Kenya, where campaign rally attendance is high. In the 2022 elections in Kenya, for instance, over 50% of the population attended a campaign rally (Lynch Reference Lynch2023: 344), underlying the cogency of claims about rallies being a principal means of communicating. Moreover, this claim strikes a chord with other cases such as India where rallies are central to campaigns and indeed with Trump’s use of rallies in the United States (Kumar Reference Kumar2022). Where rally attendance is much lower, however, the rally-centric label fits less well. But the great merit of the scholarship of Lynch and Paget and colleagues (Reference Paget, Beardsworth and Lynch2023) is to underline that link between rallies and other modes of campaigning, especially social media.

Muñoz (Reference Muñoz2019) ties together the informational function of large-scale party gatherings with clientelism. She argues that politicians use the offer of inexpensive consumer goods to buy the participation of voters at campaign events in order to grab their attention. The large crowds do not just persuade voters of candidates’ desirability, they signal electoral viability to voters and donors. Muñoz challenges the conventional wisdom of clientelism in a number of ways. Prevailing approaches to clientelism as an electoral strategy, for instance, see the main goal of parties or candidates who distribute goods during campaigns as buying votes or impacting turnout directly on election day. Muñoz’s Buying Audiences (Reference Muñoz2019), however, highlights the unfolding dynamics of the campaign itself and stresses the indirect effects that early investments in electoral clientelism have on electoral choices, pointing to the importance of timing in campaigning.

Muñoz’s stress on the distribution of goods to those individuals who attend rallies and the wider communities where the rallies take place finds echoes not just elsewhere in Latin America, but in other parts of the globe where such offers of tea, tuna, televisions and the promise of cataract surgeries have all been used in campaigns (Bleck and van der Walle Reference Bleck and van der Walle2018; Koter Reference Koter2017; West Reference West2020). Nonetheless, the importance of clientelism needs to be qualified. Recent research on sub-Saharan Africa, for instance, suggests that poor voters reject these kinds of exchanges if they occur outside of a broader relationship with a candidate – that is, it is not possible to buy attendance or votes – but they can be encouraged once the candidacy is seen to be suitable and viable. Thus the election campaign cannot be divorced from the wider set of relationships between a candidate and a community (Cheeseman et al. Reference Cheeseman, Lynch and Willis2021). Moreover, we should not ignore the entertaining and alluring aspects or rallies beyond the clientelistic. Paget’s (Reference Paget2019) study showed nearly 69% of all Tanzanian voters attended a political rally during the last month of the 2015 campaign, lured as much – if not more – by the prospect of music as by gifts. Nevertheless, in both cases resources are needed. As many prospective voters can be met at once, rallies can be cheaper than other forms of campaigning (Brierley and Kramon Reference Brierley and Kramon2020: 583–584), but crucially a series of rallies can quickly rack up significant costs, making it much easier for incumbents, especially in more rural areas, who can marshal resources from natural resource rents or from a growing economy (Bleck and van de Walle Reference Bleck and van der Walle2018; Brierley and Kramon Reference Brierley and Kramon2020).

A message for the swingers

In a similar vein to Lynch, Muñoz and Paget, using the case of Kenya, Horowitz (Reference Horowitz2022) illuminates the importance of rallies. In the 2007, 2013 and 2017 presidential elections the main candidates held an average of 83 rallies over the three months prior to election day (Horowitz Reference Horowitz2022: 65). It is not just the volume, but the location of campaign rallies that matters in a country where more than 20% of voters changed their voting intentions in the run-up to election day. Horowitz’s theory of campaign strategy challenges what he dubs the ‘core mobilization model’ that proposes in ‘diverse societies parties have little incentive to court voters outside their core basis or to share resources beyond them’ (Horowitz Reference Horowitz2022: 65, emphasis in the original). In contrast, he offers the swing voter-targeting model, which suggests that in highly diverse societies ‘parties will have incentives to concentrate their campaign efforts on courting potential swing voters outside their ethnic strongholds and to opt for universal policies in place of those that favour core ethnic clientele’ (Horowitz Reference Horowitz2022: 4, 5). Key to Horowitz’s study, Multiethnic Democracy, is the distinction between those voters from groups that have a co-ethnic among the major contenders and those that do not; the former were two and half times more likely to shift their vote than the latter. In contrast to countries with highly institutionalized party systems and strong partisan attachments where the focus of campaigning is more about getting out the vote, Horowitz highlights the importance of persuasion in Kenyan elections. He provides details on the location of rallies to show that the major presidential aspirants who enjoy strong co-ethnic support at the beginning of the race opt for extensive strategies including in areas where opponents hold an advantage.

It is not just the mode of campaigning that matters, but also the message. Horowitz shows how any candidate seeking to win needs to eschew the narrow particularistic appeals that are often associated with electorates where ethnicity is salient and instead opt for more universal appeals. His argument is not that presidential candidates avoid targeted appeals related to concerns in specific localities, but rather that parties employ ‘particularistic promises as part of a nationally-orientated electoral strategy, offering targeted pledges across localities and ethnic communities’, in what he dubs a ‘universal approach to particularism’ (Horowitz Reference Horowitz2022: 93, emphasis in the original). Horowitz’s coding of nearly a hundred speeches in the 2007 Kenyan election battle between Mwai Kibaki and Raila Odinga indicates around two-thirds of the appeals were universal messages. Moreover, where such speeches emphasized promises to specific localities, they were used to shine a spotlight on national policies.

There are, however, two main limitations to Horowitz’s account. First, he is aware that his arguments will not travel everywhere. In Kenya, the largest ethnic group, the Kikuyu, make up fewer than one-fifth of voters, so there is a clear incentive to reach out to other groups in order to garner sufficient votes to win an election, an incentive that would be diminished the greater the proportion a candidate’s/party’s ethnic group accounts for in the overall population. Second, his research was conducted largely before the widespread use of mobile phones. Place and the blend of universal and particularist appeals matter in campaigning, but as Lynch’s scholarship reminds us, given mediatized messages about place, we cannot simply equate rally location with a target audience. Moreover, given the different investments of time and hence the prior attachments of voters, we might also postulate that the message for the swingers is even stronger in the online clips of the rally than to the physically assembled crowd.

Ultimately, whatever modes or messages are employed in campaigns, their aim is to impact electoral choices, both whether to cast a vote and for whom. One of the great strengths of the work of Horowitz and Kefford are the links they make between campaigning and theories of voting behaviour, but even their books could have offered more reflections on how campaigning interacts, enhances or diminishes the process of voting choice. Indeed, much of the scholarship from beyond the Anglo-American world provides not just raw material for challenging a widespread assumption from the US experience that campaigns are ‘epiphenomenal’ (Johnston and Lachance Reference Johnston and Lachance2022: 1), but it also points to the limitations of voting theories based on group identities and partisan attachments in many countries across the globe. Not only are attachments between parties and voters now weaker in longer-established democracies, but in younger democracies such ties are harder to form in the first place, a challenge only reinforced by the ‘evanescent’ nature of many new parties (Aguilar and Conroy-Krutz Reference Aguilar, Conroy-Krutz, Suhay, Grofman and Treschsel2020: 772). Central and Eastern Europe, for instance, provides many examples of parties, especially new parties, experiencing rapid increases in the weeks running up to polling day. Regardless of how voters received certain messages from the campaigns, it was the content of those messages, particularly around anti-corruption appeals and a projection of the party as the best tool to defeat an unpopular incumbent or dominant figure of the country’s politics, that played a key role in mobilizing and persuading voters (e.g. Haughton et al. Reference Haughton, Krasovec and Cutts2024, Reference Haughton, Rybář and Deegan-Krause2022). Such examples point to the increased role of persuasion and the goal of conversion in campaigning. Jane Austen may have written in her novel Persuasion, ‘How quick come the reasons for approving what we like’ (Austen Reference Austen2011: 15), but ‘the persuasive challenge’ in campaigning is ‘the ability to say the right thing at the right time to the person you wish to persuade’ (Madsen Reference Madsen2019: 38). Such persuasion relies on ethos, pathos and logic, linked to the character and trustworthiness of the speaker, the delivery of the message and an appeal to reason and logic. All of these elements are difficult to isolate, measure and test, but should be at the heart of any research on campaigning.

Researching the clicks and mortar of campaigning

Many of the books mentioned in this review employ a variety of different methodological tools, including surveys, survey experiments, participant observation, interviews and focus groups, provoking us to think more carefully about how to research political campaigns. Advances in technology make both the campaigning and the study of it easier and harder. Social media tracking and scraping tools, coding software and mapping of geo-locations of campaign events permit researchers the opportunity to know much more about the conduct of a campaign. But the exclusive nature of some social media apps makes it much harder to map and measure campaigning and its impact. Indeed, how can we capture the role of encrypted and closed communication tools like WhatsApp or study ardent fans, bots, trolls and influencers, and in general the less formal social media ecosystem (Cheeseman et al. Reference Cheeseman2020; Sehat et al. Reference Sehat, Prabhakar and Kaminski2021; Sinpeng et al. Reference Sinpeng, Gueorguiev and Arugay2020)? Moreover, that the power of a social medium like Twitter/X lies less in active users and more in the ‘secondary exposure’ as tweets are picked up by mainstream broadcasters, underlines again the connections between older and newer forms of communication (Jones Reference Jones2021: 149). We cannot and should not divorce the two.

Although we might expect that the analysis of data-driven campaigning would need to employ heavy-duty quantitative methods, refreshingly Kefford’s (Reference Kefford2021) and Dommett and colleagues' (Reference Dommett, Kefford and Kruschinski2024) books show the leverage that can be achieved from interviews and participant observation. Such an approach affords greater granularity, reducing the risk of missing important realities. The meaningful may not always be something that can be so easily measured. Big data can often tell us what is happening, ‘but it cannot [always] tell us why’ (Tett Reference Tett2021: 224, emphasis in the original). Nonetheless, using methodological tools such as interviewing poses problems. Not only are there challenges of securing access to, and eliciting responses from, the key individuals given the sensitivity of campaigns in process, but some political operatives may brag and overemphasize their roles. Moreover, politicians keen to stress their agency and suffering from a bout of wishful thinking may emphasize that their pressing of the flesh matters, in the absence of solid empirical evidence.

Much can still be leveraged from traditional surveys and polling, but experiments are increasingly the tool of choice for many contemporary political scientists. Such tools have the great virtue of helping to isolate and test variables, but they are often carried out on small scales and, if framed in a way close to reality and conducted in real time in close proximity to an election, pose ethical questions. Moreover, given the dynamic and competitive nature of campaigns, and the shifting of voters’ views and priorities, measuring the impact of a message is tricky (Hewitt et al. Reference Hewitt2024). Experiments, therefore, are best suited for pointing towards the likely impacts of specific messages rather than what determines vote choice.

Given the way in which digital tools in particular can enhance the ground war of campaigning, in order to grasp the conduct and impact of campaigning there is a clear need for quantitative and qualitative researchers to embrace the tools of the other tradition and blend them. More broadly, though, if we seek to understand the conduct and the impact of campaigning, political scientists need to embrace not just methodological pluralism, but also a genuine interdisciplinarity, encompassing psychology, sociology and the study of language. The last of these is particularly salient. Although campaigning tools can target, convey and broadcast, the heart of any campaign remains the message. No amount of technological sophistication or organizational prowess can have an impact unless the core message has resonance. More research, therefore, is needed on the messages, but also on the interplay between messages, modes and the messenger (Schmidt Reference Schmidt2025). As discussed above, modes carry meanings about modernity, virility and connection to a place. Some messages, particularly those linked to an increasingly common theme in campaigning, anti-corruption appeals, appear easier to make and more plausible if conveyed through new technology or ones not seen to be controlled by the mainstream – that is, they lend themselves to grassroots campaigning and/or social media (Haughton et al. Reference Haughton, Rybář and Deegan-Krause2022; Kramer Reference Kramer2022).

A secret sauce? How to win votes and persuade people

While there is no secret sauce for success that works for all parties and candidates and in all systems, drawing together some of the findings of recent scholarship on campaigning are we closer to understanding the recipe for a successful campaign?

Horowitz’s research, for instance, underscores the significance of the size and relative balance of different core groups in society and hence what degree of persuasion rather than mere mobilization is needed. Moreover, thanks to Dommett, Kefford and their co-authors, in particular, we now have a better appreciation of the processes and role of targeting voters particularly using DDC. The nerds clearly can make a difference, especially in the identification of voters and in the use of social media. New technology can convey a message further and faster than older forms of campaigning, but the weakness of modern technology is that it can miss the human touch. Muñoz, Horowitz and Lynch point to how important the congregating of voters and politicians can be. The very fact of physical presence not only impacts voters, but sends messages about priorities, policies and places. The power of that assembled mass can be enhanced when sharing on digital devices and in coverage in the mainstream media, highlighting again that it is not about one or other mode. Effective campaigning is about combinations.

Nevertheless, recent scholarship suggests the need for further research. Although Muñoz does touch on the timing of rallies in her account, for example, we still need to know more about timing. Simply put, a campaign needs to aim for peak support when it matters – that is, when voters cast their ballots – but critically there are many steps to reach that point. We need a better understanding not just of what happens in campaigning, but also of when that happens. Some narratives, for instance, are established early and become hard to shift, but other messages fade quickly, or appear as bolts from the blue to change the dynamic of a campaign. Much is linked to kairos, the ancient Greek concept designating the right or opportune moment to say or do something. But crucially, timing can matter in different ways for different parties and candidates in different contexts. In some more fluid party systems it may be beneficial to arrive on the scene late in the day, benefit from a blaze of publicity and reduce the time for your opponents to dig up dirt on you. In other systems this is not an option. In the US primaries, outsider candidates like Barack Obama or Jimmy Carter have to establish themselves early on as viable and need to steal a march on their opponents by generating momentum.

Above all, we need to know more about which messages work. Although all effective messages are tailored to the specific context and their resonance with particular sets of voters, comparative studies can uncover common patterns. Returning to a theme from the introductory section of this article, research can assess, for instance, how far the seemingly perennial messages of ‘time for a change’ and ‘don’t risk it’ form the basis for successful campaigns across a range of countries in the 21st century. Elections and the choices they involve need to be framed. Narratives and images need to be crafted to show that the candidate or party is the best to address the central issues placed at the heart of the campaign. The message may be enhanced, modulated, muffled or muted by the utilization of different modes of campaigning, but modes by themselves cannot mobilize or persuade.

Campaigning in the rest of the 21st century

In 2024 nearly 2 billion voters in 50 countries had the chance to cast ballots in elections. To what extent do those elections and ones in coming years challenge the insights from the literature discussed in this article drawing on experiences from several years ago? Given the usual time lags for academic publication, but also the sensitivity of tactical and strategic insights when campaigns are live, at present we have only an impressionistic sense of what seemed to be effective in 2024 and more broadly how campaigning is evolving.

Three aspects are worthy of mention. First, in many cases the last set of national elections was the first since the emergence of COVID-19. A few years on from the outbreak of the pandemic and clearly the world has not returned to the exact status quo ante, but neither did it bring the irrevocable changes some of the doomsday pundits were claiming. In terms of campaigning, perhaps the pandemic’s greatest impact was to accelerate shifts towards a greater use of online and digital tools.

Second, initial accounts of several of 2024’s elections highlighted a changing social media landscape, with different platforms, particularly those that share videos like TikTok and YouTube, playing more significant roles (e.g. Srinivas and Kamra Reference Srinivas and Kamra2024), suggesting perhaps visual forms of messaging are becoming more important. But the past couple of years have also been striking for transformation of platforms – like Twitter into X – and the use of platforms such as Telegram by far-right and pro-Russian politicians in Europe.

The adoption by political parties and candidates of platforms like TikTok is partly the product of herd mentality where rivals quickly start using whatever their opponents seem to be using effectively (well illustrated by the rapid adoption of TikTok by UK political parties in the run-up to the 2024 election), but it is also reflective of the evolving nature of society and the media landscape and the (elusive) search for modes that can best communicate messages. That search for the most effective means of communicating messages given the limitations of time and resources is also likely to be behind decisions in some elections not to use microtargeting so much, but rather to convey messages with broader appeal to more voters (Waterson Reference Waterson2024).

Third, much of the journalistic noise in recent times has been around Artificial Intelligence (AI). It is too soon to draw definitive judgements on the speed and impact of generative AI on campaigning. Artificial Intelligence offers threats and opportunities. It provides politicians and political parties with the chance to process information at higher speed, make more informed decisions and be more responsive to voters’ concerns. Nevertheless, generative AI, in particular, poses threats, the greatest of which is to citizens’ trust as they are bombarded with text and images, the authenticity of which they cannot judge. But other dangers lie in a deepening and reinforcement of polarization. Indeed, combining AI-generated propaganda with new capacities for microtargeting ‘could revolutionize disinformation campaigns’ (Kreps and Kriner Reference Kreps and Kriner2023: 126). With strong echoes of the discussion above of the impact of DDC, early indications from some of 2024’s elections pointed much more to AI’s potential rather than an actual impact. It reminds us that even those who have written about the dangers of new technologies to democracies warn us against over-emphasizing its power: ‘one should always take stories about technological innovation winning elections with a spoonful of salt’ (Moore Reference Moore2018: 56). Nonetheless, given the rapid developments in AI, that potential could become real and be a game-changer. What we can say for certain is that if that technological change facilitates the effective conveyance of particular messages to specific groups of voters it may indeed transform the practice of campaigning.