Hybrid rallies and a rally-centric campaign: the case of Kenya’s 2022 elections

ABSTRACT Scholars tend to present either face-to-face or mediatised audiences as the principal target for election campaign rallies. However, a close eye on the staging of, and popular engagement with, campaign rallies during Kenya's 2022 elections reveals that they constituted a hybrid form of political communication that simultaneously targeted face-to-face and mediatised audiences with tailored messages. Not only were rallies at all levels characterised by such hybridity but also at the presidential level, rallies came to dominate candidates' diaries and the traditional and social media coverage of them leading to what I coin a rally-centric campaign. This paper analyses these empirical realities and the implications for how we should study and conceptualise election rallies and campaigns. It does so by focusing on the relationship between rallies, media coverage and popular engagement with a particular focus on social media.

On 9 August 2022, Kenyans went to the polls to vote for their president, governor, senator, women's representative, member of parliament (MP) and member of the county assembly (MCA).The elections were highly competitiveif not in the general elections, then in the party nomination processes and candidates at all levels worked hard to mobilise support for themselves and against their main opponents over an extended campaign period.Investments included door-to-door canvassing and mass messaging; posters and billboards; coverage in newspapers and on radio and TV; and widespread use of political consultants and opinion polling.It also included an unprecedented investment in social media as aspirants and campaign teams took advantage of the affordances of this relatively new technology.Closed social media groups (WhatsApp, Telegram and Signal) were used to help organise campaign activities, andwhen people were confident that there were no 'moles' in a groupto plan for vote protection and discuss messaging.Aspirants at all levels recruited social media communicators to post their activities (usually on Facebook and WhatsApp, and sometimes on Twitter and other platforms), to respond to attacks against them, and to decampaign or personally attack their main opponents (cf.Lynch et al., 2022).
At the same time, aspirants invested heavily in campaign rallies.The prominence of rallies was most pronounced at the presidential level as the two front-runnersthe then Deputy President William Ruto vying on a United Democratic Alliance (UDA) ticket and Raila Odinga vying on an Azimio La Umoja ticketundertook a dizzying and exhausting schedule that started well before the official campaign period (29 May to 6 August 2022) and saw them visit almost every corner of the country.Rallies dominated these tours, but they often also included meetings with elders and key stakeholders, attendance at places of worship, and/or walking and driving visits (cf.Bob-Milliar & Paller, 2023).Lower-level candidates also held rallies as part of their ground campaignsor efforts to reach and mobilise constituents through personalised interactionsand attended rallies in or near their constituency organised by co-partisans vying for bigger seats.
At first glance this investment in social media and rallies may seem unrelated and even contradictory.Rallies in sub-Saharan Africa are usually analysed as face-to-face events with target audiences considered equatable with location (Horowitz, 2016;Paget, 2019;Brierley & Kramon, 2020;Gadjanova, 2021), which raises questions about whyespecially at the tail end of the COVID-19 pandemicpoliticians would invest so heavily in rallies when they could reach most citizens through canvassing and traditional and social media.However, a closer eye reveals Kenya's campaign rallies to constitute a hybrid form of political communication whereby 'new technology and techniques are adapted and blended together with pre-existing modes of campaigning' (de la Torre & Conaghan, 2009, p. 337) with rallies stage-managed to simultaneously target face-to-face and mediatised audiences with tailored messages.More specifically, Kenya's 2022 campaign rallies, which cumulatively enjoyed large immediate audiences, were carefully staged to attract, entertain and mobilise assembled crowds through, for example, the use of locally popular entertainment and talking points.At the same time, campaign teams posted pictures and live or edited videos of rallies on social media, while local traditional media was awash with images and summaries of presidential rallies, ensuring an even larger audience were presented, often daily, with mediatised rallies that emphasised a candidate's popularity across different areas and groups.As a result, while ideas of the relevant community, candidates' strengths and nature of the political contest were communicated to rallies' assembled audiences (Paget et al., 2023), the mediatised rally tended to focus largely (albeit not solely) on the construction of a popular and exciting candidate.This rally hybridity and differential messaging is critical to understanding the role of rallies and how, with new technology, they have becomenot obsoletebut a valued media and valued media content.Moreover, while aspirants at all levels held rallies, biases in media coverage and the greater value of band wagoning at the presidential level (given the need to reach out beyond core supporters) (Gadjanova, 2021), ensured that at the presidential level campaigns also became rally centric.The term rally centric is coined to denote a campaign in which rallies not only constitute a hybrid means of political communication, but a principal means of communicating with face-toface and mediatised audiences.
This paper analyses these empirical realities and the implications for how we study and conceptualise election rallies and campaigns by focusing on the relationship between rallies, media coverage and popular engagement with a particular focus on social media.In adopting this focus, I recognise that most Kenyans still gain their political news from, and largely follow rallies on, traditional media; most notably radio and TV (Afrobarometer, 2022;TIFA, 2023).However, given that traditional media has long covered national rallies and the increasing use, and even wider reach, of social media (Gadjanova et al., 2022), I pay more attention to social media as a newer media to see how its rise has helped to shape rallies and their intended audience, and how the prominence of rallies has impacted upon social media strategies.The paper starts with a brief overview of some of the key literature on political communications to which the paper contributes before turning to contemporary Kenyan rallies and their staging in the context of increasing social media use, popular attendance at, and popular engagement with mediatised, rallies, and the place of rallies in social media strategies.
The paper draws on insights gained from primary research conducted ahead of Kenya's 2007Kenya's , 2013Kenya's and 2017Kenya's elections, and 2005 and 2010 referenda, as well as from research on social media use in the country's 2022 election campaigns.Across all four general elections I observed (as an academic, rather than official observer) the campaigns and attended rallies for different elective posts and parties, and interviewed candidates, party officials, campaigners, journalists and civil society actors.In 2022, I also interviewed social media communicators andtogether with colleaguesconducted a survey in the final weeks of the campaigns in which we posed questions about social media use to candidates, party officials and party activists via the Qualtrics phone application.This app allowed us to send anonymised survey links, together with a participant information sheet, to a targeted sample via WhatsApp.Using well-established networks to source contacts we double-checked participants' eligibility before sending them a survey link; with participants sent KSHS 500 (about GBP 4) of phone credit on completion of the survey as a token to thank them for their time.In total, 348 people across the country completed this 'political actors survey': 133 of whom were aligned to Kenya Kwanza, 184 to Azimio, 28 to other parties/independents, one did not answer, and two erroneously took the survey (as they were not actively involved in campaigns) and so their responses have been excluded.Finally, data is drawn from Afrobarometer surveys, while questions were added to a 2,065 person nationally representative and face-to-face household survey conducted by TIFA Research between 11 and 19 March 2023.

Conceptualising political campaigns: from modernisation to hybridity
Drawing largely on evidence from the United States and Europe, many scholars have characterised campaigns in evolutionary terms with changes over timesuch as the use of TV advertising, political consultants and social media -'considered as part of the modernization process rooted in technological and political developments common to many postindustrial societies' (Norris, 2000, p. 140).In this way, Pippa Norris (2000) talks of pre-modern, modern and post-modern campaigns; Gibson and Rommele (2001) talk of a first, second and new era; and Blumler and Kavanagh (1999) talk of a first, second, and third age.These scholars present the first campaign period as heavily reliant on face-to-face interactions such as canvassing and smallscale meetings organised by local party structures.The second period is said to have followed the introduction of limited television channels and declining levels of partisanship and to have seen more centralised parties make use of innovations in mass media and canvassing of public opinion.The third period is associated with a multiplication of television channels and media outlets, the introduction of mass messaging and the internet, and increasingly professional campaigns in which voters are targeted through new media in increasingly competitive environments (for example, see Blumler & Kavanagh, 1999;Farrell & Webb, 2000;Norris, 2000;Gibson & Rommele, 2001).More recently, there has been talk of a fourth 'data driven' era in which 'Digital technology and data are now hardwired into the campaign organization and operation' (Rommele & Gibson, 2020, p. 597).
Modernisation theorists recognise that older forms of political communication, such as canvassing and rallies, 'have essentially been supplemented, but not replaced' (Norris, 2000, p. 142) by technological developments.Indeed, Pippa Norris recognises how 'some of the more localized and interactive forms of communication'such as rallies and canvassingmade a 'return' during 'postmodern campaigns' (2000, p. 149).Nevertheless, these scholars present a progressive story in which newer modes of communication are cast as dominant, or, at least in which older methods are cast as largely supportive.In this way, Rachel Gibson and Andrea Rommele talk of how, 'The second wave of campaigning … saw a shift from communication via the party organization to mass media communication between parties and voters' such that 'Citizens do not receive their information directly from party meetings or rallies, but through the mass media ' (2001, p. 33).Similarly, Jay Blumler and Dennis Kavanagh posit how, with each 'era', newer media 'became the dominant medium of political communication ' (1999, p. 212).In a more nuanced analysis, Pippa Norris presents older methods of political communication as 'ancillary' to newer ones (2000, p. 142).In this vein, Norris recognises how politicians continued to hold rallies in the 'modern campaign' era but argues that these 'became increasingly focused on achieving favourable coverage on the mainstream evening news, current-affairs programs, and leadership debates on television', rather than on communicating with face-to-face audiences (2000, p. 145).The implication is that, with each new campaign period, older modes of political communication become defunct, relatively unimportant or are overtaken, or at least shaped, by newer media logics such that rallies in second or third era campaigns become primarily 'a device to win media coverage' (Paget, 2019, p. 445).
In contrast, it is often assumed that the audience for rallies in sub-Saharan Africawhere aspirants tend to hold rallies across their constituenciesis synonymous with the one physically assembled.In this vein, Jeremy Horowitz argued that 'Because ethnic groups are geographically concentrated in Kenya, data on the location of campaign rallies provide a useful indicator of which groups the candidates targeted ' (2016, p. 335).Similarly, Brierley and Kramon (2020) draw conclusions about the role of rallies in party campaign strategies by mapping whether parties hold most of their rallies in core or swing constituencies, and Elena Gadjanova views rallies in 'given areas' as 'attempts to make appeals to voters resident in those areas ' (2021, p. 664).In the early post-colonial years, this association between rally location and audience was explained by the limited reach of traditional media and importance of localised campaigns (Miles, 1989); today it is explained largely by the latter.In line with this literature, Dan Paget emphasises the 'directness of party-voter communication' in Tanzania's rally intensive campaigns (2019, p. 458); with a rally intensive campaign defined as neither media intensive nor hybrid, but as one 'in which the ground campaign is rally-intensive and the campaign as a whole is ground campaign-intensive' (2019, p. 454; emphasis added).At the same time, Paget extended analysis of the role of rallies from generic efforts at political mobilisation to local party-building by touring leaders, local branches and 'lone organisers' (2022), and has increasingly come to recognise the role of mediatised audiences (Paget et al., 2023) even if these are still presented as secondary (Paget, 2023).
However, other bodies of literature provide ideas of how to think more explicitly of both rather than either/or face-to-face and mediatised audiences.The evolutionary progression suggested by modernisation theorists has been questioned by other scholars of election campaigns in Western democracies who have shown how new technology is often 'layered' onto, and can even reinvigorate, older practices.In this vein, Rasmus Kleis Nielsen details how innovations such as accessible online databases have been used to develop lists 'of potentially 'valuable' targets' for door-to-door canvassing and phone banking in the United States (2012, p. 28).A development that has ledin contrast to those who 'have long predicted a decline in personalized, live interaction as a tool for campaigns' -'a recent increase in the number of people who are contacted in person by political parties' (Nielsen, 2012, p. 16).Nielsen provides an excellent example of how 'Different practices of political communication seem to coexist and to be mixed and matched by campaigns on the basis of their own perceived interests, the ideas and know-how they have, the resources at their disposal, and the communications environment around them' (2012, p. 17).The implication of such a 'layered approach' is that, while campaign 'priorities and defining technical tools do change over time, such a process is not linear and does not affect all communication channels in the same way' (Vaccari, 2013, p. 11).
Like the layered approach, hybrid theories of political communication reject 'simple dichotomies' and ''either/or' patterns of thought' and encourage ''not only, but also' patterns of thought' (Chadwick, 2013, p. 4).However, they do so by building on the idea of a hybrid media system in which 'older and newer media logics in the fields of media and politics blend, overlap, intermesh, and coevolve' (Chadwick, 2013, p. 5).In so doing, scholars of hybridity highlight the ways in which older communication methods are shaped by newer ones, and how newer methods are shaped by older ones as 'each country's political culture and past shape the ways in which' new logics are adapted (de la Torre & Conaghan, 2009, p. 348).In this vein, Carlos de la Torre and Catherine Conaghan show how the personalisation of campaigns in Ecuadora development associated by modernisation theorists with second or third era campaignswas 'itself an example of hybridization' as 'two maverick candidates mounted … media-centric strategies' and used 'rhetoric that harked back to the classic populist leaders of the past ' (2009, p. 338).It is this interaction, rather than mere layering, of media in historical, cultural and political contexts that leads to hybridisation 'or a merger of traditional country-and culture-specific campaign practices with select transnational features of modern campaigning' (Plasser & Plasser, 2002, p. 350).
The following sections unpack how the ongoing prominence of rallies in Kenya's 2022 electoral campaigns despite the uptake of newer methods of political communication can best be understood as a result of such hybridity in the sense of a blending of old face-to-face rallies that seek to mobilise a local audience with rallies as traditional and social media content that seek to mobilise a broader publicand how, at least at the presidential level, this resulted in a rally-centric campaign.The term rally-centric is used instead of Paget's rally-intensive campaign to explicitly recognise such hybridity and to emphasise how rallies were not only central to ground campaigns but also to mediatised campaigns.

Kenya's 2022 election campaigns
Two alliances dominated Kenya's 2022 elections -Ruto's Kenya Kwanza (Kenya First) and Odinga's Azimio La Umoja (Declaration of Unity)with most political parties, and even most independent candidates, throwing their weight behind one or other of the presidential candidates.As Deputy President, Ruto benefited from being a state insider, but he was not the establishment candidate.Instead, President Kenyatta backed Odinga, a long-time opposition leader who Kenyatta had entered an informal alliance with back in 2018.In terms of campaign messaging, Ruto drew upon his humble beginnings to present himself as a 'hustler' who understood ordinary Kenyans and could help elevate them through the provision of small grants and economic opportunities, and by standing up to those who controlled Kenya's wealth.In so doing, Ruto cast Kenyatta and Odingaas sons of the first President and Vice-President respectivelyas 'dynasties' who needed to be ousted if ordinary Kenyans were to progress.In contrast, Odinga drew upon his long track-record of struggle for multi-party politics and constitutional reform, and Ruto's association with corruption, electionrelated violence, and authoritarian practices, to cast himself as the change that Kenya needed, and to present Ruto as a dictator in waiting.In terms of policies, Odinga also narrowed in on the economy, but focused on assistance through social welfare programmes to be funded by a war on corruption (Cheeseman et al., 2023).Both candidates and alliances sought votes across the country with a particularly intense battle amongst Kenyatta's Kikuyu co-ethnics in Central Kenya from where Ruto and Odinga's running mates were drawn (Opalo, 2022).Ultimately, Ruto secured a narrow first-round victory with 50.49 per cent against Odinga who came a close second with 48.85 per cent.Azimio disputed the results, but their failure to provide evidence of rigging in the context of unprecedented transparencywith 97 per cent of polling station level results available online for public scrutiny within 24 h of the polls closing (IEBC 10 August, 2022)led the Supreme Court to reject their petition (Cheeseman et al., 2023).
Over the extended campaign period, aspirants and their teams at all levels used a wide range of methods to mobilise support and decampaign opponents.Roads and market centres were bedecked by posters, banners and billboards.Presidential candidates paid for radio, TV and newspaper adverts, and prominent supporters joined panel shows and wrote regular opinion pieces.Better-resourced lower-level aspirants facilitated journalists to accompany them on the campaign trail or paid for radio slots.At the same time, social media teams advertised aspirants' activities and messages, responded to attacks, and launched their own decampaigns.Investment in social media usually involved a relatively small salaried communications team, and sometimes a separate dedicated 'attack team', who collaborated with a larger group of 'online warriors', 'bloggers' or 'influencers' who worked in exchange for a small stipend and/or hope of a position (or other forms of assistance) if the candidate was elected (Gadjanova et al., 2019).
Campaign teams also sought to canvass voters through mass messaging and 'door-to-door' campaigning, which often involved meetings with small groups as well as household visits.Finally, candidates hired vehicles bedecked with loudspeakers to move around their constituencies as they toured the same.These tours involved a range of methods including attendance at places of worship when aspirants might not necessarily speak to congregations, but be blessed; small 'fireside' or 'townhall' meetings when aspirants would engage in a conversation with voters; walks or 'caravans' through local neighbourhoods when aspirants would wave and occasionally stop to have conversations or hold mini (and sometimes large) rallies; and rallies in public places where speakers would 'address an audience face-toface for the ostensible purpose of politically mobilizing it' (Paget, 2019, p. 451).
Our 'political actor survey' focused on social media use, but we also asked respondents: 'besides work on social media what else do you do as part of the campaigns'?The responses highlight the importance of ground campaigns with 207 saying that they went door-to-door, 199 that they distributed leaflets or posters, 165 that they organised rallies, 86 that they helped fundraise, and 57 that they distributed gifts or electoral incentives, while 45 said that they appeared on radio and 35 on TV.A lot of time was also invested in social media.For example, when we asked, 'as someone actively involved in the campaigns, which of the following statements best describes your use of social media?',27 said that it was their main job, 216 said that it was an important part of their work, 68 said that they largely delegated it to others, and only 29 said that they rarely or never used it; 6 gave no answer at all.When we asked respondents how important they thought social media was for winning the 2022 presidential election, 80.1 per cent said it was very or extremely important, and only 16.5 per cent and 1.5 per cent said it was somewhat or not at all important, respectively; with 2.6 per cent failing to answer.
The importance attached to social media reflects its reach and significance.An increasing number of Kenyans regularly access social media usually on their phones.According to the TIFA survey (2023), only 24.3 per cent of Kenyans never use 'social media such as Facebook or WhatsApp', while 45.9 per cent said that they use it multiple times a day, 14 per cent at least once a day, 7.1 per cent at least once a week, 2.2 per cent at least once a month, and 2.1 per cent less than once a month.Journalists also source many of their stories from social media (Interview, journalist, Kisumu, 1 March 2022), while online debatesfrom revelations of 'fake news' to arguments between elitesare regularly taken up as stories by the mainstream media (for example, Nation 7 July, 2022).Just as importantly, direct social media users share social media stories through 'pavement media' creating regular or occasional indirect social media users.Informed by Stephen Ellis's discussion of 'pavement radio' or the 'popular and unofficial discussion of current affairs' in marketplaces, places of worship, bars, and the like (1989, p. 321), 'pavement media' includes 'various non-conversational forms of everyday communication such as songs, sermons, graffiti, flyers, and street theatre' (Gadjanova et al., 2022, pp. 163-164).As a party activist explained with regards to WhatsApp: if you send a message, it reaches people in various WhatsApp groups, but it also spreads offline -'if I get the message, my mum and dad might not be on social media, but I spread it to them' (Interview, Nakuru, 7 November 2021).This ability for information to move, often remarkably quickly, between traditional, social and pavement media has created a complex media ecosystem in which the influence of social media is far greater than user rates suggest (Gadjanova et al., 2022).This ecosystem helps to explain social media's appeal as campaigners sought to take advantage of its affordances to reach direct and indirect users alike in the context of highly competitive elections.The key affordances were an ability to speak directly to audiences in ways that had a broad geographic reach (but could also be targeted), were relatively cheap, and which enabled official figures to delegate, and thus distance themselves, from negative campaigning (Lynch et al., 2022;Abboud et al., 2023).
Rallies were thus one of a menu of mobilisation tactics, but they were nevertheless a central one particularly at the presidential level.Not only did Ruto and Odinga's campaign teams dedicate a lot of manpower to organising ralliesfrom working out a schedule and finding and fitting out venues to mobilising and often facilitating a crowd, sourcing 'talking points', and coordinating speakers -Ruto and Odinga also dedicated significant resources and much of their own time to rallies.Even back in 2018, Ruto's predilection for holding rallies led him to be mocked by critics as Team Tangatanga (or Team roam around and around).However, while his opponents predicted that he would wear himself out (Interview, ODM activist, Nairobi, 15 November 2021), his schedule continued to pick up pace as the election neared.Ruto held many classic rallies where he spoke from a stage inside a public ground, but more frequently he opted to play with the rally genre (Paget et al., 2023) and to hold shorter rallies where he talked through the sunroof of an SUV to a crowd gathered in a local town or trading centre.This approachtogether with Ruto's remarkable energy, significant 'war chest', and help of lowerlevel Kenya Kwanza candidateshelped him to hold rallies in towns and centres across the country and it was not uncommon (when in more rural areas during the tail end of the campaigns) for interviewees to explain how, while Odinga had recently spoken in a nearby town, Ruto had 'even come here' (Interview, Rongai, 6 August 2022).Odinga in contrast tended to hold more conventional mega rallies where large crowds were gathered and entertained at a public ground before a slate of party dignitaries arrived amidst much pomp and ceremony and gave a series of speeches culminating with Odinga himself.
While rallies dominated presidential candidate timetables, they also featured at lower levelswith those who had more money and larger constituencies, and/or who faced a more serious challenge tending to hold more rallies than those who had fewer resources, smaller constituencies and/or were more confident of victory.For example, when in the ODM stronghold of Kisumu County during the last week of the campaign I was struck by the number of (often small) rallies held by the ODM gubernatorial candidate, Anyang Nyong'o, in the face of a relatively strong independent candidate.This contrasted with previous elections when Nyong'o had been visible at national rallies alongside Odinga, but largely invisible at the local level following his securement of an ODM party ticket.Even at the ward level, where it was more feasible for campaign teams to go door-to-door or to hold meetings with small groups, aspirants usually held at least a couple of rallies.As one incumbent MCA in a geographically small ward in Eldoret town explained a week before the polls, he had held one big rally early in the campaigns and was considering holding another even though the event would cost him at least Ksh200,000 (around USD 1,600), as his constituents expected it of him (Interview, Eldoret, 2 August 2022).
Not only did rallies absorb much of the presidential candidates' time and resources; it also absorbed traditional media coverage of them with occasional references to rallies held by lower-level candidates.For example, the front page of newspapers often depicted presidential rally crowds, while articles summarised some of the key statements with a tendency to focus on crowd size and anything controversialfrom a slur against an opponent to the absence or defection of a local leader or heckles from the crowd (for example, Nation 17 June, 2022; The Standard, 27 March, 2022).Similarly, the evening news provided daily coverage of presidential rallies and excerpts from leaders' speeches with attention sometimes also given to smaller meetings attended; with relatively little attention given to manifestos, past performance, and lower-level races.
This obsession with rallies extended to social media.For example, of the 60 posts added to Ruto's official Facebook page in the last week of the presidential campaigns, 49 focused on UDA rallies.In contrast, while only 34 of the 106 posts on Odinga's Facebook page during the same time period focused directly on Azimio rallies, many of Odinga's other posts (which largely consisted of campaign posters and videos) featured images, short clips and/or audio of, or adverts for, rallies.For example, in the last few days of the campaigns, Odinga's feed regularly featured short videos calling upon people to attend Azimio's final rally in Nairobi.Rally images were then shared by social media 'armies' who posted them on public platforms, shared them through closed groups, and/or commented on and liked them.
Social media was even more important for advertising the admittedly less intense rally activities of lower-level candidates, which rarely made the mainstream news.To facilitate such coverage, aspirants at all levels ensured that a team of 'bloggers' followed them on the campaign trail to capture, and then share, images and sometimes short videos.This work was then reinforced by other 'bloggers' and supporters who posted their own images of campaign rallies and/or shared or boosted those taken by others, as a way of supporting their preferred candidate and/or of coming to the attention of campaign teams in the hope of gaining employment.
Not only did traditional and social media provide extensive coverage of rallies, but the public engaged with rallies as both face-to-face and mediatised audiences.Indeed, 50.8 per cent of Kenyans attended a campaign rally ahead of the 2022 elections (TIFA, 2023) up, despite COVID, from 47 per cent ahead of the 2017 elections (Afrobarometer, 2019) putting Kenya's 2022 election campaigns amongst the most rally intense in the world (Paget, 2019).Attendance reflected a mix of the importance attached to positions (Cheeseman et al., 2019, pp. 224-225) and intensity of campaigning with 27.6 per cent of respondents saying that they had attended a rally organised for a presidential candidate, 22.6 per cent for a governor, 26.2 per cent for an MP and 24.1 per cent for an MCA, as compared to only 8.7 per cent who attended a rally for a senator, and 7.8 per cent for a women's representative (TIFA, 2023).As in many other sub-Saharan African countries, people attended rallies for a multitude of reasonssome wanted to show their support or hear what candidates had to say, some were attracted by the entertainment or opportunity to socialise, and some were paid or promised rewards (Kramon, 2017).
However, while it is impressive that almost a quarter of Kenyans came faceto-face with a presidential candidate at their rallies, it is even more striking how many, and how regularly many, watched, listed or read about campaign rallies through traditional mediawhich, from observations, focused primarily on rallies by presidential candidatesand through social media (Figure 1).
The fact that the numbers of people regularly following rallies through these media was about the same and sometimes slightly higher than those who said that the same media constituted one of their main sources of political news also provides further evidence for how dominated these media were by images, clips and stories of rallies during the height of the campaigns.For example, while 66 per cent of respondents said that they watched rallies on TV at least once a day during the final month of the campaigns, 64.5 per cent listed the TV as one of their three main sources of news; while 56.1 per cent listened to rallies on the radio at least once a day, 57.1 per cent listed the radio as one of their three main sources of news; while 38.6 per cent followed rallies on Facebook at least once a day, 30.8 per cent listed Facebook as one of their three main sources of news; and while 45 per cent said that they followed rallies on WhatsApp at least once a day, only 24.3 per cent listed WhatsApp as one of their three main sources of news (Figures 1 and 2).
These incredibly high levels of reported regular popular engagement with rallies through various media are supported by observations.It was common during the campaigns to see people in bars and restaurants turn their attention to rallies as captured on the evening news; to see traders and taxi drivers listen to rallies on the radio; and to see people huddled around newspaper stands looking at the front-page images of rallies.It was also common to see people watching rallies on their phones and for them to share videos or images of rallies, often with quotes, through various platforms.Interestingly, this engagement was often social andtogether with the number of people for whom 'pavement media' is one of their three main sources of media (including friends and family (9.9 per cent), chiefs (0.7 per cent), religious leaders (1.2 per cent) and street parliaments (0.8 per cent))helped to ensure that social media coverage of rallies extended from direct to indirect social media users.For example, whilst having lunch with a member of a gubernatorial candidate's social media team, I noticed a group of young men at the next table huddled around one of their smart phones.It soon became clearfrom the animated discussion of the size and energy of competing crowdsthat they were watching footage of rival presidential campaign rallies in Central Kenya most likely through Facebook Live (Notes, Kisumu, 31 July 2022).This was neither the first nor the last time that I noticed peoplefrom taxi drivers to people in restaurants and barswatching rallies on a phone and discussing content with those around them.It was also a phenomenon noted by others.As one interviewee explained, social media has 'penetrated every village … [we] have at least one person with a smart phone and [that] is the one who shows others where Baba [Raila Odinga] is today or where the DP [William Ruto] is' (Interview, civil society actor, Nakuru, 16 February 2022).
As a result of such dynamics, rallies, which are themselves a form of direct communication, became part of Kenya's complex media ecosystem as rally organisers and speakers spoke directly to face-to-face audiences, and as other audiences watched, listened and/or heard about rallies through a mix of traditional, social and/or pavement media.The following section unpacks the relationship between rally staging and audiences in more detail to show how, while rallies are local events that target face-to-face audiences, they are simultaneously mediatised events; with newer media also shaped by rallies.

Playing to multiple crowds
It has long been recognised that the rally audience in sub-Saharan Africa is not only that which is gathered, but that which follows through mass media (Jourde, 2005), while several papers in this special issue draw attention to the importance of other audiencesfrom the mediatised public (Paget et al., 2023) to political opponents (Tendi, 2013;Wilkins & Vokes, 2023) and principals gathered on the stage (Waddilove, 2023).However, while the reality of multiple audiences is recognised, the implications for how we understand campaigns and their communication methods have largely been ignored and remain under-theoriseda fact that helps to explain why political scientists of Western democracies tend to focus on rallies as media events, while those of sub-Saharan Africa tend to equate rally location with audience.
Campaign rallies during Kenya's 2022 elections were clearly targeted at immediate audiences.Significant effort went into mobilising face-to-face rally audiences and into organising rallies across an aspirant's constituency.Organisers also tried to ensure that rallies were engaging for, and appealing to, local audiences.Rallies were staged in towns and marketplaces to facilitate attendance.Music and other forms of entertainment, which rarely formed a focus of media coverage, were organised to draw people to the grounds and enliven the crowd.Politicians staged their arrival to further foster a sense of excitement.Local politicians (who were unlikely to get national airtime) started off the speeches.National politicians used their campaign teams to source 'talking points' on local issues and priorities that would speak directly to those assembled to persuade them that they were well placed to protect and promote their interests.Rallies were often also used as opportunities to distribute patronage and display generosity to those gatheredfrom the issuance of direct handouts to promises of future development projects.They were also used to display popularity and electoral viability to those assembled.
The importance attached to the engagement of immediate audiences was perhaps clearest when attending rallies deemed to be poorly organised.For example, during a rally organised for a gubernatorial candidate in Seme in Kisumu County attended by a few hundred people, I spoke with a visibly agitated local party leader.The man's complaints were multiple: the rally was organised by people from Kisumu who did not know the place and had failed to send money to local party officials to mobilise attendance; it was held on the outskirts of the market centre, so people had to get a motorbike taxi there; no proper entertainment had been provided; people had been kept waiting; and the MC and speeches were lacklustre.The event, in short, was a failure and lacked the audience and energy that one expected of a gubernatorial level rally and was likely to do little to persuade people in the area to vote for a candidate who already enjoyed a reputation for being 'stingy' (Interview, Seme, 29 July 2022).
The localism of Kenya's campaign rallies follows a long tradition of political meetings and representation where politicians are expected to be visible, accessible and to assist constituents in the form of handouts, development projects and/or defence of local interests (Barkan, 1976;Haugerud, 1995;Kramon, 2017).However, as Cheeseman et al. (2020) have detailed, Kenyan politicians are not only expected to be good patrons in this broad sense, but to also be civic-minded leaders who support the rule of law and contribute to national goods such as development, security and stability.The layered performance intrinsic to ralliesincluding the ability to switch between languages, delegate more exclusive messaging to less prominent politicians, and project different identities, for example, through the dancers hired, notables gathered, and attire selected (Beardsworth, 2020;Paget et al., 2023) ensure that they are excellent means for relaying complex and often contradictory messages to assembled crowds.
However, immediate audiences were not the only ones targeted by Kenya's 2022 campaign rallies, which ensured that complex and multilayered rally performances could be edited and tailored for different audiences.Politicians with the means often facilitated journalists to attend to ensure traditional media coverage, and TV, newspaper and national radio news were also obsessed with presidential rallies, while local radio stations often covered lower-level rallies.Candidates at all levels also brought social media teams to their rallies and ensured that they had a good vantage pointfrom atop of an SUV or truck, or via a dronefrom which to capture the crowds gathered.Photographers and videographers used wide-angle lensesand often some photoshoppingto exaggerate crowd size and enthusiasm levels.The need to not only record the events but prepare for them (from mobilising a crowd to sourcing talking points) and then post and share images and footage of them, meant that candidates' communication teams sometimes spent most of their time covering rallies (Interview, Kisumu, 31 July 2022).Additional resources often then spent to 'boost' a particular post on Facebook within the geographic boundaries of the relevant constituency (Interview, party activist, Njoro, 8 May 2022).
Rally staging and scripting also took mediatised audiences into account.A successful rally not only drew a big crowd but gathered one in such a way that it photographed well.Ruto proved particularly adept at playing with the rally genre to achieve this.He gathered people in town centres where roads could easily be filled and even 'small crowds look large' (cf.Gunning & Baron, 2014, p. 214 on Egypt), spoke to them from an SUV, and posted photos taken by drones, ensuring that he looked thronged by, and comfortable with, people across the country.As one friend in Nakuru explained, when Ruto had visited the town earlier that day he had spoken to people on Kenyatta Avenue, which 'as you know, is a fairly small street', but the Facebook posts made it look like a mega rally on a 'grand avenue' (Notes, 5 August 2022).
Politicians also used mediatised versions of the events to send adapted messages to online audiences.For example, while presidential candidates toured the country speaking to people in stronghold, swing and opposition areasand adapted their performances through, for example, local talking points and appearance of local candidates and dignitariesonline followers were often already loyal supporters.Thus, while rallies as a face-to-face media focused on trying to persuade a local crowd to vote for a candidate or party through tailored performances, social media coverage tended to focus on popularity and thus electoral viabilitythrough things like crowd size and reception warmthand the threat posed by principal opponents to both sway undecideds and encourage supporters to turn out to vote.This more simplistic messaging, which was developed by coverage of individual rallies and accumulatively by coverage of rallies across the country and over time, is evident, for example, from a series of Facebook posts uploaded onto Ruto's official page that accompanied photos of rallies held on the penultimate day of his campaigns.
What a day!We thank all our supporters.We draw this energy from your passion for a new Kenya that believes in uplifting ordinary people, Engineer, Nyandarua County.
The love for Hustler Nation is astonishing!Thank you the people of Ol Kalau, Nyandarua County.
We ask our supporters to turnout in large numbers to vote.It is your right.Ignore coercion and intimidation from state agents.Make your future count with Kenya Kwanza.
We don't need billboards of fake opinion polls.We are with the people and we are winning.Nakuru town.(5 August 2022, https://www.facebook.com/williamsamoei) The efforts made to advertise Odinga's final rally or #FirimbiFest (#WhistleFest) through social media are similarly telling with emphasisfrom calls to attend to encouragements for people to make personalised posters of themselves standing next to a smiling Odinga and online coverageplaced on a display of popularity rather than substantive programmes.
However, the social mediatised rally could also be used to send more specific messages.For example, a member of a parliamentary candidate's campaign team in Nakuru County explained how they made sure to share and locally boost footage and photos of their candidate at Ruto's national rallies via Facebook and WhatsApp groups.The idea was that the candidate's reputation for being 'close' to the presidential front-runner would help to mobilise support in the local parliamentary race given understandings of how the state works and the benefits of being close to power.The activist was quick to explain how they balanced this message to minimise criticism that the candidate was too focused on national-level politics and largely absent from his constituency.To this end, the team carried a collection of shirts with them when their candidate toured his home constituency so that he could change clothes in-between public meetings.The idea: to use social media posts of the candidate in different places in different clothing to suggest that he had been travelling around the constituency over the course of several days, rather than a single afternoon (Interview, Nakuru County, 7 August 2022).
This shirt-changing 'trick' constitutes what Wardle (2017) refers to as 'misleading content'; a form of 'fake news' or dis-information.While a relatively minor example, it is notable that much of the online dis-information that I was told about or saw during the 2022 election campaigns focused on rallies.The most dangerous and irresponsible example of such rally-related 'fake news' was a doctored video of Ruto, which showed him inciting interethnic hatred in Uasin Gishu that was shared by leading figures in Azimio (Citizen Digital 2 August, 2022) even though, in the original, 'He was reassuring the other communities that they are safe living in the region and should go about their business' (Orero, 2023).Much more common however, was 'false context' or 'manipulated' or 'fabricated content' (Wardle, 2017) that sought to exaggerate a candidate's popularity or question an opponent's support.To give just a few examples.On 16 January 2022, Ruto's digital strategist, Dennis Itumbi, tweeted a picture of a massive crowd gathered at a UDA rally in Embakasi as evidence, according to the caption, of how the 'Hustler Nation Strong!!'; the image subsequently revealed to be from Zimbabwe (Business Daily 1 February, 2022).Similarly, a 13 July 2022 Twitter image of a huge crowd at a Kenya Kwanza rally in Meru was subsequently revealed to be of an earlier rally held in a different part of the country (Pesa Check 20 July, 2022).
Indeed, a range of tricks were used to misrepresent rallies.For example, campaign teams played with audio to increase the sound of the applause at their own rallies or to add heckles or alter audience responses at an opponent's rally (Interview, Nairobi, 10 November 2021; AFP Fact Check 13 July, 2022).Campaign teams photoshopped images to suggest that an opposing candidate and their supporters had been chased from a venue (Interview, social media communicator, Kisumu, 9 November 2022), that a presidential aspirant had snubbed a gubernatorial candidate (AFP Fact Check 27 July, 2022), or even that a dangerous snake or drone-grabbing eagle had disrupted Odinga's final rally (AFP Fact Check 8 August, 2022; AFP Fact Check 9 August, 2022).However, not all rally misinformation relied on technological tricks.For example, one social media communicator in Kisumu explained how they were organising for youth to attend an upcoming Ruto rally with banners for their main parliamentary opponent and to then take and post photos of the same to give the impression that the candidate had decamped to the Ruto side and thus kill the man's electoral prospects within this Odinga stronghold (Interview, 9 November 2022).These extensive efforts to misinform people about rallies are interesting in themselves, but they also highlight politicians' awareness of multiple audiences, the centrality of rallies to the campaigns, and provide an excellent example of how rally logicsnamely, of displaying local engagement through faceto-face performances and electoral strength through mediatised coverage of enthusiastic crowdshave come to shape the use of newer media in Kenya's hybrid campaigns.
However, the social mediatisation of rallies did not only foster 'fake news'.Indeed, the fact that everyone knew that rallies were often covered by traditional media and were almost always recorded and uploaded by social media teams and citizen 'prod-users' (Bruns, 2008), and that they were being watched by both supporters and opponents alike, seems, in many ways, to have made politicians and campaign teams more guarded in their speech and actions.The fear was two-fold.First, politicians and campaign teams knew that it was possible that their acts might be passed on to the National Cohesion and Integration Commission or police, and that they might face legal proceedings (Interview, social media communicator, Kisumu, 9 November 2021).Second, and more importantly (given the fact that such legal proceedings rarely progressed very far) was the possibility that their acts might be picked up by their opponents and used to decampaign them as irresponsible and violent leaders.Certainly, questionable statements by prominent politicians such as the use of the term madoadoa (or 'spots')historically associated with the incitement of ethnic clashes in the Rift Valley in the early 1990sby Senator Linturi at a rally in Eldoret in January 2022 (The Standard 10 January, 2022) were jumped upon by opponents and seem to have backfired in the context of cross-ethnic campaigns (cf.Lynch et al., 2022).

Conclusion: hybrid rallies and a rally-centric campaign
Kenya's 2022 campaign rallies had multiple audiencesthis included those assembled, but also those who followed the rallies through traditional, social and pavement mediawith more people engaging, and more regularly, with mediatised than physical rallies.Messages were also tailored for these audiences.The face-to-face media of a rally was used, first and foremost, to try and persuade those present that a candidate(s) or party were the best representatives available by constructing and communicating ideas of the community to be represented, the candidate, and the contest faced (Paget et al., 2023).Traditional and social media coverage of rallies in turn used, first and foremost, to mobilise support and encourage turnout through displays of a candidate's popularity and electoral viability.This distinction should not be exaggerated however: rallies were also used to display popularity to those present, while mediatised rallies were also used to relay key messages across ideas of community, candidature and contest from the diversity of a political constituency and candidate's national, rather than tribal, outlook to the critical and problematic nature of the contest faced.
These findings have important implications for how we research rallies and for how we analyse them.In terms of methods, multiple audiences means that to fully understand rallies we need to look at both their face-to-face and mediatised versions.In terms of analysis, it means that we cannot simply equate rally location with a target audience.On the contrary, presidential rallies in swing and opposition areas were, for example, staged in ways that also helped to mobilise loyal supporters in other parts of the country to get out and vote through choreographed traditional and social media coverage.
This reality of multiple audiences is not only important for how we study rallies; it is also important for how we conceptualise campaigns.The prominence of rallies and their adaptation to local audiences, and time dedicated to ground campaigns by campaign teams, can encourage a sense that Kenya's campaigns are simply ground intensive.However, by looking at the ways in which rallies are hybrid and staged for both assembled and media audiences, and at the ways in which media coverage of campaigns is shaped by a long tradition of political rallies, we can develop a better sense of the role and importance of different media and forms of political communication.As a result, we see how the affordances of newer media helped to shape rally performances, and an embedded political culture of local rallies helped to shape how the affordances of newer media, such as social media, were used, and indeed abused, by politicians.In turn, it becomes unsurprising that rallies have not fallen by the wayside with the rise of traditional and social media but have instead become a valued form of media content.
This closer look at rallies and their mediatisation and audiences has also shown how, at least at the presidential level, Kenya's campaigns were rally centric in the sense that rallies were not only characterised by media hybridity but were a principal means of communicating with face-to-face and mediatised audiences.This contrasts with the idea of a rally-intensive campaign, which also views rallies as a critical form of political communication, but which focuses solely or predominately on rally communications with immediate audiences (Paget, 2019(Paget, , 2023)).Finally, while the research and paper have focused on Kenya, it seems likely that findings have broader applicability.This includes other countries in sub-Saharan Africa where rallies are often analysed as purely or largely face-to-face events, but where rallies also enjoy much traditional and social media coverage.It also includes other parts of the world where rallies are often assumed to primarily constitute efforts to attract media coverage, but where rallies also assemble and communicate directly with immediate audiences.

Figure 1 .
Figure1.Thinking about the last month of the 2022 election campaigns how often did you watch or listen to, or read about, campaign rallies on the following media?Source: TIFA, 2023.

Figure 2 .
Figure 2. What are your three main sources of information about political affairs?Source: TIFA, 2023.