1Social media offers significant benefits for political parties and candidates in meeting the aims of election campaigning  (Caers et al., 2013). One key benefit for parties is to “bypass mainstream media gatekeepers” (Bimber, Davis, 2003; Brundidge, Rice, 2009; Lin, 2016), and directly engage their supporters (Steiner, 2009). Supporters in turn can be mobilised to promote parties and their platforms online and offline (Branch, 2009), thereby reaching demographics beyond party supporter communities, particularly younger and less political interested groups (Shippert, 2009; Williams, Gulati, 2008). Increasingly, parties focus on fostering grassroots participation for message diffusion (Enli, Skogerbø, 2013), relying on “likes”, “shares” and “comments” to accelerate the reach of their messages (Karlsen, 2015; Hersh, 2015; Karpf, 2016; Nielsen, 2012). They thus encourage their supporters to act as transmitters (De Bruyn, Lilien, 2008). “Likes” are the most common form of online political engagement (Bonsón et al., 2014) awarding visibility, greater awareness of the party and its policies. “Likes” support gaining donations and changing attitudes even for users becoming accidentally exposed to party messages (Hanson et al., 2010). Hence parties seek the benefits of going viral (Klinger, Svensson, 2015).
2Barack Obama’s campaign in the United States in 2008 demonstrated the value of a social media for mobilisation. Supporters met within online social spaces and self-mobilised; the identity of core activists blurred with broader members of the network. Campaign managers globally have subsequently attempted to adapt Obama’s strategy and tactics in order to mobilise their supporters, obtain donations and extend their reach within online networks (Karlsen, 2015; Lilleker et al., 2015). Research demonstrates some significant impact: Dutch parties successfully employed Twitter to attain broader publics (Vergeer et al., 2013); the 15-M movement in Spain organised mass protests and mobilised voters (Sianpetro, Ordaz, 2015); progressive social movements have also leveraged social media to gain activists (Fowler, Hagar, 2013). Similarly, Donald Trump’s campaign in 2016 used social media to control the news agenda, simultaneously securing support through targeted communication and dominating coverage among the mainstream media commentators (Baugham, Cali, 2017). While outliers in the broad political scene, these notable cases demonstrate the long recognised potential that social media may have for political organisations (Foot et al., 2007).
3This paper aims at building a broader understanding of the extent to which political parties benefit from the potential offered by the most popular social media platform, Facebook. We ask whether, during the 2014 European Parliamentary (EP) electoral campaign, parties attracted online communities, and if so how active these communities were. During the 2014 EP campaign, parties used social media predominantly for self-promotion hoping to benefit from platform affordances (Ceron, Curini, 2018). As is the norm during election campaigns (Ross et al., 2015), party strategy tends to focus almost entirely on “informing, targeting advertising, recruiting, engaging, and fundraising” (Spyridou, Veglis, 2011). These activities require a direct channel to supporters, circumventing media agenda setting and gatekeeping mechanisms (Schroeder, 2018). Therefore, social media is a valuable, tactical tool for meeting campaign objectives (Baldwin-Philippi, 2015; Jungherr, 2016; Lilleker et al, 2015; Vaccari, Valeriani, 2015). In this paper, we propose a new taxonomy of the politically engaged users on social media according to their level and character of activity. We put forward three different groups. The Clicktivists are those who only like the content. The Loyal activists are those who like as well as comment on party profiles, and are similar to positive opinion leaders. This classification is based on previous literature. We also introduce a new category, the Deliberators, i.e. those who comment on the profile but never like content. These three classes will be conceptualised in context in the next section prior to presenting our methodology and discussing our data.
4Facebook facilitates heavy consumers of political information, who due to their partisan attachments follow political parties, to instantly disseminate information through their networks. This constitutes an increasingly common practice in a variety of contexts (Lee et al., 2014). Tentative evidence shows that having a large and active followership can be beneficial. Social media can be seen as a credible source of political information (Johnson, Kaye, 2014). Parties that gain high levels of remediation seem to obtain high levels of popularity on a platform (Anstead, O’Loughlin, 2015). They thus employ a combination of paid optimisation of posts and fan promotion. Greater numbers of likes equate to a stronger likelihood of being widely seen due to the Facebook algorithm. Comments can equally give a post prominence, but they are beyond the control of the party and can therefore be positive or negative. Commentators can ask questions or make suggestions accelerating or diluting campaign objectives. Yet, as with media coverage, visibility may be sufficient. Bond et al’s (2012: 297) United States-based experiment found that “online political mobilization works. It induces political self-expression, but it also induces information gathering and real, validated voter turnout […] Social mobilization in online networks is significantly more effective than informational mobilization alone”. Bond and colleagues’ data show that organisations can mobilize supporters, and, more importantly, that those within supporters’ networks are more susceptible to influence through third-party endorsement effects. The relationship is reciprocal: through their followers’ reactions, supporters feel empowered and are encouraged to further promote party content (Weeks et al., 2017). While viewing does not simply equate with becoming more supportive, there is evidence that seeing content circulated within one’s network increases the likelihood of elaborating on that content independent of any partisan bias towards the original source (Lilleker, 2014). Peer-to-peer content dissemination may thus reduce the extent to which parties and their supporters co-exist within bounded, ideologically homogeneous communication environments (Messing, Westwood, 2014).
5Within this frame, remediation reveals to be important. Those who subscribe to party profiles on Facebook are most likely supporters (Norris, 2006) and therefore represent a minority with high interest in politics. Yet, their interest and likely greater knowledge about politics means that they may enjoy influence among the less political engaged in their networks (Karlsen, 2015). The concept of remediation is at the heart of the classic two-step flow model of political communication (Katz, Lazarsfeld, 1955). Classically, journalists alone awarded political parties’ prominence and credibility. In the digital age, a range of players have roles in the political information cycle, with party supporters being potentially pivotal during elections. A study of the 2005 United Kingdom general election (Norris, Curtice, 2008) found party website visitors conformed to predictions from previous scholar research: male, older and highly informed (Norris et al., 1999). Yet, content remediated by influential users through online networks was of a more heterogeneous character than the one found within the party supporter community. With the increased diversity of those who are politically active, it is likely that those accidentally exposed are even more heterogeneous. Bond et al’s (2012) experiment demonstrated that users receiving a “reminder to vote” message endorsed by a random selection of their friends were significantly (0.4%) more likely to have voted than those who received a neutral non-endorsed message. Remediation with an accompanying endorsement can have a real-world political impact.
6However, remediation is the preserve of the few. Supporter communities are mostly passive containing only a “cyberactivist” elite minority (Scarrow, 2014). Cyberactivists are the most active and committed party supporters (Wojcieszak, Rojas, 2011). They are probably the online and offline opinion leaders that Karlsen (2015) deemed critical for directing the flow of political communication in society. Yet, research has found that in general parties do not succeed in obtaining a high return from social media campaigning (Ross et al., 2015). They fail to directly request remediation from their communities and do not interact with supporters. Party mobilisation strategies do not achieve to empower their supporters as “co-producers” of the campaign (Jensen, 2017). Although incidental contact with diverse political content and perspectives is more likely to occur online (Brundidge, 2010; Kim, 2011; Colleoni, Rozza,Arvidsson, 2014; Lee et al., 2014; Choi, Lee, 2015), parties may not benefit hugely from this accelerated pluralism. (Jung, Sundar, 2016).
7Research has also shown that the intensity of campaigning affects voter turnout (Green, Schwam-Baird, 2016). If such an effect can be found in a tight state race in the United States, will the frequency of posting on social media also correspond to the activities of followers? Arguably, staunch party supporters should be keen to support a campaign, yet pushing them up the ladder of activism or loyalty (Lilleker, Jackson, 2014) requires strategy and effort. Building a community that interacts and self-mobilises may be key. Rojas and Puig-i-Abril (2009) found that the more individuals express themselves politically using digital technology, the more likely they are to attempt to mobilize others through social networking sites, which in turn translates into greater offline engagement. Therefore, the dynamics of a community, and the extent to which supporters and party leaders interact, may have an influence on aggregate participation rates (Thrall et al., 2014). Simple strategic differences matter, variations in the tone, timing, and content of posts shape significantly rates of likes and comments (Xenos et al., 2017). In this sense, the overall communication style is important (Vaccari, Nielsen, 2013; Williams, Gulati, 2008). Equally, the network size is positively correlated with the number of likes and comments.
8Interaction has long been seen as the gold standard for a community. However, parties tend not to encourage comments or respond to them. Thus, why users offer comments, especially when that is the only action they perform, is potentially complex and multifaceted. Commenting can evidence trolling, or alternatively a desire to interact with the like-minded in an ideological echo chamber. Conservatives, it is argued, seek confirmation bias and are more likely than liberals to prefer an echo chamber environment; liberals seek debates (Jost et al., 2003; Jost, Krochik, 2014). However, communication variables may be the most important driver of user responses. Research demonstrates higher engagement when the host or other users express opinions, share information and links to other media. Basically, users respond when communication corresponds to norms of behaviour on social media, eschewing interaction with blunt campaign demands such as fundraising efforts. Some researchers suggest that opinion-led content, consistent with emerging styles of participatory engagement is more appropriate for online political discourse (Loader et al., 2014). Overall community remediation is deemed important, and increasingly so by parties; yet explaining the dynamics of remediation is complex. In the next section we set out our hypotheses.
Theorising community gain and engagement
9Social media offers significant opportunities but also adds new costs and social pressure to parties seeking to maximise their potential (Nitschke et al., 2016). Research shows that parties with a high chance of increasing their share of seats, or winning power, tend to be more active on Facebook. On the contrary, parties with slimmer chances appear less incentivised (Lev-On, Haleva-Amir, 2018). Disparities in effort influence follower interest and engagement, with the benefits of a proactive strategy deemed significant. However, offline dynamics, mass media prominence and historical elements, may be the overarching variable that determines which parties benefit most from the affordances of social media platforms. The politics-as-usual or normalization thesis (Margolis, Resnick, 2000) predicts that because of their greater resources, major parties and candidates will have a more sophisticated and regularly updated web presence than minor parties and candidates. This explanatory hypothesis has remained powerful until very recently (Gibson, 2015). In contrast, the equalization hypothesis (Margolis et al., 1999) regards the web as a democratising technology that levels the political playing field by reducing the gaps in campaigning costs between minor and major parties (Lilleker, Jackson, 2010). Social media platforms are found to be particularly helpful in increasing the visibility of more electorally marginal parties (Gueorguieva, 2007). Research equally shows that minor parties can not only be more active online but may gain visibility comparable to that of more established parties (Larsson, 2016). But the normalisation versus equalization debate overlooks more nuanced and complex explanatory patterns (Wright, 2012). The factors understood as indicators of normalisation may actually result from large, catch-all broadcasting messages which draw a voluminous, interested but inactive followership (Nah, Saxton, 2013). In contrast, activist groups seek to mobilise and interact with existing followers rather than build a mass following. Consequently, debates around normalisation are hotly contested. Cardenal (2013: 87) argues that “large parties that can realistically expect to win elections and occupy the government may have an extra incentive to campaign on the internet to win additional votes than small parties”. To frame our exploration of supporter behaviour we hypothesise that larger parties enjoy a higher dividend, with each follower representing an endorsement and a potential activist, so suggesting that smaller, more marginalised parties may attract incrementally smaller online communities especially during the campaign (H1).
10Ideological differences in communication style appear to have diminished since 2010, signalling a levelling of the playing field between different types of parties (Klinger, 2013; Vergeer, Hermans, 2013). However, we hypothesise that while mass parties overall gain greater likes, more ideologically extremist parties may attract a more committed following (H2). Indeed, better resourced parties may be more active on social media, and have larger follower communities (Larsson, 2017). Yet, ideologically marginalized parties may gain more cyberactivists (Gunnarsson Lorentzen, 2014) because they attract a more cynical, disenfranchised and more active followership who are empowered by less politically-correct and more extremist political actors (Gil de Zúñiga et al., 2010). Crucial is the evidence that social networks represent a potential source of organizational coherence for fringe political parties, allowing them to develop relationships by promoting “low threshold activities” (Vaccari, 2013). However, we recognise that independent of resource or ideology, campaign intensity can influence follower behaviour. Hence, we hypothesise that those parties who are most active on social media, and so provide a plethora of ways to engage, will enjoy higher levels of engagement (H3). In particular, we draw on evidence that suggests that parties most active on social media “may broaden party-related engagement beyond party members, allowing ‘citizen campaigners’ to play a greater role in the activities and organizational lives of parties” (Vaccari,Valeriani, 2015).
11Larger parties tend to have the largest online networks (Vergeer et al., 2013), so it is likely their higher numbers of influential followers will award them greater remediation (Anstead, O’Loughlin, 2015). However, social media no longer conforms purely to the rules of politics-as-usual. The behaviour of the profile host and the form of their posts can determine the extent to which content results in accelerated reach. Our data allows us to explore these dynamics and to study the impact of communication strategies on the actions and interactions of party supporters. We can assess whether parties with the highest support and resources still develop the most sophisticated offline and online campaigns and achieve the largest and most engaged communities (Schweitzer, 2011). Alternatively, we may find followers of smaller parties work in smaller but more committed communities that seek to mitigate the lower resources and media attention (Gibson,Ward, 1998). We can also identify patterns of community growth and engagement driven by party ideology. This factor might prove particularly important in the context of EP elections, with some parties gaining a dividend for their stance on European integration (Hirzalla et al., 2011). We ask whether engagement is a factor of a parties’ campaigning intensity. The data will also examine links between the forms of communication, in order to determine whether patterns of engagement within a party’s community can be predicted by the type of content produced.
12The study is based on data from the Facebook official profiles of the 253 political parties standing across the 28 EU member states for election to the 2014 EP. It was collected by Sotrender  during the two weeks before the EP 2014 elections.
13Community gain: number of new followers of the profile gained during the two weeks before the campaign (N= 253 parties). The overall Facebook community developed on average by 1967 (Standard Deviation [SD]=5478, Max=47176). We define a “follower” as a person who subscribes to the party page, not only to a post. The reason for following may vary, from being a supporter to having a civic interest in one or multiple parties or a professional need to be informed about party activities. Followers may include journalists, students or activists for opposing parties. They do not necessarily engage in any further activity on the profile, however they have higher chances of being notified in their news feed when a party posts content.
14Community activity: The number of times content posted by a party received a like or comment. We model those who engage across three distinct categories users a detailed description is in the following section of the paper. In order to overcome the strong over-dispersion, we use negative binomial regression (Generalised Linear Models) (Hilbe, 2011).
15Party characteristics: Party years of existence (continuous) number of years since party was established (M=30.7, SD=34, Max=182); Party size (dummy) categorization of the parties according to their vote share in the last national elections and number of seats in the national parliament: major parliamentary (above 20% of vote at the last national election, N=48), minor parliamentary (other parties present in parliament, N=123), fringe parties (N=82, reference category, not represented in parliament). Party ideology (dummies): Right leaning (N=76), Left (N=114), Centre (N=28) and Single issue/other (N=35) parties, EU positioning (dummies): Pro-EU (N=161), Neutral (N=44) and EU-sceptics (N=48). Party in government (dummy, 1= in government N=64). Community size t1 (continuous) measure of the size of the community on Facebook profiles at the beginning of campaign (M=25102). Party ideology and EU positioning variables are based on the data delivered by the EU profiler study (Garzia et al., 2015).
16Country characteristics: Countries are included as fixed effects in the regression, with Ireland being a reference (as the country with average Internet penetration (M=78.2%) closest to the overall EU average (M= 77.8%).
17Communication strategy - characteristics of the posts of political parties: posts are put into one exclusive category: Video, Photo, Link and Status (text only, is a reference category). This categorization is similar to that employed by Facebook algorithms which attribute each category a hierarchy of importance and thus visibility. Our categories are exclusive, that is to say that ‘video’ can be accompanied by “text”, however it is attributed only to the highest importance (‘video’) category. Our data set does not allow us to control for double or triple codification of the post, we thus adopt the classification employed by the Facebook algorithms.
18In our sample, the parties produced and posted 2065 videos (M=8.16, SD=10.11, Max=94), 7260 photos (M=28.7, SD=34.4, max=331), 6435 links (M=25.4, SD=43.6, Max=557) and 1036 text posts (M=4, SD=11.6, Max=113). The values in regressions are logged.
The European Facebook party communities
19Consistent with our hypotheses, we firstly explore the extent to which European parties, especially during election campaigns, exploit the potential offered by the general Facebook penetration rate in each country. Then, we explore the gain in new community members during the campaign (H1). We find that the size of the social media communities differs according to countries and their population. Not surprisingly, the largest European Union countries (Italy, United Kingdom and Germany) gather the largest Facebook communities. Facebook general statistics allow us to calculate country comparisons in terms of the size of political communities in relation to Facebook users per country. We do so by dividing the number of all parties’ followers in a country at the time of EP elections by the number of Facebook users in that country (Table 1). Malta is the definite leader, as political parties attract 22% of the potential Facebook followership. Hungary (12%), Cyprus and the Czech Republic (10%) are next, below are Luxemburg (8%), Sweden and Austria (6%), Denmark, Slovakia, Belgium, Portugal, Greece and Poland (5-4%), with parties from other countries gaining less than 3% of the potential followers. France (1.6%) and Latvia (.04%) attract the smallest ratio. This data demonstrates significant differences in the online political community across European nations.
Table 1. EU party communities by country
Table 1. EU party communities by country
Community building during the elections
21Given party strategies to attract and mobilise supporters, it is important to determine the factors which attract new followers during the campaign period (Table 2). The regression analyses indicate that three factors dominate in explaining the expansion of party Facebook communities. First, systematic and frequent updating of engaging content; if parties post videos they are more likely to attract new community members. This finding partially confirms H3 and is consistent with research which indicates that videos are most likely to have the potential of going viral on Facebook, as well as of gaining accelerated reach into the external network of community members (Koc-Michalska et al., 2016). Second, the size of the initial community is also important. The larger the community, the greater the number of cyberactivists during the campaign is. Therefore, data confirms our intuitive hypothesis; that the campaign environment is not built during the electoral period but through permanent communication and long-term loyalty building strategies. But the strongest factor confirms our first hypothesis and the continued relevance of normalization. Major parties attract on average the most new-community members, and this effect is reinforced if we introduce the interaction term: Major parties * Community size t1. The data indicates that major parties have the largest communities prior to a contest and in turn obtain most additional followers during the campaign. Thus, overall, we find the rich parties get richer in terms of gaining followers.
Table 2. New community gain during the electoral campaign
Table 2. New community gain during the electoral campaign
22However, there are outliers that counter this pattern. A small number of parties outside the traditional parliamentary system attracted an exponentially larger number of followers during the campaign. These are parties described as populist (Inglehart,Norris, 2016), namely Podemos (Spain, populist-left), having attracted 47000 newcomers, Congress of the New Right (Poland, populist-right), with 40000 new followers, and UKIP (United Kingdom, populist-right) having gained 35000 new followers during the two weeks before election day (see Table 5). These data confirm that parties in opposition are likely to attract new followers, probably interested in alternative or non-establishment programs (Governmental parties β = -.597). Other party characteristics, including political ideology or positioning towards the European Union show no statistically significant value.
23In our approach to the taxonomy of the social media community, we propose three unique groups: Clicktivists, Loyal Activists and Deliberators.
24Clicktivists, are those followers who only like  party content and who do so once or frequently. This form of activity, described by some as clicktivism (Halupka, 2014), indicates a low-level form of political engagement. However, if political parties seek to maximise their visibility and virality, liking could be the most attractive and desirable form of follower behaviour. Hypothetically then, the latter represent a very important group as they may evidence success for party strategy. In our data, they constitute 86% of those performing activities on party posts across European countries. It is possible that not all Clicktivists are followers of the party profile. A like simply indicates that the post was appealing and relevant at the point of viewing and can be performed by any Facebook user accidently exposed to a party’s post through their networks (Norris, Curtice, 2008), or through targeted communication by the political party or algorithms.
25Loyal Activists intensively like content while also commenting on posts, and so probably represent a more committed group who seek to influence their network (Weeks et al., 2017). These are constant promoters, perhaps party members or highly mobilized supporters, and are of high value to a party as they award credibility (Anstead, O’Loughlin, 2015). They are possibly opinion leaders (Norris, Curtice, 2008). Loyal Activists are followers who frequently engage in a suite of actions to support the campaign, spreading viral marketing, endorsing campaign posts or participating in discussions, most probably contributing positive or constructive content. Loyal Activists represent 9% of a party’s followership.
26Deliberators comment but never like the content of the post and constitute 5% of our sample. Possibly, this category includes opponents’ supporters, or a strategically less desirable group who may demand responses to questions or post critical comments. It is likely that some within this category are keen to engage parties on vital social issues, but they are not, currently, party supporters or sympathizers who wish to (neutrally) contribute to online conversation. Most probably, deliberators represent a potential problem as they perform no act that is supportive of party aims and may question, critique, challenge or insult the party (for evidence of this behaviour see Zurutuza-Muñoz, Lilleker, 2018). They may also be politically or financially motivated trolls working for an opponent to undermine arguments or accelerate and negatively inflame debates.
27The regression results (Table 3) show that in order to meet strategic objectives, i.e. attracting Clicktivists or those Loyal activists, parties need to post content regularly; this confirms H2. Frequency of posting plays a stable, significant explanatory function: the more parties post, the more likely they are to receive engagement; the more attractive the format (link, picture or video over text threads) the more significant the likelihood of triggering responses. This finding is similar to Cvijikj and Michahelles (2013), who in their analysis of the top 100 FTSE (Financial Times Stock Exchange) companies found that entertaining content gains the greatest levels of engagement from communities. Deliberators are equally attracted by pictures and videos. It is quite surprising, as one could imagine that this specific group would be most interested in textual information, the latter being the basis for their questions or challenges to the party. Qualitative analysis, outside of the remit of this paper, is needed to understand if this suggests that their comments are shallow, failing to elaborate on the subject of the content.
28Political variables offer very limited explanations for different engagement patterns. Again, there is an indication of the power of the normalization thesis (H1). Parties with voluminous electoral base gain the highest number of engaged followers within their community. We find limited effects from party ideology or from the party’s position towards the European Union. There is no difference in terms of attracting Clictivists or Loyal Activists, however Deliberators seem to gravitate to the content of parties with a right-wing or pro-EU stance. As at European parliamentary contests these parties represent polar opposites, we hypothesize there may be two ideologically opposed groups of Facebook users. These users might give support to the party whose stance they share, while challenging those they oppose by posting critical comments without ever liking the profiles’ posts. This requires further investigation, however, as we could imagine a scenario where virulent pro and anti-European activists vent political frustrations against their opponents. The same phenomenon was evident on Facebook during the United Kingdom’s subsequent referendum on European Union membership (Lilleker, Bonacci, 2017).
29Table 3 also indicates country differences in Facebook engagement patterns. We find that Sweden and Ireland are outliers with the highest levels of community engagement independent of content. French political parties attract the least active communities, well below the European average. However, there are no clear national explanatory factors.
Table 3. Regression analysis. Community taxonomy
Table 3. Regression analysis. Community taxonomy
31For a more granular understanding of the community activities, table 4 indicates the average scores for community sizes in terms of taxonomy groups, party’s size and political standpoint. Interestingly, we find that fringe major parties - those which earned above 1% in the last national parliamentary election but, often due to the electoral system, did not receive representation in the parliament -, are more likely to attract loyal activists. Table 5 indicates the top ten parties which received the most attention in each of the community categories.
Table 4. The size of community by taxonomy groups by Party size and Political standpoint
Table 4. The size of community by taxonomy groups by Party size and Political standpoint
33Additionally, and similarly to community gain analysis, we looked at the basic numbers of activists by party. Again, we find similar results; among the top 10 are the populist parties who attract the largest engaged communities: Northern League (Italy), UKIP (United Kingdom Independence Party), Alternative for Germany (AfD), National Democratic Party (NPD) (Germany), Congress of the New Right (Poland) or Five Star Movement (Italy). Yet, these parties also attract the largest number of deliberators suggesting a polarized conversation may take place on those profiles (Table 5).
Table 5. Top ten parties having the most active community groups
Table 5. Top ten parties having the most active community groups
35Overall, our data shows that normalisation dominates when it comes to community size and followership growth (H1). We find that the levelling of the playing field between different types of parties is largely not realised (Klinger, 2013; Vergeer, Hermans, 2013). Our findings regarding fringe parties (Table 4 and 5) partially demonstrate that parties with a stronger ideological stance attract a more committed following (Gil de Zúñiga et al., 2010). Largely, however, H2 is not proven in the regression. Parties with clear right wing or pro-EU positions do attract larger numbers of deliberators suggesting, counter to H2, that they may also attract more opponents on Facebook and that their pages become sites for debate and argument. We also know that a specific group of populist parties are among the top parties to attract Clicktivists as well as Loyal Activists. Therefore, while normalisation is the dominant explanatory model, H3 is partially supported as we find that those parties who post engaging content (video and photo) are rewarded by their communities with higher levels of engagement (Vaccari, Valeriani, 2016). While the evidence may suggest that a more granular series of explanations are at play, it is difficult to make a strong case for the mobilisation thesis. Larger parties still tend to have the most voluminous online networks (Vergeer et al., 2013), as well as higher numbers of influential followers who remediate their content (Anstead, O’Loughlin, 2015). They benefit from more resources, and develop more sophisticated content, attracting larger and more engaged communities (Schweitzer, 2011). The only counter argument is that followers of fringe parties may operate in smaller but more committed communities (Gibson, Ward, 1998). Yet, while they may have a greater percentage of loyal activists, it is unlikely they earn similar reach as the more engaging content of their larger rivals. The numerical superiority of the communities of major parties gives them a significant advantage. Hence, largely, on Facebook, the power laws of politics-as-usual tend to remain dominant in the 2014 EP election contests with Facebook mirroring the offline dynamics of the campaign.
The study was possible due to support from the Audencia Foundation.
Sotrender.com is an academic-led company running the application analyzing social media. For the purpose of the project the data delivered is a real time archive of the posts and reactions to them by the public. The data were archived just after the election, thus any changes made after the campaign are not taken into account (e.g. additional likes clicked after the campaign). Sotrender does not control for the possible bots or so called “likes farms” but makes a scan of profiles as they are visible to the follower.
In the time of the data gathering, other reactions (love, angry, sad, etc. emojis) were not yet introduced by Facebook.