An online electoral connection? How electoral systems condition representatives’ social media use

Abstract

This article analyses the impact of electoral institutions on the re-election campaigning and outreach strategies of Members of the European Parliament on the Twitter social media platform. Social media offers politicians a means to contact voters remotely and at low cost. We test the effect of diverse national proportional representation electoral institutions in European elections on a possible online electoral connection. We draw upon an original dataset of Members of the European Parliament Twitter activity before, during, and after the 2014 European elections. Our results confirm that variation in electoral institutions leads to meaningful differentiation in representatives' social media campaigning, which is further affected by national party, voter and legislator characteristics. Representatives make constructive use of Twitter, but there is no sustained online electoral connection.

Keywords

Electoral system European Parliament re-election campaigns social media Twitter

Introduction

Social media allows politicians to interact with colleagues, journalists and the general public, who can all help representatives’ re-election quests by spreading their message. In particular, representatives can foster an electoral connection at essentially no cost by engaging with voters (Lassen and Brown, 2011; Mayhew, 1974). By contrast, the traditional campaign toolbox is costly in terms of resources. Personal meetings with constituents cost time, while leaflets and other forms of advertisement cost money. Social media therefore constitutes an attractive alternative to these potentially expensive forms of outreach and communication. Next to Facebook, Twitter is one such popular platform. It allows users to send messages (tweets) to their followers as well as to share (‘re-tweet’) the tweets of others. Tweets are limited in length to 140 characters and thus resemble text messages. Users are identified by their ‘handle’ (e.g. @MartinSchulz). So-called hashtags (e.g. #EP2014) allow users to engage in debates and to identify trending topics. Social media thus carries the promise of the electoral benefits of personalisation and interaction (Kruikemeier, 2014; Kruikemeier et al., 2013), but at the very least it allows representatives to broadcast their message directly to voters (Golbeck et al., 2010; Graham et al., 2013).

Such qualities are particularly germane within the context of the EP, which is marked by a weak ‘electoral connection’ between MEPs and their electorate. This is because European elections are ‘second-order national elections’ that are typically of low salience to voters and fought on national, rather than European issues (Hix and Marsh, 2011; Reif and Schmitt, 1980; Van der Eijk and Franklin, 1996). Social media might help invigorate the electoral connection by allowing representatives to reach everyday citizens more directly and to interact with voters across large constituencies. However, this is not to suggest that social media is a panacea for political actors. User numbers of Twitter, in particular, remain relatively low. Moreover, if politicians make use of social media, they risk exposure to abuse given the lower inhibition towards impolite and uncivil communication online compared to offline (see e.g. Papacharissi, 2004). While there is thus a general sense that social media matters, we do not know under what conditions and to what extent this is the case.

Accordingly, this article investigates the extent to which parliamentarians use social media platforms, such as Twitter, to foster an online ‘electoral connection’ with voters (Lassen and Brown, 2011; Mayhew, 1974). Carey and Shugart (1995) demonstrate that the ballot structure and institutional design of electoral systems determine parliamentarians’ incentives to invest in cultivating a ‘personal vote’. We develop hypotheses based upon this reasoning, in order to test whether MEPs use Twitter to build such an electoral connection and campaign for a personal vote via social media. In particular, we assess to what extent this ‘electoral connection’ is sustained during a three-month period before and beyond the immediate elections.

The possible emergence of an online electoral connection within the EP and the degree to which it is shaped by national proportional representation electoral institutions has important implications for both the literature on comparative institutional design, as well as the broader literature on campaigning for re-election. Integrating social media campaigns within a more traditional comparative institutions framework, we seek to contribute to the debate on the implications of the internet on politics (Farrell, 2012). Furthermore, given social media’s ease of use and free-of-cost nature, we seek to address whether social media as a new tool upsets traditional campaigning dynamics (see Small, 2008). Therefore, we examine whether established, major parties are more active on social media than smaller ones. Moreover, we investigate to which extent social media campaigning may be demand driven based upon the composition of party voters. Yet, we consider our main contribution as testing whether standard theories of electoral institutions can explain social media activity and whether established findings on a weak electoral connection hold for online, social media campaigns.

Two important limitations should be noted. First, electoral institutions only vary across member states within the scope of Article 1 of Council Decision 2002/772, which stipulates that MEPs ‘shall be elected on the basis of proportional representation’. European elections may be an ideal laboratory for testing different variants of proportional representation within a single legislature (Bowler and Farrell, 2011; Farrell and Scully, 2005), but they do not allow us to study majoritarian systems. Second, while our focus on the electoral connection means that we investigate re-election campaigning, incumbents and their campaigns might be substantively different from challengers and their campaigns. While this does present a potential drawback that our results are not generalisable to all campaigns, we do believe that our examination of all re-election seeking MEPs and their use of Twitter permits us sufficient analytical purchase to draw important conclusions.

In order to examine the role of Twitter in MEP campaigns, we analyse the extent to which outgoing MEPs from the seventh session of the EP use Twitter during the 2014 EP campaign. To do so, we draw upon original data on their Twitter usage, taken over a roughly three-month period between March and July 2014. These data are collected through the automated programme interface with the social media platform, using the Chorus Tweetcatcher software package (Brooker et al., 2015; Chorus Analytics, 2014). Raw Twitter data are then matched with relevant information about electoral institutions and the personal background of each MEP.

Our findings indicate a strong relationship between traditional electoral system variables, such as list type and district magnitude, and Twitter usage – a relationship that we might expect to observe in an ‘offline’ campaign. By contrast, established, major parties and resource-strapped minor ones equally seem to use social media for electoral campaigning, even though we might expect the former to conduct more professional, multifaceted campaigns. Our findings also provide novel insights that the more citizens MEPs represent, the more active they are on social media. This means that they use Twitter constructively where it would otherwise be resource intensive to campaign. They are particularly active on Twitter where demand for social media campaigning may be high, e.g. if parties target younger voters and where social network usage is high, even though this finding does not extend to the internet usage of party supporters specifically. While this suggests that social media might help invigorate the weak electoral connection, the results demonstrate that Twitter activity is largely limited to vote seeking during the campaign rather than constituting a sustained attempt to create an online ‘electoral connection’ (see also Vergeer et al., 2013).

Electoral politics and social media

MEPs face serious challenges when campaigning. The 2014 post-election survey found that only 57% of the electorate felt that they had all the information they needed in order to cast their vote (European Parliament/TNS Opinion, 2014: 16). Only a minority of 47% thought it was very important which candidates were elected in their country (European Parliament/TNS Opinion, 2014: 67). Even fewer citizens (42.54%) turned out to vote, marking the lowest ever turnout in European elections. This was despite the elections being advertised as ‘different’, given that the major party groups nominated lead candidates for the first time (see Hobolt, 2014; Peñalver García and Priestley, 2015; Schmitt et al., 2015).

Accordingly, European elections are considered ‘second-order national elections’ (Reif and Schmitt, 1980; Van der Eijk and Franklin, 1996). Given their low salience, voters cast their ballot largely based upon national considerations, leading to a weak electoral connection between representatives and the electorate (Hix and Marsh, 2007, 2011; Hobolt and Spoon, 2012). Nevertheless, Hobolt and Høyland (2011) demonstrate that rather than simply punishing parties in national government, voters reward the experience of candidates. If such personal characteristics matter, there may be incentives for MEPs seeking re-election to invest in creating a long-term electoral connection. MEPs might be well advised to signal their accomplishments and look to voters in keeping with Mayhew’s (1974) three strands of an electoral connection: advertising, credit claiming, and position taking. Therefore, we analyse whether MEPs use social media and how prolific they are in doing so.

In this respect, we assume that Twitter activity is indeed aimed at the three objectives outlined by Mayhew and that it is beneficial to fostering an electoral connection, despite the risks of online abuse and incivility (Gervais, 2015; Papacharissi, 2004). Indeed, there is strong evidence that most politicians use Twitter for one-way ‘broadcasting’ rather than interactive dialogue (Golbeck et al., 2010; Graham et al., 2013). No matter whether a MEP tweets a link to his or her speech, re-tweets an article in a newspaper, or potentially replies to a follower, these actions merit consideration as campaigning, because they will arguably reflect their preferences, views, or constituency service (Jackson and Lilleker, 2011; Lilleker and Koc-Michalska, 2013). In essence, this means that we consider each tweet simply as a message broadcast by the representative, and equivalent to an election leaflet (Hix and Hagemann, 2009) or direct e-mail (Bowler and Farrell, 2011). We consider this a valuable step in studying the impact of electoral institutions on the electoral connection.

The impact of social media on the electoral connection

Social media lends itself to building and maintaining an electoral connection, since it allows for politicians to conveniently contact voters. However, in a small study of Dutch candidates in the European election, Vergeer et al. (2013: 477) find that ‘most candidates in 2009 still used Twitter reluctantly’. Twitter was still a new phenomenon then, and these early adopters used Twitter for electoral rather than continuous campaigning. With social media and Twitter in particular being popular beyond early adopters today, social media platforms might provide important benefits when campaigning for personal votes. In particular, it comes at a low cost when compared to the traditional electioneering toolbox of election leaflets, advertisements, and events. This low cost might enable representatives to sustain an electoral connection remotely without facing a trade-off between effective legislating and constituency work (Farrell and Scully, 2010; Scully and Farrell, 2003).

Therefore, the larger the number of citizens represented, the more attractive it might become for representatives to use Twitter. In the EP, the number of seats per national delegation is allocated on the basis of regressive proportionality, whereby larger member states have more seats than smaller ones, but are relatively under-represented. Whereas in large member states such as Germany, France, or the United Kingdom there are more than 800,000 inhabitants per MEP, the ratio falls to less than 100,000 inhabitants per MEP in small states such as Malta and Luxembourg (Corbett et al., 2011: 29–30). MEPs from larger national delegations are thus tasked with representing more citizens, making the formation of an electoral connection all the harder.

Therefore, if MEPs make constructive use of social media, then we might expect MEPs representing more citizens to have a stronger incentive to use social media.

H1: The more citizens represented per MEP, the more active the MEP will be on Twitter.

The impact of electoral institutions on the electoral connection

In spite of their second-order nature, there is strong evidence to suggest that electoral institutions shape the behaviour of legislators in the parliamentary and electoral arena. Even though the EU has passed legislation stipulating a ‘uniform procedure’ based upon proportional representation, important national differences remain (Farrell and Scully, 2005, 2010). Accordingly, the electoral threshold varies between 0 and 5%, the ballot structure may allow for preferential forms of voting, and the district magnitude (the number of representatives elected in an electoral district) depends upon a combination of total delegation size and national rules. This has important implications for the incentives of MEPs and electoral competition (Bowler and Farrell, 2011).

For our purposes, the two most important differences in European elections relate to district magnitude and whether the ballot structure allows voters to express a preference for candidates. In preferential systems (such as open lists, ordered lists, or single transferable vote systems), voters can choose between different individual representatives of the party. As a consequence, candidates and parties compete against each other. In non-preferential (closed list) systems, voters cast their ballot for a party only, so electoral competition is between parties. Therefore, preferential systems are more candidate- than party-centred and politicians have an incentive to campaign for a personal vote that differentiates themselves from their colleagues (Bowler and Farrell, 1993). We expect that this quest will lead MEPs to seek out venues such as Twitter to promote themselves as individuals.

Carey and Shugart (1995) provide nuance to our expectations related to ballot structure by taking district magnitude into account. Their seminal work demonstrates that in candidate-centred (i.e. preferential) systems, an increase in district magnitude is an additional incentive to build a personal reputation. The more representatives are elected in a district, the more candidates there are, and the more important individual reputation. In contrast, an increase in district magnitude denotes a decrease in incentives to cultivate a personal reputation under party-centred (i.e. non-preferential) systems. The more representatives are elected, the less important the individual candidate and the more important the party label. Therefore, we expect that ballot structure and district magnitude will influence MEPs’ social media usage to cultivate a personal vote.

Previous research provides evidence that electoral system features influence MEPs’ understanding of their role, that this influences their work, and that the results make a marked difference for voters. In line with this, Farrell and Scully report that ‘more open systems are associated with a greater constituency focus by elected representatives’ (2010: 51), based upon MEPs’ responses to a survey asking whether they hold clinics for voters (see also Bowler and Farrell, 1993, 2011). The impact of electoral institutions on the representatives’ efforts to cultivate a personal vote is confirmed by Hix and Hagemann (2009) in a study on campaigning during the 2004 European elections. They focus on traditional ‘offline politics’, using the Eurobarometer post-election survey, which asked respondents whether they had received any election leaflets. The results demonstrate that in preferential electoral systems, more citizens had received leaflets. With regard to district magnitude, they find the expected negative relation of district size and proportion of voters who had received information. This suggests despite their second-order nature, standard insights on electoral politics do apply to ‘offline campaigning’ in European elections.

In contrast, Bowler and Farrell’s (2011) findings on the same elections are more mixed. They demonstrate that electoral institutions perform poorly in predicting campaign effort, but well in predicting party- or candidate-focused campaign activities. Vergeer et al. (2013) more narrowly study online campaigning of candidates running in the 2009 European election from a communications angle. They fail to find any significant impact of electoral system variables on website features such as newsletters, personal information, and social network links. A sophisticated study on the campaigns of 1336 candidates in these elections by Giebler and Wüst (2011) shows that preferential voting did not influence campaign intensity as measured by time and money spent, but that it had a significant positive impact on the extent to which candidates used post-modern campaign tools, such as websites or social networks. Most recently, Lorenzo Rodríguez and Madariaga (2016) demonstrate that the ballot structure affected online campaign intensity in the 2014 European elections.

This raises the question of whether electoral institutions affect MEPs’ campaigning on Twitter. In line with Carey and Shugart (1995), we test whether the connection between preferential voting and district magnitude may serve to mitigate or to exacerbate a given MEP’s need to cultivate a personal vote via Twitter. From this follows our second set of hypotheses:

H2a: In non-preferential voting systems, the larger the district magnitude, the less active the MEP will be on Twitter.

H2b: In preferential voting systems, the larger the district magnitude, the more active the MEP will be on Twitter.

The electoral formula also has implications for the competitiveness and predictability of re-election. In non-preferential systems, those at the top of a party list can often be certain that they will be re-elected, whereas the fate of lower ranked politicians is less secure. Preferential systems that allow for voters to support individual candidates within a list or to reorder the list entirely contribute an additional aspect of uncertainty. Therefore, we also include the safety of a candidate’s electoral position into our analysis.

Giebler and Wüst (2011) differentiate conceptually between hopeless, unsafe, and safe seats in their study of candidates in the 2009 European election. They suggest that candidates in the latter two categories exhibit similar behaviour in terms of campaign intensity and tools used (Giebler and Wüst, 2011: 58), but that hopeless candidates significantly lag behind, including in the use of electronic and online campaign tools. While this may not be surprising in the case of hopeless candidates, Lassen and Brown (2011) more generally find that sitting Members of Congress from the most contested seats (i.e. those in the weakest position) were less likely to build connections with voters via the use of Twitter than those relatively more certain of their re-election.

If we are able to replicate this surprising result in European elections, where electoral lists are drafted by parties in order to insulate party leaders with ‘safe’ positions at the top (where ballot structure allows), then we should expect the most prominent party members and effective figureheads of national parties’ campaigns to be the most active on Twitter. While the mechanism of seat contestation is certainly distinct from that of the U.S. Congress, the position of list leaders may also suggest that the ‘safest’ politicians have the responsibility to take to social media more frequently, in the capacity of their leadership role. Accordingly, our third hypothesis focuses on the impact of list safety:

H3: The safer the list position, the more active the MEP will be on Twitter.

Does social media upset traditional power relations?

There are also competing expectations as to how resources, usually operationalised by party size, influence professionalisation of campaigns and the uptake of social media as a campaign tool (Gibson and Römmele, 2009; Lilleker et al., 2011). A debate in media and communication studies pits a so-called equalisation thesis against a normalisation thesis (Small, 2008). The equalisation thesis posits that innovations such as websites and social media present a cheap avenue for small parties to make up ground on larger, better funded ones. Adopting social media would allow representatives from smaller parties to bypass the gatekeeper function of traditional media (Lassen and Brown, 2011), in which MEPs from smaller groups are mentioned less frequently than those from large groups (Gattermann and Vasilopoulou, 2015). For instance, Lilleker and Koc-Michalska (2013) find that MEPs from minor national parties achieve higher performance scores in different categories of online campaigning than those from major parties. Equalisation then refers to social media levelling the playing field between resource-rich and resource-strapped parties. If we find support for these claims, it would suggest that social media upsets traditional campaigning dynamics and, if effective, might upset electoral results.

In contrast, the normalisation thesis stipulates that social media would form a natural part of election campaigns, regardless of party stature. As a consequence, the use of social media as a campaign tool would reflect the general professionalism and resources available to the candidate. Established, major parties can afford more professional campaigns, including social media as one of multiple channels (Lilleker et al., 2011). One direct way to observe party resources in a broadly comparative context would be to focus on their dominance in national elections. Thus, major parties, as measured by their seat share in national parliaments, should be expected to have more funds, staff, and training at their disposal, thereby allowing them to campaign more professionally. Another way is to identify government parties, whose candidates seem to commit more funds to campaigning (Giebler and Wüst, 2011). In line with the normalisation thesis, we hence investigate the extent to which national power ought to correlate with party resources and thus favour the usage of Twitter as a campaign tool.

H4: The more dominant the national party in the national system, the more active the MEP will be on Twitter.

Voter demand for an ‘online’ electoral connection: The national party’s target group

The incentives to use Twitter may not only depend upon national-level internet permeation of countries, that many studies control for, but also on the national party’s target group. To account for this variation, we introduce a variety of measures meant to proxy the demand for social media usage by supporters of the MEP’s national party. These measures, discussed in the following section, each capture various aspects that may leave a politician more inclined to turn to Twitter in his or her campaign and include aspects, such as the party voters’ mean age and their usage of the internet and social media to seek political information. In keeping with this line of thought, we expect the following:

H5: The greater the affinity for internet and social media among party supporters, the more active the MEP will be on Twitter.

Data and method

In order to test determinants of social media usage and the conditional impact of electoral institutions, we collected data on the Twitter usage of all MEPs during the lead-up to the May 2014 elections. Using an automated tool for data capture (Brooker et al., 2015; Chorus Analytics, 2014), information from the official Twitter handles of all MEPs using the application was collected,¹ beginning two months prior to the elections and concluding after the seating of the eighth EP in July 2014. Captured information relevant to this article includes the volume of MEPs’ tweets, as well as the number of their Twitter followers throughout the period.

We then pair the newly collected social media data with an original dataset on the career behaviour of MEPs during the seventh EP (Daniel, 2015). The resulting dataset is fully comprehensive for all outgoing MEPs from 2009 to 2014 EP.

Dependent variables

Of the 856 MEPs present in the legislature at some point during the seventh EP session, 341 had active Twitter accounts during the election period that were listed on the EP's public Twitter list, and 245 of those sought re-election. Tweets measures a given MEP’s total number of tweets emitted at various points of the re-election campaign.² We focus on specific time points to observe the volume of Twitter activity throughout the campaign. The periods observed allow for us to capitalise on variation in Twitter usage over time.

The Baseline Tweets version, using data from 25 March 2014, measures the total number of tweets that an MEP had ever made from their official Twitter handle, as of two months prior to the election. Although there is no ‘official’ start to the EP campaign, we do expect this variable to reveal valuable information about which MEPs are ‘typically’ the most active on Twitter. The T-2 Months version then captures the number of Tweets per MEP between 25 March and 24 April, which was the last month of relatively ‘normal business’ in the EP, as the last plenary of the session concluded on 17 April. The T-1 Month version between 24 April and 26 May traces activity during the final month of the campaign, when MEP activity on Twitter would be expected to be the most directly linked to campaign appeals. These two variables are evenly spaced and allow for direct comparison of the MEPs’ level of activity during the months of April and May, respectively. Then, using the T-1 Week version, we examine only the final week of the campaign – when Twitter usage should be almost entirely related to the elections, as members are fully engrossed with the campaign in their constituencies. Finally, the T+1 Month version examines the denouement period between 26 May and the seating of the new eighth EP on 6 July 2014, when the campaign has definitively come to a close. While we do not derive formal hypotheses for the effect of the electoral cycle on Twitter usage, our data nonetheless allow us to consider the dynamic usage of Twitter at various points in the campaign, adding useful context to our specific findings.

Independent variables

A number of independent variables are coded for each MEP observation, pertaining both to their structural and to individual situations in the election. Citizens Represented measures the total number of citizens (in tens of thousands) represented per MEP, averaged by nationality.³ As discussed in H1, MEPs that represent more citizens are expected to use Twitter more actively; this grants them a relatively costless and efficient platform for contacting large swaths of voters.

H2a states that for MEPs from countries without preference voting in EP elections, higher district magnitude will have a negative effect on MEP tweeting. H2b is the corollary hypothesis that states that MEPs from countries with preferential voting, larger district magnitude will have a positive effect on MEP tweeting. To capture this possibility, we construct a dichotomous Preferential Vote indicator that captures those EU member states that elect MEPs using any form of preferential voting. This includes lists that can be reordered or selected from, given a party’s suggested order, as well as preference lists where the party order is not suggested – such as in purely alphabetical lists. Average District Magnitude is then calculated on a national basis for the EP elections and is interacted with this measure.⁴

H3 posits that list safety positively affects social media usage. To operationalise this, we use polling data to identify an MEP’s List Safety on the electoral list.⁵ One particularly sophisticated method for identifying the safety of MEPs at election time was created by Giebler and Wüst (2011: 58), who identify ‘hopeless’ candidates during the 2009 election period. We begin by replicating their coding scheme to identify candidates from ordered lists (preferential or otherwise) who are not expected to win the election, given 2014 PollWatch projections (Cunningham and Hix, 2014), and assign them a value of 0.⁶ We then consider three additional categories of MEPs that Giebler and Wüst (2011) refer to as ‘promising’ candidates.

Candidates from ordered lists that are featured as list leaders are assigned the maximum value of 3 and considered to be the most likely to be re-elected, unless they come from national parties that were not projected by PollWatch to win any seats at all (i.e. splinter candidates that have been excluded from their original party or run under a new list). Our list leader category also includes lists that are constructed at the regional level, such as in France, as well as lists where the party has suggested an order, but where voters can prefer specific candidates, such as in Poland.

The next value, 2, is given to MEPs that are featured on party-ordered lists in positions that are likely to be elected, given PollWatch forecasts. While such candidates are not as secure as list leaders, we consider them ‘safe’. The value 1 is then given to MEPs from purely preferential lists, given without an order, where voters cannot simply default to the party’s suggestions.⁷ In such situations, all spots are theoretically equally ‘competitive’, although less so than both categories 2 and 3. Given the expectations of H3, we expect that the safest MEPs will be among the most active users of Twitter during the election period, given that they are also list leaders.

H4 posits that MEPs from dominant national parties that might be endowed with greater resources will be more active on Twitter. We use two measures based upon ParlGov data (Döring and Manow, 2015) to capture this dimension of party strength, National Party Seat Share and National Party in Government. The former is the per cent of seats held in the lower house of the national legislature by the national party. The latter is a dichotomous indicator of national political parties serving in national government at the time of the May 2014 elections.

Finally, H5 posits that MEPs from national parties that are more likely to have a demand for social media usage will be more likely to use Twitter as a mainstay in their campaigning strategies. To account for this possibility, we draw on data from the 2014 European Election Study (Popa et al., 2015; Schmitt et al., 2015) to measure the percentage of national party voters who responded that they sought information on 2014 EP elections either ‘sometimes’ or ‘often’ using the internet or social media. We use this National Party Voter Internet Affinity variable alongside the mean National Party Voter Age to proxy for parties that face greater incentives to incorporate Twitter into their campaigns.

A host of control variables are also included in the analysis. Naturally, the decision of MEPs to use Twitter during the campaign may also be predicated upon broad national differences in the usage of online social networks. To account for these differences, we use data from the autumn 2013 Eurobarometer survey (European Commission, 2014) to construct Social Network Usage, which is the percentage of people in a given country that use online social networks at least once a week. To account for the possibility that MEPs from more extreme national parties may use Twitter differently than their mainstream counterparts, we use expert survey data from the 2014 UNC Chapel Hill Expert Survey (Bakker et al., 2015) to construct three measures of MEPs’ national party extremeness: EU Integration, Left-Right, and GAL-TAN. Each variable measures the national party’s squared distance from the midpoint of the scales, with higher values signifying more extreme party positions on the relevant aforementioned category.

Finally, the difference of an MEP’s number of Twitter Followers between time periods is captured for each of the periods discussed above and is lagged by one observation period (i.e. April’s followers may impact May’s tweets), in order to account for the possibility that MEPs with more social media followers may choose to interact with their voting base more in the application. MEP Age and Seniority (measured as the number of terms served in the EP) may also impact their decision to use new forms of technology in the campaign. MEPs that have a higher profile within the legislature, serving as an EP Leader (any internally elected position, such as EP President, Vice President, Quaestor, and party group leadership positions) or a Committee Leader (Chair or Vice Chair) may also behave differently during the campaign. Female is used as a dichotomous control for MEP gender. Lastly, EP party group dummies are also used, to account for potential variation in usage by group.⁸

Modelling choices

Since the distribution of the dependent variable measuring the number of tweets emitted by MEPs during the campaign period is skewed to the right and marked by a variance that is larger than its mean, a negative binomial estimation of the models in Table 1 would technically be appropriate. However, given the high frequency of tweets, the substantive and statistical significance remains consistent when using a linear model, which allows for direct interpretation by the reader. We therefore use linear multi-level mixed effects models with random country intercepts.

Table 1.

Frequency of MEP Tweets during and after 2014 Elections (March–July 2014).

	(1)	(2)	(3)	(4)	(5)
DV: # of Tweets	Baseline	t-2 Months	t-1 Month	t-1 Week	t+1 Month
IVs of interest:
H1. Citizens Represented	62.497**	3.758***	6.788***	2.663***	0.907
	(24.11)	(0.96)	(1.51)	(0.41)	(1.17)
H2. Preferential Vote	−1512.570	7.062	7.178	−49.842*	−52.650
	(1577.43)	(52.26)	(82.93)	(22.21)	(101.13)
H2. Avg. District Magnitude	4.418	−1.855***	−3.353**	−1.352***	−0.863
	(16.01)	(0.55)	(1.02)	(0.39)	(0.88)
H2. Pref. Vote X Avg.	123.186*	0.672	2.828	3.239*	1.033
Dist. Mag.	(65.38)	(2.13)	(3.55)	(1.33)	(2.33)
H3. MEP List Safety	−194.369	−16.755*	−28.908*	−3.620	−6.953*
	(141.36)	(10.09)	(14.37)	(7.11)	(4.17)
H4. National Party	14.726	−0.968	−0.728	−0.229	0.746
Seat Share	(14.78)	(1.36)	(2.22)	(0.56)	(0.69)
H4. National Party in Gov’t	−1156.810*	9.468	14.422	−14.286	−39.538**
	(596.04)	(38.76)	(55.34)	(16.42)	(13.35)
H5. Nat. Party Voter	−4432.368	58.993	55.297	−146.363	−14.885
Internet Affinity	(5577.75)	(267.08)	(444.31)	(138.90)	(175.03)
H5. National Party	−124.078**	−1.683	−4.619	−2.119	−4.007*
Voter Age	(41.90)	(2.99)	(4.93)	(1.39)	(1.94)
Additional controls:
Member State Social	111.879**	4.703**	7.842**	3.638***	0.546
Network Usage	(41.15)	(1.50)	(2.53)	(0.85)	(1.42)
National Party Extreme	−180.782	−12.319**	−14.602*	−7.103*	−4.551
(Left–Right)	(110.98)	(4.43)	(7.98)	(3.24)	(2.96)
National Party Extreme	184.287*	7.596**	10.291*	5.322**	3.007*
(GAL-TAN)	(85.32)	(2.85)	(4.44)	(1.91)	(1.22)
National Party	204.895	13.196	15.740	0.619	−0.999
Extreme (Integration)	(188.67)	(8.69)	(15.11)	(3.63)	(4.63)
MEP Twitter	0.031**	0.000	0.028***	0.013*	0.003
Followers (lagged)	(0.01)	(0.00)	(0.01)	(0.01)	(0.00)
MEP seniority	−97.277	−21.032*	−31.967*	−10.617	−1.352
	(147.76)	(10.15)	(18.31)	(6.60)	(4.10)
MEP age	−100.537***	−4.224*	−7.236*	−2.819***	−1.901*
	(21.96)	(2.22)	(3.46)	(0.82)	(0.89)
Female	61.553	55.534	84.296	12.004	11.386
	(494.45)	(37.33)	(59.93)	(18.88)	(14.66)
EP leader	217.321	−0.319	7.475	−8.798	−9.497
	(711.74)	(31.39)	(55.77)	(17.38)	(24.38)
Committee leader	−215.588	−19.337	−28.844	−22.870	−26.864
	(520.96)	(35.90)	(65.84)	(22.93)	(25.76)
Party group fixed effects	Yes	Yes	Yes	Yes	Yes
Constant	6555.249	−107.124	−104.455	181.380	344.153
	(6112.14)	(363.90)	(640.97)	(192.13)	(234.78)
Intercept variance	870.123	0	0	0	47.036
Residual variance	3425.731	245.823	400.637	135.139	113.749
Intraclass correlation	0.061	0	0	0	0.146
N (country groups)	245 (28)	245 (28)	245 (28)	245 (28)	245 (28)

Note: R.S.E. in parentheses; robust clusters by country delegation; t denotes time of EP election; authors’ own calculations.

Full table including party group fixed effects in the online appendix.

Results and analysis

Table 1 displays linear regression coefficients for a series of multi-level mixed effects models. Robust standard errors are reported in parentheses below each beta coefficient. Although the mixed model considers the possibility for random country-level effects to impact MEP twitter usage, the fixed effect coefficient values in the table can be directly interpreted, which is a major strength of the linear model. Each of the five models uses a version of the Tweets variable, as described in the discussion of the dependent variables.

As relates to our hypotheses, we find that the number of an MEP’s Citizens Represented is shown to be a significant predictor of tweeting – although only up until the election (Models 1–4) – with each additional 10,000 constituents represented corresponding with a 6.79 tweet bump during the month preceding the elections. This confirms H1. With respect to H3, we find that list leader MEPs actually tweeted about 50.28 (3 × 16.76) fewer tweets than the ‘hopeless’ MEPs during the early campaign model (Model 2), once multivariate regression controls are introduced. This relationship contradicts H3 and is also confirmed in Model 3, where list leaders tweeted an average of 86.73 fewer tweets than ‘hopeless’ MEPs in the final month of the campaign. Interestingly enough, the significant effect of the list safety is not present during the final week of the campaign. This may be due to the fact that all politicians involved in the campaign are fighting for attention at insignificantly differing levels.

One area that is less obvious to directly analyse relates to H2a and H2b. As each of these hypotheses is conditional upon the existence of a preferential voting system in the MEP’s home country, the Preferential Vote dummy variable is interacted with the Average District Magnitude measure. While the independent effect of preferential voting is not confirmed, except for in Model 4, we do find that the interaction of preferential voting systems with district magnitude is significant at various points during the campaign (Models 2–4). In the absence of a preferential vote (i.e. the variable is set to zero), we see a negative effect of increased district magnitude levels during the campaign (see only the Average District Magnitude variable in Models 2–4), which would seem to offer support to H2a. Graphical interpretation can also be used to more directly interpret the conditional nature of preference voting and its mitigating effect on district magnitude.

Figure 1 uses known values of district magnitude to predict the volume of tweets during the final week of the campaign (Model 4), in both the presence and absence of a preferential voting system. Here, the relationship posited by H2a and H2b is much more apparent. In systems with a preferential vote, there is a sharp increase in the number of tweets: as the district magnitude increases, MEPs from the largest districts in a preferential voting system (e.g. The Netherlands) were shown to tweet more than twice as often as MEPs from the smallest districts (e.g. Northern Ireland, Malta, and Luxembourg). Conversely, we do observe a steady decline in Twitter usage, as district magnitude increases in these strictly closed list systems. It seems that, to the extent that Twitter is used by MEPs to make personal and direct appeals to voters, this is more likely to occur in more open and preference-based systems.

Figure 1.

The conditional effect of district magnitude on tweeting. Week prior to May 2014 elections.

With respect to H4, we find no support for the hypothesis that MEPs from larger parties at the national level tweet more often in EP elections. However, we do find that MEPs whose national party is in government actually tweet less during the periods preceding and following the European election campaign (Models 1 and 5). This interesting result suggests that perhaps Twitter plays a special role in ‘opposition’ politics during the normal legislative calendar.

H5 examines the ‘demand’ for Twitter usage in the campaign. Here again, we find mixed results. Whereas parties with generally older voters do use Twitter less often outside of election time (Models 1 and 5), the effect disappears during the election period. The extent to which party voters use the internet to seek out campaign information (National Party Voter Internet Affinity) has no independent effect on the demand for additional tweets from MEPs. However, our control variables suggest that an increase in the degree to which an MEP’s country of origin is permeated by Social Network Usage positively affects Twitter activity. Likewise, the individual MEP’s number of Followers influences social media usage at some points.

The additional control variables in the model offer further insights into the Twitter usage of MEPs during the campaign period. In particular, more extremist MEPs on the traditional left–right economic scale are shown to use Twitter less often during the campaign, whereas MEPs that are extreme on the GAL-TAN dimension – which refers to the progressive-traditional dimension of politics – were more likely to use Twitter. At the individual level, both seniority and MEP age have negative effects on Twitter usage.

Finally, it is also worth reiterating from the results in Table 1 that the passage of time clearly has an impact on the behaviour of MEPs during the campaign season. We might conclude from the baseline model that Twitter activity in the EP is relatively idiosyncratic to the individual member – a reasonable conclusion, given people’s very different personal attitudes towards social media usage. However, Models 2–4 clearly indicate that Twitter becomes an active component of many MEP re-election campaigns. So active, in fact, that a dummy variable for MEPs seeking re-election is dropped during the estimation – simply put, the only reason to be active on Twitter as an MEP, during the election, is because of the election itself! Lastly, we include Model 5 to demonstrate that almost all systematic effects on MEP Twitter usage disappear entirely, once the campaign concludes. There is clear evidence from the collective models to suggest that Twitter has not only taken hold of the EP, but that it has done so for a very specific reason: campaigning.

Conclusions

This article has demonstrated that re-election campaigning using new social media can still be interpreted by standard theories of comparative political institutions. Using a novel dataset of MEP activity on Twitter, collected before, during, and after the 2014 elections, we show that there are systematic effects present that both predict an MEP’s decision to use Twitter in the first place, as well as the degree to which they use Twitter during the election period. Thereby, we are able to trace the evolution of traditional modes of campaigning into the present.

Our findings demonstrate that MEPs representing more citizens are more likely to use Twitter and to use it more frequently during the campaign. This means that MEPs make constructive use of Twitter when it might otherwise be particularly resource intensive in terms of time and money to reach voters. These findings a priori suggest that social media might help to invigorate the weak electoral connection in the EU. Future research should investigate how representatives make use of the interactive features of social media to foster an electoral connection.

EU member states use different variants of proportional representation for European elections (Farrell and Scully, 2005). One major point of differentiation is the presence of preferential voting systems. In keeping with the classic expectations of the comparative politics literature, we find that MEPs from preferential voting systems are far more likely to be active on Twitter when they hail from districts of a larger magnitude. The reverse is found to be true of systems in which there is no voting preference allowed: as district magnitude increases, Twitter usage decreases. These findings are consistent with H2a and H2b. Future research should analyse the content of tweets, which would be particularly desirable in order to further test the extent to which candidates use Twitter to seek a personal rather than a party vote. Moreover, the uniform electoral procedure of the EU prevents us from investigating the effect of majoritarian electoral systems, in which incentives for seeking a personal vote also vary (Carey and Shugart, 1995). It would be intriguing to study how these fit in with our findings.

In addition, we observe MEPs who are in jeopardy of losing their seats are most active on Twitter. We show that the ‘safely seated’ party leadership is less active on social media, compared with those in list positions classified as ‘competitive’ or even ‘hopeless’, based upon ballot structure and polling data. While this finding runs opposite to the originally expected direction, it demonstrates the potential usefulness of Twitter as an innovative campaigning tool. Particularly for candidates that have lost the support of their party or their voters and have been poorly placed on electoral lists, Twitter may serve as a remedy for under-funded and under-supported traditional campaign activities.

While we might expect major, established parties with a strong base in national parliaments to lead more multi-faceted, professional campaigns including social media, we do not find support for this. Our only related finding is a higher level of social media activity for national opposition parties outside the heat of the campaign. Unlike the incumbents that we study, challengers from the same national parties might be younger and generally more willing to embrace new campaign tools such as social media, so one avenue for future research would be to test whether these findings hold for all candidates.

Finally, we investigated the extent to which social media use by representatives may be demand driven, given the internet affinity of voters for the national party. We find that MEPs with younger national party voters are more active on Twitter outside of the campaign season, and that Twitter activity depends on general country-level online social network usage rather than an MEPs’ national party voters’ internet usage.

In sum, the findings suggest that MEPs value social media to engage with voters. The decision to use Twitter is an innovative and essentially costless way to remotely spread ones message to voters. However, it is important to note that many of the effects described are structured by the election cycle. This reminds us not to overstate the importance of social media, in particular in absence of evidence to its effectiveness in the quest for re-election. Twitter activity seems to be targeted at vote seeking during the campaign, rather than constituting a sustained attempt to create an online ‘electoral connection’ (Vergeer et al., 2013). It would be interesting in future research to gain a better understanding of MEPs’ attempts to use social media to create a long-term electoral connection in absence of an imminent election.

Footnotes

Acknowledgements

The authors are grateful for the helpful comments of participants at the Fifth Annual Conference of the European Political Science Association in Vienna, the Second Annual PADEMIA conference and the anonymous reviewers. They would also like to thank Simon Hix for sharing comprehensive PollWatch data (Cunningham and Hix, 2014).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Bakker R, Edwards E, Hooghe L, et al. (2015) 2014 Chapel Hill Expert Survey. Chapel Hill: University of North Carolina, Chapel Hill. Available at: www.chesdata.eu (accessed 1 October 2015).

Bowler

Farrell

(1993) Legislator shirking and voter monitoring: Impacts of European parliament electoral systems upon legislator-voter relationships. Journal of Common Market Studies 31(1): 45–69.

Bowler

Farrell

(2011) Electoral institutions and campaigning in comparative perspective: Electioneering in European Parliament elections. European Journal of Political Research 50(5): 668–688.

Brooker P, Barnett J, Cribbin T, et al. (2015) Have we even solved the first ‘big data challenge?’ Practical issues concerning data collection and visual representation for social media analytics. In: Snee H, Hine C, Morey Y, et al. (eds) Digital Methods for Social Science: An Interdisciplinary Guide to Research Innovation. Houndmills, Basingstoke: Palgrave Macmillan, pp.32–50.

Carey

Shugart

(1995) Incentives to cultivate a personal vote: A rank ordering of electoral formulas. Electoral Studies 14(4): 417–439.

Chorus Analytics (2014) Chorus Tweetcatcher Desktop. Available at: http://chorusanalytics.co.uk (accessed 5 March 2015).

Corbett

Jacobs

Shackleton

(2011) The European Parliament, 8th ed. London: John Harper Publishing.

Cunningham K and Hix S (2014) PollWatch.eu Project, March–May 2014. Unpublished dataset.

Daniel

(2015) Career Behaviour and the European Parliament: All Roads Lead Through Brussels?, Oxford: Oxford University Press.

10.

Döring H and Manow P (2015) Parliaments and governments database (ParlGov): Information on parties, elections and cabinets in modern democracies. Development version. Available at: http://www.parlgov.org (accessed 25 September 2015).

11.

European Commission (2014) Eurobarometer 80.1 (2013) ZA5876 Data file Version 1.0.0, Cologne: GESIS Data Archive.

12.

European Parliament/TNS Opinion (2014) European elections 2014: An analytical overview. Available at: http://www.europarl.europa.eu/pdf/eurobarometre/2014/post/post_2014_survey_analitical_overview_en.pdf (accessed 15 June 2015).

13.

Farrell

Scully

(2005) Electing the European Parliament: How Uniform are ‘Uniform’ Electoral Systems? Journal of Common Market Studies 43(5): 969–984.

14.

Farrell

Scully

(2010) The European Parliament: One parliament, several modes of political representation on the ground? Journal of European Public Policy 17(1): 36–54.

15.

Farrell

(2012) The consequences of the internet for politics. Annual Review of Political Science 15(1): 35–52.

16.

Gattermann

Vasilopoulou

(2015) Absent yet popular? Explaining news visibility of Members of the European Parliament. European Journal of Political Research 54(1): 121–140.

17.

Gervais

(2015) Incivility online: Affective and behavioral reactions to uncivil political posts in a web-based experiment. Journal of Information Technology and Politics 12(2): 167–185.

18.

Gibson

Römmele

(2009) Measuring the professionalization of political campaigning. Party Politics 15(3): 265–293.

19.

Giebler

Wüst

(2011) Campaigning on an upper level? Individual campaigning in the 2009 European Parliament elections in its determinants. Electoral Studies 30(1): 53–66.

20.

Golbeck

Grimes

Rogers

(2010) Twitter use by the U.S Congress. Journal of the American Society for Information Science and Technology 61(8): 1612–1621.

21.

Graham

Broersma

Hazelhoff

(2013) Between broadcasting political messages and interacting with voters. Information, Communication and Society 16(5): 692–716.

22.

Hix

Hagemann

(2009) Could changing the electoral rules fix European Parliament elections? Politique Europénne 28: 28–42.

23.

Hix

Marsh

(2007) Punishment or protest? Understanding European Parliament elections. The Journal of Politics 69(2): 495–510.

24.

Hix

Marsh

(2011) Second-order effects plus pan-European political swings: An analysis of European Parliament elections across time. Electoral Studies 30(1): 4–15.

25.

Hobolt

(2014) A vote for the president? The role of Spitzenkandidaten in the 2014 European Parliament elections. Journal of European Public Policy 21(10): 1528–1540.

26.

Hobolt

Høyland

(2011) Selection and sanctioning in European Parliamentary Elections. British Journal of Political Science 41(3): 477–498.

27.

Hobolt

Spoon

(2012) Motivating the European voter: Parties, issues and campaigns in European Parliament elections. European Journal of Political Research 51(6): 701–727.

28.

Jackson

Lilleker

(2011) Microblogging, constituency service and impression management: UK MPs and the use of Twitter. The Journal of Legislative Studies 17(1): 86–105.

29.

Kruikemeier

(2014) How political candidates use Twitter and the impact on votes. Computers in Human Behavior 34: 131–139.

30.

Kruikemeier

van Noort

Vliegenthart

(2013) Getting closer: The effects of personalized and interactive online political communication. European Journal of Communication 28(1): 53–66.

31.

Lassen

Brown

(2011) Twitter: The electoral connection? Social Science Computer Review 29(4): 419–436.

32.

Lilleker

Koc-Michalska

(2013) Online political communication strategies: MEPs, E-representation, and self-representation. Journal of Information Technology and Politics 10(2): 190–207.

33.

Lilleker

Koc-Michalska

Schweitzer

(2011) Informing, engaging, mobilizing or interacting: Searching for a European model of web campaigning. European Journal of Communication 26(3): 195–213.

34.

Lorenzo Rodríguez

Garmendia Madariaga

(2016) Going public against institutional constraints? Analyzing the online presence intensity of 2014 European Parliament election candidates. European Union Politics. 17(2): 303-323..

35.

Mayhew

(1974) Congress: The Electoral Connection, New Haven, CT: Yale University Press.

36.

Papacharissi

(2004) Democracy online: Civility, politeness, and the democratic potential of online political discussion groups. New Media and Society 6(2): 259–283.

37.

Peñalver García

Priestley

(2015) The Making of a European President, Houndmills, Basingstoke: Palgrave Macmillan.

38.

Popa

Schmitt

Hobolt

(2015) EES 2014 Voter Study Advance Release Notes, Mannheim: MZES, University of Mannheim.

39.

Reif

Schmitt

(1980) Nine second-order national elections – A conceptual framework for the analysis of European election results. European Journal of Political Research 8(1): 3–44.

40.

Schmitt

Hobolt

Popa

(2015) Does personalization increase turnout? Spitzenkandidaten in the 2014 European Parliament elections. European Union Politics 16(3): 347–368.

41.

Schmitt

Popa

Hobolt

(2015) European Parliament Election Study 2014. ZA5160, Cologne: GESIS Data Archive.

42.

Scully

Farrell

(2003) MEPs as representatives: Individual and institutional roles. Journal of Common Market Studies 41(2): 269–288.

43.

Small

(2008) Equal access, unequal success major and minor Canadian parties on the net. Party Politics 14(1): 51–70.

44.

Van der Eijk

Franklin

(1996) Choosing Europe? The European Electorate and National Politics in the Face of Union, Ann Arbor, MI: University of Michigan Press.

45.

Vergeer

Hermans

Cunha

(2013) Web campaigning in the 2009 European Parliament elections: A cross-national comparative analysis. New Media and Society 15(1): 148–148.

46.

Vergeer

Hermans

Sams

(2013) Online social networks and micro-blogging in political campaigning: The exploration of a new campaign tool and a new campaign style. Party Politics 19(3): 477–501.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.04 MB

0.00 MB

0.48 MB