The EU Parliament on Twitter—Assessing the Permanent Online Practices of Parliamentarians

ABSTRACT

Although conceptual efforts have often suggested that the Internet harbors considerable possibilities to revolutionize political participation, empirical studies have often presented rather limited impacts in this regard. Nevertheless, novel online services such as Twitter are still pointed to as having potential to be employed by citizens and politicians alike. Utilizing state-of-the-art data collection methods, this study builds on the suggestions of previous research and gauges the degree to which EU parliamentarians make use of Twitter for so-called permanent campaigning. Specifically, the paper seeks to assess the degree to which Twitter use by European Parliament representatives can be described as being characterized by permanence—a concept related to the professionalization of political campaigns. Thus, by examining these uses outside of election periods, the study provides useful insights into the day-to-day uses of Twitter, contributing to the limited body of work focusing on the everyday online practices of politicians.

KEYWORDS:

• EU Parliament
• permanent campaign
• political communication
• professionalization
• social media
• Twitter

INTRODUCTION

Although the notion of a crisis for democracy is difficult to discuss in more specific terms, a multitude of scholars, pundits, and public commentators have nonetheless traced such tendencies—such as declining parliamentary-political engagement and lower voting attendance among the citizenry in several Western countries (e.g., Hermans & Vergeer, 2012; Lilleker & Malagón, 2010). It would seem that such claims are particularly valid, then, among younger citizens whose countries are part of the European Union (EU). Perhaps due to what has been described as high levels of democratic deficit within the supranational organization (e.g., van Os, Jankowski, & Vergeer, 2007), interest as well as trust in the EU among its younger citizens are often reported at low levels (Schweitzer, 2009; Vergeer, Hermans, & Sams, 2011a). Such disinterest becomes especially alarming when one considers the increasing levels of influence that the institution at hand has over the everyday lives of most Europeans.

Roughly corresponding to these developments toward parliamentary disillusionment, the introduction and expansion of the Internet has, much like previously novel technologies, been pointed to as harboring possible remedies for these tendencies by allowing “a Europeanized public dialogue to emerge” (Schweitzer, 2009, p. 21), where citizens and their elected officials would be able to connect and discuss (e.g., Anduiza, Cantijoch, & Gallego, 2009; Bentivegna, 2006). Although the bulk of empirical research has largely disproven such arguably overtly optimistic claims, voices are still lauding the potential of the Internet, supposedly as it has moved from a 1.0 paradigm—in this context largely denoting “Burkean, Top-Down” (Jackson & Lilleker, 2009, p. 246) dissemination of information from the politician to the citizen—to a 2.0 rationale for Web publishing, suggesting that the interactive elements of the Web be given a more important role, often through a variety of so-called social media applications (e.g., Jackson & Lilleker, 2009; Vergeer & Hermans, 2013). In relation to this, although the notion of the permanent campaign, suggesting blurred lines between periods of campaigning and governing, was first coined in the 1970s (e.g., Blumenthal, 1980; Ornstein & Mann, 2000), the “always-on” logic of social media has led to suggestions that such continuous endeavors by politicians might flourish as a result of the technical developments briefly described above (e.g., Klinger, 2013; Tenscher, 2013).

The empirical focus of the present study is placed on the social media platform Twitter. Arguably one of the currently most-discussed online services, Twitter allows its users to send short updates of up to 140 characters each—tweets—to a network of nonreciprocal followers. Although use of the service is far from universal, and differs considerably between different parts of the world—for example, between the United States and many European countries (e.g., Beevolve, 2012)—services such as Twitter are nevertheless important to study, primarily because of the many high expectations held regarding its supposed impact. Twitter has been studied in a variety of contexts, perhaps most famously during uprisings in totalitarian states (e.g., Ameripour, Nicholson, & Newman, 2010). Our interests here, however, are directed toward the uses of Twitter at the hands of political actors elsewhere. Adopting an exploratory approach, the study provides an overarching, structural assessment of the everyday uses of Twitter at the hands of EU parliamentarians, posing the following research question: to what degree can the Twitter use of European Parliament representatives be described as being characterized by permanence? In short, the notion of permanence in this regard is often understood as campaign-like activities undertaken by politicians outside of election seasons. Specifically, the study employs a series of statistical analyses to gauge the influences of individual (such as age or gender) and contextual (such as country or party characteristics) variables on such ongoing Twitter use.

Indeed, because most research on the online efforts of politicians has focused on election campaigns, the emphasis on “daily political communication practices” (Tenscher, 2013, p. 243) championed here provides a useful contrast to the bulk of work performed (see also Schweitzer, 2009, p. 38). Moreover, although previous scholars have employed dichotomized approaches—testing for adoption or nonadoption of various online services by politicians (e.g., Hermans & Vergeer, 2012; Larsson, 2011; Schweitzer, 2011), this study utilizes continuous data regarding Twitter use in order to provide more detailed insights.

By gauging Twitter practices among parliamentary representatives from across the EU, the study presented here makes a clear contribution to the apparent dearth of comparative work across different countries (Bruns & Stieglitz, 2012; Tenscher, 2013). From a conceptual standpoint, the paper is informed by the previously mentioned notion of permanent campaigning, which broadly suggests that activities formerly primarily related to election periods are taking place outside of such periods of presumed heightened interest as well. Given the empirical material used, our focus is placed on how such permanence, often thought of as an indicator of the professionalization of politics, plays out in the online setting described above. Although it might be difficult to sustain discussions on complex pan-European political issues given Twitter’s limitations regarding message length, previous studies have suggested that activity on the platform can lead to a view of politicians as being more approachable (Vergeer et al., 2011a), something that could perhaps be particularly valuable when considering outreach attempts to younger citizens (e.g., Porten-Cheé, 2013). With this in mind, the subsequent section details the conceptual design of the study at hand.

ONLINE PERMANENCE AS AN EXAMPLE OF POLITICAL PROFESSIONALIZATION

If citizens are often suggested to be “disenchanted with … dated political systems” (Christensen & Bengtsson, 2011, p. 2), the undertakings of their elected officials are often thought of as undergoing professionalization in various regards. Although the term has become something of a “catchall phrase” (Lilleker & Negrine, 2002, p. 99), we can nevertheless distinguish a few tendencies that are often associated with the term at hand, tendencies that are of specific interest with regard to the topic of the present study. As such, while other texts deal with this terminology in a broader manner, suggesting several dimensions (e.g., Gibson & Römmele, 2001; Lisi, 2013; Strömbäck, 2007; Tenscher, 2013), our focus here is placed on two interrelated concepts often thought of as valuable parts of professionalized political activity: uses of communication technology, specifically in combination with so-called permanent campaigning.

For professionalization, suggestions that novel media services—be they 1.0 or 2.0—would hold importance for the professional procurance of political campaigns have been aired since the mid-1990s (e.g., Farrell, 1996), showing little signs of diminishing over time. In large part due to the apparent social media prowess of the 2008 Obama campaign (e.g., Hargittai & Shaw, 2013; Kalnes, 2009; Tumasjan, Sprenger, Sandner, & Welpe, 2010), there is considerable hype also in other contexts regarding the political possibilities of online tools such as Twitter (e.g. Lisi, 2013; Vergeer & Hermans, 2013). Although the influences of older media such as radio or TV are still felt in most political contexts, the need for politicians to assess and master new media forms has been suggested to be of the utmost importance (e.g., Druckman, Kifer, & Parkin, 2007). Consider how previous political leaders have grasped the technological novelties of their time—the role of the radio for Roosevelt’s presidency, or similarly, of Kennedy’s apparent understanding of the television medium (Vergeer, 2012). It would appear, then—at least from popular hype—that the Internet has a similar part to play in today’s political environment.

However, given the previously mentioned “always-on” rationale of social media such as Twitter, the degree to which the Internet is employed by politicians outside of election periods is of specific interest. As shown by research looking into politicians online, such activity tends to be rather limited (e.g., Gibson, 2004; Graham, Broersma, Hazelhoff, & van ′t Haar, 2013; Larsson, 2011; Vergeer et al., 2011b). As suggested by Vaccari (2008b), there is a need for political actors to “mobilize resources not only during campaigns, but also outside of them” (p. 6). Such mobilization, then, is often referred to as permanent campaigning.

Although it is difficult to suggest precise measurements regarding exactly what a permanent campaign would entail in an operationalized online setting, the explorative approach employed here aims to provide a way forward by presenting empirical data detailing how these practices play out—data that could ideally serve as starting points for future comparative studies. With this in mind, the literature suggests that the concept of permanent campaigning has been a discussion point for some time. Blumenthal (1980) traces its origins to the 1976 U.S. presidential campaign, when an advisor to president-elect Carter suggested the necessity to provide campaign-like efforts also outside of election periods. Although such practices were employed to varied extents in the 1970s and onward, the online era has supposedly refreshed the interest among political actors to apply a more continuous mode of campaigning. The potential of the Internet to usher in more efforts of permanence has been touched upon by several researchers. For example, Vergeer and co-authors (2011b) suggest that “with the advent of the Internet, permanent campaigning … becomes easier” (p. 485), while Vaccari (2008b) similarly claims that online activities by political actors are “maintained for extended periods and long-term objectives” (p. 6). Indeed, suggestions that such new platforms for communication would allow politicians to interact directly with their respective constituents without interference from the media, and that those types of connecting activities would hold strong also outside of election season is an important part of the discourse regarding these issues (e.g., Calenda & Meijer, 2009; Chadwick, 2003; Larsson, 2013).

In sum, the study follows previous scholars in viewing online activity and permanence in campaigning as substantial, yet not isolated, subsets of professionalized politics (Gibson & Römmele, 2001; Strömbäck, 2007). By assessing these two perspectives in tandem in an empirical setting, the study provides useful insights into how Twitter is being put into play, and what factors seem to influence more continuous activity. Given that much of the research on topics such as these has primarily dealt with UK and U.S. contexts (e.g., Lilleker & Malagón, 2010; Needham, 2005) the study presents a different focus: the European Union.

THE EU ONLINE

The European Union is a political and economic union currently comprised of 27 member states. Tracing back to the European Coal and Steel Community, formed in the 1950s, the EU of today operates through a series of independent, supranational institutions, one of which is under specific scrutiny here. As the name implies, the European Parliament (EP) functions as the directly elected parliamentary institution of the EU. Although interest in partaking in EP elections is generally low, perhaps related to widespread “euroskepticism” among the citizenry (Lubbers & Scheepers, 2010), the same musings regarding the role of the Internet and its supposed democratic role as discussed previously are heard in relation to EP elections. Even though citizens may be skeptical, the EP has matured into an increasingly important actor regarding a number of policy issues, most of which have real, tangible outcomes on the national level, affecting the everyday lives of Europeans. As such, adopting permanent campaigning tactics could be a suitable way for EP representatives to make themselves heard—and possibly viewed in a better light—among the citizenry.

A series of studies have engaged with this group of politicians and their uses of online services, more often than not in relation to EP elections. Although our focus here is not on elections, studies employing such emphasis will serve as examples of preceding scholarship regarding topics relevant for our current purposes.

Starting with the EP elections in 2004, Schweitzer (2009) studied actor and issue salience of uniquely European perspectives on German political Web sites. As such, her efforts mostly dealt with specific aspects of the content and outgoing links to sources dealing with European issues provided on such sites, finding a “near absence of … the European Union” (2009, p. 37) in her material. Of specific interest to our current purposes is the result that features inviting interaction with politicians, such as online chat (found on 9% of the sites) or Weblogs (found on 4% of sites), were employed to limited extents, indicating a predominant one-way communicative mind-set from the side of the representatives. Beyond the 2004 events, a series of scholarly efforts report on various aspects of the subsequent 2009 elections. Indeed, while the index of political professionalization favored by Tenscher (2013) deals only partly with the topics specifically under study here, the results from his study on German national and EP elections during that year provide useful insights. Specifically, although employment of the aforementioned index largely found that “the 2009 EP campaign structures were more professional than in 2005” (Tenscher, p. 248), indicating an increase in online engagement at the hands of the politicians, the suggestion is made by Tenscher that this result could in large part be due to a spillover effect of sorts, given that national elections were held the same year. Nevertheless, this could be seen as an indication that novel practices, such as online narrowcasting or campaigning through Weblogs, were being employed at a larger scale than during the previous election. Taking the influence of Twitter into account, Vergeer et al. (2011a) focused on the uses of the service at the hands of Dutch politicians during that same election. Testing for influences on Twitter use as well as the vote-gaining possibilities of the platform, the authors found that candidates from parties in opposition were early adopters of Twitter and tended to use it more frequently, a result somewhat similar to the “underdog effect” found in other contexts (e.g., Larsson and Kalsnes, 2014). Although Vergeer and co-authors rightfully point out that the limited adoption rate of Twitter should be kept in mind when interpreting the results regarding votes gained from tweeting activity, their findings did indeed indicate a positive effect in this regard, suggesting that online activity such as this can pay off at the polls.

Moving beyond single-country case studies, another effort by Vergeer, Hermans, and Cunha (2012) presents a cross-national comparative analysis of 1,026 party or candidate Web sites from 17 countries participating in the 2009 EP elections. Overall, the results indicated that although parties and representatives from Sweden, Ireland, and Belgium made more extensive use of their Web sites, their colleagues from Portugal, the Netherlands, and Greece showed less prowess in utilizing the full potential of the Web site as a platform for communicating with potential voters. Although some geographic patterns regarding such activity could be noted, the authors argue that support for a “North–South divide across Europe for web campaigning” (Vergeer et al., 2012, p. 142), where the latter end would be characterized as significantly less active in this regard, cannot be maintained as evident in their data.

Dealing with that same election, Lilleker et al. (2011) likewise made use of Web site–feature analysis, albeit with a different sampling procedure: placing their focus on all parties from France, Germany, Great Britain, and Poland that competed for places in the parliament. Although their findings indicated that the dominant mode of campaigning mostly consisted of a more traditional variety of one-way communication, parties had indeed started to embrace some of the more interactive features often associated with the Web 2.0 dictum, mirroring results from other contexts suggesting that a “Web 1.5” rationale to Web campaigning is being employed, adding more interactive features incrementally and adapting those features to established campaign rationales (e.g., Jackson and Lilleker, 2009). More conservative approaches to interactive features were found among smaller parties, especially in Poland, while those parties who tended to utilize more of such features were found to be major and emanating from “nations with longer traditions of democratic engagement” (Lilleker et al., 2011, p. 206).

As such, although we can perhaps discern tendencies toward a ”Western European Style of Campaigning” (Tenscher, 2013, p. 250) from reviewing the results of previous studies, it would seem that differences regarding professionalization in campaigning do not necessarily respect nation borders. Indeed, in his study of campaigning practices in Portugal, Lisi (2013) suggests that “new democracies present particularly favourable conditions for the professionalization of political campaigns” (Lisi, 2013, p. 260), somewhat contradicting the results from the previously discussed comparative studies. As such, although we might find national differences among the EPs studied here, nationality alone cannot be expected to fully explain variations pertaining to permanence as defined here. Issues pertaining to digital divides among the diverse citizenries across Europe might be at play here, but previous scholarship also suggests other possible factors for clarification. Thus, the next section presents the steps taken to bring other potentially explanatory variables into the fold, also providing details regarding data collection.

METHOD

The study made use of both firsthand and secondhand data. First, in order to construct a comprehensive list of EU parliamentarians on Twitter, the official home page of each politician in Parliament as of May 2013 was visited and checked for links to Twitter accounts. The presence of a Twitter link was taken as a sign of official sponsorship or even ownership of the account linked to (following Kalnes, 2009, p. 255), allowing us to effectively exclude “Twitter Fakers” (e.g. Wilson, 2011) from our prospective sample. Following the work performed by Vergeer and Hermans (2013) and Vergeer et al. (2011a), each identified Twitter handle was queried against the Twitter Application Programming Interface (henceforth API) in order to download publicly available account activity data. The interface at hand can be employed to gather data from the platform in a comparably straightforward fashion. The available data can revolve around specificities regarding the tweets themselves, or deal specifically with the user accounts (Giglietto, Rossi, & Bennato, 2012). For our current purposes, the latter of these options was especially relevant. Specifically, the GET users/lookup command was employed to gather data from the Search API, yielding results similar to those one might get by performing a Twitter search. The process itself is described in detail at https://dev.twitter.com/docs/api/1.1/get/users/lookup (accessed on December 17, 2013).

Specifically, the following information was noted for each identified parliamentarian: the date that the Twitter account was registered; the number of tweets sent since that date; the number of friends (i.e., the number of other Twitter users followed by each politician), and the number of followers (the number of users following the politician).

This gathering of firsthand data was performed on May 5th, 2013, and as such, the analyses performed and results presented must be understood as snapshots of supposedly continuously ongoing activities at the hands of politicians (e.g., Brugger, 2012). Although “freezing the flow” (Karlsson & Strömbäck, 2010) of variables such as these is associated with some methodological difficulties, the selected approach will nevertheless allow for important insights into the drivers of Twitter use by EP representatives.

Suitable independent variables for the study were identified by consulting the literature regarding adoption and continued use of ICTs at the hands of politicians. The variables included, and the rationales for including them, are discussed in the subsequent sections.

Candidate Variables

The initial group of variables employed deal specifically with the individual member of Parliament. First, as adoption of new technology in general is associated with younger age groups (e.g., Rogers, 2003), we could expect younger cohorts of politicians to adopt new ways to communicate with their respective constituencies. Although there certainly are exceptions (e.g., Chi & Yang, 2010; Hargittai & Litt, 2012), the bulk of scholarly efforts suggests that younger age is an influencing factor behind adoption as well as continued use when assessing the social media habits of politicians (Carlson & Strandberg, 2008; Strandberg & Carlson, 2007). For our current purposes, the respective ages of the parliamentarians could be gathered from their respective home pages on the EP Web (http://www.europarl.europa.eu/portal/en).

Second, although the claim posed by Druckman et al. that “gender matters in the sense that it generates different approaches to campaigning” (2007, p. 429) has been disproven in the context of elections to the European Parliament (Vergeer et al., 2012), the division between male and female candidates has been reported as valid in other political contexts, such as Finland (Carlson, 2007), the United States (Kahn, 1996; Puopolo, 2001) and Scandinavia (Larsson & Moe, 2013b). Because research dealing specifically with gender differences in Internet use have found that women are more likely to use social network sites (Hargittai, 2007) and to do so more sociably than men (Pujazon-Zazik & Park, 2010; Wei & Lo, 2006), we could expect to see larger amounts or at least differing types of activity when comparing the sexes. Thus, data regarding the gender of each parliamentarian was gathered from the European Parliament Web site.

Third, the characteristics of the political organization to which the individual politician belongs need to be taken into account. Specifically, incumbency and ideology have both been used as predictors of online activity sophistication. Regarding the former of these, because more established political actors could be expected to gain more coverage in established media such as print or broadcast (Vergeer and Hermans, 2013), challengers have been found to try to counter this dominance by employing new communication technologies to larger extents (Carlson, 2007; Druckman et al., 2007; Gibson, Lusoli, & Ward, 2008; Jackson & Lilleker, 2009), a finding that appears to have been at least partly challenged in the supposed 2.0 era (Williams & Gulati, 2012). As for the latter, the influence of ideology has been reported as fluctuating. Although comparably early efforts found differences based on ideological directions (e.g. Gibson, 2004; Kalnes, 2009; Lilleker & Malagón, 2010; Vaccari, 2008a), more recent publications appear to find a decreasing effect of ideology on the online performance of politicians (Lilleker et al., 2011), suggesting an ongoing process of “deideologization” (Schweitzer, 2008)—especially tangible, it would seem, in the context of Twitter use (Vergeer & Hermans, 2013; Vergeer et al., 2012). To test for the influence of incumbency and ideologies on the permanence of Twitter use, the status of the eight party groups to which each representative belongs was noted in the data set.

Fourth, the networking aspects of Twitter are taken into account by assessing the influence of the number of other users following the politician, as well as the quantity of other users followed by the politician on the activity undertaken. The procedure for the construction of these variables involved accessing the Twitter API, as discussed previously.

General Internet Use

Beyond individual variables, more contextual predictors—such as the degree to which different populations of different countries make use of novel technology (e.g., Hermans & Vergeer, 2012; Jensen, Danziger, & Venkatesh, 2007; Williams & Gulati, 2012)—could be expected to have importance in this regard. As such, in order to assess the level of general Internet use in the European Union, two variables extracted from the 2012 Eurostat survey (http://ec.europa.eu/eurostat) were employed. First, in order to gauge the varying levels of Internet penetration in the countries comprising the EU, we employed a variable denoting the percentage of the populations (age 16 to 74) who were frequent users of the Internet, defined as use “every day or almost every day on average within the last three months before the survey.” Second, because no reliable cross-EU data regarding the uses of Twitter could be unearthed, a more inclusive variable was employed in order to capture the degree to which EU citizens make use of such online innovations. Employing the same age and time frames as for the previous variable, this latter inclusion tallied the percentage of “individuals using the Internet for participating in social networks” in each country.

Political Internet Use

Although more general aspects of Internet use could be expected to have influence on the practices under scrutiny here, specific political uses of the Internet at the hands of constituents might also have influence on the degree to which elected officials employ services such as Twitter in their daily routines. In order to account for such uses, two variables from the aforementioned 2012 Eurostat survey were employed. First, the degree to which citizens had been “taking part in online consultations or voting to define civic or political issues (e.g., urban planning, signing a petition)” was employed as a measure of more general online political activity. Second, the propensity to engage with politicians online was tested by means of a variable measuring the percentage of “individuals using the Internet for interaction with public authorities” in each country.

Political Trust

As discussed in the introduction of this paper, it would appear that citizen interest in the EU is regularly found at less than impressive levels. As such, levels of trust toward the same institution might have some influence over the processes of interest here; citizens who have high confidence in the EU might be more prone to contact their officials for suggestions or discussions, potentially leading to more activity at the hands of the latter group. Conversely, a reverse scenario could be possible: low levels of trust among the citizenry might lead to extended use of social media such as Twitter at the hands of politicians in attempts to enter into dialogue with citizens to present themselves as more likeable or approachable (Vergeer et al., 2011a). Data regarding political trust were taken from the European Social Survey (available at http://www.europeansocialsurvey.org/). Specifically, the mean score calculated for each country on the Likert Scale variable assessing “trust in the European Parliament” (0–10, where “0” indicated “no trust at all”) was employed for this purpose.

In sum, descriptive statistics for the independent variables described above are available in Figure 1.

FIGURE 1. Activity index medians per party group.

Error bars indicate 95% confidence intervals.

As indicated in Table 1, some of the variables, such as data on users followed and followers on Twitter, proved to be highly skewed. In order to make them more suitable for further analysis, a log transformation was employed in order to normalize them (following Hair, 2010; Vergeer, Hermans, & Sams, 2011a).

TABLE 1. Descriptive Statistics for the Independent Variables Employed

Variables	Descriptives	Data gathered from
Politician’s Demographics
Age	M = 55, Mdn = 56,SD = 11	http://www.europarl.europa.eu/portal/en
Gender	65% male	http://www.europarl.europa.eu/portal/en
Politician’s Twitter Use
N of followers	M = 3926, Mdn = 987,SD = 12,422	Twitter API
N of users followed	M = 458, Mdn = 193,SD = 734	Twitter API
Population’s General Internet Use
Frequent Internet Use (% of population)	M = 58, Mdn = 60,SD = 13	http://ec.europa.eu/eurostat
Social Media Use (% of population)	M = 40, Mdn = 37,SD = 10	http://ec.europa.eu/eurostat
Population’s Political Internet Use
Online Consulting	M = 8, Mdn = 7SD = 4	http://ec.europa.eu/eurostat
Interact w. Public Authorities	M = 42, Mdn = 35,SD = 14	http://ec.europa.eu/eurostat
Population’s Political Trust
Trust in EP (Scale of 0–10, 10 highest level of trust)	M = 4, Mdn = 4,SD = 1	http://www.europeansocialsurvey.org/

As for the dependent variable, the current approach was inspired by previous research (e.g., Schweitzer, 2005, 2011; Vaccari, 2008a, 2008b). An activity index was constructed by dividing the total number of tweets sent by each parliamentarian by the number of days that each politician had been on Twitter at the time of data collection. As such, the index provides us with a straightforward measurement of the average number of tweets per day sent by each included politician, giving some indication of the degree of permanence. Indeed, although such an arguably simplistic ratio cannot provide any deeper insights into the often-intricate relationships and communication habits that take place on and are formed by Twitter (e.g., Marwick & boyd, 2010), it can provide us with more overarching insights regarding the degree to which politicians use Twitter in a continuous, permanent fashion.

RESULTS

To begin with, out of 754 members of the European Parliament, 413 (55%) provided a link to a Twitter account at the time of data collection. Descriptive statistics show considerable variation regarding the activity within this group. To illustrate, the mean number of days that each account had been active was 1037 (SD = 501, Mdn = 1,198), while the mean number of tweets sent from each account was 1,315 (SD = 2,641, Mdn = 473). The aforementioned activity index suggested that although on average, each account had sent about one tweet per day (M = 1.3, SD = 2.3), the median, reported at .55, indicated that the level of activity is perhaps better understood at more modest levels. Similar results were also reported for the other Twitter use variables. The mean number of followers emerged as 3,926 (SD = 12422, Mdn = 987), while the mean number of users followed was 459 (SD = 734, Mdn = 193).

Given the quality of the data, medians rather than means were chosen for further analysis. First, given the previously mentioned assumed influence of ideology and incumbency on the permanence of online activities by politicians, Figure 1 provides a comparison of the medians of the activity index for each party group in the parliament.

The bars indicating the median value for each party group are presented in descending order. As such, we can tell that although politicians in the nonattached group—parliamentarians previously affiliated with one of the established party groups—appear as the comparably most ardent everyday users of Twitter (Mdn = 1.13), the associated error bar indicates a rather large spread around the reported median. It follows that we should expect to find both high-end and low-end users among politicians in this group. Similar tendencies, although perhaps not as stated, can be seen for the Group of the Greens (Mdn = .81), European Conservatives (Mdn = .65), Confederal Group of the European United Left (Mdn = .48), and the Europe of Freedom and Democracy Group (.37), suggesting considerable variation regarding the Twitter activity of the politicians in each group. Although the error bars for the remaining three party groups—Group of the Progressive Alliance (Mdn = .63), Group of the European People’s Party (Mdn = .56), and Group of the Alliance of Liberals and Democrats (Mdn = .41)—are still substantial, their relatively smaller size indicate less variation and hence less uncertainty regarding the reported median values.

Finally, as no significant differences between the reported medians could be found for differing ideologies or for incumbency status (Independent Samples Kruskal-Wallis = p > .05 across groups), we can conclude that although the results reported in Figure 1 provide some insights into the Twitter activities of different party groups, more solid explanatory variables must be sought elsewhere.

As previously discussed, research has indicated that online activity might vary based on the political cultures of different countries. Figure 2 presents the median values of the activity index for politicians per country.

FIGURE 2. Activity Index Medians per country.

Error bars indicate 95% confidence intervals.

Utilizing the same guidelines for interpretation as for Figure 1, politicians from the Netherlands appear to utilize Twitter in a more permanent fashion—over two tweets per day (Mdn = 2.47)—than representatives from other countries. Continuing down the list of descending medians, we can discern that countries such as Spain (Mdn = 1.48), Slovenia (Mdn = 1.11), Ireland (Mdn = .93), the UK (Mdn = .88), Poland (Mdn = .83), and Italy (Mdn = .81) tend to feature politicians maintaining a Twitter activity of around one tweet per day. At the lower, right-hand side of the scale, we see that representatives from Slovakia (Mdn = .03), Romania (Mdn = .06), Cyprus (Mdn = .09), and Lithuania (Mdn = .12) apparently take a less permanent approach to the service under scrutiny.

Much as with the results presented in Figure 1, the error bars visible in Figure 2 make a clear mark on some of the reported medians. Gauging the size of these confidence intervals in relation to their corresponding medians, we can see that although politicians representing established democracies such as Denmark and Cyprus exhibit some variety in their Twitter activity, the bulk of countries with comparably larger error bars tend to be so-called “new democracies” (Lisi, 2013) such as Portugal, Czech Republic, Bulgaria, Lithuania, Romania, and Slovakia. Even though the medians for these countries tend to be lower, the spread around those medians suggest that some of these representatives are highly active, employing Twitter in modes characterized by permanence, while some are admittedly not so active.

In comparison with the results presented in Figure 1, the differences between the medians depicted in Figure 2 emerged as statistically different (Independent Samples Kruskal-Wallis = p < .05 across groups), suggesting that although variations are indeed abundant, different countries appear to nurture different political cultures, which in turn apparently influences online practices.

In order to further assess the influences of the independent variables on the activity index, hierarchical multiple regression analysis was employed (see Table 2). Roughly corresponding to the order in which the variables were previously presented, the results are introduced blockwise, as they were entered into the analysis.

Focusing first on the variables dealing with representative demographics (Block I: Politician’s Demographics), the negative effect of age—the younger the representative, the more likely he or she is to use Twitter more adamantly—should perhaps come as no surprise. Because younger generations in general tend to be more well versed in employing online services for political purposes (e.g. Bakker & de Vreese, 2011; Hargittai & Shaw, 2013), such tendencies should indeed also be visible among those in the younger generation who pursue a career in politics. Conversely, gender emerged as a nonsignificant predictor. As such, the previously discussed gender-based differences pertaining to more permanent Twitter use do not appear as salient in this context. This result could be interpreted as a somewhat general effect of political professionalization, given that research gauging comparable topics has come to similar conclusions regarding this variable (e.g., Vergeer & Hermans, 2013; Vergeer et al., 2011a), perhaps the effect of gender is undermined by such processes of professionalization.

TABLE 2. Multiple Regression Analysis Predicting Twitter Activity Index

	Block I: Politician’s Demographics	Block II: Politician’s Twitter Use	Block III: Population’s General Internet Use	Block IV: Population’s Political Internet Use	Block V: Population’s Political Trust	Incremental R^2 (Adj. R^2)
Age	−.21***
Gender*	−.06					.06 (.05)
N of Followers		.05
N of Friends		.33***				.21 (.20)
Frequent Internet Use			−.22**
Social Media Use			.18**			.26 (.24)
Online Consulting				.17*
Interact w. Public Auth.				.16*		.27 (.25)
Trust in EP					−.09	.28 (.26)

Note: Standardized beta coefficients presented.

*** = p < .001, ** = p < .01, * = p < .05

* = Dichotomous variable: Female = 0, Male = 1

Block II tests for the influences of Politician’s Twitter Use. The results suggest that although ample numbers of followers do not persuade the politicians to maintain a more permanent Twitter presence, representatives who themselves follow other Twitter users more extensively also tend to be more active in sending tweets on a continuous basis. This suggests that having an “imagined audience” (e.g., Marwick & boyd, 2010) to cater to is not enough in this regard. Rather, it seems reasonable that a sizeable self-selected group of what could be described as Twitter “followees”—a group of users with whom the individual politician would interact more frequently—would result in a higher score on the employed activity index.

Block III provides details regarding the Population’s General Internet Use in each country. Curiously, although the variable gauging the extent to which citizens across the EU make frequent use of the Internet emerges as a negative, significant predictor for the activity index—the less frequent Internet use, the more permanent Twitter activity at the hands of politicians—the measurement testing for social media use among the populace proves to be likewise significant, albeit with a positive effect, suggesting a reinforcing effect on our dependent variable. One tentative interpretation of these results could be derived by again assessing the visualization in Figure 2. The countries to the right of the scale, featuring the most active EU parliamentarians, are not necessarily the same countries that are characterized by higher levels of Internet use—consider examples such as Spain, Slovenia, Ireland, Poland, Italy, or Hungary. The results presented here suggest that in countries such as these, EU parliamentarians are more adamant in upholding their day-to-day Twitter activity. Combined with the significant and positive effect of social media use among the populations, the data seems to cautiously suggest what could be labeled as an “elite effect” of social media, whereby EU representatives for countries with lower degrees of Internet penetration employ novel technologies to interact with those already connected, supposedly following other users and recruiting followers among those already privileged (e.g. Enjolras, Steen-Johnsen, & Wollebaek, 2012; Hargittai, 2007; Hargittai & Litt, 2012), thereby presumably deepening the already existing digital divides (e.g. Rogers, 2001; Van Dijk, 2005). Such an elite effect is indeed discernible also in other contexts, largely suggesting a reinforcement of established societal power structures (e.g., Larsson, 2013).

This suggested “elite effect” is seemingly discernible also when assessing the fourth group of measurements (Block IV: Population’s Political Internet Use), dealing with political Internet use in the case countries; both variables emerge as significant, positive predictors of the activity index. Previous research has shown that online consultations and interactions with elected officials usually involve members of various societal elites (e.g., Coleman & Shane, 2012; Conover et al., 2011), and as such, a tentative interpretation of these findings could be that similar influential groups are involved in contacting their respective EU parliamentarians on Twitter, leading to higher degrees of activity from the politicians. Following suggestions that digital media could function as a “magic elixir” (Stromer-Galley, 2000, p. 113) to usher previously unengaged citizens into more active roles (e.g. Chadwick, 2003; Christensen & Bengtsson, 2011), such an interpretation would need to be corroborated by future research, but could nonetheless function as a working hypothesis of sorts.

Block V, then, contains the final variable employed, measuring the Population’s Political Trust in relation to the EP. As evident from the results presented in Table 2, the variable emerged as negative and nonsignificant. Although it is difficult to provide any definitive statement for why trust seems unimportant in this regard, one possible interpretation is that it could be related to the aforementioned disinterest in the EU exhibited by many citizens.

DISCUSSION

Based on their study of the Web sites used by political parties during the 2009 European Parliamentary elections, Lilleker et al. (2011) suggest that such online presences “can no longer be described as static or boring” (2011, p. 208). The results presented here, then, would call that assessment partly into question when it comes to the Twitter activities of EU Parliamentary representatives. Although it is entirely possible for a political party to maintain a lively Web presence while the online activities of associated individual politicians remain at rather low levels, the median amount of tweets per day (measured at .55) gives some credence to the claim that “the very nature of Web 2.0 technologies may make them less compatible” (Kalnes, 2009, p. 251) with the established work routines of politicians. Indeed, they are sometimes seen as resistant to change (e.g., Downs, 1957) and in favor of more established (as in controllable) modes of citizen outreach (e.g., Stromer-Galley, 2000; Vaccari, 2008b). Although we should perhaps be wary of labeling this result as an example of “de-professionalization” (e.g., Tenscher, 2013), the findings do indeed speak to the notion that the uptake and permanent use of Twitter at the hands of politicians are taking place at slower paces than popular debate would have us believe. Indeed, although King’s (1982) focus was on technology use in local governments, the same “10-year lag period between introduction … and acceptance and routinization” (King, p. 25) is perhaps applicable here also, a suggestion that could help explain the co-occurrence of the sophisticated Web pages found by Lilleker et al. (2011) and the relative inattention to engaging on Twitter in a permanent fashion. Indeed, the results presented here suggest that these processes are at somewhat of a starting point, where early adopters of permanent practices lead the way for others. Moreover, although the results provided here provide useful insights on their own, they could hopefully also serve as a point of comparison for future comparative efforts.

The results presented here found online permanence on Twitter to be rather limited. However, this overarching result needs to be understood in combination with the influences of the independent variables. Because the differences between different party groups proved to be nonsignificant (as reported in Figure 1), we should be careful to draw any larger conclusions from the results presented therein. With this important caveat in place, it is nonetheless interesting to note that according to the activity index, politicians associated with the nonattached group appear more zealous in their day-to-day Twitter activity, a result that could be related to their status as relative outsiders when compared to more traditionally oriented parliamentarians. With such more incumbent groups in mind, it is noticeable that the two groups enjoying the majority of parliamentary mandates (Group of the European People’s Party [275 mandates after the 2009 election] and Group of the Progressive Alliance [195 mandates]) exhibit comparably limited spread around the medians as expressed by the error bars in Figure 1. Again, although caution must be applied, this result could be related to the tendencies found by Tenscher (2013), suggesting that established parties might choose to settle on strategic, seemingly less than permanent levels of online campaigning effort from which they deviate only slightly.

Regarding variations between politicians from different countries, the findings suggest a blurring of the previously proposed North–South barrier across Europe regarding the online activities of politicians (e.g., Norris, 2000a). Although there are exceptions, the outcomes presented in Figure 2 indicate the politicians who appear to more wholly adopt notions of permanent campaigning in the context dealt with here are spread all over Europe. The results almost suggest a “leapfrog” effect of sorts, indicating that relative newcomers to the Web—or to democratic modes of government, for that matter—might be able to more easily adopt novel would-be paradigms such as Web 2.0. Consider, for example, the result suggesting that parliamentarians from a comparably new democracy such as Poland—suggested by Lilleker et al. (2011) to “lag behind in almost all respects” (p. 206) of traditional, Web-site–based campaigning—appear as comparably more permanent in their use of Twitter than representatives from countries more often associated with online sophistication. Continuing with the aforementioned dichotomy of “new” and “old” democracies in mind, as pointed out, politicians from countries in the former of these two groups tend to exhibit more variation in their Twitter activity, something that could be interpreted as a more enthusiastic or perhaps experimenting approach. Although it is difficult to draw any such detailed conclusions based on the overarching, structural data collected for this study, the findings reported in Figure 2 indicate that we might need to reconsider previous geopolitical dividing lines when dealing with this particular aspect of the professionalization of politics.

The results emanating from the multiple regression analyses presented in Table 2 suggested that younger parliamentarians, representing constituencies characterized by high levels of political Internet use who follow comparably more other users, tend to maintain a more permanent presence on the platform under analysis. However, the results dealing with more general aspects of Internet use proved to be somewhat different. One possible interpretation for these combined results sees a combination of the aforementioned tendencies toward “elite” and “leapfrog” uses of online services. Although politicians representing countries traditionally characterized by high levels of Internet use might indeed host Web sites characterized by higher degrees of sophistication and professionalization (e.g. Hermans & Vergeer, 2012; Lilleker et al., 2011), social media platforms such as Twitter are apparently more permanently used by those politicians who hail from countries traditionally associated with limited levels of Internet penetration. As such, the professionalization of politics would appear to take on different shapes in different contexts (Lisi, 2013), at least when considering the particular subset of related concepts under scrutiny here. Given the nature of the data employed for the study at hand, any claims regarding “innovation” (Margolis & Resnick, 2000) would need to be corroborated by more in-depth studies of these practices (see also Larsson & Svensson, 2014). With this in mind, the following and final section of this paper deals with such suggestions for future research efforts, simultaneously taking the limitations of the work performed into account.

LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH

Although the study at hand has made important inroads in charting out day-to-day social media use by EU parliamentarians, it has limitations that need to be addressed. First, although querying the Twitter API provides us with overarching use data in the form of status updates per user, this particular process does not provide insights into the specifics of message types sent, such as @replies that allow users to directly communicate with others, or retweets where posts sent by other users are redistributed to one’s own network of followers. Future research efforts should attempt to take a more refined view of these messages, thereby providing findings regarding the contents of tweets—something that this structural study cannot deliver.

Second, because the specific focus of the work at hand was on Twitter, the findings reported here arguably miss out on any activity carried out by parliamentarians on other online—or offline, for that matter—platforms. As such, future research should make attempts to adopt multimodal perspectives, taking several services into account (e.g. Kim, Painter, & Miles, 2013; Vergeer & Hermans, 2013).

Third, although the utilization of trustworthy secondhand data can help raise levels of validity and reliability in research efforts, the degree to which such data is malleable to new contexts beyond the original must be described as limited (e.g., Moeller & de Vreese, 2013). Because the variables borrowed from the Eurostat and European Social Survey were thematically aligned with the topic dealt with here, this is perhaps less of a problem in the current study, but worth noticing nonetheless.

Fourth, although the independent variables employed here provided useful insights into the mechanisms behind continuous Twitter use, future research should seek to implement novel measurements. Because the supposed shift from a 1.0 to a 2.0 paradigm of Web publishing and campaigning has coincided with the development of more individualized campaigning efforts (e.g., Bimber & Davis, 2003; Vergeer, 2012), explanatory factors pertaining specifically to the individual politician might be of interest. As suggested by Vergeer and Hermans (2013), such large-scale collection “might be very difficult to organize, requiring a survey among politicians and candidates” (p. 16). Perhaps a suitable starting point could be more qualitative efforts looking into the characteristics of individual politicians of particular interest when it comes to permanence in online activities. For example, it could be suitable to further scrutinize how the national political characteristics of each parliamentarian affect their efforts in the EP. Similarly, deeper understandings into the motivations and thought processes of nonusers of Twitter and similar services would be helpful in order to further develop our knowledge regarding not only drivers of adoption, but also of nonadoption. Qualitative insights such as these could prove helpful not only on their own, but also when variables are to be constructed for future quantitative studies.

Fifth, and finally, the results presented here suggest that the permanent use of Twitter by EP representatives was rather limited. Such a statement begs the question: low levels of use compared to what? As already mentioned, the hope is that the results presented here can serve as cases for comparison for similar, future studies. As such, although this paper has provided important insights regarding current communicative practices among EP representatives, future studies might help provide nuance to the reported findings.

In conclusion, we must keep the raised interest surrounding services such as Twitter in mind when discussing its political uses. Although little comparable cross-country data is available regarding the use of Twitter in Europe or elsewhere, insights from political Twitter use in contexts such as the Scandinavian countries (Larsson & Moe, 2012, 2013a, 2013b), Austria (Ausserhofer & Maireder, 2013), the United States (Bekafigo & McBride, 2013), and Germany (Tumasjan et al., 2010) have indicated that high-end users tend to be privileged societal “elites” (e.g. Hargittai & Litt, 2012), something that again resonates with the findings provided here. As such, Twitter is perhaps best understood in this context as a channel for politicians to communicate and network with those of equal privilege, rather than as a means to counter the decline in political participation and interest. Although the results presented here suggested that politicians from countries not usually associated with high levels of Internet use were among the more ardent permanent Twitter users, we cannot say anything about what they say on the platform or with whom they communicate. Future studies looking into these issues, perhaps also in other, non-Western contexts (e.g., Hsu, Park, & Park, 2013; Otterbacher, Shapiro, & Hemphill, 2013), will provide further insights into these matters. Because a permanent campaign by definition is something that takes place over time, further longitudinal research looking into these uses should provide us with more insights regarding the degree to which professionalized online campaigns are indeed “characterized by being permanent, although varying with intensity” (Strömbäck, 2007, p. 54), as well as how the speed and characteristics of these ongoing processes might vary in different political contexts.

Additional information

Notes on contributors

Anders Olof Larsson

Anders Olof Larsson is a postdoctoral fellow at the Department of Media and Communication, University of Oslo. His research is mainly focused on organizational uses of the Internet. Larsson’s web site can be accessed at andersoloflarsson.se.