Scholarly article on topic 'Triumph of the Underdogs? Comparing Twitter Use by Political Actors During Two Norwegian Election Campaigns'

Triumph of the Underdogs? Comparing Twitter Use by Political Actors During Two Norwegian Election Campaigns Academic research paper on "Media and communications"

Share paper
Academic journal
OECD Field of science

Academic research paper on topic "Triumph of the Underdogs? Comparing Twitter Use by Political Actors During Two Norwegian Election Campaigns"


Triumph of the Underdogs? Comparing Twitter Use by Political Actors During Two Norwegian Election Campaigns

1 2 Anders Olof Larsson and Hallvard Moe


Social media are often discussed in terms of online novelties. However, especially within the broader field of political communication, the uses of such services, as Twitter, at the hands of political actors such as politicians and the parties to which they belong, have become something of a fixture of research in recent years. Although the study of political Twitter use has provided a series of insightful case studies, often focused on one single election or country, this article presents a comparative study looking at Twitter use at the hands of political actors during two Norwegian elections, 2011 and 2013. We are interested in what overarching tendencies can be discerned from these uses—specifically, if differing usages can be found between the two elections, suggesting developments pertaining to the normalization and equalization hypotheses respectively. This is examined by focusing on two main analytical areas: The level and type of activity undertaken by those up for election, and the repercussions that this activity appears to have in terms of popularity on the studied platform. In short, the results suggest that although Twitter largely remains an "elite" medium in the Norwegian context, smaller political and other actors are making use of the platform at hand to higher degrees than their more well-known peers. Tendencies of both hypotheses are traced in the data, and although the findings could signal an opening for "outsiders" in this regard, the sheer amount of traffic driving the tweets sent by high-end politicians suggest otherwise.


political communication, politicians online, social media, Twitter, Norway, normalization, equalization


October-December 2014: 1-13 © The Author(s) 2014 DOI: 10.1177/2158244014559015


Although social media such as Twitter are often viewed as novelties, studies of such services within political communication have been a fixture for quite some time. Indeed, research on the political uses of Twitter has provided a series of single-country case studies from a variety of geographical settings. At the same time, this very basic observation suggests a dearth of research comparing uses between different time points or contexts (Bruns & Stieglitz, 2013). The present article makes a contribution in this regard. Specifically, we compare data on Twitter use during two Norwegian election campaigns, in 2011 and 2013. Although the time span is not that long, we argue that two elections set apart by 2 years constitute a valuable basis for analysis when it comes to the fast-paced developments of online services and their uses in society in general, and parliamentary-political settings in particular. Moreover, as the case studies that make up the field almost exclusively provide findings from one single election or similar event, our current effort provides a comparative contrast.

The current article is focused on Twitter use by political actors (understood here as individual politicians as well as the parties that they represent) in a small European nation

state, featuring party-centered politics (Karlsen, 2010) and advanced Internet users (Vaage, 2012). Our case country Norway has just more than 5 million inhabitants and is often described as a Nordic welfare state (e.g., Hilson, 2008). It features a multi-party parliamentary system with universal voting privileges. National as well as local and regional elections are held on fixed, but separate, dates. The present study, then, deals with two such political events—the 2011 local and regional elections, and the 2013 national elections. Both elections took place in similar national political contexts, as Norway had been ruled by a Left-Centre coalition since 2005. The 2011 election saw the Conservative party and Labour as overall winners, in addition to, albeit on a smaller scale, the Green Party. The 2013 election saw a change of government to a minority coalition of the Conservative and the right-wing populist Progress Party, with support from

'University of Oslo, Norway University of Bergen, Norway

Corresponding Author:

Anders Olof Larsson, Department of Media and Communication, University of Oslo, P.O. Box '093, Blindern, Oslo 0317, Norway. Email:

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License ( which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page (

r r 6 r Downloaded from by guest on Jufie 5, 2016 r&vr &r r )

smaller Center parties. It follows from this that although the elections studied are certainly different in terms of size and scope, the opportunity to examine diachronic data from the same political context arguably needs to be assessed. As such, although we cannot make any exact comparisons between the 2011 and 2013 events, our current efforts are nevertheless relevant to those interested in the developing nature of political social media use.

From a conceptual viewpoint, the study presented here is framed by the two competing normalization and equalization hypotheses. The former of the two essentially suggests that those political actors who are in favorable positions in the offline world would enjoy a similar status and amount of attention in the online environment—here represented by Twitter. The latter hypothesis suggests the opposite: In essence, the Internet in general and social media in particular would usher in tendencies of equalization between larger and smaller political actors, suggesting that the less fortunate group would use various online techniques to higher degrees than their competitors.

These hypotheses have certainly been up for discussion in previous scholarly efforts. The study at hand, however, takes a diachronic, comparative approach to the propositions made—gauging the degree to which these hypotheses remain valid or appear to change during the course of two elections. Our basic assumption is that during the period of study, Twitter went from creating buzz as the newest political communication channel in what was dubbed Norway's first Twitter election in 2011 (Enli & Skogerbo, 2013), to becoming something of a stable of political campaigning, with what could perhaps be described as more established pattern of use, more mature for politicians and their campaigners, for activists, as well as for politically interested citizens.

To explore these supposed developments, we focus empirically on two main analytical areas. First, we scrutinize the relations between the most and the least active users during both elections to assess the degree to and the ways in which political actors take up space in Twitter-based political contexts. Second, to get a more detailed understanding of the ways in which Twitter's specific communicative modes for dialogue and for redistribution are used among the most prolific users, we also compare the networks constituted by Twitter's dialogic mode of address (@replies) and its function for redistribution of messages (retweets, RTs) with a specific focus on those who self-identify on their profile pages as politicians or political parties.

The empirical analysis is based on archived tweets, and their metadata, related to the most prolific hashtags for both elections (2011: N = 29,423; 2013: N = 60,612). Although the problem of changing technical circumstances, individual data gathering tools used by different research groups, and the lack of shared data between such groups have hindered much comparisons across cases in the field (e.g., Bruns, 2013; Bruns & Stieglitz, 2013), data for the present analysis were collected following a similar set up and with a shared

methodological approach for both periods, providing the basis for a meaningful comparison. Using statistical and network analysis, we combine approaches often associated with "big data" research with an analytical focus on "small data" (cf. boyd & Crawford, 2012; Bruns, 2013). As such, the aim is to compare how Twitter use by political actors compare between two elections, and whether these uses can be best understood along the lines of the two hypotheses introduced above and discussed further in the subsequent section.

Social Media in the Longer Perspective—Normalizing or Equalizing?

Online developments are fast-paced, especially when it comes to the suggested political employments of "hyped" social media services such as the one under scrutiny. As previously mentioned, the current article seeks to trace changes in Twitter use by politicians and political parties during two recent Norwegian elections. Specifically, election-related Twitter data are gauged for tendencies of normalization or equalization with regard to usage at the hands of the aforementioned actors.

A broad field of inquiry, political communication could be said to encompass a multitude of perspectives and methodological preferences as well as a series of subfields—effectively rendering the field akin to what Whitley (2000) has referred to as a "fragmented adhocracy." Nevertheless, there are certainly concepts used within the field that might be familiar to the vast majority of scholars, regardless of what specific research interests they pursue. This, we argue, is certainly the case with the conceptual framework taken into use in the current study. Indeed, one could perhaps question the need for or even suitability of scholars lamenting so-called positive or negative results with regard to the uses of Information- and Communication Technologies (ICTs) by more or less established actors (e.g., Bekafigo & McBride, 2013; Hermans & Vergeer, 2013). Nevertheless, these and other, similarly themed dichotomies of perspectives into broadly "Internet-pessimist" or "Internet-optimist" have become a mainstay when discussing not only our current theme (e.g., Larsson & Svensson, 2014) but also in relation to more general inquisitive inroads into online participation among more or less privileged groups of citizens (Dahlgren, 2005; Freelon, 2010). Although Wright (2012) suggests such a dichotomized perspective might lead researchers to be "too pessimistic in their analysis of the impacts of technology on politics" (p. 249), we argue that a rigorous employment of these concepts to trace tendencies in empirical data can still provide interesting insights regarding the development of political social media use. Indeed, we attempt to take both sides into account, rather than erring on caution or hype. As such, these conceptual tools can serve as useful sensitizing devices with which to approach our results.

Early—especially purely conceptual or theoretical—work on the topic at hand largely fell along the lines of the equalization hypothesis, suggesting a view of the Internet as an "inherently democratizing technology" (Coleman & Blumler, 2009, p. 166). Indeed, optimism was abundant as proponents suggested that those actors who could be regarded as under-or unprivileged in some regard in the "broadcast democracy" would come to have certain advantages by adopting Internet-based applications for campaign purposes. For instance, marginal actors would essentially be able to gain access to the public through the Internet—exposure that they lacked in a broadcast setting, given the restrictions on access to speak through traditional mass media, carefully guarded by editorial gatekeepers. With the relatively inexpensive platforms provided by Internet technologies (Strandberg, 2013), these novel practices were assumed to lead to success in terms of attention gained and—possibly—a follow-up in terms of triumphs at the ballots (see Strandberg, 2009 for an overview).

Comparably later work, largely informed by empirical studies, have for the most part reached conclusions suggesting tendencies according to what Margolis and Resnick (2000) earlier labeled "Politics as usual" (see also Larsson, 2013). By contrast to its equalization counterpart, the normalization hypothesis suggests a "rich-get-richer" effect with regard to the uses of the Internet at the hands of politicians and political parties. Essentially, this entails that much as major political actors have tended to dominate in society at large, so will they come to overshadow the activities of minor actors also on the plethora of Internet platforms available. Indeed, "cyberspace does not exist in a vacuum" (Gibson, Lusoli, & Ward, 2008, p. 17), and scholars suggesting the merits of the normalization hypothesis often point to how parties rich in resources come to dominate in online as well as offline settings (e.g., Klinger, 2013).

As opinions and results have fluctuated with regard to these contrasting viewpoints, scholars have begun to trace the contours of a third possible mode of Internet adoption by political actors (e.g., Larsson & Svensson, 2014). Sometimes discussed as the "ebb and flow thesis" (Lilleker et al., 2011) or "web 1.5" (Jackson & Lilleker, 2009) advance to the web by political actors, such a middle road appears to suggest a cyclical approach to the above discussed hypotheses. Similarly, such a cyclical approach has also been hinted to by some authors (see Gibson & McAllister, 2014) as a result of the influx of social media such as Twitter. Given their general potential to provide a "ready-made" platform for use by smaller actors, Twitter and similar services could serve as an equalizing factor (e.g., Strandberg, 2013). In specific relation to social media services such as the one under scrutiny here, previous research from the Scandinavian context has largely found that although relatively unknown "regular citizens" are certainly taking part in hashtagged political conversations, the bulk of traffic is performed in relation to—often as retweets of—established politicians (Larsson & Moe, 2013).

However, the bulk of the studies performed are construed as case studies, focusing on one election or similar event. As such, our current efforts provide a somewhat novel approach to the study of political Twitter use.

To be precise, we focus on assessing two specific aspects of political parties' and their representatives' presence: first, the level and type of activities undertaken by those up for election. Following the equalization hypothesis, smaller actors would make greater use of Twitter to disseminate their messages. The normalization hypothesis, then, would suggest the opposite—as larger parties have more resources and staff, they will be able to dominate not only the established media but also the electoral Twittersphere. Second, we look at the repercussions that the activity undertaken by the politicians and political parties appear to have in terms of popularity on the studied platform. In essence, the equalization hypothesis would then suggest that diminutive actors enjoy greater spread in the setting under scrutiny here. The converse must be said for the normalization variety, proponents of which would point out that as major political actors remain in the public's eye throughout a campaign—through televised content, for example—so will they gain the most attention also in the online setting.

The comparative diachronic aspects of the study at hand provide insights into how these tendencies develop between two elections in an advanced Internet society. Although we cannot make any firm statements based on data from two time points, we argue that the results presented here help us understand much-hyped ongoing developments.

Research Approach

Data collection was undertaken utilizing YourTwapperKeeper, an open-source tool, which uses the Twitter stream and search Application Programming Interfaces (APIs) to collect public tweets and their corresponding metadata (e.g., Bruns & Burgess, 2011a; Bruns & Stieglitz, 2013). Specifically, a hashtag-based approach was deemed suitable to "identify the relevant streams of information" (González-Bailón, 2013, p. 154) regarding political activity on Twitter. On Twitter, hashtags "represent a way of indicating textually keywords or phrases especially worth indexing" (Halavais, 2014, p. 36). Although they are sometimes used as inside jokes, to express sarcasm, or to add metacommentary on a tweet, we assume that for the users of the hashtags selected here, the original intent of improving searchability and allowing third parties to track the conversation on a topic remains key. The approach is suitable as we concentrate on those who use Twitter to express themselves in a public communication on politics during an election campaign. With regard to our specific focus on studying developments in online practices such as these, we might also expect hashtags to become more widely used over the period of study. As such, our current efforts differ from previous research, where the study of politicians on Twitter has largely been focused on tracing their

Table 1. Summary of Archives of Tweets for Both Studied Elections.

Election year Hashtag archives Total N of tweets archived Time frame

2011 #valg20ll (election20ll) 29,423 l2 August-l5

#valgll (electionll) September

#kommunevalg (municipality


2013 #valg20l3 (election20l3), #valgl3 60,612 9 August-l2

(electionl3) September

respective accounts (e.g., Vergeer & Hermans, 2013). By approaching our topic in this way, we can show the degree to which certain political actors use Twitter in what could be labeled a "mature" fashion—including hashtags to make their messages visible in certain thematic settings.

A combinatory approach of two or more hashtags per election was used (see Table 1). For both studied elections, a similar time frame was used to capture tweets during the intensive "short campaign" (Aardal, Krogstad, & Narud, 2004)—that is, the month leading up to election day. Also for both cases, archiving of tweets was terminated 3 days after the election, so as to catch the immediate post-electoral activities.

As similar, election-themed hashtags had prevailed in previous Scandinavian elections (Larsson & Moe, 2013), the focus on such popular, themed tweets seemed reasonable. Moreover, the focus on tweets posted by users into a self-selected public political context makes the ethical considerations of studying online political utterings slightly less complicated (e.g., Lewis, Zamith, & Hermida, 2013; Zimmer & Proferes, 2014).

The collected data were subjected to a series of analyses. For statistical examinations regarding the quantity of tweets sent by different users, Excel and SPSS were used. Gawk scripts were utilized to control data quality, filter the data sets, and extract information from them (Bruns, 2011; Bruns & Burgess, 2011b; Bruns & Stieglitz, 2013). The network analysis and graphing software Gephi was used to map out the relationships between the identified high-end users (Bastian, Heymann, & Jacomy, 2009; Bruns, 2011). Furthermore, the classification of these high-end users was undertaken according to a twofold rationale. First, with regard to the identities of high-end Twitter users, their profile pages were visited and the self-disclosed information provided there was taken into account. The short profile presentations provided by each user were classified according to a rationale inspired by previous, similar efforts (Ausserhofer & Maireder, 2013; Graham, Broersma, Hazelhoff, & van 't Haar, 2013). For our purposes, it was deemed suitable to distinguish between five types of users: Media (journalists, writers, entertainers, and so on, affiliated with established media organizations or accounts operated by the media organization centrally); Political actors (accounts operated by politicians or political parties); Communication/PR (Public Relations; professionals working in the communications/PR

industry); and Citizens (users who do not present themselves as affiliated with any specific organization or interest group). This approach, then, allowed us to detail the degree to which political parties or individual politicians were among these top users, or whether indeed other groups—societal elites or not—were more plentiful.

Second, high-end users in terms of sending of @replies and retweets were assessed by applying classification rationales featured in similar, previous research efforts (Larsson & Moe, 2012, 2013). Specifically, for the practice of sending and receiving @replies, we can distinguish between Senders (characterized by sending many but receiving few), Receivers (receives many, sends few), and Sender-Receivers (exhibits a comparably reciprocal approach with regard to the specified functionality). For the practice of retweeting, users are understood according to a similar threefold classification— Retweeters (active in redistributing messages sent by others), Elites (retweeted often, but not active in retweeting messages sent by others), and finally Networkers (exhibits a reciprocal approach to the retweet functionality). These classifications helped us gauge the degree to which the identified political actors adopted the medium at hand with all functionalities in mind, and the degree to which they enjoyed popularity in the electoral twittersphere—through receiving @replies and retweets.

For both coding rationales, reliability was assessed by involving both authors in the work process. The first author made the initial efforts of classification, which were then assessed and agreed upon by the second author. Adopting an iterative approach, both actors subsequently discussed and judged the classifications made. As both authors agreed on the judgments made, the coding provided was considered reliable (e.g., Kirk & Miller, 1986).


Before turning to the specific findings based on our two points of comparison, Figure 1 presents a temporal overview of the data collected for both elections. The presented timelines illustrate the setting for the activities undertaken by the political actors found in the sample.

Figure 1 features a series of bars (denoting the number of tweets sent) as well as lines (representing the number of users responsible for sending those tweets). Black color for both indicators represents the 2011 election, whereas gray indicates tweets and users active during the 2013 events. In both cases, data were collected during a month-long period before each election day, also including some of the postelection period. This collection rationale is visible in Figure 1 as the graph starts 31 days from both election days, allowing for direct comparison between the two periods.

Considering both elections, two main tendencies can be discerned from Figure 1. First, by pure quantitative measures, the election-themed hashtags studied in 2013 attracted more users, also producing more tweets than the 2011

Figure 1. Distribution of tweets and users for the 20ll (black bars and lines) and 20l3 (gray bars and lines) elections, 3l days preelection to 3 days post election.

varieties. Although the gray and black bars denoting tweets sent during both elections at times exhibit somewhat similar patterns, indicating comparable levels of use, the difference between the two increases as the campaigns progress. On election day, this difference is substantial, with 5,988 users producing 14,066 tweets in 2013 compared with 3,015 users yielding 9,163 tweets in 2011.

Although it is difficult to provide solid evidence for such a claim, this difference in scale is most likely dependent on a general rise in Twitter use from the former election period to the latter. In connection, the awareness of social media conventions, such as the hashtag, among users as well as the general public had most likely been heightened since the 2011 election. By 2013, not only was the hashtag part of the communicative modes on other services (e.g., the photo sharing platform Instagram), but the hash sign denoting such content pertaining to thematic categorizations also seeped into other contexts, such as online and offline newspaper articles or adverts attempting to increase interest on Twitter for their specific product or service. Moreover, as the 2013 events concerned national rather than local or regional events, and as voter engagement in national elections tend to be higher than for regional counterparts— approximately 10% to 15% lower voter turnout in Norwegian local elections compared with national ones (Statistics Norway, 2014)—the increase in use is not unexpected.

A second tendency visible in Figure 1 concerns differences in intensity of use during the periods. The lines and bars are characterized by a series of clearly discernible increases, indicating heightened levels of active users and corresponding tweet traffic. Although the principal escalations occur on election day for both years—with increasing levels leading up to those specific days—the other rises are related to political, election-related media events such as party leader interviews or political debates. Closer scrutiny of the data for both years discloses that the two studied election periods are remarkably similar on this point: The first spike in each case is related to increased activity surrounding the Norwegian public service broadcaster Norwegian Public Service Broadcaster (NRK's) televised party leader debate— an event that effectively signals the opening of the most intensive phase of election campaign (on Day 28 for both years). The remaining spikes all correspond with similar media events, both time and content wise. The findings presented in Figure 1 thus suggest that the influence of television is tangible also in a "Post-Broadcast Democracy" (Prior, 2007). With this in mind, these introductory remarks corroborate the findings provided by previous scholarship, largely suggesting that "the Internet reflects and amplifies other events" (Lilleker & Jackson, 2010, p. 93) rather than initiates them—a claim that finds support also in other, comparable contexts (Bruns & Highfield, 2013; Graham, Jackson, & Broersma, 2014; Larsson & Moe, 2012, 2013).

Figure 2. Users of 2011 election hashtags with more than 3 tweets per day on average (99 tweets or more during the period). Note. Number of tweets per user.

Assessing the Activity of Political Actors

With this overall contextualization in place, we now turn to our two main empirical-analytical efforts. First, Figures 2 and 3 regard the assessment of the activity of political actors on the platform. Placing our focus on the more active users of Twitter during the studied election campaigns, Figures 2 and 3 together detail this information for the high-end users—defined here as Twitter accounts that produced at least three tweets on average per day —during the periods specified for each election.

As discussed, previous scholarship suggested distinguishing between four types of accounts. These user types are represented in the figures below according to the following scheme: Media (journalists affiliated with established media organizations or accounts operated by the media organization centrally—represented by light gray bars in Figures 2 and 3); Political actors (accounts operated by politicians or political parties—represented by dark gray bars); Communication/PR (professionals working in the communications/PR industry—represented by black bars); and Citizens (users who do not present themselves as affiliated with any specific organization—visible as white bars in Figures 2 and 3).

When comparing the distributions presented in Figures 2 and 3, we can first note that the sampling criteria used with regard to average activity per day yielded more Twitter users using the service to higher degrees during the latter of the

studied elections. As such, among the most active users, we find the distribution to be slightly more evenly distributed in 2013 than in 2011. Much like before, this development most likely has to do with general trends of increasing uses of social media in the Norwegian context. Moreover, as noted above, the fact that the 2013 election focused on national rather than regional government probably played a part as well. Although the figures presented here only can be said to illustrate a tendency—one among a fraction, albeit an elite, of the users—it still indicates a less dominant position for a few early adopters in terms of volume.

As the scale of "high-end" users involved has increased, we can also notice changes with regard to the self-reported identities of users involved. With the previously introduced color scheme in mind, we can tell that although the distribution for the 2011 election features journalists, politicians, and communication professionals among the very most active users, this has changed slightly for the 2013 event. Here, the comparably larger presence of white bars indicates a strong citizen presence. The top Twitter account for both studied years—VALG2011 and VALG2013, respectively—appears to be operated by the same niche actor fervently airing support for a greater focus on multilingualism in Norwegian politics. Perhaps somewhat surprisingly, media representatives are not found among these top users, and accounts from politicians appear scattered throughout the figures. Both Figures 2 and 3 almost exclusively feature what could be described as "underdog" politicians (Larsson & Kalsnes, 2014)—those

VALG2013 Partiet LisaCharleneH jonhaugan dekristne HavardJohansen Nagulens OyvindRein PiratPartiet_No KSteigen stmarthinsen AtleSognli RadneyThomsen Yrkeskverulant RuneAaH

Figure 3. Users of 2013 election hashtags with more than 3 tweets per day on average (99 tweets or more during the period). Note. Number of tweets per user.

who did not enjoy incumbency, had roles outside government, or held comparably low-key individual positions within their respective party organizations. For 2011, we can point to Oterhaug and TomStaahle, both local representatives for the Progress Party (FrP). A bit further down, we find LarsMDG and Partiet. A green party politician operates the former, whereas the latter is the official party account of the same party. The only other official party account to be found in Figure 2 is Hoyre, representing the conservative party.

Similar tendencies of mostly minor political actors being active are also found in Figure 3. Indeed, the only representative of a major political party here is AtleSognli, regional politician for Labour (Ap). The Green party account, Partiet, is seen in the 2011 data and is also visible here. Other highly active political users for the 2013 election are dekristne (fringe Christian party) and PiratPartietNo (the Pirate party).

Taken together, the results presented in Figures 2 and 3 indicate that although prominent, high-level politicians did indeed make comparatively extensive use of Twitter during the two elections, considerable levels of tweeting were mostly associated with "underdog" actors. Such a claim appears valid regardless of type of actor—we could point to the aforementioned VALG2011 and VALG2013 special interest group accounts, attempting to push their specific agenda in the Twittersphere. Among political party-oriented accounts, we can conclude that the most ardent ones in this regard were operated by candidates or parties that are best described as minor or marginal. As such, although findings from other contexts seem to suggest otherwise (e.g., Strandberg, 2013; Vergeer & Hermans, 2013), the results

presented here give precedence to the findings presented by previous studies looking into the Norwegian political environment, suggesting that comparably smaller parties would be more keen in their employment of social media such as Twitter (Kalnes, 2009).

In sum, closer scrutiny of the relations between the most active and the remaining users shows a relatively even distribution, but more so in 2013 than in 2011. In terms of activity levels, Twitter remained marked by "underdogs" also during the 2013 election campaign. As such, the identification of these "underdog" politicians and parties seemingly refutes the normalization hypothesis, speaking to the previously mentioned cyclical nature of the hypotheses under scrutiny.

Leverage and Distribution of Political Actors

Turning to our second main analytical focus point, we seek to establish which political actors appear to enjoy the most spread or attention in the hashtagged communicative networks for each election. Although the volume of traffic emanating from specific accounts can provide details regarding the overall activities of top users, we here need to analyze the specific uses of Twitter for entering into dialogue (by means of @replies) and for redistributing messages (retweeting). The argument is made here that a user receiving ample amounts of @replies and retweets should be considered a popular one within a specified context, and therefore identifying such users is of relevance here.

Utilizing the Degree Range functionality available in Gephi to delimit our sample, we focus on the top users of

Figure 4. Top users of @replies in 2011 (to the left) and 2013 (to the left). Note. Degree range > 35.

Twitter-specific functionalities—starting with the @replies variety.

Figure 4 features a number of nodes, each representing a specific Twitter user. The color of the node is indicative of the degree to which each user has been active in sending @ replies—the darker the node, the more active the user— whereas the size of the node and label suggests the volume of @replies received by each user. Finally, the relative thickness of the lines between the nodes is indicative of the amount of traffic undertaken.

With these guidelines for interpretation in place, we build on previous work to sort the identified high-end users into three different, broad categories (Larsson & Moe, 2012, 2013). First, Senders are visible in Figure 4 as smaller, darker nodes, suggesting that although they appear quite active in sending @replies, they receive comparably few such messages. Second, Receivers show opposite tendencies—visible in the figure as comparably large, lighter colored nodes, these users receive plenty of messages, but are not as active in sending themselves. Third, Sender-Receivers can be characterized as more reciprocal in their approach to the @reply functionality, as they both send and receive a comparably large number of messages. They are depicted in Figure 4 with larger, darker nodes. Table 2 provides notable examples of identified user types, combining the approach discussed here with the classification scheme presented in conjunction with Figures 2 and 3.

The presence of anonymous users in the Sender category is felt for both years. In the case of one user (Pederen), he

appears to be taking on a similar role during both elections. Beyond this particular user group, we note the presence in 2011 of VALG2011, an account, as noted, operated in close ties with the online service

For users identified as Receivers, similar tendencies can be seen for both years. The presence of accounts operated by politicians and political parties is clear during both elections, but increased in 2013. In 2011, the politicians identified as receivers are notably well-known figures. As discussed above, this could be understood in terms of the latter election being more important for a variety of actors mostly operating on the national level—but also in terms of a general increase in Twitter use from 2011 to 2013. The fact that two accounts clearly related to mainstream media outlets (Mariesimonsen—a well-known political journalist and commentator for a national tabloid—and NRKvalg—the public service broadcaster's election coverage tag) receive comparably more tweets than they themselves send out is also noteworthy, as this could again be pointed to as an example of the influence of established, traditional media in a Twitter setting.

Finally, the category of Sender-Receivers also sees simi-lar—and in some cases, the same—accounts showing up for both studied years. The official account for the environmental party, Partiet, is visible for both years, whereas the Pirate Party account (PiratPartiet No) was active in this fashion only during the 2013 elections. Following up on the VALG2011 account mentioned earlier, VALG2013 is operated in relation to the same website, now updated for the

Table 2. Categorizations of Top @Reply Users.

User type

Senders MQueseth, Pederen (Anonymous) VALG20II (Citizen)

Receivers SVKristin, jensstoltenberg, Erna_Solberg, KAHareide (Political actors) Vampus, Orjaz

(Communication/PR) nicecap (Citizen)

Sender- Voxpopulinor (Citizen) receivers Stmarthinsen (Communication/ PR)

kjetilloset (Media) Partiet (Political actors)

Pederen, BaksidenavAP

(Anonymous) KristianVea, kagjerde (Citizen) Liberaleren (Media) KAHareide, Siv_Jensen_FrP, TrineSG, audunlysbakken, jensstoltenberg, erna_solberg, SVKristin, jonasgahrstore, Senterpartiet, KrFNorge, Arbeiderpartiet, Hoyre (Political actors)

Mariesimonsen, NRKvalg (Media) KSteigen (Citizen) Stmarthinsen (Communication/PR) VALG20I3 (Citizen) Partiet, PiratPartiet_No (Political actors)

HavardJohansen (Communication/ PR)

Note. NRK = Norwegian Public Service Broadcaster

latter election. As the node for the 2013 account is slightly larger than the one present in the 2011 data, the operator of these two accounts apparently saw a change in 2013, as the account gets more @replies directed to it than in 2011.

Using a similar mode of visualization as above, Figure 5 introduces the top users for the retweet functionality.

For the nodes in Figure 5, color is indicative of the degree to which the user has engaged in redistributing tweets sent by others—retweeting. Size, then, reveals the frequency with which each user has been retweeted. As before, the lines between the nodes represent the volume of traffic between the identified accounts.

Utilizing similar guidelines as discussed above, we can distinguish between three main groups of users. Retweeters, characterized by comparably smaller, darker nodes, are highly active in retweeting messages sent by other users, but enjoy comparably few retweets themselves. This characteristic is somewhat reversed for the Elites, who are represented as larger, lighter colored nodes—users who do not retweet others, but get their own messages redistributed to larger degrees. Finally, Networkers apparently take a more communal approach to the functionality under scrutiny. Visible in Figure 5 as comparably larger, darker nodes, they both retweet and are retweeted. Notable examples of users categorized in this regard can be found in Table 3.

First, although the rather few notable examples of Retweeters can be found, the presence of politically inclined users in 2013 suggests a difference from the previous election. Again, this could be seen as a result of the parliamentary election holding more weight than the regional and local ones. Beyond mammathessy, MartinovicEmma introduces herself as engaged in a local branch of the Social Democratic Youth Organization, whereas OyvindArum is employed as a communications advisor for the same party. As such, two of our three identified top retweeters can be understood as

"underdog" political actors (although OyvindArum is not employed as a politician, he works for the Labour party)— using Twitter's potential for redistribution to position themselves in a network of more established politicians. In this way, the practice of retweeting could be seen as an attempt to "piggy-back" along on issues identified by other actors, thereby presumably raising interest in one's own perspectives and standpoints. This is a tendency also found among "underdog" actors in other political contexts (Christensen, 2013).

Second, the makeup of the Elites category similarly appears to have changed between the two studied elections. Although media actors and politicians dominated the category in 2011, the 2013 roster of tweeters sees citizens and media celebrities having their hashtagged messages redistributed by others to a larger extent. The largest node visible in the 2013 data represents the user MarteRS, who introduces herself on her profile page as a trainee for a Norwegian banking organization. It might seem surprising to see a non-celebrity gain such traction. By looking closer at the traffic generating these results, we can see that her popularity is based on other users redistributing one of her tweets. Marte_ RS appears to have been the first using the specified hashtag to comment on a somewhat crude statement made on live television during election night by Progress Party Leader Siv Jensen, directed to Norwegian Prime Minister Jens Stoltenberg. This tweet,1 then, was the only one sent by the particular user to be picked up to this degree—but it still gave Marte RS a clear influence, at least in terms of being retweeted the most during election night.

Third, although the 2011 nodes representing Networkers are rather small compared with their 2013 counterparts, two users from the former election can still be singled out. The heading nicecap is used by a citizen who, through extensive use of Twitter, has managed to make a name for himself in the Norwegian political landscape. The characteristics of the VALG2011 have been discussed above. Overall, then, the two most notable networkers making their mark on Twitter traffic pertaining the 2011 election were citizens, suggesting that the role of the electorate might be limited to such a redistributing rather than creating function in this regard. This result is contrasted with the findings for the latter of the two elections. Here, the dominant Networkers are all political party accounts—whereas no accounts operated by individual politicians can be discerned in this regard. This is perhaps to be expected given the national focus of this particular election, but the fact that no individual politicians appear to have made extensive use of Twitter to the extent that they have taken on a networker role is worth noting.

From the analysis of Figures 4 and 5, we can identify a shift with regard to the types of actors getting the most attention in the hashtagged communication about Norwegian elections. Although we are only studying two time points here, such tendencies still merit attention. Politicians are present to a larger degree in 2013, although most leading

Figure 5. Top users of retweets in 2011 (to the left) and 2013 (to the left). Note. Degree range > 35.

Ole Berget







yn kietiltoset








5000 martingiuner


Maisws kietiilba


PiralPi@ieLNo PemilleTHuseby

googlegoesi^n g


FerdQwner SanneKviTies



elecltßnista P<Î0iet kjejjba WWI®lorge Afte




Thorn asCEallaksen

irijajssii Stja[jjg|ipp



entoQurerkittJiyy Hgjarte"l®stheim antiblnQgeren



, „__, . StiaffiCallaTi


C"* I jifiVcl


Toriii^jaigo SenteQiartiet


JanEv@nsen2 S^5

Table 3. Categorizations of Top Retweet Users.

User type 2011 2013

Retweeters Helgestad, JanFredrikB (Citizen) mammathessy (Citizen) MartinovicEmma (Political actors) OyvindArum (Communication/PR)

Elites Voxpopulinor (Citizen) Marte_RS, kimfyy, KSteigen

stmarthinsen (Communication/PR) (Citizen)

Partiet, Rotevatn (Political actors) Kvalshaug, TufteJo, StianBlipp

politiskno, kjetilloset, kjetilba (Media)


Networkers Nicecap, VALG20II (Citizen) Venstre, Partiet, PiratPartiet_No, Hoyre, Arbeiderpartiet, Senterpartiet (Political actors)

political actors found in Figure 5 are classified as Receivers, as addressees. For retweets, the corresponding figure indicates that although activity during the 2011 election was largely a mixed affair—politicians, journalists, bloggers, and so on taking part in the fray—the 2013 election saw more established political parties join their "underdog" competitors in adopting the role of Networkers, conspicuously using retweets to gain leverage in the Twittersphere. Again, this

could be seen as a result of a more important election being at hand, or as a case of organizational learning or profession-alization (Tenscher, 2013), with parties having tried out Twitter during the 2011 election and using the service more fully during the latter election.

The 2013 election also saw the potential that one well-timed tweet can have for an otherwise unknown user. The example of MarteRS and her comment sent during election night can serve as an example that anonymous Twitter enthusiasts can indeed make their voice heard—even if, as in this case, the popularity must be understood as a "one-off' rather than as a continuous phenomenon for the user at hand.


The starting point for this article was the aim of contributing to our understanding of how novel online tools are used for political communication in a comparative, diachronic fashion. Our conceptual framework was made up of the two opposite hypotheses—normalization and equalization— which allowed us to identify, map, and discuss changes in uses of Twitter over two different election campaigns.

Our analytical foci points were selected in a contextual setting where the sheer number of users and tweets had risen

considerably from the 2011 election campaign period to the events in 2013. In isolation, this speaks to the rising popularity of Twitter, as well as to the increased use of hashtags, and signals a more central place for Twitter among other online communication platforms. Moreover, the fact that regional elections, such as the one held in 2011, generally draw scarce amounts of attention when compared with national elections, such as in 2013, should be acknowledged. Therefore, the insights presented here must be understood with this caveat put firmly in place: We are indeed comparing Norwegian elections, but different types of Norwegian elections. Nonetheless, the results presented here speak to the general tendencies of Twitter use for political purposes, as well as point to the differences between the often-studied national context and the relative dearth of insights regarding local or regional uses of novel information and communication technologies (e.g., Lisi, 2013). Regardless of electoral focus, Twitter use is still dwarfed by other, somewhat similar ser-vices—mainly Facebook. Moreover, it remains an elite medium, primarily understood as used by an urban, well-educated "twitterati." As such, changes in employment patterns and who these users actually are (or at least claim to be) are interesting questions for researchers to continue to ask.

In sum, we did not find clear aspects of change in accordance with the normalization thesis: Although patterns of retweet usage (from Elites to Networkers) changed between the elections, we found no clear-cut overall growth in terms of volume and use of Twitter-specific communicative modes by dominant political actors. Although more established political parties and candidates can rely on a more or less steady supply of opportunities to convey their message across more traditional media outlets, minor actors arguably do not have easy access to such channels. As such, it might be suitable for minor actors to invest in "alternative tools for delivering their message" (Lisi, 2013, p. 272)—a tendency that seems particularly valid for the Environmental party, as pointed to above. As environmental parties elsewhere were known during the "web 1.0" era for innovative uses of the Internet (e.g., Gibson, 2004), perhaps the results presented here could be understood in terms of such priorities repeating themselves also in the supposed "web 2.0" phase of the web. With regard to the comparably widespread presence of this particular party in the material presented here, we could point to the findings as supportive of the equalization hypothesis, suggesting that less established political actors would make innovative use of novel technologies. At least in terms of activity levels, Twitter was still in 2013 an arena for "underdog" politicians and parties, a finding that seems to refute the normalization hypothesis, thus speaking to the previously mentioned cyclical nature of the hypotheses under scrutiny (e.g., Gibson & McAllister, 2014).

The network analysis allowed us to see the development over time of another aspect of Twitter use. Specifically, the employment of Twitter-specific modes of communication was found to have risen among major parties, which would

serve as a support for the normalization hypothesis. Yet, the dominant national politicians were still to a large degree absent from the hashtagged dialogical communication—a finding that could underline again the relative lack of importance ascribed to Twitter during an election campaign, even in 2013.

Finally, the results indicated that the often-proposed potential for an anonymous citizen to gain leverage and attention in a politically themed online setting cannot be completely disregarded (e.g., Bekafigo & McBride, 2013; Bruns & Highfield, 2013). However, we should not expect unrealistic transformations of public discussion in this regard. As shown above, although the user Marte_RS only produced one tweet baring relevant hashtags, this one tweet—featuring a simple, straightforward "one-liner" type of a message, one might add—secured this particular user the central, most influential node in a network map of retweets as presented in the rightmost part of Figure 5. The chain of events associated with this particular tweet gaining high amounts of traction clearly must be regarded a "one-off' of sorts. Yet, this finding nevertheless serves as a subtle reminder of the participatory potential of social media—any and all users do indeed have the potential to make a name for himself or herself, be it through a series of messages sent— or by means of a what could best be described as viral fluke.

Be that as it may, our current efforts cannot make substantive inroads into the mind of the individual Twitter user— querying them on their motivations, ideas, and expectations in relation to these types of Twitter use. As suggested by Lomborg and Bechmann (2014), the type of data presented here "say very little about the meanings that users ascribe to their social media use" (p. 260). As such, although the "overall picture" provided here is certainly of use, varying forms of more qualitative inquiry are duly needed to provide different perspectives. In relation to this, we should encourage scrutiny of the content political social media use. Given language obstacles and the need for an analysis embedded in cultural contexts, such studies might be even more challenging to conduct as comparisons. Yet, to give insights into how political actors actually behave when they use social media, we need to take up the challenge.

In conclusion, tendencies of both studied hypotheses were unearthed in our analyses, giving some merit to the previously discussed claims made regarding a "middle road" between the two extremes of normalization and equalization. Although smaller actors are indeed making themselves heard in the political Twittersphere, their voices are largely overshadowed by more established or well-to-do competitors— most likely due to the presumed larger amount of resources available to the latter group. Nevertheless, an interesting tendency was found regarding the environmental party, who appear to follow in the proverbial digital footsteps of previous subscribers to this ideology (e.g., Gibson, 2004). Given the media attention or hype surrounding many of these online novelties, perhaps smaller political parties can gain indirect

mass media attention by attempting to gain journalistic interest through fervent online activity. Future research could be of use here, to delve deeper into how minor parties and politicians plan out and execute campaign efforts on different platforms.

Although this study has focused on comparing two elections, future comparative work will surely be able to shed more light on the further influences of the "established political commentariat" (Bruns & Highfield, 2013, p. 672) with regard to electoral social media practices. Moreover, the study at hand has provided comparative insights into political Twitter use in conjunction with elections with a focus on the structure of the communication taking place. Although it could be feasible to use the current approach to provide further insights over time, future efforts might find it equally useful to take the content of what is being tweeted into account.

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.


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

1. Available at 823655425, accessed August 21, 2014.


Aardal, B., Krogstad, A., & Narud, H. M. (2004). I valgkampens hete: strategisk kommunikasjon og politisk usikkerhet [In the heat of the election campaign: Strategical communication and political uncertainty]. Oslo, Norway: Universitetsforlaget. Ausserhofer, J., & Maireder, A. (2013). National politics on Twitter.

Information, Communication & Society, 16, 291-314. Bastian, M., Heymann, S., & Jacomy, M. (2009, May 17-20). Gephi: An open source software for exploring and manipulating networks. Paper presented at the Third International ICWSM Conference, San Jose, CA. Bekafigo, M. A., & McBride, A. (2013). Who tweets about politics? Political participation of Twitter users during the 2011 gubernatorial elections. Social Science Computer Review, 31, 625-643.

boyd, d., & Crawford, K. (2012). Critical questions for big data.

Information, Communication & Society, 15, 662-679. Brans, A. (2011). How long is a tweet? Mapping dynamic conversation networks on Twitter using Gawk and Gephi. Information, Communication & Society, 15, 1323-1351. Brans, A. (2013). Faster than the speed of print: Reconciling "big data" social media analysis and academic scholarship. First Monday, 15(10). Retrieved from index.php/fm/article/view/4879 Brans, A., & Burgess, J. (2011a). #Ausvotes: How Twitter covered the 2010 Australian federal election. Communication, Politics & Culture, 44(2), 37-56.

Brans, A., & Burgess, J. (2011b). Gawk scripts for Twitter processing. Retrieved from

Bruns, A., & Highfield, T. (2013). Political networks on Twitter. Information, Communication & Society, 16, 667-691.

Bruns, A., & Stieglitz, S. (2013). Towards more systematic Twitter analysis: Metrics for tweeting activities. International Journal of Social Research Methodology, 16, 91-108.

Christensen, C. (2013). Wave-riding and hashtag-jumping: Twitter, minority "third parties" and the 2012 US elections. Information, Communication & Society, 16, 646-666.

Coleman, S. S., & Blumler, J. G. (2009). The Internet and democratic citizenship: Theory, practice and policy. New York, NY: Cambridge University Press.

Dahlgren, P. (2005). The Internet, public spheres, and political communication: Dispersion and deliberation. Political Communication, 22, 147-162.

Enli, G. S., & Skogerb0, E. (2013). Personalized campaigns in party-centred politics. Information, Communication & Society, 16, 757-774.

Freelon, D. G. (2010). Analyzing online political discussion using three models of democratic communication. New Media & Society, 12, 1172-1190.

Gibson, R. K. (2004). Web campaigning from a global perspective. Asia-Pacific Review, 11(1), 95-126.

Gibson, R. K., Lusoli, W., & Ward, S. (2008). Nationalizing and normalizing the local? A comparative analysis of online candidate campaigning in Australia and Britain. Journal of Information Technology & Politics, 4, 15-30.

Gibson, R. K., & McAllister, I. (2014). Normalising or equalising party competition? Assessing the impact of the web on election campaigning. Political Studies. Advance online publication. doi:10.1111/1467-9248.12107

González-Bailón, S. (2013). Social science in the era of big data. Policy & Internet, 5(2), 147-160.

Graham, T., Broersma, M., Hazelhoff, K., & van 't Haar, G. (2013). Between broadcasting political messages and interacting with voters. Information, Communication & Society, 16, 692-716.

Graham, T., Jackson, D., & Broersma, M. (2014). New platform, old habits? Candidates' use of Twitter during the 2010 British and Dutch general election campaigns. New Media & Society. Advance online publication. doi:10.1177/1461444814546728

Halavais, A. (2014). Structure of Twitter: Social and technical. In K. Weller, A. Bruns, J. Burgess, M. Mahrt, & C. Puschmann (Eds.), Twitter and society (pp. 29-42). New York, NY: Peter Lang.

Hermans, L., & Vergeer, M. (2013). Personalization in e-campaign-ing: A cross-national comparison of personalization strategies used on candidate websites of 17 countries in EP elections 2009. New Media & Society, 15, 72-92.

Hilson, M. (2008). The Nordic model. Scandinavia since 1945. London, England: Reaktion Books.

Jackson, N. A., & Lilleker, D. G. (2009). Building an architecture of participation? Political parties and Web 2.0 in Britain. Journal of Information Technology & Politics, 6, 232-250.

Kalnes, 0. (2009). Norwegian parties and Web 2.0. Journal of Information Technology & Politics, 6, 251-266.

Karlsen, R. (2010). Fear of the political consultant: Campaign professionals and new technology in Norwegian electoral politics. Party Politics, 16, 193-214.

Kirk, J., & Miller, M. L. (1986). Reliability and validity in qualitative research. Beverly Hills, CA: SAGE.

Klinger, U. (2013). Mastering the art of social media. Information, Communication & Society, 16, 717-736.

Larsson, A. O. (2013). "Rejected bits of program code": Why notions of "Politics 2.0" remain (mostly) unfulfilled. Journal of Information Technology & Politics, 10, 72-85.

Larsson, A. O., & Kalsnes, B. (2014). "Of course we are on Facebook—"Social media adoption in Swedish and Norwegian parliaments. European Journal of Communication.

Larsson, A. O., & Moe, H. (2012). Studying political microblog-ging: Twitter users in the 2010 Swedish election campaign. New Media & Society, 14, 729-747.

Larsson, A. O., & Moe, H. (2013). Twitter in politics and elections—Insights from Scandinavia. In A. Bruns, J. Burgess, K. Weller, C. Puschmann, & M. Mahrt (Eds.), Twitter and society ,319-330. New York, NY: Peter Lang.

Larsson, A. O., & Svensson, J. (2014). Politicians online— Identifying current research opportunities. First Monday, 1P(4).Retrieved from article/view/4897

Lewis, S. C., Zamith, R., & Hermida, A. (2013). Content analysis in an era of big data: A hybrid approach to computational and manual methods. Journal of Broadcasting & Electronic Media, 57, 34-52.

Lilleker, D. G., & Jackson, N. A. (2010). Towards a more participatory style of election campaigning: The impact of Web 2.0 on the UK 2010 general election. Policy & Internet, 2(3), 67-96.

Lilleker, D. G., Koc-Michalska, K., Schweitzer, E. J., Jacunski, M., Jackson, N., & Vedel, T. (2011). Informing, engaging, mobilizing, or interacting: Searching for a European model of web campaigning. European Journal of Communication, 26, 195-213.

Lisi, M. (2013). The professionalization of campaigns in recent democracies: The Portuguese case. European Journal of Communication, 28, 259-276.

Lomborg, S., & Bechmann, A. (2014). Using APIs for data collection on social media. The Information Society, 30, 256-265.

Margolis, M., & Resnick, D. (2000). Politics as usual: The cyberspace "revolution." London, England: SAGE.

Prior, M. (2007). Post-broadcast democracy: How media choice increases inequality in political involvement and polarizes elections. New York, NY: Cambridge University Press.

Statistics Norway. (2014). Elections. Retrieved from http://www.

Strandberg, K. (2009). Online campaigning: An opening for the outsiders? An analysis of Finnish parliamentary candidates' websites in the 2003 election campaign. New Media & Society, 11, 835-854.

Strandberg, K. (2013). A social media revolution or just a case of history repeating itself? The use of social media in the 2011 Finnish parliamentary elections. New Media & Society, 15, 1329-1347. doi:10.1177/1461444812470612

Tenscher, J. (2013). First- and second-order campaigning: Evidence from Germany. European Journal of Communication, 28, 241-258.

Vaage, O. F. (2012). Norsk mediebarometer 2011 [Norwegian media barometer 2011]. Oslo: Statistisk sentralbyra (Statistics Norway).

Vergeer, M., & Hermans, L. (2013). Campaigning on Twitter: Microblogging and online social networking as campaign tools in the 2010 general elections in the Netherlands. Journal of Computer-Mediated Communication, 18, 399-419.

Whitley, R. (2000). The intellectual and social organization of the sciences. Oxford, UK: Oxford University Press.

Wright, S. (2012). Politics as usual? Revolution, normalization, and a new agenda for online deliberation. New Media & Society, 14, 244-261.

Zimmer, M., & Proferes, N. (2014). A topology of Twitter research: Disciplines, methods, and ethics. Aslib Proceedings, 66, 250-261.

Author Biographies

Anders Olof Larsson is postdoctoral research fellow at the

Department of Media and Communication, University of Oslo.

Hallvard Moe is professor at the Department of Information

Science and Media Studies at the University of Bergen.