Coherent clusters or fuzzy zones Understanding attention and structure in online political participation
First Monday

Coherent clusters or fuzzy zones  Understanding attention and structure in online political participation by Anders Olof Larsson

Social media and their uses are in an almost constant flux, and the need for comparative approaches — across platforms and time points — appears as urgent. The study at hand presents a dual comparative approach looking into political communication as undertaken on social media. Presenting data from Twitter and Instagram use during the 2013 and 2017 Norwegian elections, the study traces developmental tendencies and suggests terminology with which to assess the ways that these activities are undertaken on the studied platforms. Results indicate that while Twitter and Instagram activity was rather differently fashioned in terms of structure and attention in 2013, these activities had grown more similar — focused on political elites rather than a broader range of users — in 2017. As such, the study argues for a normalized view of online political participation, wherein professionalized political actors appear to increasingly orchestrate the studied activities.


Hashtags beyond the ‘Twittersphere’
Study design
Discussion and conclusions




With empirical foci ranging from discussion forums to hashtagged tweets, a considerable amount of scholarly work has gauged the potential for participation in online settings. Often working from a deliberative ideal of participation and ‘publicness’ — wherein multi-actor deliberation between representatives from all ‘walks of life’ is held in high regard — results have been mixed as to the potential of the Internet to increase or indeed decrease participatory practices. Tendencies in comparably recent studies looking into these as well as related issues has suggested that online milieus are increasingly characterized by normalization (Kjeldsen, 2016) — hypothesizing that rather than leading to a democratization of the public sphere, the influx and continued use of new, digital media will reinforce pre-existing structures of participation and power. The study presented here, then, engages with the issue of online deliberation by means of empirical analysis of two online fora — Twitter and Instagram — from two different political events — the 2013 and 2017 Norwegian parliamentary elections. Specifically, we will focus on gauging issues of attention — what types of users emerge as particularly important in these fora — and structure — how these users relate to each other to create “networks of political communication” [1].

A great deal of scholarly work has looked into the uses of social media during political events such as those studied here. However, the majority of efforts in this regard have focused on Twitter. Indeed, a series of research reviews looking detailing the research field of online political communication have suggested that Twitter “is overrepresented in studies of networked communication” [2]. In the same vein, Lomborg (2017) argues that the broader field of social media study is characterized by “a focus on the uniqueness of the single service” and by “a tendency of seeing each service in isolation” [3]. With this in mind, the present study seeks to expand the field by means of a dual comparative approach — comparing participation across social media as well as across elections. With respect to the first comparative aspect, while Twitter is often pointed to as a key platform for online political participation in several contexts (e.g., McGregor, et al., 2017), Instagram is a relative newcomer and caters to a somewhat different — younger — audience in our case country of Norway (Ipsos/MMI, 2017). As such, comparing social media platforms will help enrichen our knowledge of what types of users emerge as particularly poignant in these different fora. For the second comparative aspect, the study features a diachronic approach in that it assesses activities undertaken on the studied platforms across two national elections. Indeed, Kreiss (2014) suggests that “media practice should be viewed as contextual and subject to change as political events unfold” [4]. Thus, such an over-time approach appears as suitable since it moves beyond “single-shot case studies” [5] and allows for insights into developmental aspects of online communication. Based on these suggestions made by previous research, the present study is guided by two research questions. Dealing first with the roles of different users, we ask: what types of users emerge as successful in gaining attention across elections and social media platforms? Second, we are interested in assessing how these users group themselves together in networks of attention-giving, and thus ask: How do the networks created by these users differ across elections and social media platforms? In complementing the often-studied Anglo-American contexts, the study presented here makes a contribution to our understanding of contemporary communication processes by presenting results spread over time, across platforms, in a small yet advanced context.



Hashtags beyond the ‘Twittersphere’

The country studied here — Norway — can perhaps easiest be distinguished as one of the Scandinavian welfare states characterized by a moderate pluralistic parliamentary system. Often reported as featuring high degrees of Internet use in general and social media use in particular, the Norwegian context features high levels of citizen adoption of novel platforms (Ipsos/MMI, 2017). In addition, high consumption of older media formats (Vaage, 2017) lay the groundwork for a decidedly hybridized media environment (Chadwick, 2013). The selected context could thus be considered as highly relevant for a study such as the one presented here. The results provided could also serve as useful for comparison with other, dissimilar contexts, teasing out differences regarding the development of social media use across political contexts.

Hashtags — thematic keywords that allow users of both Twitter and Instagram to find posts related to their specific interests — were utilized as starting points for the study at hand. Including hashtags in messages sent on either platform allows for users to partake in public conversations about specific events, potentially reaching outside of their own networks of followers and joining the “virtual loungeroom” [6]. While hashtags could be considered as temporary publics or communities, political varieties such as the ones under scrutiny here have been suggested to be particularly persistent over time (e.g., Kjeldsen, 2016; Romero, et al., 2011). At least for Twitter, then, the diachronic approach featured here was deemed suitable. Given the dearth of previous, similar work on Instagram, the research field is currently arguably not in a position to gauge the longevity of hashtags on the specified platform. Besides the specific theme of the study at hand, the empirical design championed here and described in the section following this one aims to make a contribution also in this more general regard.

Although the often-employed hashtag approach to studying social phenomena allows us to “filter messages that users posted with the clear intention of contributing to the political discourse” [7] it has certain limitations. Hashtag use has been seen as presupposing a certain level of user skill or experience when it comes to social media use (e.g., McKelvey, et al., 2014). Moreover, any message sent that might have been relevant for the topic at hand but that did not feature the studied hashtags were effectively omitted, resulting in a loss of posts that might have been relevant to include (Lorentzen and Nolin, 2017). Nevertheless, the hashtag-based approach was deemed as suitable as the alternative of searching for non-hashtagged keywords would likely result in a plethora of unrelated posts being gathered (Jungherr, 2014b).

While hashtags are a common feature on both Twitter and Instagram, the two social media services nevertheless differ with regards to other, general functionalities “ but also when it comes to what we know regarding their use in Norway. As previously alluded to, Twitter has been studied extensively in the context under scrutiny (e.g., Karlsen and Enjolras, 2016; Larsson and Ihlen, 2015; Larsson and Moe, 2013). These studies have often found Twitter employment to be primarily related to a “Twitterati” (Bruns and Burgess, 2011) of urban, elite and media-savvy users, featuring citizens with clear access to public life, such as journalists, politicians and PR consultants. Because of the connection between Twitter and users holding such societal positions, the specified platform has been pointed to as particularly important to study in order to gauge contemporary — elite — communication trends and structures (e.g., Chadwick, 2013).

While Twitter thus holds a specific status in Norwegian society, the other platform under scrutiny here — Instagram — has a less obvious position in terms of public deliberation. Comparably novel, the user base for this latter service has nevertheless grown larger than that enjoyed by the former. While 5.2 per cent of Norwegians over 18 years of age use Twitter on a daily basis, nearly half of citizens in the same age group reported to have an Instagram account, with around 55 percent of account holders using the service daily (Ipsos/MMI, 2017). This popularity has primarily manifested itself among younger citizens, which could be the reason for its apparent growing popularity among those up for election in recent Norwegian elections (Larsson, 2017) — as well as in political events elsewhere (e.g., Filimonov, et al., 2016). Thus, while Twitter use has been characterized as being by and for societal elites, Instagram appears to be primarily used by comparably younger cohorts. Given these differing characteristics in combination with the increased importance of social media services for political campaigning in the studied country (e.g., Enli and Skogerbø, 2013), the argument is made that the comparative efforts featured in the study at hand will allow for interesting insights into overtime developments regarding online political participation.



Study design

Researchers embarking on studies based on social media data — such as the one at hand — need to familiarize themselves with the limitations often placed on data access by the platforms under scrutiny. Data gathering from both services entail querying different application programming interfaces (APIs), a method that has been described as ‘black boxing’ certain details of the process (e.g., Bruns, 2011). As such, it is important to describe the data collection processes undertaken with a certain degree of detail.

For Twitter, the relative ease with which its APIs can be accessed has led to its mentioned popularity among researchers. For this study, Twitter data from the 2013 election (hashtags #valg13 and #valg2013 — Norwegian for #election13 and #election2013) had previously been collected by means of the yourTwapperKeeper service (please refer to Larsson and Moe [2014] for further details about the original use of the 2013 data) — for a long time considered “the preferred tool for capturing #hashtag or keyword tweets” [8]. For the 2017 data (hashtags #valg17 and #valg2017), the Digital Methods Initiative Twitter Capture and Analysis Toolset (DMI-TCAT) was employed (Borra and Rieder, 2014). As the Twitter activity generated by the hashtags under study was judged to be within the boundaries of what tools such as the ones employed can offer access to (e.g., Driscoll and Walker, 2014; Giglietto and Selva, 2014), these approaches were deemed suitable.

As for Instagram, the broader research field largely lacks precedence — at least in comparison with Twitter. Much like for Twitter, a series of tools for data collection have come and gone (e.g., Rieder, 2015). A tangible limitation of data collection possibilities occurred in 2016 when Instagram revised their API policies. While free-of-charge access to the Instagram API have since become difficult, commercial entities were up until recent time available that could provide similar services. One such enterprise — now defunct — was employed for the study at hand (MagiMetrics, 2017). While relatively little is known about the specific inner workings of MagiMetrics, the completeness of the exported data spanned across both elections and was assessed by a series of manual Instagram searches for the queried hashtags (see above). For both elections and both platforms, data was gathered during a one-month period preceding each election — a period often seen as characterized by higher degrees of activity of the type we are gauging here (e.g., Gaber, 2017).

As mentioned, our analytical focus is placed on the structure of hashtagged activity as well as on what users emerge as enjoying comparably higher amounts of attention across platforms and elections. For attention, we follow Bruns and Burgess (2012) and use the indegree measurement available in the network analysis software Gephi (Bastian, et al., 2009). Essentially, the size of the node representing each user in the network graphs to follow will correspond the amount of attention they’ve enjoyed. While attention can be provided and defined in a number of different ways on the two platforms under scrutiny, and while such modes of attention-giving are subject to change [9], these processes commonly involve the ‘@’-character (Marwick and boyd, 2010). While mentions utilizing the ‘@USERNAME’ standard is practiced on both platforms, Twitter also offers an integrated function to redistribute — retweet — messages originally provided by other users. For such a retweeted message, the ‘@USERNAME’ standard is included as well, so as to indicate the original poster. Instagram, however, did not offer an integrated solution for the redistribution of messages at the time of data collection. Indeed, while “regramming” is possible, it entails a series of manual practices or ‘add-ons’ to the basic platform. Such approaches tend include an encouragement to indicate the original poster — again, often by means of aforementioned ‘@USERNAME’ expressions. As such, larger nodes will indicate comparably popular users in this regard — while nodes with diminutive sizes denotes the contrary. Given these guidelines, we can talk of attention-giving as being more centralized or more decentralized. For the former, attention by means of the @ character emerges as concentrated to relatively fewer (thus larger) nodes, whereas in the latter, a relative multitude of smaller nodes would suggest more evenly distributed attention-giving.

For structure, we assess how the nodes — the users — are grouped based on who they give attention to. Specifically, the community detection algorithm modularity (Newman, 2006) will be employed to visualize how users group themselves based on their attention-giving practices. These groups of users will be identified by different shades visible in the graphs presented later. By focusing on visual inspection, characteristics of structure will be easier to assess — are we seeing overlapping groups of users — indicating attention being given across groups — or are these groupings more aptly described as being largely independent of each other — a finding that would indicate limited common ground across groups? Of course, these delimitations within and between different groupings of users are seldom clear-cut, with some researchers suggesting “fuzzy” borders between groups [10] and others employing terms like “zones” rather than using the often-used term “clusters” — the latter term “suggesting too much coherence” [11]. Building on these suggestions, we can thus think of the groups that emerge from the Gephi graphs as either zones that are characterized by more overlapping or indeed fuzzy borders — or as clusters, characterized by delimited structures.

Integrating our perspectives on attention and structure as detailed above, Table 1 presents four ideal types of networks.


Table 1: Four ideal types of networks characterized by structure (zones or clusters) and attention (centralized or decentralized).
StructureZoneslarger nodes, overlapping structuresmaller nodes, overlapping structure
Clusterslarger nodes, delimited structuresmaller nodes, delimited structure


For example, a network of users characterized by a few users receiving plenty of attention (larger nodes) with little overlap between groups of users (delimited structure) would be referred to as centralized clusters, while a graph featuring a series of comparably smaller nodes the activities of which place them in a series of overlapping groups would be labeled as characterized by decentralized zones. These principles for analysis will be employed in the following chapter and will assist us in assessing the developments evident from the 2013 to the 2017 elections.




While 59,933 tweets were archived for the 2013 elections, the 2017 amount had dropped to 20,085. It is difficult to say exactly why such a dearth occurs — while previous studies comparing Twitter use across elections in similar environments found the volume of hashtagged tweets to increase (Larsson and Moe, 2016), perhaps the aforementioned declining popularity of the service at hand is related to this initial result. For Instagram, a diverging but weak trend can be discerned. Here, the 2013 hashtags yielded 5,782 posts, while the latter of the studied events resulted in a total of 6,133 posts. While the circumstances regarding data collection as described above makes it difficult to provide any steadfast claims regarding the statistical significance of these numerical differences, this initial comparison nevertheless suggests a shift in terms of social media attentiveness to political affairs.

Employing the terminology introduced in the previous section, we now move on to assess the activity undertaken in terms of structure and attention on the two studied platforms across both elections. In order to facilitate comparison between platforms and elections, the same ‘starting point’ for graph construction was employed — using the Force Atlas 2 algorithm and the Giant Component filter and utilizing the same ranges for node sizes to visualize the degree to which users received attention (as noted, based on the indegree measurement — min size: 10, max size: 200). Color, then, is based on modularity class — essentially, nodes featuring the same shade belong together based on shared links to the same nodes. As the number of users and therefore connections between users will vary across elections and platforms, four different figures will be presented — two for Twitter, two for Instagram. We will first look at the developmental tendencies found for Twitter.

Twitter — From decentralized zones to centralized zones

Figures 1 and 2 present the network charts depicting activity from the 2013 and 2017 elections respectively.


Hashtagged Twitter activity during the 2013 Norwegian elections
Figure 1: Hashtagged Twitter activity during the 2013 Norwegian elections.
Note: Larger version of Figure 1 available here.



Hashtagged Twitter activity during the 2017 Norwegian elections
Figure 2: Hashtagged Twitter activity during the 2017 Norwegian elections.
Note: Larger version of Figure 2 available here.


In terms of structure, the groups visible in both Twitter graphs (discernible by differing shades assigned by the modularity class algorithm) emerge as overlapping rather than featuring delimited clusters of users. Indeed, the nodes making up the identified groups of uses intersect and integrate, suggesting that Twitter featured activity spanning across a series of groups during both elections. Although the structure essentially remains the same when moving from 2013 to 2017, we can identify some differences regarding attention. Specifically, while no clearly dominating nodes can be identified as enjoying the bulk of attention in the 2013 data, the presence of several large nodes in the 2017 depiction suggests a development towards more centralized patterns. While some users enjoyed considerable amounts of attention in 2013 — thereby resulting in comparably larger nodes in Figure 1 — such larger varieties of nodes emerge as more dominating for the 2017 Twitter graph. As such, while both graphs feature zone-like, overlapping groups, attention within these zones appear to have moved from a decentralized to a more centralized variety, characterized by larger nodes — popular users — receiving the majority of attention.

While a multitude of users — too many to describe in detail — are identifiable by their node sizes in the two Twitter graphs, we can point to some differences with regards to how attention appears to shift from one election to another. The largest zone for the 2013 data can be identified in the upper left section of Figure 1. While political actors can be found within this zone (social democrat party account arbeiderpartiet as well as their corresponding youth organization aufnorge), the two main nodes here — ksteigen and kvalshaug — both belong to media professionals. Similarly, in the zone visible to the lower left of Figure 1, we can identify tuftejo, representing Norwegian comedian Bård Tufte Johansen. This focus on media professionals and celebrities to some extent confirm the findings regarding a “Twitterati” (Bruns and Burgess, 2011) gaining attention on the platform at hand. However, this attention given to media professionals and even citizens (as visible by the size of the marte_rs node, representing a telecom professional) visible in the same cluster as tuftejo) appears to shift when we move from 2013 to 2017. While a zone centered around the Environmental party (partiet) is visible in both graphs, the 2017 network is clearly dominated by constellations related to political party accounts (such as the aforementioned arbeiderpartiet, their party leader jonasgahrstore in addition to their main competitor, the conservative hoyre party and their leader erna_solberg). While the zones related to partiet during both elections emerge as consisting largely of party members and others engaged in environmental issues, the other zones visible in Figure 1 but primarily in Figure 2 do not appear as ideologically coherent as the previously discussed partiet varieties. Instead, these zones are made up of users subscribing to different political agendas — a result suggesting contact being made over ideological lines.

Instagram — From centralized clusters to decentralized zones

Figures 3 and 4, then, depict the activities undertaken on Instagram in the 2013 and 2017 elections respectively.


Hashtagged Instagram activity during the 2013 Norwegian elections
Figure 3: Hashtagged Instagram activity during the 2013 Norwegian elections.
Note: Larger version of Figure 3 available here.



Hashtagged Instagram activity during the 2017 Norwegian elections
Figure 4: Hashtagged Instagram activity during the 2017 Norwegian elections.
Note: Larger version of Figure 4 available here.


As Instagram activity was undertaken to smaller extents across both elections compared to Twitter, and as we are employing the same algorithmic settings across all figures as mentioned previously, both Instagram graphs appear as sparser than those depicting Twitter activity. Regardless of such sparsity, the overtime tendency for Instagram suggests a development from clusters of users (considering the comparably delimited structures of nodes visible in Figure 3) to a depiction more akin to the Twitter varieties discussed previously for the 2017 Instagram data — indeed, Figure 4 feature overlapping tendencies, suggesting a zone-like structure as defined in Table 1.

As for the users that yield higher amounts of attention within these structures, some interesting discrepancies can be noted when comparing Instagram to Twitter. First, while the Twitter graphs representing activity from 2013 and 2017 showed political actors to be present, other types of users — media professionals, pundits, celebrities and even citizens — could be discerned for both elections. By contrast, the data presented for Instagram in Figures 3 and 4 appear as less diverse in this regard, as virtually all of the dominant nodes in both Figures represent political parties or their leaders. Similarly, while the figures depicting Twitter activity during the two studied elections suggested a centralization of activity to a few larger nodes when moving from 2013 to 2017, the graphs representing Instagram activity seem to suggest the opposite.

Second, these mainly politically professional users also relate to each other in different ways on Instagram than on Twitter. For the 2013 elections especially, the Instagram clusters shown appear as more ideologically coherent than the Twitter zones. For the specified election, this can be exemplified by focusing on the lower left side of Figure 3, where a cluster of social democratic users emerge, circling their activities around the party account arbeiderpartiet, then party leader jensstoltenberg and youth organization aufnorge. On Twitter, these and other users emerged as intertwined across ideological borders. The results presented for Instagram, then, suggest a more delimited mode of interaction.

As the structure of Instagram use evolves from centralized clusters to decentralized zones from 2013 to 2017, so does the coherence of the identified groupings evolve towards more overlapping patterns of attention-giving. While exceptions can be found (consider, for example, the party-party leader pairings of social democrats arbeiderpartiet and jonasgahrs or conservatives hoyre and erna_solberg found to the right of Figure 4), hashtagged Instagram activity undertaken during the 2017 election is clearly different in this regard when compared to the earlier election. Consider, for instance, the middle section of Figure 4, suggesting a zone wherein accounts operated by liberal party, centre party senterparti, christian democrats krfnorge and right-wing populists fremskrittspartiet all emerge as centers of attention. Indeed, these zones and their contents arguably paints a very different picture when compared to the clearly delimited clusters shown in the 2013 Instagram depiction.

Finally, while the environmental party appeared as rather isolated on Twitter, the patterns related to their Instagram account — degronne — indicates a more open approach to attention-giving on the service under discussion here, especially for the 2013 elections. Such diverging use patterns could be seen as related to the oft-reported willingness of ‘green’ or indeed environmental parties to adopt original ways of utilizing previously novel online technologies as shown in a series of contexts (Vergeer, et al., 2011).



Discussion and conclusions

Esser and Hanitzsch (2012) point out that a comparative approach in communication research offers “a valuable tool for advancing our understanding of communication processes, and [...] opens up new avenues of systematic research” [12]. The study presented here adopted a double comparative design — across elections as well as across social media services — in order to assess the changing nature of online political discussion. Detailing structures of participation and patterns of attention-giving, the study has built on previous scholarship and developed terminology that can hopefully be useful for future research efforts on activities undertaken on social media platforms — also beyond the political settings studied here. Using these techniques and terminologies, the results presented in the paper at hand serves as examples of the differences, similarities and indeed developmental tendencies that can be teased out when utilizing a comparative research design. What follows in this final section is a discussion of some of the key aspects of the empirical results.

First, for Twitter, the change in attention-giving from decentralized to centralized characteristics must be considered as one of the key developments uncovered — especially when one considers the drop of overall Twitter use from the 2013 to the 2017 elections. Essentially then, fewer messages are being sent, and these messages generally tend to circle around a decreasing cast of users being mentioned. Specifically, while popular users identified for 2013 included a comparably wider representation of users belonging to “the chattering classes” [13] — societal elites such as journalists, pundits and politicians — the centralized results emanating from the 2017 election data see the bulk of attention provided in relation to various political actors — parties and party leaders. Indeed, while non-elite actors succeeded to make an impact in 2013 (consider the marte_rs node visible in Figure 1), this opportunity appears to have vanished in 2017. Although the tweet sent by marte_rs must be considered a ‘one-off’ — the user was commenting on a rather crude utterance by right wing populist party leader Siv Jensen during election night — this possibility of non-elite users to yield influence, to make their voices heard has always been intimately tied in with novel technologies whether digital or not (Brecht, 2001; Polat, 2005). The data presented here, then, suggest that at least for the Norwegian case, such opportunities for regular citizens are decreasing. This could be due to the diminishing popularity of Twitter, a platform that, while characterized by elites, has consistently featured previously unknown users being lifted from relative obscurity as a result of their activities on the platform (Moe and Larsson, 2012). Second, for Instagram, the results presented in Figures 3 and 4 suggested developments from centralized clusters of users in 2013 to decentralized zones in 2017. Regardless of structure, the attention given on Instagram emerges as almost uniquely focused on political elites — party and party leader accounts — during both studied elections. The service thus differs from Twitter, where different types of users succeeded to gain a place in the proverbial spotlight — especially for 2013 — as discussed earlier.

The combined results for Twitter and Instagram can be understood along the lines of the aforementioned normalization hypothesis. Indeed, as political parties and leaders come to dominate the attention-gaining game on both platforms and over time, as the influences of other elites appear to diminish (if indeed it ever was present) and as non-elites appear to be playing similarly diminishing roles in the studied processes, social media like the ones studied here appear to be increasingly mimicking the characteristics of public debate in general — focused on elites. While the supposed dynamic between the normalizing of political activity and the competing equalizing perspective (suggesting, among other things, increased influence by smaller, non-elite actors) has been shown to be cyclical (Gibson and McAllister, 2015), the results presented here suggest that over time, the normalized perspective holds more water. This dominance of established parties and party leaders can also be seen as related to processes of political professionalization. A broad term with many interpretations (e.g., Lilleker and Negrine, 2002), the definitions available nevertheless often feature some reference to organized use of the Internet (e.g., Farrell, 1996; Lisi, 2013; Strömbäck, 2007). Taking this perspective, the results presented here support the comparably early claim by Mancini that parties are transforming into “communication machines” orchestrated by “skilled professionals who are able to use all of the opportunities offered by technological innovation” [14]. As previous studies from Norway (Kalsnes, 2016) as well as from other contexts (Kreiss, 2014) has shown an increasing awareness of the need for centrally organized online activities, the focus on political actors found here could be the result of political professionals going about their business, orchestrating supporters to boost the mentioning rates represented in the Figures presented previously. While the innermost workings of the APIs guiding the services under scrutiny remains largely unknown, rallying supporters to give attention to a specific party or party leader in the suggested way is likely to have a positive effect on the visibility of the specified account. Such strategies of attention-giving in order to increase visibility has previously been suggested in relation to minor, often extremist parties and political groups (e.g., Larsson, 2014). The results presented here, then, could be interpreted as suggesting that such techniques have now been adopted by more mainstream political actors. Future research can perhaps shed more light on the prioritizations of political parties when it comes to online campaigning, perhaps building on the insights provided here.

Assessing these results within the broader context of media and communication studies, we might ask ourselves where the previously hashtagged discussions and citizen activity move now that we can conclude that such activities on Twitter — arguably a platform previously used for such purposes to a comparably large extent — appears to be dwindling. While it is beyond the scope of our current efforts to make firm claims about if and if so where these types of activities take place, some interesting tendencies can be noted regarding these issues as they have played out in Norway. Specifically, the role of Facebook as a platform for debate about societal issues appears to have increased following popular discussions in legacy media outlets about ‘what debaters to follow’ on the specified platform, as well as news stories about the discussions taking place on the personal profile pages of some of these recommended debaters (Hansen and Aanstad, 2016). As of this writing, such lists of recommended profiles to follow tend to be largely made up of established opinion leaders such as journalists and other media personalities — users who “routinely appropriate the genres of social media sites and hybridize these with their preexisting routinized, professional practice” [15]. Granted, while some such high-profile users and supposed debate curators — such as pundit and musician Kjetil Rolness (Aldridge, 2016) — emphasize that their Facebook presences and the debates contained therein are open for the public to engage with, the differences between engaging in a discussion hosted under a hashtag on the one hand and on a profile page belonging to an established societal actor on the other needs to be further scrutinized. For now, the tendencies uncovered in the empirical results presented here as well as in the popular debates paraphrased briefly above do indeed indicate a concentration of attention towards societal elites, and that such elites to a higher extent appear to have a hosting and perhaps even controlling role over debates. The oft-repeated “capacity of social media for giving voice to ordinary citizens’ interests, opinions and critique” [16] appears to be decreasing, with such activities instead largely centered around various types of opinion leaders and media elites. Future research should look more clearly into the roles and functionalities of such hosts or indeed potential gatekeepers of online discussion. How do these supposed gatekeepers view their own roles and responsibilities as de facto hosts of public discussion? As traditionally influential actors succeed to navigate the novel media landscape and emerge with clear and important roles to play it is useful to remember Karlsen and Enjolras’ suggestion that “although power is not disrupted, it is in some ways transformed” [17]. While comparably novel actors also succeed in making their respective marks in these contexts, older media logics and the types of institutions primarily related to such logics still appear to dominate.

In conclusion, the repercussions following a decrease of hashtag use for the possibility of citizens to engage in political or indeed other types of discussion was brought up previously. Such an apparent reduction also carries with consequences for scholars who are interested in the matters dealt with here. If, as discussed above, public discussion and debate is shifting from being organized around thematic hashtags to circling around accounts and profile pages hosted by certain key users, this development poses challenges not only to the ways that debate is structured and possibly also who feels empowered to take part. From a scholarly standpoint, such a shift carries with it a series of ethical and methodological issues — too plentiful to be discussed at length here (e.g., Zimmer, 2010; Zimmer and Proferes, 2014) — that must be taken into account by those interested in contributing to our increased common understanding of the issues discussed here. For instance, while the hashtag criterion for ‘publicness’ employed here is rather common in these types of research projects, issues of data openness, data access and end-user expectations about privacy must be considered as these tendencies develop. As the history of these and other services clearly shows us, online platforms, apps and the ways that these services are employed by end users are in an almost constant flux (Lomborg, 2017). As the ways that such online uses change, so must the scholarly community strive to adapt their methods and approaches in order to efficiently study these uses. The current study has drawn on suggestions by previous research in order to systematically map the ways that social media are used across elections and platforms, thus attempting to consolidate research approaches in what must be considered a sprawling field of scholarship. As a community, engaged researchers should indeed allow diverse research efforts and approaches to blossom and come to fruition, but in order to avoid developing into a “fragmented adhocracy” (Whitley, 2000), those interested should make efforts to consolidate their approaches. Such efforts will hopefully help us identify a series of what Lomborg (2017) refers to as “mainline research trajectories for social media studies” [18] where interested researchers would ideally contribute in ways that cumulatively build on what came before. Relatedly, it would appear that the ephemeral nature of these platforms make up the core problem with regards to bringing such efforts forward. Indeed, Lomborg (2017) suggests that “researchers of social media seem to accept change, rather than continuity, as a condition for the study of social media” [19]. While the technical changes related to social media platforms are largely out of the hands of the research community, efforts can nevertheless be undertaken to reach more continuous, cumulative modes of scholarly conduct. The study at hand, for instance, used the @ character as a symbol of attention-giving across time periods and platforms. Identifying and focusing on such commonalities across platforms could be one way forward for the research field. As technical commonalities will undoubtedly change, it might instead be more suitable to detail the common affordances of social media across platforms, time periods and contexts (e.g., Evans, et al., 2017; Larsson, 2017). Such an approach — although obviously not the only possible one — could serve as a fruitful road forward for empirically-driven social media research. End of article


About the author

Anders Olof Larsson is Professor at Kristiania University College in Oslo, Norway.
E-mail: andersolof [dot] larsson [at] kristiania [dot] no



1. Bruns, 2011, p. 1,332.

2. Enli, 2017, p. 3.

3. Lomborg, 2017, p. 7.

4. Kreiss, 2014, p. 1,477.

5. Jungherr and Theocharis, 2017, p. 102.

6. Harrington, et al., 2013, p. 405.

7. Jungherr, 2014a, p. 244.

8. Bruns, 2011, p. 1,333.

9. Lomborg, 2017, p. 7.

10. Lorentzen, 2016, p. 7.

11. Gerlitz and Rieder, 2013, p. 7.

12. Esser and Hanitzsch, 2012, pp. 3–4.

13. Kalsnes, 2016, p. 158.

14. Mancini, 1999, p. 243.

15. Chadwick, 2013, p. 13.

16. Lomborg, 2017, p. 10.

17. Karlsen and Enjolras, 2016, p. 17.

18. Lomborg, 2017, p. 12.

19. Lomborg, 2017, p. 7.



Ø Aldridge, 2016 “Kjetil Rolness: — Facebook har vært viktig når det gjelder å gjøre den jobben som norske medier ikke har gjort,” Aftenposten (23 July), at, accessed 6 July 2018.

E. Borra and B. Rieder 2014. “Programmed method: Developing a toolset for capturing and analyzing tweets,” Aslib Journal of Information Management, volume 66, number 3, pp. 262–278.
doi:, accessed 24 July 2019.

B. Brecht, 2001. “The radio as a communications apparatus,” In: M. Silbermann (editor). Brecht on film and radio. London: Methuen, pp. 40–48.

A. Bruns 2011. “How long is a tweet? Mapping dynamic conversation networks on Twitter using Gawk and Gephi,” Information, Communication & Society, volume 15, number 9, pp. 1,323–1,351.
doi:, accessed 24 July 2019.

A. Bruns and J. Burgess, 2012. “Researching news discussion on Twitter: New methodologies,” Journalism Studies, volume 13, numbers 5–6, pp. 801–814.
doi:, accessed 24 July 2019.

A. Bruns and J. Burgess, 2011. “#Ausvotes: How Twitter Covered the 2010 Australian federal election,” Communication, Politics and Culture, volume 44, number 2, pp. 37–56.

A. Chadwick, 2013. The hybrid media system: PoIItics and power. Oxford: Oxford University Press.

K. Driscoll and S. Walker 2014. “Working within a black box: Transparency in the collection and production of big Twitter data,” International Journal of Communication, volume 8, pp. 1,745–1764, and at, accessed 24 July 2019.

G. Enli, 2017. “New media and politics,” Annals of the International Communication Association, volume 41, numbers 3–4, pp. 220–227.
doi:, accessed 24 July 2019.

G.S. Enli and E. Skogerbø, 2013. Personalized campaigns in party-centred politics: Twitter and Facebook as arenas for political communication,” Information, Communication & Society, volume 16, number 5, pp. 757–774.
doi:, accessed 24 July 2019.

F. Esser and T. Hanitzsch, 2012. “On the why and how of comparative inquiry in communication studies,” In: F. Esser and T. Hanitzsch (editors). The handbook of comparative communication research. London: Routledge, pp. 3–22.

S.K. Evans, K.E. Pearce, J. Vitak and J.W. Treem 2017. “Explicating affordances: A conceptual framework for understanding affordances in communication research,” Journal of Computer-Mediated Communication, volume 22, number 1, pp. 35–52.
doi:, accessed 24 July 2019.

D.M. Farrell, 1996. “Campaign strategies and tactics,” In: L. LeDuc, R.G. Niemi and P. Norris (editors). Comparing democracies: Elections and voting in global perspectives. London: Sage, pp. 160–183.

K. Filimonov, U. Russmann U and J. Svensson, 2016. “Picturing the party: Instagram and party campaigning in the 2014 Swedish elections,” Social Media + Society (9 August).
doi:, accessed 24 July 2019.

I. Gaber, 2017. “Twitter: A useful tool for studying elections?” Convergence, volume 23, number 6, pp. 603–626.
doi:, accessed 24 July 2019.

C. Gerlitz and B. Rieder, 2013. “Mining one percent of Twitter: Collections, baselines, sampling,” M/C Journal, volume 16, number 2, at, accessed 24 July 2019.

R.K. Gibson and I. McAllister, 2015. “Normalising or equalising party competition? Assessing the impact of the Web on election campaigning,” Political Studies, volume 63, number 3, pp. 529–547.
doi:, accessed 24 July 2019.

F. Giglietto and D. Selva, 2014. “Second screen and participation: A content analysis on a full season dataset of tweets,” Journal of Communication, volume 64, number 2, pp. 260–277.
doi:, accessed 24 July 2019.

S.S. Hansen and K.H. Aanstad, 2016. “Disse debattantene bør du flge på Facebook,” Aftenposten (6 July), at, accessed 6 July 2018.

S. Harrington, T. Highfield and A. Bruns, 2013. “More than a backchannel: Twitter and television,” >Participation, volume 10, number 1, pp. 405–409, and at, accessed 24 July 2019.

A. Jungherr, 2014a. “The logic of political coverage on Twitter: Temporal dynamics and content,” Journal of Communication volume 64, number 2, pp. 239–259.
doi:, accessed 24 July 2019.

A. Jungherr, 2014b. “Twitter in politics: A comprehensive literature review” (27 February), at, accessed 24 July 2019.

A. Jungherr and Y. Theocharis, 2017. “The empiricist’s challenge: Asking meaningful questions in political science in the age of big data,” Journal of Information Technology & Politics, volume 14, number 2, pp. 97–109.
doi:, accessed 24 July 2019.

B. Kalsnes, 2016. “The social media paradox explained: Comparing political parties Facebook strategy versus practice,” Social Media + Society (17 May).
doi:, accessed 24 July 2019.

R. Karlsen and B. Enjolras, 2016. “Styles of social media campaigning and influence in a hybrid political communication system: Linking candidate survey data with Twitter data,” International Journal of Press/Politics, volume 21, number 3, pp. 338–357.
doi:, accessed 24 July 2019.

L. Kjeldsen, 2016. “Event-as-participation: Building a framework for the practice of ‘live-tweeting’ during televised public events,” Media, Culture & Society, volume 38, number 7, pp. 1,064–1,079.
doi:, accessed 24 July 2019.

D. Kreiss, 2014, “Seizing the moment: The presidential campaigns’ use of Twitter during the 2012 electoral cycle,” New Media & Society, volume 18, number 8, pp. 1,473–1,490.
doi:, accessed 24 July 2019.

A.O. Larsson, 2017. “Top users and long tails: Twitter and Instagram use during the 2015 Norwegian elections,” Social Media + Society (15 June).
doi:, accessed 24 July 2019.

A.O. Larsson, 2014. “Everyday elites, citizens, or extremists? Assessing the use and users of non-election political hashtags,” MedieKultur. Journal of Media and Communication Research, volume 30, number 56, pp. 61–78.
doi:, accessed 24 July 2019.

A.O. Larsson and Ø. Ihlen, 2015. “Birds of a feather flock together? Party leaders on Twitter during the 2013 Norwegian elections,” European Journal of Communication, volume 30, number 6, pp. 666–681.
doi:, accessed 24 July 2019.

A.O. Larsson and H. Moe, 2016. “From emerging to established? A comparison of Twitter use during Swedish election campaigns in 2010 and 2014,” In: A. Bruns, G. Enli, E. Skogerb, A.O. Larsson and C. Christensen (editors). Routledge companion to social media and politics. London: Routledge, pp. 311–324.

A.O. Larsson and H. Moe, 2014. “Triumph of the underdogs? Comparing Twitter use by political actors during two Norwegian election campaigns,” SAGE Open (8 December).
doi:, accessed 24 July 2019.

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

D.G. Lilleker and R. Negrine, 2002. “Professionalization: Of what? Since when? By whom?” >Harvard International Journal of Press/Politics, volume 7, number 4, pp. 98–103.
doi:, accessed 24 July 2019.

M. Lisi, 2013. “The professionalization of campaigns in recent democracies: The Portuguese case,” European Journal of Communication, volume 28, number 3, pp. 259–276.
doi:, accessed 24 July 2019.

S. Lomborg, 2017. “A state of flux: Histories of social media research,” European Journal of Communication, volume 32, number 1, pp. 6–15.
doi:, accessed 24 July 2019.

D.G. Lorentzen and J. Nolin, 2017. “Approaching completeness: Capturing a hashtagged Twitter conversation and its follow-on conversation,” Social Science Computer Review, volume 35, number 2, pp. 277–286.
doi:, accessed 24 July 2019.

P. Mancini, 1999. “New frontiers in political professionalism,” Political Communication, volume 16, number 3, pp. 231–245.
doi:, accessed 24 July 2019.

A.E. Marwick and d. boyd, 2010. “I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience,” New Media & Society, volume 13, number 1, pp. 114–133.
doi:, accessed 24 July 2019.

S.C. McGregor, R.R. Mourão and L. Molyneux, 2017. “Twitter as a tool for and object of political and electoral activity: Considering electoral context and variance among actors,” Journal of Information Technology & Politics, volume 14, number 2, pp. 154–167.
doi:, accessed 24 July 2019.

K. McKelvey, J. DiGrazia and F. Rojas, 2014. “Twitter publics: How online political communities signaled electoral outcomes in the 2010 US House election,” Information, Communication & Society, volume 17, number 4, pp. 436–450.
doi:, accessed 24 July 2019.

H. Moe and A.O. Larsson, 2012. “Twitterbruk under valgkampen 2011,” Norsk Medietidsskrift, volume 19, number 2, pp. 151–162, and at, accessed 24 July 2019.

M.E.J. Newman, 2006. “Modularity and community structure in networks,” Proceedings of the National Academy of Sciences, volume 103, number 23 (6 June), pp. 8,577–8,582.
doi:, accessed 24 July 2019.

R.K. Polat, 2005. “The Internet and political participation: Exploring the explanatory links,” European Journal of Communication, volume 20, number 4, pp. 435–459.
doi:, accessed 24 July 2019.

J. Strömbäck, 2007. “Political marketing and professionalized campaigning: A conceptual analysis,” Journal of Political Marketing, volume 6, numbers 2–3, pp. 49–67.
doi:, accessed 24 July 2019.

M. Vergeer, L. Hermans and S. Sams, 2011. “Is the voter only a tweet away? Micro blogging during the 2009 European Parliament election campaign in the Netherlands,” First Monday, volume 16, number 8, at, accessed 24 July 2019.
doi:, accessed 24 July 2019.

R. Whitley, 2000. The intellectual and social organization of the sciences. Second edition. Oxford: Oxford University Press.

M. Zimmer, 2010. “‘But the data is already public’: On the ethics of research in Facebook,” Ethics and Information Technology, volume 12, number 4, pp. 313–325.
doi:, accessed 24 July 2019.

M. Zimmer and N.J. Proferes, 2014. “A topology of Twitter research: Disciplines, methods, and ethics,” Aslib Journal of Information Management, volume 66, number 3, pp. 250–261.
doi:, accessed 24 July 2019.


Editorial history

Received 15 March 2019; accepted 24 July 2019.

Creative Commons License
This paper is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

‘Coherent clusters’ or ‘fuzzy zones’ — Understanding attention and structure in online political participation
by Anders Olof Larsson.
First Monday, Volume 24, Number 8 - 5 August 2019

A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2019. ISSN 1396-0466.