Social TV and the WWE: Exploring the fan-to-brand relationship in a highly engaged,live-viewing,interactive online space

Abstract

Social television (TV) engagement has become more commonplace as viewers seek alternative ways of engaging with TV shows and other viewers. This is especially true with televised professional wrestling; 119,506 tweets were analyzed using social network analysis during the four World Wrestling Entertainment telecasts. Results show that brand-affiliated users primarily interact among one another and not the fans themselves, despite fans reaching out to the brand, resulting in significant social stratification and low interactivity within the community. The findings suggest that when fans think they are able to join and contribute to the brand’s ongoing conversation, those fans might still be highly motivated to communicate with the brand, even if the brand does not reciprocate.

Keywords

Consumer engagement social network analysis social TV television viewers Twitter

As social media continue to evolve, so do the ways in which people use them to enhance other media consumption experiences. Nowhere is this more common than with social television (TV), the simultaneous phenomenon of engaging with real-time TV broadcasts through social media platforms (Lin et al., 2016). This is especially true with live, televised World Wrestling Entertainment (WWE) programs, two of which are consistently ranked at the top of the Nielsen Social Content ratings, the metric that evaluates program-related social TV measurement across Twitter, Facebook, and Instagram (Nielsen, 2019). In fact, two of the WWE’s weekly live shows featuring scripted narratives and matches between their wrestlers (aka Superstars) – ‘WWE Raw’ and ‘WWE Smackdown’ – together consistently dominate weekly social content ratings. During these two telecasts, fans have the opportunity to engage with the WWE brand, its affiliates, and each other, in a firestorm of Instagram and Facebook posts and tweets through Twitter.

Online fandom in the digital age, including through social media, has itself been studied in a variety of contexts, including celebrity gossip among teenage women (Redmond and Cinque, 2017), as a connection between musician and fan (Cinque, 2015) and as a re-imagined construct of fandom itself in a Web 2.0 society (Hills, 2013). In the world of the WWE, there exists an enormity of interactive engagement between fans and the WWE brand and its affiliates in real time during these telecasts, a phenomenon that goes beyond fandom and entertainment value. Here, through Twitter, WWE fans have a front row seat along with thousands of other like-minded viewers who share a common connection – interacting online about the evening’s telecast. It is this real-time consumer engagement that makes the WWE more than just a series of televised live events. It speaks to the power of online exchanges as a marketing tool that was not possible before social media became the juggernaut it is today. Why social TV ratings for these televised WWE shows are so far out in front of other programs whose viewers also engage in social TV (Nielsen Weekly Top 10) is a question worth exploring to understand how engagement works in the universe of wrestling fans and the brands they engage with. Using Twitter as the social media platform of analysis, and through the lens of online consumer engagement, this study seeks to explore the tripartite relationship between consumers (WWE fans), the WWE organization, and its affiliates (the wrestlers and other affiliated accounts) and understand how the organization has somehow been able to harness the power of these fans through these real-time online interactions.

Review of the literature

World Wrestling Entertainment

World Wrestling Entertainment, Inc., more commonly known as WWE, is an entertainment and media organization that is most widely known for professional wrestling, much of which is televised live. Two of these televised programs are the weekly ‘WWE Smackdown!’ and ‘WWE Raw’, both of which broadcasted on the USA Network during the time of this study, December 2018 (‘WWE Smackdown’ has since moved to the Fox Network, while ‘WWE Raw’ remains on USA). Both of these programs consistently rank in the Nielsen weekly top 10 series and specials for social content ratings. For example, during the weeks ending on December 16 and 23, 2018, when data were collected for this study, ‘WWE Smackdown!’ and ‘WWE Raw’ ranked #1 and #2, respectively, both weeks, with a total two-week combined social media interactions of 5.034 million across Twitter, Facebook, and Instagram. Comparatively, the third-ranked program both weeks was NBC’s ‘The Voice’, each week with approximately 800,000 interactions (Nielsen Social Content Ratings, 2018).

Consumer engagement online

Throughout marketing research, the concept of consumer engagement in branded communities – those which are ‘based on a structured set of social relations among admirers of a brand’ (Muniz and O’Guinn, 2001) – has been conceptualized in various ways, with the underlying notion that the manifestation of customer-based approaches in engaging with brands has trended toward behaviors that go beyond legacy metrics such as purchase (and repurchase) performance (van Doorn et al., 2010), to a more dynamic brand interaction (Araujo et al., 2020). Both positive and negative consumer engagement can be the result of customer brand exchanges; however, valence with lower customer brand involvement can have a stronger effect on customer satisfaction than with those more highly involved (Brady et al., 2006), possibly providing support for levels of brand engagement as an independent variable on customer fulfillment.

Notably, participants in online brand-related conversations do not merely interact with the brand alone. Rather, they engage with the brand as well as fellow consumers in a large-scale discursive space, negotiating their own role within the network of interactions among users (Barger et al., 2016; Maslowska et al., 2016). Consequently, individuals can distinguish themselves from their peers through their contributions to the discourse. For instance, some prolific users might make a large number of contributions to forcibly command attention by dominating the conversation, while others might make a smaller number of especially insightful comments that inspire responses or that their peers share to a broader audience, thereby carrying their voices to a uniquely wide population (e.g. Britt et al., 2020; Cho et al., 2015). Either way, such individuals are able to rise to prominence as a result of their submissions, adopting a central role in the conversation such that their future contributions receive heightened attention and they are therefore positioned to exert especially great influence over others’ thoughts and ideas about the topic at hand.

The practice of connecting to like-minded others online is certainly not a new concept. The levels of access, however, have changed quite a bit since the computer-assisted groupminds that grew out of teleconferencing. Usenet’s computer bulletin-board systems, Programmed Logic for Automatic Teaching Operations’ (PLATO) computer-based education platform, Turoff’s Electronic Information Exchange System, and the Advanced Research Projects Agency Network (ARPANET), for example, have been considered testing grounds for what would become known more commonly as computer-mediated communication. These tools would ultimately allow citizens to help each other solve problems (Rheingold, 1993). As the Internet became more mainstream, and modern online social media emerged, users began to rely on these spaces for sharing their own feelings, building relationships, and living out both personal and intimate parts of their lives, which can result in what Ito et al. (2010) refer to as intimacy interactions (Kennedy, 2016), connecting not only divergent individuals, but those with similar interests and ideals.

As well, these technologies and newer ways of communicating raise questions as to the benefits afforded to organizations from a marketing perspective, particularly as many consumers interact more and more through their mobile devices, engaging with companies, including through social media (Grewal and Stephen, 2019). The ability to influence consumer behavior through social media might also depend on how these interactions take place – who, for example, is contributing content (customer or organization), and which types of posts might be more effective than others (Grewal et al., 2019). As Araujo et al. (2020) points out, brands and consumers interact continuously in a variety of different contexts providing value for consumers as well as the brand in the short- and long term; therefore, it is paramount that brands understand how engagement behaviors impact their online community immediately and in the future. One thing is clear, though: Online engagement allows consumers to interact not only with a brand directly but also with other consumers, like minded or otherwise, incorporating both purchase and nonpurchase behaviors (Libai, 2011). Inside these online communities, not only is information exchanged, but friendships are developed among those with common interests, resulting in reference groups of people who can influence individual behaviors (de Valck et al., 2009).

Bagozzi and Dholakia (2002), in conceptualizing participation in online communities, found two determinants of what they refer to as ‘intentional social action’-- positive anticipated emotional desires, and social identity. They suggest that the virtual communities’ interactions share two goals – functional (exchange of information) and hedonic (positive experiences through these virtual interactions). Hayes and King (2014), further, found evidence that consumers share brand content for both functional and hedonic purposes in the social media viral advertising context.

Some research has focused specifically on online engagement related to fans of TV fiction programs (Baym, 2000; Booth, 2008; Bury, 2016; Lacalle and Simelo, 2017). For example, Lacalle and Simelo (2017) studied online activity of feminine online social TV communities of fictional programming and found that active online contributors defined themselves through their own emotional ties with the programs, often basing their activity on their own relationship with the narrative storylines.

Brodie et al. (2013) conceptualized a consumer/customer engagement model in an online, virtual brand community that spans from nonengaged to highly engaged. Their working definition of engagement in such communities identified five themes: (1) a fundamental theme which recognizes the importance of the interactive experiences between consumers and other actors in the brand network; (2) consumer engagement as a context-dependent, motivational state of engagement at a specific point in time; (3) that transient engagement states happen within broader engagement processes; (4) engagement as a multidimensional model made up of cognitive, emotional, and behavioral elements; and (5) recognition of the central role of consumer engagement in the relational exchange.

Gummerus et al. (2012) examined how customer engagement behaviors affect perceived relationship benefits and outcomes in a Facebook brand community. Their findings distinguished between two types of engagement behaviors, community and transactional: Community behaviors include actions such as liking, reading, and commenting and transactional behaviors include actions more akin to exchanging money for goods and services. Moreover, they proposed that along with satisfaction and loyalty associated with this engagement, customers can experience different relationship benefits, such as entertainment.

Nambisam and Baron (2007) studied virtual customer environments (VCEs) – organizations’ online domains that provide services to brand communities – and found that customer views of the benefits of interactions with a product shaped future interactions with a brand, thus having an effect on attitude toward the product. They suggested that the level of positive experiences consumers have with these VCEs can have the same positive effect on the customer brand relationship, making the virtual interactive experience as crucial to the relationship as their off-line one.

To be sure, customer-based approaches to connecting with brands have become more dynamic, as computer-mediated tools have progressed toward allowing consumers to more easily interact with each other, and not necessarily just the brands, building relationships that go beyond legacy customer-to-brand interactions. With TV viewing, virtual co-viewing – that is engaging with others through smartphones and other connected devices about what they are watching – has elevated those connected experiences (Anderson and Jiang, 2018; Nee and Barker, 2019). Now more than ever, shared experiences in these more highly engaged environments permit consumers to not only be influenced by the brand, but by each other. Thus, and in the case of entertainment TV, these connections between fans have pushed the engagement envelope to a point where brand communities can be conceptualized less as a two-way discussion between customer and brand and more as a fluid series of relationships that also keeps the consumers engaged with each other through these virtual environments.

Social TV

The emergence of smartphones as second screens – the term used to describe an additional device (typically a smartphone or tablet) used simultaneously while watching TV (Weimann-Saks et al., 2020) – and the increased popularity of social media platforms have given way to a recent convergence between the two – social TV. Simply put, social TV allows viewers to engage with each other and a program about the televised content through social media platforms such as Twitter, Facebook, and Instagram (Nielsen Social), all while synchronously watching the same TV programs as they are broadcast in the real time (Quintas-Froufe and González-Neira, 2014). This gives ordinary audiences access to the public stage (Salomaa and Lethinen, 2018) by providing a simultaneous backchannel during a telecast, essentially virtually connecting viewers with each other, as well as at times the program itself, in real time. This can provide crucial insight to audience reactions to specific, temporal moments of a telecast in ways legacy viewership measurements, such as Neilsen household ratings for audience size and demographic composition, cannot match (Harrington et al., 2013). This social soundtrack provides a supplement to the larger screen, where viewers can share their own reactions with others, often synchronously (Nagy and Midha, 2014).

Nielsen, through their own metric – Social Content Ratings – analyzes TV-generated social media engagements such as likes, comments, and reposts across Twitter, Facebook, and Instagram. This allows content producers and advertisers alike the ability to gauge TV audiences beyond legacy viewership measures, creating a more active TV audience profile (Nielsen Social, n.d.).

According to Nielsen (2018), in the United States, 45% of respondents use their digital device either always or very often while watching TV. As well, 71% look up information related to the TV content they watch. The use of Twitter alongside TV shows adds a new dimension to the audience experience, providing an alternative channel for engaging with other like-minded viewers (Harrington et al., 2013). Studies examining the relationship between like- and disparate-minded contributors to social media platforms have been carried out on a number of polarizing issues such as political views (Himelboim et al., 2013; Rathnayake and Suthers, 2019), scientific evidence (Cote and Darling, 2018), and gun policy (Merry, 2016). Much of this research focuses on the echo chamber effect, whereby individuals routinely engage with those who support their views, versus divergent views.

Additionally, social TV activity studies have included constructs such as virtual co-presence of other viewers (Gross et al., 2008; Hwang and Lim, 2015), viewer engagement (Pynta et al., 2014), levels of interactivity (Gantz and Lewis, 2014), social gratification (Kramer et al., 2015), expression of viewer opinions (Giglietto and Selva, 2014), connectedness to a larger community, (Shirra et al., 2014) need for belonging (Cohen and Lancaster, 2014), seeking others with similar interests (Wohn and Na, 2011), and, of course, engagement with others about a show (Han and Lee, 2014). Nagy and Midha’s (2014) ‘earned audience’ approach – those viewers exposed to Tweets about TV shows and its sponsors – uncovered that not only are audiences engaged with program content, but also brand sponsors within the show. This adds a marketing dimension to their findings, and they delineate between two different types of social TV audiences: those responding to the program (and advertising) and those merely exposed to the conversations.

Research questions

Drawing from the literature on social TV and consumer engagement, the following research questions are presented to further explore the phenomena of online engagement between WWE fans and each other, as well as the WWE brand and its affiliates overall:

Research Question 1: What types of brand engagement relationships are formed among different types of users?

Research Question 2: To what extent do highly prolific contributors engage with each other versus with comparatively casual contributors?

Research Question 3: How does the network self-organize to differentiate types of users?

Methods

Data collection

Twitter-based social TV conversations occurring during the live broadcasts of ‘WWE Raw’ (December 10 and 17, 2018, 8:00–11:00 p.m., ET) and ‘WWE Smackdown’ (December 11 and 18, 2018, 8:00–10:00 p.m., ET) were collected and analyzed for this study. This time frame represents two episodes of each show and was chosen to avoid irregularities in the conversation owing to special events (e.g. pay-per-view events). First, all four telecasts were recorded and then content analyzed to identify hashtags used within the show. These hashtags were displayed as on-screen graphics. Then, for each episode, the queries consisted of hashtags plus tweets authored by the brand accounts or its affiliates, as well as tweets engaging with the brand accounts or its affiliates. For the ‘WWE Raw’ telecast on December 10, the hashtags were #WWE, #WWEUniverse, #WWERAW, #RAW, #TagTeamTitles, and #WWETLC; for the ‘WWE Raw’ telecast on December 17, the hashtags were #WWE, #WWE Universe, #WWERAW, #RAW, #Fatal4Way, and #GauntletMatch; for the ‘Smackdown’ telecast on December 11, the hashtags were #WWE, #WWEUniverse, #Smackdown, #SDLive, and #WWETLC; for the ‘Smackdown’ telecast on December 18, the hashtags were #WWE, #WWEUniverse, #Smackdown, #SDLive, and #WomensTitle. Additionally, variations of these hashtags were included in the search to account for differences in how they were spelled in the tweets.

Crimson Hexagon (now Brandwatch Consumer Research), a subscription-based social media data library that provides access to the full Twitter firehose, was used to gather all tweets occurring during the broadcasts via a Boolean key word search. Key words were identified in two steps. First, open coding of the four broadcasts within the chosen time frame was administered to collect hashtags promoted with the shows. Second, Twitter handles for the WWE brand (@WWE) as well as individual Twitter handles for all characters associated with both telecasts were gathered via the WWE website. These key words were used to gather and export tweets made during the respective live broadcasts. Exported data from each show were then imported into R 3.5.1 using code adapted from the cRimson package (Jolley, 2016) and were subsequently combined into a single data set representing all 119,506 relevant tweets (i.e. tweets engaging with or mentioning either the WWE or the broadcasts themselves such that they were retrieved via the aforementioned query) made during this period.

User type classification

The collection of Twitter users in this community was classified into three fundamental user types based on their objectively defined roles in relation to WWE. The first group, ‘Brand’, included only the official WWE Twitter account, @WWE. The second group, ‘Affiliates’, consisted of 126 other Twitter user handles that were affiliated with the ‘WWE Raw’ or ‘WWE Smackdown’ programs, primarily including wrestlers participating in relevant WWE events as well as a few other affiliated accounts (such as @WWEUniverse) that were distinct from the core @WWE brand handle. The complete list of user handles in the ‘Affiliates’ group is provided in Appendix 1. Finally, all users who did not fall into either the ‘Brand’ or ‘Affiliates’ groups, who therefore did not represent the brand or current participants in either of the two relevant programs, were classified as ‘Fans’.

Network construction

There are three primary ways in which one Twitter user can directly engage with another in a public tweet: by replying to or retweeting a prior tweet made by that other user or by mentioning the other user. As such, three directed weighted networks were constructed representing these three dyadic communication mechanisms: a reply network, a retweet network, and a mentions network, with individual users represented as vertices in each network. In the reply network, an edge was formed between two vertices if one user replied to a tweet made by the second. The number of times that the first user replied to the second was used as the weight of the edge pointing from the vertex representing the first user to the one representing the second.

The retweet network was constructed in the same way, but with edges representing instances in which the first user retweeted the second. Similarly, edges between vertices in the mentions network represented cases in which the first user mentioned the user handle of the second in either a reply or an original tweet. (Retweets were excluded from the mentions network, as they do not constitute original content submitted by the poster, so they would not necessarily represent a deliberate attempt to mention another specific user by name.)

These three networks were constructed using R code adapted from the cRimson package (Jolley, 2016). Finally, since different users are likely to have different preferred methods of Twitter engagement, an omnibus network was created that combined the vertices and edges from the reply, retweet, and mention networks, to represent the entire communication system within the community surrounding these WWE broadcasts.

Analysis

To assess RQ1, which dealt with the tendency for different types of users to interact with one another, crosstab tables were created showing the number of edges in each network (reply, retweet, mentions, and omnibus) created between each user type (from the brand to the affiliates, from the brand to fans, etc.). These tables were used to describe the interpersonal connections forged among and across different user types.

The extent to which prolific contributors to the community engaged with one another as opposed to other users (RQ2) was evaluated using the aforementioned crosstabs in conjunction with the assortativity coefficient (Newman, 2002) for all users in each of the four networks. Prolificness was operationalized as the number of relevant tweets made by a given user during the data collection period. Consequently, pairwise similarity was assessed using the number of tweets made by the pair of users connected by each edge.

The various roles that different users adopted in the community (RQ3) were identified using a cluster analysis. Three standard network metrics were calculated for each user in each directed network: inbound degree centrality, representing the cumulative weight of all edges pointing toward a given vertex; outbound degree centrality, or the summed weights of all edges pointing away from a given vertex; and betweenness centrality, the proportion of all pairs of other vertices in the network for which a given vertex falls along the shortest path connecting the dyad. This yielded a total of 12 classification parameters used in the cluster analysis.

Since the variances of different parameters widely varied, each of these 12 variables was normalized with mean = 0 and standard deviation = 1 prior to any further analysis. Afterward, a dendrogram was constructed using agglomerative clustering with the complete linkage method. By default, agglomerative clustering algorithms continue until all users are contained in a single group. Thus, it is important to identify the point during the process at which each cluster is meaningfully different from each other cluster. For this purpose, a scree plot featuring the distance between the two groups merged at each iteration of the clustering algorithm was developed. The bend or ‘elbow’ in this plot represented the iteration at which the algorithm transitioned from forming groups of highly similar users to merging relatively dissimilar clusters. Thus, this was treated as the point in the clustering process at which each remaining cluster significantly differed from each other cluster, and the clusters that were present at this point in the analysis were considered to represent conceptually distinct groups of users in the network.

Results

RQ1: What types of brand engagement relationships are formed among different types of users?

The crosstabs for each network are presented in Tables 1 to 4. As is immediately evident from Table 1, neither the brand nor any affiliate had a single reply in the data set. The brand and affiliates were a bit more active in engaging with fellow users through retweets (Table 2) and mentions (Table 3), however. The brand tended to retweet affiliates quite often (147 retweets), and affiliates did the same for the brand (314 retweets), in both cases with much greater frequency than their retweets of fans (18 retweets and 29 retweets, respectively). Affiliates were also frequently mentioned by both the brand itself (310 mentions) and fellow affiliates (129 mentions). Interestingly, though, affiliates almost never mentioned the brand in their original tweets (2 tweets). This might suggest that there is little purpose in a WWE affiliate mentioning the @WWE Twitter handle, as anyone reading an individual affiliate’s tweets is likely already intimately familiar with the @WWE account as well.

Table 1.

Crosstabs for connections among users in the reply network.

			Receiver
		Brand	Affiliates	Fans
	Brand	0	0	0
Sender	Affiliates	0	0	0
	Fans	4771	994	3686

Table 2.

Crosstabs for connections among users in the retweet network.

			Receiver
		Brand	Affiliates	Fans
	Brand	0	147	18
Sender	Affiliates	314	6	29
	Fans	17336	7767	30558

Table 3.

Crosstabs for connections among users in the mentions network.

			Receiver
		Brand	Affiliates	Fans
	Brand	0	310	74
Sender	Affiliates	2	129	22
	Fans	6403	24,100	11,328

Table 4.

Crosstabs for connections among users in the omnibus network.

			Receiver
				Brand	Affiliates	Fans
	Brand	0	457	92
Sender	Affiliates	316	135	51
	Fans	28,510	32,861	45,572

In contrast, fans used all three forms of dyadic engagement – replies, retweets, and mentions – to connect with the brand, WWE affiliates, and fellow fans alike. They contributed nearly 10,000 replies during the data collection period (4771 replies to the brand, 994 replies to affiliates, and 3686 replies to fellow fans), along with over 50,000 retweets (17,336 retweets of the brand, 7767 retweets of affiliates, and 30,558 retweets of other fans) and 40,000 original mentions (6403 mentions of the brand, 24,100 mentions of affiliates, and 11,328 mentions of other fans). The sheer size of this disparity between fan engagement and that of the brand and affiliates is exhibited in Table 4, which combines the totals for Tables 1 to 3. This is likely due, in part, to the fact that there are more fans to collectively make replies, retweets, and mentions than the number of brand and affiliate accounts associated with WWE. Yet it bears repeating that WWE-affiliated entities tended to retweet and mention one another rather than engaging with fans, even though there were far more fan accounts that could have been targeted with retweets or mentions.

To put these findings in perspective, although the number of followers associated with each account changed across the data collection period (and even from one tweet to the next), the @WWE account had over 10,000,000 followers throughout this period. Individual affiliate accounts were also heavily followed, with figures such as @SashaBanksWWE (over 1,700,000 followers) and @RonKillings (over 1,500,000 followers) playing central roles.

RQ2: To what extent do highly prolific contributors engage with each other versus with comparatively casual contributors?

The assortativity coefficients for each network are given in Table 5. All four coefficients were negative (ranging from −0.140399 to −0.110915), indicating that all four networks featured a mild degree of disassortativity. However, this is likely due in large part to fans’ efforts to reply, retweet, and mention the brand, which single handedly made 364 relevant tweets during the data collection period. This was the eighth-highest tweet count among all 18,650 users in the data set. (For reference, the most prolific tweeter in the data set contributed 641 tweets but, like many other especially active users in the conversation, held no formal association with the brand. This demonstrates that the most active contributors were third parties pursuing their personal interests rather than the brand-related entities who, by definition, already represented the core of the discussion.)

Table 5.

Assortativity coefficients for each network based on the number of tweets made by each user.

Network	Assortativity
Reply	−0.140399
Retweet	−0.116194
Mentions	−0.110915
Omnibus	−0.112964

Consider, for instance, the reply network, which was the most disassortative of all four networks (assortativity = −0.140399). Neither the brand nor affiliates made a single reply to any tweet in the entire data set, so this result was entirely due to the fans engaging in disassortative mixing. Much the same was true of the other networks, as the brand and affiliates generally preferred to engage with one another, while fan replies, retweets, and mentions were disproportionately directed at the few WWE-affiliated accounts rather than the thousands of fellow fans in the conversation. In other words, the observed disassortative mixing pattern across all four networks was largely caused by casual contributors reaching out to the most prolific users, not by those highly prolific contributors choosing to engage with their less-invested peers.

RQ3: How does the network self-organize to differentiate types of users?

Figure 1 illustrates the dendrogram for the cluster analysis, which provides a tree-like visual representation of the groups of users in this community. As is evident from this figure, many merge operations formed groups whose users had highly similar characteristics. Several of the last merge operations, however, united largely dissimilar users, as the cluster analysis progressed until all users were contained in a single group.

Figure 1.

Dendrogram of the cluster analysis for users in the WWE Twitter network.

Figure 2 provides a scree plot showing the distance between each pair of clusters during the last 30 merge operations. There is a relatively clear ‘bend’ in the graph that begins before the final seven merge operations, suggesting substantially larger differences between the clusters that were merged during these last iterations of the cluster analysis. Eight clusters remained prior to the last seven merge operations, so these were considered to represent conceptually distinct groups of users. In other words, there were, at most, eight distinct clusters of users in the community.

Figure 2.

Scree plot of the distance between clusters merged during the last 30 steps of the cluster analysis. The graph has a visual bend prior to the last seven plotted points, suggesting that the eight clusters that were retained prior to the last seven merge operations were conceptually distinct from one another.

The list of users in each cluster is presented in Table 6, and each cluster’s means and standard deviations for all metrics used in the cluster analysis are given in Table 7. Additionally, a plot of the network that shows membership in different clusters is provided in Figure 3. Note that user handles for accounts not associated with the brand have been de-identified for the sake of preserving privacy where possible, while those for brand-related accounts (listed in Appendix 1) are provided in order to provide context for the community conversation, following the principles set forth by boyd and Crawford (2012).

Table 6.

Clusters of users identified via cluster analysis.

Cluster	Users
1	@WWE
2	@WWEUniverse
3	[1 account unaffiliated with the brand]
4	[3 accounts unaffiliated with the brand]
5	[4 accounts unaffiliated with the brand]
6	@RonKillings[1 account unaffiliated with the brand]
7	@NaomiWWE@SashaBanksWWE[13 accounts unaffiliated with the brand]
8	[All other users]

Table 7.

Means and standard deviations of network metrics for clusters identified via cluster analysis.

Cluster	Reply inbound degree centrality	Reply outbound degree centrality	Reply between centrality	Retweet inbound degree centrality	Retweet outbound degree centrality	Retweet between centrality
1	1870.0 (N/A)	0.0 (N/A)	0.0 (N/A)	2925.0 (N/A)	23.0 (N/A)	2,591,254.3 (N/A)
2	268.0 (N/A)	0.0 (N/A)	0.0 (N/A)	1107.0 (N/A)	6.0 (N/A)	3,374,668.1 (N/A)
3	5.0 (N/A)	0.0 (N/A)	0.0 (N/A)	62.0 (N/A)	3.0 (N/A)	2,054,490.0 (N/A)
4	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	8.0 (2.6)	72.3 (117.5)	3,365,682.2 (67,523.3)
5	0.3 (0.5)	1.0 (0.8)	9.0 (18.0)	2.0 (1.4)	81.3 (96.2)	2177.4 (2552.4)
6	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	3.5 (4.9)	11.0 (1.4)	24,556.2 (34,727.8)
7	9.9 (26.5)	3.9 (9.4)	237.9 (662.7)	116.6 (269.2)	24.9 (33.9)	172,636.4 (223,920.0)
8	0.2 (1.7)	0.3 (1.0)	1.9 (47.0)	1.2 (12.2)	1.5 (4.6)	444.9 (7519.3)
Cluster	Mentions inbound degree centrality	Mentions outbound degree centrality	Mentions between centrality	Omnibus inbound degree centrality	Omnibus outbound degree centrality	Omnibus between centrality
1	3357.0 (N/A)	112.0 (N/A)	429,061.8 (N/A)	6873 (N/A)	121 (N/A)	17,119,727.9 (N/A)
2	237.0 (N/A)	65.0 (N/A)	17,387.9 (N/A)	1502 (N/A)	71 (N/A)	19,963,007.5 (N/A)
3	25.0 (N/A)	2.0 (N/A)	62.0 (N/A)	91.0 (N/A)	5.0 (N/A)	14,236,743.8 (N/A)
4	2.0 (3.5)	4.7 (5.7)	87.0 (150.7)	10.0 (5.0)	76.0 (114.4)	15,462,652.4 (248,648.9)
5	3.5 (3.1)	15.0 (14.1)	50.6 (59.7)	5.5 (3.1)	96.3 (89.7)	3,453,270 (497,485.4)
6	54.0 (73.5)	11.5 (16.3)	137.6 (194.6)	57.5 (68.6)	21.5 (16.3)	5,745,756.3 (345,451.2)
7	42.2 (92.7)	21.4 (32.1)	418.0 (1033.4)	162.0 (308.6)	46.7 (50.6)	1,130,011.1 (296,324.8)
8	1.3 (18.9)	1.5 (4.22)	9.0 (380.4)	2.6 (24.5)	3.1 (6.7)	2838.1 (27,250.7)

Figure 3.

Plot of the Twitter network around WWE, accounting for all forms of mutual user interaction (replies, retweets, and mentions) with different user clusters represented by different vertex colors (cluster 1 = orange, cluster 2 = light blue, cluster 3 = green, cluster 4 = yellow, cluster 5 = dark blue, cluster 6 = red, cluster 7 = magenta, cluster 8 = gray). WWE: World Wrestling Entertainment.

First, to provide a point of comparison, cluster 8 consisted of all 18,623 users who did not fall into one of the other seven clusters. This cluster was exemplified by low all-around metrics, on average. In the omnibus network, for instance, users in cluster 8 had a mean inbound degree centrality score of only 2.6; outbound degree centrality was little better, with a mean of 3.1. Similarly, this cluster’s betweenness centrality in the omnibus network averaged 2838.1, far below the 1,130,011.1 of the next-smallest cluster.

As one might expect, @WWE ended up constituting its own unique cluster (cluster 1). After all, Tables 1 to 4 demonstrate how high the inbound degree centrality was for the @WWE brand account, and it also had relatively high betweenness centrality across the four networks, but particularly in the mentions network, where @WWE’s betweenness centrality of 429,061.8 was more than 20 times as high as that of the next-most central cluster (cluster 2; betweenness centrality = 17,387.9).

Speaking of cluster 2, this cluster was also entirely comprised by one user: @WWEUniverse, one of several alternative accounts run by the WWE; @WWEUniverse is devoted to WWE’s worldwide fan base. This account featured high values in many of the same metrics as @WWE, such as inbound degree centrality in the retweet network, where the 1107 value for cluster 2 was second only to the 2925 figure for cluster 1, representing @WWE. These both stood far beyond the 116.6 value for the next-closest cluster. In fact, clusters 1 and 2 held the top two mean scores on many metrics, including inbound degree centrality in all four networks, outbound degree centrality in the mentions network, and betweenness centrality in the mentions and omnibus networks.

@WWE and @WWEUniverse had especially high inbound degree centrality scores in the retweet network, which suggests that they acted as hubs driving conversation in the network. After all, those accounts with high inbound degree centrality in the retweet network achieved that status by being retweeted often, which means that others were spreading their messages such that they generated broad reach.

In contrast with these two official Twitter accounts, cluster 3 represented a smaller scale account devoted solely to the cruiserweight division in WWE and which operated on a small enough scale that it was not deemed to be directly affiliated with the WWE brand as a whole. Unlike @WWE and @WWEUniverse, this account earned only moderate inbound degree centrality across all networks (such as, for instance, inbound degree centrality = 91 in the omnibus network), and it had middling scores on many other measures, as well. The one exception was its betweenness centrality in the retweet network (2,054,490.0), which was nearly as high as that for @WWE (2,591,254.3) and @WWEUniverse (3,374,668.1).

While this smaller scale brand account did not have many connections to others, its few connections were important, as they allowed this user handle to serve as a bridge that connected otherwise disparate parts of the network. Much the same was true of the three user accounts in cluster 4, whose mean betweenness centrality in the retweet network (3,365,682.2) was second only to cluster 2 (3,374,668.1). The main distinction between clusters 3 and 4 was that, while the sole account in cluster 3 was highly retweeted (retweet network inbound degree centrality = 62) but did not highly retweet others (outbound degree centrality = 3), cluster 4’s mean inbound degree centrality score (8.0) was much lower than its mean outbound degree centrality (72.3). In short, while the users in cluster 4 also acted as bridges in the community, they achieved that stature by being especially active in retweeting others rather than by being highly retweeted themselves.

The four users in cluster 5 similarly devoted most of their energy to retweeting others, as they held a mean outbound degree centrality of 81.3, which was far beyond any of their other degree centrality scores. Yet unlike cluster 4, cluster 5’s efforts did not result in particularly high betweenness centrality in the retweet network (2177.4). Thus, while these users were highly engaged with others, they did not achieve especially central positions in the network as a result.

Next, the two members of cluster 6 were much more active in terms of mentions than replies or retweets. Yet they achieved their network positions in different ways. One user, a WWE fan whose name has been redacted for the sake of anonymity, had relatively high outbound degree centrality (23) in the mentions network but comparatively low inbound degree centrality (2). In contrast, @RonKillings, the account for wrestler Ron Killings, had exceedingly high inbound degree centrality (106) despite not making a single tweet mentioning another user in the network. Thus, both of them gained a degree of prominence in the mentions network, and they consequently achieved moderately high levels of betweenness centrality in the omnibus network (mean betweenness centrality = 5,745,756.3), but through different means. Both the aforementioned fan and @RonKillings received several retweets using the data collection period (12 and 10, respectively), further illustrating their role as highly engaging users in the community.

Finally, cluster 7 consisted of 15 different users who achieved relatively high betweenness centrality in the omnibus network (mean betweenness centrality = 1,130,011.1) through a variety of means. @NaomiWWE and @SashaBanksWWE, the user accounts for two WWE wrestlers, received large numbers of mentions (263 and 266, respectively). The same was true for the 83 mentions attributed to a third account, a wrestler associated with the ‘NXT’ faction rather than ‘Raw’ or ‘Smackdown’, the two programs analyzed in this study, who also received a large number of replies (105). (The account handle for this user has been redacted since this was not an account explicitly associated with the brand analyzed in this study.) Two fan accounts both received zero mentions but were highly active in retweeting others (with outbound degree centrality of 99 and 100, respectively, in the retweet network), while a third fan had all-around high outbound degree centrality spanning the reply network (37), the retweet network (50), and the mentions network (97). These varied paths all led to respectable betweenness centrality levels in the omnibus network ranging from 764,410.5 to 1,692,240.9. Broadly speaking, one might say that these individuals represented a second tier in the WWE Twitter community, maintaining an intermediate role between top-tier users like @WWE and @RonKillings, and the comparatively casual contributors who comprised cluster 8.

It is worth recognizing that the fan account that was previously recognized as the most prolific individual with respect to the sheer number of tweets made during the study period fell into cluster 8 as a casual contributor. This empirically demonstrates that simply tweeting in high volume is not sufficient for users to meaningfully distinguish themselves from their peers.

In short, the WWE community self-organized into several distinct user roles: sought-after hubs who held uniquely great inbound degree centrality across all network types; bridges who connected other groups of users through replies, retweets, and/or mentions; highly engaging users who were especially prominent in the mentions network; moderately prominent users who fell into an intermediate tier between top influential users and the masses; especially active users who nonetheless failed to gain substantial traction in the community; and the thousands of casual contributors who lightly participated and received minimal attention in return.

Discussion

Over the course of the 2 weeks of data collection – weeks ending December 16 and 23, 2018 – the number of social media interactions across both ‘WWE Raw’ and ‘WWE Smackdown’ were 5.034 million. Of those, 1.035 million were through tweets.

As the results for RQ1 demonstrated, the WWE and its affiliates hardly engaged with fans at all through general tweets, mentions, comments, replies, or retweets, instead preferring to retweet and mention one another. Yet the WWE fanbase is nonetheless highly engaged, to such an extent that this community serves as an exemplar for live tweeting during televised events. It is possible that since the WWE and its affiliates created a conversation among one another, fans who replied, retweeted, and mentioned affiliates and the brand felt as though they were part of that conversation, even if the interpersonal connections were never (or rarely) actually reciprocated by the brand and affiliates. In other words, these findings suggest that brand-related parties don’t have to actually engage with lay users to foster a dedicated fanbase. Rather, they can simply engage in a public dialogue with the overall community, or among one another, and allow lay users to participate in the same interactive space, thus giving the illusion of collective engagement uniting all parties.

What is particularly interesting when considering the amount (or lack thereof) of direct engagement from the brand to its fanbase is the level of disassortivity within the network. This further supports the idea that the brand itself does not need to engage directly with its fans to maintain its incredibly high degree of brand-specific Twitter activity during these telecasts. This could be, at least in part, a consequence of the sheer amount of tweets throughout the telecasts, and the implausibility of the WWE being able to sort through and keep up with this activity and respond to a substantial portion of it in a timely fashion, considering this incredibly high number of interactions takes place over an abbreviated period of time (telecasts run between 2½ h and 4 h). The fact that the activity here is nearly exclusively one way (fans to the brand, without any reciprocation) further drives the point that actual engagement from the brand is not paramount to the fan.

This same lack of reciprocation is also evident when looking at how the entirety of the Twitter activity organically creates the individual clusters of user types. As might be expected, there exists a concrete disassociation between WWE brand and affiliate accounts and fans. The cluster analysis clearly shows accounts such as @WWE, @WWEUniverse, and the sole account attributed to cluster 3 – all Twitter accounts connected in varying forms to the WWE organization – representing their own respective user types. Beyond that, the remaining clusters are composed mainly of fans who, while they contribute a large amount of content through tweets, are still overall not acknowledged by the WWE organization.

In the case of cluster 3 in particular, the account that comprised it achieved exceptionally high betweenness centrality in the retweet network despite making only three retweets and being retweeted just 62 times in the entire data set. This suggests that the account filled structural holes in the network, connecting users who otherwise engaged in wildly disparate parts of the Twitter conversation. As suggested by Burt (1992), and reinforced by Nee and Barker (2019), such individuals are more likely to be engaged in novel communication (whether they are privy to important, nonredundant insights or are disseminating unique knowledge themselves), and they accrue more social capital as a result of their position in the conversation. Importantly, typical social media metrics would treat this account as a relatively middling entity compared with @WWE and even @WWEUniverse, as it exhibited minimal activity and featured fewer than 80,000 followers at the time of data collection (vs. over 10,000,000 and over 5,000,000 followers for @WWE and @WWEUniverse, respectively). Yet because of its ability to engage across the entire conversation, the account attributed to cluster 3 held an important unifying role and significant power within the discursive space as a whole.

To be sure, the WWE fanbase is a loyal, passionate one, particularly when it comes to online social media engagement, as is indicated by the sheer constant volume of social media activity on a weekly basis as defined by Nielsen social TV metrics. Thus, the ability for the WWE to let the fans carry the weight of the social TV activity while the brand itself (and its affiliates) merely stokes the fire to propagate such activity speaks to the tremendous power that the organization holds over its fans. This phenomenon is akin to the participation dilemma, where viewers gain influence over a production – in this case, the social TV arm of the program that further emotionally ties themselves to the show itself (van Es, 2016).

Further supporting the power of the WWE brand as a social TV juggernaut, and televised professional wrestling overall, is the growing number of weekly telecasts that also attract immense social TV activity. Besides the legacy shows ‘Raw’ and ‘Smackdown’, the WWE has since introduced ‘WWE NXT’, also a weekly show on the USA Network, while the upstart ‘All Elite Wrestling’ (not affiliated with the WWE) has since been added to the list of weekly televised wrestling shows (TNT Network).

Limitations of the study and future research

To be sure, this study, in its exploratory nature, is a prime jumping off point for further evaluating several aspects of this fan-to-brand online relationship. The nature of the data analysis leaves further opportunities to go beyond the ‘how’ of these multidirectional connections between all active parties and investigate the ‘why’. For example, the valence of the tweets – positive, negative, or neutral – could add a motivational component to the results, introducing variables that might further branch fans into subgroups based on sentiment. Anecdotally, an inspection of a subset of the data suggested that the majority of tweets expressed positive sentiments toward the brand, but a content analysis of the tweets could prove useful to distinguish between users expressing favorable attitudes and those who were disgruntled due to the lack of reciprocal engagement or other factors.

While this data set is robust in number, a more linear approach to data collection (beyond two weeks and four telecasts) would offer a more complete data set, one that would allow the researchers to follow more closely the most active fan base – those who tweet week in and week out. And, while ‘WWE Raw’ and ‘WWE Smackdown’ are two consistent and very high-profile WWE telecasts, there are others, including pay-per-view events and other telecasts offered only on the WWE’s own premium TV network. And because of the paywall behind this network, it would allow for a more in-depth look at the WWE’s most ardent fans – those willing to pay the extra amount to have unlimited access to the WWE 24/7. It should be noted, too, that while this study captured the entirety of primary Twitter engagement by the WWE, its affiliates, and fans, it does not include measurements of ‘likes’ due to the volatility inherent in those data, which could otherwise add another dimension to the research addressing a comparatively low level of interactive engagement. Finally, while it is acknowledged that Twitter is a driving force in social TV activity, the two other large players – Facebook and Instagram – are forces in their own right, and much could be gleaned from including both of these platforms in future studies.

Footnotes

Author biographies

Stephen McCreery (PhD, University of Georgia) is an associate professor in the Department of Communication, Electronic Media/Broadcasting at Appalachian State University. His research focuses on mass media consumption, particularly within the intersection of legacy and new media content.

Brian C. Britt (PhD, Purdue University) is an assistant professor in the Department of Advertising and Public Relations and the Associate Director of Data Analytics in the Public Opinion Lab at the University of Alabama. His research focuses on the intersection between organizational communication and new media, with a particular emphasis on the study of large-scale online organizations and the development of computational social scientific research methods to assess them.

Jameson Hayes (PhD, University of Georgia) is an associate professor in the Department of Advertising and Public Relations and director of the Public Opinion Lab at the University of Alabama. His research specialization is emerging media marketing communication, and social media analytics, specifically examining the intersection of brand and interpersonal communication within emerging media and the resulting implications for advertising/marketing communication theory and practice.

Appendix 1

References

Anderson

Jiang

(2018) Teens, social media & technology. Pew Research Center. Available at: http://www.pewinternet.org/2018/05/31/teens-social-media-technology-2018 (accessed 8 February 2021).

Araujo

Copulsky

Hayes

, et al. (2020) From purchasing exposure to fostering engagement: Brand–consumer experiences in the emerging computational advertising landscape. Journal of Advertising 49(4): 428–445.

Bagozzi

Utpal

(2002) Intentional social action in virtual communities. Journal of Interactive Marketing 16(2): 2–21.

Barger

Peltier

Schultz

(2016) Social media and consumer engagement: A review and research agenda. Journal of Research in Interactive Marketing 10(4): 268–287.

Baym

(2000) Tune In, Log On: Soaps, Fandom, and Online Community. Thousand Oaks, CA: Sage Publications.

Booth

(2008) Rereading fandom: Myspace character personas and narrative identification. Critical Studies in Media Communication 25(5): 514–536.

boyd

Crawford

(2012) Critical questions for big data. Information, Communication & Society 15(5): 662–679.

Brady

Voorhees

Cronin

Jr , et al. (2006) The good guys don’t always win: The effect of valence on service perceptions and consequences. Journal of Services Marketing 20(2): 83–91.

Britt

Hayes

Britt

, et al. (2020) Too big to sell? A computational analysis of network and content characteristics among mega and micro beauty and fashion social media influencers. Journal of Interactive Advertising 20(2): 111–118.

10.

Brodie

Ilic

Hollebeek

(2013) Consumer engagement in a virtual brand community: An exploratory analysis. Journal of Business Research 66(1): 105–114.

11.

Burt

(1992) Structural Holes: The Social Structure of Competition. Boston, MA: Harvard Press.

12.

Bury

(2016) Technology, fandom, and community in the second media age. Convergence: The International Journal of Research into New Media Technologies 23(6): 627–642.

13.

Cho

Keum

Shah

(2015) News consumers, opinion leaders, and citizen consumers: Moderators of the consumption–participation link. Journalism & Mass Communication Quarterly 92(1): 161–178.

14.

Cinque

(2015) Digital shimmer: Popular music and the intimate nexus between fan and star. In: Marshall

Redmond

(eds) A Companion to Celebrity. Australia/Oceania: Wiley Blackwell, pp. 440–455.

15.

Cohen

Lancaster

(2014) Individual differences in in-person and social media television coviewing: The role of emotional contagion, need to belong, and coviewing orientation. Cyberpsychology, Behavior, and Social Networking 17(8): 512–518.

16.

Cote

Darling

(2018) Scientists on Twitter: Preaching to the choir or singing from the rooftops. Facets 3: 682–694.

17.

de Valck

van Bruggen

Wierenga

(2009) Virtual communities: A marketing perspective. Decision Support Systems 47(3): 185–203.

18.

Gantz

Lewis

(2014) Sports on traditional and newer digital media: Is there really a fight for fans? Television & New Media 15(8): 760–768.

19.

Giglietto

Selva

(2014) Second screen and participation: A content analysis on a full season dataset of tweets. Journal of Communication 64(2): 260–277.

20.

Grewal

Stephen

(2019) In mobile we trust: The effects of mobile versus nonmobile reviews on consumer purchase intentions. Journal of Marketing Research 56(5): 791–808.

21.

Grewal

Hulland

Kopalle

, et al. (2019) The future of technology and marketing: A multidisciplinary perspective. Journal of the Academy of Marketing Sciences 48(4): 1–8.

22.

Gross

Fetter

Paul-Stueve

(2008) Toward advanced social TV in a cooperative media space. International Journal of Human-Computer Interaction 24(2): 155–173.

23.

Gummerus

Liljander

Weman

, et al. (2012) Customer engagement in a Facebook brand community. Management Research Review 35(9): 857–872.

24.

Han

Lee

(2014) Motivations for the complementary use of text-based media during linear TV viewing: An exploratory study. Computers in Human Behavior 32: 235–243.

25.

Harrington

Highfield

Bruns

(2013) More than a backchannel: Twitter and television. Journal of Audience & Reception Studies 10(1): 405–409.

26.

Hayes

King

(2014) The social exchange of viral ads: Referral and coreferral of ads among ads among college students. Journal of Interactive Advertising 14(2): 98–109.

27.

Hills

(2013) Fiske’s ‘textual productivity’ and digital fandom: Web 2.0 democratization versus fan distinction? Participations: Journal of Audience and Reception Studies 10(1): 130–158.

28.

Himelboim

McCreery

Smith

(2013) Birds of a feather tweet together: Integrating network and content analyses to examine cross-ideology exposure on Twitter. Journal of Computer-Mediated Communication 18: 154–174.

29.

Hwang

Lim

(2015) The impact of engagement motives for social TV on social presence and sports channel commitment. Telematics and Informatics 32: 755–765.

30.

Ito

, et al. (2010) Hanging Out, Messing Around, and Geeking Out: Kids Living and Learning with New Media. Cambridge, MA: The MIT Press (The John D. And Catherine T. MacArthur Foundation Series on Digital Media and Learning).

31.

Jolley

(2016) ccjolley/cRimson. Available at: https://github.com/ccjolley/cRimson (accessed 8 September 2019).

32.

Kennedy

(2016) Post, Mine, Repeat: Social Media Data Mining Becomes Ordinary. London: Palgrave Macmillan.

33.

Krämer

Winter

Benninghoff

, et al. (2015) How ‘social’ is social TV? The influence of social motives and expected outcomes on the usage of social TV applications. Computers in Human Behavior 51(A): 255–262.

34.

Lacalle

Simelio

(2017) Television fiction and online communities: An analysis of comments on social networks and forums made by female viewers. Critical Studies in Media Communication 34(5): 449–463.

35.

Libai

(2011) Comment: The perils of focusing on highly engaged customers. Journal of Service Research 14(3): 275–276.

36.

Lin

Sun

Chen

(2016) Social television: Examining the antecedents and consequences of connected TV viewing. Computers in Human Behavior 58: 171–178.

37.

Maslowska

Malthouse

Collinger

(2016) The customer engagement ecosystem. Journal of Marketing Management 32(5-6): 469–501.

38.

Merry

(2016) Making friends and enemies on social media: The case of gun policy organizations. Online Information Review 40(5): 624–642.

39.

Muniz

O’Guinn

(2001) Brand community. Journal of Consumer Research 27: 412–432.

40.

Nagy

Midha

(2014) The value of earned audiences: How social interactions amplify TV impact. Journal of Advertising Research 54(4): 448–453.

41.

Nambisan

Baron

(2007) Interactions in virtual customer environments: Implications for product support and customer relationship management. Journal of Interactive Marketing 21(2): 42–62.

42.

Nee

Barker

(2019) Co-viewing virtually: Social outcomes of second screening with televised and streamed content. Television & New Media 21(7): 712–729.

43.

Newman

MEJ

(2002) Assortative mixing in networks. Physical Review Letters 89: article 208701.

44.

Nielsen Social Content Ratings (2018) Weeks Ending December 16 and December 23. Nielsen Weekly Top 10 Social TV Ratings. Available at: https://www.nielsensocial.com/socialcontentratings/weekly/ (accessed 8 February 2021).

45.

Nielsen (2018) Juggling act: Audiences have more media at their disposal and are using them simultaneously. Available at: https://www.nielsen.com/us/en/insights/article/2018/juggling-act-audiences-have-more-media-at-their-disposal-and-are-using-them-simultaneously (accessed 8 February 2021).

46.

Nielsen (2019) Tops of 2019: Social TV. Available at: https://www.nielsen.com/us/en/insights/article/2019/tops-of-2019-social-tv/ (accessed 8 February 2021).

47.

Nielsen Social (n.d.). Available at: https://www.nielsensocial.com/ (accessed 8 February 2021).

48.

Pynta

Seixas

SAS

Nield

, et al. (2014) The power of social television: Can social media build viewer engagement? Journal of Advertising Research 54(1): 71–80.

49.

Quintas-Froufe

González-Neira

(2014) Active audiences: Social audience participation in television. Comunicar 22(43): 83–90.

50.

Rathnayake

Suthers

(2019) Enclaves of exposure: A conceptual viewpoint to explore cross-ideology exposure on social network sites. The Social Science Journal 56(2): 145–155.

51.

Redmond

Cinque

(2017) Talking Miley: The value of celebrity gossip. In: Harrington

(ed) Entertainment Values. London: Palgrave Macmillan, pp. 71–90.

52.

Rheingold

(1993) The Virtual Community: Homesteading on the Electronic Frontier. Massachusetts, MA: Addison-Wesley.

53.

Salomaa

Lehtinen

(2018) Congratulations, you’re on TV!: Middle-space performances of live tweeters during the FIFA World Cup. Discourse, Context & Media 25: 131–142.

54.

Tien

Rivas

AAA

Liao

(2018) Examining the influence of customer-to-customer electronic word-of-mouth on purchase intention in social networking sites. Asia Pacific Management Review 23(4): 1–12.

55.

Van Doorn

Lemon

Mittal

, et al. (2010) Customer engagement behavior: Theoretical foundations and research directions. Journal of Service Research 13(3): 253–266.

56.

van Es

(2016) Social TV and the participation dilemma in NBC’s The Voice. Television & New Media 17 (2): 108–123.

57.

Weimann-Saks

Ariel

Elishar-Malka

(2020) Social second screen: WhatsApp and watching the World Cup. Communication & Sport 8(1): 123–141.

58.

Wohn

(2011) Tweeting about TV: Sharing television viewing experiences via social media message streams. First Monday 16(3). Available at: https://firstmonday.org/ojs/index.php/fm/article/view/3368 (accessed 8 February 2021).