How and Why Some IssuesSpread Fast in Social Media

This paper brings together current insights from various disciplines into the spreading of social media posts.By astructured literature review of peer-reviewed literature 39 recent articles on diffusion in social media were located. The search spanned 10 years, although all the papers found were published after 2009, indicating that the examination and observation of spread patterns in social media is still in an early stage. The analysisfocused on spread patterns and factors that may explain how rapidly issues spread in the online environment. Based on the findings, from an organizational perspective a model is constructed and directions for future research are suggested.The model focuses on characteristics of the issue, of the social media involved, actor resources and general factors. A better understanding of how social media posts spread helps organizations to be prepared for upcoming issues and crises, such as launching early rumour detection to prevent losses in organizational and brand image.


Introduction
The aim of this paper is to clarify the diffusion patterns and factors that explain the spread of issues in social media. This can help organizations to interpret the results of social media monitoring and decide on the necessity for social media interventions. To this end, the paper adopts a broader perspective, collecting insights from various disciplines on diffusion in social media.
Social media allow people to build social networks using internet applications that provide users with a variety of exchange platforms (Nadeem, 2012;Wang, 2012). This has led to an explosive growth of social media posts. For organizations, the rapid development of social media brings not only insights into customer opinions and new ways to spread their own viewpoint (Kumar & Mirchandani, 2012), but also rapid diffusion of possibly unexpected topics, such as negative electronic word-of-mouth messages (Zhang, Jansen, & Chowdhury, 2011). Nowadays, the universal use of social media has become apriority in order to improve organizational performance and enhance communications with users (Fan, Geddes, & Flory, 2013). Social media may strengthen organizations' ability to reach a large audience. However, alongside opportunities,social media interaction also brings challenges, e.g. "the advent of consumer-generated content and its rapid diffusion takes much of the control over messages initiate the spread of issues themselves. In the following sections, we summarize how issues spread in social media, as described in the literature.

Cascading
In the literature, spread patterns are seen as directed by the paths along which messages travel in social media. Various authors (e.g. González-Bailón, Borge-Holthoefer, & Moreno, 2013) have conceptualized this as cascading, a process by which a particular message is passed to a first group of receivers who then pass it on to the next, and so on, until an extensive network is built up. Generally, by standing out in social media, cascading allows users to contribute to a virtually unlimited process of diffusion. In micro blogs such as Twitter, people have followers, and therefore any message emitted from a node will immediately be available to anyone following the tweet sender (Borge-Holthoefer, Rivero, & Moreno, 2012). By a simple retweet, messages spread, embarking on numerous different paths (Bosley et al., 2013;Stefanidis, Crooks,&Radzikowski, 2013;Stieglitz,& Dang-Xuan, 2013;Zhang et al., 2011).
The concept of cascading suggests that posts that are passed on may multiply. However, not all posts will be shared by cascading. Empirical research on Facebook did not show evidence of cascading as such, but rather the colliding of shorter chains while a threshold amount of start-nodes were needed in order to spread apost wider (Rogers et al., 2012). time' of individuals (Doerr, Blenn & Mieghem, 2013). For example, Coombs (2002) uses the concept of 'issue contagion' to address the spread of issues. In epidemiology, the spreading of viruses is, for example, related to contact probability and frequency, based on a model developed by Reed and Frost in the 1920s. In recent years, this model has inspired the development of mathematical models for the spread of messages in social media (see a later section).
Like the spread of viruses, the diffusion of social media messages may start in a particular location and then spread to others. To analyse the diffusion of messages, starting for example in a local event, geo-location can be used, as was done in the case of political communication on Twitter (Stieglitz,& Dang-Xuan, 2013). However, currently only a small proportion of social media messages is geo-located.

Comparing Message Reach to Adoption of Innovations
In the literature, the number of people reached by a social media message is often explained by reference to the model of diffusion of innovations as developed by Rogers (1995), who defined diffusion as a process by which an innovation is communicated through certain channels over time amongst the members of a given social system. His model shows a normal (bell) curve with successive groups of people adopting the innovation. The model was also applied by communication scholars, for example, to investigate the diffusion of news among the public (Valente,& Rogers, 1995). To apply this model to hypes in social media, the consultancy firm Gartner extended the curve after the peak of inflated expectations to show a steady plateau of productivity (Fernando, 2010).
The literature shows that spread patterns do not always follow the normal curve, as the structure of networks and their paths for diffusion differ, as also do the positions of those who trigger the diffusion and help disseminate the message (González-Bailón et al., 2013). Time intervals also need to be considered as, particularly in the initial stage, there may be a time lag in the passing of a message (Fan, Geddes, & Flory, 2013). Similarly, long power outages Online Journal of Communication and Media Technologies Volume: 5 -Issue: 1 -January-2015 may also result in unexpected time lags. More importantly, a study on the reputation of the brand Toyota showed that the content of a social media postchanges as it is passed on, and may become more positive, neutral or negative (Fan et al., 2013). This adaptation in the process of passing on social media posts was also noted in the activism engendered during the so called Arab spring, when resistance leaders reconstructed messages to suit their needs, after which the local message was recreated for a global audience (Newsom & Lengel, 2012).
This shows that the spread of social media posts, rather than passing on a package, can be seen as an interaction between various actors.

Network Patterns and Roles
The way issues spread depends on the roles of the actors in the network. According to Castells (2008, p. 152), a network can basically be seen as "a set of interconnected nodes".Some individuals are more connected than others, so connectivity is not equally distributed across the network. In the network of micro blog followers, only a few highly connected nodes act as hubs (Borge-Holthoefer et al., 2012). A hub has a dominant position in a network, as it functions as a gatekeeper (Gruzd, Wellman, & Takhteyev, 2011), deciding whether to pass on or not pass on social media posts to other users.
It has been suggested that, in particular, weak links are important for diffusion in the online environment (Granovetter, 1973). However, in social media the role of weak links needs to be specified. Empirical research in the social media environment showed that weak links do not speed up diffusion, but "act as bridges to connect isolated communities" (Zhao, Wu,& Xu, 2010, p. 2). Diffusion may stagger to a halt if connections are bounded; in such instances, wider links with other networks are important for the growth of an issue.

Mathematical Models
To predict the spread of social media posts, various authors (e.g. Laskela, 2010) have developed mathematical models to describe the relationships between the variables that influence diffusion. Such quantitative models often focus on the speed and number of nodes  Zhao et al. (2010), such models, based on relational data, provide an estimation of random diffusion, although it has proved difficult to include all the complexities of real world exchange on the Internet.
According to Kumar and Mirchandani (2012) only few attempts have been made to define message spread, influence and impact in relation to marketing or communication management. They tried to predict the ability of influencers to generate viral spread, based on e.g. the number of times a message was forwarded and the number of comments or replies received. Along similar lines, Li and Shiu (2012) designed a diffusion support mechanism for selecting endorsers in social media, and tested its performance in measuring user preference through click-through rate, network influence by re-post rate, and propagation strength by exposure rate. The above sections show the variety of approaches that exist for determining how issues spread in social media.

Factors Influencing Dissemination in Social Media
To better understand what influences the spread of social media posts on the web, we further analysed the selected literature. This yielded various factors that may enable or hinder dissemination in social media. We then organized these factors, following a research model of communication in issue arenas (Vos, Schoemaker, &Luoma-aho, 2014) according to whether they concerned characteristics of issues,media or actors.

Characteristics of the Issues Involved
In a network or micro blog, all users have their own friends or followers. What posts are passed on also depends on what people like to share with each other. According to Wang (2012, p. 309), "If the message itself is valid and possesses high social value, it is likely that the message will be shared by many different users on multiple occasions, thereby increasing the instances of exposure". In the literature, various characteristics of an issue were expected to promote dissemination in social media. for example, with health-related messages. Also favoured is content that increases knowledge (Desai et al., 2012), as it may provide a solution to a particular problem or offer the receiver "true value and benefit" (Bates & Riedy, 2012).
-Expresses needs or emotions. If a post relates to needs or emotions it is more likely to be passed on. "Emotionally charged Twitter messages tend to be retweeted more often and more quickly" (Stieglitz & Dang-Xuan, 2013). This may also apply to emotional experiences related to a product or service, for example, expressed in blogs, micro-blogs such as Twitter, and on YouTube ornetworks such as Facebook.
-Has entertainment value or imparts a positive sentiment. Qualities like humour enable the sharing of posts, for example allowing "participants to move from initial nervousness into more relaxed and comfortable conversations" (Byron et al., 2013, p. 40). Meanwhile, positive messages are generated and spread easily. For example, Desai et al. (2012, p. 4) noted in a study on re-tweeting messages about a conference, that a positive tweet "leaves a good impression with the reader and increases the likelihood that future tweets will be amplified by that reader".
-Has news value. A message that has news value, for example includes eye-witness accounts during a crisis, is more likely to be passed on (Hiltz, Diaz, & Mark, 2011). News content of social media messages may be related to a well-known organization (Williams et al., 2012) and the content positive, negative or neutral (e.g. Fan et al., 2013). It may also concern a well-known person or celebrity, as Sanderson & Cheong (2010) showed in a study on how -People want to be identified with it. Consumers use social media to engage with brands, products and services they want to identify with (Williams et al., 2012). What users reinforce, for example by re-tweeting or 'likes', is often shown on their homepage. Since it adds to their identity, people may be opportunistic in what they wish to show and with whom they want to be seen to belong. How fast a message travels also depends on societal factors, which basically turns users into sensors (Stefanidis et al., 2013). Therefore, issues shared in social media are less likely to include topics related to taboos, as users do not wish to invite gossip or bullying, as "social media content is incorporated into broader practices of self-presentation and identity management" (Byron et al., 2013, p. 41).
The way issues take form in social media differs widely from that in news media, as it seems that people use social media especially from a personal perspective, to express their views, depict their experiences and share what they perceive around them, for example in eye-witness reports. Consequently, such motives then influence how an issue is communicated and takes form in the online environment.

Characteristics of the Media Involved
The particular features of the individual social media may facilitate diffusion of an issue on a smaller or broader scale. It should, for example, be noted that Facebook and Whatsapp are based on strong ties and emphasize the strengthening of friendships, while Twitter is primarily based on weak ties and is suitable for factual exchange (Zhao, Wu,& Xu, 2010).
Social media have various features that enable fast dissemination, including ease of searching, sharing, and connecting with other users. Therefore, it matters from which social media platform the issue discussion has originated, although transfer to other social media platforms is possible and is more easily initiated in some social media than others. Mainly, however, diffusion depends on ease of sharing, ease of finding what one is looking for, and ease of connecting in the social media used. on links to YouTube videos (Robichaud et al., 2012). Facebook in turn facilitates active involvement with friends, e.g. through 'likes' that strengthenrelationships (Rogers et al., 2012). The use of "likes" has also been used in campaigns to create weak ties.
Ease of finding what one is looking for, or posting matters that can easily be found by others, for example by using a hashtag on Twitter, is also related to ease of dissemination (Kiernan and Wigglesworth, 2011).
Ease of connecting may be higher in some social media services than in others (Bronstein, 2013). Users of Twitter are free to follow others, which also results in weak ties while, for example, WhatsApp is a more closed friendship environment. Consequently, Gruzd et al. (2011Gruzd et al. ( , p. 1294 note, that "connections on Twitter depend less on in-person contacts, as many users have more followers than they know". When ease of connecting is high, this may result in connections to an undeterminable degree and "constantly shifting clusters of conversations that have collapsed the traditional boundaries of space and time" (Farshid et al., 2011, p. 222).

Characteristics of Actor Resources forSocial Media
The actors involved in the issue-spreading process may be more or less connected, and more or less active in interaction on the web. Therefore, organizationsthat are successful in their use of social media will devote considerable resources to laying the foundation for their social media activities and involving other actors (Nah, & Saxton, 2013). They may do this by building platforms, content and followers, and developing ongoing monitoring and multi-channel approaches. In social media campaigns, organizations may want to spread matters widely, rapidly and/or to targeted groups by involving key-stakeholders (Suarez-Almazor, 2011), including not only policy makers and various public groups but also intermediaries through which relevant public groups may be reached. An organization can be supported in its online activities by cooperating with its (business) partners who may, for example, retweetimportant messages. How many are reached depends on the interconnectedness of the actors that provide the post or pass it on (Gruzd et al., 2011). Organizations can form links with partners to increase their interconnectedness; this may include other organizations or, for example, bloggers.
Actors who often pass on social media posts to others are known as influencers. Such actors have the knowledge and willingness to support dissemination, for example, through tweet amplification (Desai et al., 2012). Following the growing interest in social media, organizations have begun to attributea profound role to influencers. Influencers with an established network in social media are also called 'social endorsers' (Li,&Shiu, 2012).
Authority is attributed to those who are highly influential because they have many links with well-connected others; for example, if a blogger is highly influential "we would expect his ideas to propagate to other blogs" (Lawrence, Melville, Perlich, Sindhwani, Meliksetain, Hsueh,& Liu, 2010, p. 3). Since some bloggers have influence within a community, while others (also) have influence outside that community, measures are being developed to help organizations select the most suitableblogs for dissemination (Lawrence et al., 2010).
In social marketing practice, identifying influencers who are highly influential, also called influentials, is organization-and case-specific (Kumar,&Mirchandani, 2012), and thus the choice of influencers will often depend on the issue at stake.In purchasing decisions, customers maybe affected by user-generated content (Stieglitz,& Dang-Xuan, 2013), often referred to as 'consumer-generated media' or 'consumer-generated advertising' (Farshid et al., 2011). In that sense, consumers can be good influencers, as "a skilled consumer may offer a more compelling message that has more credibility than a company-generated message" (Williams et al., 2012, p. 129). However, Freberg (2012) found that user-generated sources are not always more effective, as their trustworthiness may be perceived differently according to users also depending on the topic. In any case, interconnectedness in the online environment is seen as a resource of actors.
Online Journal of Communication and Media Technologies Volume: 5 -Issue: 1 -January-2015

A Model ShowingFactors that Enable Online Issue Spread
Below, based on the findings reported in the previous section,we present a model of the keyfactors that influence the diffusion of issues in social media. First, the model shows the characteristics of an issue that influence its rapid dissemination on the web, as it is thesecharacteristics that make it more or less attractivefor users to pass the message on.
Second, the model shows the socialmedia characteristics that may also facilitate the rapid spreading of an issue. It matters from which social media platform the issue discussion originates, although there may also be transfer among different social media platforms.Third, the model shows the organizational resources for social media,influencing the preparedness of the organizationfor social media monitoringthat,depending on the issues management policy, may be geared towards a better understanding of stakeholder points of view or towards influencing the spreading of issues.
Next to these factors that relate specifically to the online environment, there are also factors of a more general nature related to societal developments and organizational reputation.For example, organizations should consider their vulnerability concerning issues and that issues related to them may travel more or less rapidly on the web. Such vulnerability could relate to societal factors or similar crises in the past, in the history of the organization or its (business) sector. Therefore, organizational reputation should be seen as an important general factor,just asdevelopments and power relations in the broader social environment that all may influence the interplay of actors in traditional as well as online issue arenas .
Together, these factors pull or push the discussion on the issue, explaining the speed at which the issue travels on the web, as shown in Figure 1. The centre of the model symbolizes the iterative process of reflection on the spread of social mediaposts to better understand the outcome of all these influences.Inspired by Rogers' (1995) model for the diffusion of innovations, we assume that in the various stages ofdissemination different actors may be active in the process, such as early adopters (or, for example, activists drawing attention to an issue), who may act and be perceived differently from the broader public, which may be case study published in conference proceedings, Yang and Counts (2010) found that tweets that came later during the observation period and those that included links often travelled further in the network. There might also be a threshold for diffusionin social media, resembling that inepidemiology, wherea minimum number of infections is required to increase the probability of a disease spreading to the whole network, oras in game theorywhereaninnovation needs to attract a minimum number of adoptees before its utility for other prospective usersis at a high enough level to induce them to adopt itas well (Song, 2013).

Directions for Future Research
Current mathematical models focus on the spread of individualmessages, for example, in random diffusion, whereas the various complexities related to the spread of organizational issues in social mediahave yet to be taken into account. Changes in messages as they are passed need to be further investigated, as some authors state that in this process the message content becomes adapted in a more negative or positive direction, or to suit a broader public (Fan et al., 2013;Newsom,&Lengel, 2012).
We also argued that transfer within different social media and with traditional news media needsto be taken into account.Moreover, interference between the traditional news media and social media is not reflected in the models. For organizations, it is relevant that issue transfer between social media and traditional news media exists (Meriläinen,&Vos, 2013), although it has been suggested that this needs to surmount a threshold in order to gain momentum, rather like the threshold described in the diffusion of posts in, for example, Facebook (Rogers et al., 2012).

Implications for Practice
When companies monitor social media, the results may reveal various issues related to organizational policies. However, monitoring in itself does not clarify what issues mostly need attention. This needs a better understanding of the factors that determinewhether an issue can be expected to develop rapidly.The model presented here brings together current insightson the diffusion of social media posts and provides input for decision-making on communication strategies and a more critical outsourcing of related monitoring services, by enhancing understandingof the principles of diffusion.In the social media environment, above all, it is interconnectedness that counts.