Risk, Crisis, and Social Media: A systematic review of seven years’ research

Social media has received considerable attention from researchers in risk and crisis communication. However, as the number of studies on social media, risk, and crisis communication has grown tremendously, it becomes all the more important to take stock before moving forward to avoid exhausting research questions and topics. While a few useful literature reviews have been published (e.g., Fraustino, Liu & Jin 2012; Ruggiero & Vos 2014; Veil, Buehner & Palenchar 2011), these must be updated and expanded, not least to include the recent experimental and comparative studies, and cover the field’s geographical focus. We thus review recent research and draw out some important conclusions. In particular, we argue that researchers and risk and crisis planners must take into account that not everyone uses social media. Particularly, citizens with low socio-economic status are those with the lowest internet availability (Zickuhr 2013), and they are also the most vulnerable in a disaster, which Hurricane Katrina tragically proved (Thiede & Brown 2013). Risk and crisis management via social media thus risks missing those who are most vulnerable and most in need of relief information.

The article is structured as follows: next, we define social media and its integration with risk and crisis communication. This section is followed by a discussion of previous literature reviews, and leads up to the formulation of our research questions. The following section on methodology describes how we conducted a database search and coded the resulting sample. Our findings are then presented in a section structured after the main variables. The findings and their implications are then discussed in the conclusion.

Researchers in the fields of risk and crisis management have increasingly come to conceptualise risk and crisis as transboundary (e.g., Boin, Rhinard & Ekengren 2014; Falkheimer 2013; Olsson 2015), meaning that the origins and effects of threats and adverse events traverse functional, national, and cultural boundaries. Acknowledging the need for concepts that are useful in such complex processes, we rely on Klinke and Renn (2002) who see risk as the “possibility that human actions or events lead to consequences that harm aspects of things that human beings value” (p. 1071). This definition does not presuppose that risk is a given object. Hilgartner (1992) adds a communicative perspective on such risk objects. He states that “changes in the definition of risk objects can redistribute responsibility for risks, change the locus of decision making, and determine who has the right – and who has the obligation – to ‘do something’ about hazards” (Hilgartner 1992: 47). Therefore, risk is not just something we communicate about. It is defined, negotiated, and shaped through communication.

Following this definition of risk, we conceptualise crisis as a variety of severe events negotiated through communication. Crisis is defined by Boin et al. (2005: 2–3) as situations where widely shared values are under immediate threat, leading both to demands for prompt action and to uncertainties regarding the full extent and consequences of the event and possible remedies.

Approaching risk and crisis as transboundary phenomena entails the bridging of relations that are somehow tied apart. Research has particularly drawn attention to the limitations in understanding risk and crisis within the confines of the nation state. Threats and adverse events increasingly develop and have effects across nations and their political and jurisdictional mandate, necessitating cooperation across nations and sectors (e.g., Olsson 2015). Whereas these limitations of a national perspective are widely recognised, we suggest that there are more boundaries – surfacing at the local or global level – that transboundary risk and crisis communication should take into account. In general terms, a boundary entails relations marked by similarity and difference. This article stresses some boundaries, including differences in social media access and use, and the partial geographical scope of the field.

We approach social media as web-based platforms whose key applications offer opportunities for users to communicate and maintain social relationships with multiple individuals. The most optimistic accounts of social media convey that platforms such as Facebook, Twitter and YouTube offer anyone with necessary knowledge and Internet-access the opportunity to engage in many-to-many dialogue across the globe (Hogan & Quan-Haase 2010). Nevertheless, empirical studies show strong patterns of homophily in social media, in that elites follow elites whereas “ordinary” citizens rarely get attention. Wu et al. (2011) demonstrate that 50 per cent of the consumed content on Twitter was produced by 20 000 elite users.

With these limitations in mind, social media still offers a range of possibilities for risk and crisis communication. Among examples mentioned in the literature is the ability to facilitate information and opinion sharing, but also to engage in dialogue with and build relationships with key stakeholders. Initiating dialogue before risks manifest and/or crises occur can help an organisation navigate a crisis (Veil, Buehner & Palenchar 2011). Listening in on the conversations that take place on social media is also valuable for organisational issues management, here defined as the attempt by an organisation to map important themes, trends or problems that in the long run represent a threat to the organisation (Heath & Palenchar 2008). Social media analysis can assist organisations in identifying and understanding issues that might eventually rise to prominence. By addressing these issues early on, organisations might be able to avoid the issue evolving into a full-blown crisis. Social media can also help empower relevant members of the public to develop and use emergency responses so that crisis outcomes can be mitigated (Palenchar 2009). Using social media in crisis communication may also prove more favourable among stakeholders than issuing press releases via traditional media (Xu & Wu 2015).

Previous literature reviews have typically focused on developing best practices. In their review, Veil, Buehner and Palenchar (2011) formulate advice for how organisations should establish policies, prepare pre-event logistics, and partner with the public as well as authoritative sources. Other suggestions relate to crisis response, such as meeting the needs of the media, communicating with compassion and empathy, and accepting uncertainty and ambiguity. Furthermore, messages of self-efficacy should be provided. Following the hurricanes “Katrina” and “Rita”, volunteers created a people-finder blog that helped blog-users finding missing relatives and friends. After the Haitian earthquake, volunteers organised donations via Twitter (Smith 2010). A particularly interesting point to us was that Veil, Buehner and Palenchar (2011) found little discussion of difference in access to technology. The latter has been dubbed “the digital divide” and can be said to encompass three distinct aspects, namely a global, social, and democratic divide. The former refers to differences in access to technology, the second concerns how a population can be seen as information rich or poor, while the last aspect relates to differences in the use of digital media to partake in public life (Norris 2001).

The literature review of Fraustino, Liu and Jin (2012) concludes that that there is substantial evidence that social media use increases during disasters (see Sweetser & Metzgar 2007), and that media choices are crucial – Twitter being ideal for quick, short, news updates (see Schultz, Utz & Göritz 2011). Repeatedly, however, they find that the studies for the most part focus on single cases and single media – mostly Twitter – but results are taken to represent social media in general. Moreover, Fraustino, Liu and Jin (2012) point out that there are several types of social media users, ranging from those who only take part in information without ever writing or sharing a post (“lurkers”), to those who have thousands of followers and sometimes even organise public events via social media during crises (Jin & Liu 2010; Paine 2007). They have various reasons for using social media during crises, such as information retrieval (Jin, Liu & Austin 2014), connecting with family and friends (Procopio & Procopio 2007), and also stress-relieving humour (Liu, Jin & Austin 2013). Importantly, we also know that some people do not use social media because of issues with privacy and security (Yates & Paquette 2011), and doubts about medium and source credibility and information accuracy (Veil, Buehner & Palenchar 2011). Fraustino, Liu and Jin (2012) argue that most of the research has been descriptive, and call for investigations of causal relationships that can guide best practices.

Ruggiero and Vos (2014) focused on monitoring in crisis communication, and found that most of the research discussed the use of YouTube, with Twitter and blogs trailing behind. Given its popularity, it is notable that Facebook was not studied. The authors synthe sise and present advice regarding, for instance, how practitioners should look for the communication patterns of emerging networks, such as how a network is developed from a tweet via re-tweets, or a blog post that is shared multiple times on social networking platforms.

The present study will comprise a comprehensive quantitative content analysis, since the above literature reviews do not cover volume trends (e.g., Alexander 2014; Fraustino, Liu & Jin 2012; Veil, Buehner & Palenchar 2011), or only include a limited number of studies (e.g., Ruggiero & Vos 2014).

Firstly, it has been postulated both that the field is growing (Austin, Liu & Jin 2012), and that “it is still quite limited” (Alexander 2014: 719). Yet another study states that research on crises and social media peaked, quantitatively speaking, in 2009–2010 (Simon, Goldberg & Adini 2015). Our first research question, therefore, is:

1) What is the volume trend regarding research on social media and risk/crisis communication?

There are indications that most of the research in this field rather focuses on “relief and rehabilitation than on risk reduction” (Gender and Disaster Network 2008). Another study criticizes the fact that the field is lacking experimental data (Fraustino, Liu & Jin 2012). Hence, our second research question is:

2) To what extent is the field drawing lessons from risk/pre-crisis situations, crisis situations, or experimental, fictitious scenarios?

It is important to note which platforms the research on risk and crisis addresses, as people of different generations use different social media (Bowe & Wohn 2015) and platform popularity fluctuates. We need to be aware that some of today’s social media may soon be yesterday’s technology. Our third research question, therefore, is:

3) What is the trend in terms of social media platforms that the research focuses on?

We know that public relations research has mainly focused on the corporate sector (Olsson 2014), but we have little knowledge of which actors and organisations the research on risks, crises, and social media addresses. Hence, our fourth research question is:

4) Which groups of social media users and/or case organisations are studied?

Communication theory and research were long developed in Europe and the US, and much less so in other parts of the world. Gunaratne (2009) stresses that the power imbalance within global academia – including media and communication studies – has been as uneven as in the sectors explained in Wallerstein’s world systems theory. So, although risks and crises are part of lived experience across the globe, and social media use is expanding fast (Kemp 2015), there may be a Western bias in terms of attention paid to certain regions. We therefore pose the following research question:

5) Where are the empirical studies geographically focused?

Following up on RQ5, we are also compelled to address how the research speaks to issues of ICT and social media access (cf., Mossberger, Tolbert & Stansbury 2013). Hence, our final research question is:

6) Do the studies mention the digital divide or other issues of social media access?

To answer the research questions, the lead author searched for peer-reviewed articles listed in the databases Communication and Mass Media Complete (CMMC) and Summon, rendering suitable results in 106 English language journals (see Appendix). We assumed that the field is wider than the discipline of media and communication studies. For example, we included articles from informatics and computer science journals if they were about social media content and users, and from health science journals if they dealt with collective risks and crises, such as epidemics, and social media. The following keywords were applied for the period 2009–2015: internet, online, social media, Facebook, Twitter, YouTube, blog*, or Instagram; all of which were combined with risk*, crisis, or disaster*.

Our methodology is best described as a systematic review, in that it includes an exhaustive search and systematisation of several research characteristics in studies that deal with a common theme, but which may be quite heterogeneous in design and focus. Select results are explained in narrative form. This contrasts another form of review, the meta-analysis, which statistically combines the results of studies that are homogenous in design and focus (Grant & Booth 2009).

To retain some similarity and comparability, book reviews, editorials, or theoretical articles without empirical data were excluded from the final sample, along with articles that did not treat social media and risk or crisis as central themes. An exclusion criterion was whether the study could have been carried out, presented, and understood without the dimension of social media and risk or crisis. The final sample consisted of 200 articles, with Public Relations Review taking the lead with 27 published pieces. Content analysis (Krippendorff & Bock 2008) was performed on these articles, with coding of six variables: 1) publication year; 2) whether the main focus is on real life risks, crises, or fictitious cases; 3) social media platform(s) studied (Twitter, Facebook, blogs, YouTube, message boards, Instagram/Flickr, wikis, other); 4) whose social media use is studied (citizens, public institutions, students, journalists, business organisations, NGOs, public relations practitioners, universities/colleges, other); 5) which geographic area the data is drawn from; and, 6) whether or not the studies mentioned the digital divide or related issues of ICT access. For variable 5 we used the same geographical units as the We Are Social reports (e.g., Kemp 2015).

After an initial round of coding, the codebook was improved. The lead author then coded 150 articles, while a research assistant coded 80, allowing for an intercoder reliability test regarding 15 per cent of the articles (n = 30). Using Krippendorff’s alpha, the result was .92 and above for each variable, which must be deemed satisfactory. Almost exclusively, the study used nominal variables. In order to identify trends over time, a multiple variable approach was employed (Riffe 2014). The use of a total sample strategy rendered no need for calculating statistical significance.

Between 2009 and 2015, an increasing number of studies dealing with risk, crisis, and social media were published. As shown above, most of the studies focused on crisis communication, more specifically, the use of Twitter, with Facebook placing second. About one-tenth of these studies used experimental scenarios and tested causal relations in social media crisis communication. By and large, the studies focused on citizens or undefined social media users. For the most part, the literature concerned real-life crisis communication in a North American setting. While we maintain that these traits and the presented numbers are interesting in themselves, they also indicate some areas that need more attention.

First, there is a need for more research on other types of social media than Twitter, Facebook, blogs, and YouTube, particularly since other types of platforms can provide further and better opportunities for risk and crisis planners. One unique study focused solely on wikis: Yates and Paquette (2011) presented a number of arguments for the usefulness of this social medium in times of crisis. A worthwhile direction for future research would be to examine wikis, and visuals such as editable maps, in crisis relief efforts. Still, we are aware that researchers also risk addressing the social media of yesterday, since technological development moves so fast. This, in turn, necessitates the identification of certain defining traits that are common across different social media, like visibility, editability, persistence, and association (Treem & Leonardi 2012).

Second, as with much research, the studies about risk, crisis, and social media focuses mainly on the United States, followed by Western Europe and East Asia. Thus, our knowledge of how social media is used and understood in connection with risks and crises in other cultures is limited. Given that social media users on a large continent (South America) and in other major regions are largely overlooked, it is important to extend the research foci to include non-Western contexts and countries in the South. Otherwise, this stream of research becomes too particularistic and Western-centred without even demonstrating an awareness of these limitations. Risk and crisis situations are indeed global, and social media use is growing worldwide. For instance, there are 103 million active social media accounts in Africa, and nearly 197 million in South America (Kemp 2015). A way forward would be to further explore risks and crises as transboundary phenomena, realising that some boundaries are reproduced in a Euro-American research tradition, and that these boundaries are not only national, but separate regions and continents. The approach to risk and crisis transboundary phenomena should include the design of the research and its geographical focus. Some conceptualisations of risk and crisis can help facilitate a cross-cultural focus, of which a few were raised in the theory chapter.

Thirdly, a rather small part of the literature focuses on risk communication and social media. We can only speculate as to why crisis scenarios are studied much more often. Crises may be found more practical to study than risk situations in that they are perceived to be temporally and spatially limited events (Berthod, Müller-Seitz & Sydow 2013). It might also be easier to get funding for the research, since crises receive much attention, raise questions about accountability and improvement, and prompt the organisations involved to protect their reputation. As a less-studied area, risk communication and social media make up a fruitful avenue for future research. Here, for instance, would be an excellent opportunity for practitioners and researchers alike to put insight from issues management into use. As mentioned, issues management could be thought of as a foresight-orientated, proactive part of public relations (Heath & Palenchar 2008). In this regard, social media can be a listening post, but also a place where organisations can participate in dialogue with engaged stakeholders and risk bearers, and float suggestions before implementing them on a full scale. Through social media, concerns can be picked up and addressed.

Fourth, an important finding of the article is that there is so little attention being given to a notion like the digital divide, and to other issues relating to the uneven use and reach of risk and crisis communication through social media. For the most part, the studies seem to take for granted that the stakeholders and risk bearers are online and engaged. Risk and crisis planners must, however, also address resource-poor stakeholders and communities. This then, gives us an opportunity to call for using social media in tandem with traditional media, and not to regard social media as a panacea for problems of risk and crisis (Kent 2010; Kent, Saffer & Pop 2013). There are, for instance, many more people who have access to the older cell phone features like text messaging compared with web features requiring a smartphone with an internet subscription (Kemp 2015). This makes us believe that mass text messages can effectively complement the Web 2.0 social media in times of crisis. This option, and several non-digital alternatives, need to be studied further. Also on the topic of access, only four studies mentioned people with disabilities and their use of social media in risk and crisis situations. In line with the UN convention on the rights of persons with disabilities, several nations have laws requiring that public authorities should promote universally accessible ICT solutions (UN General Assembly 2007) including social media platforms. An untapped venue for future research would thus be to study how social media in risk or crisis situations meets, or needs improvement in order to meet, the information needs of these groups.

Limitations: given that we solely focused on research in English, our review does not claim to cover all research in the area, and complementary reviews, including more languages, are needed. Our systematic review has also incorporated several themes, which implies that each one can only be dealt with briefly. Given this limitation, we want to encourage more reviews that go into one or more of the issues in greater depth. Selected themes could be the recent experimental studies and causal relationships in social media crisis communications, or non-Western research results and how they contrast with some of the Western research.