Viruses Going Viral: Impact of Fear-Arousing Sensationalist Social Media Messages on User Engagement

Abstract

Social media presents a promising yet competitive communication landscape for health practitioners and organizations during the rapid spread of an emerging infectious disease. The current study examined the association between the level of fear-arousing sensationalism and user engagement in 800 Facebook posts regarding the 2016 Zika virus outbreak. Results revealed that the majority of nonnews posts completely lacked sensationalist elements, and that user engagement increased significantly as the level of fear-arousing sensationalism increased from low levels. We discuss the potential value of message design that can engender an appropriate level of public concern and increase user engagement on social media.

Keywords

crisis communication health communication infectious disease risk communication social media

Infectious diseases have been among the foremost causes of morbidity and mortality around the world for centuries and remain so in current times (Fauci, Touchette, & Folkers, 2005; Nii-Trebi, 2017). Newly emerging or reemerging infectious diseases (EIDs) are defined as “infections that are rapidly increasing in incidence or geographic range” (Morse, 2009, p. 198). The most recent example is the Zika virus (a mosquito-borne infectious disease first identified in Uganda, Africa, in 1947), whose 2016 outbreak in the Americas was declared a public health emergency of international concern by the World Health Organization (WHO; Petersen et al., 2016).

After initial reports of local transmission in Brazil in May 2015, the Zika virus subsequently spread rapidly across the Americas (Petersen et al., 2016). During the peak of the outbreak in 2016, active transmission of the disease had been reported across 48 countries and territories (Pan American Health Organization/World Health Organization [PAHO/WHO], 2016b). Furthermore, cumulative estimates show that travel-associated and locally acquired cases of the Zika virus numbered more than 180,000 (PAHO/WHO, 2016a). Although many of those infected were asymptomatic, the virus was alarming due to its linkage with a significant rise in the number of infants born with microcephaly and other neurological complications (Focosi, Maggi, & Pistello, 2016; Petersen et al., 2016). In addition, the “lack of vaccines and reliable diagnostic tests, broad geographical distribution of mosquito species that can transmit the virus, and absence of immunity in newly affected countries” were causes for concern (Focosi et al., 2016, p. 227).

Social media offers new opportunities for health crisis and risk communicators to disseminate health information efficiently to broad audiences during public health emergencies such as EID outbreaks. Today, social media forms a major source of information and provides a singular platform for both consumption and generation of news and opinion (Gottfried & Shearer, 2016; Hermida, Fletcher, Korell, & Logan, 2012). Interest has been growing in academia on the use of social media during public health emergencies, with studies investigating topics like the impact of pregnancy on choice of social media as a source of knowledge (Chan, Farhadloo, Winneg, Jamieson, & Albarracin, 2018), correlation of volume of information on risk perception (Chan, Winneg, Hawkins, Farhadloo, Jamieson, & Albarracín, 2018), impact of discussion topics on user attitudes (Farhadloo, Winneg, Chan, Jamieson, & Albarracin, 2018), ideal timing for health messaging post crises (Black, Dietz, Stirratt, & Coster, 2015), and disease surveillance and forecasting (Hossain, Kam, Kong, Wigand, & Bossomaier, 2016; McGough, Brownstein, Hawkins, & Santillana, 2017; Rodriguez-Morales, Castañeda-Hernández, & McGregor, 2015).

The many-to-many format of social media provides a unique interactive setting where information can be received and retransmitted by users in near real time, allowing instant propagation of health risk–related information to ever-wider audiences (Bucher, 2002; Vos & Buckner, 2016). Harnessing this power of the medium depends on message design that motivates people to spread the message through their networks and beyond. Few studies, however, have examined the specific relationship between social media message attributes and increased social media engagement in the context of public health emergencies (e.g., Strekalova, 2017); however, a nascent, yet larger, body of research has started to investigate this phenomenon in other health contexts (Rus & Cameron, 2016; Strekalova & Krieger, 2017). Still, health communication in times of crisis and emergency risk diverges from traditional health communication in various aspects, including intensity, and therefore warrants special attention (Glik, 2007; Holmes, 2008).

The present study thus aims to add to the current body of research on EIDs and social media by providing a content analysis of a particular message attribute, namely, fear-arousing sensationalism, in messages on Facebook during the 2016 Zika virus outbreak in the Americas. Facebook was chosen as the social media platform in this study because it is the most popular social media platform worldwide, with a current base of 2.32 billion monthly active users (Statista, n.d.). This study also aims to examine the relationship between fear-arousing sensationalism and social media user engagement behaviors during this time. In the following literature review, we describe our underlying theoretical model, explaining how exposure to fear-arousing sensationalist messages about an EID may lead to heightened risk perceptions, which may in turn affect user engagement. The findings of this study offer valuable insights that may improve message design and dissemination efficiency of social media–based risk communication campaigns during public health emergencies.

Literature Review

Social Amplification of Risk Framework and Sensationalist Media Messages

Health communication during EID outbreaks can be significantly enhanced by incorporating basic principles and frameworks outlined in the field of crisis and risk communication (Glik, 2007; Holmes, 2008). One such framework, Social Amplification of Risk (SARF), denotes the phenomenon by which psychological, social, institutional, and cultural phenomena shape the social experience of risk (by amplifying or attenuating public risk perception), thereby contributing to public response behaviors and risk consequences (R. E. Kasperson et al., 1988). SARF identifies two stages of risk perception amplification: during the transmission of information regarding the risk and in the response mechanisms of the public (R. E. Kasperson et al., 1988). The present study focuses on the first stage. Within transmitted information, SARF identifies the following four information attributes that affect amplification of risk perceptions: sensationalism of risk-related information, the degree to which individuals or groups dispute risk-related information, symbolic connotations present in risk information, and the volume of information regarding a risk (R. E. Kasperson et al., 1988).

For the purposes of this study, we examined the first attribute identified by the model: sensationalism. Sensationalism is often defined as emotional or sensory arousal in viewers, aimed at provoking attention toward the media message in question (Grabe, Zhou, & Barnett, 2001). As per this definition, sensationalism is a construct involving two main dimensions: one that is emotional and the other that is sensory (Ihekweazu, 2017). The emotional dimension of sensationalism may be especially important within the given context of EID outbreaks. Messages that can cause emotional arousal are frequently used in the field of health communication, having been shown to yield better recall and greater persuasive effectiveness (Dunlop, Wakefield, & Kashima, 2008). In addition, health communication often relies on negative emotional appeals in its persuasive messages (Dunlop et al., 2008). Influencing behaviors through fear appeals, in particular, is the most known and frequently used persuasive message tactic in health communication campaigns, news reports, and advertisements (Chang, 2012; Cho & Salmon, 2006; Fung, Namkoong, & Brossard, 2011; Muthusamy, Levine, & Weber, 2009). Consequently, in line with prior research, which has examined media sensationalism during public health emergencies (Gorney, 1992; Ihekweazu, 2017), we focus on the emotional dimension as it pertains to fear arousal.

Fear-Arousing Sensationalist Media Messages

The current study followed Ihekweazu’s (2017) sensationalism definition, which characterized the construct as a message attribute, which provokes “emotional stimulation, particularly fear, above what is deemed appropriate by societal standards” (p. 743). Some topics may be considered inherently sensational such as crime, celebrities, and scandals (Grabe et al., 2001). However, by introducing certain elements and/or manipulating the message, other topics that are viewed as lacking this inherent quality, such as public health messages, may nonetheless become sensationalized because of their way of presentation to the public (Ihekweazu, 2017; Slattery, 1994). Prior research indicates that there are various ways in which sensationalism in media messages has been typified. One prominent example is that of a study by Grabe et al. (2001), in which the authors explain sensationalism as it applies to television broadcasts, such as specific camera maneuvers and angles, decorative effects, and audio techniques. However, these aspects of televised sensationalism are not necessarily fear inducing. In addition, online social media messages “represent the convergence of all forms of traditional mass media transmissions” which include text, picture, video, or any combination of these (Chung, Nam, & Stefanone, 2012, p. 175). As such, these audiovisual production techniques may not be applicable to some messages in the realm of social media, particularly those which are purely textual.

Given this, the current study would benefit from focusing on sensationalism in the message itself. As Ihekweazu (2017) noted, “sensationalism in messages can involve the use of fear-evoking words or graphic imagery” (p. 742). Such a typification has been applied to account for sensationalism across various forms of traditional mass media (Gorney, 1992; Ihekweazu, 2017). We hence follow prior literature, which has operationalized sensationalist media in terms of emotional stimulation through the use of intense or extreme fear-inducing word choices and linguistic packaging (in that the message uses emotionally charged rather than neutral language) and through the use of fear-inducing or threatening graphics or visuals (Gorney, 1992; Ihekweazu, 2017). Such messages may increase attention and curiosity, as well as anxiety and fear in audiences. When fear-inducing messages are constructed in this highly intense manner, using vivid language and images to describe negative health consequences, they are by definition highly sensational as well (Rhodes, 2017). However, such fear-arousing sensationalist messages differ from their nonsensational counterparts, as the latter may depict the negative consequences that individuals will experience unless they start preventive behavior or stop risky behavior, without the use of the intense or extreme dramatization (Cho & Salmon, 2006).

Source “Amplification Stations”

Within SARF, health risk information is communicated through sources, also called “amplification stations,” which disseminate such messages to wider audiences (J. X. Kasperson, Kasperson, Pidgeon, & Slovic, 2003). These sources may include news outlets, research scientists, government agencies, and regular individuals, among others (J. X. Kasperson et al., 2003). The proliferation of information from these sources is possible in the modern Web 2.0 environment, which has removed the substantial technological skill and financial investment barriers posed by traditional mass media for creating and sharing news and information (Metzger, Flanagin, Eyal, Lemus, & McCann, 2003; Metzger & Flanagin, 2013; Thackeray, Neiger, Hanson, & McKenzie, 2008). According to SARF, the salience of certain aspects of health risk information can increase depending on which of these source “amplification stations” is communicating the message (J. X. Kasperson et al., 2003).

Along these lines, news organizations have long been well-known for introducing sensational aspects into their media messages. This is due to the market-driven nature of the journalism industry, with the primary goal of capturing audience attention for increased viewership ratings, rather than eliciting protective/preventative behaviors (Arkin, 1990; Leask, Hooker, & King, 2010). In fact, public health organizations, experts, doctors, and other members of the health community, often try to mitigate media hype, fear, and anxiety in a population by focusing on creating factual messages that do not appeal to emotion (Dalrymple, Young, & Tully, 2016; Glik, 2007). We therefore posit that the sensationalizing of messages attributed to news sources in traditional mass media is likely to manifest in the online medium as well. In addition, it may also be instructive to examine if various types of nonnews sources, such as those mentioned above, differ in terms of the amount of sensationalism in their online media messages. Although such sources may not have as much incentive to sensationalize their messages as news outlets, they may have distinct goals in terms of content creation, which may lead to varying levels of sensationalism (J. X. Kasperson et al., 2003).

Hypothesis 1: There will be higher levels of fear-arousing sensationalism in Zika-related Facebook posts from news sources compared with Zika-related Facebook posts from nonnews sources.

Research Question 1: Will there be a difference between nonnews sources in terms of the level of fear-arousing sensationalism in their Zika-related Facebook posts?

“Amplified” Risk Perceptions and User Engagement on Social Media

The above described fear-arousing sensationalist media messages are likely to influence audience perceptions of risk. According to SARF, sensationalism in the transmission of information “amplifies” or heightens risk perceptions about the particular hazard in question, the rest of this section discusses this dynamic in light of theory and research from the field of psychological risk analysis. According to Slovic (1987), risk perceptions refer to “people’s subjective assessments of the possibility that negative outcomes or diseases may occur” (p. 281). The affect heuristic and risk-as-feelings hypothesis refer to a process requiring low cognitive effort that people are prone to using during risk assessment, in which emotional arousal is posited to affect risk perceptions (Finucane, Alhakami, Slovic, & Johnson, 2000). Previous research has supported this assertion by showing that aroused emotions are significant determinants of risk perception, rather than being solely based on cognitive aspects (Loewenstein, Weber, Hsee, & Welch, 2001; Rundmo, 2002; Slovic, Finucane, Peters, & MacGregor, 2005). Within this realm, the role of negative emotional arousal in shaping risk perceptions has been identified as a particularly influential factor (Loewenstein et al., 2001). All negative emotions, however, do not work in the same way to affect risk perceptions, and in fact may have opposite effects. For example, research has shown that fear amplifies risk perceptions, while anger seems to attenuate them (Lerner, Gonzalez, Small, & Fischhoff, 2003; Lerner & Keltner, 2000). Thus, it appears that the emotional arousal of fear forms a part of the affective component of people’s risk perception and can be used to increase risk perceptions by appeals directed at feelings of vulnerability, which are formed through appraisals of uncertainty and situational control (Dunlop et al., 2008).

Previous research suggests that people are more likely to engage in interpersonal communication when exposed to messages that elicit negative emotions, and more recent research reveals that fear-arousing news messages may increase interpersonal communication regarding a risk by amplifying risk perceptions (Dunlop et al., 2008; Luminet, Bouts, Delie, Manstead, & Rimé, 2000; Paek, Oh, & Hove, 2016; Rimé, 1995). These studies support the idea that people learn how to deal with fear and other negative emotions by ways of interpersonal communication, which can help people internalize the threats they fear and figure out how to cope with them (Dunlop et al., 2008; Paek et al., 2016; Rimé, 1995). Amplified risk perceptions may thus lead to an increase in social media activity, which is essentially a form of “human communication, possessing characteristics of participation, openness, conversation, community, and connectedness,” and is a measure of public reaction to crises (Schultz, Utz, & Göritz, 2011; Veil, Buehner, & Palenchar, 2011, p.110). Within the realm of social media activity, the current study is particularly focused on “user engagement,” which can be conceptualized as activity resulting from motivational drivers, and can be addressed from the perspective of measuring actions undertaken by a user, with different measures being applied depending on the possibilities offered by the platform (Lehmann, Lalmas, Yom-Tov, & Dupret, 2012; Van Doorn et al., 2010). On Facebook in particular, users can engage with messages (called “posts”) by (a) commenting on existing posts, (b) indicating interest in a post by clicking the “reaction” button, and (c) sharing the post on their profile or a friend’s profile. Our literature review thus far makes it a plausible a priori assumption that fear-arousing sensationalist messages about a risk will increase user engagement on Facebook, as depicted in the model presented in Figure 1, which is thus assumed to be the underlying mechanism for the following hypotheses of the study:

Hypothesis 2: Higher levels of fear-arousing sensationalism in Zika-related Facebook posts will be associated with higher number of post reactions.

Hypothesis 3: Higher levels of fear-arousing sensationalism in Zika-related Facebook posts will be associated with higher number of post comments.

Hypothesis 4: Higher levels of fear-arousing sensationalism in Zika-related Facebook posts will be associated with higher number of post shares.

Figure 1.

Underlying model for a priori hypothesis formulation.

Method

Sample and Procedure

We used the search term “#Zika,” one of the most popular hashtags about the Zika virus in 2016 according to Google Trends, and based on the results, selected an 8-month period with the highest #Zika internet search–related activity. We then used this hashtag in Facebook’s graph search feature to find posts about the Zika virus in each of the 8 months. Since the total number of search results was not reported by Facebook, it was not feasible to determine a random selection from within these results. Consequently, using the natural sorting order Facebook employs for its search results, a convenience sample of the first 100 English language posts from each month were selected for analysis.

To achieve intercoder reliability, four undergraduate students were selected to independently code the sample of posts. Coders were given a coding manual to guide categorization of content and were given a 6-hour training in which they were familiarized with the coding manual. After the training period, they were all given a subsample of Facebook posts from the main sample to code in isolation from one another. Since there were more than two coders in this content analysis, we used Krippendorff’s alpha for intercoder reliability calculation, as suggested by Hayes and Krippendorff (2007). The alpha coefficient values ranged from .70 to .99.¹ After this, coders received an evenly distributed share of Facebook posts to complete coding of the main sample.

Measures

Source Type

For each Facebook post included for analysis in the study, we measured a nominal level variable that captured whether the author/source was a public figure (e.g., politician, celebrity), news resource (e.g., news, media, broadcasting companies), private sector organization (e.g., pharmacy, clinic), government organization (e.g., Centers for Disease Control and Prevention), individual (private citizen), expert (e.g., doctor, research scientist), nonprofit organization (e.g., WHO), academic institution (e.g., university), or none of these.

Fear-Arousing Sensationalism

Ihekweazu’s (2017) sensationalism measure was adapted to include elements relevant to the context of the current study. Specifically, fear-arousing sensationalism was measured as an ordinal-level variable (Low, Moderate, or High) based on the presence or absence of the following two items:

use of loaded, fear-inducing words (e.g., “Zika crisis,” “Zika emergency,” “Killer virus”) in the Facebook post

use of provocative, fear-arousing graphics and visuals (e.g., close-ups of blood-filled mosquitoes, crying infants with microcephaly) in the Facebook post

This means that the variable fear-arousing sensationalism would have a value of Low if both fear-inducing terminology and provocative imagery were missing from the Facebook post, a value of Moderate if either was present in the post and a value of High if both were present in the post.

User Engagement

On Facebook, users can engage with messages (called “posts”) by (a) commenting on existing posts, (b) indicating interest in a post by clicking the “reaction” button, and (c) sharing the post on their profile or a friend’s profile. An important point to note is that the conceptual distinctiveness and motivations of users to use each method of engagement has not yet been established in the context of health-related messages (Rus & Cameron, 2016). Due to this, we refrained from summing them up together to form an overall engagement measure but rather measured each one of them separately as three ratio level variables: number of post reactions, number of post comments, and number of post shares on a post.

Efficacy (Covariate)

Self-efficacy refers to the degree to which people believe in their ability to perform particular behaviors to influence events that may affect their lives (Bandura, 1997). Efficacy information in messages thus instructs individuals about ways to effectively and appropriately respond to an event, and hence may aid in reducing associated risks (Viel, Reynolds, Sellnow, & Seeger, 2008). Prior research has found that efficacious information tends to receive higher levels of online user engagement (Cappella, Kim, & Albarracín, 2015). For this reason, Ihekweazu’s (2017) efficacy measure was adopted for the current study. Individual efficacy information was thus measured as an ordinal level variable (None, Low, Moderate, High) based on the following presence or absence of the following three items in the Facebook post:

identification of the specific signs and symptoms of the Zika virus (e.g., fever, rash, joint pain)

personal protective or diagnostic measures to take in order to reduce one’s risk of contracting the disease or to detect the Zika virus (e.g., mosquito repellent, blood tests)

information about how the Zika virus is transmitted (e.g., sexual transmission)

This means that the efficacy variable would have a value of None if no efficacy information was present in the Facebook post, a value of Low if any one type of efficacy information was present in the post, a value of Moderate if two types of efficacy were present in the post, and a value of High if all three types of efficacy information were present in the post.

Data Analysis

Kruskal–Wallis H test was used to determine the differences between groups (low, moderate, and high fear-arousing sensationalism) in terms of user engagement. One-way analysis of variance was not appropriate as normality tests (Shapiro-Wilk and Kolmogorov-Smirnov) of all three dependent variables (reactions, comments, sharing) were significant even after log transformations of the data. In addition, 14, 3, and 5 outlier values were excluded from analysis for reactions, comments, and sharing, respectively. This was done by calculating standardized scores and eliminating those data points which were more than 3 standard deviations away from the mean.

Results

We first carried out a descriptive analysis of the covariate in the study: efficacy. For the efficacy variable, results showed that the None category accounted for 58.4%, Low category accounted for 30.9%, Moderate category accounted for 7.5%, and the High category accounted for 3.3% of all posts (N = 467, N = 247, N = 60, N = 26, respectively). A series of chi-square tests were subsequently performed to check the distribution of covariates between the levels of fear-arousing sensationalism. No significant difference was found between the levels of fear-arousing sensationalism in terms of efficacy information, χ²(6) = 10.08, p > .05.

A descriptive analysis was also conducted for the source type variable. There were nine possible categories for this variable: public figure, news resource, private organization, government organization, individual, expert, nonprofit organization, academic institution, and other. Of these, news resources were the most significant source within the sample of posts about the Zika virus on Facebook within the study time; 51.1% of all analyzed posts originated from them, followed far behind by public figures (13.1%), private organizations (10.5%), governmental organizations (8.0%), and nonprofit organizations (7.1%). All other sources each accounted for less than 5% of the posts analyzed.

A descriptive analysis of the fear-arousing sensationalism variable showed that the Low category accounted for 39.9%, Moderate category accounted for 45.7%, and the High category accounted for 14.4% of all posts (N = 319, N = 365, N = 114, respectively). After the deletion of outliers, another descriptive analysis was run on the continuous variables of total number reactions (N = 786, M = 186.87, SD = 419.56), comments (N = 797, M = 15.64, SD = 41.44), and shares (N = 795, M = 111.93, SD = 377.76), respectively.

To address Research Question 1, a chi-square test was conducted to examine whether the eight nonnews sources (public figure, private organization, government organization, individual, expert, nonprofit organization, academic institution, and other) differed in terms of the level of sensationalism in their Zika-related Facebook posts. No significant difference was found in the level of fear-arousing sensationalism among nonnews sources, χ²(14) = 16.12, p > .05. We hence decided to collapse these categories into one group, thereby transforming the categorical variable of source type into one that had only two groups: nonnews sources (N = 391) and news sources (N = 409).

To test Hypothesis 1, a chi-square test was conducted crossing the variables source type and fear-arousing sensationalism. It was found that there was a significant difference in the levels of fear-arousing sensationalism according to source type, χ²(2) = 50.92, p < .05. Table 1 presents a post hoc analysis, which shows that a significantly smaller percentage of Facebook posts from news sources were categorized as having low fear-arousing sensationalism (37%), compared with Facebook posts from nonnews sources (63%). Conversely, a significantly larger percentage of Facebook posts from news sources were categorized as having moderate or high fear-arousing sensationalism (57% and 72%, respectively), compared with Facebook posts from nonnews sources (43% and 28%, respectively). Hence, Hypothesis 1 was supported.

Table 1.

Chi-Square Post Hoc Analysis of Source Type and Fear-Arousing Sensationalism.

Source_Type × Fear_Arousing_Sensationalism Cross-tabulation
		Fear_Arousing_Sensationalism
			Low	Moderate	High	Total
Source_Type	Nonnews	Count	201	157	32	390
		Expected count	155.7	178.2	56.1	390
		% Within Source_Type	51.50	40.30	8.20	100.00
		Adjusted residual	6.5	−3	−4.9
		p Value	.000	.003	.000
	News	Count	118	208	83	409
		Expected count	163.3	186.8	58.9	409
		% Within Source_Type	28.90	50.90	20.30	100.00
		Adjusted residual	−6.5	3	4.9
		p Value	.000	.003	.000
Total		Count	319	365	115	799
		Expected count	319	365	115	799
		% Within Source_Type	39.90	45.70	14.40	100.00

In testing Hypothesis 2, statistical analysis by means of Kruskal-Wallis H test revealed a significant difference in number of reactions between levels of fear-arousing sensationalism, χ²(2) = 21.78, p < .05, with a mean rank number of reactions of 351.85 for low, 408.09 for moderate, and 460.46 for high level of fear-arousing sensationalism. Post hoc analysis (Table 2) showed that the mean rank of the number of reactions increased significantly, at the p < .05 level, between the low and moderate, and from low and high fear-arousing sensationalism levels. However, there was no significant difference between the moderate and high level of fear-arousing sensationalism in terms of the mean rank of the number of reactions. Hence, this study found partial support for Hypothesis 2.

Table 2.

Post Hoc Analysis for Number of Reactions.

Pairwise comparisons
Sample 1-Sample 2	Test statistic	SE	Standard test statistic	Significance	Adjusted significance
Low-Moderate	−56.235	17.500	−3.213	0.001	0.004
Low-High	−108.606	25.026	−4.340	0.000	0.000
Moderate-High	−52.371	24.605	−2.128	0.033	0.100

Note. Each row tests the null hypothesis that the Sample 1 and Sample 2 distributions are the same. Asymptotic significances (two-sided tests) are displayed. The significance level is .05. Significance values have been adjusted by the Bonferroni correction for multiple tests.

Similarly, in testing Hypothesis 3, statistical analysis by means of Kruskal-Wallis H test revealed a significant difference in number of comments between levels of fear-arousing sensationalism, χ²(2) = 16.11, p < .05, with a mean rank number of comments of 364.52 for low, 410.11 for moderate, and 456.61 for high level of fear-arousing sensationalism. Post hoc analysis (Table 3) showed a significant increase, at the p < .05 level, in the mean rank of the number of comments between the low and moderate, and low and high levels of fear-arousing sensationalism. However, there was no significant difference between moderate and high levels of fear-arousing sensationalism in terms of the mean rank of the number of comments. Hence, this study found partial support for Hypothesis 3.

Table 3.

Post Hoc Analysis for Number of Comments.

Pairwise comparisons
Sample 1-Sample 2	Test statistic	SE	Standard test statistic	Significance	Adjusted significance
Low-Moderate	−45.588	17.128	−2.662	0.008	0.023
Low-High	−92.088	24.353	−3.781	0.000	0.000
Moderate-High	−46.499	23.961	−1.941	0.052	0.157

Note. Each row tests the null hypothesis that the Sample 1 and Sample 2 distributions are the same. Asymptotic significances (two-sided tests) are displayed. The significance level is .05. Significance values have been adjusted by the Bonferroni correction for multiple tests.

Finally, in testing Hypothesis 4, statistical analysis by means of Kruskal-Wallis H test showed a significant difference in the number of shares between levels of fear-arousing sensationalism, χ²(2) = 16.56, p < .05, with a mean rank number of shares of 361.86 for low, 410.87 for moderate, and 454.47 for high level of fear-arousing sensationalism. Post hoc analysis (Table 4) revealed significant increase, at the p < .05 level, in the mean rank of the number of shares between the low and moderate, and low and high levels of fear-arousing sensationalism. However, no significant difference was found between moderate and high levels of sensationalism in terms of the mean rank of the number of shares. Hence, this study found partial support for Hypothesis 4.

Table 4.

Post Hoc Analysis for Number of Shares.

Pairwise comparisons
Sample 1-Sample 2	Test statistic	SE	Standard test statistic	Significance	Adjusted significance
Low-Moderate	−49.012	17.298	−2.833	0.005	0.014
Low-High	−92.617	24.569	−3.770	0.000	0.000
Moderate-High	−43.605	23.172	−1.804	0.071	0.214

Note. Each row tests the null hypothesis that the Sample 1 and Sample 2 distributions are the same. Asymptotic significances (two-sided tests) are displayed. The significance level is 0.05. Significance values have been adjusted by the Bonferroni correction for multiple tests.

Discussion

In an increasingly populated and globalized world, EIDs pose an ever-present and heightened risk of crossing borders and rapidly spreading among diverse populations (Holmes, 2008; R. D. Smith, 2006). Unless they are contained, infectious disease outbreaks may not only threaten the physical health of millions of people but also negatively affect the social fabric of affected communities and impose severe and unequal economic burdens (Morens & Fauci, 2013). Additionally, the worldwide incidence of EIDs, such as the West Nile virus, Avian and Swine Influenza, Ebola virus, and severe acute respiratory syndromes, has increased in recent times (Nii-Trebi, 2017).

Since the uncertainty associated with EID outbreaks in today’s world presents unique and complicated challenges, the need for efficient dissemination of news and information addressing the health risk is of paramount importance (Glik, 2007; Holmes, 2008). Social media presents a powerful tool to achieve this; however, messages are likely to get lost in this increasingly dynamic and information-saturated environment. Even authoritative voices may get drowned out in the noise of the many-to-many communication model, thereby necessitating that public health professionals design messages in ways that they receive public attention even though there are many others competing for the very same. In this vein, some advocate for the use of strong emotional or scary appeals to help break through the cluttered plethora of messages (W. A. Smith, 2000), although such propositions must be based on solid empirical foundations (Freberg, 2012).

Our study provides some support toward the validity of this suggestion within the given context of health risk messages on social media. The results indicate that about 60% of the sample Facebook posts regarding the Zika virus contained moderate or high fear-arousing sensationalism levels. It was also discovered that increasing levels of fear-arousing sensationalism increases engagement in terms of reactions, comments, and shares, but only to an extent—when the level of fear-arousing sensationalism was high, there was no significant difference in these engagement behaviors compared with the moderate level. This may be explained by the limited capacity model of mediated message processing (Lang, 2000). In line with the model, it is plausible that fear arousal through sensationalism may result in initial attention gains; however, continued provocation may overwhelm the message receiver’s cognitive resources and result in a decreased effectiveness of message attentiveness and understanding (Lang, 2000).

The leveling off of social media activity for higher levels of fear-arousing sensationalism may also be explained by the original inverted U-shaped fear drive function (Janis, 1967), which explains individuals’ reactions to fear appeals. Janis (1967) posited that a moderate level of fear arousal is required to engender a motivational drive state for adaptive behaviors, but when fear levels get too high, people might engage in maladaptive behaviors, such as defensive avoidance, in which individuals might become inattentive or suppress thoughts about the communication. Our findings, which use real-world data, contradict many experimental studies which negate Janis’s U-shaped fear drive function by showing a monotonic effect of fear appeal strength on behavior; however, this may be because achieving high levels of fear, such as those experienced in real-world threat situations, is likely not viable in laboratory settings (Hartmann, Apaolaza, D’Souza, Barrutia, & Echebarria, 2014).

It is, however, important to note that such fear-arousing sensationalized news and information, if used as an isolated strategy, may cause unintended adverse consequences in the concerned population. Opposition for the use of strong emotional and fear-arousing communication stems from both moral and pragmatic arguments (Hastings, Stead, & Webb, 2004). The moral argument states that the manipulative use of strong emotional appeals infringes on people’s autonomy, thus violating principles of ethics (Guttman, 2011). Also, such media messages exaggerate and inflate the probability and severity of the risk in question, therefore, failing to meet stipulations for accuracy and correctness (Guttman, 2011). Past research has shown that such types of media representations that sensationalize health issues may lead to increased stress and panic among the general public (Pfefferbaum et al., 2014; Vasterman & Ruigrok, 2013; Vasterman, Yzermans, & Dirkzwager, 2005) and potentially minimize the public’s intentions to take recommended health actions to safeguard against health threats (Ihekweazu, 2017). Caution must therefore be exercised to be both ethical and pragmatic, and as such, communication interventions that aim to arouse risk perceptions by appealing to the emotion of fear should also include information that enhances people’s capacity to protect themselves against the risk (Guttman, 2011). The vital role played by individual self-assessment in the ability to effectively address a risk, called self-efficacy, is emphasized in theory on the use of persuasive fear-arousing messages (Guttman, 2011). The extended parallel process model predicts that individuals enact risk-ameliorating or “danger control” behaviors if fear arousal increases perceived threat levels, and people feel efficacious enough to combat the risk (Rimal, 2001; Witte, 1992). In this case, people become cognitively engaged and motivated to make deliberate efforts to protect themselves against the risk (Witte, 1994).

However, the results of the current study found that more than half of the posts did not contain any efficacy information. This is not an isolated case, as other studies have also shown low efficacy information on social media in times of public health emergency, such as an infectious disease outbreak (e.g., Vos & Buckner, 2016). In such cases, individuals may engage in maladaptive or “fear control” behaviors if fear arousal increases their perceived threat levels sufficiently, but their perceived ability to combat the risk is low (Rimal, 2001; Witte, 1992). Resultant behaviors, in this case, may include suppressing thoughts about the risk (defensive avoidance) or deliberate reactions against the communicator or message (Witte, 1994). Research supports this theoretical argument, showing that in order for fear-based campaigns to be effective, they must induce fear while also inducing a higher level of perceived efficacy for message acceptance to occur (Witte & Allen, 2000). When the audience feels that there is a higher level of fear than efficacy, message rejection and in some cases, boomerang responses occur (Witte & Allen, 2000).

In terms of the source “amplification stations” examined in this study, the results revealed an interesting trend in the relative amount of posts from various sources. Specifically, even though individuals form a clear majority of the Facebook user base, they barely functioned as sources of the posts about Zika. This, however, does not necessarily imply that individuals did not participate in the discourse on Zika. Rather, their mode of participation or engagement may have been different, in the forms of comments, likes, and shares of original content posts. This finding is in line with general individual user behavior on Facebook, where their preferred mode of engagement is liking, commenting, and sharing content that they have encountered, rather than creating completely new content (Pew Research Center, 2012). Conversely, organizations and experts can easily disseminate content with their large numbers of followers on social media, and thereby are more likely to primarily create new content, as opposed to engaging in liking and sharing behaviors.

Our results also revealed that news sources were significantly more likely to incorporate fear-arousing sensationalist elements in their Facebook posts about the Zika virus, whereas nonnews sources were more likely to steer clear from such dramatization entirely. However, referring to the latter, these messages may, in fact, create a disconnect between health communicators and the concerned population because of the difference in their perceived risk and the perceived risk that the organization is portraying (Dalrymple et al., 2016). Health crisis and risk communication demands complex considerations in both conveying information and managing fear, which may sometimes result in contradictory objectives. For instance, the need to assure the public against panic borne of the perception of personal risk, may result in an undervaluation of the national risk and subvert the subsequent justification for intervention, and such contradictions in communication may diminish public trust (Dalrymple et al., 2016). Speaking to this, some researchers suggest that public health messages should be designed in a way which engenders an appropriate level of concern (Dalrymple et al., 2016), which will, in turn, inspire necessary message dissemination and preparedness without causing panic.

The results of the current study contribute to this line of work as they indicate that introducing a moderate level of fear-arousing sensationalism in health-risk messages on social media may be able to increase dissemination efficiency. However, work remains to be done before such message design can be recommended. Specifically, future research is needed to test for causal associations between the independent and dependent variables under consideration. In addition, the theoretical mediation model underlying this study (Figure 1) remains to be tested, which would measure the affective responses and other psychological motivations to engage with content. Future research may experimentally test this model, which accounts for the mediating roles of emotional arousal and risk perceptions in explaining the effect of fear-arousing sensationalism in message content regarding the risk on social media user engagement. Given the earlier discussion on the importance of efficacy information in such content, future studies should also test the moderating effect of efficacy information on the relationship between fear-arousing sensationalism and user engagement.

Another critical question remains regarding the usefulness of the widespread dissemination of such messages on social media in effectively producing the desired attitudes and behaviors from the population. Message dissemination may be especially powerful in the current context of social media platforms like Facebook because, when users engage with health-related content, they increase the relevance of the topic for other members within their social networks. Thus, not only do they operate as disseminating agents for the original message but they also increase the message effect (Strekalova, 2017). Additionally, message propagation and dispersion through such platforms may increase the likelihood of message recipients to heed the advice of risk-related communication because of two notable factors known to drive behavioral change. The first being that individuals in social media are likely to be exposed to the same message multiple times (Centola, 2010), and the second, that they are likely to be exposed to information coming from a trusted source (such as a friend or family member within their social network; Liu, Fraustino, & Jin, 2015). However, the dissemination of health information may help increase target behaviors if the message is being endorsed through online engagement (Southwell & Yzer, 2007). If engagements are negative instead, such as comments criticizing the message, then unintended backlash effects can occur (Southwell & Yzer, 2007). Hence, to gain a better understanding of what impact such messages might have, future research should go a step beyond studying the number of likes, shares, and comments, by using manual content analysis or automated sentiment analysis to more deeply examine the online engagement of individuals (Cabling et al., 2017; Feinberg et al., 2015; Vos & Buckner, 2016).

Since this study limited its scope to explore sensationalism only, future research on SARF should test how uncertainty, symbolic connotations, and message volume may also contribute to increased user engagement, to gain a better understanding of the framework as a whole. In addition, other message attributes such as post length, sentiment, message framing, and potential interactions between various attributes should also be tested. Future research should also explore how message attributes may contribute to increased user engagement on other social media platforms and seek to compare results between platforms to understand potential commonalities. Moving beyond health communication related to EIDs, we suggest that future research broaden the horizon to other public health emergency contexts, in order to compare the effect of various attributes that may increase user engagement on social media across diverse health crisis and risk communication topics. Last, the current study was limited by the use of a convenience sample. Although it is very difficult to obtain a simple random sample of social media posts, future research may use data mining approaches to gather a more representative sample.

In conclusion, our results indicate that health communicators may stand to benefit from including a moderate level of fear-arousing sensationalism to disseminate their message through increased user engagement on social media. Our discussion highlights the ethical and pragmatic considerations in doing so and points out the need for future research in this area. The findings of this study are important for health crisis and risk practitioners, from national and international organizations, to deploy messages in the competitive social media landscape. We hope these findings will help increase the number of at-risk individuals exposed to, and potentially responding to, public health and safety warnings during EID outbreaks and other health emergencies.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

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

ORCID iD

Khudejah Ali

Notes

Author Biographies

Khudejah Ali is a doctoral candidate in the School of Communication at the University of Miami. Her primary research program centers on testing the effectiveness of social media as a communication tool for health promotion and behavior change, with an emphasis on health risk and crisis communication in times of public health emergency.

Khawaja Zain-ul-abdin (PhD, Florida State University) is an assistant professor in the Suleman Dawood School of Business at the Lahore University of Management Sciences. His primary research interests center on developing and testing new media communication methods and metrics for social media marketing, cocreated advertising, viral communication engineering, health marketing, and online consumer behavior.

Cong Li (PhD, University of North Carolina at Chapel Hill) is an associate professor in the School of Communication at the University of Miami. His primary research program centers on testing the effectiveness of personalized information in a computer-mediated communication environment. He has authored two books and published on a variety of journals including Human Communication Research, Media Psychology, Communication Research, and others.

Lisa Johns is a doctoral candidate at the University of Miami’s Leonard and Jayne Abess Center for Ecosystem Science and Policy. Her research focuses on the media practices and production strategies of environmental nongovernmental organizations, as well as the examination of formal and narrative aspects employed in short-form, digital video as components of broader Internet-based environmental awareness-building campaigns.

Ayesha Aziz Ali is a medical student at the Florida International University’s Herbert Wertheim College of Medicine. She is interested the prevention and treatment of infectious diseases in both developing and developed countries. In particular, her clinical and research interests revolve around vertically transmitted diseases that affect maternal and child health alike.

Nicholas Carcioppolo is an assistant professor in the School of Communication at the University of Miami. His research focuses on persuasion and health communication largely in the context of cancer prevention and screening. He is particularly interested in understanding the cognitive mechanisms through which exposure to persuasive health communication messages can affect health outcomes. Much of his cancer-related research involves reducing excessive UV exposure (via outdoor or indoor tanning), encouraging HPV (human papillomavirus) vaccination, and increasing compliance with colorectal health recommendations.

References

Arkin

E. B.

(1990). Opportunities for improving the nation’s health through collaboration with the mass media. Public Health Reports, 105, 219-223.

Bandura

(1997). Self-efficacy: The exercise of control. New York, NY: Freeman.

Black

D. R.

Dietz

J. E.

Stirratt

A. A.

Coster

D. C.

(2015). Do social media have a place in public health emergency response? Journal of Emergency Management (Weston, Mass.), 13, 217-226.

Bucher

H. J.

(2002). Crisis communication and the Internet: Risk and trust in a global media. First Monday, 7(4). doi:10.5210/fm.v7i4.943

Cabling

M. L.

Turner

J. W.

Hurtado-de-Mendoza

Zhang

Jiang

Drago

Sheppard

V. B.

(2017). Sentiment analysis of an online breast cancer support group: Communicating about Tamoxifen. Health Communication, 33, 1158-1165.

Cappella

J. N.

Kim

H. S.

Albarracín

(2015). Selection and transmission processes for information in the emerging media environment: Psychological motives and message characteristics. Media Psychology, 18, 396-424.

Centola

(2010). The spread of behavior in an online social network experiment. Science, 329(5996), 1194-1197.

Chan

M. P. S.

Farhadloo

Winneg

Jamieson

K. H.

Albarracin

(2018). Sources affecting knowledge and behavior responses to the Zika virus in US households with current pregnancy, intended pregnancy and a high probability of unintended pregnancy. Journal of Public Health, 40, 776-786.

Chan

M. P. S.

Winneg

Hawkins

Farhadloo

Jamieson

K. H.

Albarracín

(2018). Legacy and social media respectively influence risk perceptions and protective behaviors during emerging health threats: A multi-wave analysis of communications on Zika virus cases. Social Science & Medicine, 212, 50-59.

10.

Chang

(2012). News coverage of health-related issues and its impacts on perceptions: Taiwan as an example. Health Communication, 27, 111-123.

11.

Cho

Salmon

C. T.

(2006). Fear appeals for individuals in different stages of change: Intended and unintended effects and implications on public health campaigns. Health Communication, 20, 91-99.

12.

Chung

C. J.

Nam

Stefanone

M. A.

(2012). Exploring online news credibility: The relative influence of traditional and technological factors. Journal of Computer-Mediated Communication, 17, 171-186.

13.

Dalrymple

K. E.

Young

Tully

(2016). “Facts, not fear” negotiating uncertainty on social media during the 2014 Ebola crisis. Science Communication, 38, 442-467.

14.

Dunlop

Wakefield

Kashima

(2008). Can you feel it? Negative emotion, risk, and narrative in health communication. Media Psychology, 11, 52-75.

15.

Farhadloo

Winneg

Chan

M. P. S.

Jamieson

K. H.

Albarracin

(2018). Associations of topics of discussion on twitter with survey measures of attitudes, knowledge, and behaviors related to Zika: Probabilistic study in the United States. JMIR Public Health and Surveillance, 4(1). doi:10.2196/publichealth.8186

16.

Fauci

A. S.

Touchette

N. A.

Folkers

G. K.

(2005). Emerging infectious diseases: A 10-year perspective from the National Institute of Allergy and Infectious Diseases. Emerging Infectious Diseases, 11, 519-525. doi:10.3201/eid1104.041167

17.

Feinberg

Pereira

J. A.

Quach

Kwong

J. C.

Crowcroft

N. S.

Wilson

S. E.

, . . . Public Health Agency of Canada/Canadian Institutes of Health Research Influenza Research Network (PCIRN) Program Delivery and Evaluation Group. (2015). Understanding public perceptions of the HPV vaccination based on online comments to Canadian news articles. PloS One, 10(6), e0129587.

18.

Finucane

M. L.

Alhakami

Slovic

Johnson

S. M.

(2000). The affect heuristic in judgments of risks and benefits. Journal of Behavioral Decision Making, 13, 1-17.

19.

Focosi

Maggi

Pistello

(2016). Zika virus: Implications for public health. Clinical Infectious Diseases, 63, 227-233.

20.

Freberg

(2012). Intention to comply with crisis messages communicated via social media. Public Relations Review, 38, 416-421.

21.

Fung

T. K.

Namkoong

Brossard

(2011). Media, social proximity, and risk: A comparative analysis of newspaper coverage of avian flu in Hong Kong and in the United States. Journal of Health Communication, 16, 889-907.

22.

Glik

D. C.

(2007). Risk communication for public health emergencies. Annual Review of Public Health, 28, 33-54.

23.

Gorney

(1992). Numbers versus pictures: Did network television sensationalize Chernobyl coverage? Journalism Quarterly, 69, 455-465.

24.

Gottfried

Shearer

(2016, May 26). News use across social media platforms 2016. Pew Research Center. Retrieved from http://www.journalism.org/2016/05/26/news-use-across-social-media-platforms-2016/

25.

Grabe

M. E.

Zhou

Barnett

(2001). Explicating sensationalism in television news: Content and the bells and whistles of form. Journal of Broadcasting & Electronic Media, 45, 635-655.

26.

Guttman

(2011). Ethics in communication for health promotion in clinical settings and campaigns: New challenges and enduring dilemmas. In Thompson

T. L.

Parrott

Nussbaum

J. F.

(Eds.), The Routledge handbook of health communication (2nd ed., pp. 632-646). New York, NY: Routledge.

27.

Hartmann

Apaolaza

D’Souza

Barrutia

J. M.

Echebarria

(2014). Environmental threat appeals in green advertising: The role of fear arousal and coping efficacy. International Journal of Advertising, 33, 741-765.

28.

Hastings

Stead

Webb

(2004). Fear appeals in social marketing: Strategic and ethical reasons for concern. Psychology & Marketing, 21, 961-986.

29.

Hayes

A. F.

Krippendorff

(2007). Answering the call for a standard reliability measure for coding data. Communication Methods and Measures, 1, 77-89.

30.

Hermida

Fletcher

Korell

Logan

(2012). Share, like, recommend: Decoding the social media news consumer. Journalism Studies, 13, 815-824.

31.

Holmes

B. J.

(2008). Communicating about emerging infectious disease: The importance of research. Health, Risk & Society, 10, 349-360.

32.

Hossain

Kam

Kong

Wigand

R. T.

Bossomaier

(2016). Social media in Ebola outbreak. Epidemiology & Infection, 144, 2136-2143.

33.

Ihekweazu

(2017). Ebola in prime time: A content analysis of sensationalism and efficacy information in U.S. nightly news coverage of the Ebola outbreaks. Health Communication, 32, 741-748.

34.

Janis

I. L.

(1967). Effects of fear arousal on attitude change: Recent developments in theory and experimental research. Advances in Experimental Social Psychology, 3, 166-224.

35.

Kasperson

J. X.

Kasperson

R. E.

Pidgeon

Slovic

(2003). The social amplification of risk: Fifteen years of research and theory. In Pidgeon

Kasperson

R. E.

Slovic

(Eds.), The social amplification of risk (pp. 13-46). Cambridge, England: Cambridge University Press.

36.

Kasperson

R. E.

Renn

Slovic

Brown

H. S.

Emel

Goble

. . . Ratick

(1988). The social amplification of risk: A conceptual framework. Risk Analysis, 8, 177-187.

37.

Lang

(2000). The limited capacity model of mediated message processing. Journal of Communication, 50, 46-70.

38.

Leask

Hooker

King

(2010). Media coverage of health issues and how to work more effectively with journalists: A qualitative study. BMC Public Health, 10(1), 535. doi:10.1186/1471-2458-10-535

39.

Lehmann

Lalmas

Yom-Tov

Dupret

(2012). Models of user engagement. In Proceedings from International Conference on User Modeling, Adaptation, and Personalization (pp. 164-175). Berlin, Germany: Springer.

40.

Lerner

J. S.

Gonzalez

R. M.

Small

D. A.

Fischhoff

(2003). Effects of fear and anger on perceived risks of terrorism: A national field experiment. Psychological Science, 14, 144-150.

41.

Lerner

J. S.

Keltner

(2000). Beyond valence: Toward a model of emotion-specific influences on judgement and choice. Cognition and Emotion, 14, 473-493.

42.

Liu

B. F.

Fraustino

J. D.

Jin

(2015). Social media use during disasters: How information form and source influence intended behavioral responses. Communication Research, 43, 626-646.

43.

Loewenstein

G. F.

Weber

E. U.

Hsee

C. K.

Welch

(2001). Risk as feelings. Psychological Bulletin, 127, 267-286.

44.

Luminet

Bouts

Delie

Manstead

A. S.

Rimé

(2000). Social sharing of emotion following exposure to a negatively valenced situation. Cognition & Emotion, 14, 661-688.

45.

McGough

S. F.

Brownstein

J. S.

Hawkins

J. B.

Santillana

(2017). Forecasting Zika incidence in the 2016 Latin America outbreak combining traditional disease surveillance with search, social media, and news report data. PLoS Neglected Tropical Diseases, 11(1), e0005295.

46.

Metzger

M. J.

Flanagin

A. J.

(2013). Credibility and trust of information in online environments: The use of cognitive heuristics. Journal of Pragmatics, 59, 210-220.

47.

Metzger

M. J.

Flanagin

A. J.

Eyal

Lemus

D. R.

McCann

R. M.

(2003). Credibility for the 21st century: Integrating perspectives on source, message, and media credibility in the contemporary media environment. Annals of the International Communication Association, 27, 293-335.

48.

Morens

D. M.

Fauci

A. S.

(2013). Emerging infectious diseases: Threats to human health and global stability. PLoS Pathogens, 9(7), e1003467. doi:10.1371/journal.ppat.1003467

49.

Morse

S. S.

(2009). Emerging infections: Condemned to repeat? In Mack

Choffnes

E. R.

Hamburg

M. A.

Relman

D. A.

(Eds.), Microbial evolution and co-adaptation: A tribute to the life and scientific legacies of Joshua Lederberg: Workshop summary (pp. 195-208). Washington, DC: National Academies Press.

50.

Muthusamy

Levine

T. R.

Weber

(2009). Scaring the already scared: Some problems with HIV/AIDS fear appeals in Namibia. Journal of Communication, 59, 317-344.

51.

Nii-Trebi

N. I.

(2017). Emerging and neglected infectious diseases: Insights, advances, and challenges. BioMed Research International. doi:10.1155/2017/5245021

52.

Paek

H. J.

S. H.

Hove

(2016). How fear-arousing news messages affect risk perceptions and intention to talk about risk. Health Communication, 31, 1051-1062

53.

Pan American Health Organization/World Health Organization. (2016a). Zika Cumulative Cases—29 December 2016. Retrieved from https://www.paho.org/hq/index.php?option=com_content&view=article&id=12390:zika-cumulative-cases&Itemid=42090&lang=en

54.

Pan American Health Organization/World Health Organization. (2016b). Zika Epidemiological Update—29 December 2016. Retrieved from https://www.paho.org/hq/index.php?option=com_content&view=article&id=11599:regional-zika-epidemiological-update-americas&Itemid=41691&lang=en

55.

Petersen

Wilson

M. E.

Touch

McCloskey

Mwaba

Bates

. . . Azhar

E. I.

(2016). Rapid spread of Zika virus in the Americas: Implications for public health preparedness for mass gatherings at the 2016 Brazil Olympic Games. International Journal of Infectious Diseases, 44, 11-15.

56.

Pew Research Center. (2012, February 24). Why Facebook users get more than they give. Retrieved from http://www.pewinternet.org/2012/02/03/part-2-facebook-activity/

57.

Pfefferbaum

Newman

Nelson

Nitiema

Pfefferbaum

Rahman

(2014). Disaster media coverage and psychological outcomes: Descriptive findings in the extant research. Current Psychiatry Reports, 16(9), 464. doi:10.1007/s11920-014-0464-x

58.

Rhodes

(2017). Fear-appeal messages: Message processing and affective attitudes. Communication Research, 44, 952-975.

59.

Rimal

R. N.

(2001). Perceived risk and self-efficacy as motivators: Understanding individuals’ long-term use of health information. Journal of Communication, 51, 633-654.

60.

Rimé

(1995). The social sharing of emotion as a source for the social knowledge of emotion. In Everyday conceptions of emotion (pp. 475-489). Dordrecht, Netherlands: Springer.

61.

Rodriguez-Morales

A. J.

Castañeda-Hernández

D. M.

McGregor

(2015). What makes people talk about Ebola on social media? A retrospective analysis of Twitter use. Travel Medicine and Infectious Disease, 13, 100-101.

62.

Rundmo

(2002). Associations between affect and risk perception. Journal of Risk Research, 5, 119-135.

63.

Rus

H. M.

Cameron

L. D.

(2016). Health communication in social media: Message features predicting user engagement on diabetes-related Facebook pages. Annals of Behavioral Medicine, 50, 678-689.

64.

Schultz

Utz

Göritz

(2011). Is the medium the message? Perceptions of and reactions to crisis communication via twitter, blogs and traditional media. Public Relations Review, 37, 20-27.

65.

Slattery

(1994). Sensationalism versus news of the moral life: Making the distinction. Journal of Mass Media Ethics, 9, 5-15.

66.

Slovic

(1987). Perception of risk. Science, 36, 280-285.

67.

Slovic

Finucane

M. L.

Peters

MacGregor

D. G.

(2005). Affect, risk, and decision making. Health Psychology, 24(4S), S35-S40.

68.

Smith

R. D.

(2006). Responding to global infectious disease outbreaks: Lessons from SARS on the role of risk perception, communication and management. Social Science & Medicine, 63, 3113-3123.

69.

Smith

W. A.

(2000). Ethics and the social marketer: A framework for practitioners. In Andreasen

A. R.

(Ed.), Ethics in social marketing (pp. 632-646). Washington, DC: Georgetown University Press.

70.

Southwell

B. G.

Yzer

M. C.

(2007). The roles of interpersonal communication in mass media campaigns. Annals of the International Communication Association, 31, 420-462.

71.

Statista. (n.d.). Most popular social networks worldwide as of January 2019, ranked by number of active users (in millions). Retrieved from https://www.statista.com/statistics/272014/global-social-networks-ranked-by-number-of-users/

72.

Strekalova

Y. A.

(2017). Health risk information engagement and amplification on social media: News about an emerging pandemic on Facebook. Health Education & Behavior, 44, 332-339.

73.

Strekalova

Y. A.

Krieger

J. L.

(2017). Beyond words: Amplification of cancer risk communication on social media. Journal of Health Communication, 22, 849-857.

74.

Thackeray

Neiger

B. L.

Hanson

C. L.

McKenzie

J. F.

(2008). Enhancing promotional strategies within social marketing programs: Use of Web 2.0 social media. Health Promotion Practice, 9, 338-343.

75.

Van Doorn

Lemon

K. N.

Mittal

Nass

Pick

Pirner

Verhoef

P. C

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

76.

Vasterman

Yzermans

C. J.

Dirkzwager

(2005). The role of the media and media hypes in the aftermath of disasters. Epidemiologic Reviews, 27, 107-114.

77.

Vasterman

Ruigrok

(2013). Pandemic alarm in the Dutch media: Media coverage of the 2009 influenza A (H1N1) pandemic and the role of the expert sources. European Journal of Communication, 28, 436-453.

78.

Veil

S. R.

Buehner

Palenchar

M. J.

(2011). A work-in-process literature review: Incorporating social media in risk and crisis communication. Journal of Contingencies and Crisis Management, 19, 110-122.

79.

Viel

Reynolds

Sellnow

T. L.

Seeger

M. W.

(2008). CERC as a theoretical framework for research and practice. Health Promotion Practice, 9(4 Suppl), 26S-34S.

80.

Vos

S. C.

Buckner

M. M.

(2016). Social media messages in an emerging health crisis: Tweeting bird flu. Journal of Health Communication, 21, 301-308.

81.

Witte

(1992). Putting the fear back into fear appeals: The extended parallel process model. Communications Monographs, 59, 329-349.

82.

Witte

(1994). Fear control and danger control: A test of the extended parallel process model (EPPM). Communications Monographs, 61, 113-134.

83.

Witte

Allen

(2000). A meta-analysis of fear appeals: Implications for effective public health campaigns. Health Education & Behavior, 27, 591-615.