ppt

Jangetal2019.pdf

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=hhth20

Health Communication

ISSN: 1041-0236 (Print) 1532-7027 (Online) Journal homepage: https://www.tandfonline.com/loi/hhth20

From Social Media to Mainstream News: The Information Flow of the Vaccine-Autism Controversy in the US, Canada, and the UK

S. Mo Jang, Brooke W. Mckeever, Robert Mckeever & Joon Kyoung Kim

To cite this article: S. Mo Jang, Brooke W. Mckeever, Robert Mckeever & Joon Kyoung Kim (2019) From Social Media to Mainstream News: The Information Flow of the Vaccine-Autism Controversy in the US, Canada, and the UK , Health Communication, 34:1, 110-117, DOI: 10.1080/10410236.2017.1384433

To link to this article: https://doi.org/10.1080/10410236.2017.1384433

Published online: 13 Oct 2017.

Submit your article to this journal

Article views: 932

View Crossmark data

Citing articles: 3 View citing articles

From Social Media to Mainstream News: The Information Flow of the Vaccine-Autism Controversy in the US, Canada, and the UK S. Mo Jang a, Brooke W. Mckeevera, Robert Mckeevera, and Joon Kyoung Kima

aSchool of Journalism and Mass Communications, University of South Carolina

ABSTRACT Despite increasing warnings about inaccurate information online, little is known about how social media contribute to the widespread diffusion of unverified health information. This study addresses this issue by examining the vaccine-autism controversy. By looking into a large dataset of Twitter, Reddit posts, and online news over 20 months in the US, Canada, and the UK, our time-series analysis shows that Twitter drives news agendas, and Reddit follows news agendas regarding the vaccine-autism debate. Additionally, the results show that both Twitter and Reddit are more likely to discuss the vaccine-autism link compared to online news content.

Social media have become an increasingly important venue for disseminating and constructing health and science knowledge (Liu, Lu, &Wang, in press). However, health communication on social media come with caveats, one of which involves the spread of inaccurate information (Tambuscio, Ruffo, Flammini, & Menczer, 2015). Health professionals and scholars have warned that the Internet, particularly, social media may boost the circu- lation of hearsay and misleading medical and science informa- tion (Wilson & Keelan, 2013). This paper addresses this issue using the vaccine-autism controversy.

Concerns about the risks of vaccines have grown in recent years among the public in the US and other western countries (McKee & Bohannon, 2016). Based on an unsubstantiated claim that vaccines cause autism, some parents delay or entirely refuse vaccination of their children against the recom- mendation of medical experts (Jolley & Douglas, 2014). Research has also found that vaccination-hesitant parents tend to rely on online information more so than parents who have their children vaccinated (Kata, 2012). The rising suspicion of this vaccine-autism link among parents poses a significant challenge to public health systems.

Despite increasing concerns about unverified health infor- mation online, little research has been done about how mis- leading health information spreads. Are social media the rumor mill of misleading health knowledge? Does informa- tion about the vaccine and autism link flow from mainstream news to social media or is it the other way around? How much does the discussion of the vaccine and autism link dominate news and social media? To answer these questions, this study looks at social media and online news content over 20 months in the US, Canada, and the UK. First, this study investigates the flow of information about the vaccine-autism controversy between social media (Twitter and Reddit

content) and mainstream online news. Second, we compare social media and online news media in terms of the degree to which media pay attention to the vaccine-autism controversy. Finally, we examine different patterns shown in the content coming from the US, Canada, and the UK.

Information flow of health and science communication

Health and science communication has been traditionally unidirectional. Health and science news outlets pick up research results and translate them into knowledge, which is essential for the public understanding of science and policy- decision processes (Brossard, 2013; Davies & Horst, 2016). Hilgartner (1990) noted that this popularization process often oversimplifies and distorts scientific knowledge, but that it also provides a useful resource for public discourse. As few people have first-hand connections to or understand- ing of complex scientific findings, mass media have func- tioned as important conduits in spreading scientific information (Schafer, 2012).

The flow of health and science communication has been diversified alongside the emerging digital media environment. As of 2016, the Internet became Americans’ primary source of information about science (National Science Board, 2016). Realizing the potential of social media, many scientists are now using social media to reach the broader public and boost public visibility (Allgaier, Dunwoody, Brossard, Lo, & Peters, 2013). Scientists are utilizing crowdfunding strategies via social media to improve funding opportunities as well (Koh, Dunwoody, Brossard, & Allgaier, 2016).

Importantly, the participatory nature of social media enables not only scientists and journalists to disseminate science and

CONTACT S. Mo Jang [email protected] University of South Carolina, 800 Sumber St., Columbia, SC 2901, USA Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/hhth.

HEALTH COMMUNICATION 2019, VOL. 34, NO. 1, 110–117 https://doi.org/10.1080/10410236.2017.1384433

medical information, but it also enables ordinary people to participate in knowledge co-construction online (Hara & Sanfilippo, 2016). Lay audiences express their opinions, share their experiences, and seek advice from peers, experts and pseudo-experts. This user-generated information broadens the scope and diversity of science knowledge as well as provides useful information to science and medical experts. For exam- ple, scientists monitor Google query data to detect influenza epidemics (Ginsberg et al., 2009). Moreover, public sentiment on social media about controversial science issues may influ- ence policy-making decisions (O’Connor, Balasubramanyan, Routledge, & Smith, 2010). Indeed, digital media afford both top-down and bottom-up flow of science knowledge between experts and lay audiences.

Research on intermedia agenda setting offers a useful the- oretical framework on the flow of information between tradi- tional and social media (Neuman, Guggenheim, Jang, & Bae, 2014). Compared to research on agenda setting that addresses how the media agenda influences the public agenda, interme- dia agenda setting considers the agenda transfer among var- ious media outlets. In particular, social media scholars have documented the agenda-setting effects of social media on news media in numerous contexts. For example, research has shown how fund-raising campaigns, such as the ice bucket challenge, went viral and originated from social media and subsequently turned into major news topics (Jang, Park, & Lee, in press). On the other hand, Conway and colleagues (Conway, Kenski, & Wang, 2015) found that the intermedia process was symbiotic between Twitter feeds of presidential candidates and newspaper reports over the primary election campaign. Overall, the literature shows the possibility of bot- tom-up intermedia agenda setting in certain contexts such as viral campaigns and political dialogues. Yet, little is known about intermedia agenda setting in science and health con- texts, where information is often considered complex, and, thus, the general public needs the mediating role of science journalists and health professionals.

The vaccine-autism myth

Suspicions about a link between vaccines and autism began to proliferate following an article that appeared in February 1998 in the British medical journal, The Lancet (Wakefield et al., 1998). Wakefield and his colleagues suggested there was an association between autism and the measles, mumps, and rubella (MMR) vaccine. Although Wakefield’s research was met with initial skepticism among scientific and medical pro- fessionals, it attracted considerable attention, and news of the supposed connection between autism and the MMR vaccine traveled quickly and alarmed many parents throughout the world. Since that time, numerous medical professionals and scholars have refuted this link, and there is considerable evi- dence that there is no connection between vaccines and autism (see e.g., Gerber & Offit, 2009; Institute of Medicine, 2004). Additionally, the GMC revoked Wakefield’s medical license, and The Lancet retracted the original article about the alleged link between autism and the MMR vaccine (Whalen, 2010).

Despite these events, however, public opinion and health- related behaviors among parents seemed to be aggravated since

the news coverage of the Wakefield article. Vaccine rates declined, both in the US and the UK (Smith, Ellenberg, Bell & Rubin, 2008; Burgess, Burgess, & Leask, 2006), and diseases like measles that were once eliminated in the US have been on the rise again in recent years. For instance, although there were fewer than 800 cases of measles reported in the US between 2000 and 2010, 667 cases of the disease were reported in 2014 alone (Centers for Disease Control and Prevention, 2016).

The myth and media

Numerous communication scholars have studied the vaccine- autism controversy, including media coverage of the topic and possible communication strategies that might mitigate the harmful effects of false beliefs about the vaccine and autism link. For example, research has shown that news coverage of autism increased over time (from the late 1990’s through early 2000’s) in accordance with the increasing frequency of autism diagnoses and the controversy surrounding Wakefield and his colleagues’ claims about a purported link (McKeever, 2012). However, the majority of news stories seemed to accurately portray Wakefield as being at least partially culpable for the controversy and its subsequent effects (Holton, Weberling, Clarke, & Smith, 2012). Several studies on news coverage of the issue have suggested that reporters trying to convey jour- nalistic balance may have inadvertently painted an inaccurate picture of the scientific evidence that shows there is no link between vaccines and autism (Clarke, 2008; Lewis & Speers, 2003). Some scholars have investigated this notion of “false balance,” suggesting that individuals’ false beliefs about the issue might be mitigated by media coverage that includes explicit statements about the weight of evidence supporting the fact that there is no link between vaccines and autism (Dixon, McKeever, Holton, Clarke, & Eosco, 2015).

Other recent research has looked at digital and social media coverage of this issue. More and more people are turning to online news, websites, blogs, and social media networks for health information and to gauge public opinion regarding divisive health topics such as the vaccine-autism controversy (Brunson, 2013; Goldstein, MacDonald, & Guirguis, 2015). Kata (2012) examined anti-vaccine content online and described the rhetoric and tactics used to support anti-vaccine arguments, including skewing science, shifting hypotheses, censoring dissent, and attacking critics. As one recent survey (N = 455) found, mothers who do not support childhood vaccinations were more likely to engage in online communication regarding the issue, including seeking, shar- ing and forwarding information on social media (McKeever, McKeever, Holton, & Li, 2016). As the authors noted, if anti- vaccine information continues to spread via digital media– without being corrected or without equal amounts of pro- vaccine information also appearing in the same spaces–a false consensus may develop, and more individuals’ beliefs and health behaviors could be affected as a result.

Present study

This study examined, how user-generated media contribute to the widespread diffusion of misleading information

HEALTH COMMUNICATION 111

about vaccines and autism. Because we focused on contro- versial public issues, rather than interpersonal relationship management issues, we selected the most popular news- oriented social media outlets, Twitter and Reddit posts. While Twitter has attracted increasing social and political attention especially in the wake of political leaders’ fre- quent use of the microblog medium, Reddit posts have received limited attention so far. Unlike most prior studies that have focused exclusively on tweets, we included both tweets and Reddit posts in our analysis. Reddit is an important social venue due to its tremendous popularity of the website (ranked 7th in the US; Alexa, 2016) and its ability to diffuse news stories and prompt online discus- sions (Shelton, Lo, & Nardi, 2015). Although 4% of US adults were regular Reddit users in 2016, 70% of Reddit users reported that they learned about news mostly through Reddit, which indicates that Reddit is one of the more-news-oriented social spaces on the Internet (Pew Research Center, 2016).

Largely, we employed two types of tests to address the role of user-generated content (Twitter and Reddit) in spreading the vaccine-autism myth. First, we tested intermedia agenda- setting hypotheses, anticipating the possibility that social media first set the agenda of the vaccine controversy and then news media reflected and followed that agenda. Specifically, we examined the temporal order of news and user-generated media in the vaccine controversy context by testing the direction of information flow between multiple media outlets. Considering previous findings, which indicated that vaccination-hesitant parents tend to use online commu- nication for vaccine information (Kata, 2012; McKeever et al., 2016), we hypothesized the bottom-up direction of intermedia agenda setting.

H1: News coverage about the vaccine-autism myth comes after tweets about the topic.

H2: News coverage about the vaccine-autism myth comes after Reddit posts about the topic.

Next, we compared tweets, Reddit posts and online news in terms of the prevalence of media attention to the vaccine- autism controversy. We focused on the daily volume of media content and compared the relative popularity of the vaccine- autism controversy within any discussion about vaccines. The ratio value was calculated based on the daily volume of media content including the vaccine-autism link as a proportion of all media posts mentioning vaccines.

H3: Media attention to the vaccine-autism myth is more prevalent on social media than in news media.

Additionally, we examined H3 within each country to see if any different patterns emerged across the US, Canada, and the UK, seeking to answer the following question.

RQ1: Within the US, Canada, and the UK, what is the relative prevalence of the vaccine-autism controversy on Twitter, Reddit, and online news media?

Finally, this study further explored the potential difference between tweets in general and tweets without external links. The purpose of probing this distinction is to address potential

confounds in our interpretation of the data, as one could reasonably contend that it is possible that a substantial por- tion of tweets function as a tool for redistributing news stories, rather than an instrument for disseminating unique, user-generated ideas. Thus, we ran separate analyses with both tweets in general and tweets without external links, assuming that tweets without any links may better capture the nature of user-generated content more so than tweets in general. Thus, we asked the following research question.

RQ2: Is the prevalence of the vaccine-autism myth different between tweets in general and tweets without external links?

Method

Data

Data for this study were collected using the Crimson Hexagon’s ForSight platform. Crimson Hexagon is a private company providing electronic databases of media content including Twitter, blogs, and online news. We also retrieved Reddit posts from Crimson Hexagon, but these data were available only for the US. Metadata attached to the media content allowed us to filter the entire archive based on time, location, language, and web domain. Then we analyzed Twitter, online news articles, and Reddit posts for a 20-month period spanning from February 1, 2015 to September 30, 2016.

Spam tweets (e.g., Twitter bots) are impediments to obtain- ing valid, high-quality data. Crimson Hexagon employs four main methods to judge, whether posts are spam or actual content run by real people (see Neuman et al., 2014, p. 198 for details of these methods). As a result, approximately 40% of what Crimson Hexagon crawls is discarded as spam. We included retweets because literature suggests that they are an effective indicator for the extent to which messages are per- ceived to be important in the network (Larsson & Moe, 2012).

Media attention to vaccines

We first measured media attention to vaccines by retrieving postings and articles that mentioned vaccine-related terms in their content. Our close examinations of sampled texts ensured that the used Boolean search term, (vaccine OR vaccines OR MMR), constituted a representative search phrase to capture relevant media content about vaccines. The retrieval process included news stories and tweets from the US, Canada and the UK, and Reddit posts from the US.

During our time period, on a typical day, approximately 4,210 (SD = 4,313) tweets, 535 (SD = 513) Reddit posts, and 1,230 (SD = 777) online news stories mentioned vaccines in the US. In the U.K, approximately 720 (SD = 440) tweets and 65 (SD = 41) online news stories focused on vaccines. In Canada, approximately 332 (SD = 376) tweets and 59 (SD = 60) online news stories referred to vaccines.

Media attention to the vaccine-autism link

Based on the same search process, we captured media atten- tion to the vaccine-autism controversy. By running a Boolean

112 S. M. JANG ET AL.

search that paired (vaccine OR vaccines OR MMR) with autism, we compiled the number of daily tweets, Reddit posts, and online news articles that met this search parameter. These search terms generated approximately 362 (SD = 851) tweets, 29 (SD = 42) Reddit posts, and 44 (SD = 82) online news stories per day about the vaccine-autism controversy in the US. In the UK, approximately 40 (SD = 79) tweets and one (SD = 4) online news story discussed the vaccine-autism link in their content. In Canada, approximately 27 (SD = 64) tweets and five (SD = 16) news stories mentioned the link between vaccines and autism during this study’s time period.

Validation of search terms

Boolean search terms representing autism and vaccines appeared to have strong face validity in signifying attention to the prevalent myths. Using procedures illustrated by Stryker, Wray, Hornik, and Yanovitzky (2006), we evaluated our search keywords based on two criteria: Precision and Recall. The average precision and recall estimate was 95% and 92% respectively. Average interceder reliability was .81 (Cohen’s Kappa) and the overall agreement reached 98% between two coders.1

Covariates

Extant literature indicates that time-series data are subject to the product of a trend and cyclical forces (Becketti, 2013). Notably, research has identified periodic cycles in the volume of media content (Neuman et al., 2014). Thus, it is necessary to control for long-term trends and cyclical patterns that might influence content production in the online news media and on Twitter and other social websites. To account for these processes, we controlled for days of the week. Additionally, we controlled for long-term changes in the volume of media content during 20 months with both linear and quadratic variables for time.

Results

To explicate the intermedia agenda-setting process between the elite and social media and to test the study’s first and second hypotheses, we conducted a time-series analysis based on the framework of Granger causality (Granger, 1969). The Granger analysis investigates whether the current value of y can be better predicted by past values of x and y together than by past values of y alone. Granger causality is best used to determine the temporal order between two time-series data.

As a first step of Granger analysis, we tested each vector autoregression (VAR) for a stationary (Becketti, 2013). The stationary test identifies whether impulsive spikes, trends, cycles, and seasonal variations create unrecoverable deviations from the average. The results confirmed that all of our VARs passed the stationary test.

Next, we set up a proper lag length, or the amount of time we would lag the independent variables in our regressions. Prior research indicates that agenda-setting effects mostly happen in a week or less (Roberts, Wanta, & Dzwo, 2002). Roberts et al. (2002) indicated that a 7-day lag produced the

most effects for the relationship between news and online content. Neuman et al. (2014) also confirmed that statistical criteria, such as AIC and SBIC, indicate that 7 days indicate an optimal lag. According to this suggestion, we included 7- day lags. Note that using 7-day lags in Granger analysis does not mean that we only captured 7-day lagged effects. Rather, we included all time-lag variables ranging from 1-day to 7-day lags (from Xt-1 to Xt-7). Then, to control for weekly periodi- city, we included six dummies that represent days of the week in our Granger analyses.

As shown in Table 1, we ran 18 Granger analyses between online news and Twitter (and Reddit for the US) across the three geographic areas. Our study’s first prediction (H1) pro- posed that social media talked about the vaccine-autism link and then news media picked up the topic trend. As illustrated in the overall analysis, we observed this reverse intermedia agenda-setting process at the aggregate level. In particular, we found strong evidence of Granger causality from Twitter to online news coverage, but not the other way around in the US and Canada. Interestingly, however, when tweets with links were excluded, such intermedia influences disappeared. On the other hand, the UK showed a pattern of mutual interac- tion where both online news and Twitter Granger-caused each other. This unique pattern in the UK may be attributable to the fact that the vaccine-autism controversy was already an established topic and actively discussed among the public and news media. Both the UK public and professional journalists have closely followed the vaccine-autism controversy along with the rise and fall of the British scientist, Dr. Andrew Wakefield (Burgess et al., 2006; O’Neill, 2006). Moreover, prior research on intermedia agenda setting (Jang et al., in press; Neuman et al., 2014) indicated that popular public debates such as gun control, same-sex marriage, and polariza- tion showed mutual influence across different media plat- forms. Taken together, these findings offer partial support for H1, as the reverse intermedia agenda-setting process was observed in both the US and in Canada. Our second hypoth- esis (H2) posited that the elite news media focused on the vaccine-autism controversy after Reddit posts had been made about the topic. This prediction was not supported, as results from the analysis indicated that Reddit posts did not Granger- cause news media content. Contrary to our expectations, results from the analysis instead offered evidence of top- down intermedia agenda-setting effects, such that news media coverage preceded Reddit content.

Table 1. Granger analysis between online news and user-generated content.

Intermedia influence US Canada UK Overall

From To Chi

square Chi

square

News Twitter 9.868 5.313 41.862*** 9.253 Twitter News 24.797*** 25.816*** 16.384* 22.01*** News Twitter (no

link) 6.986 13.271 17.307* 6.754

Twitter (no link)

News 9.607 9.179 24.56*** 9.142

News Reddit 18.267* NA NA NA Reddit News 11.429 NA NA NA

Note. *p < .05., **p < .01. ***p < .001.

HEALTH COMMUNICATION 113

We next examined the relative prevalence of the vaccine- autism controversy in three media platforms, Twitter, reddit. com, and online news websites. We calculated the ratio of media content specifically about the vaccine and autism link out of media content about vaccines in general. Figure 1 depicts the proportion of social and mainstream news media content about the vaccine-autism link in the US, Canada, and the UK.

H3 predicted that the vaccine-autism controversy would be discussed more frequently in social media than in mainstream online news. To test this hypothesis, a series of t-tests (employing Bonferroni-corrected significance criteria) were performed to determine whether there were differences in the average daily volume of tweets (with and without links) and online news stories discussing the vaccine-autism link. Findings from our overall analysis indicated online news showed a lower ratio of the daily volume of content about the vaccine-autism link (M = 3.82, SD = 4.90) when compared to tweets in general (M = 7.35, SD = 5.21), t = 15.20, p < .001. This was also the case when online news was compared to tweets without links (M = 8.15, SD = 5.90), t = 16.10, p < .001. Taken together, these findings offer robust support for H3.

Next, we examined whether the daily volume of news and social media content about the vaccine-autism link varied across the US, UK, and Canada (RQ1). In the US, online news showed a lower ratio of the daily volume of the vaccine-autism link (M = 3.75, SD = 4.88) compared to tweets in general (M = 7.78, SD = 5.30, t = 16.79, p < .001) and tweets without links (M = 8.87, SD = 6.13, t = 18.02, p < .001). Analysis of social and elite media content from the UK revealed a similar pattern in the prevalence of discussions related to the vaccine-autism link. Within the UK media, online news was found to contain less discussion of the vaccine-autism controversy (M = 2.01, SD = 5.26) when com- pared to both tweets in general (M = 5.42, SD = 5.40, t = 13.26, p < .001) and those without links (M = 5.28, SD = 5.78, t = 12.19, p < .001). Intriguingly, analysis of social and elite media in Canada revealed that the prevalence of the controversy in online news (M = 6.99, SD = 12.24) did not differ from the controversy discussion in Twitter content (M = 6.98, SD = 6.17, t = 0.021, p = .98) and tweets without links (M = 7.55, SD = 7.26, t = 1.06, p = .29) at a level that was statistically significant.

Recall that RQ2 asked whether tweets in general and tweets without links showed any different patterns. Our results showed that the vaccine-autism controversy tended to be more popular among tweets without links than among tweets in general. The US (t = 4.76, p < .001) and Canada (t = 2.18, p < .05) followed this pattern, although the UK showed no significant difference between tweets in general and tweets without links (t = 0.63, p = .53). The results largely supported our findings that the vaccine controversy was more popular on social media where users can generate their own content, which may or may not be supported by news media stories provided through links.

Discussion

Health and science communication has previously been thought of as unidirectional from elites to the public. The bottom-up direction of information flow appeared to be impossible not only because scientific knowledge was too complex for the public to produce and disseminate, but also because media channels available for user-generated informa- tion were limited. However, with the rise of social media, technical constraints have been eliminated, and it becomes questionable whether health and science communication also flows upward from the public. Although ample evidence sug- gests that social media conversations may influence main- stream media agendas in political contexts (Conway et al., 2015; Neuman et al., 2014), little research has been done in health and science contexts. This study therefore explored this issue by tracking the information flow in the context of the vaccine-autism link, one of the most controversial science issues of late. The findings provided initial evidence of reverse intermedia agenda setting in science communication.

Theoretical implications

This study contributes to the existing body of literature on mass communication theory and media processes in several important ways. In some aspects, the findings related to reverse intermedia agenda setting in the current study both

US Canada UK Overall

P ro

po rti

on o

f C on

te nt

a bo

ut V

A L

in k

(% )

Online News

Twitter

Twitter No Link

Reddit.com

Figure 1. Proportion of media content about vaccine and autism link by countries. Note. Reddit content was only retrievable from U.S. sources.

114 S. M. JANG ET AL.

replicated and expanded upon results from Neuman and colleagues’ (2014) previous research examining the influence of topics discussed on Twitter in driving the agenda of elite media. As has been found in past studies examining inter- media agenda setting in contexts such as the ice-bucket chal- lenge (Jang et al., in press) and the dissemination of political news during the presidential primary (Conway et al., 2015), Granger causality analysis showed that tweets about the now- discredited link between the MMR vaccine and autism led the online news media agenda in covering this topic. The reverse direction of information is somewhat surprising in that health and science communication has been considered a translation process by science journalists and health professionals whose role is to make complex scientific results accessible to the public (Brossard, 2013). The current findings indicate that social media users can generate and share health and science information, eventually influencing the mainstream news agenda.

However, contrary to similar expectations derived from the aforementioned literature, the current study also found the top-down flow of intermedia agenda setting where main- stream media agendas led Reddit content. While this finding contradicts our predictions, it is consistent with findings in related research (e.g., Groshek & Groshek, 2013; Weeks & Southwell, 2010), which suggests that despite the proliferation of new media, in many cases elite media still maintain the dominant power of agenda setting, and that the direction of intermedia agenda setting may vary depending on the topic (Neuman et al., 2014). In addition to these possible interpre- tations of the top-down agenda setting effect observed in Reddit content, this discrepant finding may relate to the fact that tweets and Reddit posts are markedly different in nature. In the case of the latter, information found on Reddit gen- erally consists of popular links shared and voted on by users rather than newly created content by users, as is typical of tweets. Reddit employs a user-participatory voting system where users indicate their preferences by voting on existing comments rather than adding their own comments (Mills, 2011). Thus, considering Reddit content reflects the popular- ity of current issues, it makes sense that the intermedia influ- ence flows from news to Reddit rather than the other way around.

The current findings also highlight the recent advances in agenda-setting theory. First, the observed prevalence of the vaccine-autism link on Twitter compared to the news media shed light on the community structure approach (Funk & McCombs, 2017). Community structure research posits that the structure of a community influences the selection of local news topics. For example, immigrant issues tend to receive increasing media attention from local newspapers in an area with a large population of immigrants compared to and area with a small population of immigrants. This approach pro- vides a useful explanation of relative prevalence of the vac- cine-autism controversy on Twitter and Reddit. For example, it may be difficult for news organizations to pay continued attention to the vaccine-autism controversy unless novel scientific evidence emerges. However, on Twitter or Reddit, users and communities (e.g., parents with autistic children) are willing to discuss the controversy, perceiving that their

imagined message recipients share similar interests and ideol- ogy (Himelboim, McCreery, & Smith, 2013). The current results illustrate the need for future research to focus on the role of the structure of digital audiences on user-generated agendas.

Second, our findings support the notion of second-level (attribute) agenda setting (McCombs & Stroud, 2014) and further hinted at the possibility of second-level intermedia agenda setting. While most previous studies have focused on intermedia agenda setting at the topic level (i.e., vaccine), the current findings provide evidence of salience transfer at the attribute level (i.e., vaccine-autism link). In other words, the ease with which autism risks are retrievable in the discussions of vaccines was transmitted across different media. However, as the current research considered only one aspect of vaccines, it is difficult to explicate the integrated picture of multiple attributes surrounding vaccines. Thus, future efforts should adopt a network approach (e.g., network agenda setting model or third level agenda setting; Vargo, Guo, McCombs, & Shaw, 2014).

Practical implications

The results also have practical implications for media practi- tioners and health and science communicators seeking to dispel unverified information spread through social media. Specifically, according to these findings, in both the US and the UK, the vaccine-autism controversy was more commonly discussed on social media platforms, which ranked among the top resources used by parents in the vaccination decision- making process (Brunson, 2013; Goldstein et al., 2015). This finding is particularly relevant to health communication prac- titioners and vaccination advocates, because previous research has found that mothers who generally supported childhood vaccinations were less likely to communicate about vaccines on social platforms (McKeever et al., 2016). If that is so, then considering the reversed intermedia agenda-setting effect observed in the current study, it is possible that those who oppose childhood vaccinations are the ones dominating the discourse surrounding the topic on social media, which may– in turn–perpetuate the presence of the link between vaccines and autism in elite news coverage. These conversations focus- ing on the unverified vaccine-autism link could eventually have downstream effects on health beliefs and behaviors.

Misleading information provided via social media is pro- blematic because, among other reasons, it can be difficult to ensure that the correct information spreads as far and wide as the original misinformation (Southwell & Thorson, 2015). The novel challenge online is that popular information may artificially inflate support for unverified claims. This false consensus leads to inaccurate representations of public opi- nion, social confusion, and opinion polarization. There is an urgent need for science and health communication profes- sionals to develop computerized mechanisms that monitor social media trends and provide fact-checking capabilities.

Although this study displayed the dark side of the inter- media agenda-setting process, the findings also indicate the potential of social media for promoting science-based infor- mation. Given that the current work found strong support for

HEALTH COMMUNICATION 115

the temporal order in which social media leads the elite media agenda surrounding the vaccine-autism controversy, it is likely that social platforms may provide a lucrative venue for future advocacy efforts in this area. In view of the robust reverse intermedia agenda-setting effect found in the current work, practitioners should consider the effectiveness of pro- moting science-based information about the safety of vaccines through social media along with traditional top-down approaches using elite media.

Limitations and future directions

Although our keyword selection was relatively straightforward and went through rigorous precision and recall tests, several limitations should be considered. First, keyword analysis does not consider different cultures and norms that prevail in social media dialogues across the three English-speaking regions included in this study, so we should be cautious when making direct comparisons. Second, ourmeasure ofmedia attention to the vaccine-autism link includedmessages both supporting and deny- ing the link. Although we contend that unnecessary media atten- tion to the false claimsmay be damaging to society, the potential of self-correcting capabilities on social media should receive further scholarly attention. Recent research shows that social media users are sometimes able to debunk or inactivate inaccurate information through the crowdsourcing process (Bode&Vraga, 2015; Zubiaga, Liakata, Procter, Hoi, & Tolmie, 2016). Future studies should also examine the effects of fact-checking or misinformation-alert sys- tems on the perceptions of the public. Finally, because the two types of user-generatedmedia we studied (Twitter and Reddit) are categorized as information-oriented social media, it remains unclear how the intermedia influence emerges between relation- ship-oriented user-generated media (e.g., Instagram, Facebook) and newsmedia (Kim& Lee, 2016). The salience transfer between information-oriented media may be more common compared to the transfer amongmedia with different motivations. Future work should examine this distinction when studying the intermedia agenda setting involving social media.

Notes

1 The detailed information about this validation procecess is avail- able upon request from the corresponding author.

Disclosure of potential conflicts of interest

No potential conflict of interest was reported by the authors.

ORCID

S. Mo Jang http://orcid.org/0000-0003-3935-7421

References

Alexa. (2016). Top sites in the United States. Retrieved from http://www. alexa.com/topsites/countries/US

Allgaier, J., Dunwoody, S., Brossard, D., Lo, -Y.-Y., & Peters, H. P. (2013). Journalism and social media as means of observing the con- texts of science. BioScience, 63, 284–287.

Becketti, S. (2013). Introduction to time series using Stata. College Station, TX: Stata Press.

Bode, L., & Vraga, E. K. (2015). In related news, that was wrong: The correction of misinformation through related stories functionality in social media. Journal of Communication, 65, 619–638.

Brossard, D. (2013). New media landscapes and the science information consumer. Proceedings of the National Academy of Sciences, 110, 14096–14101.

Brunson, E. K. (2013). The impact of social networks on parents’ vacci- nation decisions. Pediatrics, 131, e1397–1404.

Burgess, D. C., Burgess, M. A., & Leask, J. (2006). TheMMR vaccination and autism controversy in United Kingdom 1998–2005: Inevitable commu- nity outrage or a failure of risk communication? Vaccine, 24, 3921–3928.

Centers for Disease Control and Prevention. (2016). Measles cases and outbreaks. Retrieved from http://www.cdc.gov/measles/cases-out breaks.html

Clarke, C. (2008). A question of balance: The autism-vaccine controversy in the British and American elite press. Science Communication, 30, 77–107.

Conway, B. A., Kenski, K., & Wang, D. (2015). The rise of Twitter in the political campaign: Searching for intermedia agenda-setting effects in the presidential primary. Journal of Computer-Mediated Communication, 20, 363–380.

Davies, S. R., & Horst, M. (2016). Science communication: Culture, identity and citizenship. London, UK: Palgrave MacMillan.

Dixon, G., McKeever, B. W., Holton, A., Clarke, C., & Eosco, G. (2015). The power of a picture: Overcoming scientific misinformation by communicating weight-of-evidence information with visual exem- plars. Journal of Communication, 65, 639–659.

Funk, M. J., & McCombs, M. (2017). Strangers on a theoretical train: Inter-media agenda setting, community structure, and local news coverage. Journalism Studies, 18, 845–865.

Gerber, J. S., & Offit, P. A. (2009). Vaccines and autism: A tale of shifting hypotheses. Clinical Infectious Diseases, 48, 456–461.

Ginsberg, J., Mohebbi, M. H., Patel, R. S., Brammer, L., Smolinski, M. S., & Brilliant, L. (2009). Detecting influenza epidemics using search engine query data. Nature, 457, 1012–1014.

Goldstein, S., MacDonald, N. E., & Guirguis, S. (2015). Health commu- nication and vaccine hesitancy. Vaccine, 33, 4212–4214.

Granger, C. W. J. (1969). Investigating causal relations by econometric methods and cross spectral methods. Econometrica, 34, 424–438.

Groshek, J., & Groshek, M. C. (2013). Agenda trending: Reciprocity and the predictive capacity of social networking sites in intermedia agenda setting across topics over time. Media and Communication, 1, 15–27.

Hara, N., & Sanfilippo, M. R. (2016). Co-constructing controversy: Content analysis of collaborative knowledge negotiation in online communities. Information, Communication & Society, 19, 1587– 1604.

Hilgartner, S. (1990). The dominant view of popularization: Conceptual problems, political uses. Social Studies of Science, 20, 519–539.

Himelboim, I., McCreery, S., & Smith, M. (2013). Birds of a feather tweet together: Integrating network and content analyses to examine cross- ideology exposure on twitter. Journal of Computer-Mediated Communication, 18, 40–60.

Holton, A., Weberling, B., Clarke, C., & Smith, M. J. (2012). The blame frame: Media attribution of culpability about the MMR-Autism vac- cination scare. Health Communication, 27, 690–701.

Institute of Medicine. (2004). Immunization safety review: Vaccines and autism. Washington, DC: National Academies Press.

Jang, S. M., Park, Y. J., & Lee, H. (in press). Round-trip agenda setting: Tracking the intermedia process over time in the ice bucket challenge. Journalism. doi:10.1177/1464884916665405

Jolley, D., & Douglas, K. M. (2014). The effects of anti-vaccine conspiracy theories on vaccination intentions. PLoS ONE, 9, e89177.

Kata, A. (2012). Anti-vaccine activists, Web 2.0, and the postmodern paradigm: An overview of tactics and tropes used online by the anti- vaccination movement. Vaccine, 30, 3778–3789.

Kim, C., & Lee, J. K. (2016). Social media type matters: Investigating the relationship between motivation and online social network heteroge- neity. Journal of Broadcasting & Electronic Media, 60, 676–693.

116 S. M. JANG ET AL.

Koh, E. J., Dunwoody, S., Brossard, D., & Allgaier, J. (2016). Mapping neuroscientists’ perceptions of the nature and effects of public visibi- lity. Science Communication, 38, 170–196.

Larsson, A. O., & Moe, H. (2012). Studying political microblogging: Twitter users in the 2010 Swedish election campaign. New Media and Society, 14, 729–747.

Lewis, J., & Speers, T. (2003). Misleading media reporting? The MMR story. Nature Reviews: Immunology, 3, 913–918.

Liu, X., Lu, J., & Wang, H. (in press). When health information meets social media: Exploring virality on Sina Weibo. Health Communication. doi:10.1080/10410236.2016.1217454

McCombs, M., & Stroud, N. J. (2014). Psychology of agenda-setting effects: Mapping the paths of information processing. Review of Communication Research, 2, 68–93.

McKee, C., & Bohannon, K. (2016). Exploring the reasons behind par- ental refusal of vaccines. The Journal of Pediatric Pharmacology and Therapeutics, 21, 104–109.

McKeever, B. W. (2012). News framing of autism: Understanding media advocacy and the Combating Autism Act. Science Communication, 35, 213–240.

McKeever, B. W., McKeever, R., Holton, A., & Li, J.-Y. (2016). Silent majority: Childhood vaccinations and antecedents to communicative action. Mass Communication and Society, 19, 476–498.

Mills, R. (2011). Researching social news: Is reddit.com a mouthpiece for the ‘hive mind’, or a collective intelligence approach to information overload? Retrieved from http://eprints.lancs.ac.uk/61646/

National Science Board. (2016). Science and engineering indicators 2016, Retrieved from https://www.nsf.gov/statistics/2016/nsb20161/uploads/ 1/10/chapter-7.pdf

Neuman, W. R., Guggenheim, L., Jang, S. M., & Bae, S. Y. (2014). The dynamics of public attention: Agenda-setting theory meets big data. Journal of Communication, 64, 193–214.

O’Connor, B., Balasubramanyan, R., Routledge, B. R., & Smith, N. A. (2010). From tweets to polls: Linking text sentiment to public opinion time series. Icwsm, 11, 122–129.

O’Neill, B. (2006). The media’s MMR share. The Guardian. Retrieved from https://www.theguardian.com/commentisfree/2006/jun/16/ whenjournalismkills

Pew Research Center (2016). How the 2016 presidential campaign is being discussed on reddit. Retrieved from http://www.pewresearch.org/fact- tank/2016/05/26/how-the-2016-presidential-campaign-is-being-dis cussed-on-reddit/

Roberts, M., Wanta, W., & Dzwo, T.-H. D. (2002). Agenda setting and issue salience online. Communication Research, 29, 452–465.

Schafer, M. S. (2012). Online communication on climate change and climate politics: A literature review. Wiley Interdisciplinary Reviews: Climate Change, 3, 527–543.

Shelton, M., Lo, K., & Nardi, B. (2015). Online media forums as separate social lives: A qualitative study of disclosure within and beyond reddit. iConference 2015 Proceedings, 1–12.

Smith, M. J., Ellenberg, S. S., Bell, L. M., & Rubin, D. M. (2008). Media coverage of the measles-mumps-rubella vaccine and autism contro- versy and its relationship to MMR immunization rates in the United States. Pediatrics, 121, e836–e843.

Southwell, B. G., & Thorson, E. A. (2015). The prevalence, consequence, and remedy of misinformation in mass media systems. Journal of Communication, 65, 589–595.

Stryker, J. E., Wray, R. J., Hornik, R. C., & Yanovitzky, I. (2006). Validation of database search terms for content analysis: The case of cancer news coverage. Journalism and Mass Communication Quarterly, 83, 413–430.

Tambuscio, M., Ruffo, G., Flammini, A., & Menczer, F. (2015). Fact- checking effect on viral hoaxes: A model of misinformation spread in social networks. Proceedings of the 24th International Conference on World Wide Web, 977–982.

Vargo, C. J., Guo, L., McCombs, M., & Shaw, D. L. (2014). Network issue agendas on Twitter during the 2012 US presidential election. Journal of Communication, 64, 296–316.

Wakefield, A. J., Murch, S. H., Anthony, A., Linnell, J., Casson, D. M., Malik, M., . . . Walker-Smith, J. A. (1998). Retracted: Ileal-lymphoid- nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children. The Lancet, 351, 637–641.

Weeks, B., & Southwell, B. (2010). The symbiosis of news coverage and aggregate online search behavior: Obama, rumors, and presidential politics. Mass Communication and Society, 13, 341–360.

Whalen, J. (2010). UK bans doctor who linked autism to vaccine. The Wall Street Journal. Retrieved from http://online.wsj.com/article/ SB10001424052748704113504575263994195318772.html

Wilson, K., & Keelan, J. (2013). Social media and the empowering of opponents of medical technologies: The case of anti-vaccinationism. Journal of Medical Internet Research, 15, e103.

Zubiaga, A., Liakata, M., Procter, R., Hoi, G. W. S., & Tolmie, P. (2016). Analysing how people orient to and spread rumours in social media by looking at conversational threads. PloS One, 11, e0150989.

HEALTH COMMUNICATION 117