Ice, Ice Bucket (Challenge): A data driven investigation into shareability


By Danielle K. Kilgo, Kyser Lough, Martin J. Riedl

The Amyotrophic Lateral Sclerosis Association (ALSA) and Muscular Dystrophy Association (MDA) had never seen anything like what happened in 2014. The non-profit organizations started receiving metaphorical buckets of donations while the people around the world dumped buckets of ice water on their head. This included politicians, celebrities, actors, and musicians such as … you guessed it, Vanilla Ice.  Both organizations were quick to align with the organic grassroots movement-turned-formal-health-campaign. The Ice Bucket Challenge (IBC) enjoyed both the exposure that accompanied the virality of the memetic message, physical donor action, free recirculation, little controversy and extensive media coverage. 

To investigate the news coverage generated by this phenomenon, we examined the elements of a story that drove social media audiences to share these articles, in an effort to reimagine the principles of health campaigns in a digital age.

Image: The Ice Bucket Challenge. Credit: Kymberly Janisch.

IBC's organic means contested the role media, and the audience have in our traditional understanding of health campaigns. Typically, campaigns are designed around five key principles: the use of theory, audience segmentation, message design, channel selection and message exposure. Traditional theory was elusive in this campaign because neither ALSA or MDA were the campaign creators.  The Ice Bucket Challenge existed before it was ever associated with ALS.  Digital records show that it wasn't until golfer Chris Kennedy accepted the challenge and selected the ALSA in his challenge that the familiar IBC legend came to fruition. The viral campaign was not hurting for audience segmentation strategies simply because it met every definition of viral. Social media users were mostly in charge of message design and channel selection. However, when it comes to message exposure, our interests fell not in how the videos were shared, but how news media outlets (traditional, digital, alternative) covered this online phenomenon.   

To examine news coverage of the event, we were interested in two concepts. The first was news values: the selection criteria of the story. We also looked at the emotional appeals evoked in the story. Emotional appeals are thought of as an engagement strategy from most communication perspectives. We know emotional appeals are consistently found in news coverage as well, from coverage of crisis and protest to coverage of local events. Scholarship suggests that in online health communities, five primary emotions regularly appear: hope, humor, anger, sadness, and fear. We used these as a guide for exploring emotional appeals in online news articles.

Our research also repositions our understanding of the impact of journalistic coverage by examining how articles were networked within the social media sphere. We conceptualized sharing as a form of message amplification, and to better understand digital coverage in the social media sphere, the correlation of these results help us examine the new media landscape in a more ecologically valid way.

We pulled a random sample of 750 articles from a database provided by the social media data collection company NewsWhip that includes over 50,000 news organizations worldwide. The web addresses of each article (and any shortened versions such as bit.ly links) are tracked across the public APIs of Facebook and Twitter to collect sharing data. These data were based on the user posting a link to the article publicly (we couldn’t include privately-shared links).

During several coder training sessions, the codebook adapted to support the complexity and variety of digital journalism article construction, variations in multimedia and to increase coder agreement among variables. The coders (one an author of the study, the other a graduate student) identified the most prominent news value, which included proximity, timeliness, oddity, human interest, unusualness, usefulness and other. Additionally, coders looked for the presence or absence of emotional appeals. Though multimedia and visual communication is regularly discarded in standard content analysis projects, our study included multimedia. Intercoder reliability testing was performed before the final sample was coded, and reliability scores were above the recommended measures. Human coded information was correlated with cumulative sharing score data provided by NewsWhip.

Our results highlight media production trends at several levels. Significant coverage was devoted to celebrity involvement in IBC. Alternative media organizations are crafting more stories using celebrities and prominent people as the news value (49.4%) than other organizations. We weren’t surprised to see high use of prominent people since many celebrities participated in the IBC. The difference in alternative media’s use follows traditional newsroom logic that stories with more impact have higher priority than those just talking about a celebrity.

Traditional and digital native media, which in this study routinely presented information similar characteristics, were more likely to emphasis the societal impact of the campaign, which ultimately provides an element of seriousness that does not always accompany stories of celebrities.  

Research has shown that hope and sadness tend to appear most often in health-specific online community discussions, and we saw a similar trend in our study. News articles were more likely to include emotional appeals of hope, followed by sadness and humor. Additionally, digitally native organizations used emotional appeals most of all, including primarily hope and sadness. While digital native organizations adhere to similar news values as traditional outlets, their flexibility to adopt new practices and to understand their digital audiences may explain the rise in emotional content.

In terms of sharing online, news values didn’t help explain what the audience valued in sharing. For example, while stories centered on prominent figure and celebrity involvement in the campaign dominated coverage, the inclusion of this news value did not increase sharing. Emotions, however, do play a role in predicting sharing. Interestingly, stories that included sad emotional appeals were shared less often than other articles. We explain this by thinking about the paradox of sadness: humans are engaged by sadness, but don’t necessarily want to encounter it. Our study finds that this psychological process may play out unconsciously in the digital sphere. Finally, sharing on the different social media networks is very different, and ultimately, Facebook users appear more likely to react to emotional appeals than Twitter users.

Future health campaigns that consider social media should create fundamentally different strategies for different platforms. Though IBC may be an outlier, it provides an ideal case for analyzing the relationship between health information and successful engagement with a world population. In addition, journalists should continue to understand the audiences they are reporting to, and the parts of news that are valued in different platforms.

About the authors

Danielle K. Kilgo (Ph.D., University of Texas at Austin) is an assistant professor of journalism in The Media School at Indiana University. Her research focuses on race, gender and disability issues in visual and digital communication. Currently, she concentrates on the international media coverage of social movements. Prior to her return to the academy, she worked as a photojournalist and public relations professional. Twitter: @danikathleen.

Kyser Lough is a Journalism doctoral student at the University of Texas at Austin and a Population Health Scholar at the University of Texas System. Prior to coming to UT, he spent ten years in journalism as a reporter and photographer, as well as in healthcare marketing. His research focuses on visual communication and solutions journalism. Twitter: @KyserL.

Martin J. Riedl is a doctoral student at the University of Texas at Austin's School of Journalism. Prior to UT, he obtained master's degrees in social sciences (Humboldt University of Berlin) and media management (Hanover University of Music, Drama and Media). His research focuses on content moderation, media sociology, internet governance and social media. Twitter: @martinriedl.

Explore the study here.

Image: Global Panorama.