YouTube: Striking Back at Pro-ana Communities

A Finnish group used a hi-tech approach to evaluate pro- and anti-pro-ana video viewers.

Reprinted from Eating Disorders Review
January/February Volume 27, Number 1
©2016 IAEDP

Pro-anorexia (pro-ana) online communities have aroused concern because they are readily accessible and promote adoption of an illness with high morbidity and mortality. Moreover, they are very frequently visited: for example, in the EU Kids Online survey, 10% of children 9 to 16 years old had viewed such sites (Livingstone et al, London: LSE, 2011).

But, what about the anti-pro-ana online sites and videos? Are they equally effective? Dr. Atte Oksanen and colleagues at two universities in Finland teamed with the Swiss Federal Institute of Technology (ETH Zurich) in Zurich, Switzerland, to study emotional reactions to pro- and anti-pro-ana online content using a technique called sentiment analysis (J Med Internet Res. 2015. 17:e256)

Pro-ana communities are found on various social media sites, including Facebook, YouTube, Twitter, Instagram, Pinterest, and Flickr. YouTube is a social media mega-star that reaches more than 1 billion viewers, and 300 hours of video are uploaded to it every minute. YouTube is also the most popular social media site that uses publicly available videos and comments; thus, its easy availability and popularity make it a significant source for information about anorexia. Dr. Oksanen and colleagues used one of the features of YouTube, the feedback by viewers of videos to either “like” or “dislike” the video or to post a comment in the comment section below the video. Any registered YouTube user can use this function. The authors selected YouTubebecause users commonly engaged in active discussion by expressing either positive or negative sentiments in their messages, so that, for example, anti-pro-anorexia content may challenge pro-ana content. The researchers asked three research questions concerning the characteristics of the pro-anorexia and anti-anorexia videos and video uploaders: the strength of the positive and negative emotional feedback, and whether uploading time or video length was related to comments.

From October 15 to 29, 2014, the authors retrieved YouTube videos by searching two terms, “pro-ana” and “anti pro-ana.” They also used an automatic web crawler using the YouTube Data Application Programming Interface, a technique used to measure video popularity during political campaigns. Video comments were assessed with the SentiStrength automatic sentiment tool, which uses an algorithm to estimate the sentiment content based on a list of approximately 3,000 sentiment words and grammatical categories. 

Anti-pro-anorexia videos were more popular among viewers

The authors collected a large amount of data, including 133 pro-ana videos and 262 anti-pro-ana videos. These 395 videos drew a total of 12,161 comments from 7,903 commenters. Only 1% of viewers commented on both types of videos. Ninety-two percent of the video uploaders were female and the rest qualified themselves as unknown or “other” genders. The viewers came from 13 different countries; 44% came from the US, but the authors reported that the country of origin had no impact on the comments. The two types of videos had more than 6 million total views, and a positive finding was that the anti-pro-ana videos were more popular than the pro-ana videos.

Unlike an earlier Facebook study in which a pro-anorexia group was found to be more active and better organized than an anti-pro-ana group (Eat Weight Disord. 2013. 18:413), in their study the authors found that pro-anti-ana videos were more popular among viewers, received more video “likes,” and were commented upon more positively. The difference between these two studies may be traced to the type of social media examined. YouTube is a publicly available global platform for distributing video content; with Facebook an individual often sets up the group and thus has more power and means to manage what is said and distributed within that group and how it is structured. The content is also commonly available only to those users who join the “group.” In contrast, YouTube material is available to all users and can spread virally to other YouTube users and other social media. 

The results clearly show that the pro-ana community has online opponents, at least within the YouTube community. The anti-pro-ana videos were more popular and produced more positive feedback than did the videos promoting anorexia. Study limitations included the use of only one online community and the fact that all comments analyzed were limited to English. However, the results should be viewed as a source of encouragement in the efforts against pro-ana sites.