Understanding Anti-Vaccination Attitudes in Social Media

"What drives people to develop and perpetuate the anti-vaccination movement?"

Through a case study of the vaccination debate, this study demonstrates how analysis of natural language expressions and social media activities can paint a multifaceted picture of attitudes around a divisive topic. Using 4 years of longitudinal data capturing vaccine discussions on Twitter, the researchers identify users who persistently hold pro and anti attitudes, as well as those who newly adopt anti attitudes towards vaccination. After gathering each user's entire Twitter timeline, totaling to over 3 million tweets, they explore differences in the individual narratives across the user cohorts. Given the strong base of conspiratorial thinking underlying anti-vaccination attitudes, they conclude by highlighting the need for alternatives to traditional methods of using authoritative sources such as the government when correcting misleading vaccination claims.

Following an introduction that briefly traces the emergence of a vaccination culture promoting anti-vaccination and its importance in light of the increasing reliance on online media for health information, the researchers provide a brief overview of 2 broad areas of research relevant to the study: attitude measurement and text analytic approaches to study user traits. They point out that the rapid growth of text-based social media has opened new opportunities to study attitudes unobtrusively, as they naturally unfold in large populations and over long time periods. Drawing on the success of studying attitudes from online textual data, they built a classifier to determine positive and negative attitudes towards vaccination. Attitudes and language are intimately related, and the researchers use a text analytic technique called the Meaning Extraction Method (MEM), complemented with Linguistic Inquiry and Word Count (LIWC) to examine differences in expression styles.

The paper explains the data collection process, which covered tweets from January 1 2012 to June 30 2015, totaling 315,240 tweets generated by 144,817 unique users. The classifier to identify pro and anti stance of the collected vaccine tweets was based on a 2-step labeling process: (1) gathering human annotations on a sample, and (2) leveraging the labeled data to annotate the remaining collection of tweets. They then segregated 3 principle actors: long-term advocates of pro and anti vaccination attitude and users newly adopting anti-vaccination attitude. The idea is that long-term anti-vaccination advocates play an important role in preventing eradication because they sustain weakness in herd immunity, and thus it is crucial to understand them and their motivations. Examining new anti-vaccination proponents allows us to understand the type of person who would adopt such a stance despite strong recommendations to the contrary from authoritative organisations like the Centers for Disease Control and Prevention (CDC).

Using MEM and LIWC, the researchers identify the characteristics of individuals holding persistent attitudes towards vaccination, the active-pro and active-anti user cohorts. They explore the questions: What topics are relevant to them? How do they present themselves? Do their social media characteristics differ? Then, to understand the characteristics of people who join the anti-vaccination movement, they ask these same questions, comparing recent adoptees (joining-anti) with those already perpetuating the movement (active-anti). Taken together, these comparisons suggest that those joining the anti-vaccination cohort are more social and less definitive - indicators of people who might join a cause or a group. Long-term anti-vaccination supporters who are concrete and complex in thought had higher attention status ratio - indicators of people who could perpetuate a cause. Those joining also posted relatively more content about vaccination (higher engagement), suggesting that for them this was, at least initially, a specific issue of interest, while for long-term anti-vaccination advocates, vaccination appears to be one in a number of government conspiracy issues of interest.

Overall, in brief, the researchers find that people holding persistent anti-vaccination attitudes use more direct language and have higher expressions of anger compared to their pro counterparts. They also show general conspiracy thinking and mistrust in the government. (For instance, themes emerging from the twitter history of anti-vaccine advocates refer to terms like "government conspiracy"; they had significantly higher mentions of "vaccine fraud" after their first mention of vaccination.) Adopters of anti-vaccine attitudes show similar conspiratorial ideation and suspicion toward government even before they start expressing anti-vaccine attitudes. This suggests that the new adoptees are already predisposed to form anti-vaccine attitudes.

As explained here, conspiratorial beliefs have a "self-sealing quality", meaning that attempts to reject the theory may backfire and that those very attempts may be characterised as further proof of conspiracy. Hence, conspiratorial beliefs are very difficult to correct. This suggests that dispelling vaccination myths among long-term anti-vaccine supporters might be very difficult, despite the amount of scientific evidence and rational arguments provided by government officials, scientific journals, or other regulatory bodies. In terms of the possible use of emotional appeals to dismiss vaccine myths, the researchers find that - complementing their cognitive concreteness with decreased positive emotion expressions - long-term anti-vaccine advocates can be identified as categorical thinkers: people whose writing is more focused on objects, things, and categories, marked by higher use of nouns, articles, and prepositions. Such categorical thinkers tend to be emotionally distant. In contrast, active-pro users are characterised by higher cognitive processing and an informal expression style signaling dynamic thinking. Thus, appealing to the emotional side of the anti-vaccination movement also is not likely to successfully change their attitudes.

Moreover, the study found that long-term anti-vaccine advocates exhibit higher sense of group solidarity. Individuals in such close-knit groups typically end up adhering to their extreme positions, often fueling beliefs in false conspiracies, and are particularly resistant to correction. One possible strategy discussed here would be to introduce informational and social diversity in the closely-knit anti-vaccination groups.

Directions for future work are proposed. For example, such work could examine cohorts who are not heavy social media users but still have strong anti-vaccination views. It could look at why pro or anti vaccine advocates exhibit certain attitudes. It could also build on the above-summarised findings to investigate vaccine discussions or more general health information debates on social media sites other than Twitter, as well as in mainstream media.