From Hesitancy Framings to Vaccine Hesitancy Profiles: A Journey of Stance, Ontological Commitments and Moral Foundations

Abstract

While billions of COVID-19 vaccines have been administered, too many people remain hesitant. Twitter, with its substantial reach and daily exposure, is an excellent resource for examining how people frame their vaccine hesitancy and to uncover vaccine hesitancy profiles. In this paper we expose our processing journey from identifying Vaccine Hesitancy Framings in a collection of 9,133,471 original tweets discussing the COVID-19 vaccines, establishing their ontological commitments, annotating the Moral Foundations they imply to the automatic recognition of the stance of the tweet authors toward any of the CoVaxFrames that we have identified. When we found that 805,336 Twitter users had a stance towards some CoVaxFrames in either the 9,133,471 original tweets or their 17,346,664 retweets, we were able to derive nine different Vaccine Hesitancy Profiles of these users and to interpret these profiles based on the ontological commitments of the frames they evoked in their tweets and on value of their stance towards the evoked frames.

Topics in Antivax and Provax Discourse: Yearlong Synoptic Study of COVID-19 Vaccine Tweets

“…Trust and safety-related topics, for example, side effects, rushed vaccine, featured prominently in the antivax tweets, which supports other work that identified safety and trust (in institutions and governments) as a key hurdle in addressing vaccine hesitancy [18,20,23]. Our stance detection approach also enabled the definitive identification of a set of dual-stance users who contributed a significant volume to both antivax and provax tweets, supporting existing findings [4,26,[119][120][121]133].…”

“…Most of these studies analyzed large tweet data sets comprising millions of tweets. In contrast, 6 studies conducted manual content analysis of a few thousand tweets [ 4 , 18 , 23 , 119 , 123 , 133 ], whereas another 6 papers used lexical and linguistic models, proprietary or otherwise, to detect vaccine hesitancy [ 7 , 20 , 25 , 121 , 129 , 132 ] using moderate-sized data sets (up to several hundred thousand tweets). Finally, a few papers classify hashtags [ 120 , 133 ] or URLs [ 2 , 126 ] to differentiate between antivax and provax tweets using large data sets with millions of tweets.…”

“…A growing body of literature has analyzed social media posts, particularly tweets related to COVID-19 vaccines. Here, we compared our work with 29 significant Twitter studies [2][3][4][5][6][7]18,20,21,[23][24][25][26][119][120][121][122][123][124][125][126][127][128][129][130][131][132][133][134]. Unsupervised methods, such as sentiment analysis and topic modeling, are the most popular methods used to classify and analyze tweets; 11 studies used some combination of sentiment analysis, emotion analysis, and topic modeling [3,6,20,21,24,124,126,128,130,131,134].…”

Topics in Antivax and Provax Discourse: Yearlong Synoptic Study of COVID-19 Vaccine Tweets (Preprint)

“…They classified users as 'anti-vaxxer' or 'pro-vaxxer' if 70% of their tweets were anti-vax or pro-vax, respectively. [9] developed a framework of Vaccine Hesitancy Framings (VHF), which involved defining specific factors behind vaccine hesitancy, identification of tweets' stances towards these VHFs using linguistic features, and classifying users based on their stances into different profiles. 22% of the users were classified into the profile 'undecided' which have almost balanced stances (accept or reject) towards VHFs and 8% are 'concerned' who are supportive of vaccines but have minor concerns [9].…”

“…We will discuss the related work in the Section II in detail. None of the papers we know, discussed the dual-stance users explicitly, however, few have hinted about their presence [5]- [9]. In our previous paper [1], we provided some preliminary results of content analysis through topic modelling but were not able to go deeper into exploring this specific cohort of dual-stance users, which we address now in this paper.…”