Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Moral and emotional influences on attitude stability towards COVID-19 vaccines on social media

Samantha C. Phillips, Lynnette Hui Xian Ng, Wenqi Zhou, Kathleen M. Carley

TL;DR

The paper investigates how moral foundations and emotional language in COVID-19 vaccine-related tweets relate to attitude stability on social media. Using data from about $10^6$ X users during a two-month period, it employs a fine-tuned BERTweet stance classifier and Netmapper-based lexicons to quantify six moral foundations and six emotions, then assesses stance variability per user. Findings show anti-vaccine tweets carry more moral and emotional language overall, with emotional language generally linked to greater stance variation, while liberty is tied to more stable attitudes and fairness/sanctity to greater variation. These results inform targeted pro-vaccine messaging and receptive-audience identification, suggesting that messages tailored to specific moral-emotional profiles may enhance public health communication and counter misinformation. The work highlights the nuanced role of moral reasoning in attitude change and offers practical directions for platform-specific outreach strategies.

Abstract

Effective public health messaging benefits from understanding antecedents to unstable attitudes that are more likely to be influenced. This work investigates the relationship between moral and emotional bases for attitudes towards COVID-19 vaccines and variance in stance. Evaluating nearly 1 million X users over a two month period, we find that emotional language in tweets about COVID-19 vaccines is largely associated with more variation in stance of the posting user, except anger and surprise. The strength of COVID-19 vaccine attitudes associated with moral values varies across foundations. Most notably, liberty is consistently used by users with no or less variation in stance, while fairness and sanctity are used by users with more variation. Our work has implications for designing constructive pro-vaccine messaging and identifying receptive audiences.

Moral and emotional influences on attitude stability towards COVID-19 vaccines on social media

TL;DR

The paper investigates how moral foundations and emotional language in COVID-19 vaccine-related tweets relate to attitude stability on social media. Using data from about X users during a two-month period, it employs a fine-tuned BERTweet stance classifier and Netmapper-based lexicons to quantify six moral foundations and six emotions, then assesses stance variability per user. Findings show anti-vaccine tweets carry more moral and emotional language overall, with emotional language generally linked to greater stance variation, while liberty is tied to more stable attitudes and fairness/sanctity to greater variation. These results inform targeted pro-vaccine messaging and receptive-audience identification, suggesting that messages tailored to specific moral-emotional profiles may enhance public health communication and counter misinformation. The work highlights the nuanced role of moral reasoning in attitude change and offers practical directions for platform-specific outreach strategies.

Abstract

Effective public health messaging benefits from understanding antecedents to unstable attitudes that are more likely to be influenced. This work investigates the relationship between moral and emotional bases for attitudes towards COVID-19 vaccines and variance in stance. Evaluating nearly 1 million X users over a two month period, we find that emotional language in tweets about COVID-19 vaccines is largely associated with more variation in stance of the posting user, except anger and surprise. The strength of COVID-19 vaccine attitudes associated with moral values varies across foundations. Most notably, liberty is consistently used by users with no or less variation in stance, while fairness and sanctity are used by users with more variation. Our work has implications for designing constructive pro-vaccine messaging and identifying receptive audiences.
Paper Structure (13 sections, 2 figures, 2 tables)

This paper contains 13 sections, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Data and Methods
  4. Dataset
  5. Stance detection
  6. Moral value and emotion extraction
  7. Results
  8. RQ1: Moral values and emotions in pro- and anti-COVID-19 vaccine tweets
  9. RQ2: Moral values and emotions in tweets by users expressing no versus any variation in stance
  10. RQ3: Moral values and emotions in tweets by users expressing some degree of variation in stance
  11. Discussion
  12. Future work and limitations
  13. Conclusion

Figures (2)

  • Figure 1: Average and 95% CI number of concepts associated with each moral foundation (left) and emotion (right) in pro- and anti-COVID-19 vaccine tweets.
  • Figure 2: Regression coefficients for the standard deviation of expressed stance across tweets. Negative values indicate the variable is associated with less variation in stance. Error bars indicate the 95% confidence intervals. All coefficients are significant ($p<0.0001$) except for anger ($p>0.1$) and care ($p=0.049$).