Emotion Regulation and Dynamics of Moral Concerns During the Early COVID-19 Pandemic

Abstract

The COVID-19 pandemic has upended daily life around the globe, posing a threat to public health. Intuitively, we expect that surging cases and deaths would lead to fear, distress and other negative emotions. However, using state-of-the-art methods to measure emotions and moral concerns in social media messages posted in the early stage of the pandemic, we see a counter-intuitive rise in some positive affect. In addition, we measure changes in emotions and moral concerns during notable events, such as the first US death and state-ordered lockdowns, and find that negativity shows a short-term increase and then a counter-intuitive decrease after COVID-19 takes hold in the US. We hypothesize that the long-term increase of positivity, as well as amelioration of short-term negative reactions, are associated with hedonic adaptation, emotion regulation, and a decrease of uncertainty. Finally, we identify a partisan divide in affect that emerged after the first US death, which resulted in differences in moral and emotional reactions of liberals and conservatives and among US regions. Overall, these results show how collective emotional states have changed since the pandemic began, and how social media can provide a useful tool to understand, and even regulate, diverse patterns underlying human affect.

Non-Binary Gender Expression in Online Interactions

“…To measure the presence of emotions in a tweet we use SpanEmo, a multi-label emotion classification system (Alhuzali and Ananiadou 2021; Guo et al 2022). SpanEmo utilizes BERT encodings of input text and emotion classes, and yields a higher F1 score than base BERT model.…”

#RoeOverturned: Twitter Dataset on the Abortion Rights Controversy

“…The advantage of social media datasets over surveybased measurements used in public opinion polls is the ability to feasibly assess issue positions of the larger public without biases inherent in survey self-reports, especially of contentious political issues (e.g., social desirability bias). Large-scale social media datasets have also made possible, the analysis of protest mobilization (Breuer, Landman, and Farquhar 2015;Steinert-Threlkeld 2017;Munn 2021), proliferation of misinformation (Nikolov, Flammini, and Menczer 2020;Rao et al 2021;Chen et al 2021), moral and emotional attitudes (Guo et al 2022;Priniski et al 2021), echo chambers and ideological biases on platforms (Barberá 2015; Wojcieszak et al 2022), among other democratically relevant phenomena. Previous studies (Yardi and Boyd 2010;Garimella et al 2018;Cinelli et al 2021) have also analyzed polarization with respect to wedge issues, including abortion rights in the US.…”

Using Emotion Embeddings to Transfer Knowledge Between Emotions, Languages, and Annotation Formats

“…They can guide our attention and influence our information consumption, beliefs, and our interactions [10,26]. Deep learning has enabled us to extract affective constructs from natural language [7,5], allowing emotion recognition from text at scale [13]. Nevertheless, the need for better performance across various metrics of interest still exists.…”