Worry, coping and resignation - A repeated-measures study on emotional responses after a year in the pandemic

Maximilian Mozes, Isabelle van der Vegt, Bennett KleinbergarXiv 2021 · OpenReview

Abstract

The introduction of COVID-19 lockdown measures and an outlook on return to normality are demanding societal changes. Among the most pressing questions is how individuals adjust to the pandemic. This paper examines the emotional responses to the pandemic in a repeated-measures design. Data (n=1698) were collected in April 2020 (during strict lockdown measures) and in April 2021 (when vaccination programmes gained traction). We asked participants to report their emotions and express these in text data. Statistical tests revealed an average trend towards better adjustment to the pandemic. However, clustering analyses suggested a more complex heterogeneous pattern with a well-coping and a resigning subgroup of participants. Linguistic computational analyses uncovered that topics and n-gram frequencies shifted towards attention to the vaccination programme and away from general worrying. Implications for public mental health efforts in identifying people at heightened risk are discussed. The dataset is made publicly available.

Introduction

Most research on pandemic-related mental health has relied on "snapshot" observational data—often scraped from social media. While useful, these studies lack "ground truth" (known emotional states of the authors) and cannot track how the same individuals change over time. This study addresses this gap by using a repeated-measures design. By surveying the same cohort twice, exactly one year apart, the researchers distinguish between general population trends and individual trajectories, investigating how emotional responses shifted as the UK moved from strict isolation toward the "new normal" of vaccination and reopening.

Methodology

The researchers employed a multi-modal analytical approach utilizing the COVID-19 Real World Worry Dataset (RWWD).

  • Participants: 1,698 individuals returned for the Phase 2 follow-up in April 2021 from an original pool of 2,491 in April 2020.
  • Psychometric Analysis: Participants rated eight emotions (anger, anxiety, desire, disgust, fear, happiness, relaxation, sadness, worry) on a 9-point Likert scale. Within-subjects t-tests, Cohen’s $d$ effect sizes, and Bayes Factors (BF) were used to quantify emotional shifts.
  • Unsupervised Clustering: $k$-means clustering was applied to the change scores (Phase 2 minus Phase 1) to identify distinct subgroups of coping.
  • Computational Linguistics: The study used Linguistic Inquiry and Word Count (LIWC), Structural Topic Modeling (STM), and $n$-gram differentiation to analyze text responses (one long, $\ge$ 500 characters; one short, "Tweet-length").

Results: Overall Emotional Shifts

Statistical analysis revealed substantial decreases in negative emotions over the one-year period. The most significant reductions were observed in worry ($d=0.86$), fear ($d=0.75$), and anxiety ($d=0.55$). Conversely, positive emotions like happiness ($d=0.32$) and relaxation ($d=0.26$) saw moderate increases. These shifts suggest a general trend toward better adjustment as the pandemic progressed and vaccination efforts began.

Mean emotional ratings and statistical comparisons between Phase 1 (April 2020) and Phase 2 (April 2021).

Subgroups of Adjustment: Well-Coping vs. Resignation

A key finding of this research is that recovery was not universal. Unsupervised clustering identified two distinct trajectories:

  1. Well-coping (43.6%): This group showed marked improvements across all emotional variables, with significant drops in fear and anxiety and increases in happiness.
  2. Resignation (56.4%): This larger subgroup reported lower fear and worry but also lower happiness and relaxation. This suggests a state of emotional exhaustion or "languishing" rather than true recovery.
Mean change scores and effect sizes for the 'well-coping' and 'resignation' clusters.

Linguistic Insights

Linguistic analysis mirrored the psychometric shifts. In 2020, dominant themes centered on "home," "family," and immediate "worry." By 2021, the discourse shifted toward "vaccines," "normality," and "looking forward." Notably, the word "vaccine" served as a powerful emotional pivot; participants mentioning vaccines showed significantly larger reductions in anxiety ($d=0.22$) and sadness ($d=0.17$). However, negative outlooks on government handling remained a constant, phase-independent theme.

Discussion and Limitations

While the study shows a general improvement in mental well-being, the identification of the "resignation" cluster highlights the complex, non-linear nature of pandemic recovery. The study's limitations include a geographic bias toward the UK and the subjective nature of unsupervised NLP interpretations. Furthermore, the researchers found that while text data is a good proxy for population-level shifts, it struggled to predict individual cluster membership, suggesting that internal emotional states are more nuanced than what is captured in short-form text. Future work should explore more sophisticated linguistic features or deep learning models to bridge this predictive gap.

Summary updated

Abstract

The introduction of COVID-19 lockdown measures and an outlook on return to normality are demanding societal changes. Among the most pressing questions is how individuals adjust to the pandemic. This paper examines the emotional responses to the pandemic in a repeated-measures design. Data (n=1698) were collected in April 2020 (during strict lockdown measures) and in April 2021 (when vaccination programmes gained traction). We asked participants to report their emotions and express these in text data. Statistical tests revealed an average trend towards better adjustment to the pandemic. However, clustering analyses suggested a more complex heterogeneous pattern with a well-coping and a resigning subgroup of participants. Linguistic computational analyses uncovered that topics and n-gram frequencies shifted towards attention to the vaccination programme and away from general worrying. Implications for public mental health efforts in identifying people at heightened risk are discussed. The dataset is made publicly available.
BibTeX
@misc{mozes2021worry,
  author = {Maximilian Mozes and Isabelle van der Vegt and Bennett Kleinberg},
  title = {{Worry, coping and resignation - A repeated-measures study on emotional responses after a year in the pandemic}},
  year = {2021},
  eprint = {2107.03466},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2107.03466},
}
Figure