Emotional geography in the digital age: a Twitter-based study of population well-being in Thailand during COVID-19

The COVID-19 pandemic profoundly affected emotional well-being worldwide, highlighting the need to assess happiness levels in Thailand, particularly from 2020 to 2021. This study analyses nationwide happiness trends through sentiment analysis of over 65 million words from Twitter posts across 77 provinces. Provincial happiness scores were calculated using a Thai sentiment lexicon of 4929 terms adapted from the LabMT system. Spearman’s rank correlation was employed to examine the relationship between happiness scores and COVID-19 cases across different phases of the pandemic, as well as their correlation with well-being indices, particularly the Thailand Green and Happiness Index (GHI). The findings indicate an average happiness score of 5.98, with notable variations across provinces and different phases of the pandemic. The highest score (6.0) was observed during relaxation periods when social activities resumed, suggesting a temporary boost in well-being. However, no statistically significant correlation was found between happiness scores and COVID-19 case numbers in 2020, 2021, or across different waves. In 2020, weak negative correlations between happiness and indicators such as population density and household numbers imply that residents in high-density areas may have experienced lower happiness, potentially due to limited resources and heightened pandemic-related stress. Conversely, weak positive correlations with crime-related indicators suggest that social coping mechanisms, such as collective support or adaptive behaviours, may have contributed to community resilience. By 2021, only a weak correlation with violent crime arrests persisted, reflecting shifts in the factors influencing happiness as communities adapted to ongoing challenges. This research highlights the potential of social media data for real-time sentiment analysis, offering an alternative methodology for studying emotional well-being. Despite limitations such as the lack of demographic information, the findings suggest that social media analytics can complement traditional methods, providing valuable insights during crises and informing future studies on emotional geography.