Online feelings and sentiments across Italy during pandemic: investigating the influence of socio-economic and epidemiological variables

During the on-going COVID-19 pandemic, online social media have been extensively used by policy makers and health authorities to quickly disseminate useful information and respond to public concerns in a timely fashion. Notwithstanding the huge amount of literature on analyzing positive and negative emotions conveyed by social media users, researchers have not widely investigated the main determinants of online sentiment during crises. To fill this gap, in this paper we analyse a large-scale dataset of over 1.7 M tweets in order to understand whether online feelings, expressed by Italian individuals on Twitter during the pandemic, have been affected by socio-economic and epidemiological variables. Leveraging both panel models and cross-section regressions at different geographical levels, we find that more pessimistic feelings are communicated by users located in areas where the virus hit more severely, with a higher mortality rate and a larger fraction of infected individuals with respect to the local population. Finally, we show that administrative units exhibiting the most positive emotions are those characterized by lower income per capita and larger socio-economic deprivation, suggesting that sentiments in online conversations could be driven by epidemiological factors and by the fear of economic backlashes in wealthier areas of Italy.