Date created: | Last Updated:

Category: Project

Description: A Bayesian Study On Social Media Language During The First Wave of the COVID-19 Pandemic. Personality traits change over time, however research on it was sparse, since previous approaches were too time-consuming and expensive. Also, the necessary methodological complexity was beyond the capabilities of classical personality researchers, which resulted in contradictory results and lack of methodological standards. In this paper, we presented a simple and cost-effective method that overcame these restrictions. We introduced a machine learning approach for daily measurements to personality research, and developed a bespoke Bayesian algorithm to analyse the observed change. This resulted in uncovering concrete points of regime-shift that overlapped with relevant exogenous events for a Japanese sample of social media users. With it, we showed that personality measures displayed significant elasticity under extreme exogenous conditions during the first wave of COVID-19 and the subsequent societal countermeasures, which can be interpreted as a temporary shift from normal expression of latent psychological traits z to their respective emergency expression ze. Concretely, we found that the group of top 25% Conscientiousness users displayed a significant change in the FFM factors Agreeableness and Extraversion. We finally compared our findings with those from similar studies in other cultures, and discussed generalisability as well as future qualitative and quantitative directions for research.