The Emotional Climate of Academia: Exploring Social Media Data as an Indicator of Well-Being

Abstract

Understanding and promoting researchers‘ well-being is crucial for successful research outcomes and a thriving scientific community. Traditional well-being assessments can be resource-intensive, but prior research suggests that sentiment analysis of social media text data can provide insights into the emotional well-being in the general population. In this proof-of-concept study, we explore whether social media sentiment can serve as an indicator of researchers‘ well-being on a group-level. Specifically, we analyze the emotional climate of the academic community by applying sentiment analysis to a dataset of more than 3.48 Million messages from 14,876 psychology researchers, posted on the platform Twitter/X between 2019 and 2021. As this period is one year before and after the onset of the coronavirus pandemic, we used the findings from COVID-19 research to derive two hypotheses: lower well-being during the pandemic is reflected in more positive/less negative sentiment in researchers‘ social media posts (H1), and a stronger impact on female researchers is reflected in even more positive/less negative sentiment in female researchers‘ posts (H2). Using structural break analysis, the impact of the pandemic was found to be statistically significant for positive sentiments. A differential effect by gender was observed descriptively, but did not reach statistical significance. The results suggest a positivity bias in scholarly communication on Twitter/X, with only changes in positive sentiments as an indicator of researchers‘ emotional well-being. Gender differences were found for general sentiment magnitude, but not in the effect of the pandemic. Exploratory analysis of cognitive well-being revealed relationships and accomplishment as the predominant PERMA + 4 dimensions in researchers‘ posts. We discuss promising expansions of our approach and highlight practical implications for policymakers.