The mental health crisis in academia

The goal of this project is to provide insights into the mental health of academic researchers affiliated to universities around the world that are active Twitter users. While surveys and interviews are the most commonly used methods to analyze the mental health of individuals and groups, these approaches have disadvantages such as their limited scale, their reliance on self-reported information, and their inability to measure public disclosure behaviours. Thus, we will make use of recent development in machine learning to track scholars on Twitter, and examine the prevalence of poor mental health disclosure (i.e., considering covert and overt signals), accounting for several socio-demographic group differences (e.g., differences amongst genders, tenured seeking vs tenured staff). More specifically, after collecting a large dataset of scholars along with their publishing and tweeting records, and developing a machine learning model to capture mental health signals in tweets, we will seek to answer the following research questions:

1. How do the academic Twitter users compare to a random set of non-academic users in terms of mental health signals captured in their tweets?
2. How do the socio-demographic characteristics of academics (e.g., country, affiliation, gender, ethnicity, academic age, scientific performance) relate to the mental health signals captured in their tweets?
3. Do the mental health signals captured on Twitter differ between research fields?
4. Can we observe temporal patterns in the mental health signals of academics throughout the calendar year so that we could identify periods where academics are most at risk?
5. Can we observe patterns in the mental health signals of academics that would coincide with major global or national events such as the COVID-19 pandemic, or the changes in the political landscape (e.g., the election of anti-science governments)?