Psychological Consequences of Environmental Change: A Longitudinal Study of Emotions Using Social Media Data

Environmental changes and natural disasters significantly impact humans in various ways, encompassing physical health, mental well-being, economic stability, and social equality. For example, the 2023 Canadian wildfires burned over 45 million acres, making it the most extensive wildfire season in Canada's history. The fires produced vast amounts of smoke, severely impacting air quality and causing respiratory issues in Canada and the United States. Thousands of people were evacuated as the fires destroyed homes, businesses, and infrastructure, leading to substantial economic losses.

Although the psychological and social impacts of environmental changes share similarities and are well recognized, each disaster presents unique challenges due to various levels of access to resources and support, uncertain duration, potential inequities, etc. These factors make the short- and long-term psychological consequences of environmental changes challenging to predict.

To understand the potential psychological consequences of environmental changes, longitudinal research provides an effective way as measuring the same subjects repeatedly over time allows estimation of within-person intrinsic relationships between factors and how these evolve. Collecting longitudinal data is often challenging and expensive, such work is valuable but scarce.

In this study, Tong will promote the use of social media data to enable longitudinal studies in this field as social media has become a daily activity for people to express views, connect with and communicate to others and enables the understanding human behaviors and emotions. Investigating human emotion changes and variations across individuals and subpopulations— potentially considering factors like gender, parental status, geography, race, and income level. This can help predict the consequences of a natural disaster and related public policies.

The goal of the study is to:

(1) meaningfully interpret emotion changes and how potential influential factors may affect emotion changes for different sub-populations during and after the environmental change, and

(2) develop and apply novel statistical methods to analyze X/Twitter data and investigate factors that influence the change of emotions using longitudinal structural equation models (SEMs).

The project will benefit a large interdisciplinary community in environmental science.