Machine Learning How Human Risk Perceptions Shape Behavior: Comparing a Diversity of Learning Approaches to Predict How Climate Risk Perceptions Lead to Green Behaviors for the US Population
Presented at the 4th Annual BayesiaLab Conference in Nashville on September 29, 2016.
Dr. Asim Zia
Associate Professor of Public Policy & Decision Analysis
Department of Community Development and Applied Economics & Department of Computer Science at the University of Vermont
Abstract
Global-scale climate change is projected to have a variety of local to regional scale impacts on human societies and ecosystems. The severity of these impacts (risk magnitude) depends upon the extent to which humans at the global scale mitigate Green House Gas (GHG) emissions through switching their fossil fuel-intensive behaviors to green behaviors, as well as adapting to adverse impacts of climatic change at local scales. While a flurry of studies and Integrated Assessment Models (IAMs) have independently investigated the impacts of switching mitigation behaviors or adaptation in response to different climate scenarios, little is understood about the feedback effect of how human risk perceptions of climate change could contribute to switching human behaviors from fossil-fuel intensive pathways to climate-friendly behaviors. Standard Global Climate Models assume a disconnect between risk/adaption and mitigation behaviors. IAMs typically ignore human risk perceptions and focus on economic dynamics to predict the effect of adaption on mitigation behaviors (if at all). This study utilizes a suite of machine learning algorithms, both supervised and unsupervised, to explore how shifts in human risk perceptions, cognition, affect/emotion, trust of scientists and past experience can influence the adoption of green behaviors. We use a mixed-pool “Climate Change in the American Mind” dataset collected between 2008 and 2014 (N=13,400) and apply three supervised and three unsupervised machine learning algorithms on the data set to predict the adoption of green behaviors. We discuss the implications of the best fitting algorithm for understanding the implicit processes by which people may translate perceptions into action.