TiCER Pilot Project 2

Status: Manuscript in advanced preparation stage
Title: Public support for environmental policy: Testing mitigation of health hazards using regression and machine learning models
Investigators: Rotem Dvir and Arnold Vedlitz

Abstract
This study explores public policy preferences to mitigate the risks of environmental hazards. While many studies assess policies to tackle general hazards, we focus on the health-related aspect of hazards and public views of proposed government solutions. We posit that for policy preferences, the main factor is people’s perceptions about the competence and degree of responsibility of the government to mitigate the risks. We also account for common factors such as risk perceptions (including health-specific concerns), knowledge and experience of health issues. We implement a combined empirical approach using regression models to estimate the relationship between the factors and observed outcome of policy support, as well as factor importance tests to ascertain the role of each element. We complement the test with a data-driven approach using a machine learning method of random forest to validate the models and gain greater predictive accuracy. Our empirical tests of national survey data (N=1207) provide ample evidence showing that views of competence and responsibility are by far the most important factors in shaping policy support. Risk perceptions, factual knowledge, and political ideology also play important roles in explaining the variations in citizens’ views. The results of this study extend existing knowledge on the factors that determine policy preferences to mitigate the health risks of environmental health hazards. The combined methodological approach offers further robustness to our ability to predict how citizens would assess different government solutions to address health-related risks from environmental hazards.