Predicting adolescents’ environmental action: From individual to national-level factors using an explainable machine learning approach

Abstract

As a key force in future environmental actions, youth play a crucial role in driving societal transformation. However, the factors influencing youth environmental actions have not been fully validated, and the role of national-level influences is often overlooked. This study aims to identify the factors that are associated with adolescents’ public-sphere and private-sphere environmental actions. Unlike prior studies, which typically use single-level analyses, we simultaneously examine individual, school, and national factors to capture the often-overlooked national context. Using PISA-2018 data on 420,339 adolescents from 66 countries, we used LightGBM and XGBoost to build predictive models. Shapley Additive Explanations (SHAP) were then applied to detect non-linear threshold effects and to quantify each feature’s contribution to environmental action. Results indicate that individual-level factors, such as environmental attitudes, the discussion of international events in school, and critical thinking, are significantly associated with adolescents’ private-sphere environmental actions. Conversely, national-level factors, such as Sustainable Development Goal (SDG) performance and country vulnerability, play a particularly strong role in shaping public-sphere environmental actions. This study underscores the importance of incorporating national-level factors, which have often been under-emphasized in research on youth environmental behavior.

Bo Hu 胡博

PhD Student

My research interests include environmental psychology and behavior intervention.