Bayesian modeling for pro-environmental behavior data: sorting and selecting relevant variables.

Area of Science:

• Environmental Psychology
• Statistical Modeling
• Climate Change Research

Background:

• Pro-environmental behaviors are crucial for climate change mitigation but challenging to measure statistically.
• Traditional survey methods often lack explicit response variables, complicating data analysis.
• Existing statistical models may not fully capture the nuances of environmental behavior data.

Purpose of the Study:

• To propose a robust methodological framework for analyzing pro-environmental behavior data from surveys.
• To address the complexities of non-explicit response variables in statistical modeling.
• To provide a reliable approach for evaluating climate change attitudes and behaviors.

Main Methods:

• Factorial analysis to establish validity and identify latent factors.
• Creation of indices from factor scores, with a latent factor summarizing the target variable.
• Application of Beta regression for modeling the derived index.
• Bayesian inference for variable selection and model interpretation using posterior probabilities.

Main Results:

• The proposed methodology successfully models pro-environmental behavior data.
• Factorial analysis provided validity evidence for the measurement of behaviors.
• Bayesian inference enabled effective variable selection and robust conclusions.
• The Chilean survey data illustrated the practical application and interpretability of the models.

Conclusions:

• The developed statistical methodology offers a powerful tool for analyzing complex pro-environmental behavior data.
• This approach enhances the understanding of factors influencing climate change attitudes.
• The study provides a replicable framework for researchers in environmental science and psychology.