Natural Experiments in Environment-Behavior Research

Chanam Lee (Texas A&M University)
Marcia Ory (Texas A&M University)
Xuemei Zhu (Texas A&M University)
Wei Li (Texas A&M University)

Establishing causality in population studies involving environment-behavior relationships is not easy. Natural Experiments (NEs) have become increasingly popular as ways to evaluate causal impacts of various environmental interventions that are not suitable for controlled experimentation. NEs refer to observational studies that examine exposures to an intervention that is not designed by the researcher, which often include changes in the environment and policies.
This research presentation highlights the unique opportunities and challenges of NEs, drawing from the specific examples of two on-going NE projects in Texas funded by the National Institutes of Health. They represent two common types of NEs: “relocation” (e.g. people moving to live or work ina different environment) and “exposure” (e.g. new grocery store or transit service in a neighborhood). Using surveys and objective measures (GPS and accelerometer), the NE projects examine short-term and long-term casual effects of the target interventions - relocation to a walkable neighborhood and exposure to a new transit system, respectively - on people’s physical and social activity levels and travel behavior.
It first presents general research design and study execution strategies commonly used in NEs, followed by the specific strategies used in the two NE project examples (e.g. propensity score matching, multiple comparison groups, multi-channel marketing, and machine learning algorithms). It then discusses expected and unexpected challenges of NEs (e.g. baseline data collection time constraint, dynamic changes accompanying the target intervention, participant recruitment/retention, and sample bias) and effective responses to those challenges.
This research presentation aims to demonstrate the strong potential for NEs to advance environment-behavior research. It offers insights on overcoming frequently encountered challenges in NEs to promote their further applications in future environment-behavior research.