Causal inference in coupled human and natural systems

Summary

This paper addresses fundamental methodological challenges in causal inference within coupled human and natural systems (CHANS)—complex, feedback-rich systems where social and environmental dimensions interact. Using marine protected areas as a case study, the authors demonstrate that two key statistical assumptions (excludability and absence of interference) are often violated in real CHANS, leading to biased causal claims. The authors argue that human and capital mobility both creates interference (biasing inferences) and moderates causal effects themselves, and that no single methodological approach can satisfy all assumptions; instead, disciplinary integration and academic-practitioner collaboration are essential for triangulating credible causal evidence.

UK applicability

Whilst the paper focuses on marine systems, its methodological insights apply broadly to UK agricultural and environmental policy evaluation, including assessment of agri-environment schemes, land-use interventions, and water management policies where human mobility and spillover effects are common. UK policy researchers and practitioners should recognise that isolated impact evaluations of farming or land-management interventions may be subject to the same interference and excludability violations highlighted here.

Key measures

Violations of excludability and interference assumptions; human and capital mobility as sources of causal bias; applicability of causal inference methods to ~200 marine protected area studies

Outcomes reported

The study examines methodological challenges in establishing causal relationships within coupled human and natural systems, using marine protected areas and a stylised marine simulation as exemplars. It identifies how violations of excludability and interference assumptions bias causal inferences and proposes that triangulation across multiple analytical approaches is necessary for credible evidence.

Topic tags