Figure 1

Directed Acyclic Graphs depicting confounding (Part A), an indirect biasing pathway (Part B), and a collider-stratification bias (Part C) in the estimation of relationships between environmental factors and recreational walking. The relationship of interest is represented in bold. A solid arrow indicates a causal effect. A dotted line indicates a spurious association generated through adjustment. A solid box around a factor indicates that it is adjusted for. A dotted box around a factor indicates that it should be adjusted for.