Causal & Structural Analysis

Establishing how and why a system produces the outcomes it does.

A chain of causes, with one point of leverage and two ruled-out alternatives Five markers run in a line, each joined to the next by an arrow, showing a chain from cause to outcome. The middle marker is ringed and drawn darker than the rest: the point where intervention has leverage. Two dashed arcs bypass that middle marker, one arching above the chain and one below, each carrying a faint marker of its own - alternative explanations that were tested and not supported.

At a glance

Produces
Claims about what causes what, carrying their evidence and the alternative explanations that were tested.
Depends on
An explicit claim, and a standard of evidence agreed before the analysis starts.
Fails when
The evidence supports an association but the conclusion is stated as a cause.

Causal and structural analysis identifies the relationships, mechanisms and structural conditions that generate observed outcomes. It moves from description and correlation towards defensible claims about cause, and from there towards an account of where intervention might actually work.

How we approach it

Causal claims are the most consequential thing research produces, so they carry the highest evidential burden.

  • We state the causal claim explicitly before testing it, including what would count as evidence against it.
  • Structure and agency are treated separately. Some outcomes persist because of incentives and feedbacks, and some persist because of decisions that could be made differently.
  • Layered analysis is used where problems resist a single level of explanation, moving between the visible event, the pattern, the structure and the underlying framing.
  • Where the evidence cannot support a causal claim, we say so and describe the association instead. This is reported as a limitation, not hidden as a caveat.