Causal Inference

Causation

Causation means that changing one thing actually makes another thing change: if X causes Y, then intervening to alter X would, on average, move Y. This is a claim about what would happen if you reached in and acted, not just about what tends to appear together.

Causal claims are the goal of most science and policy, because we want to know whether an action (a drug, a price cut, a new law) produces an effect. The hard part is that we usually only observe the world as it happened, so separating real causation from coincidence or shared causes requires careful design or strong assumptions.

A useful test: would the effect still happen if you forced the cause to change by hand, holding everything else fixed?