Comment by mitthrowaway2

You don't necessarily need to establish causality in one grand step; it can still be effective to accumulate evidence of causality, even if this evidence cannot distinguish between other competing non-causal explanations.