We are always warned that a statistical correlation should not be used to infer causation because either the causal relationship can go either way or both effects may be due to a third cause. The only way to truly determine the direction of causality is to design experiments that are specifically meant to tease it out. But sometimes we can be fooled by multiple, seemingly independent correlations that strongly suggest causality. A recent study of vitamin D deficiency and health illustrates this danger. [Read more…]