Statistical Thinking

A headline blares: 'Coffee drinkers live longer — study finds p < 0.05!' Before you brew a tenth cup, ask: 0.05 of WHAT? Was the sample 10 people or 100,000? Did they control for smoking, income, sleep? Was it one study or twenty (and did nineteen find nothing and vanish into a drawer)? A p-value is not a truth-ometer — it's a small piece of a bigger picture, and reading it well is the difference between being informed and being manipulated.