An online study recruits 50,000 self-selected volunteers. A rival study randomly samples 400 people from the target population. Why does the smaller study support better generalisation, despite having one hundred and twenty-five times fewer participants?

A. Because size controls random error and method controls systematic error.

Only one kind of wrongness responds to sample size: random error shrinks as n grows and systematic error does not, so 50,000 self-selected volunteers locate a wrong answer with impressive precision. The 1936 Literary Digest poll had 2.4 million responses and called the election for the wrong candidate; Gallup got it right with a fraction of that.