The technique of learning about a whole population by carefully questioning a small, representative sample — a counterintuitive idea that lets a few thousand well-chosen people stand in for millions, and whose results depend entirely on choices about who is asked, how, and in what words.
In 1936, the Literary Digest mailed survey ballots to ten million Americans — a number never before attempted in the history of opinion polling. The magazine predicted a landslide victory for Republican Alf Landon over Franklin Roosevelt. Landon lost in one of the biggest electoral defeats in American history. That same year, a young pollster named George Gallup surveyed roughly fifty thousand people, chose them carefully, and called the election correctly. The Digest had more than two million responses. Gallup had fifty thousand. The Digest was wrong. This is the founding paradox of the survey: a small sample, chosen well, beats a vast sample chosen badly. What makes the difference is not how many people you ask, but who they are.