Gallup Predicts an Election by Asking Fewer People, Correctly

Survey methodology is the discipline of constructing a representative sample and a validated questionnaire so that responses collected from a limited number of people can be generalized, within a stated margin of error, to a much larger population. The American pollster George Gallup demonstrated the method's power in 1936, correctly predicting Franklin Roosevelt's re-election using a scientifically selected sample of a few thousand respondents, a result that stood in sharp and widely publicized contrast to a national magazine's poll of ten million respondents, drawn from an unrepresentative sample, which predicted the opposite winner and got it wrong.

Most of what is publicly known about how many people hold a given religious belief, attend a given service or identify with a given tradition comes from surveys, and survey methodology is what determines whether a given poll's headline number actually means what it is reported to mean, how the sample was drawn, how the question was worded, and what margin of error surrounds the result. The atlas checks a cited religious survey's methodology before repeating its figures, since two surveys asking superficially similar questions with different wording or sampling can produce very different, equally publicized numbers.