FDR, Gallup, and the Dangers of Sampling Bias

3 segments available

88 years ago, a leading national magazine committed a major statistical blunder that ultimately caused it to go bankrupt. Since we're in the full swing of election season, here's a cautionary note on the power of sample bias... The year is 1936. FDR is up for reelection, and the Literary Digest had sent out 10 million "straw" ballots polling people on their choice for president. They received back 2.4 million ballots. The Digest predicted that Alf Landon would beat Franklin Delanor Roosevelt 57% to 43%. But a rising challenger emerged — the man who'd later be called the 'Babe Ruth of the polling profession'. George Gallup, founder of the Gallup poll, brashly announced that the Digest would be wrong. He was convinced that these millions of Digest postcards were mailed on the basis of telephone and auto registration lists and took no account of the low-income voters who were backing the New Deal. So Gallup polled a few thousand, albeit random, people and used their data to predict FDR's victory. And guess what, FDR won by a landslide: 62% to 37%. Our guest Ben Orlin recounts this story, and one thing is clear: Sample bias is a killer. Watch the full episode here: https://www.youtube.com/watch?v=o4GgOBeXISk