The 1936 Poll — Data Village · Data Stories
A magazine asked millions of people and got it wrong. One young man asked far fewer and got it right. How? Data Stories · MyLeoNes.
A magazine asked millions of people and got it wrong. One young man asked far fewer and got it right. How?
The Story
In 1936, in the United States, Literary Digest magazine received about 2.4 million replies and predicted Alf Landon would win the election. But Franklin D. Roosevelt won with about 61% of the vote.
The magazine mostly picked people from phone and car owners’ lists, who at that time were usually richer. George Gallup used a much smaller sample, but it was more like the whole country, and he got the winner right. The lesson: just getting more answers isn’t enough, your sample needs to truly represent everyone.
Key Ideas
- Sample: A small group that you pick to tell you about a big group.
- Representative: A sample that’s similar to the whole group, with people of all types.
- Bias: When a sample always leans one way and gives a wrong answer.
- Margin of error: A poll is never exact: it gives an approximate answer.
Step by Step
- Decide what the big group is.
- Pick a mixed sample.
- Ask everyone the same question.
- Count the answers.
- Say the result with the margin of error.
Try it Yourself
Ask five people what their favourite animal is. Then ask another ten. Did the result change?
Keep exploring
Other languages
Loading MyLeoNes™…
