Population and Sample
Level 4 · Samples and links
How a poll of 1,000 people can describe a whole country. Why a huge but lopsided sample can still get it badly wrong, and how random picks fix that.
Course contentsLesson 7 of 10
Level 1 · Parts and chances
Level 2 · Chance over many tries
Level 3 · Describing data
Level 4 · Samples and links
Level 5 · Testing ideas
After this lesson you can
- Tell a population from a sample
- Explain why random picks matter
- Spot a lopsided sample
11 mincovers 3 glossary entries
Worth reading first: Normal Distribution
How can a small sample speak for a whole country?
The population is the whole group you want to know about. A sample is the smaller group you actually measure. Sampling is how you pick it. Say each member has the same chance of being picked. Then a random sample of about 1,000 people can describe a whole country to within about 3 percentage points, most of the time.
That sounds too good to be true, but it follows from the law of large numbers. A random sample is a fair mix of the whole. Doubling it from 1,000 to 2,000 only shrinks the error a little. Size helps, but how you pick matters far more.
A lopsided sample can be huge and still wrong. In 1936, a US magazine collected over 2 million poll replies and predicted the president would lose. He won with about 61 percent of the vote. Its lists leaned toward richer people, who in those years were more likely to vote against him.
Easy samples are the usual trap. Asking people who walk past one shop, who follow one account, or who choose to fill in a survey gives you the views of those people. People with strong feelings answer most, so online polls swing to the extremes.
Random picking fixes this by giving everyone the same chance. Good polls also fix their samples so that age, region and other groups match the whole. Without that care, a bigger sample just repeats the same tilt more loudly.
Samples have limits even when done well. A sample of 1,000 says little about a small group within it, such as 40 people from one town. The error grows as the group you look at shrinks.
The common mistake is to trust a big number of responses. Ask how the people were picked before you ask how many there were. Ten thousand self-selected answers can be worse than one thousand random ones.
Worked example
A survey on a website
A gym asks its members to rate it online. Of 1,200 members, 90 answer, and the average is 4.8 out of 5.
The manager then calls 50 members picked at random. Their average is 3.9. The happiest members had been the most keen to answer the first time.
The words for this lesson3
Check yourself
Which sample is most likely to describe a whole town well?
Finished this lesson?
Marking it complete also marks its 3 glossary entries learned, which updates your roadmap and your saved page. 0 are already marked.