Samples, populations and justified comparisons · Foundation
Introducido| English | Significado |
|---|---|
| sample/ˈsæmpl/ | A subset of a population selected for observation, measurement or analysis. |
| population/ˌpɒpjʊˈleɪʃn/ | población |
| bias/ˈbaɪəs/ | A systematic tendency to favour certain outcomes or produce unrepresentative results. |
The poll of 2.4 million
- In 1936 the magazine Literary Digest asked 2.4 million readers how they would vote and predicted that Alf Landon would beat Franklin Roosevelt. Roosevelt won easily. A much smaller poll of about 50 000 by George Gallup got it right, because his sample was chosen better.
- A sample is a smaller group used to learn about the whole population 总体.
Population and sample
- The population is everyone you want to know about, such as all 600 students in a school.
- A sample is the group you actually ask. It is quicker and cheaper than asking everyone.
A school wants to know how students travel. What is the population?
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
Everyone they want to know about.
Random samples
- In a random sample, every member has the same chance of being chosen — for example, names drawn by a computer.
- Different random samples give slightly different results. That is normal.
Two different random samples from the same population can give different results.
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
Sampling variation.
Worked example: estimating from a sample
In a random sample of 80 students, 24 walk to school. Estimate how many of the school's 600 students walk.
- Proportion: $\dfrac{24}{80} = 0.3$.
- Scale up: $0.3 \times 600 = 180$.
- Say: "about 180" — another sample might give 170 or 195.
Samples, populations and justified comparisons
Define the target population and variables before sampling
Classify the worked-case statements, then explain the units or invariant that justifies each decision.
In a random sample of 50 people, 12 are left-handed. Estimate how many of 1500 people are left-handed.
Dificultad 2/5 · 140 monedas + un pequeño bono en el primer intento de hoy
0.24 × 1500.
Bias
- A sample is biased if some groups are more likely to be chosen. Asking only a sports club about exercise overcounts active students.
- The 1936 poll asked car and phone owners — richer people in the 1930s.
Which samples are likely to be biased for "How much do students exercise?"
Dificultad 2/5 · 140 monedas + un pequeño bono en el primer intento de hoy
Only the random list gives everyone an equal chance.
Bigger is not always better
- A larger random sample gives a more reliable estimate.
- A larger biased sample is still biased: 2.4 million replies did not fix the 1936 poll.
A larger biased sample removes the bias.
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
The 1936 poll had 2.4 million replies.
A voluntary online poll is not a random sample. Only people who choose to answer are counted, and they often feel strongly. Adding more replies does not remove that bias.
A voluntary online poll is a random sample.
Dificultad 1/5 · 100 monedas + un pequeño bono en el primer intento de hoy
People choose themselves.
A sample estimates a population: use its proportion, scaled up, and call the result an estimate. Choose members at random to avoid bias. Bigger random samples are more reliable, but size never cures bias.
A shop asks the first 20 customers on a Monday morning whether they like a new product. Give two reasons the sample may not represent all customers, and suggest a better method.
Sampling method.