Samples and Populations
We are going to learn about what it means to sample from a population. While a population is a group of people or things that you are interested in (e.g. university students), a sample is a subset of that group that you can actually measure (e.g. the people who take part in my experiment). The goal of inferential statistics is to use that sample to learn something about the population as a whole.
Watch the video and then answer the questions below.
Nine-minute video
You can also view this video on YouTube
You can find the slides here and also as .odp.
Key Points
- We use a sample to infer knowledge about a population.
- Inference in statistics is about constructing a convincing argument.
- A population is the group of people or things that we’re interested in. This might be every human being, or every output that an algorithm will ever give.
- A sample is the group we actually measure: the people who do our experiment, or the algorithm outputs that we collect.
- Validity is how convincing your inference argument is. Did you measure what you think you measured (e.g. physiological reaction times in humans)? Or did you actually measure something else (physiological reaction times in over-worked undergraduate students)?
Questions
1. Check your understanding
1. Roulette
Roulette is a casino game where a numbered wheel is spun by a croupier. A ball is spun in the opposite direction and eventually lands in one of the 37 (or 38) numbers.
If you suspected this wheel was biased and ran a study to prove it, what would your population be?
What would your sample be?
What does it mean for a sample to be biased?
Summary
In this section we have learned about sampling from a population and threats to validity involved. Once you’ve completed the questions, you can move on to the next section about hypothesis testing.