ACT® · Math: Integrating Essential Skills
Informal Inferential Statistics
A sample proportion, measured on a small random group, can be scaled up to estimate a count in the whole population. The trick only works if the sample was chosen randomly, not conveniently.
Weighing 50 cafeteria trays works because those 50 trays, chosen randomly, behave like a mini version of the whole cafeteria. Whatever fraction of those trays has leftover food, you expect roughly that same fraction across all 2,000 students. You can't weigh every tray, so you let a manageable sample stand in for a population you can't fully measure.
The mechanism is just scaling a ratio. Find the proportion in the sample, k out of n, then apply that same proportion to the population size N. The estimate is (k/n) x N. This assumes the sample proportion is close to the true population proportion, which is only reasonable when the sample was drawn randomly from the whole population, not grabbed from one convenient spot.
This idea sits right next to unit rates and percent problems on the test: it's the same scaling move, just applied to a sample instead of a map or a recipe. The catch that trips people up is randomness versus size. A sample surveyed from a single classroom can be large and still useless, because every member shares whatever makes that class unusual. Fix the bias by sampling randomly across the whole population; fixing the size alone changes nothing.
Key terms
- Sample
- A smaller group measured directly, used to learn about a larger group that can't be fully measured.
- Population
- The entire group you actually care about, such as every student in the school.
- Parameter
- A true value describing the whole population, like the actual number of left handed students in the school.
- Sample proportion
- The fraction of the sample with a given trait, used to estimate that same fraction in the population.
- Bias
- A systematic skew in a sample caused by how it was chosen, which larger sample size cannot correct.
Watch it explained
- How it worksHow does a statistic measured on a sample become an estimate of a population parameter?
- Worked exampleFind the estimated number of left handed students in a school of 1,200 when 14 of a random 200 are left handed.
- Common mistakeWhy isn't a bigger sample automatically a better one when it all came from a single class?
Check yourself
End of topic.