Allday Education

Random Sampling & Representative Samples

You can't ask every student in a school, every voter in a state, or test every part off an assembly line. So statisticians ask a sample — a smaller group — and use its answers to describe the whole population. That shortcut only works if the sample actually looks like the population.

A sample is representative when every member of the population has an equal chance of being chosen. The reliable way to get that is random selection — names from a hat, every 10th customer, a random draw from the full roster. A random sample lets you generalize to the whole population; a biased one doesn't.

Representative vs. biased samples

Ask one question about any sample: did everyone in the population have an equal chance of being picked? If yes, the sample is representative. If some group was more likely to be included — or left out entirely — the sample is biased.

Surveying only the soccer team about the new schedule leaves out everyone who isn't on the team. Surveying the first 3030 students to arrive favors early risers. Surveying one classroom leaves out every other class. Each of those samples overrepresents one kind of student, so what it says may not match the school.

Sample size doesn't fix bias. Asking 500500 soccer players is still a soccer-team opinion. A random sample of 3030 from the full roster beats a biased sample of any size.

Predicting a population from a sample

Once you have a random sample, you can scale its results up to the population, because the fraction in a representative sample should match the fraction in the population.

If 1212 of 3030 sampled students ride bikes, the sample fraction is 1230\displaystyle \frac{12}{30}. Set it equal to the population fraction and solve: 1230=x300\displaystyle \frac{12}{30} = \frac{x}{300}. It's a proportion, and everything you know about solving proportions applies.

Remember these are estimates. A prediction of 120120 riders means about 120120 — another random sample would give a slightly different number, and that's normal.

Worked examples

Example 1: pick the representative sample

A school wants students' favorite sport. Which sample is representative: asking only the basketball team, or drawing 5050 names from a hat containing every student's name?

Check the basketball team: only athletes can be chosen, so most students have no chance of being picked — biased
Check the hat: every student's name is in it, so every student has an equal chance — random

Answer: Drawing 5050 names from the hat is the representative sample

Example 2: predict from a sample

In a random sample of 3030 students, 1212 ride bikes to school. The school has 300300 students. Predict how many students ride bikes.

Write the sample fraction equal to the population fraction1230=x300\dfrac{12}{30} = \dfrac{x}{300}
The population is 1010 times the sample, so scale the count by 1010x=1210x = 12 \cdot 10
Simplifyx=120x = 120

Answer: About 120120 students ride bikes

Example 3: a messier proportion

In a random sample of 2525 students, 1010 pack their lunch. The school has 400400 students. Predict how many students pack lunch.

Set up the proportion1025=x400\dfrac{10}{25} = \dfrac{x}{400}
Simplify the sample fraction1025=25\dfrac{10}{25} = \dfrac{2}{5}
Take 25\displaystyle \frac{2}{5} of the populationx=25400=160x = \dfrac{2}{5} \cdot 400 = 160

Answer: About 160160 students pack their lunch

Try one yourself

Common questions

Why does random selection make a sample representative?

Random selection gives every member an equal chance, so no group is systematically favored or left out. On average, the sample ends up with the same mix as the population — same fraction of athletes, early risers, and everyone else.

Is a bigger sample always better?

Bigger helps only if the sample is random. A larger random sample gives steadier estimates. But a huge biased sample just repeats the same bias louder — 500 soccer players still only tell you what soccer players think.

Is surveying every 10th customer really random?

It's called a systematic sample, and at this level it counts as representative: every customer has an equal chance, since who arrives 10th, 20th, 30th is effectively chance. The samples to reject are the ones favoring a group — volunteers, friends, one class.

Will the prediction match the real count exactly?

Almost never, and it isn't supposed to. A prediction of 120120 bike riders means about 120120. Different random samples give slightly different predictions, all clustered near the truth — that cluster is what makes sampling useful.

Want the video version?

Allday Everyday Math has video lessons, practice, and an AI tutor for every topic, Pre-Algebra through Algebra 2.

Try it for $1