A random sample is a subset of a population selected so that every individual has an equal chance of being chosen, which reduces bias and is the condition AP Statistics requires you to check (along with the 10% condition) before doing inference for proportions, means, chi-square tests, or slopes.
A random sample is a group pulled from a population using chance, so every individual has an equal probability of being selected. The point is fairness. When chance does the picking instead of a researcher or a volunteer sign-up sheet, the sample tends to look like the population, and any difference between the sample statistic and the true parameter comes from random variation rather than bias.
In AP Stats, "random sample" is more than a vocabulary word. It's a condition you must verify before almost every inference procedure. The CED's essential knowledge for confidence intervals and significance tests (6.2.B, 6.4.C, 8.2.E, 9.2.B, and more) says the same thing each time. To check for independence, data should be collected using a random sample or a randomized experiment, and when sampling without replacement, n must be at most 10% of the population. Random sampling is also what makes generalization legal. If your sample was random, your conclusion can extend to the whole population it came from. If it wasn't, your conclusion stays stuck with the people you actually measured.
Random sampling is the single most repeated condition in the entire inference half of the course. It shows up in the conditions checks for one-sample z-intervals and z-tests for proportions (LOs 6.2.B and 6.4.C), two-sample proportion tests (6.10.C), chi-square goodness-of-fit and two-way table tests (8.2.E and 8.5.C), and t-intervals and t-tests for slopes (9.2.B and 9.4.C). The interpretation of a confidence interval (6.3.A) literally depends on it, since the whole "in repeated random sampling, about C% of intervals capture the parameter" logic assumes the sample was random. Earlier in the course, the binomial setup in Topic 4.11 also leans on random selection to justify treating trials as independent. If an FRQ asks you to verify conditions and you skip "the problem states this was a random sample," you lose easy points on something you knew.
Keep studying AP Statistics Unit 4
10% Condition (Units 6-9)
Random sample and the 10% condition travel as a pair. The random sample gets you unbiased selection, but when you sample without replacement, each pick slightly changes the population left over. Checking n ≤ 10% of N keeps that effect small enough to treat observations as independent anyway. On the exam, the independence check is always both pieces together.
Stratified Sampling (Units 3 & 8)
A stratified random sample splits the population into groups first, then randomly samples within each group. This matters in Unit 8 because the chi-square test for homogeneity specifically requires a stratified random sample or randomized experiment, while the test for independence requires one simple random sample. Knowing which sampling design was used is how you tell those two chi-square tests apart.
Sampling Error (Units 5 & 7)
Even a perfectly random sample won't match the population exactly. That sample-to-sample wobble is sampling error, and it's the random variation Topic 7.1 warns can cause inference errors. Random sampling doesn't eliminate this variation. It just makes the variation predictable, which is exactly what margins of error and p-values quantify.
Binomial Random Variable (Unit 4)
The binomial model in Topic 4.11 assumes independent trials with a fixed success probability. A random sample (plus the 10% condition) is what justifies that assumption when you're counting successes in a sample, which is why the binomial machinery later powers the normality check for proportion inference (np₀ ≥ 10 and n(1-p₀) ≥ 10).
Random samples are everywhere on the AP Stats exam, usually as the setup line of a problem. Released FRQs hand you the phrase directly, like the 2017 FRQ about a random sample of 207 men and women treated for schizophrenia (chi-square) or the 2018 FRQ where a grocery store manager selected a random sample of 11 customers (regression inference). Your job is to spot that phrase and cite it when verifying conditions. Multiple-choice questions test it two ways. Some give you a scenario and ask which condition is met or violated, like a question asking when it would be inappropriate to construct a confidence interval for a proportion. Others embed it in a calculation stem, like "a random sample of 400 voters" before asking for a margin of error. On inference FRQs, write the condition check explicitly, something like "the problem states the data come from a random sample, and 400 voters is less than 10% of all voters." Graders want to see you connect the stated random sample to independence, not just list "random ✓."
Random sampling is how you SELECT individuals from a population; random assignment is how you SORT already-selected individuals into treatment groups. They unlock different conclusions. Random sampling lets you generalize results to the population. Random assignment lets you make cause-and-effect claims. The CED's independence condition accepts either one ("a random sample or a randomized experiment"), but the scope-of-conclusion question on the exam hinges on which one the study actually used.
A random sample gives every individual in the population an equal chance of being selected, which reduces selection bias and lets you generalize results to that population.
Random sampling is half of the independence condition for nearly every inference procedure in Units 6 through 9; the other half is the 10% condition when sampling without replacement.
On FRQs, explicitly cite the words "random sample" from the problem when verifying conditions, because graders award points for connecting the stated design to independence.
Random sampling supports generalization to a population, while random assignment in an experiment supports cause-and-effect conclusions; they are not interchangeable.
The chi-square test for independence requires a simple random sample, while the test for homogeneity requires a stratified random sample or randomized experiment, so the sampling design tells you which test to run.
A random sample still produces sampling error from sample to sample; that's not a flaw, it's the predictable variation that margins of error and p-values are built to measure.
A random sample is a subset of a population chosen by chance so that every individual has an equal probability of selection. In AP Stats it's the standard way to satisfy the independence condition for confidence intervals and significance tests in Units 6-9.
No. A random sample eliminates selection bias, but sampling error still makes every sample statistic differ from the true parameter. That's why a confidence interval either captures the parameter or it doesn't, and why we attach a margin of error instead of claiming exactness.
Random sampling selects who is in the study from the population, which lets you generalize findings. Random assignment sorts participants into treatment groups within an experiment, which lets you conclude causation. The CED's independence check accepts either, but the conclusions you can draw differ.
When you sample without replacement, each selection slightly changes the remaining population, so observations aren't perfectly independent. Verifying n ≤ 10% of N keeps that effect negligible. The CED lists random sample and the 10% check together as one independence condition.
No. A simple random sample draws from the whole population at once, while a stratified random sample randomly samples within predefined groups. The distinction matters in Unit 8, where a chi-square test for independence requires a simple random sample but a test for homogeneity requires a stratified random sample or randomized experiment.
Connect this key term to the AP exam workflow: review the course, practice questions, and check related study tools.
Review units, study guides, and course resources.
Check this vocabulary in multiple-choice context.
Apply key concepts in written AP responses.
Estimate the exam score you are working toward.
Review the highest-yield facts before practice.
Put the full course together before test day.