Independent random samples in AP Statistics

In AP Statistics, independent random samples are two (or more) samples, each randomly selected from its own population, so that which individuals land in one sample has no effect on which land in the other. This is the randomness and independence condition behind two-sample z and t procedures and chi-square homogeneity tests.

What are independent random samples?

Independent random samples means you take separate random samples from separate populations, and the two selection processes don't talk to each other. Say you survey 120 adults from City A and 150 adults from City B. Who ends up in the City A sample tells you nothing about who ends up in the City B sample. That's independence between the samples.

There are really two ideas packed into the phrase. Random means each sample is chosen by a chance process (often a simple random sample), so you can generalize to each population. Independent means the samples are unrelated to each other. They aren't the same people measured twice, and they aren't twins split into two groups. Think of it as running two totally separate lotteries. The winners of one lottery have nothing to do with the winners of the other. When you can say that, you can use the two-sample formulas.

Why independent random samples matter in AP® Statistics

This term is the gatekeeper condition for comparing two populations. It shows up in proportions inference (Topic 3.10, confidence intervals for p1 minus p2, and Topic 3.12, setting up the two-proportion z-test). It shows up again in means inference (Topics 4.7, 4.9, and 4.10, the two-sample t-interval and t-test). It also comes back in Topic 3.14, where a chi-square test for homogeneity needs separate independent random samples from each population.

Here's why it matters mathematically. When two samples are independent, their variances add. That's the only reason the standard error for a difference is the square root of (variance of statistic 1 + variance of statistic 2). If the samples aren't independent, that formula is wrong and the whole procedure falls apart. So on the exam, stating this condition (and checking it from the context) is how you earn the conditions part of an inference FRQ.

How independent random samples connect across the course

10% Condition (Units 3-4)

Independent random samples gives you independence between the two samples. The 10% condition gives you independence within each sample, so sampling without replacement doesn't mess things up. For a two-sample procedure you need both, and you check the 10% condition separately for each population.

Standard Error of a Difference (Units 3-4)

Independence is why you add variances, never standard errors. That's where the familiar two-sample formula comes from, the square root of p1(1 - p1)/n1 + p2(1 - p2)/n2 for proportions or s1²/n1 + s2²/n2 for means. No independence, no adding.

Chi-Square Homogeneity vs. Independence (Unit 3)

The sampling design picks the test. Several independent random samples, one from each population, point to a test for homogeneity. One single random sample classified on two categorical variables points to a test for independence. The 2021 soft drink FRQ used three independent samples, which is the homogeneity setup.

Pooled Proportion and the Success/Failure Condition (Unit 3)

Once you have independent random samples from two populations, a two-proportion z-test pools the successes to estimate a common p under the null. You still check the success/failure condition in each sample separately, because each sample's sampling distribution needs to be approximately Normal on its own.

Are independent random samples on the AP® Statistics exam?

You'll see this phrase in two ways. Sometimes it's given to you in the stem ("Independent random samples of 120 adults from City A and 150 adults from City B..."). In that case, your job is to cite it as the randomness and independence condition and then check the rest, meaning the 10% condition and success/failure (or Normality) for each sample. Other times you have to recognize it from the design and choose the right procedure.

Multiple-choice questions often ask which conditions must be met for a two-proportion interval or test. They also give you a finished interval, like a 99% interval of (-0.12, 0.04) for a difference in proportions, and ask what it means. Because 0 is inside, there's no convincing evidence of a difference.

Released FRQs use the term directly. 2026 FRQ Q1 compares goat weights using independent random samples of 14 goats from each of two breeds, a two-sample t setup. With samples that small, you also need to address Normality (watch for skewness or outliers). 2021 FRQ Q5 uses three independent samples of teens and soft drink consumption, which leads to a chi-square test for homogeneity. On FRQs, write the condition in context ("the samples of Breed H and Breed J goats were selected independently and at random"). Just writing "random" doesn't earn the point.

Independent random samples vs Paired (matched pairs) data

With independent random samples, the two groups are separate individuals with no link between them, so you use two-sample procedures and add the variances. With paired data, each value in one group is naturally matched to one value in the other, like the same person before and after, or twins. Then you take the differences and run a one-sample procedure on them. Quick test: ask yourself whether you could line up observation #1 in group A with observation #1 in group B in a meaningful way. If yes, it's paired. If no, the samples are independent.

Key things to remember about independent random samples

• Independent random samples are separate random samples from separate populations, chosen so that selecting one sample has no effect on selecting the other.

• This condition is what allows you to add variances, which is the basis of the standard error formula for any difference between two proportions or two means.

• For two-sample procedures, you still check the 10% condition and the success/failure (or Normality) condition separately for each sample.

• If the data are naturally matched, like before-and-after measurements on the same subjects, the samples are not independent and you should use a paired procedure instead.

• In chi-square inference, separate independent random samples from several populations call for a test for homogeneity, while one random sample classified two ways calls for a test for independence.

• On FRQs, state the condition in the context of the problem by naming both populations, because a generic "the data are random" usually won't earn credit.

Frequently asked questions about independent random samples

What are independent random samples in AP Stats?

They are two or more samples, each randomly chosen from its own population, where the selection of one sample doesn't influence the other. For example, the 2026 FRQ Q1 took independent random samples of 14 goats from each of two breeds to compare weights.

Is before-and-after data on the same people independent random samples?

No. Measuring the same individuals twice creates paired data, because each "after" value is linked to a specific "before" value. You analyze the differences with a one-sample t procedure, not a two-sample t.

Does independent random samples mean I can skip the 10% condition?

No. Independent random samples covers independence between the two samples, but you still need the 10% condition within each sample. Check that each sample is less than 10% of its own population.

How are independent random samples different from independent groups in an experiment?

Independent random samples come from random sampling of separate populations, so you can generalize to those populations. Independent groups in an experiment usually come from random assignment of one set of subjects to treatments, which supports cause-and-effect conclusions. Both satisfy the independence condition for two-sample procedures.

Do independent random samples mean I should use a chi-square test for homogeneity or independence?

Use homogeneity when you have separate independent random samples from each population, like the three independent samples of teens in 2021 FRQ Q5. Use the test for independence when you have one random sample and classify each individual on two categorical variables.