A simple random sample (SRS) is a sampling method where every individual, and every possible group of n individuals, has an equal chance of being selected from the population. On the AP Stats exam, an SRS is the condition that lets you generalize from sample statistics to population parameters.
A simple random sample (SRS) is the gold-standard way to pick a sample. Every individual in the population has an equal chance of being chosen, and so does every possible group of n individuals. Think of it as putting every name in the population into a hat, mixing thoroughly, and drawing n names. No judgment calls, no volunteers, no convenience. Pure chance does the choosing.
Why is that the gold standard? Because random selection is the only thing that removes selection bias and makes the sample representative of the population on average. Every formula you use in Units 5-9, from the standard deviation of the sampling distribution (σ/√n or √(p(1-p)/n)) to chi-square tests, assumes the data came from random sampling. An SRS isn't just a vocabulary word. It's the foundation that makes statistical inference legitimate, which is why "data come from a random sample" shows up as a condition you must check in almost every inference procedure.
The SRS is the thread connecting Unit 5 (Sampling Distributions) and Unit 8 (Chi-Square Inference). In Topics 5.5 and 5.7, the formulas for the mean and standard deviation of sampling distributions (AP Stats 5.5.A and 5.7.A) only hold when samples are selected randomly, and the 10% condition kicks in because real SRSs sample without replacement. The Central Limit Theorem in Topic 5.3 (AP Stats 5.3.A) requires sample values to be independent of each other, and random sampling is how you get that independence. Then in Topic 8.5, the CED is explicit (AP Stats 8.5.C): a chi-square test for independence requires data collected using a simple random sample, while a test for homogeneity uses stratified random samples or a randomized experiment. That distinction alone is a classic exam question. If you can't justify why an SRS matters, you can't earn full credit on inference FRQs, because checking conditions is part of the scoring.
Keep studying AP Statistics Unit 5
Stratified Sampling (Unit 1, applied in Unit 8)
Stratified sampling splits the population into groups first, then takes an SRS within each group. The payoff shows up in Topic 8.5, where a chi-square test for homogeneity expects stratified samples or separate samples from each population, while a test for independence expects one single SRS. The sampling method literally tells you which chi-square test to run.
Sampling Distributions for Sample Proportions and Means (Unit 5)
The formulas μp̂ = p, σp̂ = √(p(1-p)/n), μx̄ = μ, and σx̄ = σ/√n all assume random sampling. The 10% condition (n ≤ 10% of N) exists because an SRS samples without replacement, which technically shrinks the standard deviation. Small sample relative to population means the difference is negligible.
Central Limit Theorem (Unit 5)
The CLT promises an approximately normal sampling distribution for large n, but it requires sample values to be independent of each other. An SRS (plus the 10% condition) is how you justify that independence. Without random sampling, the CLT's guarantee doesn't apply to your data.
Chi-Square Test for Independence (Unit 8)
The independence test asks whether two categorical variables are associated in one population, so it needs one SRS from that single population. If you see separate samples from several populations instead, that's a homogeneity setup. Spotting the sampling design is step one of setting up the test.
Simple random samples appear two ways on the exam. First, MCQs test whether you know when an SRS is required and what it guarantees, like "Why is a simple random sample important for a chi-square test of independence?" The answer is that it gives independent observations and a sample that represents the population, which justifies generalizing your conclusion. Second, FRQs (including the 2025 exam, Q4) hand you a study description and expect you to verify conditions before inference. You'll write something like "the problem states the data come from a simple random sample, so the random condition is met," then check the 10% condition and large counts. A common trap is a problem that describes stratified samples from multiple populations. That signals a homogeneity test, not an independence test. Misidentifying the design costs points even if your calculations are perfect.
An SRS draws one sample where every group of n individuals is equally likely, so the sample's makeup is left entirely to chance. A stratified sample divides the population into groups (strata) first, then takes an SRS within each stratum, guaranteeing representation from every group. They're not interchangeable on the exam. A stratified sample is NOT an SRS of the whole population, because not every combination of n individuals is possible (you can't end up with all individuals from one stratum). In Unit 8, this distinction decides your test. One SRS from one population means a chi-square test for independence, while stratified or separate samples from multiple populations means a test for homogeneity.
In a simple random sample, every individual AND every possible group of n individuals has an equal chance of being selected, not just every individual.
An SRS is the "random" condition behind nearly every inference procedure, because random selection removes selection bias and justifies generalizing to the population.
The sampling distribution formulas in Unit 5 (like σx̄ = σ/√n) assume random sampling, and the 10% condition handles the fact that an SRS samples without replacement.
For a chi-square test for independence, the CED requires data from a single simple random sample; for homogeneity, you need stratified random samples or a randomized experiment.
An SRS reduces bias but does not eliminate sampling error, since different random samples will still produce different statistics by chance.
On FRQs, explicitly state that the data come from a random sample when checking conditions. Skipping the conditions check costs points.
A simple random sample (SRS) is a sample chosen so that every individual, and every possible group of n individuals, has an equal chance of being selected from the population. It's the standard sampling method assumed by the inference formulas in Units 5-9.
No. An SRS eliminates selection bias, but sampling error (the natural sample-to-sample variability in a statistic) still exists. That variability is exactly what sampling distributions in Unit 5 describe, and increasing n shrinks it by a factor of √n.
An SRS picks n individuals from the whole population at once, so the sample's composition is entirely up to chance. A stratified sample splits the population into groups first and takes an SRS within each one, guaranteeing every group is represented. A stratified sample is not an SRS of the overall population.
Per the CED (AP Stats 8.5.C), the independence test examines whether two categorical variables are associated in one population, so the data must come from a single SRS of that population. If the data come from stratified samples of multiple populations, the correct test is homogeneity instead.
No. Random sampling (like an SRS) is how you select individuals from a population, and it lets you generalize results to that population. Random assignment is how you place subjects into treatment groups in an experiment, and it lets you make cause-and-effect conclusions. An experiment can have one, both, or neither.