Population means (μ₁ and μ₂) are the true average values of a quantitative variable in two complete populations. On the AP Stats exam, you estimate them (and especially their difference, μ₁ - μ₂) using sample means, sampling distributions, and two-sample t confidence intervals.
A population mean, written μ (mu), is the actual average of a variable across an entire population. It's a parameter, a fixed number you almost never get to see directly because measuring everyone is usually impossible. When a problem talks about "population means" (plural), you're working with two populations at once, with means μ₁ and μ₂, and the real question is almost always about their difference, μ₁ - μ₂.
Here's the core logic of AP Stats in one sentence. You can't observe μ₁ - μ₂, so you estimate it with the difference in sample means, x̄₁ - x̄₂, and use the sampling distribution of that statistic to say how good your estimate is. Per the CED, the sampling distribution of x̄₁ - x̄₂ has mean μ₁ - μ₂ and standard deviation √(σ₁²/n₁ + σ₂²/n₂), and it's approximately normal if both populations are normal or both sample sizes are at least 30. Everything in Topics 5.8, 7.6, and 7.7 builds on that setup.
Population means anchor two big chunks of the course. In Unit 5 (Topic 5.8), learning objectives 5.8.A through 5.8.C have you find the mean and standard deviation of the sampling distribution of x̄₁ - x̄₂, decide whether it's approximately normal, and interpret probabilities in context. In Unit 7 (Topics 7.6 and 7.7), objectives 7.6.A-7.6.D and 7.7.A-7.7.C have you build and interpret a two-sample t-interval for μ₁ - μ₂ and use it to justify a claim. The whole point of inference for means is recovering information about unknown population means from sample data, so if you're fuzzy on what μ actually is, every interpretation sentence you write will be slightly wrong. Graders specifically look for language like "the true difference in population means" rather than "the difference in sample means."
Keep studying AP Statistics Unit 7
Sample Mean (Units 1, 5, 7)
The sample mean x̄ is your window into μ. The population mean is the fixed target; the sample mean is the estimate that bounces around from sample to sample. Confusing these two is the single most common interpretation error in Unit 7.
Sampling Distribution (Unit 5)
The sampling distribution of x̄₁ - x̄₂ is centered exactly at μ₁ - μ₂. That's not a coincidence, it's why x̄₁ - x̄₂ is an unbiased estimator and why the whole inference machinery in Unit 7 works.
Difference in Two Population Means (Units 5, 7)
Two-sample problems almost never ask about μ₁ and μ₂ separately. They ask about μ₁ - μ₂, because one number answers the real question, which group's average is bigger and by how much.
Confidence Interval (Units 6, 7)
A two-sample t-interval gives a range of plausible values for μ₁ - μ₂. The big payoff is whether 0 is in the interval. If 0 is excluded, you have convincing evidence the population means actually differ.
Multiple-choice questions give you population standard deviations and sample sizes and ask for the standard deviation of x̄₁ - x̄₂, or give you μ₁ - μ₂ and ask for a probability using the normal model (like finding the chance that x̄₁ - x̄₂ exceeds some value when μ₁ - μ₂ = 8). Another classic MCQ hands you a confidence interval for μ₁ - μ₂, say (3.2, 9.8), and asks which conclusion is valid. On FRQs, two-sample mean comparisons show up regularly. The 2022 exam (Q5) compared blood pressure reduction between two groups, and the 2025 exam (Q6) compared reading comprehension by time of day. You're expected to define the parameters in context ("let μ₁ be the true mean reduction for..."), check independence and normality conditions, compute or interpret the interval, and justify a claim by checking whether 0 falls inside it. Defining μ₁ and μ₂ clearly at the start is an easy point that lots of people leave on the table.
The population mean μ is a parameter, a fixed but unknown number describing the entire population. The sample mean x̄ is a statistic, a value calculated from your data that changes with every sample. You never "calculate" μ from data; you estimate it with x̄. On interpretations, saying you're 95% confident about the difference in sample means is wrong (you know that difference exactly). The confidence is about the difference in population means.
A population mean (μ) is a fixed, unknown parameter, while a sample mean (x̄) is the statistic you actually calculate and use to estimate it.
The sampling distribution of x̄₁ - x̄₂ has mean μ₁ - μ₂ and standard deviation √(σ₁²/n₁ + σ₂²/n₂), so the difference in sample means is an unbiased estimator of the difference in population means.
That sampling distribution is approximately normal if both populations are normal or both sample sizes are at least 30.
Since σ₁ and σ₂ are almost never known, you use s₁ and s₂ and a t-distribution, giving the two-sample t-interval (x̄₁ - x̄₂) ± t*√(s₁²/n₁ + s₂²/n₂).
If 0 is not inside a confidence interval for μ₁ - μ₂, you have convincing evidence that the two population means differ.
Increasing the sample sizes shrinks the standard error, which makes the confidence interval for the difference in population means narrower.
It's the true average value of a quantitative variable across an entire population, written μ. It's a parameter, meaning it's fixed but usually unknown, so you estimate it using the sample mean x̄ from a random sample.
The population mean μ describes everyone in the population and never changes; the sample mean x̄ comes from your data and varies from sample to sample. Inference works in one direction, using the known x̄ to estimate the unknown μ.
No, and on the AP exam you almost never will. When population standard deviations are unknown, you substitute the sample standard deviations s₁ and s₂ and use a t-distribution instead of z, which is why two-sample mean problems use t* critical values.
It doesn't prove anything, but it does give convincing statistical evidence. Since every value in the interval is positive and 0 is excluded, plausible values for μ₁ - μ₂ are all greater than zero, supporting the claim that μ₁ is larger than μ₂.
When both population distributions are normal, or when both sample sizes are at least 30 (the central limit theorem condition). You have to verify one of these before building a two-sample t-interval on an FRQ, or you lose the conditions point.