The population mean, written μ, is the true average value of a quantitative variable for every individual in a population. It's a parameter (a fixed but usually unknown number), and most of AP Stats inference exists to estimate it or test claims about it using the sample mean, x̄.
The population mean (μ, pronounced "mu") is the average you'd get if you could measure every single member of a population. Mean blood pressure of all adults, mean amount of gold on every necklace a machine produces, mean calcium intake of all adolescents in a city. In real life you almost never get to measure everyone, so μ is a parameter: a fixed number that exists, but that you usually don't know.
That "fixed but unknown" status is the whole reason inference exists in AP Stats. You take a random sample, compute the sample mean x̄ (a statistic), and use it to estimate μ or test a claim about μ. Unit 5 tells you how x̄ behaves around μ (the sampling distribution of x̄ is centered at μ with standard deviation σ/√n, per LO 5.7.A). Unit 7 then builds the tools: a one-sample t-interval gives a range of plausible values for μ, and a one-sample t-test checks whether a hypothesized value μ₀ is believable. Every hypothesis you write in Unit 7 is a statement about μ, never about x̄, because you already know x̄ exactly. There's nothing to test about a number you computed.
The population mean is the central parameter of Unit 5 (Sampling Distributions) and Unit 7 (Inference for Means), and it echoes into Units 8 and 9. In Topic 5.7, LO 5.7.A says the sampling distribution of x̄ has mean μx̄ = μ and standard deviation σ/√n, which is the mathematical guarantee that sample means cluster around the population mean. Topic 5.8 extends this to differences in means (μ₁ - μ₂). In Unit 7, LOs 7.2.A through 7.5.C are all built on μ. You construct x̄ ± t*(s/√n) to estimate μ, write hypotheses like H₀: μ = μ₀, and interpret p-values "assuming the true population mean equals the value in the null." Even matched pairs reduce to inference about a single population mean of differences, μd. If you can't keep μ (parameter) and x̄ (statistic) straight, you lose points on hypothesis statements, interval interpretations, and conclusion sentences across half the exam.
Keep studying AP Statistics Unit 8
Sample Mean, x̄ (Units 1, 5, 7)
The sample mean is your best single guess for the population mean. x̄ is the point estimate, μ is the target. The CED makes this exact in LO 5.7.A, where the sampling distribution of x̄ is centered at μ, meaning x̄ is an unbiased estimator of the population mean.
Sampling Distributions for Sample Means (Unit 5)
Topic 5.7 describes how x̄ dances around μ from sample to sample. The center of that dance is exactly μ, and its spread is σ/√n. This is also where the Central Limit Theorem kicks in, letting you use a normal model for x̄ when n ≥ 30 even if the population isn't normal (LO 5.7.B).
t-Distributions and t-Procedures (Unit 7)
Because you almost never know the population standard deviation σ when you don't know μ, you substitute s and use a t-distribution with n-1 degrees of freedom (LO 7.2.A). Every confidence interval and significance test for a population mean on the exam runs through t, not z.
Population Regression Slope, β (Unit 9)
The parameter-vs-statistic logic of μ and x̄ repeats with regression. The sample slope b estimates the unknown population slope β, just like x̄ estimates μ, and the interval b ± t*(SEb) in Topic 9.2 has the exact same structure as x̄ ± t*(s/√n).
Multiple choice loves to hand you μ and σ and ask about the sampling distribution of x̄, like the question giving a population mean calcium intake of 800 mg with σ = 120 mg for samples of 36, or asking which statement about a sample drawn from a population with μ = 50 and σ = 15 is false. Know μx̄ = μ and σx̄ = σ/√n cold, and know when the normal model applies. On FRQs, population means show up constantly in inference problems. The 2023 FRQ about gold coating on necklaces involved a normally distributed amount of gold, and the 2021 walking-and-cholesterol study and 2023 omega-3 matched-pairs study both required hypotheses and conclusions about a population mean (or mean difference μd). The graders want three specific things from you. Write hypotheses about μ, not x̄. Interpret a confidence interval as capturing the population mean, with context ("we are 95% confident the interval captures the true mean cholesterol reduction for adults in this population"). And interpret the p-value as computed assuming the true population mean equals the null value.
The population mean μ is a fixed parameter describing everyone in the population; the sample mean x̄ is a statistic computed from one sample, and it changes from sample to sample. You never write hypotheses about x̄ (you already know it exactly), and a confidence interval is for μ, not for x̄. A quick mental check helps. If you could recalculate it by taking a new sample, it's x̄. If it's the unknown truth you're chasing, it's μ.
The population mean μ is a parameter, a fixed but usually unknown true average for the entire population, while the sample mean x̄ is a statistic that varies from sample to sample.
The sampling distribution of x̄ is centered exactly at μ with standard deviation σ/√n, which is why x̄ is an unbiased estimator of the population mean.
Hypotheses are always written about μ (H₀: μ = μ₀), never about x̄, because the sample mean is known and the population mean is what's in question.
Since σ is almost never known when μ is unknown, inference for a population mean uses t-procedures with n-1 degrees of freedom, giving the interval x̄ ± t*(s/√n).
A confidence interval interpretation must say you are C% confident the interval captures the true population mean, with units and context, not that μ has a C% probability of being in the interval.
Matched-pairs problems reduce to inference about a single population mean of differences, μd, so define your order of subtraction and proceed like a one-sample problem.
The population mean, μ, is the true average of a quantitative variable across every individual in a population. It's a parameter, so it's fixed but usually unknown, and you estimate it with the sample mean x̄ using t-intervals and t-tests in Unit 7.
μ describes the whole population and doesn't change; x̄ comes from one sample and varies every time you sample. On the exam, hypotheses and confidence intervals are about μ, and writing H₀: x̄ = 50 instead of H₀: μ = 50 will cost you points.
Yes. When you sample randomly, the mean of the sampling distribution of x̄ equals μ exactly (μx̄ = μ), no matter the sample size. What sample size affects is the spread, which shrinks to σ/√n.
Use a one-sample t-test. The CED is explicit that since σ is typically unknown for quantitative variables, you substitute the sample standard deviation s, which makes the test statistic follow a t-distribution with n-1 degrees of freedom.
No. Each interval either contains μ or it doesn't, because μ is a fixed number. The 95% refers to the method, meaning about 95% of intervals built this way from repeated random samples would capture the population mean.