AP Statistics sampling distributions cheat sheet

By Jude Wallis · Published

The sampling distribution of a sample proportion has mean pp and standard deviation p(1−p)/n\sqrt{p(1-p)/n}. The sampling distribution of a sample mean has mean μ\mu and standard deviation σ/n\sigma/\sqrt{n}. The central limit theorem makes the sample mean approximately normal for large nn.

In the Fall 2026 AP Statistics course, sampling distributions are introduced in Unit 2 (topic 2.12) and their formulas are developed in Unit 3 for proportions (topics 3.2 and 3.9) and Unit 4 for means (topics 4.1 and 4.6).

Three distributions to keep apart

DistributionWhat it describesTypical symbols
Population distributionThe values of the variable for every individual in the populationμ\mu, σ\sigma, pp
Distribution of a sampleThe values actually collected in one samplexˉ\bar{x}, ss, p^\hat{p}
Sampling distributionThe values of a statistic over all possible samples of the same size from the populationμxˉ\mu_{\bar{x}}, σxˉ\sigma_{\bar{x}}, μp^\mu_{\hat{p}}, σp^\sigma_{\hat{p}}

A sampling distribution describes a statistic. Its standard deviation measures how much that statistic varies from sample to sample. See sample distribution vs sampling distribution.

Mean, standard deviation, and shape

StatisticMeanStandard deviationApproximately normal when
Sample proportion p^\hat{p}μp^=p\mu_{\hat{p}} = pσp^=p(1−p)n\sigma_{\hat{p}} = \sqrt{\frac{p(1-p)}{n}}np≥10np \geq 10 and n(1−p)≥10n(1-p) \geq 10
Sample mean xˉ\bar{x}μxˉ=μ\mu_{\bar{x}} = \muσxˉ=σn\sigma_{\bar{x}} = \frac{\sigma}{\sqrt{n}}The population is normal (any nn), or n≥30n \geq 30
Difference p^1−p^2\hat{p}_1 - \hat{p}_2p1−p2p_1 - p_2p1(1−p1)n1+p2(1−p2)n2\sqrt{\frac{p_1(1-p_1)}{n_1} + \frac{p_2(1-p_2)}{n_2}}n1p1n_1p_1, n1(1−p1)n_1(1-p_1), n2p2n_2p_2, n2(1−p2)n_2(1-p_2) all at least 10
Difference xˉ1−xˉ2\bar{x}_1 - \bar{x}_2μ1−μ2\mu_1 - \mu_2σ12n1+σ22n2\sqrt{\frac{\sigma_1^2}{n_1} + \frac{\sigma_2^2}{n_2}}Both populations are normal, or both n1≥30n_1 \geq 30 and n2≥30n_2 \geq 30

p^\hat{p} ("p-hat") is the sample proportion, xˉ\bar{x} ("x-bar") is the sample mean, pp and μ\mu are the population proportion and mean, and σ\sigma is the population standard deviation.

For a difference, the variances of the two statistics add under the square root. The standard deviations themselves are never added, and nothing is subtracted under the root.

Conditions

ConditionCheckWhat it justifies
RandomThe data come from a random sample (for two samples, independent random samples or a randomized experiment)The sampling distribution is centered at the parameter
10%When sampling without replacement, n≤0.10Nn \leq 0.10NThe standard deviation formula, which assumes independent observations
Large counts (proportions)np≥10np \geq 10 and n(1−p)≥10n(1-p) \geq 10The normal shape for p^\hat{p}
Normal population or large sample (means)The population is normal, or n≥30n \geq 30The normal shape for xˉ\bar{x}

The mean of the sampling distribution equals the parameter for any sample size. The 10% condition matters for the standard deviation, and the last two rows matter only for the shape. A population with extreme skew may need a sample much larger than 30.

The central limit theorem

The central limit theorem says that the sampling distribution of the mean of a random sample can be approximated by a normal distribution, and that the approximation improves as the sample size grows. It holds even when the population is not normal.

The theorem does sayThe theorem does not say
The distribution of xˉ\bar{x} over many samples becomes approximately normal as nn growsThe population becomes normal
The approximation is better for larger nnThe data in one large sample look normal
It applies to populations that are not normalA small sample from a strongly skewed population gives an approximately normal xˉ\bar{x}

If the population is normal, xˉ\bar{x} is normally distributed for every sample size and the theorem is not needed. For a proportion, use the large counts condition to justify a normal shape. See central limit theorem and the sampling distribution simulator.

What sample size changes

ChangeCenter of the sampling distributionStandard deviationShape
Larger nnNo changeSmaller, in proportion to 1n\frac{1}{\sqrt{n}}Closer to normal, if it was not already normal
nn multiplied by 4No changeCut in halfSame as above
nn multiplied by 9No changeCut to one thirdSame as above

Because nn is under a square root, cutting the standard deviation in half takes four times the sample size.

A statistic is an unbiased estimator when the mean of its sampling distribution equals the parameter. Both p^\hat{p} and xˉ\bar{x} are unbiased: μp^=p\mu_{\hat{p}} = p and μxˉ=μ\mu_{\bar{x}} = \mu. A larger sample lowers variability. It does not correct bias that comes from how the sample was collected. See bias vs variability.

Standard deviation versus standard error

The formulas above use the parameters pp and σ\sigma. In inference those are unknown, so sample values replace them and the result is called a standard error.

StatisticStandard deviation (parameters known)Standard error (estimated from the sample)
p^\hat{p}p(1−p)n\sqrt{\frac{p(1-p)}{n}}p^(1−p^)n\sqrt{\frac{\hat{p}(1-\hat{p})}{n}}
xˉ\bar{x}σn\frac{\sigma}{\sqrt{n}}sn\frac{s}{\sqrt{n}}

The standard deviation ss of a sample describes how individual values vary. The standard error sn\frac{s}{\sqrt{n}} describes how the sample mean varies from sample to sample, and it is smaller. See standard error vs standard deviation.

Sampling distribution of a sample proportion

In a large city, 30% of adults have a library card. A random sample of 200 adults is taken. Describe the sampling distribution of the sample proportion p^\hat{p} with a library card, and find the probability that p^\hat{p} is 0.35 or more.

  1. Mean: μp^=p=0.30\mu_{\hat{p}} = p = 0.30.

  2. Conditions for the standard deviation: the sample is random, and 200 adults are less than 10% of the adults in a large city.

  3. Standard deviation: σp^=0.30(0.70)200=0.00105=0.0324\sigma_{\hat{p}} = \sqrt{\frac{0.30(0.70)}{200}} = \sqrt{0.00105} = 0.0324.

  4. Shape: np=200(0.30)=60np = 200(0.30) = 60 and n(1−p)=200(0.70)=140n(1-p) = 200(0.70) = 140 are both at least 10, so p^\hat{p} is approximately normal.

  5. Standardize: z=0.35−0.300.0324=1.54z = \frac{0.35 - 0.30}{0.0324} = 1.54.

  6. Area to the right of z=1.54z = 1.54: 1−0.9382=0.06181 - 0.9382 = 0.0618.

p^\hat{p} is approximately normal with mean 0.30 and standard deviation 0.0324. P(p^≥0.35)≈0.062P(\hat{p} \geq 0.35) \approx 0.062 from the table. Technology with the unrounded z-score gives 0.0614.

Sampling distribution of a sample mean

The wait times at a large clinic have mean 50 minutes and standard deviation 12 minutes, and the distribution is skewed right. A random sample of 36 visits is taken from the many thousands of visits this year. Find the probability that the sample mean wait time is less than 47 minutes.

  1. Mean: μxˉ=μ=50\mu_{\bar{x}} = \mu = 50 minutes.

  2. Standard deviation: the sample is random and 36 is less than 10% of the visits, so σxˉ=1236=126=2\sigma_{\bar{x}} = \frac{12}{\sqrt{36}} = \frac{12}{6} = 2 minutes.

  3. Shape: the population is skewed, but n=36≥30n = 36 \geq 30, so xˉ\bar{x} is approximately normal by the central limit theorem.

  4. Standardize: z=47−502=−1.5z = \frac{47 - 50}{2} = -1.5.

  5. Area to the left of z=−1.5z = -1.5: 0.0668.

P(xˉ<47)≈0.0668P(\bar{x} < 47) \approx 0.0668. About 6.7% of random samples of 36 visits would have a mean wait time below 47 minutes.

Frequently asked questions

Does the central limit theorem apply to sample proportions?

The course states the central limit theorem for the mean of a random sample. For a sample proportion, the normal shape is justified by the large counts condition: np≥10np \geq 10 and n(1−p)≥10n(1-p) \geq 10.

Why does the standard deviation of the sample mean have a square root of n?

Averaging nn independent values divides the variance by nn, so the variance of xˉ\bar{x} is σ2n\frac{\sigma^2}{n}. Taking the square root gives σn\frac{\sigma}{\sqrt{n}}. That is why four times the sample size is needed to cut the standard deviation in half.

Is n at least 30 needed when the population is normal?

No. When the population distribution is normal, the sampling distribution of xˉ\bar{x} is normal for every sample size, including small ones.