What is the difference between a sampling distribution and a sample distribution?

A sample distribution shows the individual data values from one single sample you collected. A sampling distribution shows the values of a statistic (like x or p) across all possible samples of a fixed size. One is made of raw data points; the other is made of statistics.

How do I tell whether a number is a parameter or a statistic?

Ask whether the number describes the entire population or just a sample. Words like "all," "every," or "the population" signal a parameter (e.g., , p). A number computed from a survey or sample of individuals is a statistic (e.g., x, p).

Does a bigger sample size reduce bias?

No. Larger samples reduce the variability of a statistic, making its sampling distribution narrower, but they do not fix bias. Bias comes from the sampling or measurement method, so it can only be reduced by improving how the sample is chosen or measured.

Why does sampling variability matter for the rest of AP Statistics?

Everything in inference — confidence intervals and significance tests — relies on knowing how much a statistic varies from sample to sample. The sampling distribution quantifies that variation, which is why this concept is the foundation for the Central Limit Theorem and the specific distributions of p and x you study next.

AP-STATS-5.1-5.4

U5.1 The Sampling Distribution

Master parameters vs. statistics, sampling variability, sampling distributions, bias, and variability for AP Statistics Unit 5 (topics 5.1-5.4).

What you'll do in this lesson

A voice-first session with the Crimsora tutor on U5.1 The Sampling Distribution, then targeted practice and FRQs — with the tutor adapting to where you get stuck.

Start U5.1 The Sampling Distribution free

What this lesson covers

Every time you take a sample and compute a mean or proportion, you get a slightly different answer. Why? Because a statistic is a random quantity — it changes from sample to sample. This lesson builds the single most important idea in inferential statistics: the sampling distribution, the pattern of values a statistic takes over all possible samples.

Before you can trust a confidence interval or a p-value, you must understand where those tools come from. Here you'll learn to tell a parameter from a statistic, describe why sampling variability is unavoidable, define the sampling distribution precisely, and judge estimators by two criteria the AP exam loves to test: bias and variability.

Parameters vs. Statistics

A parameter is a fixed number that describes a population. A statistic is a number computed from a sample used to estimate that parameter. The parameter is usually unknown and constant; the statistic is known once you have data, but it varies depending on which sample you happened to draw.

Getting the notation right is worth easy points. Population values use Greek or capital letters; sample values usually use Latin letters or hats.

Quantity	Parameter (population)	Statistic (sample)
Mean	$\mu$	$\bar{x}$
Proportion	$p$	$\hat{p}$
Standard deviation	$\sigma$	$s$

A common misconception is that the sample statistic equals the parameter. It almost never does exactly — it is an estimate. The whole point of Unit 5 is to describe, mathematically, how far off that estimate tends to be.

When a question describes a number, ask: does this describe everyone in the population, or just the people who were sampled? "The mean height of all 2,400 students is 66 inches" is a parameter. "In a survey of 50 students, the mean height was 65.2 inches" is a statistic. The word "all" signals a parameter; a described sample signals a statistic.

Sampling Variability

Sampling variability is the fact that a statistic changes from one random sample to the next, even when the population stays the same. If you drew ten different random samples of size 50 from the same school and computed ten sample means, you would get ten different numbers clustered around — but rarely equal to — the true population mean.

This variability is not an error or a mistake. It is a natural, predictable consequence of random sampling. Because it is predictable, statisticians can model it, and that model is exactly the sampling distribution.

Two factors control how much a statistic varies. First, sample size: larger samples produce statistics that vary less, because extreme individual values get averaged out. Second, population variability: a more spread-out population produces more spread-out statistics.

Students often confuse three distinct distributions. The population distribution describes all individuals. The distribution of a single sample describes the data points you actually collected. The sampling distribution describes a statistic over all possible samples of a fixed size. The AP exam frequently tests whether you can keep these three straight — read carefully whether a question asks about individuals or about a statistic.

Defining the Sampling Distribution

The sampling distribution of a statistic is the distribution of values that the statistic takes over all possible samples of the same size

n

from the same population. You can imagine repeating the sampling process endlessly, recording the statistic each time, and graphing all the results.

Like any distribution, a sampling distribution has a shape, center, and spread. The center tells you what value the statistic estimates on average. The spread tells you how much the estimate typically bounces around. The shape tells you which values are likely versus rare — and later topics (the Central Limit Theorem) describe when that shape becomes approximately normal.

A key clarification: a sampling distribution is theoretical. You do not usually build it from real repeated sampling. Instead, statistical theory gives you its mean and standard deviation directly. What you can do in class is approximate it with a simulation, taking many random samples and plotting the resulting statistics — the histogram of those results approximates the true sampling distribution.

On the exam, expect to describe a sampling distribution using the SOCS framework applied to a statistic rather than raw data: shape, outliers/unusual features, center, and spread. Always state clearly that each dot or bar represents one sample's statistic, not one individual observation.

Bias and Variability of Estimators

We judge a statistic as an estimator using two independent criteria. An estimator is unbiased if the mean of its sampling distribution equals the true parameter — on average, it hits the target. It has low variability if the values in its sampling distribution are tightly clustered — repeated estimates agree with one another.

Think of a dartboard. Bias is whether your darts center on the bullseye; variability is how tightly grouped they are. These are separate: an estimator can be unbiased but highly variable, or precise but consistently off-target.

Estimator	Bias	Variability	Quality
Centered on parameter, tight	Unbiased	Low	Best
Centered on parameter, spread	Unbiased	High	Imprecise
Off-center, tight	Biased	Low	Consistently wrong

The sample mean

\bar{x}

and the sample proportion

\hat{p}

are both unbiased estimators of

\mu

and

p

. Importantly, bias does not depend on sample size — a biased sampling method (like a voluntary-response survey) stays biased no matter how large

n

is. Variability, however, decreases as $n$ increases. A frequent AP trap: increasing sample size fixes variability but never fixes bias. Only a better sampling design can reduce bias.

Key terms

Parameter.: A fixed numerical value describing a population, such as $\mu$ or $p$ , usually unknown.
Statistic.: A numerical value computed from a sample, such as $\bar{x}$ or $\hat{p}$ , used to estimate a parameter.
Sampling variability.: The natural variation in a statistic from one random sample to another drawn from the same population.
Sampling distribution.: The distribution of a statistic's values over all possible samples of a fixed size $n$ from a population.
Unbiased estimator.: A statistic whose sampling distribution has a mean equal to the parameter it estimates.
Bias.: A systematic tendency for a statistic to over- or underestimate the parameter; not reduced by larger samples.
Variability of an estimator.: How spread out a statistic's sampling distribution is; decreases as sample size increases.

Worked example

A city has 40,000 registered voters, and exactly 55% support a ballot measure. A polling firm plans to survey random samples of 200 voters and record the sample proportion

\hat{p}

who support the measure. (a) Identify the parameter and the statistic. (b) Explain what the sampling distribution of

\hat{p}

represents. (c) Is

\hat{p}

an unbiased estimator here, and what happens to its variability if the firm surveys 800 voters instead?

Part (a): The parameter is the true population proportion of all 40,000 voters who support the measure,

p = 0.55

. This value is fixed. The statistic is

\hat{p}

, the proportion supporting the measure within any one sample of 200 voters. Its value changes from sample to sample.

Part (b): The sampling distribution of

\hat{p}

is the distribution of all the

\hat{p}

values we would get if we took every possible random sample of size 200 and recorded each sample proportion. If we graphed thousands of such

\hat{p}

values, they would cluster around 0.55, with some samples slightly above and some slightly below due to sampling variability.

Part (c): Because the samples are random,

\hat{p}

is an unbiased estimator of

p

— the mean of its sampling distribution equals 0.55. If the firm increases the sample size from 200 to 800, the bias stays zero (it was already unbiased), but the variability decreases: the sampling distribution becomes narrower, so individual sample proportions land closer to 0.55 on average. Larger

n

improves precision, not accuracy.

Practice questions

A researcher takes many random samples of size 60 from a large population and computes the sample mean each time. Which statement best describes the resulting collection of sample means?

It is the population distribution, showing every individual value
It is a single sample distribution, showing the 60 collected data points
It approximates the sampling distribution of the sample mean
It shows the bias of the sampling method

Answer: It approximates the sampling distribution of the sample mean

Each sample produces one

\bar{x}

. Collecting many of these means and graphing them approximates the sampling distribution of

\bar{x}

. The population distribution describes individuals, and a single sample distribution shows one sample's 60 data points — neither matches a graph made of many sample means.

An estimator is described as biased. A student claims that taking a much larger sample will remove the bias. Explain whether this claim is correct.

Answer: The claim is incorrect; increasing sample size reduces variability but does not remove bias.

Bias is a systematic error built into the sampling or estimation method — for example, a poorly worded question or a nonrandom selection process. Increasing

n

makes the sampling distribution narrower (less variability), so estimates become more consistent, but they still center on the wrong value. Only fixing the sampling method itself can eliminate bias.

A polling agency reports that 62% of a random sample of 500 adults own a pet, while the true percentage of all adults in the region who own a pet is 60%. Identify which value is a parameter and which is a statistic, using correct notation.

Answer: The statistic is $\hat{p} = 0.62$ from the sample; the parameter is $p = 0.60$ for the whole population.

The 62% comes from the sample of 500 adults, so it is a statistic, written

\hat{p} = 0.62

. The 60% describes all adults in the region, so it is the population parameter

p = 0.60

. The difference between them, 2 percentage points, reflects sampling variability.

FAQ

What is the difference between a sampling distribution and a sample distribution?: A sample distribution shows the individual data values from one single sample you collected. A sampling distribution shows the values of a statistic (like $\bar{x}$ or $\hat{p}$ ) across all possible samples of a fixed size. One is made of raw data points; the other is made of statistics.
How do I tell whether a number is a parameter or a statistic?: Ask whether the number describes the entire population or just a sample. Words like "all," "every," or "the population" signal a parameter (e.g., $\mu$ , $p$ ). A number computed from a survey or sample of individuals is a statistic (e.g., $\bar{x}$ , $\hat{p}$ ).
Does a bigger sample size reduce bias?: No. Larger samples reduce the variability of a statistic, making its sampling distribution narrower, but they do not fix bias. Bias comes from the sampling or measurement method, so it can only be reduced by improving how the sample is chosen or measured.
Why does sampling variability matter for the rest of AP Statistics?: Everything in inference — confidence intervals and significance tests — relies on knowing how much a statistic varies from sample to sample. The sampling distribution quantifies that variation, which is why this concept is the foundation for the Central Limit Theorem and the specific distributions of $\hat{p}$ and $\bar{x}$ you study next.

Learn this with a teacher, not a page

The Crimsora tutor teaches U5.1 The Sampling Distribution live — explaining on a whiteboard, asking you questions, and adapting to where you get stuck.

Try a live session free