Not sure which lesson to do first? Find your level and get a plan →

Statistics · Grade 9-12

Sampling Distribution

The sampling distribution is the probability distribution of a statistic (such as the sample mean xˉ) computed from all possible random samples of a given size n drawn from a population.


Quick Answer

The sampling distribution is the probability distribution of a statistic (such as the sample mean xˉ) computed from all possible random samples of a given size n drawn from a population. It describes how that statistic varies from sample to sample. Use it to judge whether a sample result is ordinary or surprising: sample means pile up around the population mean μ, with a spread of σn called the standard error. For heights with σ=4 inches, the means of samples of 50 people have a spread of only about 0.57 inches.

Why This Matters

The sampling distribution is what turns one sample into a reasoned claim about a population. It tells you how far a sample mean typically falls from μ, which is the basis of every margin of error, confidence interval and hypothesis test. Without it, you cannot say whether a sample of batteries averaging 485 hours is bad luck or a bad batch.

Intuitive Explanation

Take a tiny population: four cards numbered 2, 4, 6 and 8, with mean 5. Draw two cards and average them. There are only six possible pairs, and their averages are 3, 4, 5, 5, 6 and 7. That list, with each value's chance (5 has probability 26, the others 16 each), is the sampling distribution of the sample mean for samples of size 2. It is centred at 5, the population mean, and it is narrower than the population: the means run from 3 to 7, while the cards run from 2 to 8.

Real populations are too large to list every sample, but the same pattern holds. Imagine drawing thousands of random samples of 50 adults and recording each sample's mean height. Individual heights have a standard deviation of about 4 inches, yet the 50-person means cluster around the population mean with a standard deviation of 450≈0.57 inches. Averaging cancels tall and short people within each sample, and larger samples cancel more, so the sampling distribution narrows as n grows.

Three distributions are easy to mix up. The population distribution is every individual's value, with spread σ. The distribution of one sample's data is the 50 heights you measured, a small, rough copy of the population with spread close to σ. The sampling distribution is made of statistics, one mean per sample, with spread σn. Only the third shrinks when you take bigger samples; the population's own spread never changes.

Think of it like this. A sampling distribution is like a record of a dart player's averages: single darts land all over the board, but the average position of each set of 10 darts stays close to where she aims. The more darts in each set, the tighter those averages cluster. The board, like the population, never changes; only the averages tighten.

Core idea. A statistic has its own distribution across all possible samples of size n: for the sample mean, it is centred on μ with spread σn, so you can tell how far one sample's result is likely to fall from the truth.


When to Use

Use a sampling distribution when a question asks how likely a sample result is, whether a sample statistic is unusual, or how much a statistic would vary across samples of a given size. Find its centre (the parameter, such as μ) and its spread (σn for a mean), then see where the observed statistic falls. If the question is about one individual's value, use the population distribution and σ; if it only notes that two samples differ, the idea of sampling variability is enough.

Tip. Ask: Is the question about how a statistic, such as a sample mean, would be spread across all possible samples of one size?

How to Recognize It

Before choosing this idea, check whether the values being described are single measurements or summaries of whole samples.

  1. Is each value in the distribution a statistic from a whole sample, such as one sample's mean?

    Yes is a sampling distribution. If each value is one person's height or one roll of a die, it is the population or data distribution, whose spread is the Standard Deviation.

  2. Does the problem fix a sample size, such as samples of 50?

    Yes fits, since the sampling distribution changes with n. If no sample size appears, the question is probably about individuals.

  3. Do you need the chance that a sample mean lands in some range, such as above 69 inches?

    Yes means using the sampling distribution's center and spread. Whether its shape is close to normal for large samples is the Central Limit Theorem's job.

  4. Is just one number for the spread asked for?

    Then the answer is the Standard Error, the standard deviation of this distribution, rather than the whole distribution.

Sampling Distribution vs Standard Deviation vs Central Limit Theorem vs Standard Error

These are all about spread and shape, which is why they blur. Ask whether the values are individuals or sample statistics, and whether you want a whole distribution, a spread number, or a rule about shape.

Sampling Distribution

Meaning
A sampling distribution is the distribution of a statistic, such as xˉ, over all possible random samples of one size.
Key test
Each value in it summarizes a whole sample.
Formula
μxˉ=μ,σxˉ=σn
Example
Means of random groups of 16 adults center on 67 inches with a spread of 1 inch.

Standard Deviation

Meaning
Standard deviation measures how far individual values typically sit from their mean.
Key test
Spread of single values, with no sample size involved.
Formula
σ=∑(x−μ)2N
Example
Single adult heights spread about 4 inches around 67 inches.

Central Limit Theorem

Meaning
The Central Limit Theorem says the sampling distribution of xˉ is close to normal when samples are large enough.
Key test
A rule about the shape of the sampling distribution.
Formula
Xˉ−μσ/n≈N(0,1)
Example
Means of 40 skewed wait times still form a near-bell shape.

Standard Error

Meaning
The standard error is the standard deviation of a sampling distribution.
Key test
One number: the spread of the sampling distribution.
Formula
SE=σn
Example
416=1 inch for averages of 16 heights.

Formula & Notation

How to read it: The sample mean is xˉ (or Xˉ as a random variable). Its standard deviation is called the standard error, SE=σn.

Worked Examples

Example 1 — Every sample from a tiny population

Easy

A population has four values: 2, 4, 6 and 8. List the sampling distribution of the sample mean for random samples of size 2 drawn without replacement. Find its centre and the probability that xˉ=5.
Show the steps and the answer
  1. Population mean: μ=2+4+6+84=5.

    The parameter is what the sample means will be compared with.

  2. List all (42)=6 samples and their means: (2, 4) gives 3, (2, 6) gives 4, (2, 8) gives 5, (4, 6) gives 5, (4, 8) gives 6, (6, 8) gives 7.

    A sampling distribution covers every possible sample of the given size, each equally likely under random sampling.

  3. Centre: 3+4+5+5+6+76=306=5.

    The mean of the sampling distribution equals μ, which is why the sample mean is an unbiased estimate.

  4. Probabilities: P(xˉ=5)=26=13; each of 3, 4, 6 and 7 has probability 16.

    Two of the six equally likely samples average 5, so the middle value is the most likely.

Answer: The sample mean takes the values 3, 4, 5, 6, 7 with probabilities 16,16,13,16,16. It is centred at 5, and P(xˉ=5)=13.

Takeaway: A sampling distribution lists a statistic's value for every possible sample; it centres on the parameter and is narrower than the population.

Example 2 — Which distribution is which?

Standard

Adult heights in a population have mean 67 inches and standard deviation 4 inches. One random sample of 50 adults has heights from 58 to 76 inches, a standard deviation of 3.9 inches, and a mean of 66.6 inches. A computer then draws 1,000 such samples and records each mean. Name the three distributions here and give each one's spread.
Show the steps and the answer
  1. Population distribution: every adult's height, spread σ=4 inches.

    This describes individuals and does not depend on any sample.

  2. Distribution of one sample's data: the 50 measured heights, spread 3.9 inches.

    A random sample of individuals looks like a rough copy of the population, so its spread is close to σ.

  3. Sampling distribution of xˉ: the 1,000 sample means, spread about 450=47.07≈0.57 inches.

    Each value is a mean of 50 heights, and averaging cancels much of the individual variation.

  4. Place the observed mean: 66.6 is 67−66.60.57≈0.7 standard errors below 67.

    Judged against the sampling distribution, not the population, 66.6 is an ordinary sample mean.

Answer: Population: spread 4 in. One sample's data: spread 3.9 in. Sampling distribution of the mean: spread about 0.57 in, against which the sample mean of 66.6 is unremarkable.

Takeaway: Only the sampling distribution is made of statistics, and only its spread, σn, shrinks with sample size.

Example 3 — Find the mistake: using the wrong spread

Application

Batteries from a factory last a mean of 500 hours with a standard deviation of 30 hours. An inspector tests a random sample of 36 batteries from a new batch and finds a mean of 485 hours. Jo writes: "The spread is 30 hours, so 485 is only half a standard deviation below 500. Nothing unusual." Find the mistake and correct it.
Show the steps and the answer
  1. Identify Jo's spread: 30 hours is the spread of individual batteries.

    Jo is judging a sample mean, so he needs the spread of the sampling distribution of xˉ, not the population's spread.

  2. Compute the standard error: 3036=306=5 hours.

    Means of 36 batteries vary six times less than single batteries do.

  3. Locate 485: 500−4855=3 standard errors below 500.

    With n=36 the sampling distribution is close to normal, and fewer than 1 in 500 samples would fall that low if the batch mean were really 500.

Answer: The sample mean is 3 standard errors below 500, not half a standard deviation, so 485 hours is very unusual and suggests the new batch's mean lifetime is below 500 hours.

Takeaway: Judge a sample mean against σn, the spread of the sampling distribution, never against σ itself.

Where You'll See It

  • Election polls of 1,000 voters

    If many firms each polled a random sample of 1,000 voters in a state where 52% back a candidate, their results would form a sampling distribution centred on 52% with a standard deviation of about 1.6 points. That spread is why a single poll of 1,000 reports a margin of error of about 3 points.

  • Weighing samples of cereal boxes

    A machine fills boxes with a mean of 500 g and a standard deviation of 8 g. Means of random samples of 16 boxes have a standard deviation of 816=2 g, so a sample mean of 494 g is 3 standard errors low, a signal that the machine needs adjusting.

  • Rolling five dice and averaging

    Single rolls spread evenly from 1 to 6 with a standard deviation of about 1.71. Averages of five rolls pile up near 3.5 with a standard deviation of about 1.715≈0.76, so the sampling distribution is much narrower than the population it came from.

Common Mistakes

  1. The slip: The sampling distribution is the same as the distribution of the data in one sample.

    Why it's tempting: Both are histograms you can draw, so they feel interchangeable, and a single sample is the only thing you actually collected.

    Instead: A sampling distribution is the distribution of a statistic (like xˉ) computed across many samples, not the distribution of individual values within one sample.

  2. The slip: Taking a bigger sample makes the population itself less variable.

    Why it's tempting: Spread does shrink as n grows, so it feels like the underlying spread changed.

    Instead: The population's variability σ is fixed; only the spread of the sampling distribution of the mean shrinks, by a factor of n.

  3. The slip: If a random sample's mean does not equal the population mean, something went wrong with the sample.

    Why it's tempting: The sampling distribution is centered at μ, so it seems every sample mean should land on μ.

    Instead: Centered at μ means the sample means average out to μ, not that each one hits it. With the cards 2, 4, 6 and 8 (μ=5), only 2 of the 6 samples of size 2 have mean 5; the others give 3, 4, 6 or 7, and that spread is ordinary sample-to-sample variation.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. A population has three values: 2, 4, 6. List the means of all 9 samples of size 2 drawn with replacement. What is the mean of these sample means?

    Hint: Write out every ordered pair, such as (2, 4) and (4, 2), and average each pair.

  2. Test scores have population mean 70 and standard deviation 12. For random samples of 36 students, find the mean and standard deviation of the sampling distribution of xˉ.

    Hint: The center stays the same; the spread is divided by the square root of the sample size.

  3. In a town, 30% of adults bike to work. For random samples of 84 adults, find the center and standard deviation of the sampling distribution of p^.

    Hint: Use p(1−p)/n for the spread.

  4. Cereal box weights are normal with mean 500 g and standard deviation 8 g. Random samples of 16 boxes are weighed. About what percent of samples have a mean weight above 504 g?

    Hint: Find the spread of the sample mean first, then how many spreads 504 is above 500.

  5. Histogram 1 shows the heights of all 2,000 students at a school. Histogram 2 shows the mean heights of 500 random samples of 25 students. Which is a sampling distribution, and which should be narrower?

    Hint: Which histogram plots one value per person, and which plots one value per sample?

  6. Commute times in a city have standard deviation 18 minutes. A student writes: "Means of random samples of 36 commuters will also have standard deviation 18 minutes." Correct the mistake.

    Hint: Does averaging 36 commutes make the result more or less variable than a single commute?

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Do sampling distributions exist for statistics other than the mean?

Yes. Any number you compute from a sample has one: a proportion, a median, a maximum, the gap between two group means, even a correlation. For a sample proportion, the center is the true proportion pp and the spread is p(1−p)n\sqrt{\frac{p(1-p)}{n}}. Some statistics, like the median, have no simple spread formula, so their sampling distributions are often found by simulation.

How can I use a sampling distribution if I only ever have one sample?

You never see it directly; it is a thought experiment about all the samples you could have drawn. Theory, such as the formula σn\frac{\sigma}{\sqrt{n}}, tells you its spread using information from a single sample, and a computer can build it by simulation. Your one sample is a single draw from it, and the theory is what lets you judge how unusual that draw is.

Why is the sampling distribution of the mean centered exactly on the population mean?

Because every member of the population is equally likely to land in a random sample, the average of all possible sample means works out to exactly μ\mu. Statisticians call such an estimate unbiased. The centering fails when the sampling method is biased, for example if taller people are more likely to be picked, which shifts the whole sampling distribution upward.

Learning Path

You need Population vs Sample (parameters and statistics), Mean as Fair Share (what an average is) and Standard Deviation (spread). The sampling distribution adds the distribution of a statistic itself: centred on the parameter, with spread σn for the mean. Next, the Central Limit Theorem shows this distribution is close to normal for large samples, and Standard Error names and estimates its spread.

See Also