Lesson 3 of 8 · 15 min
The central limit theorem
See sample means from skewed, uniform and two-peaked populations become approximately normal as grows, and standardise a sample mean with for .
What you will be able to do
- Describe how the shape of the distribution of changes with for populations of different shapes.
- State the central limit theorem and decide when a normal model for is reasonable.
- Standardise a sample mean with when is unknown and .
Last lesson's batteries had normally distributed lifetimes, so their sample means were normal too. Most real populations are not normal. Waiting times on a helpline are right-skewed: most calls are answered quickly, a few wait a very long time. Incomes, house prices and insurance claims are skewed too. Can you still use the normal distribution for their sample means?
Draw many samples with , then , , and . When does the histogram of start to look like the normal curve? Then switch the population to uniform or bimodal and repeat.
Whatever population you start from, the same three things happen as grows: the histogram of sample means stays centred at , it narrows like , and its shape becomes more and more like a normal curve. Skewed populations take longer than symmetric ones, but they get there.
Key idea
The central limit theorem
For random samples of size from any population with mean and standard deviation , when is large
The larger is, the better the approximation. If the population itself is normal, the result is exact for every .
| Population | Distribution of |
|---|---|
| Normal | Exactly normal for every . |
| Symmetric with one peak (e.g. uniform) | Close to normal for quite small , often by . |
| Two peaks, or moderately skewed | Needs a larger ; usually fine by . |
| Strongly skewed (e.g. waiting times) | Needs , and more for extreme skew. |
Household water bills in a city are strongly right-skewed. A random sample of 100 bills is taken. Which statement is correct?
There is a practical snag. The formulas use , the population standard deviation, which you almost never know. For a large sample, the sample standard deviation is a good estimate of , so you replace by . The syllabus form of the central limit theorem says that for large samples ()
is approximately standard normal, . Here varies from sample to sample too, so this is a different statistic, and the widget below lets you check it.
With the standardised means sit close to . Reduce to 5: the tails get heavier, because from a small sample is an unreliable estimate of .
Worked example
Using the central limit theorem for a skewed population
Waiting times for a council's phone line are strongly right-skewed, with mean minutes and standard deviation minutes. A random sample of calls is taken.
(a) State the approximate distribution of the sample mean waiting time , and justify it.
(b) Find the approximate probability that the sample mean is more than minutes: (i) without technology, using the 68–95–99.7% rule, and (ii) with technology.
(c) Explain why the same method can't be used to find the probability that one call waits more than minutes.
Step 1: Find the mean and standard deviation of the sample mean
and .
Step 2: Justify the normal approximation
The population is skewed, but is large, so by the central limit theorem is approximately normal: .
Step 3: Estimate the probability without technology
is two standard deviations above the mean. About 95% lies within two standard deviations, so .
Step 4: Calculate it with technology
.
The 68–95–99.7% rule rounds to , which is why the two answers differ slightly.
Step 5: Explain why one call is different
One waiting time has the skewed population distribution, not a normal one. The central limit theorem applies to the mean of a large sample, so a normal model for one call would give a wrong probability.
Try it yourself first, then check the working one step at a time.
Answer
(a) by the central limit theorem, since is large. (b) (i) about ; (ii) . (c) A single waiting time is skewed, not normal; only the sample mean is approximately normal.
Method
Find and .
Justify normality: large (central limit theorem) for a non-normal population.
Count standard errors for the 68–95–99.7% rule, or use normCdf.
Watch out for
Claiming the waiting times themselves become normal.
Using instead of .
A random sample of values from a population with mean has and . Calculate the standardised value .
Common mistake
The population does not become normal
The central limit theorem is about the distribution of sample means. One helpline call still has a right-skewed waiting time, so you cannot use a normal model to find for one call. Only probabilities about , with large, use the normal approximation.
For which of these is a normal model for the sample mean least justified?
Exam tip
Justify the normal model
When you use a normal distribution for , say why in a few words: " is normally distributed, so is normal" or " is large, so by the central limit theorem is approximately normal". If neither holds, say that the normal model may not be reasonable.
Summary
For large , is approximately whatever the population shape: the central limit theorem.
The more skewed the population, the larger the sample needed; n ≥ 30 is the usual guide.
For , is approximately , so can stand in for an unknown .
Individual values keep the population's shape; only the sample mean becomes normal.
Always justify the normal model: normal population, or large .
Finished the lesson?
Try the 3 quick checks above and the lesson completes itself, or mark it complete now.