What does the central limit theorem say, and why does it work?
The central limit theorem says that the means of many random samples of size n form a distribution that gets closer to normal as n grows, centered on the population mean μ with standard deviation σ/√n. It holds for almost any population shape, as long as the population has a finite standard deviation.
Why averaging builds a bell
Roll one die and every face from 1 to 6 is equally likely: a flat distribution. Roll two and average them, and 3.5 is now the most likely average, because many combinations produce it (1 and 6, 2 and 5, 3 and 4, in either order) while only one produces an average of 1. Average ten dice and the distribution is already close to a bell.
The reason is that extremes need everything to go the same way at once, while middling results can happen in many ways. Highs and lows in a sample tend to cancel, so averages crowd toward the center, whatever the original shape.
The numbers, with dice
One fair die has μ = 3.5 and σ ≈ 1.71. For samples of n = 25 rolls, the sample means have mean 3.5 and standard deviation 1.71 / √25 ≈ 0.34. Since that distribution is approximately normal, about 95% of 25-roll averages land within two standard deviations of 3.5, between about 2.82 and 4.18.
The √n matters. To halve the spread of the sample mean, you need four times the sample size, not twice.
What exams check
- The theorem is about the sampling distribution of the mean, not the data. A large sample does not make the data themselves normal.
- n ≥ 30 is a rule of thumb, not a law. A strongly skewed population needs a bigger n; a population that is already normal gives normal sample means at any n.
- Observations must be independent. When sampling without replacement, the 10% condition (n at most 10% of the population) keeps that close enough.
- The standard deviation of the sample mean is σ/√n, not σ/n.
This is the result that makes confidence intervals and the normal tail areas behind a p-value work even when the population is not normal.
Watching the means pile up
Type "Why does the central limit theorem work?" into Isotoon and you get a short animated video. A good one draws samples from a lopsided population, drops each sample's mean onto a histogram, and lets you watch the bell form and narrow as n grows. For the course structure, see AP Statistics and statistics.
Questions
- Does the central limit theorem make my data normal?
- No. It describes the distribution of sample means across many samples. The individual values in your sample keep the shape of the population they came from.
- Why is the sample size 30 rule not always enough?
- It is a rough guide for populations that are not too skewed. Heavily skewed populations or ones with extreme outliers need larger samples before the sample mean's distribution looks normal.