Why n ≥ 30 Is a Rule of Thumb, Not a Theorem

You’ll be able to

State the Central Limit Theorem precisely and identify what it does and does not claim
Explain why the required sample size depends on the population shape
Check the conditions for Normality of a sampling distribution

What the theorem actually says

The Central Limit Theorem states that as the sample size grows, the sampling distribution of the sample mean approaches a Normal distribution, whatever the shape of the population — provided the population has a finite standard deviation. Two things it does not say. It says nothing about the shape of the population, which is unchanged. And it says nothing about the shape of an individual sample, which continues to resemble the population. Only the distribution of the statistic becomes Normal.

Why 30 is a convention

The theorem is a statement about a limit, so it gives no specific sample size at which the approximation becomes adequate. The threshold n ≥ 30 is a working convention, and how good it is depends entirely on how far the population departs from Normal. For a symmetric population, samples of 5 or 10 may already give a nearly Normal sampling distribution. For a strongly skewed population — incomes, waiting times, insurance claims — even n = 30 can leave visible skew, and n in the hundreds may be needed. The more skewed the population, the larger the n required.

The conditions you must actually check

For inference about a mean, the Normality condition is met if any of these holds: the population is stated to be Normal; the sample size is large (the n ≥ 30 convention); or a graph of the sample data shows no strong skew and no outliers. That last route is the one students forget, and it is often the only one available. For a proportion, the analogous condition is the large-counts check, np̂ ≥ 10 and n(1 − p̂) ≥ 10, which fails for rare events even at large n.

Watch out

The Central Limit Theorem does not rescue a biased sample. It describes how a statistic varies across random samples; it says nothing about a sampling method that systematically misses part of the population. A large voluntary-response sample has a beautifully Normal sampling distribution centered on the wrong value.

Worked example

A hospital records emergency-room waiting times, which are strongly right-skewed with a long tail. A researcher plans to take samples of size 15 and use a t procedure. Evaluate the plan.

1.The population is strongly right-skewed, so the sampling distribution of x̄ will inherit some of that skew at small n.
2.n = 15 falls below the usual n ≥ 30 convention, and that convention is itself least reliable exactly when the population is strongly skewed.
3.The remaining route is to graph the sample: if a dotplot or boxplot of the 15 observations shows strong skew or outliers, the Normality condition is not met.
4.The appropriate response is to increase n substantially, or to report that the condition cannot be verified and interpret results cautiously.

Answer: The plan is not adequate. With a strongly right-skewed population, n = 15 is too small for the Central Limit Theorem to deliver an approximately Normal sampling distribution, and the n ≥ 30 guideline would itself be optimistic here. The researcher should increase the sample size well beyond 30, and should in any case graph the sample data and report what the graph shows.

Checkpoint

The Central Limit Theorem guarantees that, for large n, which distribution is approximately Normal?

Checkpoint

Compared with a symmetric population, a strongly skewed population requires —

Answer the 2 checkpoints as you read.