A central limit theorem calculator shows how the mean of a random sample behaves. Enter the population mean, the standard deviation and the sample size, and it gives the standard error and the chance that the sample mean falls above, below or between values. It also handles sums and sample proportions.
How to use the central limit theorem calculator
- Choose the mean, the sum or a proportion.
- Enter μ, σ and the sample size n, or the population proportion p and n.
- Enter one or two values to get probabilities, then press Calculate.
What the theorem says
The Wikipedia article on the central limit theorem says that the distribution of a normalised version of the sample mean converges to a standard normal distribution.[1] In the Lindeberg–Lévy form, for independent identically distributed variables with mean μ and finite positive variance σ², for large enough n the distribution of the sample mean gets arbitrarily close to the normal distribution with mean μ and variance σ² ÷ n.[1] So whatever the shape of the population, the sample means are approximately normal when n is large.
The standard error
The article on standard error gives the standard error of the mean as σ divided by the square root of n, the standard deviation of the sampling distribution, and says that when σ is unknown you replace it with the sample standard deviation s.[2] It points out that to reduce the error by half you need four times as many observations.[2] With σ = 15 and n = 36, the standard error is 15 ÷ 6 = 2.5, which is what the calculator shows. The page on sampling distributions adds that a sampling distribution depends on the underlying population, the statistic, the sampling procedure and the sample size.[3]
The sum
The page on the central limit theorem says that the sum of n variables has mean nμ and variance nσ².[1] The standard deviation of the sum is therefore σ times the square root of n. For μ = 100, σ = 15 and n = 36, the sum has mean 3,600 and standard deviation 90. Choose the sum mode to see this and the probabilities for the total.
Finding probabilities
The calculator uses z = (x̄ − mean) ÷ standard error and the normal distribution. The page does not write this z formula out, but it follows from the normal limit described above.[1] For the sample mean with μ = 100, σ = 15 and n = 36, the chance of a mean above 104 has z = 1.6 and a probability of about 5.48 percent. The chance of a mean between 96 and 104 is about 89.04 percent. These results are our own calculations.
Proportions and p-hat
For a sample proportion, the standard error is the square root of p(1 − p) ÷ n. The pages we read do not give this formula, so it is our own working from the standard result for a binomial count divided by n. For p = 0.4 and n = 100, the standard error is 0.049, and a sample proportion of 0.45, from 45 successes, has a z score of about 1.02. Enter the number of successes to see p-hat directly.
Sampling error
The calculator also shows a typical sampling error: 1.96 standard errors on either side of the mean, the range that holds about 95 percent of sample means under the normal approximation. This is our own convention for a sampling error, which is the amount by which a sample statistic is likely to differ from the population value. For the example, it is about 4.9 on either side of 100.
How large must n be?
The standard error page says the Student t distribution is approximated well by the normal when the sample size is over 100.[2] The page on the central limit theorem says convergence needs a very large number of observations to reach the tails, and mentions a rate of the order of 1 over the square root of n for the error in the approximation when the third moment is finite.[1] A rule of 30 is often taught, but the pages we read do not state it, so use judgement: a skewed population needs a larger sample.
A worked example with proportions
A poll finds that 40 percent of a large population prefer a product. Suppose that you take a sample of 100 people. The standard error is 0.049, so about 95 percent of such samples would give a proportion between 0.304 and 0.496. If your sample shows 45 successes out of 100, the proportion 0.45 sits about one standard error above 0.4, which is not unusual. A sample of 400 would cut the standard error in half, to 0.0245. These are our own examples worked from the formula in the previous section.
Using the three modes together
Use the mean mode for averages such as test scores, the sum mode for totals such as the weight of a load of items, and the proportion mode for yes-or-no outcomes. A load of 36 items with a mean weight of 100 grams and a standard deviation of 15 grams has a total of 3,600 grams with a standard deviation of 90 grams, so a total above 3,780 grams, two standard deviations up, is unlikely, at about 2.3 percent.
Common mistakes
- Using σ instead of the standard error. A sample mean varies less than one observation.
- Applying the result to one observation. It describes means and sums.
- Ignoring dependence. The theorem needs independent observations.
What next
Use this often? Add Central Limi to your Home screen.
Frequently asked questions
What is a central limit theorem calculator?
A tool that gives the mean and standard error of a sampling distribution and the probability that a sample mean, sum or proportion falls in a range.
What is the standard error of the mean?
σ ÷ √n, the standard deviation of the distribution of sample means. With σ = 15 and n = 36 it is 2.5.
What are the mean and standard deviation of a sum?
The mean is nμ and the standard deviation is σ√n.
How do I find the probability that a sample mean is above a value?
Compute z = (value − μ) ÷ (σ ÷ √n) and find the upper tail of the normal distribution. For μ = 100, σ = 15, n = 36 and a value of 104, z = 1.6 and the chance is about 5.5 percent.
What is p-hat?
The sample proportion, the number of successes divided by the sample size. Its standard error is √(p(1 − p) ÷ n).
Was this tool helpful?
What went wrong?
We only receive the reason you pick, never what you typed into the tool.
Thanks for your feedback.