Central Limit Theorem Calculator
Enter a population mean, population standard deviation, and sample size to instantly get the standard error of the mean and the sampling distribution parameters. Switch to Probability mode to find P(X-bar below or above a value), or use Quantile mode to find the sample mean that cuts off any target probability. All three modes show step-by-step working.
Formula
Worked example
Population mean 50, standard deviation 10, sample size 30. SE = 10 / sqrt(30) = 1.8257. To find P(X-bar <= 54): z = (54 - 50) / 1.8257 = 2.1909. From the standard normal table, P(Z <= 2.1909) = 98.58%. To find the 95th percentile of X-bar: z = 1.6449, so x = 50 + 1.6449 * 1.8257 = 53.00.
What is the Central Limit Theorem?
The Central Limit Theorem (CLT) is one of the most powerful results in all of statistics. It states that if you draw repeated samples of size n from any population with a finite mean (mu) and finite variance (sigma squared), the distribution of sample means will approach a normal distribution as n increases, regardless of the shape of the original population. The resulting bell-shaped sampling distribution has a mean equal to the population mean and a standard deviation equal to sigma divided by the square root of n. That special standard deviation has its own name: the standard error of the mean, or SE. The practical consequence is enormous: you can apply normal-distribution tools (z-scores, confidence intervals, t-tests) to data from skewed, bimodal, or even uniformly distributed populations, as long as your sample is large enough.
Three modes: statistics, probability, and quantile
This calculator covers all three common CLT problems. In Statistics mode you supply the population mean, standard deviation, and sample size, and the calculator returns the standard error and confirms that the sampling distribution mean equals the population mean. In Probability mode you also specify a sample mean threshold and a tail direction (left, right, between two values, or outside two values) and the calculator standardises your value to a z-score, then uses the standard normal CDF to return the probability. In Quantile mode you enter a target probability (for example 0.95) and the calculator inverts the process: it finds the z-score corresponding to that cumulative probability, then transforms it back to a sample mean using x = mu + z * SE. The Show Your Work panel mirrors the exact arithmetic for each mode so you can reproduce every number on paper.
The standard error and why it matters
The formula SE = sigma / sqrt(n) captures a fundamental trade-off: collecting more data makes your estimates more precise in proportion to the square root of the sample size, not linearly. Doubling n shrinks SE by a factor of sqrt(2), which is roughly 29 percent. Going from n = 10 to n = 40 halves the standard error. This is why large clinical trials and large-sample surveys pin their results down much more tightly than small pilot studies. A narrow SE means the sampling distribution is concentrated around the population mean, so any particular sample mean is unlikely to stray far. The chart below shows how SE falls as n grows for your inputs.
Conditions for the CLT and sample size guidance
The rule of thumb most textbooks cite is n >= 30. For a roughly symmetric population that threshold often produces an excellent normal approximation. For highly skewed or heavy-tailed populations you may need n >= 50 or even n >= 100 before the approximation is reliable. Two additional conditions are worth checking: the observations must be independent (sampling with replacement, or the sample is less than 10 percent of the population), and the population variance must be finite (so the theorem does not apply to Cauchy-distributed data, for example). When n is small and the population is clearly non-normal, use an exact test or a bootstrap method instead of relying on the CLT approximation.
Common z-score cutoffs for the standard normal distribution
| Probability (left tail) | Two-tailed confidence level | z-score |
|---|---|---|
| 0.8000 | 60% | 0.842 |
| 0.8413 | 68.27% (1 SD) | 1.000 |
| 0.9000 | 80% | 1.282 |
| 0.9500 | 90% | 1.645 |
| 0.9750 | 95% | 1.960 |
| 0.9772 | 95.45% (2 SD) | 2.000 |
| 0.9900 | 98% | 2.326 |
| 0.9950 | 99% | 2.576 |
| 0.9987 | 99.73% (3 SD) | 3.000 |
| 0.9999 | 99.99% | 3.719 |
These cutoffs are used in hypothesis testing and confidence interval construction. The same z values apply to the sampling distribution of X-bar via z = (x - μ) / SE.
Frequently asked questions
What is the standard error of the mean?
The standard error of the mean (SE) is the standard deviation of the sampling distribution of the sample mean. It equals the population standard deviation divided by the square root of the sample size: SE = sigma / sqrt(n). A smaller SE means individual sample means cluster more tightly around the true population mean, so your estimates are more precise. The SE is the key quantity in constructing confidence intervals and in z-tests for population means.
How large does a sample need to be for the CLT to apply?
The most common rule of thumb is n >= 30. For populations that are roughly symmetric and not too heavy-tailed, n = 30 usually gives a good normal approximation. For strongly skewed distributions (such as income data or claim sizes) you may need n >= 50 or more. For populations that are already normally distributed, the sampling distribution is exactly normal for any n, so even n = 5 is fine. There is no universal minimum: the appropriate size depends on how far the population deviates from normality.
Why does the sampling distribution always look normal even if the population is skewed?
When you average many independent observations, the extreme values in a skewed population become diluted. Positive and negative deviations from the mean tend to cancel across observations, so the average concentrates near the true mean. Mathematically, the moment generating function of the sum of independent random variables converges to the normal MGF as n grows. The CLT formalises this intuition: no matter what shape the population has (as long as it has finite variance), the sum or mean of a large enough random sample will be approximately normally distributed.
What is the difference between a z-score and the standard error?
The standard error is the standard deviation of the sampling distribution, measured in the same units as the original variable. A z-score is dimensionless: it tells you how many standard errors a particular sample mean sits away from the population mean. The formula is z = (x-bar - mu) / SE. Once you have the z-score you look it up in the standard normal table (or use a CDF function) to get the probability. The SE quantifies precision; the z-score tells you how unusual a specific sample mean is under the null distribution.
Can I use this calculator for proportions?
Yes, with a minor substitution. For a sample proportion p-hat from a Bernoulli population with true proportion p, the population standard deviation is sqrt(p * (1 - p)). Enter that as sigma, use n for your sample size, and the calculator returns SE = sqrt(p*(1-p)/n), which is the standard error of the sample proportion. The normal approximation is reliable when both n * p and n * (1 - p) are at least 10.
What does P(X-bar <= x) actually mean?
It is the probability that a randomly drawn sample of size n will have a mean at or below x, assuming the population has the mean and standard deviation you entered. For example, if mu = 50, sigma = 10, n = 30, and x = 54, the probability is about 98.6 percent: if you repeatedly sampled 30 items and computed the mean each time, about 98.6 percent of those means would be 54 or lower. This is the foundation of p-value calculation in one-sample z-tests.