How this instrument works
Every statistic you compute from a sample would come out a little different if you happened to draw a different sample instead. That's obviously true of any single value, but it's also true of the mean itself: if you took another eight quiz scores from the same larger pool of students, you'd get a slightly different average. Now imagine repeating that — drawing sample after sample of the same size, computing each sample's mean, and collecting all those means into their own distribution. That imagined distribution is called the sampling distribution of the mean, and this instrument estimates its standard deviation: literally how spread out that distribution of sample means would be, not how spread out your original data values are.
This is the Central Limit Theorem's home turf. The theorem states that as sample size n grows, the sampling distribution of the mean approaches a normal (bell-shaped) curve — regardless of the shape of the original population — centred on the true population mean, with a standard deviation equal to the population's standard deviation divided by √n. Since you rarely know the population's true standard deviation, this instrument substitutes your sample's own standard deviation, s, giving s ⁄ √n as the best available estimate.
Notice what the √n in the denominator does: it shrinks as your sample size grows. A sample mean built from 100 values is a far more stable estimate of the population mean than one built from 8 values, even if the underlying data is exactly as spread out in both cases — because averaging more numbers together cancels out more of the sampling noise. That shrinking behaviour, not the spread of the raw data itself, is what this figure is built to capture.
- Enter your numbers into Data set — separate them with commas, spaces or new lines in any order.
- Read Standard deviation of the sample mean beneath the data field — it estimates how much this sample's mean would vary across repeated sampling, not how spread out the raw values are.
- Compare it against this site's Standard deviation instrument if you want the spread of your raw data instead of the spread of the sampling distribution of its mean.
- Watch how the figure shrinks if you add more values with a similar spread — that shrinking with larger n is the Central Limit Theorem in action.
- Enter at least two numbers; a single value gives no usable estimate of s, and the formula is undefined for n = 1.
Worked example — the eight quiz scores, resampled in theory
Enter 2, 4, 4, 4, 5, 5, 7, 9 into Data set. This instrument first finds the sample standard deviation, s ≈ 2.138090, from the eight scores' spread around their mean of 5. With n = 8, √n ≈ 2.828427.
Dividing gives Standard deviation of the sample mean ≈ 0.755929 — the figure this instrument returns for this data. Interpreted through the Central Limit Theorem, that ~0.76 estimates how much the average of an eight-score sample like this one would bounce around from one hypothetical batch of eight quiz-takers to the next, which is a much tighter figure than the ~2.14 spread of the individual scores themselves, precisely because averaging eight values together cancels out most of their individual noise.
Questions
What does 'standard deviation of the sample mean' actually mean?
It's the standard deviation of the sampling distribution of the mean — the theoretical spread you'd see if you repeatedly drew same-sized samples from the same population and plotted all of their means. It answers 'how much would my sample's average bounce around if I resampled?', which is a different question from 'how spread out are my individual data values?'
How is this different from the standard deviation of my raw data?
The standard deviation of your raw data (s) describes how spread out the individual values are around their mean. This figure, s ⁄ √n, describes how spread out the mean itself would be across repeated samples — and it's always smaller than s, because averaging multiple values cancels out some of their individual variability.
Why does the Central Limit Theorem divide by the square root of n?
Because averaging n independent values reduces variance by a factor of n, and standard deviation is the square root of variance, so the reduction in standard deviation works out to a factor of √n. That's why doubling your sample size doesn't halve this figure — it shrinks by a factor of √2, about 1.41, instead.
Does the Central Limit Theorem require my original data to be normally distributed?
No — that's the theorem's most useful feature. Whatever shape the underlying population has, the sampling distribution of the mean trends toward a normal shape as n grows, typically a good approximation by around n = 30 for most reasonably behaved data. Smaller samples from a very skewed population may need larger n before the approximation is reliable.
Is this the same number as 'standard error of the mean'?
Yes, arithmetically it's the identical calculation, s ⁄ √n. This page frames it through the Central Limit Theorem, as the standard deviation of the theoretical sampling distribution of the mean — the concept behind why the formula works. A practically framed version of the same figure, aimed at reporting precision in an experiment or poll, lives on this site's Standard error instrument.
How many data points do I need for this calculation?
At least two, since it depends on the sample standard deviation, s, which is undefined for a single value. In practice, the Central Limit Theorem's normal-approximation guarantee gets stronger with more data — a handful of values gives a valid estimate of s ⁄ √n, but the sampling distribution it describes is a closer fit to a true normal curve once n is reasonably large.