Standard Deviation Calculator

How to Use the Standard Deviation Calculator

Standard deviation measures how spread out a set of numbers is, in the same units as the numbers themselves. Two datasets can share a mean and describe completely different situations, and this is the figure that tells them apart.

1. Paste your data into the box. Commas, spaces, tabs and new lines all work as separators, so a spreadsheet column goes in without editing.
2. Choose sample or population. Population is for when you have every member of the group — all twelve months of the year, every employee. Sample is for when your data is a subset of something larger, which is nearly always.
3. Read the standard deviation as a distance from the mean in the same units as your data. If the values are prices in dollars, so is the standard deviation.
4. Check the variance if you need it, but do not try to interpret it directly. It is in squared units — dollars squared, which is not a thing — and exists mainly as the step before the square root.
5. Look at the standard error when your data is a sample. It estimates how much the mean itself would wobble if you drew a different sample of the same size.

Everything runs in your browser. A column of salaries, test scores or clinical measurements is never uploaded anywhere.

Standard Deviation Formula

Four steps, and the only contested one is the divisor.

The n − 1 in the sample formula is called Bessel's correction. A sample mean sits slightly closer to its own data than the true population mean does, so the squared deviations come out a little too small; dividing by a smaller number compensates. The effect matters at small n and becomes negligible as the dataset grows.

Example

Twelve values: 42, 58, 47, 61, 53, 49, 58, 44, 72, 51, 58, 46

1. Mean: 639 ÷ 12 = 53.25.
2. Deviations: 42 − 53.25 = −11.25, then 58 − 53.25 = +4.75, and so on for all twelve.
3. Square each: 126.5625, 22.5625, 39.0625, 60.0625 …
4. Sum of squares: 806.25.
5. Sample variance: 806.25 ÷ 11 = 73.295455. Sample standard deviation: 8.561276.
6. Population variance: 806.25 ÷ 12 = 67.1875, giving a standard deviation of 8.196798.

What the number means

A standard deviation of 8.56 around a mean of 53.25 says that typical values sit roughly eight and a half units either side of the average — between 44.69 and 61.81. Nine of the twelve values, or 75%, fall inside that band. Widening to two standard deviations gives 36.13 to 70.37, which captures eleven of the twelve, or 91.7%.

Sample against population

The two answers here are 8.561276 and 8.196798 — the sample figure is 4.4% larger. On twelve values that gap is noticeable; on a thousand values the two would agree to three decimal places. Use the sample formula unless you genuinely have every member of the group, which in practice is rare enough that most software defaults to it.

What the Number Actually Describes

The first few values, their deviations, and those deviations squared.

Look at the 72. Its deviation of 18.75 squares to 351.56, which is 43.6% of the entire sum of squares from one value out of twelve. That is squaring doing its job: a single distant value dominates the measure, which is exactly what you want when the question is how spread out the data is.

For data that follows a normal distribution, standard deviation has a very specific meaning: about 68% of values fall within one standard deviation of the mean, 95% within two, and 99.7% within three. This dataset gives 75% and 91.7%, which is reasonably close for twelve values — small samples rarely match the theoretical proportions exactly, and expecting them to is a common error.

The coefficient of variation is the figure worth reaching for when comparing datasets in different units. It is the standard deviation as a percentage of the mean — 16.08% here — which lets you say that one process is more variable than another even when one is measured in seconds and the other in kilograms. Raw standard deviations cannot be compared across units at all.

Five Things Standard Deviation Does Not Tell You

Five things standard deviation does not tell you.

The shape of the distribution. A dataset clustered tightly around the mean and one split into two distant groups can share a standard deviation entirely. The number summarises spread and says nothing about whether the data has one peak, two, or a long tail — which is why a histogram is worth more than any summary statistic.

Whether outliers are real. The 72 in this dataset raises the standard deviation substantially. Whether that is genuine variation or a mis-transcribed 27 is a question about the data, not about the arithmetic, and no formula will answer it.

Anything about causation. Two processes with the same variability are not related, and a change in standard deviation over time does not explain itself.

Whether the mean is the right centre. On heavily skewed data the mean sits away from the bulk of the values, and a standard deviation calculated around it describes spread around a point where little of the data actually is. Check the mean against the median with the average calculator before trusting either.

Whether your sample is representative. Standard error estimates sampling variation on the assumption that the sample was drawn fairly. If the sampling was biased, a small standard error simply means you have measured the bias precisely.

Where the measure earns its keep is in process control. A production line, a support queue or a delivery time with a stable mean and a rising standard deviation is deteriorating even though the headline figure has not moved — customers experience the variability, not the average, and a service that is sometimes excellent and sometimes terrible reads as unreliable however good the mean looks.

That is the argument for plotting spread alongside centre as a matter of routine rather than reaching for it only when something has already gone wrong.

Used carefully, it remains the most informative single number you can put beside a mean. A monthly average with no measure of spread is half a report, and it is the half that hides whether the process is under control or merely averaging out.