Definition
Standard deviation is a measure of how spread out the values in a dataset are around their mean. It is the square root of the variance, and it is reported in the same units as the original data — which makes it more interpretable than the variance for everyday use. A small standard deviation means the observations cluster tightly around the average; a large one means they are scattered widely.
Standard deviation appears in nearly every formula in inferential statistics — standard errors, confidence intervals, test statistics, and effect sizes all build on it. Together with the mean, it is the most compact possible summary of a roughly bell-shaped dataset.
Why it matters
How it works
The calculation has a clean intuition. Take each observation, subtract the mean, and square the result; average those squared deviations to get the variance; take the square root to return to the original units. Squaring is the key step — without it, positive and negative deviations would cancel and the average distance from the mean would always be zero. Squaring also amplifies large deviations relative to small ones, which is why a single outlier can pull the standard deviation up dramatically. The choice to use the mean as the centre and squared distance as the metric is what makes standard deviation analytically convenient: most theorems in classical statistics rely on those two choices.
For inference, the standard deviation of a population is rarely known, so it is estimated from the sample. The sample standard deviation divides the sum of squared deviations by n − 1 rather than n, a correction known as Bessel's correction that compensates for the fact that the sample mean is itself estimated from the data. This sample value then drives the standard error — the standard deviation of the sampling distribution of the mean — which in turn powers confidence intervals and hypothesis tests. Almost everything in classical inferential statistics traces back through this number.