Standard Deviation Solver

What Standard Deviation Tells You About Your Data

Published 2026-01-12

A mean tells you where the center of your data sits. A standard deviation tells you how spread out the values are around that center. This guide explains how to read a standard deviation, what the 68-95-99.7 rule says and when it applies, and where interpretation goes wrong. To get the number itself for your own data, use the calculator on our homepage, then come back here to interpret it.

The core idea: typical distance from the mean

Standard deviation is built from the distances between each value and the mean. Deviations are squared, averaged, and square-rooted, which puts the result back in the original units of your data. (For the full arithmetic, see how to calculate standard deviation by hand.)

You can read the result as a typical distance from the mean, with one nuance: because deviations are squared before averaging, points far from the mean pull the statistic up more strongly than nearby points pull it down. So standard deviation is best read as a scale of spread, not a literal average distance.

Compare two groups:

Both have the same center, but Group B has a much larger standard deviation. Same mean, very different data. That difference in consistency, risk, or reliability is exactly what standard deviation captures.

A worked example

Consider this dataset of eight values:

10, 12, 23, 23, 16, 23, 21, 16

The mean is 18. The population standard deviation is about 4.899, and the sample standard deviation is about 5.2372. (Which one you report depends on whether these eight values are your whole group of interest or a sample from a larger one; see sample vs population standard deviation.)

How do you read “about 4.899”? It means the values typically fall within roughly 5 units of the mean of 18. The exact one standard deviation band runs from about 13.10 to about 22.90. Three of the eight values (16, 16, and 21) fall strictly inside it, and the three 23s sit just outside, about 1.02 standard deviations above the mean. The low values, 10 and 12, are about 1.63 and 1.22 standard deviations below the mean. No value is far outside the band, which tells you the spread is moderate and there are no extreme outliers.

If another dataset had the same mean of 18 but a standard deviation of 0.5, nearly all its values would sit within a unit or so of 18. If it had a standard deviation of 12, values would be scattered widely, and the mean alone would describe the data poorly.

The 68-95-99.7 rule

For data that is approximately normally distributed, meaning it follows the classic symmetric bell curve, standard deviation has a precise interpretation:

This is the empirical rule, often called the 68-95-99.7 rule. It lets you turn a mean and standard deviation into concrete statements. If adult heights in some population are approximately normal with a mean of 170 cm and a standard deviation of 8 cm, then about 95 percent of adults fall between 154 cm and 186 cm.

Critical caveat: the rule only applies to approximately normal data

This is where many explanations stop too soon, and where real mistakes begin. The 68-95-99.7 rule is a property of the normal distribution. It does not apply to data in general.

Skewed data: Income, house prices, and city populations are right-skewed: a long tail of large values stretches the distribution. For skewed data, more than half the values typically sit below the mean, and the one and two standard deviation bands do not capture 68 or 95 percent of the data. Applying the rule here gives wrong conclusions.

Heavy-tailed data: Financial returns and many natural measurements have heavy tails, meaning extreme values occur far more often than a normal distribution predicts. In heavy-tailed data, observations four, five, or ten standard deviations from the mean happen regularly. Treating such events as near-impossible, as the 99.7 percent figure would suggest, has caused famous failures in risk modeling.

Bounded or multimodal data: Percentages capped at 0 and 100, or data with two distinct clusters, also break the rule.

If your data is not approximately normal, use percentiles and the interquartile range to describe spread, or use Chebyshev’s inequality, which guarantees that for any distribution at least 75 percent of values fall within two standard deviations and at least 88.9 percent within three. Those guarantees are weaker than 95 and 99.7 percent, but they are always true.

How to check whether your data is roughly normal

Before using the empirical rule:

  1. Plot a histogram. A roughly symmetric, single-peaked, bell-shaped histogram is a good sign.
  2. Compare the mean and median. In symmetric data they are close together. A big gap signals skew.
  3. Count actual proportions. Check what share of your data really falls within one and two standard deviations of the mean. If the counts are far from 68 and 95 percent, do not quote the rule.

Standard deviation in real contexts

Consistency: A basketball player averaging 20 points per game with a small standard deviation scores reliably near 20. A teammate with the same average but a large standard deviation swings between quiet nights and huge ones.

Risk: In investing, volatility is standard deviation of returns. Two funds can have the same average return while one takes a far bumpier path to get there.

Quality control: A machine filling bottles to 500 ml with a small standard deviation is precise. A growing standard deviation is an early warning that the process is drifting.

In every case, the pattern is the same: the mean describes the target, and the standard deviation describes how tightly reality hugs that target.

To compute the statistic yourself, follow how to calculate standard deviation by hand. To choose the right formula, read sample vs population standard deviation. To avoid misreading what you compute, see common standard deviation mistakes, and run your own data through the standard deviation calculator.