The mean tells you where the centre of your data is. The standard deviation tells you how spread out the data is around that centre. Both are essential — one without the other gives you an incomplete picture.

What standard deviation measures

Standard deviation is the average distance between each individual value and the mean. A small standard deviation means values cluster tightly around the mean. A large standard deviation means they're scattered widely.

Consider two classes of students, both with a mean test score of 72:

  • Class A scores: 69, 70, 71, 72, 73, 74, 75 — standard deviation ≈ 2
  • Class B scores: 30, 45, 60, 72, 85, 92, 100 — standard deviation ≈ 24

Same mean. Completely different distributions. Class A is tightly clustered; nearly everyone is near the average. Class B has a wide range; the "average" of 72 doesn't describe most students' experience well at all.

How to calculate it

You rarely need to calculate standard deviation by hand — any spreadsheet or data tool will do it instantly. But understanding the formula clarifies what it's measuring:

  1. Find the mean of all values.
  2. For each value, calculate the difference from the mean (value − mean).
  3. Square each difference (this removes negative signs and amplifies large gaps).
  4. Find the average of the squared differences.
  5. Take the square root (to undo the squaring and get back to the original units).

The result is in the same units as the original data. If you're measuring revenue in pounds, the standard deviation is in pounds. If you're measuring temperature in degrees Celsius, so is the standard deviation.

Population vs sample standard deviation

You'll encounter two versions: the population standard deviation (σ, using n in the denominator) and the sample standard deviation (s, using n−1).

If your data is the entire population you care about (all employees in a company, all products in a catalogue), use the population formula. If your data is a sample from a larger population (1,000 survey responses out of millions of customers), use the sample formula. Most statistical analysis uses samples, so n−1 is the default in most software.

Interpreting standard deviation in context

There's no universal threshold for what counts as "large" or "small." Standard deviation is always relative to the mean and to the domain.

A standard deviation of £5 is large for a loaf of bread (mean price £1.20) but tiny for a house purchase (mean price £250,000). A standard deviation of 2°C is large for a temperature-controlled laboratory setting but trivial in the context of seasonal weather variation.

The coefficient of variation (CV = standard deviation / mean × 100%) standardises the spread as a percentage of the mean, making it useful for comparing variability across datasets with different scales.

The empirical rule (for normal distributions)

When data follows a normal (bell-curve) distribution, the empirical rule gives you useful benchmarks:

  • About 68% of values fall within 1 standard deviation of the mean.
  • About 95% fall within 2 standard deviations.
  • About 99.7% fall within 3 standard deviations.

This means a value more than 3 standard deviations from the mean is extremely rare in a normal distribution — less than 0.3% chance. This is the basis of many outlier detection methods and quality control systems.

Standard deviation vs variance

Variance is the square of the standard deviation (step 4 in the calculation above, before taking the square root). Variance is mathematically convenient for many statistical formulas, but it's in squared units, which makes it hard to interpret directly. When communicating results to non-statisticians, standard deviation is almost always preferable because it's in the same units as the data.

When to use standard deviation

Standard deviation is most meaningful when your data is roughly normally distributed. For heavily skewed data (incomes, house prices, website traffic), the interquartile range (IQR) — the difference between the 75th and 25th percentiles — is a more robust measure of spread because it isn't affected by extreme values.

Whenever you report a mean, report its standard deviation alongside it. "Our average delivery time is 3.2 days (SD = 0.4)" tells a very different story from "Our average delivery time is 3.2 days (SD = 4.1)".