Two classes both average 80 on a test. In one class, nearly everyone scored between 75 and 85. In the other, half the room failed while the rest aced it. The average can't tell these classes apart — standard deviation can. It's the single most useful number in statistics after the mean, and calculating it takes five steps you can do on paper.

What standard deviation actually measures

Standard deviation (SD, or the Greek letter σ) measures how spread out your data is around the mean. A small SD means the values huddle close to the average; a large SD means they're scattered. Same units as your data — if you're measuring test scores, SD is in points; if it's heights in cm, SD is in cm. That's what makes it readable: "average 80, SD 3" instantly tells you most scores sit within a few points of 80.

How to calculate it: 5 steps

Take five test scores: 70, 74, 78, 82, 96.

  1. Find the mean. (70 + 74 + 78 + 82 + 96) ÷ 5 = 80. (The average calculator handles this part.)

  2. Subtract the mean from each value. −10, −6, −2, +2, +16.

  3. Square each difference. 100, 36, 4, 4, 256. (Squaring kills the minus signs and punishes big outliers extra hard.)

  4. Average the squares. (100 + 36 + 4 + 4 + 256) ÷ 5 = 80. This number is the variance.

  5. Take the square root. √80 ≈ 8.94. That's the standard deviation.

So: mean 80, SD ≈ 8.9 — the typical score sits about 9 points from the average, and that lone 96 is what's stretching the spread. The standard deviation calculator runs all five steps for any dataset and shows the variance along the way.

Sample vs population: when to divide by n − 1

The one place everyone gets tripped up. In step 4:

  • Divide by n if your data is the entire population — every value you care about (all 5 students in a 5-student class).

  • Divide by n − 1 if your data is a sample standing in for a bigger group (5 students representing the whole school).

Why? A sample tends to underestimate the true spread — its own mean is always suspiciously close to its own values. Dividing by n − 1 inflates the result just enough to correct the bias. For our five scores as a sample: 400 ÷ 4 = 100, and √100 = 10 instead of 8.94. With large datasets the difference nearly vanishes; with small ones it matters a lot.

The 68–95–99.7 rule

For roughly bell-shaped (normal) data, SD turns into a prediction machine:

  • ~68% of values fall within 1 SD of the mean

  • ~95% within 2 SD

  • ~99.7% within 3 SD

Adult male height is famously close to normal: mean ~175 cm, SD ~7 cm, so about 95% of men are between 161 and 189 cm. This is also how "unusual" gets quantified — a value 3 SD from the mean happens naturally only about 0.3% of the time, which is why labs, factories, and researchers all flag results beyond that line.

Where you'll actually use it

  • Comparing consistency: two investments with the same average return are not equal if one has triple the SD — in finance, SD literally is the standard measure of risk.

  • Grading on a curve: your "z-score" (how many SDs you are from the mean) is what standardized tests convert into percentiles.

  • Spotting outliers: a data point several SDs out is either an error or the most interesting thing in your dataset.

  • Quality control: "Six Sigma" is exactly this — engineering processes so defects live six standard deviations away from normal.

The bottom line

The mean tells you where the center is; the standard deviation tells you whether the center means anything. Whenever you see an average — test results, salaries, returns, delivery times — ask for the spread before you trust it. Paste your numbers into the standard deviation calculator and you'll have the mean, variance, and SD in one shot — and if you need the result as a share of the mean, the percentage calculator finishes the job.