What Is Standard Deviation?
Two datasets can share the same average and describe completely different realities: a class of steady B students and a class of A students and failures both average 70%. Standard deviation is the number that tells those stories apart — it measures how far values typically stray from the mean.
The idea before the formula
Take each value's distance from the mean. Some distances are positive, some negative, and they cancel to zero if you just add them — so square them first (making everything positive), average the squares, then square-root back to the original units. That's the whole recipe: root of the mean of the squared deviations.
Step by step
For the classic dataset 2, 4, 4, 4, 5, 5, 7, 9:
- Mean: 40 ÷ 8 = 5
- Deviations: −3, −1, −1, −1, 0, 0, 2, 4
- Squared: 9, 1, 1, 1, 0, 0, 4, 16 — sum = 32
- Variance (population): 32 ÷ 8 = 4
- Standard deviation: √4 = 2
Variance (the squared version) is mathematically convenient but lives in squared units — “dollars squared” means nothing — which is why results are reported as the standard deviation.
Sample vs population: the n−1 question
The single most common exam mistake. If your data is the entiregroup you care about, divide by n (population, σ). If it's a samplestanding in for a larger group — the usual case in research — divide by n−1 (sample, s). The correction (Bessel's) exists because a sample's spread around its own mean slightly underestimates the true spread; n−1 compensates. For the data above, the sample SD is √(32/7) ≈ 2.14 versus the population's 2.
The 68–95–99.7 rule
For roughly bell-shaped data, the standard deviation slices the world into predictable bands around the mean:
- ~68% of values within 1 SD
- ~95% within 2 SD
- ~99.7% within 3 SD
If adult heights average 170 cm with SD 7 cm, about 95% of people fall between 156 and 184 cm. A value 3+ SDs out is rare enough to investigate — the logic behind quality control charts and outlier detection.
Where you meet it in real life
Investment risk (volatility is standard deviation of returns), manufacturing tolerances, test score curves, weather variability, lab measurement error. In every case the question is the same: is a typical value close to the average, or is the average hiding chaos?
Skip the arithmetic
The standard deviation calculator returns both sample and population SD, both variances, the mean and the sum of squared deviations — the intermediate number you need to find where a manual calculation went wrong. Start with the average calculator if you only need the central measures.