What Standard Deviation Measures
Standard deviation measures how spread out numbers are around their mean. A dataset of 12, 15, 18, 20, 25 has a mean of 18, and the numbers sit 6, 3, 0, 2 and 7 points away from it. Standard deviation turns those distances into a single number: about 4.95 for this set. Two datasets can share a mean and still be completely different — 17, 18, 18, 18, 19 also averages 18, but its standard deviation is under 1. That is why the mean alone is never enough: the mean says where the centre is, and the standard deviation says how far the numbers typically wander from it. In quality control, in test scores, in experiment results, the spread is often the interesting number.
Worked Example: Five Test Measurements
An engineer records five test measurements: 12, 15, 18, 20, 25. Step one, the count and sum: n = 5, sum = 90. Step two, the mean: 90 ÷ 5 = 18.00. Step three, the deviations from the mean: −6, −3, 0, +2, +7. Step four, the squared deviations: 36, 9, 0, 4, 49 — squaring stops the negatives cancelling the positives. Their sum is 98. Step five, the variance: divide the sum of squares by n − 1 for a sample, so 98 ÷ 4 = 24.50. Step six, the standard deviation: √24.50 = 4.95. Two related numbers come free: the mean absolute deviation (MAD) averages the absolute deviations (6 + 3 + 0 + 2 + 7) ÷ 5 = 3.60, and the population standard deviation — dividing by 5 instead of 4 — is √(98 ÷ 5) = √19.6 = 4.43. The sample figure (4.95) is bigger than the population figure (4.43), and that gap is the whole point of the next section.
Sample vs. Population: Bessel’s Correction
Divide by N when you measured the entire population — every member, no sampling. Divide by n − 1 when you measured a sample drawn from a larger population. The n − 1 is Bessel’s correction, and it exists because a sample’s spread around its own mean is always smaller than the population’s spread around the true mean: the sample mean sits in the middle of the sample by construction, so the deviations look smaller than they really are. Dividing by n on a sample systematically underestimates the true variability — by a factor of √((n − 1)/n), which is large for small samples (a sample of 5 underestimates by about 11%) and negligible for large ones. In practice: exam questions and lab reports say which one they want, and if they say “sample standard deviation” they mean n − 1. When in doubt, use the sample formula — it is the honest estimate of the population from the data you have.
Variance, Standard Deviation and MAD
Variance is the average squared deviation — 24.50 in the worked example. Its units are squared (points², dollars²), which makes it hard to interpret directly, so we take the square root to get the standard deviation, back in the original units. MAD (mean absolute deviation) skips the squaring and averages the absolute distances: 3.60 here. MAD is easier to explain — “numbers are typically 3.6 points from the mean” — and it is less sensitive to outliers, because squaring gives extreme values disproportionate weight. A single reading of 60 in the example set would push the standard deviation far more than the MAD. Use SD when the maths needs it (it is the standard in statistics, and it feeds directly into the normal distribution); use MAD when you want a robust, plain-language measure of typical spread.
The Normal Distribution: 68–95–99.7
When data follow a normal (bell-shaped) distribution, the standard deviation has a famous meaning: about 68% of values fall within one standard deviation of the mean, about 95% within two, and about 99.7% within three. With a mean of 100 and a standard deviation of 15, about 68% of values lie between 85 and 115, about 95% between 70 and 130, and almost all between 55 and 145. That is why test scores are often reported this way: a score of 130 with mean 100 and SD 15 is two standard deviations above the mean — roughly the 98th percentile. But the rule only holds for roughly normal data. Income, house prices and waiting times are skewed, and their standard deviations do not carry the same meaning. Always look at the shape of the data before quoting the 68–95–99.7 rule.
Common Mistakes
The most common mistake is dividing by n when the data are a sample — the default on many calculators and spreadsheets, and the wrong choice for most lab and survey work. The second is reporting the standard deviation without the mean: “SD = 4.95” means nothing on its own, because 4.95 points of spread around a mean of 18 is a lot, while 4.95 around a mean of 10,000 is nothing. The third is comparing standard deviations across different units — a 4.95-point spread in test scores and a 4.95-dollar spread in prices are not comparable; use the coefficient of variation (SD ÷ mean) for that. The fourth is forgetting that standard deviation is sensitive to outliers: one bad measurement can double it, so check the data for errors before trusting the number.