Skip to content
K Knidox Search…
Statistics · Descriptive

Coefficient of Variation

Paste a dataset to get the coefficient of variation — the standard deviation as a percentage of the mean.

Separate values with commas, spaces, or new lines. Non-numeric entries are ignored.
Standard deviation type
Sample data — tap to load
Coefficient of variation
40%
5
Mean
2
SD (population, ÷ n)

The coefficient of variation (CV) measures relative variability — the standard deviation expressed as a percentage of the mean: CV = SD ÷ mean × 100. Because it is unitless, it lets you compare the spread of datasets measured on different scales. For 2, 4, 4, 4, 5, 5, 7, 9 the mean is 5 and the population SD is 2, so CV = 40%.

What the coefficient of variation tells you

Standard deviation alone can mislead: an SD of 5 is tiny on a scale that averages 1,000 and enormous on one that averages 10. The coefficient of variation fixes this by dividing the standard deviation by the mean, turning spread into a relative, unitless percentage. A CV of 40% means the standard deviation is 40% of the average value. Because the units cancel, you can compare the variability of test scores, prices, and reaction times side by side — something raw standard deviations can’t do.

CV = SD ÷ mean × 100

use the sample SD (÷ n − 1) or population SD (÷ n) as appropriate

Worked example

Eight values: 2, 4, 4, 4, 5, 5, 7, 9, using the population standard deviation.

  1. 1
    Find the mean. (2 + 4 + 4 + 4 + 5 + 5 + 7 + 9) ÷ 8 = 40 ÷ 8 = 5.
  2. 2
    Find the standard deviation. The squared deviations sum to 32; population SD = √(32 ÷ 8) = √4 = 2.
  3. 3
    Divide SD by the mean, then × 100. 2 ÷ 5 = 0.4, and 0.4 × 100 = 40%.

A rough guide to CV values

Interpretation is context-dependent — a “high” CV in one field is normal in another.

CVRelative spreadReading
< 15%LowValues cluster tightly around the mean
15% – 30%ModerateA typical, middling amount of spread
> 30%HighValues are widely dispersed relative to the mean

Interpreting and using the result

It compares spread across scales. The whole point of the CV is to compare variability between datasets with different units or very different means — heights in centimetres versus weights in kilograms, for example. Standard deviation can’t make that comparison; the CV can.

It breaks down near zero. Because you divide by the mean, the CV becomes unstable when the mean is close to zero — tiny changes in the mean swing it wildly — and it is undefined when the mean is exactly 0.

It needs a meaningful zero. The CV assumes a ratio scale with a true zero and positive values. It is not meaningful for data that can be negative (like temperatures in °C or profit-and-loss figures), where the mean may sit near or below zero.

What is a good coefficient of variation?
There is no universal cutoff — it depends on the field. As a rough guide, under 15% is low, 15–30% is moderate, and above 30% is high, but a “high” CV in one discipline can be perfectly normal in another.
Should I use the sample or population standard deviation?
Use the population SD (÷ n) only when your numbers are the entire group. If they’re a sample standing in for something larger — the usual case — use the sample SD (÷ n − 1). The tool lets you switch between the two.
When is the coefficient of variation not useful?
When the mean is at, near, or below zero. Dividing by a mean close to 0 makes the CV explode, and it is undefined at a mean of exactly 0. It also assumes a ratio scale, so avoid it for data like Celsius temperatures.
Why express it as a percentage?
Multiplying SD ÷ mean by 100 turns the ratio into a percentage that’s easy to read — a CV of 40% simply means the standard deviation is 40% of the mean.
How is CV different from standard deviation?
Standard deviation is absolute spread in the original units; the coefficient of variation is relative spread as a unitless percentage of the mean, which makes it comparable across datasets measured on different scales.
Can the coefficient of variation be greater than 100%?
Yes. When the standard deviation exceeds the mean — common in highly skewed or widely dispersed data — the CV rises above 100%.