Coefficient of Variation: Comparing Variability Across Different Units and Scales
October 21, 2026
The standard deviation tells you how spread out values are, but in the units of the data. A standard deviation of 5 grams is large for a mean of 20 grams and tiny for a mean of 2,000 grams. To compare variability between datasets with different means or units, divide the standard deviation by the mean. The result is the coefficient of variation (CV), usually written as a percentage: CV = (standard deviation ÷ mean) × 100%.
A machine that fills bags with a mean of 50 grams and a standard deviation of 5 grams has a CV of 10%. Another machine filling cartons with a mean of 200 grams and a standard deviation of 20 grams also has a CV of 10%, even though its absolute spread is four times larger. Relative to their targets, they are equally consistent.
Skip the math and get your answer instantly:
Open the Standard Deviation Calculator →Comparing Different Quantities
Suppose a group of adults has an average height of 170 cm with a standard deviation of 9 cm, and an average weight of 70 kg with a standard deviation of 12 kg. Which varies more? The standard deviations cannot be compared directly because cm and kg are different units. The CVs are 5.3% for height and 17.1% for weight, so weight is more variable relative to its mean, about three times as much.
In a lab, a measurement method with a CV of 2% is considered precise for many assays, while 15% or more would be poor.
| Dataset | Mean | Standard deviation | CV |
|---|---|---|---|
| Bag fill (grams) | 50 | 5 | 10% |
| Carton fill (grams) | 200 | 20 | 10% |
| Adult height (cm) | 170 | 9 | 5.3% |
| Adult weight (kg) | 70 | 12 | 17.1% |
| Daily shop sales ($) | 1,200 | 360 | 30% |
Risk Per Unit of Return
In finance, the CV of returns is the inverse of the Sharpe-like idea: risk per unit of return. An investment with an average annual return of 8% and a standard deviation of 16% has a CV of 2.0. Another with a 5% return and a standard deviation of 5% has a CV of 1.0, meaning it delivers its return with less relative variability. Lower is better in this framing, although investors care about absolute return as well. Compare options with a similar objective using CV as one input, not the only one.
Quality Control and Science
Manufacturers use CV to compare consistency across products and sites. Laboratories report intra-assay and inter-assay CVs to describe measurement precision. In biology, CV of gene expression helps identify noisy genes. In ecology, CV of population counts over time indicates stability. A CV below about 10% is often considered low variability, 10% to 30% moderate, and above 30% high, though interpretation depends on the field.
Cautions
The CV only makes sense for data measured on a ratio scale with a true zero, such as weight, length, and counts. It is meaningless for temperatures in Celsius or Fahrenheit, because the zero point is arbitrary: the CV of the same temperatures in Fahrenheit would differ from that in Celsius. It also becomes unstable when the mean is near zero, since dividing by a tiny number produces huge or misleading values. And it assumes variation scales with the mean. For small samples, the sample CV is a biased estimate. Always report the mean and standard deviation along with the CV.
Frequently Asked Questions
What is the coefficient of variation?
The standard deviation divided by the mean, often expressed as a percentage.
When should I use CV instead of standard deviation?
When comparing variability across datasets with different means or units.
What is a good CV?
It depends on the field. Below 10% is often considered low variability.
Can I use CV for temperatures in Celsius?
No. CV requires a ratio scale with a true zero.
Why is CV unreliable when the mean is near zero?
Dividing by a tiny mean inflates the ratio.