Coefficient Of Variation
Calculator
Results
- Coefficient of variation (%)
- 15.625
Statistical results
| Coefficient of variation (%) | 15.625 |
formula-map diagram
- Coefficient of variation (%)
- 15.625
Statistical relationship
Formula
CV = σ / μ × 100= 15.625
Note
This is a simplified model: it applies the displayed standard formula to the summary values you entered and assumes their underlying conditions (independence, normality, correct sampling) hold. It does not analyse a real data set. Check the assumptions before relying on the result.
More in Statistics and probability
See all →Frequently asked questions
What does the coefficient of variation measure?+
It expresses standard deviation as a percentage of the mean, giving a unitless measure of relative variability. A CV of 20% means the standard deviation is one-fifth the size of the average value.
Why use CV instead of just comparing standard deviations?+
Standard deviations aren't directly comparable across data sets with very different means or units, like comparing the variability of employee salaries to shoe sizes. CV normalizes for scale, so you can meaningfully compare which data set is relatively more spread out.
Is a low or high coefficient of variation better?+
It depends on context, but generally a lower CV indicates more consistency relative to the average, which is desirable in quality control or measurement precision. In finance, however, investors sometimes accept a higher CV in exchange for higher expected returns.
Can CV be used on data with a mean of zero or negative values?+
No, CV becomes unstable or meaningless when the mean is zero or close to it, since you'd be dividing by a number near zero. It's also unreliable for data that can be both positive and negative, like temperature in Celsius, because the mean can be small or shift sign.
What's a typical CV threshold used in practice?+
In many lab and quality-control settings, a CV under 10-15% is often considered low variability and acceptable, though the right threshold depends entirely on the field and the tolerance for variation. There's no universal cutoff, so it should be interpreted against domain-specific norms.