Coefficient of Variation & Repeatability Calculator

Compare the spread of measurements taken at different levels on scales with a meaningful zero. Calculate sample standard deviation in original units and coefficient of variation as a percent of the mean. Plot normalized raw readings on one horizontal scale and compare a user-entered CV limit across up to six groups.

Relative variability

Coefficient of variation & repeatability

Compare spread across groups measured at different levels.

Measurement groups

Use a measurement scale with a meaningful zero, such as length or mass.

Group 1

Separate 2–100 values with commas or whitespace.

Group 2

Separate 2–100 values with commas or whitespace.

Compare relative spread

The coefficient of variation is sample standard deviation divided by mean, expressed as a percent.

Small components

10% · Above limit 5%

Large components

3% · Below limit 5%

CV limit 5%

Readings as percent of mean

Each reading is divided by its group mean. A larger circle indicates repeated readings at the same position.
Raw readings as percent of the group mean1. Small components, 2. Large components. The mean reference line is 100%. The normalized readings are available in the CSV.1Small components: 90% (1 reading)Small components: 100% (1 reading)Small components: 110% (1 reading)2Large components: 97% (1 reading)Large components: 100% (1 reading)Large components: 103% (1 reading)88%112%
  • 1. Small components
  • 2. Large components
  • Group mean 100%

Repeatability comparison ledger

Sample standard deviation is in each group’s measurement units. CV is relative to that group’s mean.

GroupSample countMeanSample standard deviationCoefficient of variationCV limit comparison
Small components310 mm1 mm10%Above limit
Large components3100 mm3 mm3%Below limit

How to use this calculator

  1. Name each measurement group and enter its optional unit label. Use a meaningful zero and comparable quantities across groups.
  2. Choose Measurements or Summary statistics per group. Paste raw values separated by commas or whitespace, or enter the count, mean and sample SD.
  3. Review Compare relative spread alongside the Repeatability comparison ledger. The ledger shows absolute sample SD and relative CV separately.
  4. Turn on Compare with a CV limit and enter your own percent threshold. Equal CV values are identified using the exact comparison.
  5. Review Readings as percent of mean for raw groups with positive means. Download the CV comparison CSV for original values and exact normalized results.

The Formula

The mathematical logic powering this calculation is:

For n measurements xᵢ, mean x̄ = Σxᵢ/n and sample variance s² = Σ(xᵢ − x̄)²/(n − 1). The sample standard deviation s is the square root of sample variance. CV (%) = 100s/x̄ for x̄ > 0. If the mean is zero, CV is undefined. A positive mean with all readings equal gives 0% CV.

Normalized raw readings are 100xᵢ/x̄. Each group's mean corresponds to 100%. Summary statistics give the group CV without supplying the shape or individual readings. The unit label describes the original readings or summaries. CV itself is dimensionless because the sample SD and mean use the same units.

An optional limit L is compared by evaluating s²/x̄² against (L/100)² exactly. This comparison describes the entered samples. Requiring a given CV for acceptance or comparing population repeatability would also depend on the study design and your chosen procedure. Meaningful ratio-scale examples include lengths, mass and duration. Offset scales such as Celsius change the denominator when units are converted.

Use 1 to 6 groups. Raw groups take 2 to 100 nonnegative decimal measurements. Summary groups take a count of 2 to 10,000, a nonnegative mean and sample SD. Measurement or summary amounts may be up to 1 billion and the limit up to 1,000,000%, with at most six decimal places. Means, variances, normalized readings and squared CV are calculated exactly. Displayed standard deviations, CV and chart coordinates use numerical approximations.

Worked Example

If you input:

  • Small components: 9, 10, 11 mm
  • Large components: 97, 100, 103 mm
  • CV limit: 5%

Result:

Small components have mean 10 mm and sample SD 1 mm, so CV is 10%, above the entered limit. Large components have mean 100 mm and sample SD 3 mm, so CV is 3%, below the limit. The group with the larger absolute spread has the smaller spread relative to its mean.

Frequently Asked Questions

What does the coefficient of variation measure?

It expresses sample standard deviation relative to the sample mean. A CV of 5% means the sample standard deviation equals 5% of the mean. The ratio can help when groups are measured at different levels.

Why do sample SD and CV sometimes order the groups differently?

Sample SD measures absolute spread in the original units. CV scales that spread by each group's mean. A group can have a larger absolute spread and a smaller relative spread if its mean is sufficiently higher.

When is CV meaningful?

Use comparable measurements on a ratio scale with a meaningful zero. Multiplying by a positive unit conversion, such as changing meters to centimeters, preserves CV. Shifting the zero, such as converting between Celsius and Fahrenheit, generally changes CV. A small positive mean can make the ratio sensitive to small changes.

Why is CV undefined for a zero mean?

The mean is the denominator. The calculator still shows the count, mean and sample SD for that group. A group with a positive mean and zero sample SD has a defined CV of 0%.

Can I use a reported mean and standard deviation?

Yes. Enter the sample count, mean and sample standard deviation calculated with n − 1 in the variance denominator. Rounded summaries give results based on the entered precision. The normalized chart uses raw measurements for the groups that provide them.

What does the CV limit comparison establish?

It compares the entered sample's CV with the percentage you specify. The exact comparison distinguishes equality from values that only appear equal after rounding. Inferring a population difference or applying a measurement acceptance rule calls for separate study criteria.

What are the NIST gear examples?

The example loads the first three complete batches of gear diameters in the NIST e-Handbook GEAR.DAT file. It shows how the observed relative spread differs from batch to batch.

Similar Calculators