This coefficient of variation calculator measures relative spread from observations or a known mean and standard deviation. It shows the CV ratio, percentage CV and relative standard deviation with explicit scale limitations.
Interpret relative spread on a meaningful scale
The coefficient of variation compares standard deviation with the mean. Its ratio is CV = SD / mean. Percentage CV multiplies that ratio by the percentage scale. Relative standard deviation uses the absolute mean in its denominator, so it is a nonnegative percentage. Relative-spread interpretation needs a measurement scale with a meaningful zero. A change of unit that multiplies every observation by the same positive factor preserves CV. Adding a constant changes the mean without changing SD, so it changes CV. Celsius and Fahrenheit have arbitrary offsets; their CV values are not meaningful unit-independent comparisons of relative temperature spread. The NIST coefficient-of-variation reference explains this ratio-scale boundary. Choose raw observations when you have individual values, or known summaries when you already have mean and SD. Match the sample or population SD basis. Numerical values do not establish a measurement scale, so the default warns that the scale has not been established. Optional settings let you identify a ratio scale or an interval/arbitrary-zero scale.
Interpret CV as relative spread only when the measurement has an appropriate ratio scale. An arbitrary offset changes the answer.
CV is not defined at a zero mean. Near-zero means make the ratio sensitive to small changes.
Use sample SD for a sample summary or population SD for a complete population. Supplied summaries keep the basis you identify.
Check mean, standard deviation and their ratio
The sample preset contains 8 values: 2 4 4 4 5 5 7 9. The engine reports mean 5, centered squared-deviation sum 32 and sample SD 2.138089935. The calculation divides the unrounded SD by the mean to get CV 0.4276179871. Percentage CV is 42.76179871%, and RSD is 42.76179871% because the mean is positive.
For the same observations treated as the complete population, the mean stays 5, population SD is 2 and CV is 0.4. Its percentage is 40%. This difference comes from the SD denominator, not from changing the observations. Sample SD uses one less than the count, while population SD uses the full count. The OpenStax spread definitions distinguish those conventions.
Separate CV, percentage CV and relative standard deviation
CV is the signed SD-to-mean ratio. Percentage CV is that same signed ratio expressed as a percentage. RSD divides by the absolute mean and therefore remains nonnegative. The labels are separate because these conventions differ when the mean is negative.
The known-summary check with mean -5 and SD 2 gives CV -0.4, percentage CV -40% and RSD 40%. These are formal arithmetic results. A negative mean or observations can make relative-spread interpretation inappropriate; the calculator shows that warning rather than assigning a good/bad band. A nonnegative RSD does not solve a measurement-scale problem. NIST Dataplot explicitly defines RSD using the absolute mean.
For known summaries, the tool does not reconstruct a dataset or infer its count. Match the supplied SD basis. No sample count is needed for the ordinary ratio; a count is required only if you request the normal-model sample adjustment.
Handle zero means, offsets and model adjustments
A zero mean has no defined CV or RSD ratio, even if SD is also zero. A nonzero constant dataset can have SD zero and CV zero. A sample needs enough observations for sample SD; a single-value complete population has population SD zero. Near-zero means can produce very large ratios, but this tool does not impose an arbitrary acceptable-variation threshold. The NIST scale and near-zero cautions explain why context matters.
Adding 20 to every value in the sample preset changes the mean from 5 to 25. SD remains 2.138089935, while CV becomes 0.08552359741. Multiplying the original observations by the positive factor 10 instead gives mean 50, SD 21.38089935 and unchanged CV 0.4276179871. These examples show why a meaningful measurement zero matters.
The optional first-order normal-model sample adjustment applies the factor 1 + 1 / (4N) to sample CV. It is off by default and assumes a normal model. In the known-summary preset, mean 50, sample SD 5 and sample count 20 give ordinary CV 0.1 and adjusted CV 0.10125. The adjusted result appears separately in details. Do not treat this formula as an exact or distribution-free promise of an unbiased estimate; the tool does not infer normality from a dataset or compute a confidence interval.
For absolute spread, the Standard Deviation Calculator supports observations, exact frequencies and grouped midpoint estimates. The Mean Median Mode Calculator compares different summaries of center. The Z-Score Calculator measures an observation’s signed distance from a reference mean. The Interquartile Range Calculator explores ordered middle spread; this CV tool retains every valid observation and does not automatically delete outliers.
Local input limits are 5,000 raw observations, 100,000 characters and 128 characters per number. Nonzero input magnitudes range from 1e-100 to 1e+100. Simple fractions, decimals and scientific notation are accepted. Nonnumeric tokens are rejected rather than silently omitted. Known-summary SD may be zero but cannot be negative.
The engine uses IEEE754 numbers and compensated mean/centered-spread arithmetic. Display rounding occurs after calculation, to 10 significant digits. Differences already lost when numbers are represented cannot be recovered. Relative variation alone does not establish clinical QC acceptance, investment outcomes, a coefficient confidence interval, or a universal threshold.
The Pooled Standard Deviation Calculator combines sample spreads with degrees-of-freedom weights. Keep groups separate; its within-group result differs from the standard deviation of all observations together.
The Mean Absolute Deviation Calculator averages absolute distances from the mean and offers a clearly labeled unscaled median absolute deviation. Its distance aggregation differs from squared-deviation SD.
The Geometric Mean Calculator averages multiplicative values or equal-period factors. It separates raw values, relative weights and percentage changes, with explicit zero conventions.