Coefficient of Variation Calculator

How to Use This Calculator

1. Enter the standard deviation and mean of your dataset.
2. Read the CV instantly — a percentage showing relative variability.
3. Use Compare Two Datasets tab to determine which is more variable in relative terms.
4. Use Sample vs Population tab to enter raw data and compute CV directly from values.
5. Professional tier adds inverse CV (signal-to-noise ratio) and QC benchmarks.

Formula

Example

Frequently Asked Questions

• What is the coefficient of variation (CV) and why is it useful? ▼
  The coefficient of variation (CV), also called relative standard deviation (RSD), expresses variability as a percentage of the mean: CV = (standard deviation / mean) × 100%. It was introduced by Karl Pearson in 1896 as a scale-independent measure of dispersion. The key advantage of CV over raw standard deviation is that it is dimensionless — it removes the unit of measurement, allowing meaningful comparison of variability across datasets with completely different scales or units. For example, if you want to compare the precision of a blood pressure measurement (mean 120 mmHg, SD 8 mmHg, CV = 6.7%) vs a blood glucose measurement (mean 5.5 mmol/L, SD 0.3 mmol/L, CV = 5.5%), the raw SDs (8 vs 0.3) are not comparable because they are in different units. The CVs (6.7% vs 5.5%) are directly comparable, telling you that blood pressure is slightly more variable relative to its scale. CV is widely used in quality control, analytical chemistry, biological research, finance, and any field where comparing relative variability across different measurement contexts is needed.
• How does CV differ from standard deviation? ▼
  Standard deviation (SD) and coefficient of variation (CV) both measure spread in a dataset, but they answer different questions. SD answers 'how spread out are the values in absolute terms, in the same units as the data?' If heights have SD = 5 cm, values typically fall within ±5 cm of the mean. CV answers 'how spread out are the values relative to their average?' A height CV of 3% means the typical spread is 3% of the average height — a scale-free, proportional measure. The critical difference emerges when comparing across groups. Consider two manufacturing processes: Process A makes bolts with mean diameter 10 mm, SD 0.5 mm. Process B makes rods with mean diameter 100 mm, SD 0.5 mm. The SDs are equal, but Process A has CV = 5% while Process B has CV = 0.5%. Process B is far more precise relative to its target size. If you only compared SDs, you would incorrectly conclude they have equal precision. SD is appropriate for absolute comparisons within the same scale; CV is appropriate for relative comparisons across different scales or when the mean differs substantially between groups.
• When should I use CV instead of standard deviation? ▼
  Use CV instead of standard deviation in three main situations. First, when comparing variability across datasets with different units or very different means — CV makes the comparison scale-free. For example, comparing measurement precision of a microgram scale (mean 50 µg, SD 2 µg) vs a kilogram balance (mean 50 kg, SD 0.1 kg): CV allows a meaningful precision comparison even though the units are incomparable. Second, when the standard deviation scales proportionally with the mean — common in biological and financial data where larger values naturally have larger spread. In this case, CV is roughly constant across different ranges, making it the natural dispersion metric. Third, in analytical chemistry and quality control, where CV (or RSD) is the standard metric for method precision — laboratory guidelines specify maximum acceptable CVs (typically < 5% for good precision, < 15% for acceptable). Use SD instead of CV when means are near zero (CV becomes unstable or undefined), when the data contains negative values that make interpretation confusing, when you need absolute deviation for practical purposes (like tolerance bands in engineering), or when comparing within a single group where scale is constant.
• What's a 'good' CV value in research vs business? ▼
  Good CV values vary dramatically by field and application. In analytical chemistry and clinical laboratory science, CV < 5% is considered excellent precision, CV 5-10% is acceptable, and CV > 15% suggests a precision problem requiring investigation. Assay validation guidelines (including ISO 5725 and FDA bioanalytical guidance) typically require CV < 15% for accuracy within batches and < 20% across batches. In financial markets, CV is used to compare risk-adjusted return across assets — lower CV means more return per unit of risk. A stock with mean return 10% and SD 20% has CV = 200%, while one with mean return 10% and SD 5% has CV = 50% (lower risk relative to return). In agriculture and biological research, CVs of 10-20% for field experiments are common and acceptable, while CVs over 30% may indicate poor experimental control. In manufacturing quality control, CVs under 2% are expected for precision machining, while 5-10% may be acceptable for less critical processes. There is no universal 'good' CV — what matters is whether the CV meets the precision requirements of the specific application.
• Why cannot CV be calculated for data with zero or negative mean? ▼
  CV is mathematically undefined when the mean is zero because it requires dividing by the mean: CV = SD / mean. Division by zero is undefined. Practically, even when the mean is near zero but not exactly zero, CV becomes extremely large and unstable — a tiny change in the mean produces a huge change in CV, making it a meaningless metric. The conceptual problem is equally important: CV measures variability relative to the magnitude of the quantity being measured. When the mean is zero, there is no meaningful magnitude to relate the variability to. For example, if a process produces measurements with mean 0 and SD 5, what does a CV of infinity or 500% tell you? Very little — it is not interpretable. With negative means, CV has an additional sign ambiguity problem. CV = SD / mean would be negative (SD is always non-negative), which does not correspond to the intuitive idea of a percentage of variability. For data with zero or negative mean, alternatives include: reporting SD directly in the original units, using the mean absolute deviation, expressing variability as a range or interquartile range, or shifting the data so the mean becomes positive (if there is a natural minimum). In quality control, when measuring deviation from zero (e.g., alignment errors), reporting absolute SD in measurement units is clearer than trying to use CV.

Related Calculators