Compare spread relative to the mean using sample standard deviation divided by the absolute mean.
Use descriptive context first, choose the method that matches the design, then keep inputs, settings and result evidence reproducible instead of copying a p-value without its analysis recipe.
The existing CV result stays first. This audit layer reuses the current raw or summary inputs, lets you choose the SD convention explicitly, reports variance and RSD, and shows the substituted calculation so a second analyst can reproduce the percentage.
CV is undefined at mean 0 and can be unstable or conceptually inappropriate near zero or on interval scales.
Use plots and descriptive summaries before formal tests. Check the sampling/design assumptions that matter for the chosen method, report effect size and uncertainty where available, and avoid treating a threshold such as p < .05 as proof of importance, causation, or truth. Browser calculations are educational/planning utilities, not domain-specific professional advice.
Use the tool first, then apply these checks to verify inputs, interpret the result, and hand it off without displacing the primary workflow.
CV inherits the standard-deviation convention used in the numerator. Use sample SD when the observations are a sample and population SD only when they represent the complete population of interest. Mixing n−1 and N denominators changes the reported CV, especially for small datasets, so record the basis with the result.
CV is most interpretable for ratio-scale measurements with a meaningful zero. It can be misleading for interval scales such as Celsius temperature, for variables that can legitimately cross zero, or when the mean is close to zero. In those cases the denominator can make the percentage unstable or conceptually meaningless even when the arithmetic is valid.
A CV without its underlying mean, SD, sample size, and calculation basis is difficult to audit. Keep those values together so another analyst can reproduce the percentage and understand the scale of the data. RSD is commonly the same relative quantity expressed as a decimal or percentage, depending on domain convention.
Whether 5%, 20%, or another CV is acceptable depends on the measurement system and domain. Laboratory precision, biological variation, process control, and financial comparisons use different expectations. Interpret the result against a domain-specific tolerance or benchmark rather than attaching a universal quality label to the percentage. For repeated measurements, also inspect outliers, skew, detection limits, censoring, instrument resolution, and batch effects because two datasets can share the same CV while having very different distributions and data-quality problems.
WebToolArc