Standard deviation summarizes how spread out values are around their mean. A small value means observations are relatively clustered; a larger value means they are more dispersed. The number only makes sense in the units and context of the data, so it should be interpreted alongside the mean, sample size and distribution.
Start with the mean
Standard deviation measures distances from the arithmetic mean. If the mean itself is not meaningful for a dataset—such as categories or heavily skewed values—standard deviation may also be a poor summary on its own.
Keep the units in mind
If the data is measured in seconds, the standard deviation is also expressed in seconds. This makes it possible to compare spread with the scale of the original measurements.
Population and sample formulas differ
A population calculation describes the complete set you care about. A sample calculation uses a correction when the values are treated as a sample from a larger population. Choose the version that matches the statistical question, not whichever produces the smaller number.
Outliers can increase spread strongly
Because deviations are squared during the calculation, extreme values can have a large effect. Inspect the raw data or a plot rather than relying on one summary statistic.
Do not compare standard deviations blindly across different scales
A standard deviation of 10 may be tiny for values around 10,000 and huge for values around 15. Consider the mean, measurement units and domain before calling variability “high” or “low.”
Use it with other summaries
Report the number of observations, mean and often the range or a distribution plot. For skewed data, median and percentile-based summaries may explain the pattern better.
Worked example
Example
Two delivery routes can both average 30 minutes. If one has a standard deviation of 2 minutes and the other 12 minutes, the second route is much less consistent even though the average is the same.
Practical checklist
Before you finish
- Choose sample or population mode intentionally.
- Interpret the value in the original units.
- Inspect outliers.
- Compare spread with the scale of the mean.
- Use additional summaries for skewed data.
Common mistakes
What to avoid
- Calling any standard deviation above 1 “high.”
- Mixing sample and population formulas without stating which was used.
- Ignoring outliers.
- Comparing datasets measured in different units as if the raw standard deviations were directly comparable.