Data

Standard Deviation vs Variance: When to Use Which

Standard deviation and variance both measure how spread out data is. They are calculated from the same steps — variance is simply the standard deviation squared — yet they are used in different places. Knowing which to use, and whether to use the sample or population version, avoids a lot of confusion in reports and analysis.

Variance: the average squared distance from the mean

To calculate variance:

Find the mean.
Subtract the mean from each value.
Square each difference (so negatives do not cancel positives).
Add the squared differences.
Divide by n − 1 for a sample or by n for a whole population.

Standard deviation: variance in the original units

Take the square root of the variance and you get the standard deviation (SD). The square root brings the measure back to the same units as the data. If your data is in dollars, variance is in “dollars squared” — hard to interpret — while SD is in dollars.

A worked example

Daily support tickets over a week: 12, 15, 11, 18, 14, 16, 13.

Mean = 99 ÷ 7 ≈ 14.14
Sum of squared differences ≈ 34.86
Sample variance = 34.86 ÷ 6 ≈ 5.81 (tickets²)
Sample SD = √5.81 ≈ 2.41 tickets

The SD says a typical day is about 2–3 tickets away from the average of 14 — immediately useful for staffing. The variance of 5.81 “tickets squared” is mathematically correct but tells a manager nothing. Check your own data with the standard deviation calculator or the variance calculator.

When to use standard deviation

Describing data to people: “average delivery time 3.2 days, SD 0.8 days.”
Setting ranges and thresholds: for roughly bell-shaped data, about 68% of values fall within 1 SD of the mean and about 95% within 2 SD.
Z-scores: how many SDs a value is from the mean. See our z-score and p-value guide.
Confidence intervals and margins of error, which are built from SD.
Investment risk (volatility), quoted as the SD of returns.

When to use variance

Combining independent sources of variation. Variances add; standard deviations do not. If two independent steps in a process have variances of 4 and 9, total variance is 13 and total SD is √13 ≈ 3.6 — not 2 + 3 = 5.
Statistical methods such as ANOVA and regression, which work with variances internally.
Portfolio theory, where variances and covariances of assets are combined.

In short: calculate with variance, communicate with standard deviation.

Sample or population?

This choice trips people up more than the SD-vs-variance question.

Population (divide by n): you have every value in the group you care about — for example, the final scores of all 30 students in a class, and you only care about that class.
Sample (divide by n − 1): your data is a subset used to learn about a bigger group — survey responses, a week of tickets used to plan the year, a batch of products tested from a production run.

Dividing by n − 1 (Bessel’s correction) slightly increases the estimate to correct for the fact that a sample tends to look less spread out than the population it came from. In business analysis you almost always want the sample version. In spreadsheets, that is STDEV.S and VAR.S; STDEV.P and VAR.P are the population versions.

Things to watch for

Outliers inflate both measures heavily because differences are squared. One delivery that took 95 hours instead of 50 can multiply the SD several times. Look at the data before trusting the number.
Skewed data such as income or order value: SD is still valid, but the 68–95 rule does not apply. Consider the median and percentiles as well.
Comparing spread across different scales: use the coefficient of variation (SD ÷ mean) — an SD of $10 is large for $20 items and tiny for $2,000 items.

Key takeaways

Variance = average squared distance from the mean; SD = its square root, in the data’s units.
Use SD to describe and communicate spread; use variance when combining sources of variation or inside statistical methods.
Use the sample formulas (n − 1) unless you truly have the whole population.
Check for outliers and skew before relying on either.