What does standard deviation measure?

Standard deviation measures how spread out a set of numbers is from the mean. A low standard deviation means most values cluster tightly around the average. A high standard deviation means the values are scattered across a wide range. It's the most common way to quantify consistency or variability in data.

How do you calculate standard deviation?

First, find the mean of the dataset. Then for each value, subtract the mean and square the result. Average all those squared differences — that's the variance. Take the square root of the variance to get the standard deviation. Formula: σ = sqrt((Σ(x - μ)²) / N) for a population, or use N-1 in the denominator for a sample.

What is a good standard deviation?

There is no universal 'good' value — it depends entirely on context. In test scores with a mean of 80, a standard deviation of 5 indicates tight clustering (most students scored 75-85), while 15 indicates wide spread (scores from 50 to 110 are common). In manufacturing, lower standard deviation means more consistent output. Compare standard deviation to the mean to judge relative variability.

What is the difference between standard deviation and variance?

Variance is the average of the squared differences from the mean. Standard deviation is the square root of variance. Variance is in squared units (e.g., dollars²), which is hard to interpret. Standard deviation is in the same units as the original data (e.g., dollars), making it more intuitive. Use standard deviation for reporting; use variance when working through formulas in statistics (like ANOVA) because it has nicer algebraic properties.

When should I use population standard deviation versus sample standard deviation?

Use population standard deviation (divide by N) when you have data for every member of the group you care about — for example, test scores for every student in a single class. Use sample standard deviation (divide by N-1) when your data is a subset drawn from a larger population and you're trying to estimate the population's spread. The N-1 correction (Bessel's correction) compensates for the fact that sample variance tends to underestimate population variance.

Home › Articles › Standard Deviation

Standard Deviation: What It Means and When to Use It

Standard deviation tells you whether a dataset is bunched tight around the average or scattered all over the place. Low standard deviation means consistency — manufacturing tolerances, test scores clustered near the class average, daily temperatures that barely move. High standard deviation means volatility — stock prices, wide variation in customer wait times, a class where half the students aced it and half failed. Here's the formula, a real worked example, and when you want variance instead.

Last updated: July 5, 2026.

What standard deviation actually measures

Standard deviation (σ for population, s for sample) measures the average distance between each data point and the mean. The bigger the standard deviation, the more spread out the data. The smaller, the tighter the cluster.

Two test-score examples:

Class A: mean of 80, standard deviation of 5. Most students scored between 75 and 85. The class performed consistently.
Class B: mean of 80, standard deviation of 15. Scores ranged from 50 to 110. Same average, wildly different distributions.

The mean alone hides that difference. Standard deviation surfaces it. That's why every serious data analysis reports both.

The formula (and why it looks like that)

For a population of N values:

σ = sqrt( (Σ(x - μ)²) / N )

For a sample (when you're estimating from part of a larger population):

s = sqrt( (Σ(x - x̄)²) / (N - 1) )

Breaking it down:

μ (mu) or x̄ (x-bar) — the mean of the dataset
x — each individual value in the dataset
(x - μ)² — the squared difference between each value and the mean. Squaring ensures negative and positive deviations don't cancel out, and it gives more weight to extreme outliers.
Σ — sum all those squared differences
N or (N - 1) — divide by the count to get the average squared deviation (variance). The N-1 version is called Bessel's correction and compensates for the fact that sample variance tends to underestimate population variance.
sqrt — take the square root to bring the units back to the original scale (variance is in squared units, which is impossible to interpret).

The result is a single number in the same units as your data. If you're measuring test scores (0-100), standard deviation is in points. If you're measuring widget lengths in millimeters, standard deviation is in millimeters.

A worked example: test scores

Ten students took a quiz. Their scores:

85, 92, 78, 88, 95, 72, 90, 65, 84, 91

Step 1: Find the mean.

(85 + 92 + 78 + 88 + 95 + 72 + 90 + 65 + 84 + 91) / 10 = 840 / 10 = 84

The average score is 84.

Step 2: For each score, find the deviation from the mean and square it.

Score (x)	x - μ	(x - μ)²
85	1	1
92	8	64
78	−6	36
88	4	16
95	11	121
72	−12	144
90	6	36
65	−19	361
84	0	0
91	7	49
Sum	—	828

Step 3: Find the variance by averaging the squared deviations.

We're treating this as a population (all ten students), so divide by N = 10:

Variance = 828 / 10 = 82.8

Step 4: Take the square root to get the standard deviation.

σ = sqrt(82.8) ≈ 9.1

The standard deviation is 9.1 points. Most students scored within roughly 9 points of the mean (84). The 65 and 95 are outliers — more than one standard deviation away from the center — but the bulk of the class landed in the 75-93 range.

Population versus sample standard deviation

The example above used the population formula (divide by N) because we had all ten quiz scores — the complete dataset we cared about. If instead those ten scores were a random sample from a 200-student class, we'd use the sample formula and divide by N - 1 = 9 instead:

s = sqrt(828 / 9) ≈ 9.6

The sample standard deviation is slightly higher. That's Bessel's correction compensating for the fact that samples tend to underestimate the true population spread. The difference shrinks as sample size grows — on 100 data points, dividing by 99 instead of 100 barely moves the needle.

Rule of thumb: if you're describing data you have in full, use population SD (divide by N). If you're estimating from a subset, use sample SD (divide by N-1). Most scientific calculators and spreadsheet functions default to the sample version.

Standard deviation versus variance

Variance is the step right before you take the square root — it's the average of the squared deviations. In the example above, variance was 82.8. Standard deviation was 9.1.

Variance is mathematically convenient. It has clean algebraic properties and shows up everywhere in advanced statistics (regression, ANOVA, portfolio theory). But it's in squared units — "points squared" for test scores, "dollars squared" for salaries — which is impossible to interpret in real-world terms.

Standard deviation fixes that by taking the square root, which brings the measurement back to the original units. You can say "the standard deviation of test scores is 9.1 points" and everyone knows what that means. Saying "the variance is 82.8 points squared" prompts blank stares.

Use variance when you're working through formulas. Use standard deviation when you're communicating results.

When low standard deviation is good (and when it isn't)

In most contexts, low standard deviation signals consistency or predictability:

Manufacturing: A machine stamping out widgets with a target length of 50mm and a standard deviation of 0.1mm is producing near-identical parts. A standard deviation of 5mm means the machine is broken or the process is out of control.
Finance: Two mutual funds with the same 8% annual return but different standard deviations are not equally risky. The fund with a 2% standard deviation had stable, predictable growth. The fund with a 12% standard deviation swung wildly year to year — it hit 8% on average, but some years lost 10% and others gained 25%.
Medicine: A drug trial measuring blood pressure reduction wants a low standard deviation in the treatment group — it means the drug works consistently across patients. High standard deviation means it works great for some people and does nothing (or harms) others.

But in some cases, high standard deviation is the point:

Innovation: A startup accelerator measuring revenue growth across cohorts expects high standard deviation. Most startups fail (zero or negative growth), a few 10x, and one occasionally 100x. Low standard deviation would mean the accelerator is picking safe, boring companies.
Sports: A basketball player with a scoring average of 20 points per game and a standard deviation of 2 is reliable but boring. A player with the same average and a standard deviation of 8 is a boom-or-bust scorer who can take over a game or go cold — higher risk, higher ceiling.

Context decides whether tight clustering is a feature or a bug.

The 68-95-99.7 rule (for normal distributions)

If your data follows a normal distribution (bell curve), standard deviation has a very clean interpretation:

About 68% of values fall within 1 standard deviation of the mean (μ ± σ)
About 95% fall within 2 standard deviations (μ ± 2σ)
About 99.7% fall within 3 standard deviations (μ ± 3σ)

On the test-score example: mean of 84, standard deviation of 9.1. If the scores were normally distributed (they're close), you'd expect:

68% of students scored between 74.9 and 93.1 (84 ± 9.1)
95% scored between 65.8 and 102.2 (84 ± 18.2)
99.7% scored between 56.7 and 111.3 (84 ± 27.3)

This rule is why "three-sigma events" get attention in manufacturing and finance. If your process is stable and something happens that's three standard deviations out from the mean, either you just witnessed a 1-in-370 fluke or something fundamental changed. Most of the time, it's the latter.

How to calculate standard deviation on a calculator

Every scientific calculator since the 1980s has a built-in standard deviation function. The key is usually labeled σ_n (population) or σ_n-1 (sample), sometimes hidden under a STAT or DATA menu.

General procedure (varies by model):

Clear any old data (usually 2nd + CLR or similar).
Enter each data point followed by the Σ+ or DATA key to add it to the list.
When all values are entered, press the σ key (or navigate to it in the STAT menu).
The display shows the standard deviation. Check whether your calculator defaulted to population or sample — if you need the other one, there's usually a toggle or separate key.

On the scientific calculator, the STAT mode handles this in a few keystrokes. Enter the dataset, compute mean and standard deviation, and you're done. Spreadsheets use =STDEV.S(range) for sample standard deviation and =STDEV.P(range) for population.

Frequently asked questions

Can standard deviation be negative?

No. Standard deviation is the square root of variance, and variance is the average of squared values — which are always non-negative. The smallest possible standard deviation is zero, which happens only when every value in the dataset is identical (no spread at all).

What does it mean if standard deviation is larger than the mean?

It means the data is extremely spread out, often with a long tail or skew. For example, household income in the US has a mean around $75,000 and a standard deviation over $100,000 because a small number of ultra-high earners pull the distribution into a long right tail. This is common in right-skewed data (income, wealth, company revenue) and signals that the mean is being distorted by outliers — median is often a better summary statistic in these cases.

How do outliers affect standard deviation?

Outliers inflate standard deviation dramatically because deviations are squared before averaging. A single extreme value can double the standard deviation of a small dataset. If you're measuring process consistency or trying to understand typical variation, consider removing obvious outliers or using a robust measure of spread like the interquartile range (IQR), which ignores the top and bottom 25% of data.

When should I use standard deviation versus range?

Range (max minus min) is simpler but fragile — it's determined entirely by the two most extreme values and ignores everything in between. Standard deviation uses every data point and is much more stable. Use range for quick gut-checks or when you genuinely care only about the extremes (e.g., temperature swing from coldest to hottest). Use standard deviation for serious analysis.

What is coefficient of variation and when do I use it?

Coefficient of variation (CV) is standard deviation divided by the mean, expressed as a percentage. It measures relative variability and is useful when comparing datasets with very different scales. Example: a manufacturing process producing 10mm widgets with a 0.5mm standard deviation (CV = 5%) is less consistent relative to size than a process producing 100mm widgets with a 2mm standard deviation (CV = 2%). Use CV when the units or scales differ; use raw standard deviation when they're comparable.

Try it yourself

Grab any dataset — test scores, daily temperatures, your last ten grocery receipts — and run the calculation by hand once to see how the formula works. Then use the scientific calculator's STAT mode to check your work. Once you've done it manually a couple of times, the intuition clicks: standard deviation is just the average distance from the mean, with squaring and square-rooting to keep the math clean. After that, let the calculator do the arithmetic.

This article is for general education and is not statistical advice. See our Terms.