Standard Deviation Calculator — sample & population, step by step

The standard deviation calculator on this page does more than spit out one number — it shows you the mean, the variance, the exact sum of squared deviations, and a full row-by-row working table so you can see precisely how each value contributed to the final result. Toggle between sample and population standard deviation depending on what your data represents, and watch every figure update instantly.

Standard deviation is one of those statistics that gets used constantly and understood rarely. This tool exists to close that gap: enter any list of numbers, choose sample or population, and get a transparent, fully worked answer rather than a black-box result. Arb Digital maintains this standard deviation calculator alongside a growing set of free tools for students, analysts, and small businesses.

What This Standard Deviation Calculator Does

Paste in any list of numbers and the calculator finds the mean, then measures how far each individual value sits from that mean, squares each of those distances, sums them, and divides by either n or n − 1 depending on whether you tell it your data is a full population or a sample drawn from a larger population. The square root of that result is your standard deviation. Every intermediate step — the mean, each deviation, each squared deviation — is displayed openly in the working table beneath the results, so you (or a student you're helping) can follow the logic from raw numbers to final answer.

How to Use It

1. Enter your data. Type or paste your numbers, separated by commas, spaces, or line breaks.
2. Choose sample or population. If your list represents every member of the group you care about, choose population. If it's a subset used to estimate a larger group, choose sample — this is the far more common real-world case.
3. Click Calculate. The standard deviation, variance, mean, and sum of squares all update immediately.
4. Review the working table. Each row shows one value, its distance from the mean, and that distance squared, so you can verify the sum yourself.
5. Compare sample vs. population. Switch the dropdown to see how the same data produces a slightly different answer depending on which denominator you use.

The Formula: Sample vs. Population Standard Deviation

Both versions start the same way. First find the mean of your dataset by adding every value and dividing by the count. Then, for every value x, compute (x − mean) and square it — squaring removes negative signs so deviations on either side of the mean don't cancel out. Add up all of those squared deviations to get the sum of squares.

From here the two formulas split. Population standard deviation divides the sum of squares by n, the total count, then takes the square root: σ = √(Σ(x − mean)² / n). Sample standard deviation instead divides by n − 1, then takes the square root: s = √(Σ(x − mean)² / (n − 1)). Dividing by the smaller number, n − 1, produces a slightly larger result, which corrects for the fact that a sample tends to underestimate the true spread of the full population it was drawn from — a correction statisticians call Bessel's correction. For a rigorous treatment of both formulas, see the NIST Engineering Statistics Handbook.

Sample vs. Population: Which Should You Use?

This is the single most common point of confusion with the standard deviation calculator, so it's worth being explicit. Use population standard deviation only when your dataset genuinely contains every member of the group you're studying — every student in a specific class's final exam scores, every product manufactured in a single completed batch, every day of a month that has already finished. In these cases there's nothing left out; the numbers you have are the entire population, full stop.

Use sample standard deviation whenever your data is a subset drawn from a larger group you're trying to draw conclusions about — a survey of 200 customers meant to represent your entire customer base, a handful of test measurements standing in for an entire production run, or a random sample of website visitors used to estimate behavior across all visitors. Because a sample almost never captures every possible extreme value that exists in the full population, dividing by n − 1 instead of n slightly inflates the result to compensate, giving a more honest estimate of the true population's variability. When in doubt, default to sample — it's the more common and more conservative choice, and it's what tools like Excel's STDEV.S function and most statistical software use by default.

Why Squaring the Deviations Matters

It's tempting to ask why we square each deviation instead of just averaging the raw distances from the mean. The answer is that raw deviations always sum to exactly zero — values above the mean and values below it cancel each other out perfectly, every single time, no matter how spread out the data actually is. Squaring converts every deviation to a positive number, so distances add up meaningfully instead of canceling. It also has a useful side effect: squaring amplifies larger deviations more than smaller ones, so a single wildly extreme value has an outsized effect on standard deviation, which is exactly the behavior you want from a measure of spread — a dataset with one huge outlier should register as more variable than one without.

Reading Variance and Standard Deviation Together

Variance is simply standard deviation before the final square root — it's the average (or n − 1-adjusted average) of the squared deviations, expressed in squared units. If your data is measured in dollars, variance is in "dollars squared," which is hard to interpret intuitively. Taking the square root brings the units back to something meaningful, which is why standard deviation, not variance, is the number most people quote and compare. Still, variance is worth watching: it's the figure that feeds directly into more advanced statistics like confidence intervals and hypothesis tests, so this calculator always shows both side by side.

Common Mistakes to Avoid

• Using population SD on a sample. Dividing by n instead of n − 1 understates variability when your data is only a subset of a larger group.
• Forgetting to square before summing. Summing raw (x − mean) deviations always gives zero — squaring is not optional.
• Skipping the square root. Variance and standard deviation are not the same number; forgetting the final square root gives you variance, not SD.
• Confusing standard deviation with range. Range only looks at the two extreme values. Standard deviation accounts for every value's distance from the mean.
• Rounding the mean too early. Round only the final answer — rounding the mean before computing deviations introduces small but avoidable errors.
• Assuming a low standard deviation means "good." Whether low variability is desirable depends entirely on context — consistent manufacturing tolerances are good, but a lack of variation in survey answers might signal a flawed question.

Related Free Tools From Arb Digital

Pair this calculator with our Mean Median Mode Calculator to see the full center-and-spread picture of your dataset. Working with fractions or ratios elsewhere in your analysis? Try the Fraction Calculator or Ratio Calculator. Need a quick percentage instead? Use our Percentage Calculator. Browse everything we offer in the free online tools hub.

What Standard Deviation Actually Means: The 68-95-99.7 Rule

A standard deviation on its own — say, "2.14" — doesn't mean much until you connect it back to the shape of the data it came from. For data that follows a roughly normal, bell-shaped distribution (think adult heights or standardized test scores), statisticians rely on a rule of thumb called the 68-95-99.7 rule, also known as the empirical rule. It states that about 68% of all values fall within one standard deviation of the mean, about 95% fall within two standard deviations, and about 99.7% — nearly everything — falls within three standard deviations. So if a class's test scores have a mean of 75 and a standard deviation of 8, roughly 68% of students scored between 67 and 83, roughly 95% scored between 59 and 91, and it would be genuinely unusual for anyone to score below 51 or above 99.

This rule is what turns an abstract number into something you can reason about. A small standard deviation relative to the mean tells you most values are clustered tightly together — a manufacturing process with a small SD on part dimensions is running consistently, and a class with a small SD on scores had students performing similarly to one another. A large standard deviation tells you the opposite: values are spread wide, whether that's a process drifting out of tolerance or a class with a huge gap between its strongest and weakest performers. The empirical rule is an approximation that applies specifically to roughly normal distributions; if your data is heavily skewed, the same percentages won't hold exactly, but the broader intuition — smaller SD means tighter clustering, larger SD means wider spread — still holds regardless of shape.

Why Sample vs. Population Actually Matters in Practice

The difference between dividing by n and dividing by n − 1 can look like a rounding footnote, but it has real consequences once a standard deviation feeds into anything downstream, like a confidence interval or a quality-control tolerance band. Bessel's correction — the technical name for using n − 1 — exists because a sample's own mean is calculated from that same sample, which makes the sample slightly "tuned" to itself. Deviations measured from a sample's own mean will, on average, be a little smaller than deviations measured from the true population mean, because the sample mean is, by construction, the value that minimizes those very deviations. Dividing by the smaller number n − 1 nudges the result upward just enough to correct for that built-in underestimate, giving a more honest guess at how spread out the full population really is.

In practice, the gap between sample and population standard deviation shrinks as sample size grows — dividing 50 by 49 barely changes anything, but dividing 5 by 4 changes the answer by a meaningful margin. That's why the distinction matters most with small samples: a pilot study with eight participants, an early batch of five manufactured parts, or a handful of quiz scores. Get the choice wrong on a small sample and you can meaningfully understate how much true variability exists, which matters if that number is about to inform a business decision or a scientific conclusion someone else will rely on.

Frequently Asked Questions

This tool runs entirely in your browser using built-in JavaScript. Nothing you enter is uploaded, stored, or sent to any server.