When should I use n − 1 vs n?

Use n − 1 when your data is a sample from a larger population (almost all real-world cases). Use n only when the data is the entire population.

Why does Bessel's correction (n − 1) exist?

Dividing by n underestimates variance because the sample mean is closer to the data than the unknown true mean. Dividing by n − 1 corrects that bias on average.

What is the standard error of the mean (SEM)?

SEM = s / √n. It estimates how variable the sample mean would be if you repeated the experiment many times.

Yes — if every value is identical, all deviations are zero, so both s and σ are zero.

How are non-numeric tokens handled?

They are skipped and the calculator displays how many tokens were ignored, so partial data does not silently inflate or shrink the result.

Standard Deviation Calculator

Compute sample SD (s, divisor n−1) and population SD (σ, divisor n) with full intermediate steps.

Written by Golam Rabbani, Founder & Lead Engineer

This interactive tool requires JavaScript. Read the formula and worked example below; you can compute the result by hand.

How to use this standard deviation calculator

Paste or type your numbers, separated by commas, spaces, or newlines.
Tick "Show step-by-step work" if you want to see each (x − mean)² contribution.
Press Calculate to get sample SD (s, divisor n − 1) and population SD (σ, divisor n) side by side.
Copy the result or Reset to start over.

About this standard deviation calculator

Standard deviation summarises how spread out a list of numbers is around its mean. Compute it by subtracting the mean from each value, squaring those deviations, summing them (call the total SS), then dividing by either n (population formula, σ²) or n − 1 (sample formula, s²). Taking the square root gives σ or s. Use the sample formula whenever the data is a sample from a larger population; the n − 1 correction (Bessel's correction) keeps the estimate unbiased.

Worked example. Enter 2, 4, 4, 4, 5, 5, 7, 9 with n = 8. Mean = 40 / 8 = 5. The squared deviations (x − 5)² are 9, 1, 1, 1, 0, 0, 4, 16 — sum SS = 32. Sample variance s² = 32 / (8 − 1) = 32 / 7 ≈ 4.5714, so s = √4.5714 ≈ 2.1381. Population variance σ² = 32 / 8 = 4, so σ = 2. The standard error of the mean SEM = s / √n ≈ 2.1381 / √8 ≈ 0.7559. Step-by-step work is available for inspection in the result panel.

FAQ

When should I use n − 1 vs n?: Use n − 1 when your data is a sample from a larger population (almost all real-world cases). Use n only when the data is the entire population.
Why does Bessel's correction (n − 1) exist?: Dividing by n underestimates variance because the sample mean is closer to the data than the unknown true mean. Dividing by n − 1 corrects that bias on average.
What is the standard error of the mean (SEM)?: SEM = s / √n. It estimates how variable the sample mean would be if you repeated the experiment many times.
Can SD be zero?: Yes — if every value is identical, all deviations are zero, so both s and σ are zero.
How are non-numeric tokens handled?: They are skipped and the calculator displays how many tokens were ignored, so partial data does not silently inflate or shrink the result.