Free pooled standard deviation calculator: enter each group's sample size, mean and standard deviation to get one combined spread and its degrees of freedom.
This calculator combines the spread of two or three independent samples into a single pooled standard deviation. Enter the number of observations, the mean and the sample standard deviation for each group; the tool returns the pooled standard deviation, the pooled variance, the grand mean and the degrees of freedom behind the estimate.
A pooled standard deviation answers a different question from an ordinary one. It is not the spread of all the observations thrown together, but an estimate of the spread inside the groups after the group means are taken out. That is why pooling works even when the groups sit at very different levels: shifting one group's mean changes the grand mean, never the pooled spread.
Each group enters the average weighted by its degrees of freedom, n − 1, not by its sample size. A group of 30 therefore pulls roughly three times as hard as a group of 10, while a group with two observations counts for a single degree of freedom. The result is the standard deviation used by the classic pooled two-sample t-test, by ANOVA as the root mean square within groups, and by meta-analyses of continuous outcomes.
Pooling assumes the samples estimate one common within-group standard deviation, so check that the group spreads are in the same ballpark before trusting the number. When one group is several times more variable than another, the pooled value hides that difference and Welch's separate-variance approach is the honest choice. Pooling is also no rescue for a tiny group: two observations add one degree of freedom and very little information.
Enter the sample standard deviation, not the standard error. If a source reports a standard error, multiply it by the square root of that group's sample size first. The result carries the same unit as your data.
It is one estimate of the spread inside several groups, formed by averaging their variances with degrees-of-freedom weights and taking the square root. It describes within-group scatter only: the gaps between the group means are deliberately left out. With 12, 10 and 8 observations and standard deviations of 3.2, 4.1 and 2.8 the pooled value is 3.4362, close to but not equal to the plain average of the three.
When the groups differ a lot in variability, or when they clearly come from different populations: the pooled number then describes no group well. A quick check is the ratio of the largest to the smallest variance. Below 2 is comfortable, 2 to 4 is borderline, and above 4 stop pooling. With unequal spreads use Welch's t-test, which keeps each group's own standard deviation instead of forcing one.
The formula has no three-group limit: each extra group adds one term to the numerator and one to the denominator. This page accepts two or three at a time, and because pooling is associative you can work in stages. Pool the first two groups, then re-enter that result as a group whose sample size is its degrees of freedom plus one, alongside the next group.
Because the two numbers answer different questions. The standard deviation of all observations also counts the gaps between the group means as spread, so it grows whenever the groups sit at different levels. The pooled standard deviation removes those means first and measures only the scatter around them, which is exactly what a pooled t-test or a within-group repeatability estimate needs.