Overview
A pooled standard deviation combines the spread of two groups into one estimate, assuming both groups have about the same true spread. It is used in t-tests and in Cohen’s d.
Enter each group’s sample size and standard deviation.
Worked examples
| Inputs | Result |
|---|---|
| Group 1 sample size30Group 1 standard deviation4.2Group 2 sample size25Group 2 standard deviation5.1 | Pooled standard deviation: 4.6293Pooled variance: 21.4302Degrees of freedom: 53 |
Show 2 more examples
| Group 1 sample size10Group 1 standard deviation3Group 2 sample size10Group 2 standard deviation4 | Pooled standard deviation: 3.5355Pooled variance: 12.5Degrees of freedom: 18 |
| Group 1 sample size100Group 1 standard deviation12Group 2 sample size20Group 2 standard deviation15 | Pooled standard deviation: 12.5317Pooled variance: 157.0424Degrees of freedom: 118 |
Pooled standard deviation by group 2 standard deviation
With the calculator's starting values, the pooled standard deviation is 4.6293. The table shows how it changes with the group 2 standard deviation.
| Group 2 standard deviation | Pooled standard deviation |
|---|---|
| 3 | 3.7051 |
| 4 | 4.1106 |
| 5 | 4.5796 |
| 5.1starting value | 4.6293 |
| 6 | 5.0945 |
| 7 | 5.6428 |
Other inputs: Group 1 sample size: 30 · Group 1 standard deviation: 4.2 · Group 2 sample size: 25. Change them in the calculator above for your own numbers.
How it works
Each variance is weighted by its degrees of freedom, n − 1, so larger groups count for more. The weighted variances are added, divided by the total degrees of freedom, and square-rooted.
The formula
| n₁, n₂ | Group sample sizes |
| s₁, s₂ | Group standard deviations |
Practical tips
- Groups of 30 and 25 with SDs of 4.2 and 5.1 give a pooled SD of 4.63.
- With equal group sizes, it is the square root of the average variance.
- If one SD is more than about twice the other, pooling may not be appropriate. Consider Welch’s t-test.
Frequently asked questions
Where is the pooled SD used?
In the two-sample t-test with equal variances and in effect sizes such as Cohen’s d.
Is it the same as averaging the two SDs?
No. It averages the variances, weighted by sample size, so the result usually differs.