About the Pooled Standard Deviation
Pooled standard deviation combines the variation from two independent samples into a single weighted estimate, under the assumption that both samples come from populations with equal variance. This calculator takes each sample's standard deviation and size and returns both the pooled variance and the pooled standard deviation. It's the standard input for equal-variance two-sample t-tests and for standardized effect sizes like Cohen's d.
How It Works
Enter the standard deviation and sample size for each of the two samples. The calculator converts each standard deviation to a variance, weights it by its degrees of freedom (sample size minus one), sums the two weighted variances, and divides by the combined degrees of freedom. Taking the square root of that pooled variance gives the pooled standard deviation.
Formula & Methodology
Each sample contributes to the pooled estimate in proportion to its degrees of freedom rather than being averaged equally, so a larger sample pulls the combined estimate more toward its own variance. When both samples happen to be the same size, the weighting cancels out and the pooled variance reduces to a simple average of the two sample variances.
Examples
Two Unequal Sample Sizes
With Sample 1 at a standard deviation of 12 across 30 observations and Sample 2 at a standard deviation of 14 across 25 observations, the pooled variance comes to about 167.55, giving a pooled standard deviation of about 12.94.
Two Equal Sample Sizes
With both samples sized at 20, one with a standard deviation of 5 and the other with a standard deviation of 7, the pooled variance is exactly the average of 25 and 49, or 37.0, giving a pooled standard deviation of about 6.08.
Advantages
- Weights each sample by its degrees of freedom automatically, avoiding the common shortcut of simply averaging the two standard deviations, which is not mathematically valid.
- Produces both the pooled variance and the pooled standard deviation in one step, both of which are needed at different points in equal-variance hypothesis testing.
- Works from summary statistics alone (standard deviation and sample size), so you don't need the raw data from either sample on hand.
Common Mistakes
- Averaging the two standard deviations directly, like (12 + 14) / 2 - because variance is the quantity that combines additively, not standard deviation, this shortcut gives a mathematically incorrect result.
- Applying pooled standard deviation to two samples with clearly different amounts of variability, when the equal-variance assumption behind the formula doesn't hold for that data.
- Entering variance values into fields meant for standard deviation - since the formula squares the input, an accidental variance-for-standard-deviation swap inflates the pooled result substantially.
Edge Cases to Watch For
- Each sample size must be greater than 1, since a sample of size 1 has zero degrees of freedom and no defined variance to contribute.
- This calculation assumes both populations share the same underlying variance - if the two sample standard deviations differ substantially, pooling them can mask a real difference in variability, and an approach that doesn't assume equal variances, such as Welch's t-test, may be more appropriate.
- With equal sample sizes, the pooled variance is exactly the unweighted average of the two sample variances, since the degrees-of-freedom weighting becomes identical for both.
Common Use Cases
- Researchers setting up an equal-variance (Student's) two-sample t-test that requires a single pooled standard deviation as input.
- Analysts calculating Cohen's d to report a standardized effect size between two groups.
- Anyone combining variability estimates from two independent samples, such as in a meta-analysis, into one weighted estimate of spread.