Skip to content
Statculator
Descriptive Statistics

Pooled Variance Calculator

Free pooled variance calculator for 2 or more groups: get sₚ², the pooled SD and df from sizes and SDs, variances or raw data, with the formula and steps.

Free, no sign-up Runs in your browserBy Hira Noor, BS MathematicsUpdated

Data input
Groups
#Sample size nStandard deviation sRemove row
1
2
3
Tip: paste rows straight from a spreadsheet.

Results

Pooled variance, sₚ²
8.04
SSW / (N − k) = 192.96 / 24
Pooled standard deviation, sₚ
2.8355
Number of groups, k
3
Total sample size, N
27
Within-group df, N − k
24
Within-group sum of squares, SSW
192.96
Hartley’s F-max (largest ÷ smallest variance)
1.5376
Mean square within, MSW
8.04

Interpretation: The pooled variance 8.04 is the weighted average of the 3 group variances, with weights equal to their degrees of freedom. In a one-way ANOVA it is the mean square within groups, the denominator of the F statistic.

Pooling table

Groupnᵢsᵢ²nᵢ − 1(nᵢ − 1)sᵢ²Weight
Group 186.25743.7529.2%
Group 2109.61986.4937.5%
Group 397.84862.7233.3%
Σ2724192.96100%

Weight = (nᵢ − 1) / Σ(nᵢ − 1), each group’s share of the pooled variance.

Step-by-step solution

  1. 1

    Square each standard deviation to get the variances

    s₁² = 2.5² = 6.25
    s₂² = 3.1² = 9.61
    s₃² = 2.8² = 7.84
  2. 2

    Multiply each variance by its degrees of freedom, nᵢ − 1

    This turns each variance back into a within-group sum of squares.

    (8 − 1) × 6.25 = 43.75
    (10 − 1) × 9.61 = 86.49
    (9 − 1) × 7.84 = 62.72
  3. 3

    Add them to get the within-group sum of squares

    SSW = 43.75 + 86.49 + 62.72 = 192.96
  4. 4

    Divide by the total degrees of freedom

    df = Σ(nᵢ − 1) = N − k = 27 − 3 = 24
    sₚ² = SSW / df = 192.96 / 24 = 8.04
  5. 5

    Read it as the ANOVA mean square within groups

    MSW = SSW / (N − k) = 8.04
    F = MSB / MSW, with df = (2, 24)

What is a Pooled Variance Calculator?

A pooled variance calculator combines the variances of two or more independent samples into one estimate of the population variance they share. The pooled variance formula is sₚ² = Σ(nᵢ − 1)sᵢ² / Σ(nᵢ − 1), so each group is weighted by its degrees of freedom. For groups of 8, 10 and 9 with SDs 2.5, 3.1 and 2.8, sₚ² = 8.04. It is the variance in the pooled two-sample t-test and the mean square within groups (MSW) in one-way ANOVA.

What the Pooled Variance Calculator does

Enter each group’s size with its standard deviation or its variance, or paste raw data. The calculator returns the pooled variance sₚ² (the pooled sample variance, read “s pooled squared”), the pooled SD sₚ, the within-group sum of squares, each group’s weight, Hartley’s F-max ratio as a quick check of the equal-variance assumption, and a step-by-step solution with your numbers.

It works for 2 groups or up to 20 groups from summary statistics (10 groups with raw data). With two groups it also gives the standard error of the difference between means and the critical t, so it doubles as a pooled variance calculator for the t-test. With raw data it shows how sₚ² fits into the ANOVA sum-of-squares decomposition.

Pooled statistics such as sₚ² are worthwhile because every group carries information about the same σ². Combining the groups gives an estimate with N − k degrees of freedom instead of only nᵢ − 1, which makes t-tests and F-tests more powerful, provided the groups really do share one variance. If you need the square root on its own, use the pooled standard deviation calculator.

How to use the Pooled Variance Calculator

  1. Under Data input, choose n and SD, n and variance, or Raw data.
  2. Fill in one row per group (or one box per group for raw data). Use Add group for more groups; each group needs n ≥ 2.
  3. Read the pooled variance and pooled SD at the top of the results. Check the notes: a large F-max ratio means pooling may not be appropriate.
  4. Open the pooling table to see each group’s weight, and the step-by-step solution for the link to the t-test or ANOVA.

How to calculate pooled variance

Calculating pooled variance by hand needs only each group’s sample size nᵢ and its standard deviation sᵢ or variance sᵢ². Follow four steps:

  1. Get each group’s variance. If you have standard deviations, square them: sᵢ² = sᵢ × sᵢ.
  2. Multiply each variance by nᵢ − 1, the group’s degrees of freedom. Each product is that group’s sum of squared deviations from its own mean.
  3. Add the products to get the within-group sum of squares, SSW = Σ(nᵢ − 1)sᵢ².
  4. Divide by the total degrees of freedom, Σ(nᵢ − 1) = N − k. The result is the pooled variance sₚ². Take its square root if you also need the pooled SD.

Pooled variance formula

Pooled variance, k groups
sₚ² = Σ(nᵢ − 1)sᵢ² / Σ(nᵢ − 1) = SSW / (N − k)
Two groups
sₚ² = [(n₁ − 1)s₁² + (n₂ − 1)s₂²] / (n₁ + n₂ − 2)
Two groups of equal size
sₚ² = (s₁² + s₂²) / 2
Pooled t-test standard error
SE(x̄₁ − x̄₂) = √[ sₚ² (1/n₁ + 1/n₂) ]
Formula symbols
SymbolMeaning
sₚ²pooled variance (“s pooled squared”)
sᵢ²sample variance of group i
nᵢsize of group i
N, ktotal sample size, number of groups
SSWwithin-group sum of squares, Σ(nᵢ − 1)sᵢ²

(nᵢ − 1)sᵢ² is exactly the sum of squared deviations of group i around its own mean, so the pooled variance is the total within-group sum of squares divided by its degrees of freedom. That is why it equals MSW in ANOVA.

The pooled variance symbol is sₚ² (also written s²ₚ, s_p² or S²pooled). Some textbooks write the two-group pooled variance equation with denominator n₁ + n₂ − 2; that is N − k with k = 2.

Worked example: pooled variance of three groups

Three fertilizer plots, from sizes and SDs

Problem. Plant heights (cm) from three fertilizers give n = 8, s = 2.5; n = 10, s = 3.1; n = 9, s = 2.8. Find the pooled variance that a one-way ANOVA would use.

  1. 01Variances: 2.5² = 6.25, 3.1² = 9.61, 2.8² = 7.84.
  2. 02Weighted: 7 × 6.25 = 43.75, 9 × 9.61 = 86.49, 8 × 7.84 = 62.72.
  3. 03SSW = 43.75 + 86.49 + 62.72 = 192.96.
  4. 04df = 7 + 9 + 8 = 24 (= 27 − 3).
  5. 05sₚ² = 192.96 / 24 = 8.04 cm², so sₚ = √8.04 = 2.8355 cm.

Answer. The pooled variance is 8.04 cm². In the ANOVA table this is MSW with 24 degrees of freedom. The groups’ weights are 29.2%, 37.5% and 33.3%, and the F-max ratio is 9.61 / 6.25 = 1.54, so pooling is reasonable.

Worked example: pooled variance for a two-sample t-test

Two classes take the same exam. Class A has n₁ = 12, mean 75.2 and s₁ = 8.1; class B has n₂ = 15, mean 68.9 and s₂ = 7.4. Using the two-group pooled variance equation, sₚ² = (11 × 8.1² + 14 × 7.4²) / (12 + 15 − 2) = (721.71 + 766.64) / 25 = 1488.35 / 25 = 59.534.

The pooled t-test standard error is √[59.534 × (1/12 + 1/15)] = 2.9883, so t = (75.2 − 68.9) / 2.9883 = 2.108 with 25 degrees of freedom. That is just above the two-tailed 5% critical value of 2.060, so the difference in means is significant at α = 0.05. The variance ratio is 65.61 / 54.76 = 1.20, well under 4, so pooling is fine.

Classnᵢsᵢsᵢ²nᵢ − 1(nᵢ − 1)sᵢ²
A128.165.6111721.71
B157.454.7614766.64
Σ27251488.35

Pooled variance vs the variance of all data combined

Merging every group into one list and computing a single variance gives a different, usually larger, number. That combined variance measures spread around the grand mean, so it includes the differences between group means. The pooled variance measures spread around each group’s own mean only.

In the raw-data example on this page the group means are 15.5, 21.2 and 14.0. The pooled variance is 60.3 / 15 = 4.02, but the variance of all 18 values merged is 12.9706, because the gap between the means adds to the spread. Use the pooled variance for t-tests and ANOVA; use the merged variance only to describe all observations as one sample.

Pooled variance, sₚ²Combined (merged) variance
Spread measured aroundeach group’s own meanthe grand mean of all values
Includes differences between meansNoYes
Degrees of freedomN − kN − 1
Raw-data example on this page4.0212.9706
Used forpooled t-test, ANOVA, effect sizesdescribing all values as one sample

When is it safe to pool variances?

Pooling is only justified if the populations have the same variance. Formal tests exist (the F-test for two groups; Levene’s test or Bartlett’s test for more), but they have low power in small samples and Bartlett’s test is very sensitive to non-normality. Many analysts use simple rules of thumb instead:

  • Largest variance ÷ smallest variance (Hartley’s F-max) below about 4, equivalent to an SD ratio below 2. The calculator reports this ratio for you; the table below shows how to read it.
  • Group sizes roughly equal. With balanced designs the pooled t-test and ANOVA are fairly robust to unequal variances; with unbalanced designs they are not.
  • If the rules fail, use Welch’s t-test for two groups or Welch’s ANOVA for several.
Largest variance ÷ smallest (F-max)What it suggests
Under 2Variances are similar; pooling is safe.
2 to 4Pooling is usually fine, especially with equal group sizes.
Over 4 (SD ratio over 2)Equal variances are doubtful; use Welch’s t-test or Welch’s ANOVA.

Pooled variance in Excel, R and Python

Few tools have a built-in pooled variance function, but the formula for pooled variance fits on one line:

  • Excel or Google Sheets: =SUMPRODUCT(B2:B4-1,C2:C4^2)/(SUM(B2:B4)-ROWS(B2:B4)), with sizes in column B and SDs in column C. Use C2:C4 instead of C2:C4^2 if column C already holds variances.
  • R: sum((n - 1) * s^2) / sum(n - 1) for vectors n and s. t.test(x, y, var.equal = TRUE) and aov() use the same pooled variance internally.
  • Python (NumPy): np.sum((n - 1) * s**2) / np.sum(n - 1) for arrays n and s.

References

The formulas and rules of thumb on this page follow these sources:

  • NIST/SEMATECH e-Handbook of Statistical Methods, section 1.3.5.3, Two-Sample t-Test for Equal Means (pooled variance and standard error).
  • Hartley, H. O. (1950). The maximum F-ratio as a short-cut test for heterogeneity of variance. Biometrika, 37(3–4), 308–312.
  • Welch, B. L. (1947). The generalization of “Student’s” problem when several different population variances are involved. Biometrika, 34(1–2), 28–35. The alternative when variances differ.

Common use cases

Assumptions and limitations

  • Assumes equal population variances. Pooling very different variances understates uncertainty for the more variable group and overstates it for the other.
  • Groups must be independent, and each must have at least two observations (nᵢ − 1 = 0 gives no variance).
  • The pooled variance is not the variance of all observations merged; that larger number also includes between-group differences (see the decomposition table in raw-data mode).
  • Like any variance, it is sensitive to outliers within groups, and a large group’s outlier carries a large weight.

Frequently asked questions

What is pooled variance?

Pooled variance is a weighted average of the sample variances of two or more groups, used when the groups are assumed to share one population variance. Each variance is weighted by its degrees of freedom, nᵢ − 1, so sₚ² = Σ(nᵢ − 1)sᵢ² / Σ(nᵢ − 1).

How do I calculate pooled variance?

Multiply each group’s variance by its n − 1, add the products, and divide by the total degrees of freedom N − k. With standard deviations, square them first. For two groups with n = 10, s = 3 and n = 15, s = 4: (9 × 9 + 14 × 16) / 23 = 305 / 23 ≈ 13.2609.

What is the formula for the pooled variance sₚ²?

For two groups, sₚ² = [(n₁ − 1)s₁² + (n₂ − 1)s₂²] / (n₁ + n₂ − 2). For k groups, sₚ² = Σ(nᵢ − 1)sᵢ² / (N − k), where N is the total sample size. With equal group sizes it reduces to the plain average of the variances.

What is pooled variance used for?

It provides one estimate of a variance shared by several groups. Its main uses are the pooled two-sample t-test, where it determines the standard error of the difference in means, and one-way ANOVA, where it is the mean square within groups that forms the denominator of the F statistic.

Is pooled variance the same as MSW in ANOVA?

Yes. The mean square within groups is SSW / (N − k), and SSW = Σ(nᵢ − 1)sᵢ², so MSW equals the pooled variance of the k groups exactly.

Why is pooled variance weighted by n − 1?

Each sample variance is based on nᵢ − 1 degrees of freedom, so it carries that much information about σ². Weighting by degrees of freedom gives an unbiased estimate of the common variance and lets larger groups count more.

What if my groups have very different variances?

Do not pool. For two groups use Welch’s t-test, which estimates each variance separately; for several groups use Welch’s ANOVA or a method that models unequal variances. Pooling in that situation can give misleading p-values, especially when group sizes are unequal.

How do I calculate a pooled proportion?

A pooled proportion is a different quantity, used in the two-proportion z-test: p̂ = (x₁ + x₂) / (n₁ + n₂), the total successes over the total sample size. For 45 of 100 and 30 of 80, p̂ = 75 / 180 ≈ 0.4167. See the Two Proportion Z-Test Calculator.

Cite this page

Statculator. (2026, November 14). Pooled Variance Calculator. Retrieved November 14, 2026, from https://statculator.com/descriptive-statistics/pooled-variance-calculator/