Welch’s t-test compares the means of two independent groups without assuming they have equal variances. It is the default two-sample test in R and is recommended by many statisticians for everyday use. Enter raw data or summary statistics; the calculator reports t, the Welch–Satterthwaite degrees of freedom, p, the confidence interval, Cohen’s d, Hedges’ g and a variance check.
Welch’s vs Student’s t-test
| Student’s t-test | Welch’s t-test | |
|---|---|---|
| Assumes equal variances | Yes | No |
| Degrees of freedom | n₁ + n₂ − 2 | Welch–Satterthwaite (usually not a whole number) |
| When variances and sizes are equal | Correct | Almost identical result |
| When variances differ, especially with unequal n | p-values can be badly wrong | Correct |
Because Welch’s test loses almost nothing when variances are equal and protects you when they aren’t, there is little reason to test variances first and then choose. Just use Welch.
Formula
t = (x̄₁ − x̄₂) ÷ √(s₁²/n₁ + s₂²/n₂) · df = (s₁²/n₁ + s₂²/n₂)² ÷ [ (s₁²/n₁)²/(n₁ − 1) + (s₂²/n₂)²/(n₂ − 1) ]
Reporting Welch’s t-test in APA style
Report the degrees of freedom to two decimals: “Welch’s t-test showed no significant difference between the groups, t(13.74) = 2.01, p = .065, d = 1.01.” The calculator writes this sentence for you.
Frequently asked questions
Why are my degrees of freedom a decimal?
The Welch–Satterthwaite formula estimates an effective number of degrees of freedom from both variances, so it is rarely a whole number. Report it to two decimal places.
Does Welch’s test need normal data?
It assumes each group is roughly normal, but with 30 or more per group it is robust. For small, skewed samples use the Mann-Whitney U test.