Parametric
t-test (independent samples)
Compares the means of two independent groups.
Variables
- Dependent variableContinuous — 1
- Grouping variableNominal · Ordinal — 1
Options
- VariantWelch · Student
- Alternative hypothesisTwo-sided · One-sided (greater) · One-sided (less)
What it reports
Welch by default, with fractional degrees of freedom; Hedges' g and its 95% CI.
scipy · Open the accuracy report
When does this apply?
For a design in which two independent groups — each observation belongs to exactly one group — are compared on one continuous outcome. The outcome sits in one column and the group label in a second. The output is t, the degrees of freedom, p, and Hedges g with a 95% confidence interval; Welch’s variant does not require equal group variances and is the default. The rank-based alternative is the Mann–Whitney U test.