Lodestar.Stats 0.4.0. This page is frozen at that release. Read the current documentation for what main says now. A link to a decision or a migration page follows main, and leaves the archive.

OneWayAnova

One-way analysis of variance: do several groups share one mean?

The k-sample generalisation of TTest.Independent with Variance.Equal: on two groups the F statistic is the square of Student's t, and the two p-values agree.

A degenerate input where every group is internally constant and the groups share that same constant computes 0.0 / 0.0 and returns a NaN statistic and p-value, propagated rather than guarded against — this matches scipy's own f_oneway on the same input, and a NaN here is an honest answer: an F statistic genuinely has no value when both its numerator and its denominator vanish. This is deliberately unlike KruskalWallis, whose analogous degenerate input (every pooled value tied) throws instead, because there the ranks the statistic is built from are provably meaningless rather than merely indeterminate.

Members

Member	What it does
OneWayAnova.Test	Compares the means of two or more groups.