Two-way mixed ANOVA: groups of subjects measured over time

Two groups of subjects — control and treated — each measured at three times. Group is a between-subjects factor (a subject is in one group only); time is within-subject (every subject is measured at every time). MinutePlot fits the univariate split-plot ANOVA: group is tested against subjects within groups, time and group×time against the within-subject error.

The data

One row per subject, the group in the first column, the times across:

Mixed layout: subjects down the rows, their group in a label column, the within-subject levels across.

In MinutePlot

Choose Two-way mixed ANOVA, paste the table, run. Mauchly’s test is computed on the repeated-measures covariance pooled within the groups (the standard mixed-design formulation), and the Greenhouse–Geisser correction is applied to the within-subject effects when needed.

The result

On the validation dataset behind this guide: time F(2, 28) = 83.8, p < 0.001; group×time F(2, 28) = 7.17, p = 0.003 (the treated group climbs faster); the group main effect on its own does not reach significance (F(1, 14) = 4.19, p = 0.060). The interaction is the finding: the treatment shows up as a steeper rise, not a uniform offset.

The figure MinutePlot produces: the two groups along the axis, one bar per time; the steeper rise in the treated group is the interaction.

Styling and exporting this figure: see Figure styling and Exporting & downloading.

Report it like this

“The response rose over time (F(2, 28) = 83.8, p < 0.001), more steeply in the treated group (group×time F(2, 28) = 7.17, p = 0.003).”