Overview
ANOVA is commonly run without post-hoc tests — producing an F-test that indicates some group differences exist without identifying which specific groups differ. An omnibus ANOVA result with p<0.05 tells the researcher nothing actionable without post-hoc comparisons. Conversely, running all pairwise comparisons without alpha correction inflates the familywise Type I error rate: with 5 groups producing 10 pairwise comparisons, the probability of at least one false positive at α=0.05 approaches 40%.
The ANOVA Framework designs the complete test sequence — omnibus F-test, appropriate post-hoc comparisons with familywise error correction, and effect size (η²) reporting that shows what proportion of outcome variance the grouping factor explains.
Key Features
ANOVA variant selection
Assumption verification with alternatives for violations
Interaction-first interpretation for factorial designs