MANOVA — Multivariate Analysis of Variance

MANOVA compares two or more groups on multiple dependent variables simultaneously. Where ANOVA tests group means on one outcome, MANOVA tests group means on a multivariate vector of outcomes.


Step 1 — Provide your data

  • One categorical grouping variable (the factor)
  • Two or more dependent variables (the multivariate outcome)

Supports Excel Import, Sample Data Generator, and Manual Entry.


Step 2 — Results

Multivariate test:

  • Wilks' Lambda with F-statistic, df, p-value
  • Partial eta-squared (multivariate effect size)

Box's M Test:

  • Tests homogeneity of covariance matrices
  • F approximation with p-value
  • Guidance on Pillai's Trace if assumption violated

Univariate F-tests (follow-up):

  • Per-variable ANOVA: F-value, df, p-value, eta-squared

A UniversalChatBot is available for discussion.


Statistical methods used

Wilks' Lambda

Lambda = det(E) / det(E + H) where E = within-groups SSCP, H = between-groups SSCP.

Rao's F approximation for p-value.

Hypothesis: H0: all group mean vectors equal. H1: at least one differs. alpha = 0.05.

Effect size: eta2_partial = 1 - Lambda^(1/s)

eta2 Label
< 0.06 Small
0.06 - 0.14 Medium
>= 0.14 Large

Box's M Test — formal test of equal covariance matrices across groups. p < 0.001 suggests violation.

Univariate follow-ups — separate one-way ANOVA per dependent variable (no Bonferroni correction applied).