Articles
Start here
- Getting started with anovakit
- Which function? Choosing an analysis
A decision guide: four questions about your data that lead to one of the eight functions, the situations each one is built for, and the mistakes the choice is meant to avoid.
- Working with results: the anovakit_fit object
What every function returns, how to print, extract and reuse it, how to go further with emmeans and the fitted model, and how to report a result.
Walkthroughs
One article per function: the data, the fit, the output, effect sizes, comparisons, diagnostics, plots, options and a model write-up.
- Welch's ANOVA: comparing means when variances differ
A walkthrough of anova_welch(): the omnibus Welch test, pairwise Welch t tests, standardised mean differences that do not assume equal variances, and the three diagnostic plots.
- Kruskal-Wallis: comparing groups by ranks
A walkthrough of anova_kw(): the Kruskal-Wallis test, epsilon squared, Dunn’s comparisons with the tie correction, missing and infinite values, and the optional diagnostics.
- ANCOVA: comparing groups at the same value of a covariate
A walkthrough of anova_ancova() on the anorexia treatment data: covariate-adjusted means, the homogeneity-of-slopes test, simple slopes, centring and robust tests.
- Repeated measures ANOVA: the same subjects under several conditions
A walkthrough of anova_rm(): mixed designs, sphericity corrections, generalised eta squared, incomplete subjects and within-subject residual checks.
- MANOVA: comparing groups on several responses at once
A walkthrough of anova_manova() on the iris measurements: the multivariate test, univariate follow-ups, canonical discriminant analysis, Box’s M and Mardia’s tests, and MANCOVA.
- Binary outcomes: logistic regression with anova_bin()
A walkthrough of anova_bin() on the Berkeley admissions data: analysis of deviance, odds ratios, marginal probabilities, frequency weights and separation.
- Counts and rates: analysis of deviance with anova_count()
A walkthrough of anova_count() on the warpbreaks and insurance-claims data: Poisson or negative binomial, rate ratios, marginal rates, exposure offsets, frequency weights and groups with no events.
- Analysis of deviance: factorial designs and non-normal responses
A walkthrough of anova_glm(): a two-way Gaussian ANOVA with Type II and Type III tests, then a Gamma model for a positive, right-skewed response, with its coefficient table, dispersion, robust standard errors and weights.
Guides
Topics that cut across every function.
- Diagnostics: what anovakit checks and what to do about it
Every assumption check the eight functions run, organised by what is being checked: how to read each one, a worked example that makes it fire, and the argument or function that helps when it fails.
- Plots: every figure anovakit returns, and how to customise them
A tour of every ggplot2 figure the eight functions build, grouped by kind, with what to look for in each; then how to change them, build your own from the result tables, combine and save them.