A Julia package for analysis of factorial experiments, including repeated measures (within-subjects), between-subjects, and mixed designs, inspired by the R packages afex, ez and emmeans.
Note: This package implements a subset of features from the above packages, focused on common ANOVA analyses. It has been tested, albeit not exhaustively, against R's aov, ezANOVA, afex, emmeans, and JASP's ANOVA interface.
- Between-subjects ANOVA - Compare groups across independent subjects
- Within-subjects (Repeated Measures) ANOVA - Compare conditions within the same subjects
- Mixed-design ANOVA - Combine between and within factors
- Estimated Marginal Means (emmeans) - Calculate marginal means with confidence intervals
- Pairwise Comparisons - Post-hoc tests with multiple comparison adjustments
- Sphericity Corrections - Greenhouse-Geisser and Huynh-Feldt corrections
- Effect Sizes - η² (eta squared), partial η², and ω² (omega squared)
- Power Analysis - Simulation-based power analysis
- Basic Plots - Interactive plots via GLMakie or publication quality outputs via CairoMakie for ANOVA/emmeans results
- Output - APA formatted tables and stats to latex, markdown, or text via PrettyTables
# From the Julia General Registry
using Pkg
Pkg.add("AnovaFun")
# Or directly from GitHub
Pkg.add(url="https://github.com/igmmgi/AnovaFun.jl")# Between-subjects ANOVA
result = anova(data, :dv, :subject, between = [:factor1, :factor2])
# Within-subjects ANOVA
result = anova(data, :dv, :subject, within = [:factor1, :factor2])
# Mixed-design ANOVA
result = anova(data, :dv, :subject, between = [:between_factor], within = [:within_factor])# Basic line plot
plot_anova(result, x_grouping = :factor1, y_grouping = :factor2)
# With faceting
plot_anova(result, x_grouping = :factor1, y_grouping = :factor2, facet_cols = :factor3)
# Plot types: :line, :bar, :violin, :boxplot, :raincloud, and custom raincloud types
# Themes changes available via Makie themes
using AnovaFun, DataFrames, CSV
# Load the bundled example dataset
data = CSV.read(AnovaFun.ExampleData, DataFrame)
# Or load your own dataset (long format)
# data = CSV.read("path/to/data.csv", DataFrame)
# Preview the first few rows
first(data, 10)| Subject | PreviousCongruency | CurrentCongruency | RT |
|---|---|---|---|
| 1 | Congruent | Congruent | 523.9 |
| 1 | Congruent | Incongruent | 576.62 |
| 1 | Incongruent | Congruent | 545.2 |
| 1 | Incongruent | Incongruent | 567.1 |
| 2 | Congruent | Congruent | 537.54 |
| 2 | Congruent | Incongruent | 609.34 |
| 2 | Incongruent | Congruent | 519.78 |
| 2 | Incongruent | Incongruent | 548.78 |
Dataset Summary:
- N = 30 subjects (each completing all conditions)
- Design: 2×2 within-subjects (PreviousCongruency × CurrentCongruency)
- Dependent Variable: Reaction time (RT)
- Total observations: 120 (30 subjects × 4 conditions)
# Run 2×2 within-subjects ANOVA
result = anova(data, :RT, :Subject, within = [:PreviousCongruency, :CurrentCongruency])result # default show method
result.table # for full anove table| Effect | DFn | DFd | F | p | sig¹ | η²ₚ |
|---|---|---|---|---|---|---|
| PreviousCongruency | 1 | 29 | 0.04 | 0.838 | n.s. | 0.001 |
| CurrentCongruency | 1 | 29 | 103.34 | < .001 | *** | 0.781 |
| PreviousCongruency × CurrentCongruency | 1 | 29 | 17.68 | < .001 | *** | 0.379 |
¹: * = p < .05, = p < .01, * = p < .001, n.s. = not significant
# Calculate estimated marginal means for all effects
result_emm = emmeans(result)| Effect | Level | N | Mean | SE | Lower | Upper | error |
|---|---|---|---|---|---|---|---|
| Grand Mean | Overall | 30 | 551.33 | 3.73 | 543.7 | 558.96 | 3.73 |
| PreviousCongruency | Congruent | 30 | 550.94 | 3.80 | 543.2 | 558.68 | 3.80 |
| PreviousCongruency | Incongruent | 30 | 551.72 | 3.80 | 543.98 | 559.46 | 3.80 |
| CurrentCongruency | Congruent | 30 | 532.85 | 3.80 | 525.11 | 540.59 | 3.80 |
| CurrentCongruency | Incongruent | 30 | 569.82 | 3.80 | 562.08 | 577.56 | 3.80 |
| PreviousCongruency × CurrentCongruency | Congruent, Congruent | 30 | 524.08 | 5.37 | 513.1 | 535.06 | 5.37 |
| PreviousCongruency × CurrentCongruency | Congruent, Incongruent | 30 | 577.80 | 5.37 | 566.82 | 588.78 | 5.37 |
| PreviousCongruency × CurrentCongruency | Incongruent, Congruent | 30 | 541.61 | 5.37 | 530.63 | 552.59 | 5.37 |
| PreviousCongruency × CurrentCongruency | Incongruent, Incongruent | 30 | 561.83 | 5.37 | 550.85 | 572.81 | 5.37 |
# Line plot showing the interaction
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :line)
# Bar plot with error bars
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :bar)
# Raincloud plot showing data distributions
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :raincloud)
# Basic line plot
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :line)
# Bar plot with error bars
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :bar)
# Raincloud plot with distributions and points
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :raincloud)
# Box plot showing distributions
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :boxplot)
# Violin plot with density estimates
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :violin)
# Custom raincloud with connected individual points
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :raincloud_custom_2x2,
individual_data = :connected_points)
# Additional custom raincloud configuration
plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :raincloud_custom)
# Custom theme with teal/coral colors
custom_theme = Theme(
palette = (color = [:teal, :coral],)
)
fig = plot_anova(result,
x_grouping = :CurrentCongruency,
y_grouping = :PreviousCongruency,
plot_type = :raincloud_custom_2x2,
theme = custom_theme
)
# Using Makie's built-in ggplot2 theme
fig = plot_anova(result,
x_grouping = :PreviousCongruency,
y_grouping = :CurrentCongruency,
plot_type = :line,
theme = theme_ggplot2()
)
custom_theme = Theme(
palette = (color = [:black, :grey],),
Axis = (
xlabelsize = 24,
ylabelsize = 24,
xticklabelsize = 20,
yticklabelsize = 20,
titlesize = 28
),
Legend = (
labelsize = 18,
titlesize = 20
)
)
fig = plot_anova(result,
x_grouping = :PreviousCongruency,
y_grouping = :CurrentCongruency,
plot_type = :bar,
theme = custom_theme,
axis_title = "Congruency Sequence Effect",
axis_xlabel = "Previous Trial",
axis_ylabel = "Mean RT [ms]",
legend_title = "Current Trial",
legend_framevisible = false,
legend_order = [:Incongruent, :Congruent]
)
If you use AnovaFun.jl in your research, please cite it:
@software{mackenzie2026anovafun,
author = {Mackenzie, Ian G. and Sonntag, Samuel and Dudschig, Carolin},
title = {{AnovaFun.jl}},
year = {2026},
url = {https://github.com/igmmgi/AnovaFun.jl},
doi = {10.5281/zenodo.18139228}
}MIT License