Magic Stat — The ultimate statistical suite for macOS, Windows & Linux

Loved by researchers

❝ What I loved most was having my favorite AI model look at the results alongside me — suggesting new comparisons, interesting hypotheses and further analyses, and even building the ideal charts inside my workflow. No other software on the market does this. ❞

❝ My nightmare was copying and pasting tables into Word and formatting them for hours. With Magic Stat they're born ready. ❞

❝ I used to suffer with the structural-equation packages in R. Magic Stat lets you draw complex models in a way that's extremely simple and fluid. ❞

❝ Making beautiful charts has never been this easy. ❞

❝ The Report section is my favorite — the sticky notes you can drop right into it make the report feel familiar. ❞

From messages we received by email.

What it does

A statistical arsenal, with the simplicity of an app

From descriptive statistics to advanced structural modeling — all in one place, no plugins, with the rigorous output your paper demands.

SEM & PLS

Structural equation modeling and PLS-SEM with a drag-and-drop interface, plus covariance-based path analysis.

Editable path diagrams
Reliability, validity & HTMT
Indirect effects & bootstrapping

Psychometrics

Reliability and construct-validity analysis for psychology and the social sciences.

Cronbach's alpha & Omega
Factor analysis (EFA)
KMO, Bartlett & item-total

Ecology

Biodiversity and community ecology with the canonical metrics.

Alpha and beta diversity
Ordination analysis
Functional & phylogenetic diversity

Species distribution modeling

From occurrence points to habitat-suitability maps in a few clicks.

Occurrence points + environmental rasters
Model ensemble with automatic evaluation
High-resolution suitability maps

Time series

Forecasting and decomposition for data over time.

Holt-Winters, Theta, ARIMA
STL decomposition
ACF/PACF & stationarity tests

Dynamical systems & chaos

Type equations on screen and explore chaos — from ODEs to discrete maps and reaction–diffusion systems.

Bifurcation diagrams & phase portraits
Lyapunov spectrum, Poincaré & basins of attraction
Chaos detection in real time series

Mixed models

Fixed and random effects for clustered and repeated-measures data.

Random intercepts
Repeated-measures ANOVA
Mixed designs

Generalized & nonlinear models

Advanced regression families for counts, categories and curved relationships.

Poisson & multinomial logistic
GAM, GEE & GLM
Nonlinear (nls) & robust (Theil–Sen)

Medical statistics

Tools built for clinical and diagnostic research.

ROC curves & AUC (DeLong)
Bland-Altman & Kaplan-Meier
Cohen's Kappa & diagnostic tests

Econometrics

The applied-econometrics toolbox for impact, panel and volatility studies.

Panel data, differences-in-differences & structural breaks
2SLS, SUR & GARCH family
Granger causality & cointegration

Multivariate

Dimensionality reduction and ordination for any field.

PCA, PCoA, NMDS
CA, DCA, RDA, CCA
PERMANOVA & MANOVA
Cluster analysis & discriminant (LDA)

Spatial analysis

Autocorrelation, clusters and regression for georeferenced data.

Moran's I, LISA & correlograms
Variograms & interpolation maps
GWR & spatial regression (SAR/CAR)

Meta-analysis

Combine and quantify evidence across multiple studies.

Forest & funnel plots
Heterogeneity (I², τ²)
Egger's test & sensitivity

Power analysis

Plan your study with the ideal sample size.

t, ANOVA, correlation, chi-square
Regression & SEM
Power curves & sample-size tables

Bayesian statistics

Inference with Bayes factors and credible intervals.

JZS BF for t, ANOVA, correlation & regression
HDI / ROPE
Posteriors

Regression & correlation

From simple to flexible.

Simple, multiple, polynomial
Logistic with OR & ROC
LOESS, splines, VIF
Dose-response & area under curve (PK)

Classics

The essentials, refined.

Cross-tabulation (χ²)
t-tests, ANOVA & post-hoc
Descriptives table for your paper

Data preparation

Clean and shape your data before the analysis — all built in, no external tools.

Variable dictionary & types
Transformations, recoding & missing values
Built-in example & synthetic datasets

Equation editor

Write beautiful equations in a WYSIWYG editor with a rich math palette.

Live typesetting as you type
Calculus, algebra, Greek & more
Export PNG, SVG, Word & LibreOffice

Circular graphs

Build complex circular plots with absolute ease.

Chord & ribbon diagrams
Flow & correlation layouts
Publication-ready export

The differentiator

An AI data scientist you already know

Plug in your favorite language model via API — OpenAI, Claude or DeepSeek — and operate the software by conversation. The assistant actually runs Magic Stat for you: it creates figures, runs analyses, builds tables, discusses the results and helps with everything.

Your model, your key (BYOK) — bring your own API key; nothing is stored or sent anywhere else.
Talk to your data — “explore this dataset”, “compare the groups”, “check the assumptions”.
It does the work — the agent opens the right window and executes real analyses, figures and tables.
It discusses with you — interprets coefficients, folds, p-values and effect sizes in plain language.
Never invents results — it only reports what it actually ran, reading the real numbers from your session.

Plug your favorite: OpenAIClaudeDeepSeek

Magic Stat Assistant (Claude via API)

Explore this data and tell me if there's a difference between the groups.

On it. I'll run a one-way ANOVA across the 5 groups, then a Tukey post-hoc.

F(4,495)=21.3, p<0.001 — significant difference, driven by groups D & A.

Great — generate the boxplot with those groups, ready for the paper.

Done.

I opened the box plot, colored by group — publication-ready. Want me to add the significance letters or export it?

Yes, add significance letters and export as PNG at 300 DPI.

Perfect.

Letters added, exported at 300 DPI. I can also paste the formatted table into the report.

Smart reports

A report ready for your paper

Forget the raw output of conventional software. Magic Stat generates a formatted report — with tables in scientific format and figures — that can be read and commented on by the AI, and exported already formatted.

✔ APA/editorial-style tables, no rework
✔ Statistically formatted output (p, d, η², BF…)
✔ Post-hoc significance letters (Compact Letter Display — a, b, c)
✔ AI comments on every section — integrated review
✔ Export formatted and keep the session for review

ANOVA + post-hocF(4, 495) = 21.3, p < 0.001, η² = 0.147

Post-hoc (Tukey) Group D: a Group E: a Group A: b Group B: bc Group C: c

Correlationr(498) = 0.978, 95% CI [0.973, 0.982], p < 0.001

Regressionβ = 6.78, SE = 0.13, p < 0.001, R² = 0.956

"Groups sharing a letter are not significantly different — the report adds the letters automatically."

Data quality

Detect outliers fast.

Find suspicious values using standard methods — then remove them automatically, from one variable or the whole sample.

IQRZ-scoreModified Z (MAD)Grubbs' test

Flagged values are highlighted in the data grid and box plot — inspect, then exclude with one click.

Color palettes

Color palettes designed for scientists.

Choose from a curated catalog of perceptually uniform and colorblind-safe palettes — so your figures stay beautiful and publication-ready, not washed out or clashing.

Qualitative

Perfect for categories and groups — distinct, harmonious colors that pop.

Sequential

Great for continuous data and gradients, perceptually uniform by design.

Diverging

Ideal for correlations and centered data — highlights direction and magnitude.

One palette selector across every figure — box plots, dot plots, scatter, bars, error plots, and more — with live color previews.

Gallery

What leaves the app.

Real figures, exactly as they leave Magic Stat — clean, crisp and labeled, ready for your paper.

Linear regression with confidence interval

Correlation matrix (Pearson) — Correlation matrix (Pearson, parametric)

Correlation matrix variant — Correlation matrix (variant)

Logistic bifurcation diagram · 1 million points plotted in 2 seconds

Structural equation modeling · PLS & covariance-based methods

Download for macOS Download for Windows Download for Linux

macOS ~490 MB · Windows ~350 MB · Linux ~575 MB · installs in a couple of clicks, ready to use.

7-day free trial: all features unlocked, totally free — no card required.

Questions

Frequently asked questions

No. Magic Stat is a window-based application, with menus and an integrated spreadsheet. You open the file, click the analysis and get a figure + table + formatted report.

Yes. Magic Stat runs on both Intel and Apple Silicon Macs. On Apple Silicon, macOS may ask to install Rosetta the first time you open the app — it's a one-click Apple feature that runs Intel apps smoothly, and you only do it once.

Yes — Windows 10/11, 64-bit, with the same full feature set as the macOS and Linux builds.

Yes. Magic Stat runs on Windows 10/11 and on 64-bit Linux, shipped as a self-contained AppImage — download it, mark it executable and run: no installation and no dependencies to chase. All three platforms (macOS, Windows and Linux) carry the same full feature set.

Yes. You get a 7-day free trial with full access to every feature — no credit card required. After the trial, activate the app with a license purchased on Gumroad.

The AI uses your own key (BYOK) with OpenAI, Claude or DeepSeek. The app itself works offline; the assistant needs a connection.

Excel (.xlsx/.xls), CSV and ODS (LibreOffice). You can also paste data straight into the spreadsheet.

Yes. Tables are born in scientific/APA format, and the report can be exported formatted — no copying raw output.

From your spreadsheet to a publishable figure in seconds