ANCOVA

ANCOVA — compare groups fairly by controlling for a covariate

Two groups can look different just because one started with a head start. Analysis of Covariance (ANCOVA) removes that head start: it compares group means on an outcome (Y) while statistically controlling for a continuous covariate (X), so every group is judged on a level playing field.

This is a tiny, beautifully designed, fully offline app that both EXPLAINS ANCOVA and RUNS it on your own numbers — no account, no internet, no clutter.

WHAT YOU CAN DO

• Enter your own data — a per-group (X, Y) editor. Type numbers directly, add or remove groups and points, and everything recomputes instantly.
• Read the full ANCOVA table — Covariate, Factor (adjusted), Error and Total rows with sums of squares, degrees of freedom, F and p. Significant effects are highlighted.
• See adjusted vs raw means — the common within-group slope and each group's mean adjusted to the average covariate level, next to the raw means.
• Check the assumption — a homogeneity-of-slopes test warns you when the slopes are not parallel and the adjusted means would mislead.
• Explore the signature plot — a scatter coloured by group with parallel fitted lines and adjusted-mean markers on the covariate-mean line.
• Simulate a design — turn the covariate strength, group difference, covariate confound, spread and noise. Type an exact value or drag the slider, then send the dataset straight to the Data tab.
• Learn the concept — what ANCOVA is, why a covariate makes comparisons fairer and the error smaller, and the adjustment formula, with mathematics rendered beautifully and entirely offline.
• Load ready-made studies — teaching method, diet and weight, fertiliser and yield, therapy and mood — each loads with a single tap.

WHY ANCOVA

When groups differ on a background variable you cannot control — a pretest score, a starting weight, baseline rainfall — a plain comparison of means is unfair. ANCOVA adjusts each group's mean to a common covariate level and tests the groups after removing the covariate's effect. The result is a fairer comparison and, usually, a smaller error term and more statistical power.

DESIGNED TO BE CLEAR

• Clean "Slate Azure" design with a colour per group.
• Plain-language verdicts, not just numbers.
• Pure-Kotlin statistics and a hand-drawn chart — no bloated libraries, tiny download.
• Works completely offline. Your data stays on your device and is never lost between sessions.

WHO IT IS FOR

Students learning the general linear model, researchers and analysts checking an adjusted comparison, teachers demonstrating covariate adjustment, and anyone who wants an honest, fair comparison of group means.

Note: this app supports learning and quick analysis. Always check the assumptions (linearity, homogeneity of slopes, independence) before reporting results.