StatsCapybara

A free, interactive statistics course for people with no stats background, built for visual learners. Every lesson pairs a written explanation with a hands-on canvas visualization you can actually move. Live at statscapybara.com.

Eight courses across two tracks — The Statistics Core (Stats 1–3 plus Machine Learning & AI) and The Research Toolkit (Methods, Data, Writing, Ethics) — alongside a shelf of tools: calculators for power, effect sizes, descriptives and correlation; exact z/t/χ²/F tables; "which test?" and "which chart?" choosers; an APA results formatter; practice problems, datasets, flashcards and a quiet quiz; printable one-page posters; and long-form guides for thesis students. The current counts live in assets/js/curriculum.js and on the homepage.

Free, no sign-up, no ads. Progress is stored in your own browser and nothing is uploaded; see privacy.html for the full account.

Architecture in two sentences

It is a plain static multi-page site: vanilla HTML, CSS and JS with no build step, no framework, no bundler, and no npm dependencies, published from main by GitHub Actions. assets/js/curriculum.js is the single source of truth for courses and lessons, and the homepage grid, every sidebar, the search overlay, and the prev/next links are all generated from it.

index.html                  Homepage (hero demo + curriculum grid)
<course>/index.html         Course landing page (nine of them)
<course>/<topic>/index.html One folder per lesson → clean URL + refresh-proof
guides/<slug>/index.html    Long-form walkthroughs for thesis students
*.html                      Tool pages (calculators, choosers, posters, quiz…)
privacy.html                What is and isn't collected
assets/css/styles.css       The whole design system (light + dark)
assets/js/curriculum.js     Single source of truth — every course & section
assets/js/site.js           Shared chrome: nav, sidebar, search, progress, print
assets/js/viz.js            Shared math/canvas helpers (exact special functions)
assets/js/{checks,software,snippets,glossary-data}.js   Per-lesson content, keyed by slug
sw.js offline.html          Service worker — the site works offline once visited
tools/                      Build/QA scripts (see below)
CLAUDE.md VOICE.md ROADMAP.md PROMPTS.md    Dev guide, editorial rules, roadmap, session prompts

Dev docs, tools/ and .claude/ live in the repo so they sync across machines, and they are not published: the deploy ships the committed tree minus that list, so they return 404 on statscapybara.com. They are of course still right here in the repository.

Working on it

Only Node is needed to preview and check the site; the two content generators are Python 3.

node tools/serve.js 8097     # local preview at http://localhost:8097/
node tools/audit.js          # site health check — required before every commit
node tools/math-check.js     # statistical regression gate — required if viz.js changes
node tools/prose-lint.js     # voice/prose budgets (editorial, not a build gate)

tools/audit.js cross-checks the whole site against curriculum.js: per-lesson coverage, SEO tags, JSON-LD, internal links, search-index freshness, and more. It exits non-zero on any error.

To add or rename a lesson, edit assets/js/curriculum.js (set ready: true once the page exists), then follow the integration checklist in CLAUDE.md — per-page SEO, sitemap.xml, glossary, quiz bank, checks, snippets, FAQs, and a rebuild of the search index.

Deploying

Commit and push to main. That fires .github/workflows/pages.yml, which assembles the artifact with node tools/make-pages-artifact.js — the committed tree minus the development docs (the root .md guides plus CONTRIBUTING.md), tools/, .claude/, .github/ and the two git dotfiles. The license files and CITATION.cff do ship, so the terms are fetchable from the domain itself — and publishes it; the live site updates in a minute or two. A verify job then checks the live domain with tools/check-live.sh, and .github/workflows/health.yml repeats that check weekly along with an external-link report. CNAME sets the custom domain. Nothing is compiled, minified or rewritten: every published file is copied byte-for-byte, and the no-build rule above still holds.

node tools/make-pages-artifact.js --list   # what ships and what doesn't
node tools/make-pages-artifact.js _site    # assemble it locally (self-verifying)
node tools/serve.js 8097 _site             # browse it exactly as the world sees it

If a deploy changes a precached shell asset (styles.css, site.js, curriculum.js, viz.js, the font, an icon, or offline.html), bump CACHE_VERSION in sw.js.

License and reuse

StatsCapybara is openly licensed, in two halves:

	License	Covers
Content	CC BY 4.0	Lesson prose, FAQs, glossary, quiz and check questions, practice problems and solutions, the guides, the printable posters, the practice datasets, the capybara artwork
Code	MIT	assets/js/, assets/css/, sw.js, tools/, the page scaffolding

Share it, adapt it, translate it, fork it, teach from it, commercially or not. Credit Hakan Karsilar / StatsCapybara, link the license, and say if you changed something. The full terms and the exact attribution wording are in LICENSE-CONTENT.md; the reader-facing version lives at statscapybara.com/license.html.

To cite the site, see CITATION.cff — GitHub renders it as a ready-made citation in the sidebar.

Contributions, and especially corrections, are covered in CONTRIBUTING.md.

Found a mistake?

Corrections are genuinely welcome, especially statistical ones. Open an issue, or use the Spotted a mistake? Tell me link at the foot of any page on the site, which fills in the page address for you.