About PsyStat Lab

PsyStat Lab is built by Moonlit Social Labs — a small, independent studio on a mission to make advanced research statistics accessible to every behavioral scientist, regardless of budget or coding ability.

The Problem

Researchers in psychology, sociology, criminology, education, and related fields need powerful statistical tools. But the options are grim: legacy desktop software with $1,000+ licenses, R scripts that require a semester of programming to learn, or point-and-click tools that cap out the moment you need something beyond a t-test.

The result? Researchers spend more time fighting their software than thinking about their science. Students learn to fear statistics instead of understanding it. Labs pay for five different tools and still end up copy-pasting numbers into spreadsheets.

Our Solution

PsyStat Lab puts 240+ statistical methods into a single platform that runs on any device — iOS, Android, and the web. No installation, no license keys, no SPSS dongle.

Every analysis comes with APA-formatted output, interactive visualizations, assumption checks, and an AI assistant that can explain what your results mean in plain language or recommend the right test for your research question.

Our Values

AI Models We Use

The AI assistant is powered by frontier language models from Anthropic. As the frontier moves, we upgrade — the model names below are current as of the date this page was updated, not permanent commitments.

• Standard model — Anthropic Claude Sonnet 4.6. Used on the Researcher tier and for the Scholar tier's day-to-day 300 monthly messages. Fast, accurate for most guidance and interpretation work.
• Advanced model — Anthropic Claude Opus 4.6. Included on the Scholar tier for its 30 monthly deep-analysis messages. Slower, higher-cost, better at long-chain reasoning about complex model choices, causal identification questions, and multi-study synthesis.

Your data is not used to train Anthropic's models. See the IRB-ready data-handling statement for the exact context bundle transmitted with each AI request.

What’s open, what’s proprietary

PsyStat Lab is open-core. We publish source code where doing so serves reproducibility and researcher trust, and we hold back the parts that constitute the hosted product itself.

• Open on GitHub: statistical implementations, the validation suite, references, datasets, and the marketing site. If you find a numerical bug or want to add a golden-number test, we take PRs.
• Proprietary: the hosted compute infrastructure (Railway), the AI orchestration layer, subscription and billing plumbing, the mobile app binaries, and any product analytics.
• License: the published parts are under a permissive open-source license (see LICENSE in the repository). Proprietary components are all-rights-reserved.

This posture is deliberate. Open-source statistical implementations mean anyone can inspect the math — the failure mode we most want to avoid is “a black box gave me a number.” A researcher who wants to independently verify a PsyStat result can do so end-to-end. The hosted business side is what pays for that work to keep happening.

Who We Are

Moonlit Social Labs is an independent software studio focused on tools for the social and behavioral sciences. We believe the best statistical software should be built with researchers, not at them.