MCPower estimates statistical power by simulation. Describe the study you plan — the model, the predictors, the effect sizes you want to detect — and the native engine generates thousands of synthetic datasets from that description, fits your model to each one, and counts how often the effect comes out significant. That count is the power: no lookup tables, no closed-form approximations, just your study run many times before you run it once.
Simulation is what lets MCPower cover designs that formula-based calculators struggle with — correlated predictors, skewed distributions, factors with unequal groups, clustered observations, real pilot data — all in one tool.
Who it's for
Researchers planning studies — an OLS regression, a logistic model, a mixed-effects design, a factorial ANOVA — who need a defensible sample size or an honest power estimate. MCPower is a planning tool, not a statistics workbench: it doesn't analyse your collected data, and it isn't a framework for building new methods.
One engine, four ports
MCPower ships as a Python package, an R package, a desktop app, and a browser app. All four are thin layers over the same compiled native engine, so a power number computed in one is produced by the exact same calculation in the others — same design, same seed, same answer. See one engine, four ports for how that works.
App or packages?
The app (desktop or browser) is the no-code path: pick a model family, fill in the panels, press Find power — best for exploring a design, teaching, or getting a quick answer. The packages are the scripted path: an analysis is a short script you can version, rerun, and paste into a grant or preregistration — best for reproducibility and pipelines. It's the same engine underneath, and a session in one maps cleanly onto the other, so switching later costs nothing.
Where next
- App, Python, or R? — which face fits your situation.
- How it compares — MCPower vs G*Power, superpower, simr, pwr, and WebPower.
- Roadmap — what's coming next and what's deliberately out of scope.
- The app — the GUI, panel by panel.
- Python — install to first analysis in minutes.
- R — the same ladder in R.
- Validation — how we know the numbers are right.