PowerMate compares its own engine against established power-analysis packages written by other researchers. If a comparison on this site informed your work, cite the package that produced the number — the reference each team asks for is given below.
Compared tools
These are the external tools in the cross-check. Versions are the ones actually executed to produce the committed comparisons, not necessarily the current CRAN release.
Where PowerMate's own formulas come from
The section above credits the external tools PowerMate compares against. This one states the provenance of the mathematics PowerMate uses itself — the formula behind every sample size it publishes, the approximation it relies on, and whether the computed power is exact or approximate. A method whose source could not be confirmed against a primary reference is shown as unverified rather than guessed.
The engine and its dependencies
Typography
Wider tool catalogue
Tools tracked for coverage and discovery. Catalogue inclusion is credit and discovery, not endorsement, and does not mean PowerMate reproduces that tool.
| Tool / package | What it covers | Role here | Benchmark state | Licence / review | Official source |
|---|
The planner uses PowerBench's own transparent reference and simulation. An external tool appears only as a comparison check. A tool can be excellent and still be “unsupported” for your exact scenario if its effect definition, allocation, tails, or model differs.
G*Power comparisons are captured by driving the application through Windows UI automation (tools/gpower/gpower_driver.py), and each capture records the exact design that was selected and the inputs that landed. G*Power needs an interactive desktop, so captures are generated locally and committed rather than run in CI — the same arrangement as the R adapters. Of the 28 canonical scenarios, 16 can be expressed in G*Power at all; the remaining 12 have no corresponding G*Power design and are declared as such rather than left blank.