Canonical scenario
The research question is translated into explicit effect, design, alpha, tails, allocation, and target power. No tool is allowed to silently change them.
Independent reference
A transparent probability-distribution calculation provides a benchmark when a defensible analytic route exists.
Fresh simulated studies
PowerBench generates datasets and runs the planned test repeatedly. The estimate includes Monte Carlo uncertainty and failure counts.
External tool comparison
An R or Python package is compared only when its inputs and assumptions mean the same thing. “Unsupported” prevents a misleading comparison.
Discrepancy review
Differences are investigated as possible assumption, parameterization, approximation, rounding, or implementation issues—never automatically called errors.
Connection to the initial planThis is PowerBench: the shared evidence layer beneath PowerMate, the Guide, and future research workflows. Its purpose is quality assurance, not another calculator menu.
Reproducible benchmark examples
Core technical credits
Analytic distributions and independent tests use SciPy. Matched external comparisons use the R pwr package. Welch planning is benchmarked against R’s t-distribution functions. See Tools & credits for versions, licences, and boundaries.