The Stata command is a port of the R package and follows its
releases; the version it ports is named in powerape version
and in every power statement. The changes of the R package are listed in
its news.
powerape for Stata 0.1.3 (2026-10-17)
- Panel designs calibrate about as fast as in the R package. A panel
powerape dgp,setape,setaie,trueapeortrueaieat the defaultnint(1e5)took about a minute and now takes one to three seconds, so robustness sweeps over panel scenarios andmdesearches, which recalibrate at every step, take seconds instead of minutes. Every result is bit-identical to 0.1.2. - The validation scripts run from a clone of the repository and are part of it; see How powerape is validated.
powerape for Stata 0.1.2 (2026-10-16)
- Every simulation command (
power,curve,n,mde,robust) prints a line beginningfit:that names who fitted the simulated studies, for exampleprobit via engine(stata), effects(closed),cfprobit via engine(stata), effects(margins)orMata engine (Fisher scoring), and stores it inr(fit),r(engine)andr(effects). -
engine()andeffects()given on a single command now apply to that command only. Before, they also changed the session default thatpowerape enginesets.
powerape for Stata 0.1.1 (2026-10-16)
- Installation with
net install(powerape.pkg,stata.toc), andpowerape version, which checks that the command and its Mata library belong together. - Declared for Stata 16 or later (frames). The control-function route
under the default engine uses
cfprobit(Stata 18 or later). -
separation(keep)underengine(stata): replications with an empty identifying cell are fitted withasisand kept, as R’sglmkeeps them. -
effects(margins)for the interaction effect on the control-function route:margins, dydx() at()aftercfprobit, with the control function held fixed.
powerape for Stata 0.1.0 (2026-10-15)
- First release: a port of the R package powerape 1.11.3 with every route (parametric, pilot data, fitted pilot model, panel, control function), estimand (APE and AIE), claim and search of the R package. Validated against the R package in four tiers; see How powerape is validated.