The revss R package for robust estimation in small samples has been completely overhauled in version 3.1.0. The entire calculation engine was rewritten from R to Fortran (with C wrappers) to handle massive Monte Carlo simulations — roughly 660 million runs — needed to compute accurate four-digit bias reduction factors for six scale metrics. New bias-adjusted functions were added, the API was broken, and LLMs (ChatGPT, Claude, Grok) were used as optimization aids during the rewrite. Due to a last-minute mistake, the flawed 3.0.0 is currently on CRAN while 3.1.0 awaits acceptance; the correct version is available on GitHub in the meantime.
Table of contents
Estimation, via Wikimedia Commons” width=”204″ height=”307″ /> The Rabbit HolesCurrent StateReferencesSort: