equivalence-testing

This package contains many functions for conducting negligible effect statistical testing (also called equivalence testing).

I constructed a simulation study to evaluate the statistical performance of two equivalence-based tests and compared it to the common, but inappropriate, method of concluding no effect by failing to reject the null hypothesis of the traditional test. I further propose two R functions to supply researchers with open-access and easy-to-use tools …

Analysis snapshot of the "Statistical testing for protein equivalence identifies core functional modules conserved across 360 cancer cell lines and presents a general approach to investigating biological systems".

Interaction Net Equivalence Testing for LLMs

Auditable COBOL-to-Rust banking modernization equivalence demo

R code for Wintermantel et al. — Equivalence testing in pesticide risk assessment: Evaluation and practical guidance for design, analysis and interpretation (Environment International).

R helpers for reproducible research — a pre-analysis-plan (PAP) template, equivalence tests, RDD utilities, results export, and a submission checklist.

Does cross-model pollination help open autoresearch? Pre-registered study of 30,687 AI-agent submissions on 10 Yukon leaderboards incl. ECDSA.Fail. Paper, code, frozen data, timestamped lock.

Gale-Shapley stable matching via NLP text similarity, with group-fairness diagnostics, on synthetic data (Python).

Publication companion: temporal Jacobians predict fitted-model interventions but do not guarantee recovery of latent cognitive dynamics.

Replication package for "Do facial-affect process data improve cognitive measurement? An adversarial out-of-sample benchmark." Lee & Wang (2026). Code and aggregate results only; no participant data.

Code, per-item results and figures for a study dissociating prompt quality from response compliance in automated prompt-engineering assessment. Four-agent MATLAB evaluator on a locally hosted Qwen 2.5-7B judge, over 498 prompts from IFEval, LMSYS-Chat-1M and WildChat.

Poisson modelling of 1.4M Seattle 911 call arrivals by time of day: MLE and Bayesian rates, entropy, TOST (UdeSA 2025)

Intel Intrinsics equivalence testing for OpenPOWER, ARM64 NEON, and RISC-V64 RVV, with reproducible checks, a local web studio, and self-contained HTML reports.

Is that eval regression real, or just noise? OpenParity does the statistics every LLM eval tool skips: paired testing, equivalence gates (TOST), cluster-robust intervals, and power analysis. Not an eval runner — it plugs in downstream of promptfoo, pydantic-evals, or your own loop.

Equivalence tests for pre-trends in Difference-in-Differences estimation. Implements max, mean, and RMS tests from Dette & Schumann (2024, JBES).