Reproducibility kit

What this release contains

This public snapshot exposes the pure calculation layer used by the interactive labs, the deterministic tests that exercise it, and the exact runtime dependency needed to run those tests. The files are generated from the same source tree as publication build ertaiba-20260828T033904Z.

Publishing the files makes the work inspectable; it does not turn this page into an independent audit. The snapshot does not include server credentials, deployment configuration, analytics settings, or account data.

Publication revision

ertaiba-20260828T033904Z

Calculation tests

17 deterministic and invariant checks

Runtime

Node.js 22.12 or newer

External calculation dependency

simple-statistics 7.10.1

Download the evidence

• 01

  Calculation source

  SHA-256 1289cc7e9d38c2f9ef94f21256e4498e939a1b258e0540b262b6bccb07423e9b

• 02

  Calculation tests

  SHA-256 39559e1c7b4399800075de3eca6cb816f4cf46599150cf5fcfce8954f0ba5299

• 03

  Runtime manifest

  SHA-256 ba4f728d6900099becc6c90b9be3714c88e4b8ff6ad2841eaab9b67eb3c7b7cd

• 04

  Checksums and scope

  Machine-readable file sizes, hashes, revision, test command, and limitations.

• 05

  Plain-text instructions

  Offline setup, run steps, file descriptions, and scope boundaries.

Run the checks locally

npm install --ignore-scripts
npm test

Place the five downloaded files in one directory before running the commands. The test file imports ./statistics.mjs, so the snapshot does not depend on this website being online after download.

What the tests cover

The checks include seed replay, distribution properties, histogram totals, sample-mean standard errors, Tukey outlier summaries, confidence-interval coverage, natural-frequency Bayes counts, weighted group rates, percentage changes, correlation examples, regression-to-the-mean selection, response mixtures, binary run lengths, axis scaling, and multiple-testing thresholds.

These are calculation and reproducibility checks. They do not prove that a teaching model is suitable for a real decision, that cited sources are error-free, or that the publication has been independently certified. Those boundaries remain part of the methodology and each guide.

Reuse and corrections

The files are public for inspection and reproduction. Public access does not grant a reuse license beyond applicable law. To report a discrepancy, email editorial@ertaiba.com with the revision, file name, expected result, and command output.