Published November 5, 2026 | Version 1.4
Software Embargoed

How Wide Should a Scenario Set Be? Replication archive (preregistered ECB SPF replication, US SPF analysis, herding and contested-axis simulations, density forecasts, post hoc checks)

Authors/Creators

  • 1. ROR icon Moscow Technical University of Communication and Informatics

Description

Replication archive for the manuscript "How Wide Should a Scenario Set Be? Compressing Expert Judgments into Scenarios, with a Preregistered Replication on Two Forecaster Panels" (E. Vakhovskii, MTUCI; prepared for Systems, MDPI).

Contents: the OSF-registered analysis plan and frozen script of the ECB Survey of Professional Forecasters replication (registration ma4jr, 22 October 2026); the deviations log with every input hash and the corrections to the registration's stated size and power; the single confirmatory output; the exploratory variants; the analysis plans, code and derived result files of the US SPF study; the plans, code and cell-level results of the herding and contested-axis simulations; the SPF density-forecast block; the figures; and the post hoc checks made after the registered run (temporal-dependence nulls; maximum-volume simplex, KKZ with an extreme start, FurthestSum and k-medoids comparators; matched-width envelopes; size calibration and power of the registered test). Supplement S1 of the manuscript is included.

Survey microdata are not redistributed; sources, versions and SHA-256 hashes are listed in README.md and data/README_data.md. Every script resolves its paths from the archive root, and run_all.sh runs the blocks in dependency order.

The files are under embargo until 1 September 2027, when the embargo of the OSF registration ends. Licences: code MIT; text, plans, logs and derived tables CC BY 4.0; Eurostat-derived files CC BY 4.0 (European Union, Eurostat).

Files

Embargoed

The files will be made publicly available on September 1, 2027.

Reason: The archive contains the plan and frozen script of OSF registration ma4jr, which is under embargo until 1 September 2027; the files will be released when that embargo ends, or earlier on publication of the article.

Additional details

Related works

Is supplement to
Other: https://osf.io/ma4jr (URL)

Software

Programming language
Python