Open Science & Transparency (cogpsych-open-science-and-transparency)

Because the contribution is a model fit to data, reproducibility here means more than open data: it means the model code and analysis scripts regenerate every reported fit. Authors are strongly encouraged to share data, model code, and materials. This skill prepares the deposits and the Elsevier declarations early, so the modeling is reproducible at submission rather than promised at acceptance. Confirm the current policy wording on the journal's guide for authors (检索于 2026-06；以官网为准).

When to trigger

• Building the data, model-code, and materials deposits
• Completing Elsevier's research-data statement and competing-interest declaration
• Deciding whether (and how) to claim a restriction on sharing sensitive/third-party data
• Linking a preregistration and reporting its status
• A reviewer or editor flagged transparency or reproducibility

What to prepare (verify current wording on the official page)

1. Open data. Deposit trial-level data with a codebook/data dictionary in a repository that mints a persistent identifier (DOI) (OSF, Mendeley Data, Dataverse, Zenodo). State availability in the Elsevier research-data statement.
2. Model and analysis code (the venue-specific must). Deposit the model-fitting code and analysis scripts with seeds and pinned versions; the reported fits, figures, and tables must regenerate in a fresh session. This is what makes a model-driven paper reproducible.
3. Materials. Deposit stimuli, item pools, counterbalancing schemes, and task code with a DOI.
4. Preregistration (where applicable). For confirmatory tests and pre-committed model comparisons, link a preregistration; keep confirmatory vs. exploratory consistent with it.
5. Elsevier declarations. Complete the declaration of competing interest, funding/role statement, author contributions, and any required ethics/consent statements per Elsevier policy.

Build-it-right checklist

• Data deposited with a DOI + a codebook/data dictionary
• Model-fitting code deposited; all fits regenerate in a fresh session (run log included)
• Analysis scripts deposited; seeds set, package versions pinned
• Materials (stimuli, item pools, task/counterbalancing) deposited with a DOI
• Preregistration linked (where applicable); confirmatory vs. exploratory consistent
• Research-data statement + competing-interest declaration + funding/contributions completed
• Any restriction on sharing clearly justified (what is withheld, why, and the access path)

Restrictions (handle honestly)

• If data/materials cannot be fully shared (sensitive populations, licensed/proprietary stimuli, IRB limits), state exactly what is withheld, why, and how others can access or approximate it (application path, synthetic data, code release). Code transparency is usually still expected even when data are restricted.

Reproducibility statement — worked draft (illustrative)

A model statement for the recognition-memory program; confirm required headings on the official page.

Open Practices
Data:        Trial-level confidence-ROC data for Experiments 1-3 are at OSF
             (DOI 10.XXXX/osf.io/abcde), with a codebook.
Model code:  The UVSD/DPSD fitting code (R/Stan) reproduces all reported fits,
             model-comparison values, and figures in a fresh session; a seeded
             run log and pinned environment are included (DOI 10.XXXX/.../fghij).
Materials:   Stimulus pools, list construction, and task code are deposited
             (DOI 10.XXXX/osf.io/klmno).
Preregistration: The model space and comparison criteria were preregistered
             before data collection (osf.io/pqrst); one post hoc bias analysis
             is reported as exploratory.
Declarations: Competing interests: none. Funding and author contributions as
             stated. (If applicable: data-sharing restriction + access path.)

Transparency-readiness decision table

Reviewer / editor pushback and the venue fix

• "I can't reproduce your model fits" → ship seeded model code + a pinned environment + a fresh-session run log; the fits must regenerate, not just the data exist.
• "No DOI, just a personal link" → swap transient links for persistent identifiers (OSF, Mendeley Data, Zenodo).
• "Statement says open but the repo is empty" → deposit first, write the statement from live DOIs.
• "Competing-interest/data statement missing" → complete the Elsevier declarations before upload.

Anti-patterns

• Sharing data but not the model-fitting code (the fits cannot be reproduced)
• "Available on request" instead of a persistent identifier
• Code that does not regenerate the reported fits in a fresh session
• Claiming a sharing restriction without a justification or an access path
• Omitting the Elsevier competing-interest / research-data declarations

Output format

【Open data】deposited + DOI + data dictionary? [Y/N]
【Model code】deposited + fits regenerate in a fresh session? [Y/N]
【Analysis scripts】seeded + versions pinned? [Y/N]
【Materials】deposited + DOI? [Y/N]
【Preregistration】linked + consistent (where applicable)? [Y/N/NA]
【Elsevier declarations】competing interest + research-data + funding? [Y/N]
【Next】cogpsych-review-process

Supplementary resources

• ../../resources/external_tools.md — OSF, Mendeley Data, Zenodo, Stan/JAGS, renv, preregistration templates
• ../../resources/official-source-map.md — Elsevier research-data and code-sharing policy