Skip to content

Reproducible Research Experiment Prompts

Six prompts for researchers using FCC as an experimental platform. They emphasize reproducibility, FAIR data principles, and transparent methods reporting. Each prompt assumes the reader has a specific hypothesis and a dataset in hand.

Personas Used

Persona ID Full Name Category Role in Prompts
RSN Reproducibility Sentinel open_science Determinism, versioning, seeds
FDS FAIR Data Steward open_science Findable, Accessible, Interoperable, Reusable
CSL Citizen Science Liaison open_science Community engagement, lay summaries
ESC Experiment Scientist open_science Hypothesis design, statistics

Prompt 1: Designing a Reproducible Simulation Study

Audience: Scientific Difficulty: advanced Personas: ESC, RSN

Context

A researcher wants to compare two prompt strategies for a diagnostic-assist persona.

Prompt

ESC drafts; RSN reviews.

Find: state the hypothesis formally (H0, H1, effect size of interest),
identify the dependent variable (CLEAR+ Efficacy by default), and
enumerate the confounders that must be controlled.

Create: write a preregistration document with (a) sample size
calculation, (b) random seed plan (master seed + derivations), (c)
analysis plan (primary test, secondary tests, corrections), (d)
stopping rule.

Critique: RSN to identify any experimenter degrees of freedom that
could inflate the false-positive rate, and propose guardrails
(blinded analysis, held-out test set, audit hooks in the event bus).

Expected Output

  • Preregistration document
  • Seed and guardrail plan

Prompt 2: FAIR Data Audit

Audience: Scientific Difficulty: intermediate Personas: FDS

Prompt

Audit the attached dataset against the 15 FAIR sub-principles. For
each sub-principle, return a score in {0, 1, 2} with a one-sentence
rationale and a concrete remediation step where the score is below 2.
Produce a summary heatmap description usable in a README.

Expected Output

  • Sub-principle scorecard
  • Remediation list
  • Heatmap description

Variations

  • Compare two datasets and highlight the larger FAIR gap

Prompt 3: Methods Section Grounded in R.I.S.C.E.A.R.

Audience: Scientific Difficulty: intermediate Personas: ESC, RSN

Prompt

Write a Methods section (600-800 words) for a journal submission
that describes a simulation experiment using three FCC personas.
Structure the section so every claim cites either (a) a specific
R.I.S.C.E.A.R. slot in the persona YAML, (b) a benchmark run ID,
or (c) a seeded code commit hash. Finish with an "Availability
Statement" paragraph.

Expected Output

  • Methods draft
  • Availability statement

Prompt 4: Replication Package Structure

Audience: Scientific Difficulty: advanced Personas: RSN, FDS

Prompt

Design the directory layout of a replication package that someone
can clone and run end-to-end in < 30 minutes on a laptop with the
mock simulation engine. Include:
- `data/` (inputs, with sha256 manifest)
- `code/` (FCC version pin, environment lockfile)
- `results/` (committed outputs, trace hashes)
- `docs/` (method note, lay summary)
- `Makefile` targets: `make env`, `make run`, `make verify`

Provide the Makefile content and the exact command a reviewer
should run to confirm byte-for-byte reproduction of `results/`.

Expected Output

  • Directory layout
  • Makefile
  • Verify command

Prompt 5: Community Engagement and Lay Summary

Audience: Scientific Difficulty: intermediate Personas: CSL, FDS

Prompt

Turn a technical preregistration into a 400-word lay summary suitable
for a citizen-science partner organization. Avoid jargon; use one
concrete analogy from daily life; invite two specific forms of
participation (data contribution and interpretation review). Include
a short FAQ covering "why this matters," "who funds it," and "how
findings will be shared."

Expected Output

  • Lay summary
  • FAQ

Prompt 6: Publishing to Open Repositories

Audience: Scientific Difficulty: intermediate Personas: FDS, RSN

Prompt

Draft a publication plan that places the replication package on
Zenodo with a DOI, the preprint on a suitable archive, and the
persona YAMLs on a protocol registry. Specify:
- License choices (code vs data vs prose)
- Metadata per target (Zenodo JSON, preprint cover letter)
- Embargo plan, if any
- A README badge list that links all artifacts

Critique the plan for any single point of failure in long-term
preservation.

Expected Output

  • Publication plan
  • Preservation critique

See Also

  • src/fcc/observability/ for audit hooks
  • Guidebook Chapter 10 (Open Science)
  • docs/tutorials/sample-prompts/open-science-research-prompts.md