demog-data-and-reproducibility

Data & Reproducibility (demog-data-and-reproducibility)

Demography asks every accepted manuscript for a data availability statement and encourages reproducible code. The journal does not maintain its own repository — you choose a FAIR repository and provide persistent identifiers. Build the materials as you go so acceptance does not stall.

When to trigger

• Building the reproducibility/replication materials and the data availability statement
• A manuscript reached acceptance and you must finalize the statement and deposit
• Data cannot be fully shared (privacy, confidentiality, proprietary/provider restrictions) and you need the exemption language
• Choosing a repository and persistent identifier

What Demography requires (verify current wording on the ethics/disclosures page)

1. Data availability statement. Accepted manuscripts must include a statement confirming whether and how the data have been shared. For publicly available data, describe the access method with a persistent identifier (DOI or accession number).
2. Persistent identifiers + FAIR repository. Because Demography hosts no repository, deposit data and materials in a FAIR-aligned repository (e.g., ICPSR/openICPSR, Harvard Dataverse, Zenodo, OSF) and cite it with a permanent link — not a personal website or generic cloud folder. Find one via FAIRsharing.org or re3data.org.
3. Reproducible code (encouraged). Provide code that regenerates the results, with software versions recorded and informative comments explaining each step. Treat this as expected, not optional, for a quantitative demographic paper.
4. Restricted data. Where access is limited to protect confidential or proprietary information, describe the conditions limiting access and how others may obtain the data, while still aligning with FAIR principles as far as possible.

Build-as-you-go checklist

• Data availability statement drafted (public / restricted / on request) with persistent IDs
• Data + materials deposited in a FAIR repository with a DOI/accession number
• One master script regenerates every table, figure, life table, and decomposition
• README documents data provenance/vintage, construction of rates/exposure, and how to reproduce each exhibit
• Code commented and software/package versions recorded (renv.lock / requirements.txt / installs)
• Seeds set and reported for bootstrap, simulation, microsimulation
• Exhibit numbers in the manuscript match the deposited output exactly
• Restricted data: exemption note + access conditions + synthetic data where feasible

Anti-patterns

• Treating the data statement as a last-minute formality at acceptance
• A personal URL or generic cloud link instead of a persistent identifier in a FAIR repository
• Undocumented, un-seeded, uncommented code that "works on my machine"
• Claiming data are restricted without describing the access conditions
• Calling the $1,000 editorial-management fee an "open-access" or "data" charge (it is neither)

What a reproducibility reviewer expects at Demography

Demography, the Population Association of America flagship published open access by Duke University Press, draws referees who routinely re-derive a denominator or re-run a Kitagawa split by hand. The deposit must let them do so. Calibrate the package against the data class you actually used.

Data class	Typical source	Deposit obligation
Fully public	HMD, HFD, IPUMS, WPP, published vital statistics	Extract code + constructed analytic file (where license permits) + persistent ID
Public-but-licensed	IPUMS, DHS, redistribution-restricted census	Extract specification + code, not raw microdata; cite the provider's persistent access path
Restricted/confidential	Linked administrative records, RDC-only census, linked mortality files	Code + synthetic stand-in file + exact access-conditions note

Referee-pushback patterns and the Demography-specific fix

• "I cannot tell which HMD/HFD vintage produced these rates." -> Record the release label and download date in the README and a code comment; vintages revise historical rates.
• "The decomposition components in the deposit do not sum to the headline number." -> Add an assertion that components reconcile to the total e0 (or TFR) change before you deposit; ship the check passing.
• "Your bootstrap intervals are not reproducible." -> Set and report a seed for every resampling, microsimulation, and graduation step; an un-seeded interval reads as un-reproducible.
• "You say data are restricted but give no way to obtain them." -> Name the data enclave or application pathway and the access conditions; a bare "restricted" reads as evasion.

Output format

【Data statement】public / restricted / on request — drafted with persistent ID? [Y/N]
【Repository】FAIR repo + DOI/accession (not personal URL)? [Y/N]
【Reproducible code】master script + comments + recorded versions + seeds? [Y/N]
【Exhibits match deposit?】[Y/N]
【Restricted data?】access conditions described + synthetic data where feasible?
【Next】demog-review-process

Supplementary resources

• ../../resources/external_tools.md — FAIR repositories and reproducibility tooling
• ../../resources/official-source-map.md — Demography data-availability and reproducible-code policy