Scholarly writing and research compliance. Use for CRediT, preregistration, Plan S, Nelson Memo, preprints, ORCID, LLM disclosure.

Academic writing and research methodology Systematic approaches for scholarly writing, research design, and academic communication. Research design fundamentals Research question development Literature review strategy Citation management Paper structure and writing IMRaD structure (scientific papers) Academic writing style Common writing problems Peer review and revision Responding to reviewers Handling criticism Grant proposals Proposal structure (NSF/NIH style) Budget justification Publishing strategy Journal selection Cover letter template Preprints and persistent identifiers Preprint servers (post a working version, get cited earlier) Preprints are author submitted manuscripts posted before formal peer review. Most journals now permit (or actively encourage) preprinting; verify the target journal's policy before posting if you're unsure. Server Field Notes arXiv (arxiv.org) Physics, math, CS, quantitative bio, econ, stats Oldest, largest; moderation by section. SSRN (ssrn.com) Social sciences, law, econ, finance Owned by Elsevier since 2016, some authors prefer SocArXiv as an alternative. bioRxiv (biorxiv.org) Biology Cold Spring Harbor Laboratory; widely cited. medRxiv (medrxiv.org) Health sciences, medicine Light moderation for safety claims; do not preprint clinical trial results without preregistration. EarthArXiv Earth, planetary, environmental sciences Community run via the OSF infrastructure. PsyArXiv Psychology + behavioral sciences OSF hosted. SocArXiv Sociology + social sciences OSF hosted; Elsevier independent alternative to SSRN. EngrXiv Engineering OSF hosted. Why preprint: Faster timestamp on priority claims, citable months before journal acceptance, and broader feedback before final review. (The 2026 OSTP Nelson Memo governs immediate public access to the peer reviewed version of federally funded US research at publication, it is not a preprint mandate. Preprinting is a separate, voluntary choice that complements but doesn't satisfy Nelson Memo compliance.) Persistent identifiers, get an ORCID and use DOIs ORCID iD (orcid.org), Free 16 digit identifier that unambiguously links you across publishers, funders, and institutions. Required by most major journals and funders. Connect it to your CV, manuscripts, datasets, software, and grant applications. DOI (doi.org), Persistent identifier for scholarly objects. Assigned automatically by journals at acceptance; you can also mint DOIs for your own datasets, code, posters, and preprints via Zenodo (zenodo.org) or your institutional repository, useful for citing non journal outputs. ROR (ror.org), Persistent identifier for research organizations, used by funders and publishers to disambiguate institutional names. Research ethics Ethical considerations checklist Avoiding research misconduct AI / LLM use in academic writing LLM assisted writing is the defining 2024 2026 ethics issue across academic publishing. Every major journal, funder, and ethics body has issued policy in this window, and policies are still tightening. Treat anything below as the floor, not the ceiling: read your target journal's current submission guidelines AND your funder's most recent policy before submitting. What's universally prohibited (as of 2026) LLM authorship. ICMJE, COPE, Nature, Science, NEJM, Cell, JAMA, Lancet, and the major university presses all explicitly prohibit listing an LLM (ChatGPT, Claude, Gemini, etc.) as a co author. LLMs cannot meet the accountability and approval criteria authorship requires. Use the acknowledgments section or a methods/disclosure statement instead, never the author byline. Generating fabricated citations. LLMs are well documented to produce plausible looking but fabricated DOIs, page numbers, and even author/title combinations that don't exist. Every citation in a manuscript must be verified against the source, the LLM induced fake citation rate has been a top retraction trigger since 2023. Generating data, results, or images. Synthesizing experimental data, fabricating figures, or using generative AI to "fill in" results that weren't actually obtained is research misconduct under COPE's definition. Undisclosed substantial use. Most journals require disclosure of any LLM use beyond trivial spell check/grammar assistance. Failing to disclose meaningful use can be grounds for retraction. What's typically permitted (with disclosure) Idea brainstorming and outlining Language polishing and grammar correction Translation of your own writing Code generation for analyses (with explicit testing) Summarization of your own notes or transcripts Generating boilerplate sections (cover letters, IRB language) that you then fact check and own Disclosure language (template) Most journals want a methods/acknowledgments statement that names the tool, version (if available), and what it was used for. Example: During the preparation of this work the author(s) used [tool name and exact version, e.g., GPT 5.6 Sol, Claude Opus 5, Gemini 3.7 Flash] in order to [specific use, e.g., language polishing of the introduction; drafting code for the cluster analysis in section 3.2]. After using this tool, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication. Adjust to match the target journal's exact required wording, Elsevier, Springer Nature, Wiley, Taylor & Francis, IEEE, and ACM each publish their own preferred language. Disclosure checklist before submission [ ] Read the target journal's AI/LLM policy in the current submission guidelines (policies updated multiple times per year). [ ] Read your funder's policy, some (Wellcome, NSF) have stricter rules than the journal. [ ] Add a disclosure statement using the journal's preferred language. [ ] Verify every citation manually, do not trust LLM output for any DOI, author list, page range, or quoted passage. [ ] Verify every numerical claim, date, and named fact. [ ] Re read the full manuscript to ensure your voice and argument structure dominate, not the LLM's. [ ] If the LLM was used for code or analysis, re run with seeds and verify all reported numbers reproduce. Detection tools (limited reliability) GPTZero, Turnitin AI Detection, Originality.ai, and similar tools have documented high false positive rates against non native English writers and against legitimately human written technical text, and high false negative rates against current models with light editing. Pangram Labs (pangram.com) has published more rigorous benchmarks and tends to outperform the older detectors at roughly equivalent settings, but is still not a definitive arbiter and shares the same fundamental limits when authors edit LLM output substantially. Detection output should never be the sole basis for a misconduct finding; journals that rely on it as a gatekeeping signal are increasingly walking that back. Treat detector results as a flag for follow up discussion with the author, not as evidence on their own. Productivity and workflow Writing routine Tools for academics Purpose Tools Reference management Zotero (recommended; open source, syncs across devices), Paperpile (web first, Google Docs integration), EndNote, Citavi. Note on Mendeley: Elsevier announced an end to Mendeley Desktop downloads in September 2022, then later restored availability. Mendeley Desktop remains legacy software with no further feature development; the Mendeley Reference Manager (web/cloud) is the actively maintained product. New users are better served by Zotero or Paperpile. Writing Scrivener, Overleaf (LaTeX collaboration), Word, Typst (newer LaTeX alternative) Collaboration Google Docs, Overleaf, Hedgedoc Version control Git / GitHub / GitLab (true version control). Word's Track Changes is collaborative editing, not version control , it lacks branching, merging, history rewrite, and atomic commits. Task management Todoist, Notion, Trello Focus Forest, Freedom, Cold Turkey Analysis R, Python, SPSS, Stata, NVivo, Julia Visualization R/ggplot2, Python/matplotlib + plotly, Observable, Tableau LLM assisted writing See "AI / LLM use in academic writing" section above for disclosure requirements before using any of these.

academic-writing