How to Prepare a Clean Source Pack for a Coauthor

A source pack is the bundle of records, files, and notes that lets a coauthor see exactly what you searched, what you kept, what you set aside, and why. A clean pack answers those questions without a meeting. It is the difference between a coauthor who can start writing and one who has to reconstruct your reasoning from scattered PDFs and half-finished notes.

The practical goal is narrower than "organize everything." You are building a handoff that a coauthor can audit and extend: a documented search, a deduplicated record set, a screening trail, and a short orientation note. The evidence base for this task is well established. Cochrane's handbook guidance on searching and selecting studies recommends planning searches from eligibility criteria, using multiple sources, combining free-text terms with subject headings, documenting the process, deduplicating records, screening in stages, linking multiple reports of one study, and rerunning searches before publication. PRISMA 2020 supplies reporting checklists and flow-diagram templates for transparent reporting of search, selection, and synthesis. NIH rigor and reproducibility guidance treats careful assessment of prior knowledge, transparent methods, validation, and reproducibility as core research practice.

This article turns that guidance into a working sequence: diagnose your starting state, assemble the artifacts, run checkpoints, and decide when the pack is ready to hand off.

Most messy source packs fail at the beginning, not the end. Before you touch a folder, decide which of these describes your project, because each needs a different first move.

• Scattered but complete. You have the searches and the PDFs, but they live in browser history, a downloads folder, and a chat thread. Your job is consolidation and documentation, not new searching.
• Complete but undocumented. You know why each paper is in the set, but that reasoning is only in your head. Your job is to write the trail down before it evaporates.
• Incomplete. You have a partial set and no clear eligibility criteria. Your job is to define criteria first, then search against them.
• Inherited. A departing student or collaborator left files behind. Your job is to verify what the records actually are before trusting the labels.

Write one sentence naming your state and the single biggest gap. That sentence becomes the first line of your orientation note and tells your coauthor what to expect.

Cochrane's guidance is explicit that searches should be planned from eligibility criteria. That ordering matters for a source pack: criteria decide what belongs, and the search is built to find it. If you search first and define criteria later, your pack will contain whatever happened to surface.

Write a short scope block with four parts:

1. Question. One or two sentences on what the pack is meant to support.
2. Include rules. Population, intervention or exposure, outcome, study type, date range, language.
3. Exclude rules. The near-misses you expect and why they are out.
4. Boundary cases. Named examples you are unsure about, with your provisional call and the reason.

The boundary-case list is the most valuable part for a coauthor. It shows where judgment was applied, so a coauthor can agree, disagree, or escalate rather than silently inheriting your call.

A search you cannot rerun is not a source pack; it is a memory. Cochrane recommends combining free-text terms with subject headings and using multiple sources, and it recommends documenting the process. Capture, for every search:

• the database or source searched;
• the exact query string, including field tags and limits;
• the date run;
• the number of records returned;
• who ran it.

Store this as a table, one row per search. If you used a scholarly search interface, keep the query text alongside the results you exported so the link between them survives. WisPaper's scholarly search interface is one place to run and capture academic paper searches; the discipline of recording the query string and date still belongs to you.

Cochrane also recommends rerunning searches before publication. Note in the pack the date of your last run and flag that a refresh is expected before submission. That single line prevents a coauthor from treating a stale set as final.

Cochrane recommends deduplicating records and screening in stages. Both steps produce artifacts your coauthor needs.

Deduplication. Export raw records, then remove duplicates before screening. Keep the raw export and the deduplicated set as separate files so the reduction is visible. Record how many records entered and how many remained.

Staged screening. Screen titles and abstracts first, then full text. At each stage, record the count in and the count out, and keep a reason for every exclusion at full-text stage. A coauthor who wants to challenge an exclusion needs the reason, not just the decision.

Study linking. Cochrane recommends linking multiple reports of one study. If the same study appears as a conference abstract and a full paper, group them under one study ID so your coauthor counts studies, not PDFs.

PRISMA 2020 provides flow-diagram templates that map naturally onto these counts. Even if you are not writing a systematic review, borrowing the flow structure gives your coauthor a one-glance view of how the set shrank.

A clean pack is a small, predictable set of files. Use the same structure every time so coauthors learn it once.

Keep file names stable and dated. A coauthor should be able to tell from the name alone whether a file is raw, screened, or final.

Once the set is screened, the included papers need a home that a coauthor can browse. Save or upload the included papers into a shared library and organize them so the structure mirrors your scope block. WisPaper's My Library is built for saving and uploading papers and organizing them in a library, which fits this step directly.

Two organizational habits pay off:

• Mirror the screening log. If your log groups papers by theme or study type, group the library the same way. A coauthor moving between the log and the library should never have to re-map the categories.
• Name the boundary cases. Put provisional or disputed papers in a clearly labeled group rather than mixing them with confirmed includes. That keeps the confirmed set clean while preserving the trail.

If your coauthor needs to ask questions of the collected papers rather than read them all, a question-and-answer pass over the library can surface what the set does and does not cover. WisPaper's Library QA supports asking questions based on papers uploaded to My Library or added from search results, which is useful for orienting a coauthor to a large set before deep reading.

A source pack usually ends in a reference list, and a reference list is where small errors become embarrassing ones. Before handoff, check that every citation in the list corresponds to a record you actually screened in, and that no citation was added from memory or a secondary mention.

For BibTeX-based workflows, citation screening can be part of the pack build. WisPaper's TrueCite checks BibTeX against real academic databases to flag hallucinated references. Run it on the pack's reference file and resolve every flag before the coauthor receives the list, so the coauthor inherits a checked list rather than a to-do.

The orientation note is the pack's front door. Keep it to one page and answer, in order:

1. What this pack is for and what it does not cover.
2. Where you started and what state the evidence was in.
3. The scope rules and the boundary calls you made.
4. The search dates and whether a refresh is pending.
5. The counts at each screening stage.
6. The open questions and who owns each one.

Write it last, after the artifacts are final, so the numbers in the note match the logs. A coauthor who reads only this page should know whether to start writing, start screening, or come back with a question.

A doctoral student is leaving a review project and handing the source pack to a faculty coauthor. Here is how the sequence plays out.

Starting state. The student has searches in browser history, PDFs in a downloads folder, and screening decisions in a notebook. The state is "scattered but complete." The biggest gap is that no search was logged.

Scope. The student writes the scope block from the review's protocol, including three boundary cases: a preprint with no peer-reviewed version, a study in a language the team cannot read, and a paper that reports the same cohort as an included study. Each gets a provisional call and a reason.

Search record. The student reruns each search, records the query string, source, date, and count, and saves the raw exports. The rerun also confirms the counts match the notebook.

Deduplication and screening. Raw exports are deduplicated, then screened in two stages. The student records counts in and out at each stage and writes a reason for every full-text exclusion. The preprint and the duplicate-cohort paper are linked or excluded according to the boundary calls, and the study map records the link.

Library. Included papers go into a shared library organized by the same themes as the screening log, with boundary cases in a separate group.

References. The reference list is exported as BibTeX and screened for hallucinated entries; every flag is resolved.

Orientation note. The student writes the one-page note, checks that the counts match the logs, and flags that a search refresh is pending before submission.

Handoff. The faculty coauthor reads the note, opens the screening log to check two exclusions, and starts drafting the methods section from the search log and flow counts. No meeting was needed to explain the pack.

Run these checkpoints in order. Each one has a clear pass condition and a stop rule.

Escalate to the coauthor or project lead when a boundary case changes the scope, when a search cannot be rerun, or when a citation cannot be traced. Those are decisions that belong to the project, not to the pack builder alone.

The pack is ready when a coauthor can answer four questions from the files alone: What was searched and when? What was kept and why? What was excluded and why? What is still open?

A practical final test is the cold-read check. Ask a coauthor who has not seen the project to open the orientation note and the screening log and state the pack's scope and its two biggest open questions. If they can, the pack is clean. If they have to ask you, the note is missing something.

A second test is the rerun test. Pick one search from the log and rerun it. If the query string and source are recorded well enough to reproduce the search, the log is doing its job.