Home > FAQ > How do I document citation chaining as a reproducible method?

How do I document citation chaining as a reproducible method?

October 26, 2026
citation chainingbackward and forward citation searchsearch documentationsystematic search methodsreference list searching

Set up citation chaining as a documented method by keeping a chaining log that records each seed paper, every backward and forward link you follow, the source used to find that link, and a keep-or-drop decision with a reason. The Cochrane Handbook treats searching reference lists and other sources as part of a planned search strategy and stresses documenting searches and managing records, so the log is the record that makes your chaining reproducible and defensible in a methods note.

Build the log before you click

Create a table with one row per linked work and these fields:

  • Seed paper: the paper you started from, with a stable identifier.
  • Direction: backward (works the seed cites) or forward (works that cite the seed).
  • Linked work: title and identifier of the work found.
  • Source used: where the link came from, such as a cited-by list, a reference list, or an API citation endpoint.
  • Decision: kept or dropped.
  • Reason: the specific criterion that drove the decision, such as topical relevance, study design, or population match.

Run the chain in order

  1. Fix your seed set. Choose the papers you will chain from and record why each is a seed. Expected result: a short, stable starting list you can rerun later.
  2. Pull backward links. Open each seed's reference list and log every candidate with the source noted. Expected result: a candidate list of earlier foundational work.
  3. Pull forward links. Use cited-by links to find later work that cites each seed. Google Scholar documents cited-by links, and the Semantic Scholar Academic Graph API provides citation data for papers. Expected result: a candidate list of later work building on the seed.
  4. Apply your keep-or-drop criteria to each candidate. Record the reason in the log at the moment of the decision. Expected result: a filtered set with an auditable trail.
  5. Repeat from kept papers if you need another round. Log the new seed and its links the same way. Expected result: a chain you can trace end to end.

Make it rerunnable and defensible

Record the source for every link so a reader can see whether a work came from a reference list, a cited-by list, or an API. Cochrane's guidance on documenting searches and managing records supports treating the log as part of your search documentation. In a methods note, describe your seeds, the directions you chained, the sources you used, and your keep-or-drop criteria; the log supplies the evidence for each step. Google Scholar also documents alerts and library links, which can help you keep a chain current, but the log remains the record of what you actually followed.

←
PreviousHow do I document a cited report removed from its publisher page but with a resolving DOI?
NextHow do I ensure a drug trial registry sweep misses no registries before handoff?
→