A university release appears. Several publications cover it. A social post says multiple sources confirm the discovery. Before counting those sources, ask whether they all point back to the same study.
This article describes our source-handling method. It is not a claim that institutional releases are inherently unreliable or that journal publication removes all uncertainty.
Count studies, not repetitions
Create one record for the underlying study, then attach its release, abstract, paper and later updates. Those documents may add information, but their existence does not automatically create independent replication.
If two articles describe the same participants and results, they should not be counted as two clinical trials merely because their headlines differ.
Record what you actually accessed
“Source reviewed” can mean several things. An abstract provides a condensed report. A full paper may provide additional methods and limitations. Supplementary materials may contain other details. A company announcement supplies the company's account of an event.
Our evidence records state which document was accessible. If a full paper could not be read, we do not imply that every table, subgroup or adverse-event detail has been checked.
Separate the event from the interpretation
A document can support “the company announced a trial” without supporting “the trial succeeded.” It can support “the authors reported an association” without supporting “the intervention caused the outcome.”
Write the smallest factual statement the source supports, then label any interpretation as interpretation. The more important the conclusion, the more important it is to preserve that boundary.
Follow versions without duplicating the discovery
A preprint and a later journal article may be versions of the same work. A correction may change a detail rather than create a new study. Record the relationship and inspect what changed before deciding whether the public summary needs revision.
Do not silently replace the old source while leaving a headline based on an earlier version. A version history is useful precisely because it shows whether the claim changed.
A practical source note
A good note contains the document title, organization or authors, publication date, identifier, access level and the exact claim it supports. Add the date you checked it. Store your own summary and links rather than assuming you may republish the entire source.
The evidence-note tool provides a local worksheet for this task. It does not fetch or validate the paper; it helps preserve what you actually know.
The point is accountability
Readers should be able to trace a claim without following an endless chain of rewritten stories. A source link is useful only when the article accurately represents the document at the other end.
The goal is not to prefer the longest document automatically. It is to know which document supports which claim—and to avoid turning repetition into evidence it does not supply.