← Home
Standing page — linked from every claim check

How we check a claim

Every entry in the log follows the same four-step process, whether the claim leans toward or against a vaccine. The process doesn’t change based on which way the answer points.

1. We capture the claim exactly as it circulated

Before we check anything, we record the claim in its original form — a quote, a screenshot, a clip — along with where it appeared and when. We don’t paraphrase the claim into something easier to check. If the original wording is ambiguous, we say so and check the most natural reading of it.

2. We rank our sources

Not all sources carry the same weight, and we rank them in this order:

Primary

Raw data and original records

VAERS raw data, clinical trial results, FDA and CDC datasets, peer-reviewed studies, court and legislative records. This is the foundation of every verdict.

Secondary

Reporting on the primary data

News coverage, agency summaries, and public statements that describe or interpret the primary data. Useful for context, never the sole basis for a verdict.

Tertiary

Commentary and opinion

Social posts, blog commentary, and opinion pieces. We track these to understand how a claim is spreading, not as evidence for or against it.

Why this matters A claim can be widely repeated by secondary and tertiary sources while the primary data says something different. Ranking sources this way keeps the verdict tied to the record, not to how often something has been repeated.

3. We state the caveats every time

Some data sources are useful but easy to misread. VAERS is the clearest example — it’s worth explaining in full, since we cite it often:

  • VAERS is a passive reporting system. Anyone — a patient, a doctor, a family member — can file a report, for any reason.
  • Reports aren’t verified before they’re logged. A report existing means someone filed it, not that a medical review confirmed a link.
  • A report count alone doesn’t establish causation. Rising report volume can reflect rising public attention as easily as a rising rate of actual events.

We repeat this context every time we use VAERS data, rather than assuming a reader has seen it explained elsewhere in the site.

4. We publish one of four verdicts

We rate every claim on the same four-point scale. We don’t use a true/false binary, because most viral claims are partially accurate once stripped of context, and a binary rating would misrepresent that.

Supported

Primary sources back the claim substantially as stated, with no material missing context.

Partially Supported

Some part of the claim holds up, but missing context changes the overall picture.

Unsupported

Primary sources contradict the claim, or don’t show what the claim says they show.

Unable to Verify

No primary source currently exists to confirm or refute the claim either way.

Corrections

Every claim check displays a “last reviewed” date. If new primary data changes a verdict, we update the article, note the change in a visible changelog at the bottom of the piece, and list it on our public corrections log. We don’t quietly edit a published verdict.

What we don’t do

  • We don’t check claims we can’t trace to a specific, sourceable origin.
  • We don’t accept sponsor input on which claims we check or how we rate them.
  • We don’t use anonymous or unverifiable sources as the basis for a verdict.