Evaluating a Source Using Authority, Evidence, Date, Purpose, and Corroboration

No source is credible in the abstract. A Pulitzer-winning newsroom can publish a claim it did not verify. A peer-reviewed journal can contain a paper whose conclusions do not follow from its data. The unit of evaluation is never the outlet. It is always the specific claim and the evidence chain that supports it.

Most people learn to evaluate a source by checking a checklist: is the author an expert, is the website reputable, is it recent enough. That approach treats the source as a standalone object with a fixed credibility score. It fails because it never asks whether this particular sentence can be traced to something real.

The five criteria that matter are authority, evidence, date, purpose, and corroboration. Apply them to a single sentence at a time, not to the masthead or the domain. A named outlet is not a traceable source if the claim does not appear in it, or if the outlet sits behind a paywall and no archive exists. The verification path for any changing fact must be stated explicitly: who measured it, when, and where the current figure is published.

Apply these criteria in sequence. A failure at any single step is enough to drop the claim from your argument. Start with authority because if you cannot identify who is speaking and why they are in a position to know, nothing else matters. But authority alone is the weakest criterion. An authoritative source can be wrong, outdated, or advancing an undisclosed agenda.

Authority: Name the Person, Not the Platform

Attribution specificity is the first test. A claim attributed to "experts say" or "studies show" has failed at the first hurdle. These are pseudo-attributions that invoke authority without accountability. You need a named researcher, a named institution, a named study with access. If the source is a publication, name the reporter who did the original work, not the outlet that syndicated it. The Reuters Handbook of Journalism requires that every fact be traceable to an original source. Apply that same standard to your own reading.

Evidence: Close the Claim-Defensibility Gap

Evidence is the reason given for accepting a claim. The claim-defensibility gap is the distance between what the evidence actually supports and what the sentence asserts. A claim can be factually true and still indefensible if the cited evidence does not reach it. A study finding a correlation in a specific population does not support a headline asserting causation in the general population. Measure this gap by reading the original source yourself. If you have not read it, you are evaluating someone else's paraphrase.

Date: State the Verification Path for Changing Facts

A fact that changes on a schedule needs a verification path: the named body that publishes the current figure and the place it appears. Cite the year of the figure you are using. Never write "currently" or "as of today" about a number revised annually. The date criterion fails when the reader cannot independently confirm that the figure is still in force.

Purpose is the criterion most readers skip because it feels subjective. It is not. Determine a source's purpose by asking two concrete questions: who paid for this to exist, and what action does the text want the reader to take. A white paper from a trade association has a different purpose than a meta-analysis in an independent journal. Both can contain true claims. Both select and frame evidence to serve that purpose. The null hypothesis default is your baseline stance: non-acceptance, not disbelief. The burden is on the text to move you with evidence and reasoning. A source that triggers an emotional response or aligns with your existing beliefs has not met a lower standard.

The charitable reading principle requires that you engage the strongest plausible formulation of a claim before you critique it. This is a practical discipline that prevents you from attacking a straw person. The motte-and-bailey structure is one of the most durable patterns in public argument: a controversial broad claim retreats to a trivially true narrow one when challenged, then re-extends. Catch it by identifying the bailey claim and testing whether the evidence cited actually supports it, or only the motte.

Corroboration means a claim is supported by multiple independent sources, not by one source quoted in multiple places. A statistic that appears in twenty articles all citing the same press release has not been corroborated. It has been amplified. Source triangulation requires that you trace each appearance of a claim back to its origin point. If three different outlets all attribute a finding to the same unnamed "study," you have one source, not three.

A fact-checking workflow for corroboration follows three steps. First, identify the original source for the claim. If no original source is named, the claim is untraceable. Discard it. Second, locate at least one independent source that gathered its own data or conducted its own analysis and reached a compatible conclusion. Third, check whether any credible source has directly contradicted the claim. The absence of contradiction is not evidence. But direct contradiction from a source with equivalent or greater authority is a red flag. Suspend judgment.

Criterion	Test Question	Failure Condition
Authority	Who is making this claim, and what gives them access to the facts?	Attribution is to a vague group ("experts") or absent entirely.
Evidence	What primary data or document supports the claim, and can I access it?	The cited source does not contain the claim, or the link is circular.
Date	When was this fact established, and who publishes the current figure?	The date is missing, or the fact changes on a schedule with no update path.
Purpose	Who funded or published this, and what action does it urge?	The funding or advocacy interest is undisclosed or obscured by neutral framing.
Corroboration	Has an independent source gathered its own data and reached a compatible conclusion?	Multiple citations all lead to the same single origin point.

Distinguish a primary source from a secondary source at the sentence level, not the document level. A single book can contain both. A primary source is the original data, document, or firsthand account. A secondary source is someone else's interpretation or summary of that primary material. When a journalist writes "according to a government report," the article is a secondary source for the claim. The government report is the primary source. If you cite the article without reading the report, you are trusting the journalist's paraphrase. The claim-defensibility gap widens every time a claim passes through another layer of interpretation.

Source traceability is an ordinal scale: a claim is either untraceable, traceable to a named source, or traceable to primary evidence. A citation to a named outlet that is inaccessible is untraceable in practice, regardless of how prestigious the outlet is. If you cannot put your finger on the sentence in the original document that supports the claim, you do not have a source. You have a rumor about a source.

When the Source and the Writer's Conclusion Diverge

A common failure in nonfiction writing is citing a source that supports a weaker version of the claim the writer actually makes. The writer draws a conclusion the source does not reach, then cites the source as though it does. Apply the source-claim distinction: excerpt the exact sentence from the source that you believe supports your claim. Place it next to your own sentence. If they say different things, you have a gap. Close the gap by narrowing your claim or finding better evidence. Do not close it by adding qualifiers that make the claim unfalsifiable.

Attribution specificity is the most reliable single indicator of whether a claim can be evaluated at all. The spectrum runs from no source, through vague attribution ("studies show"), to named institution, to named researcher, to named study with access. Each step to the right makes verification possible. Each step to the left signals that the writer either could not find the source or does not want you to check it. A text that relies heavily on "research shows" constructions asks you to accept the writer's authority in place of evidence. That is sometimes necessary in daily journalism. It is never acceptable in your own nonfiction writing.

The evidence-to-claim latency is another structural signal worth watching. This is the number of sentences between the introduction of evidence and the claim it supposedly supports. When the gap is short, the reader can hold both in working memory and judge the connection. When the latency is long, intervening material reshapes what the reader thinks the evidence established. A writer who buries the evidence three paragraphs before the conclusion is exploiting this effect, whether consciously or not.

Use this repeatable sequence on any claim you plan to cite or publish. It takes less time than fixing a correction later.

1. Extract the claim. Write it as a single declarative sentence with no qualifiers. This is what you are verifying.
2. Locate the original source. Follow every citation to its origin. If the trail ends in another secondary source that also does not cite primary evidence, the claim is not verified.
3. Read the source. Do not skim the abstract or the executive summary. Find the sentence, table, or data point that directly supports the claim. If it is not there, the claim fails.
4. Check the date and update path. For any fact that changes, identify who publishes the current figure and where. Cite the year of the figure you are using.
5. Triangulate. Find one independent source that gathered its own data and reached a compatible conclusion. If none exists, state that the claim rests on a single source.

This workflow is the minimum standard for evidence-based editing. A manuscript that has not been through it is not ready for publication.