What Does "Contested" Mean? How WasThatTrue Labels Uncertain Evidence

TL;DR: A contested fact check means credible sources or experts disagree about a claim, so a single true-or-false verdict would mislead readers. It is not the same as unverifiable, which describes the opposite problem: too little evidence rather than too much disagreement. Good fact-checkers use a contested label to keep their work honest, and viewers should treat it as a signal to read the sources for themselves.

Some claims do not have a clean answer.

A fact-checker who says "true" on a real disagreement is bluffing. A fact-checker who says "false" on the same claim is bluffing in the opposite direction.

That is why the contested label exists.

When credible experts disagree, the honest verdict is to say so. Not in a footnote. Not buried under caveats. As the actual label, with the conflicting sources attached.

This post walks through what a contested fact check is, when to use it, how it differs from unverifiable, and how WasThatTrue applies it. The goal is simple: give viewers a label they can trust, even when the evidence is messy.

What Does "Contested" Mean in a Fact Check?

Contested is a fact-check label that flags claims where credible sources reach different conclusions from the same evidence. It sits between true and false on purpose: it tells the reader the disagreement is real, not that the fact-checker could not make up their mind. Reaching for true or false on a contested claim is the actual mistake.

That last sentence is the whole idea.

Plenty of fact-checkable claims have one clear answer. Some do not. The labels we use should reflect that, instead of pretending every question is a coin flip.

Contested is not the same as "mostly true" or "mostly false". Those ratings still pick a side. Contested says the credible evidence does not pick a side at all.

It is also not a polite way to avoid a verdict. A good contested rating still does the work: it identifies the disagreement, names the sources on each side, and explains what the split is about.

Why Refusing to Pick True or False Is the Honest Choice

Fact-checking has a precision problem.

A binary verdict is satisfying. It feels decisive. But applied to a claim where serious researchers disagree, it manufactures certainty the evidence does not support.

The International Fact-Checking Network sets standards that signatories must meet, including a commitment to fair, non-partisan reporting and a transparent methodology. That standard cuts both ways. Hiding genuine disagreement behind a single verdict fails fair reporting just as much as ignoring strong evidence does.

The Duke Reporters' Lab tracks active fact-checking organisations worldwide and now counts hundreds of active outlets across more than 100 countries. That growth has put fact-check labels in front of more readers than ever, which makes accuracy in the labels themselves more important, not less.

A wrong "false" on a real disagreement also has a long-term cost. Once a reader sees a confident verdict that turns out to be premature, the next contested-but-checked claim looks like another guess. The label loses its power.

Contested protects that trust. It says: we did the work, we found a real split, here are the sources.

When Should a Claim Be Labelled Contested?

A claim should be labelled contested when well-resourced experts arrive at opposite answers, when the underlying evidence is genuinely mixed, or when the question rests partly on values or definitions, not only facts. If any of those triggers fire, forcing a single verdict misrepresents the state of the evidence.

Three triggers, in plain terms.

The first is direct disagreement among credible sources. Two well-resourced research teams using sound methods reach opposite conclusions. Neither is fringe. The disagreement is the finding.

The second is genuinely mixed evidence. Some studies say yes. Some studies say no. A meta-analysis might exist, but the underlying studies vary in quality, population, or method enough that a clean summary is not yet possible.

The third is the trickiest. A claim can sound factual but rest on a definition that experts argue about. "Crime is rising" depends on how you measure crime and over what window. A fact-check that picks one definition and rules on the claim is really just ruling on the definition.

When any of those triggers fire, a contested label is the honest one.

Contested vs Unverifiable: What Is the Difference?

Contested means experts have studied the claim and disagree. Unverifiable means the evidence needed to evaluate it does not exist yet or cannot be reached. Both labels are honest, but they describe opposite situations: one has too many credible answers, the other has none.

The two get confused often, so the line matters.

A claim like "the long-term health effects of a brand new additive are minimal" is unverifiable today, because the long-term studies do not exist. There is no disagreement to summarise, just an absence of data.

A claim like "a particular minimum-wage rise reduced employment" is often contested, because well-respected economists reach different conclusions from overlapping data sets. The evidence exists. It just points in more than one direction.

Using unverifiable on a contested claim hides the disagreement. Using contested on an unverifiable claim invents a debate that does not really exist. The labels are not interchangeable.

How Other Fact-Checkers Handle Uncertainty

Plenty of established fact-checkers already publish nuanced labels, even if they call them something else.

Snopes uses a Mixture rating for claims that contain a blend of accurate and inaccurate information. It also has Mostly True, Mostly False, and Outdated ratings for finer distinctions. Mixture is the closest cousin to contested in widespread use today.

PolitiFact runs a six-point Truth-O-Meter scale that includes Half True and Mostly True ratings, alongside a Pants on Fire verdict for ridiculous falsehoods. Half True is meant for claims that are accurate but leave out important context, which overlaps with the definitional trigger for contested.

Reuters publishes a standards and values document that emphasises sourcing, transparency, and independence. Its fact-check write-ups tend to spell out where the evidence is mixed rather than always reach for a single tag.

Google's ClaimReview structured data spec also accepts custom ratings, so publishers can mark something as "mixed" or "disputed" in the structured data Google indexes for fact checks. The mechanic for surfacing nuance already exists. The harder question is whether each publisher uses it.

The picture across the industry: nuanced verdicts are common, but they often hide inside a five or six step rating system. A clean standalone contested label keeps the disagreement visible at a glance.

Real Examples of a Contested Fact Check

Three quick examples make the criteria concrete.

Nutrition."Saturated fat causes heart disease." Mainstream dietary guidance has long pointed in that direction, but newer reviews of the underlying randomised trials have softened the link or pushed back. Credible nutrition researchers genuinely disagree about the strength of the relationship. That is a contested claim, even if a strong opinion dominates the public conversation.

Economics."A higher minimum wage costs jobs." Well-cited economists publish careful studies on both sides, using different regions, time windows, and methods. Summarising the entire literature as either true or false hides the split. The evidence exists. The split is the finding.

Policy outcomes."This city's housing policy reduced rent." The data might be partial, the time window short, and the comparison cities arguably different. Two honest analysts can reach different conclusions from the same dataset. That is the third trigger from earlier: the question depends partly on definitions and choices, not only facts.

None of these are dodges. Each is a place where a binary verdict would do more damage than good. If you are watching a claim like this in a long video, our post on why YouTube misinformation is harder to fact-check covers why the platform makes the problem worse.

How WasThatTrue Decides a Claim Is Contested

WasThatTrue marks a claim contested when its source retrieval surfaces credible publications reaching different conclusions on the same point. The verdict card shows the conflicting sources side by side, with publication, headline, and date, so the viewer can read both before deciding. The label is never used to dodge a verdict where the evidence is one-sided.

The principle behind that is the same one that runs through the rest of the product: the viewer decides.

A contested verdict in the extension does three things. It signals that this is not a clean call. It shows the sources that disagree. And it explains what the disagreement is about, so the viewer can tell whether it is a methodological split, a definitional difference, or something more political.

If you want the broader workflow that sits behind every verdict, our step-by-step YouTube fact-checking workflow covers it. If you want to see the product itself, the Chrome extension that surfaces contested verdicts is the entry point.

For data handling, the how source data is handled page explains what gets processed and what does not. The free and Pro plans set the limits and what each tier includes.

The contested label is not a hedge. It is the verdict that respects the reader enough to tell them when the evidence is actually split.

Frequently Asked Questions

Is contested just a way of avoiding a fact check?

No. The label only applies when credible publications genuinely land on different sides of a claim. If the evidence points one way and the fact-checker still refuses to commit, that is dodging, not contested. A real contested verdict attaches the disagreeing sources to the rating, so readers can verify the split exists.

What is the difference between contested and disputed?

In everyday use the words overlap, but in fact-checking contested usually means credible experts genuinely disagree. Disputed can also describe a single side rejecting an otherwise well-supported claim, which is a different situation. Most fact-checkers prefer contested because it implies the disagreement runs deeper than one party objecting.

Can a claim move from contested to true or false over time?

Yes. As new studies replicate, methods improve, or definitions sharpen, a contested claim can settle into a clearer verdict. The opposite also happens: a claim once treated as settled can become contested when fresh research challenges the consensus. Verdicts are snapshots of the evidence, not permanent rulings.

Why do not all fact-checkers use a contested label?

Some publishers prefer a sliding scale like Snopes' Mixture rating or PolitiFact's Half True, and some only publish clear true or false verdicts to keep their work simple. Both approaches work, but they can flatten real disagreement. A dedicated contested label keeps that nuance visible instead of folding it into a softer rating.

How does WasThatTrue label a contested claim?

On a contested claim, WasThatTrue surfaces the credible publications that disagree, then presents them on a single verdict card with publication, headline, and date for each. You read both sides on the same screen, with no forced verdict in between.