National’s classroom comparison: $1.2m versus $620,000

We select specific, testable assertions from attributable statements and reader submissions. National and Labour face the same criteria. This is selected coverage, not a party scorecard or a claim that every campaign statement has been checked.

Evidence and ratings

We prioritise original statistics, legislation, agency documents and parliamentary records. A party’s statement establishes what it said; it is not independent proof against another party. For claims about a published plan, a reproduced calculation establishes its arithmetic, not whether its forecasts will happen.

True

The material factual assertion is supported by strong evidence.

Mostly True

Substantially correct, with relatively minor qualifications.

Mixed

Contains materially accurate and inaccurate elements.

Misleading

Missing context, framing or an inappropriate comparison materially distorts the assertion.

False

The material assertion is contradicted by reliable evidence.

Unverified

Reliable evidence is insufficient to establish a conclusion.

Context matters

Checks identify the period, publication date and definition of each measure. We distinguish headcount from full-time equivalents, price levels from inflation, adjusted from unadjusted statistics, and forecasts from results.

Blame, credit and the full message

We read the surrounding words and images, including implied claims that a party caused a problem or deserves credit for an improvement. A correct number does not by itself prove that attribution. We label our interpretation, show the source context and explain what the evidence can establish. The same test applies to criticism by either party and to claims about its own achievements.

An attack is a reason to examine responsibility, not automatic proof of a falsehood or an unsupported motive. A recorded vote or directly implemented policy can support a specific claim. For wider outcomes, we look for evidence of the proposed causal link and consider other influences. Missing evidence does not prove a policy had no effect. Where accurate figures create a materially unsupported impression of blame or credit, the full message can be Misleading; where a conclusion cannot be established, we use Unverified. Future consequences remain predictions.

Each entry states its scope. “Full message assessed” includes the selected claim’s material factual implications, not every other assertion in the source. “Factual component checked” limits the rating to a documented action or assertion and explains broader implications separately. Promises, predictions and opinions receive context without an accuracy verdict.

Future promises, predictions and opinions

These entries explain what was proposed or argued. A future promise is a commitment whose delivery remains to be assessed; a prediction describes possible outcomes; an opinion expresses a judgement. We provide sources and context without assigning an accuracy verdict. Any testable factual assertions within these statements can be checked separately. A factual claim with insufficient evidence is labelled Unverified.

AI-assisted research and independence

The initial reports use AI-assisted research and calculations. Each report explains the sources read and the limits of its verification. External human editorial review has not been recorded for this initial batch. Sponsorship and newsletter distribution cannot change a verdict or replace the evidence requirements.

Corrections and incomplete information

Unresolved assertions remain Unverified or private drafts. A changed public conclusion retains the earlier rating, explanation and dated revision reason in its correction history. Readers can submit evidence, report an error or appeal a rating using the form below. Submissions require review before publication.

Checks describe the evidence available on their verification date. New statistics or a later policy announcement may change the answer; a later document does not establish what was true on an unknown earlier posting date.