A myth is a finding that travelled

The situation

Claims about learning reach a school system with a number attached. Two sigma. Ten per cent of what they read. They arrive in procurement meetings, training calendars and business cases, and they are expensive. A confirmation bias costs an argument. A learning style costs a procurement cycle, a year of professional development and the lesson time the alternative would have used.

The people repeating them are not careless. Around nine in ten teachers surveyed across eighteen countries hold the learning-styles claim. Plenty of people in edtech have quoted an effect size they never traced. The claims last because they are held by people who care, and that makes them hard to move.

I have put a number in a business case because everyone knew it. Nobody asked where it came from, least of all me.

The question

What if the field’s problem is not that it believes false things? Lay the catalogue out and only six of 103 claims are inventions with nothing behind them. The rest began as a real finding, correctly reported, then inflated, generalised and cut loose from the conditions that produced it. The original is usually defensible. The derivative is not.

That changes what a correction is for. Debunking is a weak instrument. A bare denial leaves the myth’s explanation in place, and the mind keeps using it to fill the gap. Meanwhile the research machinery keeps producing more. Small trials report roughly twice the effect of large ones. Tests built by the researchers report roughly twice the effect of standardised ones. Replications are 0.13 per cent of the published literature. A field with those habits will produce the next myth while you are correcting this one.

The move

Measure the distance. For any claim, find what the source actually said, then ask how far the claim has moved from it: which learners, what dose, for how long, measured how. The same four questions work on every entry, which makes the skill teachable.

Ask six questions before anything is bought or mandated. Compared to what? Measured how? How big, with what interval? Who paid, and who else has replicated it? At what scale, and what discount applies to ours? What does it cost per student, and what else could that money buy? None is new to a methodologist. Asked routinely at the point of adoption, they would have kept most of the catalogue out of classrooms.

Never take a claim away without naming what replaces it. Match instruction to a style becomes match it to the content and to prior knowledge. Flip the classroom becomes protect contact time and add quizzing. Each replacement is narrower than the claim it displaces, which is what makes it usable: the conditions say when it applies. A correction that only subtracts leaves a vacuum, and in education a vacuum is filled by whatever is marketed next.

Correct by name, and more than once. Teaching a myth beside the evidence against it works better than quietly teaching the right content, and no single-dose correction has yet been shown to last.

Hold your own claims to the same standard. A guide about overreach that gave every entry the same confidence would be committing the fault it describes, so each entry carries a rating of its own: high, medium, contested or low.

What it produced

The first was a fifth field guide, and the only one of the five whose subject is education itself. It holds 103 claims, 49 tells in the evidence machinery and 13 reasons none of it gets corrected. Each claim is tagged with what kind of wrong it is (fabricated, refuted, overstated, misapplied, unfalsifiable, contested or premature), with what the source actually said and the version that can be defended.

The second was a one-page instrument for reading an efficacy claim. Each tell ends with the question that catches it, following a number from the design of the trial to the meeting where it is quoted. Extracted, the questions fit on a page, and a page is what gets used.

The third was a short list of what survives the same standard. Retrieval and spaced practice. Worked examples for novices. Metacognitive strategies taught inside a subject. Systematic phonics for early reading. Structured tutoring and in-class coaching, which work and cost, and coaching loses about half its effect at scale. The list is short. That is the finding.

The fourth was discomfort. This is the one guide in the set whose readers are implicated, and I am one of its readers.