METHOD
How we examine a claim
We begin with the precise claim being made. We trace it to its sources, examine the evidence, identify its limits and consider how language or presentation may shape interpretation. We then separate what is established from what remains uncertain or unknown.
Our purpose is not to tell readers what to believe, but to make clear what the evidence does — and does not — establish.
The five stages
01
The Assertion
We state the exact claim in neutral terms. We distinguish a claim about what is true from a recommendation about what someone should do.
02
The Evidence
We trace the claim back to primary sources wherever possible. We ask what was measured, against what comparison, in whom, for how long, and who funded the work.
03
The Limits
We examine study design, missing data, uncertainty and applicability. Relative figures are translated into absolute numbers when the data allow.
04
Examine the framing
We examine wording, images, defaults, omitted context and presentation choices that may influence interpretation. We describe the effect of the framing without assuming why it was used.
05
What Remains Unknown
We distinguish established fact from inference, modelling and speculation. Questions the evidence cannot answer are left open, not filled with certainty.
Our standards
Primary sources
Wherever possible, we link directly to original studies, data, official documents and archived material.
Fair representation
We present the strongest reasonable version of a claim and include material evidence that points the other way.
Right of reply
Where we examine a claim made by an identifiable person or organisation, we may invite them to provide evidence, clarification or correction. Material responses are added to the record.
Numbers in context
We report absolute numbers alongside relative claims and make denominators, timescales and comparisons clear.
Evidence, not motives
We examine claims and methods. Where incentives or conflicts are relevant, we document them as context without inferring motive or intent.
Clear labels
We distinguish observed fact, inference, modelling, opinion and speculation.
Visible corrections
Material errors and revisions are corrected openly, with the date and nature of the change recorded.