How We Reach Conclusions

1. What "Scientific" Means for a Policy Project

This project doesn't run laboratory experiments. It can't split Britain into a treatment group and a control group. When we call the approach scientific, we mean something narrower: applying the habits that make science work to how claims are formed, checked and revised, even though tax, health, housing and trade can't be tested the way a hypothesis in physics can.

Those habits come down to a few specific practices, and it is worth naming them rather than using "evidence-based" as a slogan.

Evidence before conclusion, not the other way round. The usual failure in political argument is to start with a preferred conclusion and pick the facts that support it. This project tries to work the other way: establish what the evidence shows, then let the conclusion follow, including when that conclusion is inconvenient for the argument being made.

Claims are provisional, not settled. A figure or argument on this site is the best current account given what has been checked so far, not a permanent verdict. It can be revised or withdrawn if better evidence appears. That isn't a weakness in the method. It is the method.

Keep observation separate from inference. A fact that can be checked against an official source, such as an ONS statistic or an OBR projection, is a different kind of claim from a conclusion this project reaches by connecting two or more such facts. Both can be right, but they carry different weight, and blurring them is a common way for arguments to quietly overstate their certainty.

2. The Practices, and What Each Looks Like on the Page

Factual claims are sourced and can be checked

Figures cited in the pillars link to a specific row in the Sources Register, giving the institution, the publication and what the figure is used for. This isn't a bibliography for its own sake. It means any reader, or any future editor, can trace a number to where it came from and check it, rather than trusting it because it is stated confidently.

The standard isn't always met, and when it isn't, the fix is recorded. In October 2026 the site's pillars and working papers were re-checked line by line. That review found and corrected dozens of figures that were out of date, wrongly attributed or without any source, from population density comparisons to the OBR's state pension projection. Each correction is logged in the Update Register with the reason for it.

The strongest version of an opposing argument is presented, not the weakest

Most political writing argues against a caricature of the other side, because caricatures are easy to beat. Each pillar on this site has a Steel Man section, because beating a straw man proves nothing except that the writer can build one.

In practice this standard means going back and fixing things when we fall short. The Trade pillar originally described the Remain campaign's position on world trade in a way that, on closer research, no mainstream figure had actually argued. The real version of the argument (that short-term economic forecasts proved too severe, not that trade with the rest of the world would stop) was weaker evidence for our side than the straw man had been, and we published the correction anyway.

Original analysis is labelled as analysis, not presented as an outside finding

When this project connects two separately sourced facts into an argument neither source made, the text says so; the connection isn't attributed to the sources as if they had made it. The Local Government pillar's section on ageing and the council tax base is an example. The underlying facts come from ONS and other published sources, and the argument linking them (that the areas facing the fastest-growing social care costs tend to be the areas with the weakest-growing tax base) is labelled in the text as this project's own analysis.

Claims are dropped or downgraded when they fail a fresh check, even after they were written

A commitment to evidence is only real if it sometimes costs you something you have already written. Two examples from August 2026. Council tax figures for Scotland were dropped from the Local Government pillar after a fresh check found them still contradictory even in supposedly reliable secondary sources; we didn't publish a number we couldn't resolve. And a detailed economic case for agrivoltaics (solar panels over crops) had grown, over several editing sessions, into one of the Agriculture pillar's core arguments, driving several proposals. On review, the technology was still at trial stage and the case didn't meet the standard the rest of the pillar was held to. It was cut back to a single paragraph, and the full case moved to a separate working paper held to a lighter standard and clearly marked as such.

Every major proposal is built through a five-stage chain, and carries a test that could prove it wrong

The practices above cover how individual claims are checked. This one covers how a claim becomes a policy proposal, the step where evidence-based analysis most often slides into ideology, because a real fact at the top of an argument doesn't on its own justify a particular policy at the bottom.

Major proposals on this site are built through five stages, shown on the page rather than kept as drafting notes:

1. Evidence: what the data shows, tagged for how solid it is (FACT, SYNTHESIS, CONTESTED or UNVERIFIED).
2. Causation: what mechanism is being claimed, and how strong the case for cause and effect is, as opposed to mere correlation.
3. Options: the range of responses the causation would justify, not only the preferred one.
4. Values: what decides between those options, stated openly rather than slipped in as if it were also evidence.
5. Proposal + Test: the specific reform, with a falsification test attached.

The falsification test is the key addition. It sets out what should happen if the proposal is right, how large the effect should be, over what time, against what counterfactual, and, most importantly, what result would make this project drop or revise the proposal. A policy argument that can't say what would prove it wrong is advocacy dressed in the language of evidence.

This approach was introduced in August 2026, and applying it to every existing pillar is a large job that isn't worth halting other work for. It was piloted on Economic Renewal's main proposal (see the chain and falsification test for the 95% inheritance tax) and every pillar that makes proposals now has one. Several of these tests bear on the same main proposal from different angles (the rate itself, the anti-avoidance design, the choice of instrument); Economic Renewal §10, "The Kill Sheet", brings them together in one place so readers don't have to find each one separately.

3. Where the Method Has Limits

Not every question this project could cover is one evidence can answer on its own, and pretending otherwise would be its own kind of dishonesty.

Some public debates are mainly about values, not facts: reasonable people who know the same evidence still reach different answers because they weigh competing goods differently. Assisted dying is a clear example, and this project deliberately doesn't cover it for that reason. The factual questions (how well safeguards work, what happened in other countries) are real and checkable, but the core disagreement is between the value of personal autonomy and the value of protecting vulnerable people, and no amount of extra evidence settles that. Forcing an "evidence-based" verdict onto a question that isn't mainly about evidence would misrepresent what evidence can do.

Even the topics this project does cover aren't value-free just because they are evidence-based. Concluding that a trade-off is worth making still rests on a judgement about whose costs and benefits should count for more. The scientific discipline here is about getting the facts right and reasoning openly from them to a conclusion, not about removing judgement altogether. Readers who weigh the same evidence differently aren't wrong for that reason; they may simply hold different values, which this project tries to state openly rather than present as facts.

4. What This Project Isn't

It isn't primary research. No claim on this site comes from an original study, survey or experiment run by this project. Every factual claim traces back to published data, government statistics or academic research. The project's own contribution is synthesis, argument and policy design built on that evidence, not new evidence. Where the synthesis goes beyond what any single source says, Section 2 describes how that is flagged.

It isn't a neutral summary either. A project that set out "both sides" without ever reaching a conclusion would be avoiding the actual work of policy analysis, not practising objectivity. Non-partisan means not aligned with any party and not choosing evidence to flatter one. It doesn't mean declining to reach a view.

And it isn't the only project doing this kind of work. Comparable Projects sets out the closest UK equivalents, including where they disagree with the conclusions reached here.

5. How to Hold Us to This

The claims in this document can be checked against the same public record as every pillar on the site.

The Sources Register shows what each factual claim across the project traces back to.

The Gaps Register is where the project publicly records the limitations in its own work, rather than leaving them unmentioned. It currently lists 35 gaps, 6 of them fully resolved, ranging from figures that need better primary sources to questions a pillar hasn't yet answered.

The Update Register records what changed in any published document, and why, including the corrections described in Section 2.

If you think a claim on this site hasn't met the standard described here (a source that doesn't check out, a straw man that survived, or the project's own argument presented as an outside finding), set out the claim and the evidence, and it will be checked against the same standard used everywhere else.