Polling & Data

This desk reads the numbers behind British political debate — opinion polls, survey fieldwork and official statistics. The emphasis is on what the data can and cannot show: how questions were asked, how representative a sample is, and where headline figures need qualifying. Coverage draws on named, published sources so that any figure can be traced back to its origin rather than taken on trust.

Numbers carry authority in political debate, and the desk’s first concern is whether that authority is earned. Each entry asks how a poll was conducted: the size and make-up of the sample, the wording of the question, the fieldwork dates, and whether the figures were weighted to resemble the population. The guide to reading an opinion poll sets out those tests in full and is the reference point for the rest of the desk.

A single headline figure rarely tells the whole story. Movement between two polls can fall within the margin of error; a result can shift with the order in which questions are asked; and the same data can support more than one honest summary. Entries draw those qualifications out rather than smoothing them over, and prefer a trend across several surveys to any one striking number.

Every figure cited is traced to a named, published source — the polling company, the survey series or the official statistics behind it — so a reader can follow it back to its origin. Where the desk covers a contested topic, such as the generational gap in support for EU membership, the point is to show what the evidence does and does not establish.

Beyond opinion polls, the desk draws on official statistics — the work of bodies such as the Office for National Statistics, the House of Commons Library and the Electoral Commission — which carry their own methods and caveats. Administrative data and survey data answer different questions, and an honest reading keeps the two apart: a turnout figure, a registration count and a poll of voting intention are not interchangeable, and the desk is explicit about which kind of number a given claim rests on.

The desk also tracks where published polling and the eventual result diverge, since the gap between them is itself instructive. A poll measures stated intention at a moment in time, not a forecast, and the reasons a survey can miss — late swings, differential turnout, the difficulty of reaching certain voters — are part of what the desk explains. Treating a poll as evidence rather than prophecy is the habit it tries to encourage.