Every figure needs its source

Figures travel further than their sources. A number is quoted in a report, repeated in an article and pasted into a slide, and by the third retelling nobody can say where it came from, what exactly it measured or when. It still sounds authoritative. It can no longer be checked.

That is why every figure in our work says where it came from. It is the simplest of our rules, and the one we would least like to give up.

What a source line should carry

A good source line lets a reader find the same number we used. A name alone does not do that. It needs:

• who published the figure, such as a statistics agency, a central bank, a company or an exchange;
• which series, table or page, specific enough to find again;
• when it was published, and which version, if the figure has since been revised;
• what we did to it, if anything, such as adjusting for inflation or taking an average.

The last of these is the one most often left out. A figure we have changed is our figure, not the publisher’s, and the reader should know which is which. The principles that researchers, librarians and publishers agreed for citing data ask for the same: enough detail to identify, open and verify the exact data behind a claim.

The first source, not the nearest

We cite the organisation that produced a figure, not the article in which we happened to find it. Secondary sources are often where a search begins, and they sometimes add useful context, but they also add errors: a misread unit, a dropped decimal, last year’s number. When the only source we can find is a secondary one, we say so.

When sources disagree

Two reputable publishers can give different figures for what sounds like the same thing. Usually neither is wrong. They define it differently, cover a different set of companies or countries, or measure at a different time. We do not pick one silently. We say which we used and why, and where the difference matters, we show both.

Why it matters to a reader

A sourced figure can be checked, updated when it is revised and challenged when it looks wrong. An unsourced one can only be taken on trust, and trust is what research should earn rather than ask for.

If you find a figure in our writing that you cannot trace, please tell us. That is a fault in our work, and we will correct it.