I used that line myself last week, at the end of a piece about what schools teach on media. I used it without opening the document, because a ranking is convenient: it arrives already shaped as a sentence.

Then I opened it. Page 5 says plainly that the index does not measure media literacy but its predictors. That is not a technical footnote. It is the difference between knowing how much a country teaches people to read the media, and estimating how well it might hold up if someone tried to deceive it.

What gets said

Rankings travel under their title, not their methodology. This one is called a media literacy index and carries a subtitle about measuring societies’ vulnerability to disinformation. Two large promises on the cover, and those are what reach the reader.

The accompanying release says Denmark, Finland, Ireland and the Netherlands lead the European ranking. From there the number detaches from the document and walks on its own. It becomes proof that one country teaches well and another teaches badly, which is precisely the thing the index states it does not measure.

In Italy, as far as I could find, almost nobody picked up the 2026 edition. But the issue is not how often a figure gets quoted. It is how it gets quoted when it does.

What the data says

Four countries share the top with 71 points: Denmark, Finland, Ireland and the Netherlands, separated into first, second, third and fourth by fractions of a point. Finland had held first place since 2017 and slips to second. Norway is fifth with 70, Estonia sixth with 69. At the bottom sits Kosovo, forty first with 16 points, with North Macedonia and Albania on 19 and Georgia on 20.

Italy is twenty third with 51 points, last in the second of five clusters. In 2023 it was twenty fourth with 47, in the third cluster. One place up, four points up, one cluster up, and the report flags it as one of the few promotions of this edition.

Estonia is worth a look too, since it is the country Europe points to whenever schools and media come up. Sixth with 69 points. In 2023 it was fourth with 71. Two places and two points lost, still in the leading group.

And it is worth going back one edition further, which is the step usually skipped. In the 2022 edition Italy was twenty third with 50 points. Twenty third, twenty fourth, twenty third. In four years Italy’s position has not moved.

What the data does not say

The index is the weighted sum of eight numbers borrowed from other people. The weights are on page 5. Media freedom, 40 per cent: half from Freedom House, half from the Reporters Without Borders index. Education, 45 per cent: PISA reading 30, PISA science 5, PISA maths 5, tertiary enrolment 5. Trust in others, 10 per cent, from the World Values Survey. E-participation, 5 per cent, from a United Nations index.

Look for the indicator measuring how much a country teaches people to spot a false story, and there is none. Not one, not even at one per cent. Not because anyone hid it, but because that number does not exist in Europe, and the index does what it can with what there is.

Then there are three things that only appear when you set the two editions side by side.

One: where Italy’s four points came from. Recomputed from the sub scores, Italy moves from 47.3 to 50.5. Of those 3.2 points, 2.4 come from a single indicator, PISA reading, which alone carries 30 per cent of the weight. Three quarters of the improvement sits there. Italy’s raw reading score went from 476 to 482, six points. Over the same interval thirty five of the forty one countries fell, by fourteen points on average. Only four rose: Montenegro, the Netherlands, Italy and Serbia. Italy did improve, slightly. What is large is not our climb, it is everyone else’s fall. In a standardised ranking the two produce the same result, and they are not the same thing.

Two: a fifth of the index is frozen at 2017. The Freedom House column is identical for all forty one countries between the 2023 and the 2026 edition. Not two or three of them: forty one out of forty one. The reason is that the cited source is Freedom House’s Freedom of the Press, a report published from 1980 to 2017 and never issued again after the April 2017 edition. The figures match one for one: Italy 31, Norway 8, Poland 34, Greece 44, Turkey 76. Twenty per cent of a 2026 index describes press freedom as it stood a decade ago.

Three: another ten per cent does not move either. Interpersonal trust is identical in thirty nine countries out of forty one, because the World Values Survey runs in waves years apart. Added up, close to a third of the 2026 score is by construction the same as in 2023.

There is one small detail that teaches a lot. On tertiary enrolment the highest score in the whole ranking, one hundred out of one hundred, belongs to Greece, which sits thirtieth overall. The Greek raw figure is 167 per cent, because it is a gross ratio: it counts every enrolled student against the theoretical age band, and breaks through one hundred when there are late finishers and mature students. Luxembourg scores zero on 21 per cent, because its young people go to university abroad. Neither number means what it appears to mean.

The part that matters

None of this is a trick. The report states on page 5 that it measures predictors. It puts the weights in a table. It publishes the raw data on page 18, column by column, so anyone can redo the arithmetic, which is exactly what I did. And on page 6 it recalls George Box’s aphorism: all models are wrong, some are useful.

A document that declares its own limits is doing its job, and doing it well. Anyone quoting it without opening it is doing a different job, and we all do it, myself first. That is not ignorance, it is economics: reading costs half an hour, quoting costs ten seconds.

So the proposal is small and practical. Three questions before using a number lifted from a ranking.

First: what does it actually measure, as opposed to what it is called. That lives on the methodology page, which is almost always the fifth.

Second: when was the data collected, not when was the report released. Those are two different dates and they can be ten years apart.

Third, the one almost nobody asks: if there is an earlier edition, what moved in between. Whether the country moved, or the source did.

And the question this index cannot answer remains. We know where Italy stands on press freedom, PISA and trust in others. We do not know how much an Italian teenager learns, across thirteen years of schooling, about checking whether a story is true. Nor does the country Europe treats as the most advanced of all: Estonia’s education ministry, in its national report of December 2025, writes that impact evaluation is hampered by the absence of transparent metrics.

If the model country admits it cannot measure its own model, then nobody has that number. Asking for it is more useful than quoting this one.

Twenty third out of forty one is a fine line. It does not mean what it looks like, and that is not the fault of whoever wrote it.

Sources

As listed in the Italian version above.