We haven’t seen much data so far on exactly how bad the spread of fake news is across the European Union. But some new research from the Reuters Institute for the Study of Journalism might help.

In a report entitled “Measuring the reach of ‘fake news’ and online disinformation in Europe” released Wednesday evening, authors Richard Fletcher, Alessio Cornia, Lucas Graves, and Rasmus Kleis Nielsen write that “with the partial exception of the United States (e.g. Allcott and Gentzkow 2017; Guess, Nyhan, and Reifler 2018; Nelson and Taneja 2018), we lack even the most basic information about the scale of the problem in almost every country.” (Nielsen is a member of the European Commission panel on fake news and online misinformation, and he’d previously called attention to the lack-of-EU-research issue in a number of tweets.)

The paper focuses specifically on France and Italy. The authors looked at “a starting sample of around 300 websites in each country that independent fact-checkers have identified as publishers of false news,” then narrowed them further.

The findings are largely reassuring, though preliminary: These sites seem to have tiny reaches compared to legitimate news sites, and audiences spend much more time with the legitimate sites than the fake news sites. In almost all cases, fake news sites didn’t generate nearly as many Facebook impressions as legitimate brands, though there were a couple exceptions. The authors stress that more research is needed.

None of the false news websites we considered had an average monthly reach of over 3.5 percent in 2017, with most reaching less than 1 percent of the online population in both France and Italy. By comparison, the most popular news websites in France (Le Figaro) and Italy (La Repubblica) had an average monthly reach of 22.3 percent and 50.9 percent, respectively;

The total time spent with false news websites each month is lower than the time spent with news websites. The most popular false news websites in France were viewed for around 10 million minutes per month, and for 7.5 million minutes in Italy. People spent an average of 178 million minutes per month with Le Monde, and 443 million minutes with La Repubblica — more than the combined time spent with all 20 false news sites in each sample;

Despite clear differences in terms of website access, the level of Facebook interaction (defined as the total number of comments, shares, and reactions) generated by a small number of false news outlets matched or exceeded that produced by the most popular news brands. In France, one false news outlet generated an average of over 11 million interactions per month — five times greater than more established news brands. However, in most cases, in both France and Italy, false news outlets do not generate as many interactions as established news brands.

The starting sample of 300 French sites came from Le Monde’s Décodex; the Italian sites were compiled using two Italian fact-checking sites, BUTAC and Bufale, and from an Italian hoax-busting site called Bufalopedia. From there, there was further narrowing: The team excluded satirical sites (which have their own category in Décodex) and also included, for comparison, Russia Today (RT) and Sputnik, the two Russian state–backed sites that have featured in European discussions of disinformation.

“Following this process, we were left with 38 false news websites in France and 21 in Italy,” they write, “allowing us to estimate average monthly reach and average monthly time spent for many of the most popular online disinformation sources in 2017. We present data here for the top 20 false news sites yielded by our search in each country.”

In France, the most popular fake news site was Santé+ Magazine, “an outlet that has been shown by Les Décodeurs to publish demonstrably false health information,” which reached around 3 percent of the French online population.”

Similarly, “in Italy, the most widely-used false news website in our sample — Retenews24 — reached 3.1% of the online Italian population.”

In both countries, the researchers found overlap between the use of fake and legitimate sites:

If we consider desktop use only (comScore is not able to provide figures for mobile overlap in France or Italy), we see that 45.4 percent of Santé+ Magazine users also used Le Figaro in October 2017, and 34% used Le Monde. This aligns with previous research showing, despite their size, audiences for niche outlets often overlap with the audiences for more popular mainstream brands (Webster and Ksiazek 2012)…

There is also evidence of sizeable audience overlap between false news sites and news sites in Italy. To take one example, in October 2017, 62.2 percent of Retenews24 users also visited the website of Il Corriere della Sera, and 52.3 percent used La Repubblica.

When it comes to Facebook interactions there may be less reason to be optimistic: “Particularly in France, some false news outlets generated more or as many [Facebook] interactions as news outlets.”

Overall, though, the authors see some preliminary good news: In Italy and France, as in the U.S., “our analysis of the available evidence suggests that false news has more limited reach than is sometimes assumed.” But remaining questions (for instance: “We have not considered the potentially ‘long tail’ of false news access. If there are many other sites that publish false news, and the degree of overlap between their audiences is low, it may be that their combined reach is greater than that implied by the low individual reach figures. This matters even more if false news sites are reaching people that news sites do not”) should be researched further.

The full report is here.