The New York Times published an article Tuesday that reports that in Germany, “towns where Facebook use was higher than average” had “more attacks on refugees.” But according to its own source material, that is not the case.
In what appears as a classic case of confirmation bias, the article covers a study  — which, by the way, was not peer-reviewed — and takes away that Facebook is driving anti-refugee violence.
this story has pulled in tens of thousands of shares and likes, huge amounts of valuable human attention. but the study doesn't show what the viral nugget is claim, nowhere near. Afaict the NYT is wrong. If so it should publish a correction pic.twitter.com/IhCjPIq7i2 
— hal (@halhod) August 22, 2018 
But, as first reported by The Economist’s Hal Hodson , “the study’s correlations don’t come from measuring general ‘Facebook use’. The correlations come from measuring posts made on the Alternative for Germany Facebook group.
The figures in the paper cited are not representative of Facebook as a whole. Rather, it represents usage within Alternative for Germany. The Times article does not mention Alternative for Germany. Instead, it talks about “Facebook usage” on the whole, not that within a group — and correlates it with violence.