[MaysonL]

Here's a statement of the research: places where the Facebook usage was one standard deviation higher than median had a 50% higher than average rate of anti-refugee violence. Based on studying all 3,335 attacks over 2 years. See cstross's tweet: https://twitter.com/cstross/status/1031873629489963008

[tropo]

This is clearly a case where "correlation is not causation" must be considered.

Suppose that refugees love Facebook. If so, then places with refugees will have more Facebook usage. Anti-refugee violence can only occur where there are refugees. We thus would expect that Facebook usage would be higher in locations with more anti-refugee violence.

Another explanation is that old people are less likely to engage in violence and also less likely to use Facebook.