So this report essentially says...
There's a correlation between absolute number of racist Tweets and absolute number of reported hate crimes. I wonder if there's also a correlation with population size? You know, places that have more people tend to generate more Twitter traffic than places with fewer people, and places with more people tend to have more crime reported. Surprisingly today we learn that there are 8,000 more theft crimes per year on the London tube network than in cities that don't have a London Underground.
I think we take it as read that scientists do take account of these confounding factors, but you can't make these assumptions. You have to state if the measures you are using are per capita or not; it's only a few extra characters.
Actually, the paper's contents are much more interesting than the summary reported. Not 100% convinced by their methods, as their algorithm was very much tuned towards "white-originating" racist language than racism in general, and they did look at the racial demographics of cities. Most interesting was the pinpointing of particular "mouth-piece" racists.