This study seems poorly done. For example, I looked in the Orlando area, since it's a purple area: while they are different URLs, they are all the same site. The articles all seem to be bot generated content, with no clearly partisan content. It just seems like a click farm/url camping scheme, but that counts as 13 of the data points. But they are considered right leaning, because I assume because the owner is right leaning, but it doesn't really show a hyper-partisan website (or 13, in this case, in a purple area). I can't say anything about the rest of the data, but those data points alone seem to point to some issues with this article and dataset.