This article would benefit from the publication of the full dataset, there are no clear evidence in support of his claims IMO. I'm not supporting either of both party, I'm just a little bit skeptical without the full dataset. The paper is almost useless and the article is behind a paywall.