The article does indeed suggest that the training difference is null, and I think the article is misleading, and based on bad research.