Thanks for the link, yes I do not know these kind of studies before, and I skim read it just now, however I also see claim of non reproducibility and different results from other studies in this page.
Personally I notice that force me to read difficult fonts does not help at all my understatement or my memory. In the end my idea, without statistical proof of course, is that such contrasting observation are simply noise. Perhaps due to a sample size not big enough, perhaps due to tested individual itself but still noise.
Also IMVHO such studies exist for a reason: teacher's mean quality is getting lower and lower due to various schools reforms and since admit errors it's not a thing these days someone try a sort of "montecarlo test" in the desperate search of a solution to stop the bleeding. Other find such tentative nice for their business because new stuff means new sales for someone.