I can't find a link to the original paper but it would be good to check the details of the analysis. Firstly, they'd have to take care doing comparisons between the 4 different groups - it's not enough just to do multiple-comparisons as the probability of a false-positive result escalates with each comparison (I would expect them to have done this). Also, the sample sizes seem low and it would be good to see how they estimated the variances (say by using boot-strapping) and shown that this is indeed not just a result by chance.
Look at the source of this article.
This should not even be on HN.
This is just spam.
Why do you say that it's spam? It looks like a press release from the institution that conducted the study. Such articles are not always great, but they tend to be a bit better than the science-farm sites that crib from them. The standard for information like this on the web is unfortunately rather low.