The disguised gender experiment reminds me of similar ones into whether or not sugar makes children "hyper". It turns out that if you tell parents that their children have had sugar, they're much more likely to describe the kids' behaviour as unruly or hyperactive, even if the kids haven't been given anything at all (http://skeptoid.com/episodes/4142 - the Skeptoid article provides a nice summary and cites a few studies).

The fact is, the human brain isn't anywhere near the objective, rational computer and recorder it likes to think of itself as. We project our expectations on to everything we see, from the facial expressions of a monkey to random clumps of noise in datasets and it's important to keep that in mind whenever you read articles like this one. The information presented in a typical piece of science news has gone through the mental filters of researchers, university PR departments, editors, and a writer before you even start interpreting the first word. Even if every person at every step understands what they're doing and has the best intentions (and isn't, for example, deliberately twisting facts to fit their linkbait article for a pop science blog), that's a lot of room for a convenient narrative to obscure the basic, potentially important facts.