This makes sense, especially as a number of people I know feel stupid because they have test anxiety. My wife included. "Non-"tests such as these seem like they could help quite a few, as they have the exact same physical location as real tests, but break the trend of habitual anxiety because little to nothing is on the line. Practice tests elsewhere don't share the classroom setting, and there's quite a bit of evidence that location influences memory / emotion, so it would seem they should be about as effective as they are at combating anxiety (ie: not much, and not for many (anecdotally)).
edit: <strikeout>Though, to be potentially inflammatory, this does seem to support the opposite stereotype of men having more control over (non-anger-based) emotions than women. Especially when you look at the original article (linked below by Locke1689: http://news.ycombinator.com/item?id=1943596), and see almost no difference in score distribution for men but huge changes for women in the B and C categories (A had almost no change - surprise, surprise, the ones without testing problems showed no gain).</strikeout>