[knowtheory]

There's a PLOS One paper which reanalyzed Ian Walker's 2007 data and claims that the effect Walker demonstrates is illusory if you chose different classification criteria for how to group data points: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjourna...

I tweeted w/ Ian Walker a bit about it, and though he hasn't written up a reply he disputes their conclusions (just for full disclosure).