A Dan "Crockus" Hodgins update

Mark Liberman at Language Log has finally posted a bit of a follow-up on Dan Hodgins, the inventor of the crockus. One of Liberman's correspondents sent him a scan of one of Hodgins' presentation slides; it is presented in all its glory at right. The follow-up post is mostly about how very wrong the information on this slide is (and it's even about a real brain part this time!), but perhaps the best part of the post is Liberman's summary judgment of Hodgins: Hodgins himself is a pathetic figure, worth debunking only as an example of the charlatanry that apparently can flourish these days in the demi-monde of pseudoscience that the education industry calls "professional development". Amen to that. One can only hope that Hodgins is seriously questioned by someone with just enough incredulity at his next presentation.

Liberman and other bloggers have taken especial aim at Hodgins and his ilk (e.g., The Gurian Institute) for their overemphasis on small sex differences, which, when they exist at all, are usually differences in the mean value of some measure (e.g., corpus callosum size, variously measured) that are incorrectly extrapolated into categorical differences. This is a hairy statistical issue, because the difference between the means of two populations of scores is one of the first places that researchers look to decide if the two populations are different. Whereas different means do suggest different populations, the amount of overlap between even clearly different populations is sometimes enormous, and taking the difference between the means as more important than the overlap, and even as an indication that the populations should be treated differently, isn't always a reasonable leap. What's more important about the two distributions at right, that the means (the blue and purplish vertical lines) are different, or that they overlap (the gray area) a lot?

Bookmark with Del.icio.us!