Okay, let me backtrack. When I wrote that, I'd read popularizations not research papers. I knew there was something wrong with the popularizations. Now, looking over the research, I still think there's something missing.

Best case scenario for Baron-Cohen: he's found correlations between all the relevant traits on his "autism spectrum" as well as autistic traits that I haven't seen mentioned in his work. And there are no major traits common to diagnosed autistics that don't fall onto this spectrum for the general population.

About gender: I wasn't thinking about PC, I was really thinking about it not making sense. What I know: there are more male than female diagnosed autistics. Men perform consistently better than women on spatial reasoning tests. Men are, of course, more common than women in technical professions. What's in question is the additional claim that these phenomena are all part of the same thing, a spectrum from empathizing to systematizing types of brains. That's an additional claim, and a bold one.

Keep in mind that it's not enough to claim that autistics tend to be more systematizing and non-autistics tend to be less systematizing. (He does have evidence to show this.) To make the kinds of claims he does in the media, he'd have to show that this is the main difference, that the systematizing/empathizing axis explains most of the variation between autistics and non-autistics.

Now I have looked at his website and papers and the papers and summaries I glanced at don't seem to indicate that he's done the work of correlating and comparing the different traits labeled as "empathizing" and "systematizing" to see if his scale is a valid concept. His main justification for using it is that the "systematizing" cluster is a list of traits found to be more common in males than females. But he doesn't cite high correlations between the systematizing or the empathizing traits. And, while systematizing and empathizing are inversely correlated, the correlation is weak (r = 0.16.)
(http://www.autismresearchcentre.com/research/project.asp?id=2)

I don't know if this is standard practice for psychologists but, at least with the survey-based studies, I think the papers confirm that he doesn't realize how much more he'd need to do to confirm his claims.