I call bull-feathers. Linking genes to behavior is a science that's in its infancy, and the popular media are notorious for over-playing it. (Also, the for-profit gene-testing companies.) Case in point, reporters insist on calling the FOXP-2 gene "the language gene," even though scientists keep trying to explain that it is no such thing.
I'm guessing Joel Stein is talking about the SNAP-25 gene. His claim that their genetic marker is "three points higher" is just incoherent. But a study on SNAP-25 and performance IQ showed that having a particular allele of the gene makes a difference of . . . three IQ points. (Actually, 2.84 IQ points, averaging over 665 people.)
[Dramatic brake-squealing sound] THREE IQ POINTS? Hardly a marker "for" non-verbal IQ.
Reference:
Gosso, de Geus, Belzen, Polderman, Heutink, Boomsma, & Posthuma (2006). The SNAP-25 gene is associated with cognitive ability: evidence from a family-based study in two independent Dutch cohorts. Molecular Psychiatry, vol. 11, pp. 878-886.