Microsatellites and epigenetics could be other confounding factors:

http://www.johnhawks.net/weblog/ ...behavior_2005.w

“Elizabeth Hammock and Larry Young (2005) report in Science that variation in behavioral traits among different species of voles is partly controlled by variation in microsatellite loci linked to neuroactive genes.”

People are using genomic arrays and looking for double signals. I know of this being done in Drosophila and mouse, but I haven't heard of it being used in humans.

RPM-

people have been using SNP arrays to look for copy-number variation in humans, but the SNP density in the regions with that sort of variation is much lower, which complicates things.

so far, there are no tiling arrays available for the entire human genome. and going from a tiling array to the detection of polymorphism isn't straightforward: kruglyak has done it in yeast, but diploids are harder.

but that sort of approach is certain to yield results in the near future.