The Korean study was done on a large population sample and the 12% were not for AS but for ASD in general.

Article is here:
http://www.ncbi.nlm.nih.gov/pubmed/21558103

But you are right, the 12% is based on superior IQ (>120, top 8%) numbers in kids *who were in general education*, representing about 2/3 of diagnosed cases (although it is a bit more complicated than that). So 12% makes sense. Kind of (the data remains really weird).