Important limitations of GWA studies should also be kept in mind. One is their enormous potential for generating false-positive or spurious associations. Because they test hundreds of thousands of statistical hypotheses—one for each allele or genotype assessed—GWA studies have enormous potential for generating false-positive results due to chance alone. At the usual p < 0.05 level of significance, an association study of one million SNPs will show 50,000 SNPs to be “associated” with disease, almost all spuriously. One response to this problem is to reduce the false-positive rate by applying the Bonferroni correction, in which the conventional p-value is divided by the number of tests performed (45). A one-million-SNP survey would thus use a threshold of p < 0.05/106, or 5 × 10−8, to identify associations unlikely to have occurred by chance. This correction has been criticized as overly conservative, but it remains the most commonly used approach to date (50). Another cause of the false-positive associations to which GWA studies are prone is population stratification. Allele frequencies vary between population subgroups, such as those defined by ethnicity or geographic
