The fundamental goal of a case-control association study is to test for an allelic frequency difference between cases and controls to find SNPs that affect disease susceptibility. Because GWAS typically involve large sample sizes to detect small effects and hundreds of thousands of polymorphisms are studied, even small artifactual differences in allelic frequency between cases and controls can generate false-positive results. Therefore, it is particularly important to avoid associations between case-control status and experimental factors that have potential effects on allelic frequency. Well-recognized artifacts occur when cases and controls differ in population structure [Cardon and Palmer 2003] or when case and control DNA samples are handled differently in ways that affect DNA quality [Clayton, et al. 2005]. Differences in DNA quality can result in differences in the frequency of missing genotype calls, which are often biased towards one genotype or another [Wellcome Trust Case Control Consortium 2007]. When studies contain related subjects, methods that implicitly assume independence of subjects may have inflated false-positive rates. False negative results may be increased by failure to control various experimental factors, leading to `noise’
