In this study, we used GWAS data from public repositories to generate common sets of SNPs and to estimate the magnitude of imputation-induced bias among European Americans and African Americans genotyped on different Illumina and Affymetrix arrays. Imputation based on the union of genotyped SNPs available on either the Illumina 1M or 550v3 array showed spurious associations for ~0.2 % of SNPs in both European Americans and African Americans, translating to ~2,000 false positives per one million imputed SNPs. SNPs in low LD regions were more prone to imputation-induced bias, as compared to SNPs in high LD. False positives remained problematic for even very similar arrays (i.e., Illumina 550v1 vs. 550v3), albeit to a lesser extent with 0.07 % of imputed SNPs having spurious association in each ethnic group. False positives were substantial for imputation across array families (Illumina and Affymetrix), amounting to 0.53 and 0.63 % of imputed SNPs (5,000–6,000 false positives per one million imputed SNPs) in European Americans and African Americans, respectively. These results are consistent with Sinnott and Kraft, who estimated an average false positive rate
