Table 2.

Data-dredging gene-dose analyses with different numbers of loci and different sample sizes

Number of the high-risk genotypesSample size (500)
Sample size (1,000)
OR*Coverage probability of the CIOR*Coverage probability of the CI
Three loci
    Number of high-risk genotypes
        0 to 11.00001.0000
        21.19990.89061.13610.8883
        31.39240.86781.26410.8627
    Type I error rate of the trend test0.16050.1596
Five loci
    Number of high-risk genotypes
        0 to 11.00001.0000
        2 to 31.28200.85771.19430.8509
        4 to 51.64850.72801.42110.7300
    Type I error rate of the trend test0.30370.2980
Ten loci
    Number of the high-risk genotypes
        0 to 31.00001.0000
        4 to 61.41520.73991.27640.7373
        7 to 102.00050.38921.63340.3808
    Type I error rate of the trend test0.63480.6436
  • NOTE: The results are based on simulated case-control data (equal number of cases and controls). The loci are unlinked or in linkage equilibrium with one another. The genotypes in each loci have been classified into the “1- genotype” and the “0- genotype.” The frequencies for the 1- genotype are: 0.4, 0.5, and 0.6 (three loci); 0.3, 0.4, 0.5, 0.6, 0.7 (five loci); 0.4 (×3), 0.5 (×4), 0.6 (×3) (10 loci), and are the same in the case group and the control group. Ten thousand simulations are done for each scenario. The α level is set at 0.05.

  • * Geometric mean of 10,000 simulations.