Table 1.

Comparison of p Value Thresholds from Four Multiple Testing Correction Procedures in WHI SHARe African American (AA) and Hispanic American (HA) Samples

	Bonferroni (# Tests)	Simulation (10,000 reps)		Analytic Approximation
		Traits	Test Statistics
AA	$9.1\times {10}^{-8}$	$2.1\times {10}^{-5}$	$2.1\times {10}^{-5}$	$2.0\times {10}^{-5}$
(95% CI)		$\left(1.9,2.3\right)\times {10}^{-5}$	$\left(1.9,2.2\right)\times {10}^{-5}$
HA	$3.1\times {10}^{-8}$	$4.3\times {10}^{-6}$	$4.5\times {10}^{-6}$	n/a
(95% CI)		$\left(4.0,4.6\right)\times {10}^{-6}$	$\left(3.9,4.9\right)\times {10}^{-6}$

For simulation-based approaches, we also provide a 95% bootstrap confidence interval. Both simulation-based approaches used 10,000 replications. The nominal genome-wide type I error rate (${\alpha }^{\ast }$) is 0.05.