Table 4.

Type I Error Rates for Global and Haplotype-Specific Score Statistics for Quantitative Traits

		Simulation P Values for
			Haplotype-SpecificScore Statisticsb
Frequency of Skipped Haplotypesa (No. Evaluated), for Trait Distribution	Nominal Type I Error Rate	Global Score Statistic	Median	Mean
Normal:
<.005 (n = 40)	.05	.063	.051	.052
	.01	.022	.011	.011
<.01 (n = 21)	.05	.048	.052	.053
	.01	.014	.011	.011
Exponential:
<.005 (n = 40)	.05	.084	.047	.047
	.01	.037	.018	.019
<.01 (n = 21)	.05	.078	.047	.048
	.01	.024	.017	.017
Exponential, log transformed:
<.005 (n = 40)	.05	.069	.051	.051
	.01	.023	.013	.015
<.01 (n = 21)	.05	.060	.050	.050
	.01	.018	.012	.012
t Distribution (5 df):
<.005 (n = 40)	.05	.077	.047	.047
	.01	.032	.019	.019
<.01 (n = 21)	.05	.064	.043	.044
	.01	.028	.015	.016

^a“Skipped haplotypes” are not evaluated but are pooled into a single baseline group when their frequencies are less than the threshold in column 1.

^bMedian and mean P values are over the number of evaluated haplotype-specific scores.