Table 3.

Statistical properties of different methods as applied to the analysis of data generated according to the model (15) but with the polygenic variables simulated according to the mixture distribution (16)

n	Model	Power1	Power2	Power3	FP	FDR
200	fixed	0.807	0.886	0.162	6.365	0.762
		[3.49,0.16]	[3.92,0.27]	[3.26,0.11]
	fixed	0.782	0.72	0.248	4.778	0.718
	PC = 3	[2.8,0.14]	[3.74,0.25]	[3.04,0.14]
	fixed	0.775	0.43	0.404	1.8	0.497
	PC = 10	[2.2,0.11]	[2.73,0.19]	[2.57,0.18]
	fixed	0.472	0.257	0.291	0.044	0.03
		[1.07,0.11]	[2.14,0.19]	[1.68,0.19]
	mixed	0.416	0.242	0.029	0.028	0.026
	μ = 0	[1.12,0.17]	[2.21,0.29]	[0.93,0.26]
	mixed	0.419	0.199	0.047	0.034	0.037
	PC = 3	[1.1,0.16]	[2.04,0.3]	[1.23,0.28]
	mixed	0.453	0.134	0.118	0.047	0.044
	PC = 10	[1.17,0.14]	[1.7,0.29]	[1.32,0.29]
400	fixed	0.99	0.985	0.507	11.802	0.814
		[2.34,0.11]	[2.45,0.18]	[1.82,0.13]
	fixed	0.984	0.912	0.565	8.601	0.761
	PC = 3	[2,0.11]	[2.28,0.17]	[1.93,0.13]
	fixed	0.985	0.733	0.776	3.684	0.566
	PC = 10	[1.17,0.1]	[1.55,0.11]	[1.44,0.11]
	mixed	0.89	0.605	0.666	0.095	0.029
	$\mu \ne$ 0	[0.52,0.07]	[1.03,0.09]	[0.96,0.09]
	mixed	0.83	0.557	0.201	0.061	0.023
	μ = 0	[0.54,0.08]	[0.98,0.16]	[0.65,0.11]
	mixed	0.847	0.504	0.288	0.084	0.03
	PC = 3	[0.54,0.08]	[0.86,0.15]	[0.72,0.11]
	mixed	0.885	0.418	0.467	0.1	0.034
	PC = 10	[0.61,0.07]	[0.88,0.13]	[0.83,0.12]

We report Power, average number of false positives (FP) and false discovery rate (FDR). The square brackets report a mean distance to a simulated QTL and a mean value of $|{\widehat{\beta }}_i-{\beta }_i|$. A fixed effects model and a mixed model with μ = 0 are additionally supplemented with the first 3 or 10 PCs of the incidence matrix X.