Table 13.

Empirical Type I error rates (for β = 0) and power (for β ≠ 0) for the multiple traits with a CS(0.3) correlation matrix; for β ≠ 0, the first five traits were associated with a causal SNP with log-ORs β_j ~ U(0.8, 1) or ~ U(0.2, 0.3), while all others had β_j = 0. An independence working correlation structure was used in GEE.

	β = 0			βj ~ U(0.2, 0.3)			βj ~ U(0.1, 1)
#traits	Score	aSPU	aSPU.Sco	Score	aSPU	aSPU.Sco	Score	aSPU	aSPU.Sco
5	0.051	0.050	0.051	0.468	0.632	0.612	0.682	0.865	0.832
10	0.060	0.051	0.059	0.573	0.456	0.537	0.684	0.795	0.777
20	0.060	0.047	0.048	0.535	0.305	0.463	0.611	0.757	0.741
30	0.032	0.049	0.042	0.456	0.257	0.394	0.528	0.703	0.693
40	0.054	0.054	0.057	0.409	0.211	0.321	0.424	0.639	0.628