Table 3.

Comparison of Power to Detect Disease-Related Haplotypes through Use of Different Haplotype Inference Strategies under Various Disease Models and Disease Prevalences at Different Type I Error Rates

		Power
		Low Ambiguity			Medium Ambiguity			High Ambiguity
Modeland αa	Base	S1	S2	S3	S1	S2	S3	S1	S2	S3
1:
.10	55.6	55.2	56.2	55.4	55.2	56.8	51.4	57	58	54
.05	45	43.6	44	43.4	44.4	44	42.4	46.6	48.2	41
.01	29.2	29.8	29.6	30.4	29.4	29.4	25.6	30.4	28.4	24
2:
.10	85.2	82.8	84.2	82.6	80.8	82	78.6	78.8	81	77.4
.05	75	73.2	74	72.4	72.2	73.4	70.8	68.6	71.2	67.2
.01	55.4	53	53.4	52.8	52.8	54.6	52.4	49.8	51.2	46.6
3:
.10	71.8	68.2	68.8	67.4	67.4	68.4	64.4	64.2	65.2	60.2
.05	56.8	55	55.2	54.4	55.8	54.6	51.2	49	50	49.2
.01	32.6	31.4	32.2	29.8	30.2	28.6	25.6	26.8	26.4	26.2

^aFor the hypothetical case-control study, we considered three different models in our simulation experiment with the frequencies listed as θ_AB, θ_Ab, θ_aB, and θ_ab. These models are (1) case group: .4, .3, .2, .1; control group: .25, .25, .25, .25; (2) case group: .4, .1, .1, .4; control group: .25, .25, .25, .25; and (3) case group: .4, .1, .2, .3; control group: .3, .1, .4, .2. α = type I error rate.