Table 1. .

Design of Replication Studies Based on the Estimated Penetrance Parameters^[Note]

	Replication Test
	Allele Based		Best Model
Sample Size and Genetic Model					Ratio
100/100:
Multiplicative	1.09	.37	1.00	.34	.87
Three parameter	1.10	.39	.96	.24	.59
Recessive	1.03	.39	.87	.28	.42
200/200:
Multiplicative	1.07	.40	.99	.38	.88
Three parameter	1.06	.40	.97	.22	.55
Recessive	1.07	.38	.91	.25	.41
400/400:
Multiplicative	1.13	.41	1.04	.38	.87
Three parameter	1.13	.41	.98	.20	.44
Recessive	1.03	.38	.95	.22	.41
800/800:
Multiplicative	1.12	.44	1.03	.42	.87
Three parameter	1.10	.41	.98	.17	.38
Recessive	1.11	.39	.99	.21	.37

Note.— Designing samples of the indicated sizes (number of cases/number of controls) were simulated assuming a multiplicative model, a three-parameter model, or a recessive model. For each replicate simulation, the sample size to obtain 80% power in a replication study was estimated. Here, is compared with the true required sample size s for two test strategies: with use of the allele-based test or with choice of the best of several test statistics. Columns 2 and 4 show the median ratio, and columns 3 and 5 display the median relative difference. The final column shows the ratio between the estimated sample sizes for the first and the second strategy.