Table 1.

Proportion of associations reported as statistically significant ($\mathrm{P}<0.0001$) by logistic regression using a likelihood ratio test.

	Naive	PCA	MDS	SUGIBS
${\mathit{F}}_{\mathit{s}\mathit{t}}=0.001$
Random	0.0002	0.0001	0.0001	0.0001
Differentiated	0.9960	0.4483	0.6370	0.5200
Causal	0.5295	0.4779	0.4865	0.4807
${\mathit{F}}_{\mathit{s}\mathit{t}}=0.005$
Random	0.0009	0.0001	0.0001	0.0001
Differentiated	0.9980	0.0002	0.0003	0.0002
Causal	0.5226	0.4249	0.4255	0.4253
${\mathit{F}}_{\mathit{s}\mathit{t}}=0.01$
Random	0.0030	0.0001	0.0001	0.0001
Differentiated	0.9971	0.0001	0.0001	0.0001
Causal	0.5166	0.4227	0.4230	0.4229

Random SNPs with no association to the disease were generated by simulating random drift with ${\mathrm{F}}_{\mathrm{s}\mathrm{t}}$ divergence. Differentiated SNPs with no association to the disease were generated by assuming population allele frequencies of 0.8 of ancestry 1 and 0.2 of ancestry 2. Causal SNPs were generated by combining a multiplicative disease risk model while simulating the random drift with the same ${\mathrm{F}}_{\mathrm{s}\mathrm{t}}$ as the random SNPs. See methods for more details on the parameters.