Table 1.

Results obtained with the best model selected from the various statistical models run for the response variables “proportion of analyses yielding homogeneous misleading clusterings” and “proportion of analyses yielding heterogeneous misleading clusterings”

Response variable	Number of loci	Factors of selected model	Estimate	Std error	z-value (df = 7499)	P
Proportion of analyses yielding homogeneous misleading clusterings
	10	N	−0.004	0.0005	−7.931	<0.0001
		log10(N/NF)	0.399	0.084	4.746	<0.0001
		N x log10(N/NF)	0.001	0.0002	5.185	<0.0001
	20	N	−0.005	0.0005	−9.658	<0.0001
	100	N	−0.0003	0.00006	−5.757	<0.0001
		log10(N/NF)	−0.194	0.259	−0.749	0.4540
		N × log10(N/NF)	−0.028	0.005	−5.190	<0.0001
Proportion of analyses yielding heterogeneous misleading clusterings
	10	N	−0.0002	0.00007	−3.496	0.0005
		log10(N/NF)	1.107	0.114	9.672	<0.0001
		N × log10(N/NF)	0.00004	0.00002	1.827	0.0677
	20	N	−0.0004	0.00006	−5.977	<0.0001
		log10(N/NF)	0.574	0.079	7.178	<0.0001
		N × log10(N/NF)	0.00009	0.00002	4.854	<0.0001
	100	N	−0.0009	0.0001	−9.666	<0.0001
		log10(N/NF)	−0.801	0.068	−11.715	<0.0001
		N × log10(N/NF)	0.0003	0.00003	11.245	<0.0001

Note: Significant P-values, for a 5% threshold of significance, are shown in bold