Table 1. Results simulation study (Nsim = 10.000) into the power to detect a genetic variant explaining .1% of the variance with 5 differently constructed phenotypic instruments (an complete scale with 27 items, a subtest with the 9 middle items, a subtest with 9 items selected to cover the entire continuum, a subtest with 5 low-extreme and 4 high-extreme items, and a subtest with 9 high-extreme items) in two designs: a population design (N = 2500) and a selected-samples design (1250 extreme subjects and 1250 subjects from the normal range).

	α = .05				α = .01				α = .001
	Population		Selected samples		population		Selected samples		population		Selected samples
	#hits	Ratio	#hits	Ratio	#hits	ratio	#hits	Ratio	#hits	ratio	#hits	ratio
All 27 items	3763		5692		1741		3324		524		1276
9 middle items	3289	.87	5119	.90	1477	.85	2780	.84	406	.77	985	.77
9 high extreme	1967	.52	4629	.81	706	.41	2340	.70	143	.27	768	.60
9 items across the scale	2606	.69	4546	.80	1009	.60	2358	.71	246	.47	780	.61
5 low-extreme+4 high-extreme	1171	.31	2589	.45	362	.21	1030	.31	56	.11	224	.18

Note. #hits denotes the number of p-values<α = .05, α = .01, or α = .001, respectively. Ratio denotes the % of hits that the 4 subscales pick up, compared to the full instrument including all 27 items.