from below the 10tth percentile and above the 60th percentile respectively on a known liability distribution. We examined the power of the high-risk design, where subjects were ascertained through the father, with 500 case fathers and 500 control fathers (Figure 2d). We also examined two modifications of the high risk design, both of which included sampling conditional on a single measured continuous environmental factor which accounted for 25% of the total variance in the underlying liability. Under the low environmental risk sampling approach (Figure 2e), 500 high-risk (case father) and 500 low-risk (control father) trios were selected from the bottom half of the measured environmental risk scale. Under the extreme environmental sampling approach, 500 high-risk (affected father) trios were selected with values on the environmental risk below the mean and 500 low-risk (control father) trios were selected with environmental risk above the mean (Figure 2f).
