Table 4.

Extremes analyses

Variable	χ2	df	P	AIC	h2(CI)	c2(CI)	e2(CI)
Understanding number
≥85%tile	4.45	12	.974	−19.55	.52(.25–.74)	.14(.00–.35)	.34(.25–.44)
Unselected+	6.66	13	.919	−19.34	.42(.31–.53)	.16(.07–.25)	.42(.38–.46)
Non-numerical processes
≥85%tile	3.55	12	.990	−20.45	.09(.00–.37)	.44(.23–.57)	.47(.36–.56)
Unselected+	9.90	13	.702	−16.10	.32(.20–.44)	.23(.14–.32)	.45(.40–.50)
Computation and knowledge
≥85%tile	8.74	12	.725	−15.26	.42(.14–.69)	.18(.00–.39)	.40(.30–.51)
Unselected+	13.63	13	.401	−12.37	.45(.32–.56)	.07(.00–.17)	.48(.44–.53)
Composite
≥85%tile	8.53	12	.743	−15.48	.53(.31–.75)	.25(.06–.42)	.22(.16–.30)
Unselected+	13.64	13	.399	−12.36	.49(.39–.59)	.19(.10–.27)	.33(.30–.36)

Parameter estimates (and confidence intervals) from the extreme math groups and unselected twins

Note: +, from Kovas et al. (2007)

Note: Parameter estimates were taken from the Scalar Model for all measures from Table 3, which constrained variance components to be equal but allowed for threshold differences between males and females