Table 1.

Bivariate GREML estimates of variance components between the (within-person) mean and two measures of within-person variability

Trait	$h_{SNP}^2$					$r_g$	s.e.	P
	mean	s.e.	variability	s.e.	P-value
A. Within-person trait mean and (absolute) trait deviation
height	0.527	0.008	0.024	0.007	2.0 × 10−4	0.094	0.053	0.081
BMI	0.256	0.008	0.054	0.007	8.8 × 10−15	0.872	0.052	1.1×10−63
weight	0.284	0.008	0.053	0.007	2.6 × 10−14	0.799	0.050	1.4×10−56
body fat %	0.242	0.008	0.000	0.007	0.966	—	—	—
waist cir.	0.226	0.008	0.011	0.006	0.103	—	—	—
hip cir.	0.246	0.008	0.009	0.006	0.157	—	—	—
waist:hip	0.200	0.008	0.000	0.006	0.969	—	—	—
sitting height	0.456	0.008	0.020	0.007	2.4×10−3	0.078	0.063	0.213
B. Within-person trait mean and rate of trait change
height	0.524	0.008	0.015	0.006	0.023	−0.130	0.072	0.072
BMI	0.254	0.008	0.035	0.007	3.4×10−7	0.163	0.059	0.006
weight	0.284	0.008	0.031	0.007	3.0×10−6	0.089	0.059	0.129
sitting height	0.452	0.008	0.010	0.006	0.125	—	—	—

Within-person variability was assessed with either A. the non-direction (absolute) trait deviation or B. the age- and sex- adjusted rate of trait change for 2 repeated measures of up to 8 anthropometric traits. Variance components are the SNP-based heritability ($h_{SNP}^2$) and genetic correlation ($r_g$) between the mean and the measure of variability, where genetic correlations are reported only when $h_{SNP}^2$ of both traits significantly greater than zero (P < 0.05). Estimates are shown with standard errors (s.e.), and chi-squared tests (with 1 df) were used to calculated unadjusted P values.