Table 2.

Failed gold-standard generalizability test in Generation R

Canonical correlations	ABCD		Generation R
	training set	test set
r1	0.23	0.12*	0.03
r2	0.22	0.07*	0.03
r3	0.20	0.05	0.02

Note. Canonical correlations in ABCD were averaged across the 30 train-test splits.

r1: *p < 0.05 in 24 train-test splits. r2: *p < 0.05 in 12 train-test splits.

r1 Generation R: p < 0.05 in 1 train-test split.