Table 2.

Latent profile analyses of executive functions—with the 4-profile solution allowing partial conditional dependence providing the best fit.

N profiles	N free parameters	Log-likelihood	Adjusted BIC a	AIC b	Entropy	LMR adjusted test c H0 log likelihood	Bootstrapped LRT d H0 log likelihood	Smallest class (% of sample)
1	10	−11401.38	22845.12	22822.77	n.a.	n.a.	n.a.	n.a.
2	16	−11190.31	22448.40	22412.63	0.86	412.86**	422.14***	12.18
3	22	−10991.97	22147.02	22027.94	0.89	387.96***	396.68***	7.26
4	28	−10901.13	22009.83	21858.27	0.83	177.68 ***	181.68 ***	6.88
5	34	−10828.74	21909.52	21725.48	0.79	141.60 *	144.79 ***	3.10
6	40	−10791.32	21879.15	21662.64	0.81	73.20 (n.s.)	74.84***	0.25
	Relaxing partial conditional independence in models for the correlation of all three cool EF
1	13	−11268.45	22591.97	22562.90	n.a.	n.a.	n.a.	n.a.
2	19	−11058.53	22197.54	22155.05	0.94	410.61***	419.85***	7.80
3	25	−10898.41	21902.72	21846.82	0.92	313.19*	320.23***	6.04
4	31	−10798.82	21728.95	21659.64	0.85	194.80 *	199.18 ***	4.47
5	37	−10753.81	21664.35	21581.62	0.85	88.04 (n.s.)	90.02***	0.40

Correlations across cool EF included: inhibition*updating, updating*flexibility, and flexibility*inhibition. ^*p < 0.05, ^**p < 0.01, and ^***p < 0.001.

^aBayesian information criterion.

^bAkaike information criterion.

^cLo–Mendell–Rubin adjusted likelihood ratio test.

^dBootstrapped likelihood ratio test.Fit indices in bold indicate models with good fit.