Table 5.

Summary of multivariate model comparisons for the VTSABD longitudinal adolescent depression scores

Model	A	E	k	−2ln(l)	Δ− 2LL	Δdf	P	Δ− 2LL	CM	Δdf	P	AIC
1	Saturated	Saturated	120*	21369.454	–	–	–	–	–	–	–	–
2	No cross- temporal covariance	No cross- temporal covariance	44	22333.922	964.47	76	<1 × 10−5	–	–	–	–	812.468
3	C + L + Q + β, Ψ, ε	C + L + Q + β, Ψ, ε	18	21486.131	116.68	102	1.52 × 10−1	–	–	–	–	−88.677
4	Drop Q from 3	Drop Q from 3	12	21500.178	130.72	108	6.8 × 10−2	14.04	3	6	2.9 × 10−2	−86.724
5	Drop L & Q from 3	Drop L, Q from 3	8	21528.741	159.29	112	2.2 × 10−3	28.57	4	4	<1 × 10−5	−65.287
6	Drop C, L & Q from 3	Drop C, L, Q from 3	6	21591.980	222.53	114	<1 × 10−5	63.24	5	2	<1 × 10−5	−6.526
7	Drop B from Model 3	Model 3	16	21486.732	117.28	104	1.75 × 10−1	0.60	3	2	7.4 × 10−1	−91.278
8	Model 3	Drop B from Model 3	16	21494.561	125.11	104	7.78 × 10−2	8.53	3	2	1.4 × 10−2	−82.893
9	Drop B from Model 3	Drop B from Model 3	14	21499.969	130.51	106	5.34 × 10−2	13.83	3	4	7.8 × 10−3	−82.515
10	Model 7	Drop Q from Model 7	13	21494.184	124.73	107	1.16 × 10−1	8.05	7	3	4.5 × 10−2	−89.270
11	Drop Q from Model 10	Model 10	10	21511.221	141.77	110	1.90 × 10−2	17.04	10	3	0.6 × 10−3	−78.233

Elements of Ψ and ε are assumed to be constant over time unless noted otherwise (“free”)

A additive genetic, E non-shared environment, C Constant, L Linear, Q Quadratic, β Beta regression coefficient, Ψ item-specific or residual variances, ε error variances for the latent depression factors, k number of free parameters in Model for covariance structure (ignoring mean parameters), CM comparison model, Δ−2LL change in −2 × Log Likelihood asymptotically distributed as a χ², Δdf change in degrees of freedom Model 10 best fitting model

^*12 parameters fixed to zero (no data available for estimation of remote correlations