Table 1.

Results of the final mixed-effects regression models for aggression and depression

Parameters	Aggression			Depression
	Estimate	S.E.	p	Estimate	S.E.	p
Intercept	2.8	0.35	< .0001	2.11	0.28	< .001
Time	− 0.27	0.26	0.3028	− 0.57	0.2	0.0053
Time^2	− 0.36	0.1	0.0003	− 0.003	0.08	0.9632
Age	0.19	0.08	0.018	0.21	0.06	0.0013
Sex	− 0.46	0.31	0.1433	− 0.19	0.24	0.4361
Group	0.35	0.31	0.2652	0.18	0.24	0.4603
Age*sex	− 0.09	0.08	0.2372	− 0.05	0.06	0.3747
Age*group	0.01	0.08	0.8732	− 0.08	0.06	0.2225
Sex*group	− 0.47	0.43	0.2775	0.68	0.33	0.0434
Time^2*age	0.03	0.01	0.0013	− 0.004	0.01	0.5721
Time^2*sex	0.18	0.11	0.1034	0.1	0.09	0.2581
Time^2*group	0.27	0.11	0.0185	0.5	0.09	< .001
Age*sex*group	− 0.12	0.11	0.2795	0.12	0.08	0.1642
Time^2*age*sex	− 0.01	0.01	0.1658	− 0.01	0.01	0.2206
Time^2*age*group	− 0.04	0.01	0.0009	− 0.07	0.01	< .001
Time^2*sex*group	0.56	0.16	0.0004	− 0.52	0.12	< .001
Time^2*age*sex*group	− 0.09	0.02	< .001	0.08	0.01	< .001
Model fit
Intercept VAR	0.13			0.14
Residual VAR	0.26			0.14
Residual SD	0.51			0.38
AIC	1339.17			1029.52
BIC	1448.5			1138.85
− 2LL	1291.17			981.52

Significant estimates are italic (p < 0.05). Aggression and depression scores were transformed to lower skewness. Time^2 is the quadratic term for time. Age was centered on the grand mean. Sex was binary coded: 0 = female, 1 = male. Group was binary coded: 0 = control group, 1 = clinical group. VAR variance, SD standard deviation, AIC Akaike information criteria, BIC Bayesian information criterion. − 2LL = negative two log likelihood