Table 3.

Model selection results with PSSI β-signatures in the left IFG as the dependent variables and MASQ^{^a}

Model	Predictors	Goodness of fit (in small-is-better form)		Omnibus test
Model	Predictors	Log likelihood	Akaike's Information criterion	Likelihood ratio χ²	df	p
Model 1: GLM
1	GDA, GDD, AA, AD	14.41	−16.83	9.478	5	0.05
2	GDA, GDD, AD	14.22	−18.43	9.08	3	0.028
3	GDA, GDD	13.42	−18.84	7.496	2	0.024

Component	Variance	Error	R²	PRESS	Predicted R² by LOOCV
Model 2: PLS regression
1	0.354	0.693	0.251	0.910	0.016
2	0.817	0.670	0.276	0.893	0.035

↵^aAA, Anxious arousal; AD, anhedonic depression; PRESS, prediction residual error sum of squares; LOOCV, leave-one-out cross validation.