Table 1. Summary of linear mixed modeling results for predictors of the quality of participants’ social functioning.

			Effect
			Fixed			Random
Variable			b	SE	t	Deviance
Step 1 (intercept-only)						794.27**
Intercept			8.89*	.21	42.42
Step 2 (level 1)						3.77
Intercept			8.97*	.26	34.12
Slope (linear Δ)			-.16	.25	-.628
Step 3 (level 2) a						60.53**
	Model 1 b (Angry)
		Intercept	6.18*	.77	8.05
		Lifetime diagnosis	.002	.38	.01
		Neuroticism	.03*	.01	3.83
		EQ	.03	.02	1.29
		PR	-3.01	1.71	-1.76
		G	2.01	1.55	1.29
		EQ X PR	-.56*	.19	-2.89
		EQ X G	-.08	.16	-.49
	Model 2 (Sad)					60.68**
		Intercept	6.18*	.81	7.62
		Lifetime diagnosis	-.08	.42	-.19
		Neuroticism	.03*	.01	3.60
		EQ	.02	.02	.76
		PR	-.53	1.74	-.31
		G	3.67	2.00	1.83
		EQ X PR	-.14	.19	-.69
		EQ X G	-.16	.26	-.61
	Model 3 (Happy)					55.27**
		Intercept	6.00*	.76	7.87
		Lifetime diagnosis	.05	.39	.13
		Neuroticism	.03*	.01	3.91
		EQ	.02	.02	1.00
		PR	3.05	1.70	1.93
		G	-.13	1.79	-.07
		EQ X PR	-.26	.20	-1.28
		EQ X G	-.23	.23	-1.02

Note. In these analyses, the intercept represented participants’ social functioning at time 1.

^a Slope was set as fixed at Step 3 due to non-significant change in model fit at Step 2. Predictors, added at Step 3, were used to explain between-subject variability in the intercept only.

^b Three models, distinguished by type of emotional stimulus, were run at Step 3. Within each model, variables were entered hierarchically (1- covariates, 2—main effects, 3—interaction effects).

EQ = empathy quotient; PR = personally-relevant; G = generic.

* p < .05

** Increase in model fit was statistically significant (p < .05) based on chi-square test of deviances.