Table 7.

Parameter Estimates from Model 2^a

	Male			Female
Coefficients	Estimate	95%CI	p	Estimate	95%CI	p
Autoregressive
ASB12→ASB13b	0.789	0.658, 0.920	<0.001	0.782	0.664, 0.900	<0.001
ASB13→ASB15	0.746	0.673, 0.819	<0.001	0.757	0.686, 0.828	<0.001
ASB15→ASB17	0.645	0.559, 0.731	<0.001	0.725	0.633, 0.817	<0.001
ALC12→ALC13b	0.670	0.564, 0.776	<0.001	0.673	0.573, 0.773	<0.001
ALC13→ALC15	0.497	0.436, 0.558	<0.001	0.498	0.443, 0.553	<0.001
ALC15→ALC17	0.507	0.436, 0.578	<0.001	0.507	0.448, 0.566	<0.001
Cross-lagged
ALC12→ASB13	−0.062	−0.229, 0.105	0.468	−0.057	−0.202, 0.088	0.446
ALC13→ASB15	0.079	−0.017, 0.175	0.102	0.019	−0.071, 0.109	0.676
ALC15→ASB17	0.156	0.062, 0.250	0.001	0.057	−0.051, 0.165	0.300
ASB12→ALC13	0.047	−0.090, 0.184	0.503	0.078	−0.051, 0.207	0.237
ASB13→ALC15	0.276	0.202, 0.350	<0.001	0.269	0.202, 0.336	<0.001
ASB15→ALC17	0.120	0.036, 0.204	0.005	0.131	0.055, 0.207	0.001
Correlationc
ASB12-ALC12	0.676	0.625, 0.727	<0.001
Residual correlationc
ASB13-ALC13	0.238	0.115, 0.361	<0.001
ASB15-ALC15	0.459	0.390, 0.528	<0.001
ASB18-ALC17	0.649	0.580, 0.718	<0.001
Fit indices	χ2 = 12,557.796, df = 5,413, p < 0.001
	CFI = 0.960, TLI = 0.961
	RMSEA [90%CI] = 0.017 [0.017, 0.018]

CFI, comparative fit index; RMSEA, root mean square error of approximation; TLI, Tucker-Lewis index.

^aIn Model 2, autoregressive and cross-lagged paths were allowed to vary between the male and female subgroups, so separate autoregressive and cross-lagged paths by sex are presented.

^bASB and ALC are factors for antisocial behavior and alcohol use, respectively.

^cCorrelations were constrained equal across male and female subgroups in Model 2. Correlations were estimated at the initial age (age 12), and residual correlations were estimated at the rest of ages because ALC and ASB factors at ages 13 to 17 were endogeneous factors (i.e., dependent latent variables).