Table 4. Hierarchical linear models examining 6-week (pre-BMI) marijuana use status as a predictor of alcohol outcomes following the BMI.

	HED Frequency				Peak BAC				Alcohol Consequences
	B	SE	t-ratio	p	B	SE	t-ratio	p	B	SE	t-ratio	p
Intercept (Baseline) (β00)	8.05	0.28	28.51	<.001	0.21	0.01	36.96	<.001	7.26	0.31	23.79	<.001
Gender (β01)	-0.68	0.53	-1.28	.20	0.004	0.01	0.41	.69	0.12	0.50	0.24	.81
Condition (β02)	-0.59	0.48	-1.22	.22	-0.02	0.01	-2.17	.03	-0.70	0.49	-1.43	.15
MJ user (β03)	1.81	0.48	3.75	<.001	0.02	0.01	2.58	.01	2.33	0.49	4.73	<.001
3-month follow-up (β10)	-0.71	0.31	-2.25	.03	-0.02	0.01	-3.75	<.001	-0.92	0.27	-3.43	<.001
Gender(β11)	-0.66	0.52	-1.26	.21	-0.01	0.01	-0.86	.39	-0.57	0.51	-1.10	.27
Condition (β12)	-0.07	0.50	-0.13	.90	0.004	0.01	0.41	.69	-0.75	0.46	-1.61	.11
MJ user (β13)	0.15	0.52	0.29	.77	0.01	0.01	1.07	.29	-0.27	0.47	-0.57	.57
MJ user*Condition (β14)	0.14	0.92	0.15	.88	0.01	0.02	0.83	.41	-0.46	0.84	-0.55	.58
6-month follow-up (β20)	-0.81	0.31	-2.64	.01	-0.03	0.01	-4.44	<.001	-1.13	0.29	-3.86	<.001
Gender (β21)	-1.37	0.49	-2.76	.01	-0.02	0.01	-1.59	.11	-0.76	0.51	-1.48	.14
Condition (β22)	0.19	0.49	0.40	.69	0.01	0.01	0.54	.59	-0.76	0.48	-1.57	.12
MJ user (β23)	-0.39	0.49	-0.80	.42	-0.01	0.01	-0.71	.48	-0.37	0.49	-0.77	.44
MJ user*Condition (β24)	1.62	0.87	1.86	.06	0.03	0.02	1.45	.15	1.08	0.89	1.22	.23
9-month follow-up (π3i)	-0.47	0.34	-1.39	.17	-0.03	0.01	-5.05	<.001	-1.04	0.28	-3.75	<.001
Gender (β31)	-2.00	0.54	-3.67	<.001	-0.02	0.01	-1.74	.08	-0.98	0.51	-1.93	.06
Condition (β32)	0.51	0.53	0.95	.34	0.01	0.01	1.34	.18	-1.18	0.47	-2.51	.01
MJ user (β33)	-0.27	0.53	-0.50	.62	-0.01	0.01	-0.69	.49	-0.51	0.47	-1.08	.28
MJ user*Condition (β34)	1.17	0.94	1.25	.21	0.01	0.02	0.40	.69	0.36	0.87	0.42	.68

Note. Degrees of freedom in all models = 388 for intercept (pre-BMI) effects and 387 for follow-up effects. BAC = blood alcohol concentration. HED = heavy episodic drinking. MJ = Marijuana. Gender was coded as 0 for males and 1 for females Example HLM model equation (predicting consequences) is shown below, with coefficients corresponding to those listed in the table.

Level-1 Model

Consequences_ti = π_0i + π_1i*(3moFU_ti) + π_2i*(6moFU_ti) + π_3i*(9moFU _ti) + e_ti

Level-2 Model

π_0i = β₀₀ + β₀₁*(GENDER_i) + β₀₂*(CONDITION_i) + β₀₃*(MJ USER_i) + r_0i

π_1i = β₁₀ + β₁₁*(GENDER_i) + β₁₂*(CONDITION_i) + β₁₃*(MJ USER_i) + β₁₄*(MJ USERxCONDITION_i) + r_1i

π_2i = β₂₀ + β₂₁*(GENDER_i) + β₂₂*(CONDITION_i) + β₂₃*(MJ USER_i) + β₂₄*(MJ USERxCONDITION_i) + r_2i

π_3i = β₃₀ + β₃₁*(GENDER_i) + β₃₂*(CONDITION_i) + β₃₃*(MJ USER_i) + β₃₄*(MJ USERxCONDITION_i) + r_3i