Table 2.

Multilevel structural equation models predicting alcohol related consequences from alcohol and cannabis use order controlling for day of the week, sex, age, and drinks on a given day

Level	Outcome	Predictor	RR	[95% CI]
Pregaming in the model
Within	Consequences	Cannabis before drinking	1.29	[1.13, 1.46]
		Cannabis during drinking	1.24	[1.14, 1.35]
		Cannabis after drinking	0.92	[0.83, 1.00]
		Drinks	1.18	[1.17, 1.19]
		Pregaming	1.24	[1.18, 1.31]
Between	Consequences	Sex	1.05	[0.95, 1.16]
		Age	0.96	[0.92, 1.00]
	Intercept		0.86	[0.72, 1.00]
Pregaming days
Within	Consequences	Cannabis before drinking	0.82	[0.58, 1.05]
		Cannabis during drinking	0.89	[0.77, 1.02]
		Cannabis after drinking	1.07	[0.93, 1.20]
		Drinks	1.16	[1.14, 1.18]
Between	Consequences	Sex	1.07	[0.95, 1.19]
		Age	1.04	[0.99, 1.08]
	Intercept		2.10	[1.86, 2.34]
Drinking days without pregaming
Within	Consequences	Cannabis before drinking	1.88	[1.67, 2.10]
		Cannabis during drinking	1.14	[0.97, 1.31]
		Cannabis after drinking	0.81	[0.70, 0.92]
		Drinks	1.22	[1.20, 1.24]
Between	Consequences	Sex	0.91	[0.78, 1.04]
		Age	0.88	[0.84, 0.93]
	Intercept		0.61	[0.45, 0.78]

Notes: The alcohol-related consequences outcome variable was specified with a negative binomial count distribution: M = 1.62, SD = 2.25, range: 0–17, Mdn = 1.0, intraclass correlation coefficient = 0.38. Drinks represents the total drinks consumed in a day: M = 4.67, SD = 3.41, range: 1–27, Mdn = 4. Bold indicates that the 95% confidence interval (CI) does not contain 1.0. RR = rate ratio. Within = daily-level effects; between = person-level effects. All within-level predictors presented were person-mean centered. Day of the week was included as six dummy-coded variables, with Sunday as the referent group. Day-of-the-week effects were omitted from the table for clarity.