. 2022 Jan 19;83(1):55–63. doi: 10.15288/jsad.2022.83.55

Table 4.

Hierarchical linear models predicting change in use of each cannabis formulation (N = 223)

Variable	Change in leaf				Change in concentrates				Change in edibles
Variable	B	OR	[95% CI]	P	B	OR	[95% CI]	P	B	OR	[95% CI]	P
Residential change from pre-closure to post-closure-1
Stayed dependent	−1.22	0.30	[0.15, 0.59]	<.001	−0.17	0.85	[0.47, 1.54]	.585	−0.94	0.40	[0.21, 0.76]	.004
Stayed independent	−0.79	0.46	[0.28, 0.75]	.002	−0.25	0.78	[0.50, 1.22]	.275	−0.39	0.67	[0.45, 1.00]	.052
Moved to dependent	−1.35	0.26	[0.15, 0.45]	<.001	−0.66	0.52	[0.32, 0.83]	.007	−0.81	0.44	[0.28, 0.70]	<.001
Residential change from post-closure-1 to post-closure-2
Stayed dependent	0.34	1.40	[0.87, 2.26]	.164	0.23	1.26	[0.75, 2.12]	.384	0.45	1.56	[0.85, 2.88]	.154
Stayed independent	−0.05	0.95	[0.62, 1.46]	.806	0.16	1.18	[0.74, 1.88]	.497	0.25	1.29	[0.80, 2.13]	.323
Moved to dependent	0.03	1.03	[0.61, 1.74]	.924	0.36	1.43	[0.77, 2.64]	.255	0.29	1.33	[0.71, 2.51]	.374
Moved to independent	0.14	1.15	[0.68, 1.97]	.602	0.35	1.42	[0.84, 2.39]	.189	0.50	1.66	[0.89, 3.09]	.113
Covariates^a
Sex (ref. = male)	−0.16	0.85	[0.50, 1.44]	.547	0.15	1.17	[0.67, 2.05]	.589	0.13	1.10	[0.69, 1.73]	.585
Age	−0.25	0.78	[0.56, 1.10]	.152	0.00	1.00	[0.71, 1.43]	.987	0.19	1.20	[0.92, 1.57]	.153
School A (ref. = not in school)	−0.02	0.984	[0.32, 3.06]	.978	−0.91	0.40	[0.09, 1.79]	.230	−1.26	0.28	[0.09, 0.87]	.028
School B (ref. = not in school)	0.95	2.58	[0.89, 7.53]	.081	−0.15	0.86	[0.21, 3.58]	.839	−0.28	0.76	[0.27, 2.15]	.602
School C (ref. = not in school)	0.06	1.06	[0.41, 2.75]	.909	−0.20	0.82	[0.26, 2.61]	.737	−0.41	0.66	[0.30, 1.47]	.308
Other (ref. = not in school)	−0.16	0.85	[0.26, 2.81]	.788	−0.49	0.61	[0.15, 2.53]	.494	−0.66	0.52	[0.16, 1.65]	.264
Cannabis legalization (ref. = no)	−0.29	0.75	[0.34, 1.66]	.474	−0.18	0.83	[0.29, 2.39]	.734	−0.03	0.97	[0.00, 16.17]	.940

Notes: Bolded effects represent p < .05. OR = odds ratio; CI = confidence interval; ref. = referent group. Although not tabled for parsimony, models also included each Level 2 variable that was entered as a predictor of Level 1 slope effects (i.e., dummy codes for change in living situation) as an additional predictor of the model intercept, to appropriately test the cross-level interactions.

^{^a}

Effects of covariates are presented from the model in which Stayed Dependent was the referent group, but remained non-significant across all models.