Table 4.

Two part model of the quantity of cigarettes smoked in a month

	Part 1: P(nb cig > 0) N = 721			Part 2: log(nb cig) N = 245
Variable	OR	β/σ		exp(β)	β/σ
Academic discipline (ref = sociology)
• English as a foreign language				1.0318	0.09
• Medicine	-			0.4828	-3.25	**
• Nursing				0.6809	-2.16
Perceived approval of regular smoking by friends (ref = strong disapproval)
• Approval	4.5358	5.08	***	2.9571	3.41	**
• Disapproval	1.6844	2.13	*	2.4925	6.28	***
Perceived proportion of friends smoking (ref = none)
• <33%	1.2195	0.52
• Half	3.4953	3.13	**	-
• >66%	2.6591	2.61	**
Smoke discomfort in university (ref = never)
• Seldom	0.4227	-2.15	*	0.6747	-2.33	*
• Sometimes	0.2606	-3.18	**	0.2849	-4.45	**
• Often	0.1572	-3.19	**	0.1264	-4.29	**
Position about smoke-free universities (ref = against)
• Indifferent	0.2288	-2.62	**	0.5718	-2.98	*
• Mostly for	0.1506	-3.99	***	0.5687	-1.99
• Totally for	0.0756	-6.54	***	0.4794	-3.52	**
Tobacco perception score (high scores = positive perceptions)	1.1635	3.63	***	-
Beliefs about tobacco industry score (high scores = negative beliefs)	-			1.0342	2.26	*
Binge drinking (ref = no)
• Occasional (< 4 times/month)	2.9303	2.54	*	-
• Weekly (≥ 4 times/month)	1.3186	0.39
Cannabis use (ref = no use)
• Occasional (≤ 1/week)	3.4140	3.62	***	-
• Regular (> 1/week)	8.2666	2.13	*
Gender (ref = male)	2.7103	2.80	**	-
Age	-			1.0366	3.13	*
Constant (exponentiated)	0.2899	-1.40		30.9910	8.22	***
Duan's smearing estimator	Not applicable			2.2494	Not applicable

*** p < 0.001; ** p < 0.010; * p < 0.050

Notes: Only independent variables with a bootstrap-estimated, backward selection inclusion probability ≥ 80% were included in each part of the model; - = variable not included; OR = odds ratio; β/SE = Wald test statistic (β = regression coefficient, σ = standard error of the coefficient). Part 1 models the probability of being a smoker using logistic regression, part 2 models the log-number of cigarettes smoked in a month by smokers using linear regression.