Table 2.

Indirect effects of mediational paths linking baseline dual use to cigarette use at T3 via nicotine dependence and cigarette use at T2^*

	Total main effect model†	Mediational path analysis model‡
	Dual use status T1 → Cigarette use T3	Dual use status T1 → Cigarette use T3	Dual use status T1 → Cigarette use T2 → Cigarette use T3		Dual use status T1 → Nicotine dependence T2 → Cigarette use T3
	Total effects	Direct effects	Indirect effects§	Proportion mediated	Indirect effects§	Proportion mediated
Dual use status Tl	b (95% Cl)	b (95% Cl)	b (95% Cl)	% (95% Cl)	b (95% Cl)	% (95% Cl)
No use	REF	REF	REF	-	REF	-
Cannabis use only	0.95 (0.36 to 1.61)	0.82 (0.28 to 1.36)	0.03 (0.01 to 0.05)	3.2 (1.1 to 5.3)	0.10 (0.06 to 0.14)	10.5 (6.3 to 14.7)
E-cigarette use only	0.96 (0.04 to 2.57)	0.79 (−0.02 to 2.46)	0.03 (−0.01 to 0.13)	-	0.14 (−0.02 to 0.45)	-
Dual use	1.77 (1.15 to 2.55)	1.26 (0.11 to 2.31)	0.10 (0.03 to 0.18)	5.7 (1.7 to 10.2}	0.41 (0.17 to 0.64)	23.3 (9.6 to 36.3)

^*Analytical sample n=1164. All models were adjusted for past 30-day cigarette use or other tobacco products and nicotine dependence at T1 and age, sex, race/ethnicity and socioeconomic status.

^†Total main effect model presented in figure 2.

^‡Mediational path onlays is model presented in figure 3.

^§Indirect effect estimates (path coefficients between T1 and T2 x path coefficients between T2 and T3; b (95% Cl) and effect size (proportion mediated: the proportion of indirect effect to the total effect) were calculated based on the mediational path model presented in figure 3. Unstandardised path coefficient (b) and 95% Cl are presented.