Table 2.

Mediation analyses including direct and indirect effects for the trimmed model.

	Direct effects
	β	95% CI
BIS → “wanting”	0.12 (p = 0.03)	(0.01, 0.23)
“Wanting” → problem uses	0.55 (p < 0.001)	(0.43, 0.69)
RDS → problem uses	0.24 (p < 0.001)	(0.14, 0.33)
Problem uses → wellbeing	−0.09 (p = 0.03)	(−0.16, −0.01)
BIS → wellbeing	−0.20 (p < 0.001)	(−0.29, −0.11)
	Indirect effects
BIS → “wanting” → problem uses	0.07 (p = 0.04)	(0.01, 0.14)
BIS → “wanting” → problem uses → wellbeing	−0.099 (p = 0.048)	(−0.02, 0.00)
RDS → problem uses → wellbeing	−0.03 (p = 0.004)	(−0.05, −0.01)

Bootstrapped confidence intervals were based on 5,000 replications and were estimated with diagonally weighted least squares. β = standardized regression weights, 95% CI bias-corrected bootstrapped confidence intervals.