Empirical studies have found that cannabis commonly precedes consumption of drugs like amphetamine, ecstasy, cocaine and heroin. As a result a causal linkage between cannabis and subsequent hard drug use has been hypothesized. Despite mixed empirical evidence and a limited understanding of possible transmission mechanisms, the causal gateway hypothesis has been influential in formulating a strict drug policy in many western countries. Individual differences in proneness and accessibility, however, provide alternative, non-causal explanations for the observed "staircase" pattern and yield potentially different policy implications. We propose a Bayesian estimation and predictive framework to analyze the effects and relative importance of previous cannabis use, proneness and accessibility factors on hard drug initiation and to explore potential policy implications, using data from a unique recent survey of young adults in Norway. Motivated by the gateway transmission channels proposed in the literature, our model allows for a constant and a heterogeneous effect of previous cannabis use on hard drug initiation and, also, a more flexible correlation pattern for the unobservables. We find that proneness, accessibility and previous cannabis use contribute to the observed higher drug use pattern among cannabis users. The latter has the largest effect and is driven by various transmission channels.