serious drug problem. It simply states that the use of one drug increases the risk of starting to
consume another, and possibly more harmful, drug later on and that the risk increases with
frequency of use. Although controversial, the hypothesis has had considerable influence on
drug policy and legislation in many countries. The contributions made in this paper are, first,
to test the hypothesis on a unique dataset that also include detailed information about the price
of drugs, and, second, to do so by developing a new approach for testing using the
Expectation Maximization method to solve the maximum likelihood problem in a latent class
model. The results indicate that there are two distinct groups within the sample; one large
group for which the gateway effect of cannabis is statistically significant and one small group
of “troubled youths” for which there are no signs of a gateway effect.