pp. 3-30

Howard S. Becker’s study of the “career” of marijuana smokers became a reference for later ethnographic study but has hardly been an inspiration for epidemiologists’ statistical studies. This article aims first to better understand this apparent incompatibility by pointing out the pitfalls epidemiologists are exposed to when they analyze drug use. It then reexamines available data, with the purpose of drawing from Becker’s theses a set of hypotheses that can be statistically validated. Lastly, these hypotheses are validated on the basis of responses to a self-administered questionnaire survey of 12 113 teenage pupils conducted in 1999 in mainland France. The results confirm the relevance of Becker’s sequential approach to cannabis use, since they show that 1) use-related factors, namely type of supply point and use among peers, vary by cannabis consumption level ; 2) the neutralization techniques Becker hypothesized may be seen to be operative in respondents’ stated opinions. More generally, the article illustrates the possible complementarity between statistical and ethnographic approaches.