Questionable conclusions on cannabis and crime.

In a paper published recently in this journal [1], Pedersen & Skardhamar examine whether cannabis users are at increased risk of being charged with criminal offences. They find an association between frequency of cannabis use in one year and risk of criminal charges in subsequent years. The association is statistically significant with respect to drug-specific crime only, and not to other types of criminal offences. The authors conclude that (i) ‘. . . use of cannabis does not seem to represent a risk factor for a general criminal involvement’ and (ii) that ‘. . . the fact that a considerable proportion of adolescents and young adults may come into contact with the penal system because of their involvement with cannabis must [. . .] give rise to concern. [. . .] The present study must be regarded as a new argument for the necessity to debate cannabis laws and the international conventions which regulate this area’. These conclusions are highly questionable, however, due to weaknesses in their data set and analyses: the model specifications and statistical power are inadequate; the outcome variable (charge with drugspecific crime) does not reveal with what type of drug the offender is charged (cannabis or hard drugs); and the vital control variable (hard drug use) does not cover the relevant time-period. Regarding the first conclusion: the models employed suffer from endogeneity problems, as both drug use and crime can be influenced by the same unobserved factors. Not controlling for unobserved heterogeneity may lead to biases in estimated coefficients. Moreover, the data set does not have the statistical power to support the stated conclusion due to the small number of observations and the complex model employed by the authors. In the multivariate analyses presented in Table 4 (p. 114), they applied a three-category variable on cannabis use in the preceding year: never, one to 10 times and 11+ times. The number of people who were charged with criminal offences other than drug-specific offences within these categories were 12, 7 and 2, respectively, whereas the number of confounding variables that were included in the models were 8, 9 and 10. It is not surprising, then, that the adjusted odds ratio (OR) is not statistically significant. Conducting a bivariate analysis, however, reveals that cannabis users, at age 20, have a statistically significantly higher risk of being charged with criminal offences (other than drug-specific offences) compared to non-users [6.5% versus 1.1%; unadjusted OR = 5.7 (our calculations are based on data presented in Table 2, p. 113 [1])]. In essence, the data suggest that cannabis users are at higher risk of being charged with criminal offences, both drug-specific offences and other criminal offences. Whether, or to what extent, it is the use of cannabis per se or characteristics of the cannabis user that may explain the elevated risk, however, cannot be determined by this study. Regarding the second conclusion: it should be obvious that illicit drug users are at increased risk of being charged with drug-specific crimes. However, whether cannabis users are at increased risk of being charged with cannabis offences cannot be determined by this study, as the crime register data do not contain any information on what kind of drugs the charges apply to. Hence, an unknown fraction of the cannabis users who were charged with a drug-specific crime may have been charged because of their use, possession or distribution of an illegal drug other than cannabis (e.g. amphetamines or ecstasy). The authors have tried to get around this problem by controlling for other illicit drug use at the age of 20. This is not a satisfactorily control, however, as it is likely that many of the cannabis users have initiated other illicit drug use in the subsequent years during which the charges occurred [2]. This means that a person who reports cannabis use at the age of 20 may very well have progressed into amphetamine use by the time of drug arrest and the criminal charge may well relate to hard drugs, not cannabis. Thus, the data cannot support that the charges are, in fact, related to cannabis offences and therefore do not support their claim that a significant proportion of young adults come into contact with the legal system because of their cannabis use. Given the above-mentioned weaknesses, we believe that Pedersen & Skardhamar’s conclusions lack empirical support. What they find is simply that cannabis drug users, many of whom have used illegal drugs in addition to other illegal drugs, have an increased risk of being charged with a drug-specific crime. It could still be the case that a non-ignorable fraction of illegal drug users in Norway have been arrested for their cannabis use but, unfortunately, the study by Pedersen & Skardhamar cannot illuminate the issue.