REPLY TO BRETTEVILLE-JENSEN & ROSSOW
Article first published online: 5 JAN 2011
Addiction © 2011 Society for the Study of Addiction
Volume 106, Issue 2, pages 450–451, February 2011
How to Cite
PEDERSEN, W. and SKARDHAMAR, T. (2011), REPLY TO BRETTEVILLE-JENSEN & ROSSOW. Addiction, 106: 450–451. doi: 10.1111/j.1360-0443.2010.03279.x
- Issue published online: 5 JAN 2011
- Article first published online: 5 JAN 2011
Bretteville-Jensen & Rossow  argue that our analyses suffer from (i) unobserved heterogeneity, (ii) a lack of statistical power and (iii) the possibility that the association between cannabis use and later criminal charges is based on the use of other drugs and other criminal charges (i.e. not cannabis-related). We will address each of these arguments.
Our study is observational, and of course we agree with their first objection in general: both drug use and crime may be caused by the same unobserved factors. We discuss this possible limitation in the paper (p. 116). Moreover, this will always be the situation with non-experimental designs. So, this objection is not contrary to what we have stated, and it applies to all observational studies.
Their second objection is related to statistical power. First, note that our main finding was that cannabis users were at risk of being charged with subsequent drug-related crimes. The argument about power is not relevant to this finding. The point estimate was very high, although the confidence intervals were wide (Table 4, column 5). In addition, we reported a non-significant association between cannabis use and charges not related to drugs (Table 4, column 4). In a commentary that was published with our paper, David Farrington argued that the low number of non-drug offenders might have made it difficult to obtain significant statistical results in this analysis . We agree, and this implies that we may have underestimated the risk of cannabis users committing non-drug-related crimes. We are somewhat puzzled, however, that Bretteville-Jensen & Rossow now see the need to repeat this point. Note also that this does not alter our main conclusion: cannabis users are at high risk of being charged with subsequent drug-related crimes.
Bretteville-Jensen & Rossow also make a point of calculating the bivariate odds ratios to show that cannabis is indeed associated with non-drug crimes (which is also evident in our Table 2). [However, we do not understand how Bretteville-Jensen & Rossow found the counts 12, 7 and 2. The correct counts reported in Table 2 are 16, 9 and 8. Thus, their calculated odds ratios are incorrect.] For cannabis users versus non-users, the bivariate odds ratio for non-drug crimes is 6.1, and for drug-specific crimes the odds ratio is 13.8 (both estimates are statistically significant). The main difference, however, is attributable to those who had used cannabis more than 11 times. Note that this mirrors the difference between the results reported in the final models in Table 4, columns 4 and 5. Thus, cannabis users are more likely to be charged with drug-related crimes than non-drug crimes.
Their third objection concerned our lack of information about (i) the kinds of drugs that prompted the charges and (ii) whether cannabis users went on to use other illegal drugs after the age of 20 years. However, we did not suggest in our paper that the charges rested on cannabis-related crime alone. On the contrary, we stated explicitly that: ‘it would be valuable to know the extent to which these crimes are related to cannabis or harder drugs’ (p. 116). Furthermore, we discussed a number of possible mechanisms linking cannabis use to later drug-related charges. Of particular importance seems to be that many cannabis users take part in small-level dealing. We concluded that the details of the mechanisms linking cannabis use to criminal sanctions in different legal jurisdictions should be investigated in future research (p. 117).
One in four 20-year-olds who had used cannabis more than sporadically in the last year were charged with a drug-related crime during the next 7 years . There are many reasons to report this association. The bulk of previous research on the consequences of cannabis use has had a biomedical perspective that is focused on mental health and dependence. However, the most important problems associated with cannabis may be the criminal behaviours, illegal subcultures and marginalization that stem from the criminalization of cannabis use.