"However, one counter-explanation for the seemingly powerful linkage between reduced drug use and reduced criminal behavior is that both outcomes are, essentially, measures of compliance with drug court, probation, or other supervision requirements. In this view, it is perfectly logical that offenders who were more influenced to become compliant with drug court requirements in one way (by reducing their drug use) also were more likely to comply in another way (by reducing their criminal behavior). Hence, the existence of a strong and direct statistical association between the two outcomes might not mean that reduced drug use brought about the reduced criminal behavior. Rather, reduced drug use and criminal behavior might be better interpreted as parallel outcomes, both involving compliance. One way of examining this counter-explanation would be to include a variable that directly taps compliance. After including such a variable, one could then test whether compliance appeared to be the essential link explaining reduced criminal behavior, or whether reduced drug use truly exerted an independent effect. For this reason, Model 3 included a variable for the number of supervision violations in the year prior to the 18-month survey. As expected, more supervision violations predicted more criminality. Yet, drug use still exerted a strong, independent effect on criminal behavior as well. Therefore, the results in Model 3 provide additional reason to conclude that a causal linkage does exist between reduced drug use and crime."

Source

Rossman, Shelli B., et al., "Final Report, Volume 4: The Multi-Site Adult Drug Court Evaluation: The Impact of Drug Courts" (Washington, DC: Urban Institute, June 2011), p. 75.
http://nij.gov/topics/courts/…
https://www.ncjrs.gov/pdffile…