The criminals have lots of violent disputes, not just over territory, and the users commit more than just property crime, e.g. robbery. People up and down the supply chain try to rip each other off. It seems that the author did not expend much imagination in looking for prohibition-driven crime.
Unless you think a violent drug dealer is less likely than a nonviolent offender to ever have been charged with a drug offense, the data doesn't bear this out. According to the author's data, it seems that most violent offenders have never been sentenced for a drug charge.
Violent offenders are younger on average where drug offenders come in a wider age range. So, on that basis alone you would expect violent offenders to be less likely to have a drug charge than the population as a whole.
If a violent offender that was also charged with a drug offense is classified in the "Violent" category, how can you draw any conclusions about how likely they are to have ever been charged with a drug offense? The data seem to only show that you are unlikely to end up in prison with a drug offense as your most serious crime.
This jives with David Simon's description of the prevailing attitude of enforcement in The Corner. The justice system is too overwhelmed to press charges for mere drug offenses, so you are less likely to end up with a conviction without a more serious crime attached.
This is discussed towards the end of Section 2, where he attempts to establish that the set of inmates charged with violent offenses is mostly disjoint with the set of inmates charged with drug offenses.
Thanks for the pointer. Are you referring to Sections II C&D? Tables 4A and 4B seem like they would have the same categorization issue I described above. As I read it, the "Never drugs" column refers to a given offender never having a drug offense as the most serious offense for a conviction. This doesn't mean that the offender was never charged with a drug offense.
