First, let's plot the raw number of arrests by age group. I tossed out the retired players (we can hardly blame the NBA for them, can we?) and grouped the arrests by age, FBI-style. I then plotted the curves on two separate y-axes for easy comparison. Just click on the graphs to bring up the full-size versions.
By the time they get to the NBA, most players are already past their peak offending years. Still, the NBA peak comes relatively early: 23 versus 19 for US men overall. This made me think about age and selection into pro ball. Sadly, there are a lot more 23-year-olds than 41- year-olds in the NBA these days. So what we really need are age-adjusted rates. Here's what happens when we standardize by the number of players in each age group (using 2003 roster data).
The age-adjusted rate shows the mid-career blip that McCann mentioned. Even without the retirees, older players do get arrested (e.g., Gary Payton's recent DUI). Moreover, McCann seems to be spot-on about the comparatively clean records of 18, 19, and 20 year-olds in the NBA . Still the first two figures are somewhat misleading, since the x-axis shows single years for 16 to 24 but 5-year increments thereafter and the NBA arrests are plotted on a different scale than the US male arrests. So, here's what the age-adjusted arrest curve looks like for NBA players versus the rest of us.
No wonder coaches keep a close eye on their 23- and 24-year-olds. This is the only piece of the distribution where ballplayers are more likely to get arrested than regular Joes. Why? I'm thinking that off-court activities of 18-year-olds center around in-room Playstation. By the early twenties, however, wealthy young athletes probably venture out of the team hotel a bit more. So, the data are a bit sparse, but I think there's enough here to draw two conclusions: (1) the age-crime curve applies to the NBA as elsewhere (it might not be invariant, but ...); and, (2) McCann seems well-justified in challenging the NBA's age floor for 18-19 year-olds.