It hardly bears saying that in any research, the final findings depend entirely upon how one defines and sets up the measured parameters. For example, the California Department of Corrections and Rehabilitation defines recidivism only as a prisoner returning to custody. This is the most restrictive way of defining it, and the best way if you want to paint the best possible picture.

Why is it the most restrictive? For a parolee to be counted as a recidivist, he must pass through a series of gates and filters, each of which in sequence reduces total recidivism numbers. They must first commit another crime, and be found during the committing of it or after (gate 1), and then be arrested with no technical fouls on the part of the arresting officer occurring along the way (gate 2). From that point forward, all kinds of things (gates and filters) can happen that divert the reoffender from re-incarceration, hence from recidivism reports.

Crimes get bargained down or away, charges get dismissed altogether on technicalities or because prosecutors are just too busy handling other, more serious cases, reoffenders avoid re-incarceration by copping to lesser offenses or agreeing to custody-avoiding diversion programs (such as drug rehab program in one form or another, or socialization, re-entry programs and other 12-Step programs), and so on. The final and most severe filter is that we stop tracking reoffenders after the third year. Why stop there? Why not follow offenders for five or even seven years? That would give us a much more accurate picture of true recidivism.

If recidivism were defined, however, as a righteous re-arrest leading to some form of directed sentence, rather than the most difficult criterion (re-incarceration in state prison), we would, I expect, find recidivism rates strikingly higher. (I would love to see some graduate student do a thesis on that question.)

But why is that question important? Simple answer: Crime rates are driven by five sets of offenders: 1) offenders who have not been arrested (at least not yet); 2) those who've been re-arrested but have somehow escaped prosecution on some technicality or weak evidence; 3) those who've been re-arrested, but who've escaped prosecution by pleading down to a nonprison disposition (they cut a deal); 4) those who've been re-arrested and have agreed to a prison diversion program to avoid a prison sentence; 5) those who've been arrested, incarcerated, and released, and then reoffend, are re-arrested and re-incarcerated. We know that the first set of offenders commit dozens of crimes before their first arrest, but that's not recidivism itself, not yet. So we'll set that group aside. And, oddly, Group 5 is the only group found in recidivism reports.

But what about groups 2, 3 and 4? For reasons unexplained, these three groups are excluded from recidivism statistics. They're criminals with records who go on to commit more crime. How is that not recidivism, and why aren't those numbers tracked and reported as measures of total recidivism? Especially with AB 109 now in place, we need to get an accurate fix on current baseline data. We need to get an accurate downstream fix on how AB 109 is working and affecting our communities in terms of crime rates then as compared to now. We need total recidivism rates, not partial ones, to understand and evaluate crime trends and patterns. How else can we measure the effectiveness of the bold, new, untried, experimental, alternative, homegrown programs our 58 counties are each creating with which to handle their influx of former state prisoners and parolees?

Each of those latter three groups of offenders is a uniquely handled population unto itself, which we need to track to understand. For example, do arrestees who cop to a lesser offense and get off with no incarceration reoffend less or more often after their sentinel crime has been closed? Each of these groups has undergone a different judicial/correctional experience -- so how do the different experiences affect their reoffending after their sentinel crime has come and gone?

We are at the dawn of a new correctional day. We have arrived here for many reasons, and right now, there is vastly more we don't know than know about how to steer ourselves safely into that new era. Even what we think we know isn't all that certain. From painful experience, we know what doesn't and hasn't worked, and we need solid answers to some pretty serious questions. Extreme caution is needed as we pick our way forward through mostly unknown and unmapped territory.

Brik McDill, Ph.D., is a senior supervising psychologist at the California Correctional Institution in Tehachapi.