Prison security classification is intended to recognize heterogeneity in the inmate population with regard to propensity to commit misconduct and to appropriately house inmates with varying levels of violent and/or antisocial behavior while they are incarcerated. The intent of security classification is to increase safety for staff and other inmates, but little is known about the effect of security classification on prison misconduct. Using administrative records of roughly 80,000 inmates in the California Department of Corrections and Rehabilitation (CDCR), this study attempts to identify the relationship between security classification and prison misconduct using a regression discontinuity design. A secondary objective is to provide an analysis of the system of mandatory minimum scores in the security classification process used to determine housing level placement for a sizable minority of inmates.

The results suggest there is little evidence that placement a higher level suppresses or exacerbates behavioral misconduct relative to the next lower level for inmates whose preliminary classification scores are fairly close to the classification score thresholds between minimum security and medium security facilities and between medium and close security facilities. There is some evidence that inmates in maximum security have more misconduct than inmates in close security. The lack of an impact of higher security level placement on behavior (and the evidence suggesting possible criminogenic effects of maximum security placement relative to close security) is interpreted as evidence that the best prediction for how inmates with scores just above the security level threshold will behave if moved to a lower security level is their behavior at the higher security level.

With this interpretation in mind, the relatively good behavior of those inmates with binding mandatory minimum scores is particularly intriguing.  Inmates with binding mandatory minimum placement scores cannot improve their security level below a certain point. They are quite well-behaved and have rates of misconduct that are notably lower than inmates with placement scores that are slightly above or below the mandatory minimum levels across all types of misconduct. These inmates are relatively older, and have relatively low average preliminary scores. Their placement scores are ratcheted up by the mandatory minimum for the purposes of determining security placement.  Given that these characteristics (age and preliminary score) are readily observable to correctional administrators and explain the relatively good behavior of these inmates, these inmates with binding mandatory minimums provide perhaps the best prospects for targeted reforms intended to transfer portions of the inmate population to lower security levels.

The findings of this paper support the following policy recommendations: 1) if CDCR is facing capacity constraints in higher security institutions, small adjustments in the score cutoffs should provide relief without compromising safety; 2) inmates facing binding mandatory minimums who have low preliminary scores are particularly good prospects for moving to lower security levels.  In fact, the system of mandatory minimums should be rethought and perhaps abandoned in exchange for a system that allows for more case-by-case discretion in establishing minimum security levels rather than blanket administrative determinations with broad applicability.