Community violence to prison assault: a test of the behavioral continuity hypothesis.

This large-scale study (N = 23,277) investigated the relationship between criminal history in the community and serious or assaultive prison misconduct, while controlling for the effects of inmate characteristics, general criminality, and custody level. Community violence variables included the rate of prior violent crime arrests and the types of prior violent crime, as well as a range of specific violent crimes of conviction. Behavioral continuity from community to prison was neither simple nor intuitively discernible, depending on the type, recency, and pattern of community criminality. Application of logistic regression models revealed that the omnibus measure of the rate of prior violent arrests was not related to either serious or assaultive prison misconduct. Prior arrests for assault and current convictions for robbery and/or assault, but not prior or current homicides, were associated with an increased risk for prison violence. Current conviction for a sexual assault had the strongest inverse relationship to prison violence, while prior arrests for sexual assault showed no relationship to prison violence. A more "nuanced" approach in assigning risk ratings based on prior criminal history and seriousness of offense is recommended.