In their 1997 book, The Criminal Recidivism Process, Zamble and Quinsey stated on page 3, “As a rough generalization, we may say that the level of correct prediction is usually about halfway between chance and perfection….” They we referring to Nuffield’s general recidivism and violent recidivism scale, the LSI, the Violent Risk Appraisal Guide, the Salient Factor Score and the assessments used in Iowa, Wisconsin and Illinois. Would anyone wish to offer a rough generalization for today’s criminogenic risk/needs assessments – specifically the LSI-R or the LS/CMI? Thanks, Russ Fry