Current evaluating programs control for factors such as age, sex, offense, and criminal history. However, it is just as important to gather information about the stage of the correctional process the offender has passed through and his particular experience in that process. To illustrate the point, a failure-rate analysis of recidivism is presented of 468 men released from the Iowa State Penitentiary at Fort Madison. In this analysis, the length of time to failure indicated by arrest, conviction, or revocation and the return to prison can be subject to various interpretations. For an offender working with a single contact, such as probation or parole, the evaluation is clear cut; but for an offender who is a revoked probationer, a successful work releasee, a failed parolee, but a successful citizen and former inmate, the evaluation is dependent on the stage at which the case is examined. The development of valid theories in corrections should be commensurate with the measurement of what works and for whom. 1 figure and 20 references (Author abstract modified)