Twenty years of operational use of a risk prediction instrument: The United States parole commission's salient factor score

Abstract

Although empirically-based recidivism prediction instruments were developed as far back as the 1920s, the adoption of the Salient Factor Score by the federal parole board in 1972—as part of a system of explicit parole decisionmaking guidelines—marked the first time that such an instrument was used in a way that had a definite, measurable impact on paroling decisions. The Salient Factor Score has been used in federal parole decisionmaking continuously for the past twenty years. It is axiomatic that a prediction instrument, particularly an instrument used in actual case decisionmaking, be revalidated periodically to ensure that it has retained predictive accuracy. In this article, the predictive accuracy of the Salient Factor Score over time is examined using data on three, large random samples of federal prisoners released in 1970–1972, 1978, and 1987. In addition, the relationship of the Salient Factor Score to the Criminal History Score of the new federal sentencing guidelines—which apply to defendants convicted of federal offenses committed on or after November 1, 1987—is discussed.