ABSTRACT
Measurement matters. Measuring the severity of prison sentences is fundamental to analysing and interpreting sentencing practices in US courts. The issue is finding a measure of sentence severity that fits the actual decisionmaking practices of judges. Since convicted offenders serve a period of incarceration usually stated in months or years, the uttered sentence offers the obvious choice. Indeed, many see the uttered sentence as the natural dependent variable for explaining the severity of prison sentences. In this article, however, we show that using months or years as a measure of sentence severity causes problems for statistical analysis, and may result in misleading and inappropriate policy conclusions.
