Trial Penalties in Federal Sentencing: Extra-Guidelines Factors and District Variation

Federal defendants who exercise their right to trial and lose receive a “trial penalty” in the form of a harsher sentence. Prosecutors impose trial penalties either by refusing to stipulate to relevant conduct that might otherwise reduce exposure to punishment or by filing more-severe superseding indictments against defendants who refuse to plead guilty. Using case-processing, trial-rate, and sentencing-outcome data from eighty-nine federal district courts between 2000 and 2002 (when application of the United States Sentencing Guidelines was still mandatory), this study constructed a multivariate regression model analyzing the effect of trial conviction on sentence length. The model also assessed the effects of Sentencing-Guidelines-based departures, including “substantial assistance to law enforcement,” “acceptance of responsibility,” and “obstruction of justice.” The study confirmed the existence of a significant trial penalty, two thirds of which could be attributed to the Guidelines departures. As the Guidelines’ minimum sentencing recommendations increased, recorded trial penalties increased proportionately. The penalty varied noticeably across district courts, offense severity and type, and defendant characteristics such as criminal history, race, ethnicity, and gender. The severity of the trial penalty is correlated with a district court’s organizational efficiency; penalties increased as caseload pressures increased but decreased as trial rates increased. The study concedes that some dimensions of the trial penalty could not be captured within the regression.

“[W]e confirm both that a substantial trial penalty exists in federal sentencing. While just under two-thirds of it is attributable to Guideline-based factors, a significant 15% sentence length difference on average separates those who plead guilty and those convicted by trial. Trial penalties are also conditioned by court context and defendant-related factors.” p. 584