We conducted a study of just two offenses over the span of a few years. We selected the two federal offenses that we believe replicate most closely their state counterparts—arson and robbery. We compare those results to the same two offenses brought in state courts in New York over the same time period. Our results are most applicable to those federal offenses that replicate state offenses—those concurrent jurisdiction offenses where the federal interest in pursuing these charges is the same as the states’ interest. However, we believe that these results give us a window into federal prosecutorial decision-making in general, at least for those classes of crimes that are not restricted to federal courts.
In Part II.A, we describe our study of federal cases from the database kept at the U.S. Sentencing Commission (USSC), and in Part II.B, we detail the New York data provided to us by the New York State Division of Criminal Justice Services, as well as the multistate statistics we obtained from the FBI’s Uniform Crime Reports. In Part II.C, we compare the federal and state statutes to ensure that our premise—that the state and federal arson and robbery offenses are essentially identical—is accurate. We describe the additional multistate data that we use from the FBI’s Crime in the United States series in Part II.D.
In Part III, we offer the results of our comparisons and some tentative conclusions about why federal prosecutors charge cases that could be brought in either state or federal courts. In Part III.A, we describe the process of combining our two data sets into a single computer program that allows us to compare and contrast variables common to both sets. In Part III.B, we are primarily looking at contingency tables and associated chi-squared or Fisher’s exact test probabilities. However, we also engage in mean comparisons and t-tests for the age and sentence variables. In Part III.C, we use a logistic regression model to analyze our data. We find that the most significant determinants of whether a case would be brought in federal court were an investigation by a federal agency or joint task force, serious recidivism of the defendant (number of total prior arrests that were for violent offenses), the high value of the items stolen during a robbery, the value of the property destroyed by arson or explosion, the use of a weapon during the crime, cooperation with the government, involvement in conspiracy, cooperation with the government after arrest, involvement of a minor victim, and age of the defendant. Murder and arrests for minor offenses were factors that might point toward state charging. Neither U.S. citizenship, gender, nor black or white race was significantly related to federal versus state involvement. When recidivism was controlled for, the fact that a defendant was either black or white did not clearly make it more or less likely that the case was brought in federal court. We also find sentences to be significantly higher at the federal level for both offenses.
While we can offer the statistical differences between the two data sets, this will, of course, not prove the motivation behind the federal prosecutors’ case selections. Ascribing those factual differences we find between state and federal cases as the rationale for the selection presumes that the federal prosecutors knew, understood, and cared about these factual differences in advance of making their selections. We believe this to be the case. As we show in Part IV, scholarship and statistical information both within our study and outside of our study suggest that sentence lengths and conviction and guilty plea rates nationally are higher at the federal than the state level. That knowledge likely motivates federal actors to bring cases with a particular federal interest (e.g., crimes involving high-dollar values and professional criminals who work in groups) against the worst offenders (the ones who have already been convicted of serious felonies at the state level but are back on the streets) to federal court, where they will get a stiffer sentence and be assured of a conviction by trial or plea.