While prosecutorial charging decisions occupy a central position in the American criminal legal system, data access has severely constrained related empirical research. In a recent paper, The Police-Prosecutor Charging Decision, Shima Baradaran Baughman et al. (BYU) report initial descriptive results that benefit from "rare" 2019 charging data from a single (anonymous) "mid-sized" U.S. county.
Specifically, the data include a "random sample of 414 unique cases files and 1340 charges [from 2019]. For every charge, the dataset links three decision points: (1) the offense(s) recommended by the arresting agency, (2) the offense(s) the prosecutor filed, and (3) the disposition for that charge, indicating how the case was concluded whether by plea agreement, dismissal, a guilty verdict of the original charge, or a guilty verdict of a lesser charge.” Linking police-recommended charges, prosecutor-filed charges, and charge-level dispositions within the same cases provides an unusual glimpse into the critical arrest-to-filing transition.
As Figure 2 (below) implies, except for weapon offenses, "police and prosecutors generally converge on similar severity levels by offense type, and that divergence in charging severity is concentrated in specific categories rather than broadly distributed across the docket."
Figure 2: Severity by charge categories for police and prosecutors
