Aggregating Probabilities across Offences in Criminal Law

Abstract

A defendant is charged with four offences, allegedly committed in four different times and places. The probability that he committed each one of the offences is 90%. Assume that the minimum threshold required for conviction is 95%. Under prevailing criminal law the defendant would be acquitted of all four charges since no offence can be attributed to him. However, a simple calculation reveals that the probability that the defendant committed no offence at all is .01% only! Consequently it seems that convicting the person for at least one offence without specifying what this offence is would be just and efficient. We argue that in such cases, and under certain conditions, efficient deterrence will be better achieved if courts would aggregate probabilities across different offences, and would be willing to convict defendants even for unspecified offences. We also show under what conditions aggregating probabilities will yield less, rather than more, convictions.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 17 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.