Police evaluation of evidence: statistical format and evidence type.

This article investigates how U.S. police officers interpret suspect guilt based on different statistical formats of evidence strength—likelihood ratio (LR) versus random match probability (RMP)—and types of evidence (DNA, fingerprint, eyewitness identification), while controlling for prior odds of guilt. Using a sample of 209 police officers, the study found that officers were generally more accurate and less likely to under-weigh evidence when it was presented in the LR format compared to the RMP format, with this effect being particularly pronounced for DNA evidence. The type of guilt judgment scale used (log frequency vs. percentage) did not significantly influence officers' accuracy. These findings highlight inconsistencies in police evaluations of evidence depending on statistical presentation and suggest that LR formats may better support accurate guilt assessments in investigations.