“…
We used signal detection modeling of a large data set to show that forensic cartridge-case examiners were highly unlikely to commit false positives or false negatives but appear to be strongly biased against true negatives (Smith & Wells, 2024). Specifically, forensic cartridge-case examiners were far more likely to claim that the results were inconclusive when the shell casings did not match than to say the results were inconclusive when the shell casings matched.
…”