Confirmation bias in criminal cases
Moa Lidén’s 2018 doctoral thesis Confirmation Bias in Criminal Cases studies Swedish police officers, prosecutors, judges, appeals, and applications for a new trial. The local PDF is the 284-page Uppsala University thesis.
The experiments found more guilt-presumptive questions after police had apprehended a suspect; less prosecutorial interest in further investigation and more guilt-confirming proposals after a charging decision; and stronger judicial assessments of prosecution evidence when the same judge had earlier ordered pre-trial detention.
The archive study estimated that far from all wrongfully convicted people who appeal or seek a new trial are acquitted, and reported that this conclusion survived a wide range of assumptions about unknown parameters. The result directly challenges the use of successful exonerations as the denominator for system error.
The thesis supports Rättssäkerhet in Swedish criminal cases, Swedish remand detention and restrictions, and False accusations as an epistemic risk.