In our lab, 331 real judges from seven major jurisdictions (USA, Argentina, Brazil, China, France, Germany, India) spent up to 55 minutes to judge an international criminal appeals case and determine the appropriate prison sentence.

The lab computer randomly assigned each judge (i) a sympathetic or an unsympathetic defendant, (ii) a precedent disfavoring, favoring, or strongly favoring the defendant, and (iii) a short, medium, or long sentence anchor, and logged their use of the materials (fictitious briefs, statement of facts; real trial judgment, statute, precedent). Even extremely strong precedents barely had a detectable effect on judges’ decisions, whereas the legally irrelevant defendant attributes and sentence anchors did. Common and civil law participants differed in their style of written reasons but not their use of the materials, and civil law judges if anything adhered more to precedent.

Joint work with Lars Klöhn, Christophe Jamin, Vik Khanna, Zhuang Liu, Pavan Mamidi, Alexander Morell, Ivan Reidel