DISCUSSION Tosummarizethefindings: • Subjects who customized the DSS agreed more with the recommendationsofaDSS,bothinabinary”yes/no”type decision and in a more continuous ”how much” decision, even though their customization had no actual impact on therecommendations. • ThisbiasledtomuchworsedecisionswhentheDSSgavea poorqualityrecommendation,andslightlybetterdecisions whentheDSSgaveagoodrecommendation. • CustomizingtheDSSdidnotleadsubjectstobemorecon- fidentintheirdecisions. • Theillusionofcontrol,ForereffectandIKEAeffectdonot explainthiscustomizationbias. • Confirmation bias is the best explanation for the bias from these data. Subjects were highly likely to agree with the