Advice Stakes and Adherence Rates
Advice stakes slightly decreased adherence, but the drop was modest (i.e., people didn’t strongly recalibrate based on stakes). In their analysis of “higher stakes vs lower stakes,” the study reports that higher stake...
Advice stakes slightly decreased adherence, but the drop was modest (i.e., people didn’t strongly recalibrate based on stakes). In their analysis of “higher stakes vs lower stakes,” the study reports that higher stakes advice was followed slightly less than low stakes advice in the experimental conditions, with a negative effect : β = −.13 (95% HPDI = [−.19; −.06]) .[:cite[1]{ln=8}], [:cite[1]{ln=12}] They also report a formal GLM result where advice stakes were negatively related to advice following : β = −.10 (95% HPDI = [−.15; −.05]) .[:cite[2]{ln=10}], [:cite[2]{ln=13}] Even at high stakes, adherence stayed fairly high: “advice rated as ‘high’ or ‘very high’ stakes was followed in more than 65% of cases” .[:cite[3]{ln=9}] The paper summarizes the “stakes calibration” implication directly: advice following remained above 60% even for severe personal issues and high stakes recommendations , suggesting users only mildly change reliance depending on potential consequences .[:cite[4]{ln=3}], [:cite[5]{ln=3}] Bottom line: higher stakes reduced adherence a bit, but adherence stayed high (over 60% and 65% in high/very high stakes), so the effect wasn’t strong.