Debiasing training reduces confirmation bias in national risk analysts

Abstract

State risk forecasts are crucial for allocating resources to address international and domestic threats such as war, pandemics, and climate change. These risk forecasts largely rely on human judgment, which is often susceptible to cognitive biases. We conducted an experiment involving the majority of national risk analysts in a European country and a matched sample of masters students to compare their susceptibility to confirmation bias and bias blind spot in judgments related and unrelated to national risk. Additionally, we evaluated the effectiveness of a one-shot debiasing training intervention across both samples. We find that analysts exhibit less confirmation bias than students within and outside risk-related judgments. Crucially, a one-shot debiasing training session reduced confirmation bias in both analyst and student groups. These findings suggest that cost-effective debiasing interventions can improve expert judgment in national risk forecasting and provide evidence that experience and expertise reduce cognitive bias more broadly than previously recognized.