Simultaneous underweighting and overestimation of rare events: Unpacking a paradox.