Fig. 3

a Illustrates miss rate as a function of rating quantile and prevalence, with green and purple representing the low and high prevalence conditions, respectively. b illustrates the LPE as a function of quartile, calculated by subtracting miss rates in the high prevalence condition from the low prevalence condition. The dashed horizontal line represents no LPE. In both figures, each panel represents an experiment, and error bars represent the bootstrapped 95% confidence intervals. Similar to results of the logistic regression, miss rates decreased as hazardousness ratings increased. Although the magnitude of the LPE for the most hazardous quartile (Q4) is smaller than the least hazardous quartile (Q1), the miss rates under low prevalence remain approximately twice as high as miss rates under high prevalence condition (see Additional file 1: Fig. S1). Moreover, removing feedback decreased the LPE such that the LPE was not significantly different from zero at any quartile, and there was no evidence of PICC. However, there was a significant LPE overall, but it may be smaller than what can be detected with the decreased power in each quartile