Fig. 5
From: Use-inspired basic research in medical image perception
Comparison of prevalence effects in non-expert and radiologist observer populations. In both cases, miss error rates rise at low prevalence while false alarms fall. Non-expert data are replotted from Wolfe et al. (2005). Radiologist data are replotted from Evans, Birdwell, and Wolfe (2013). Differences are statistically significant, as detailed in the original papers