Fig. 2
From: Comparison of Bayesian and frequentist methods for prevalence estimation under misclassification
Estimation errors. (a) Estimation error distributions of the Rogan-Gladen point estimate and the Bayesian estimate (MCMC mean) across all data sets (top left), and across data sets as classified according to the non-truncated Rogan-Gladen estimate (case 1, top right). The Bayesian estimator shows adequate error distributions for the data sets with a truncated RGE (cases 2 and 3, bottom row). (b) Comparison of the estimation errors of the Bayesian mean and the Rogan-Gladen estimate for all data sets. Hexagonal binning is used to deal with overplotting, and the hex gray scale codes for the number of data sets that fall within it. The dashed black line shows a Deming regression of the Bayesian estimation error on the Rogan-Gladen estimation error. Its slope is 0.939 with a confidence interval of (0.938, 0.941)