Figure 6
The precision of the means submitted to the DARR. This is represented as a deviation of the reported means from the median mean in percent for each population for each dataset, separated by number- (blue) and volume-weighting (red), each referenced to their respective median means ( and ). These data are represented as boxplots (showing quartiles, with the whiskers indicating the 95% confidence interval) accompanied by the individual datapoints. The plots are shown on a symmetrical logarithmic scale with the lighter region representing a linear region covering −10 ≤ (%) ≤ 10. Both populations of each dataset contain their absolute median mean and number of entries in the inset box. A larger spread of the submitted values can be observed for the more challenging powder-based samples compared with the dispersed samples. |