Figure 7

Relative bias of the target concentration estimates under theoretical assumptions and misclassification. The relative bias is calculated as $\frac{\widehat{\lambda }-\lambda }{\lambda }$ for 1000 simulated experiments of 8 replicates without any additional sources of variation. As results are relative to the true concentration λ, the precision of different dilutions of the same sample can be assessed on the same scale. Results were plotted for different misclassification probabilities (FPR, FNR) with FPR = false positive rate = 1-specificity and FNR = false negative rate = 1-sensitivity. False positives have considerable influence on the estimates for low concentrations, while false negatives substantially influence the results for highly concentrated samples.