Fig. 3From: Analysis of contamination in cluster randomized trials of malaria interventionsRelative bias (upper plots) and empirical standard error (lower plots) for model fits. The graphs on the left illustrate the errors for the parameter estimation \(\hat {E}_{s}\) for each of the three models and the graphs on the right for the contamination range \(\hat {\theta }\) that only exists for the sigmoid models \(\mathcal {S}\) and \(\mathcal {S}_{RE}\). Depicted on the x-axis are 10 different levels of parameter variation for Es and θBack to article page