Fig. 2
From: Analysis of contamination in cluster randomized trials of malaria interventions
Illustration of the sigmoid model function for an example data set of the subsequently described simulation study. Households are arranged based on their distance to the nearest discordant household stratified by intervention status on the x-axis (Δij) and the expected prevalence is shown on the y-axis. The black line indicates the model fit and the gray rectangle the contamination range in both arms. The model is fitted to the same data set as is used in Fig. 1, the detailed parameters are listed there. The patterns in the expected prevalence (such as the approximately linear grouping in the control arm with low expected prevalence) arise from the location of households