Fig. 7From: Analysis of contamination in cluster randomized trials of malaria interventionsReanalysis of the Navrongo trial, where for each cluster, households far away from the intervention boundary were randomly excluded to vary the percentage of households in core ω. The estimated parameters (upper plots) and width of the 95%CI (lower plots) for a GEE, \(\mathcal {S}\) and \(\mathcal {S}_{RE}\) analysis in terms of the percentage of households in core are visualizedBack to article page