Fig. 1

a The spatial distribution of clusters (n = 5839) in the 20 surveys in nine countries. b A scatterplot of the intra-class correlation coefficient and the modelled estimate of prevalence for the three child morbidity indicators [fever prevalence, ITN use and malaria prevalence based on rapid diagnostic testing (RDTs)]. Each data point in the scatter represents a survey. ρ refers to the variability between clusters and shows that surveys with low prevalence exhibit small between-cluster variance. c A scatterplot showing the decrease in ρ as Effective Sample Size increases with a Bayesian 95% credible interval