highly significantP < 0.0001
The results depicted in Figure 3 - for the TBRs, mf prevalence breakpoints and R 0 - show not only that a wide distribution of values is consistent with the data in each study community when the model is fitted to mf age prevalence but also that they can vary significantly between the study communities [kruskal-wallis tests of differences in estimated values of each variable indicating highly significant study effects in each case ( P < 0.0001 in each case)].