Fig. 2

Estimates of the parameters in each component of Model 1 (blue) and Model 2 (purple): A Autoregressive component, B Neighbourhood component, C Endemic component, D Other coefficients. The y-axis. Unvax corresponds to the effect of u_{i, t}, the mean proportion unvaccinated over the 3 years before t in i; incid1 and incid2 correspond to the effect of \({N}_{i,t}^1\) and \({N}_{i,t}^2\) the category of incidence in the 3 years before t in i compared to the level of reference (i.e. \({N}_{0,t}^0\)); pop corresponds to the effect of m_{i, t} the number of inhabitants at t in i; area corresponds to the effect of the surface; sin and cos correspond to the effects of seasonality; distance and population correspond to the spatial parameters of the connectivity matrix ω (δ and γ); overdisp is the estimate of the log-overdispersion parameter in the negative binomial distribution of Y_{i, t}. Dots show the mean values associated with the parameters; arrows show the 95% confidence interval. Note different y-axes between graphs