Line plots showing the post-warmup MCMC traces for the logarithmic posterior probabilities across two independent chains for models fitted to the filtered strain and serotype data from child carriage and disease.

Each panel corresponds to a model with a different method of associating isolates with an invasiveness estimate: (A) serotype-determined, Poisson-distributed invasiveness; (B) serotype-determined, negative binomially-distributed invasiveness; (C) strain-determined, Poisson-distributed invasiveness; (D) strain-determined, negative binomially-distributed invasiveness; (E) serotype-determined, strain-modified Poisson-distributed invasiveness; (F) serotype-determined, strain-modified negative binomially-distributed invasiveness; (G) strain-determined, serotype-modified Poisson-distributed invasiveness; (H) strain-determined, serotype-modified negative binomially-distributed invasiveness; (I) strain- and serotype-determined Poisson-distributed invasiveness; (J) strain- and serotype-determined negative binomially-distributed invasiveness.

(PNG)