Ignoring the influence of observations on disease dynamics results in misleading inference on infection incidence trends.

<p>True infection prevalence for each county (Acrelandia – AC, Placido de Castro – PC, and Senador Guiomard – SG) is depicted in the top six panels (black lines), together with the estimated 95% credible interval for infection prevalence (red polygons). The bottom panels depict the true and inferred infection incidence (black lines and red polygons, respectively). Because simulations and the fitted models assume that the three counties have the same infection incidence, incidence results are displayed in a single panel. Left and right panels show results from the alternative model (<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1003312#pcbi.1003312.e107" target="_blank">eqn. 10a</a>) and original model (<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1003312#pcbi.1003312.e059" target="_blank">eqn. 10</a>), respectively.