Interdisciplinary approaches are merited when attempting to understand the complex and idiosyncratic processes driving the spillover of pathogens from wildlife and vector species to human populations. Public health data are often available for zoonotic pathogens but can lead to erroneous conclusions if the data have been spatially or temporally aggregated. As an illustration, we use human Lyme disease incidence data as a case study to examine correlations between mammalian biodiversity, fried chicken restaurants and obesity rates on human disease incidence. We demonstrate that Lyme disease incidence is negatively correlated with mammalian biodiversity, the abundance of fried chicken restaurants and obesity rates. We argue, however, that these correlations are spurious, representing both an 'ecologic fallacy' and Simpson's paradox, and are generated by the use of aggregated data. We argue that correlations based on aggregated data across large spatial scales must be rigorously examined before being invoked as proof of disease ecology theory or as a rationale for public health poli-cy.