Abstract
Statistical disease mapping is a valuable public health tool, as it identifies spatial patterns of disease occurrence. However, the Modifiable Areal Unit Problem (MAUP) poses challenges to disease mapping, as the aggregation of geographic units can impact statistical inferences. The effect of the MAUP depends on contextual factors, for example the geographic structure, aggregation level, choice of model, and the underlying data-generating process. We conducted a comprehensive simulation study to understand the role of these factors on the MAUP in the context of Australian disease mapping. We aggregated and rezoned disease count data at a fine geographic scale before fitting spatial and non-spatial regression models to assess the impact of the MAUP on coefficients. To aid the exploration of simulation results, we developed an interactive Shiny application that enables detailed and interactive exploration of the simulation results. This study highlights the need for disease mapping researchers to analyse sensitivity with rezoning and aggregation tools.