Addressing bias in the use of buffers for focal and geographically weighted analyses

Focal analyses (also known as ‘buffer’ or ‘neighbourhood’ analyses) seek to characterise a location based on its surroundings and are commonplace in GIS applications across many fields and disciplines. However, the implicit assumptions made by researchers in these analyses result in an unintended bias towards the periphery of the focal window, which we term Focal Area Bias (FAB). FAB can have a substantial impact upon the resulting values, and in the most extreme cases can result in paradoxical outcomes. Where geographical weighting functions are used, the interaction between the weighting function and FAB means that it will not have the expected effect, leading to the misinterpretation of results. This research characterises the issue of FAB, before presenting a corrective function to remove it. The efficacy of the proposed corrective function and the spatial characteristics of FAB are then evaluated to demonstrate the importance of this issue. We recommend that researchers and practitioners should consider the impact of FAB when undertaking focal analysis and make use of the corrective functions presented here to remove this issue, particularly where geographical weighting is desired.