Network distance and travel times are two popular methods of measuring potential geographic accessibility and networks are also used in gravity model-based approaches such as floating catchment area (FCA) techniques. Although some research has been conducted to assess the effectiveness of the representation of demand- (population) or supply- (destinations) side characteristics within such models, there have been few attempts to assess the implications of using alternative sources of network data.
This study employs a sensitivity analysis approach to assess accessibility to GP surgeries in south Wales using proprietary and open sources of network data. Results suggest that there are significant differences between access scores derived from the use of networks which purport to portray the same features. Furthermore, the pattern of differences varies between urban and rural areas. Case studies are used to show that the actual representation of network-based features, often overlooked in previous research, can have important implications for the findings from such studies.
We conclude by suggesting that the use of sensitivity analysis to assess geospatial data usability has a wider relevance for studies that involve the use of a range of GIS-based techniques in different application areas.