Is there really a "wrong side of the tracks" in urban areas and does it matter for spatial analysis?

Mitchell, R. and Lee, D. (2014) Is there really a "wrong side of the tracks" in urban areas and does it matter for spatial analysis? Annals of the Association of American Geographers, 104(3), pp. 432-443. (doi: 10.1080/00045608.2014.892321)

Full text not currently available from Enlighten.


Sharp socioeconomic differences between adjacent neighborhoods run counter to Tobler's first law (TFL) of geography and call into question the blanket application of smoothing techniques designed to handle spatial autocorrelation. In a recent project, large socioeconomic differences between adjacent neighborhoods were observed coinciding with physical features at the neighborhood boundary such as rivers, parks, railroads, and highways. Literature on urban form suggests mechanisms by which these features might create or maintain socioeconomic differences. We therefore test whether the presence of physical features on neighborhood boundaries is associated with greater socioeconomic disparity between the neighborhoods and whether the types of features less easily crossed are more strongly associated. The study area was the city of Glasgow, Scotland. We used vector data to determine which of N = 1,914 neighborhood boundaries coincided with physical features, a well-validated measure of multiple deprivation to assess differences in socioeconomic character across these boundaries, and linear regression to assess associations. The presence of physical features was weakly associated with greater socioeconomic difference across neighborhood boundaries (B = 0.193, p = 0.006). Water (rivers/canals; B = 0.378, p = 0.005) and open spaces (B = 0.283, p = 0.016) were most strongly associated. The presence of physical features, however, was neither necessary nor sufficient for large interneighborhood differences in socioeconomic character. We thus confirm that TFL is not infallible and suggest that spatial analysts need to be concerned about the blanket application of spatial smoothing. Physical features do not hold influence of sufficient size or consistency to guide when and when not to smooth values in spatial analysis, however.

Item Type:Articles
Glasgow Author(s) Enlighten ID:Mitchell, Professor Rich and Lee, Professor Duncan
Authors: Mitchell, R., and Lee, D.
College/School:College of Medical Veterinary and Life Sciences > School of Health & Wellbeing > Public Health
College of Science and Engineering > School of Mathematics and Statistics > Statistics
Journal Name:Annals of the Association of American Geographers
Publisher:Taylor and Francis
ISSN (Online):1467-8306

University Staff: Request a correction | Enlighten Editors: Update this record