| |
The presence of spatial pattern in a mapped variable due to geographical proximity. The most common form of spatial autocorrelation is where similar values for a variable (such as county income levels) tend to cluster together in adjacent observation-units or regions, so that on average across the map the values for neighbours are more similar than would occur if the allocation of values to observation-units were the result of a purely random mechanism. More general and complicated forms of autocorrelation can also be defined. The presence of spatial autocorrelation is very widespread, and violates a basic assumption (that the observations be \'independent\' or non-autocorrelated) of many standard parametric statistical tests (cf. general linear model). A variety of tests for spatial autocorrelation in raw data and regression residuals is available, and a group of techniques known as \'spatial econometrics\' has been developed to allow the inclusion of autocorrelation in statistical models. These techniques are now being employed in many different social sciences. (LWH)
Suggested Reading Anselin, L. 1988: Spatial econometrics. Dordrecht and Boston: Kluwer Academic. Bailey, T.C. and Gatrell, A.C. 1995: Interactive spatial data analysis. Harlow, Essex: Longmans. Haining, R. 1990: Spatial data analysis in the social and environmental sciences. Cambridge: Cambridge University Press. |
|
