Impact of polygon geometry on the identification of economic clusters

Autor: Neil Reid, Joseph Frizado, Michael C. Carroll, Bruce W. Smith
Rok vydání: 2008
Předmět:
Zdroj: Letters in Spatial and Resource Sciences. 2:31-44
ISSN: 1864-404X
1864-4031
Popis: Economic clusters have been delineated using Local Moran’s I and Getis-Ord G * because they distinguish relationships across areal unit boundaries within a specified neighborhood. A problem using spatial statistics with U.S. county data are the great variations in county sizes. We examined the relationship between the values for Local Moran’s I and G * , in groups of counties of differing size. The impact of county size on both spatial statistics using a contiguity spatial weights matrix and an inverse centroid distance matrix are assessed. In small counties, the choice in spatial weight matrices is immaterial, especially when using Local Moran’s I. For large counties the differences between the spatial weights methodologies is more apparent, due to edge effects being more prevalent. Selection of an optimal combination of spatial weight methodology and clustering statistic should depend on the study’s purpose, the distribution of county sizes, and the industry being studied.
Databáze: OpenAIRE
Externí odkaz: