This panel presents an overview of the concepts and methods used in
operationalizing a multilevel model of spatial differentiation for the analysis of social and
economic patterns in contemporary China. In this first paper we introduce the theoretical
concepts behinds the model and review the spatial and statistical data prepared to
construct it. The papers that follow in this panel will demonstrate how these data sets are
used in central place analysis, constructing an urban-rural continuum, delineating regional
systems and core-periphery structures, classifying settlement types; and positioning
households in hierarchical regional space. A second panel will introduce substantive
applications of this model to research in economics, anthropology, and sociology.
We refer to the spatial framework described by this model as Hierarchical
Regional Space or HRS (Skinner, 1994). HRS elaborates on some fundamental elements
of modern geographical thought, including central place theory, spatial autocorrelation,
regional systems theory, and diffusion theory. For agrarian societies, Christaller’s (1933)
central place theory predicts the emergence of a hierarchy of settlements, with each level
of the hierarchy providing distinctive services and attaining corresponding levels of
development. Individuals on the landscape orient their economic activities to specific
central places at each hierarchical level in accordance with the services provided there: to
a nearby town for daily purchases, to cities for products like furniture needed with less
frequency, and to metropolises for specialized services such as higher education. It
follows that the hierarchical levels providing less common services require a
correspondingly wider hinterland, and that for reasons of transportation efficiency the
hinterlands at each level become nested. Economic activities in this hierarchy develop
hand in hand with a web of social networks, making a central place analysis a useful
starting point for an investigation of social patterns.
In this investigation we take advantage of another critical feature of agrarian
societies, namely, that most economic, social, and demographic variables are spatially
autocorrelated, i.e., that they co-vary systematically through regional space. Any number
of social and economic variables collected from regions across the landscape can be
shown to be autocorrelated, a fact that confounds many standard statistical procedures.
Our approach to this problem is to make explicit the spatial relationships among regions