A key concern of geography and other disciplines which make use of spatiallyreferenced data is with differences between places. Whether the object of study is human populations or geomorphology, space is often of fundamental importance. We may be concerned with, for example, factors that effect unemployment or factors that influence soil erosion; traditionally, global methods have often been employed in quantitative analyses of datasets that represent such properties. The implicit assumption behind such methods is that properties do not vary as a function of space. In many cases, such approaches mask spatial variation and the data are under-used. The need for methods which do allow for spatial variation in the properties of interest has been recognised in many contexts. In geography and cognate disciplines, there is a large and growing body of research into local methods for spatial analysis, whereby differences between places are allowed for. This book is intended to introduce a range of such methods and their underlying concepts. Some widely-used methods are illustrated through worked examples and case studies to demonstrate their operation and potential benefits. To aid implementation of methods, relevant software packages are mentioned in the text. In addition, a summary list of selected software packages is provided in Appendix A. The book is intended for researchers, postgraduate students, and profession-
als, although parts of the text may be appropriate in undergraduate contexts. Some prior knowledge of methods for spatial analysis, and of Geographical Information Systems (GISystems), is assumed. Background to some basic concepts in spatial data analysis, including elements of statistics and matrix algebra, is provided by O’Sullivan and Unwin (304), de Smith et al. (104), and Lloyd (245). There is a variety of published reviews of local models for spatial analysis (23), (127), (128), (367), but each has particular focuses. One concern here is to bring together discussion of techniques that could be termed ‘local’ into one book. A second concern is to discuss developments of (relatively) new techniques. This chapter describes the remit of the book before introducing local models
and methods. Then, the discussion moves on to issues of spatial dependence, spatial autocorrelation, and spatial scale. The concept of stationarity, which is key in the analysis of spatially or temporally referenced variables, is also outlined. Finally, key spatial data models are described, and the datasets used for illustrative purposes are detailed.