Regression Analysis for Spatial Data (Boulder, CO)
Spatially-referenced data add important contextual and locational information to the social and behavioral sciences, such as sociology, anthropology, political science, and public health. A geographic or spatial-analytic framework, which will be taught in this workshop, can be used to explore the importance of spatial relationships in a variety of social and behavioral processes. However, spatial data and spatial relationships necessitate the use of analytic tools beyond those provided by standard statistical methods. This course introduces the spatial-analytic framework, and explores the range of issues that generally must be dealt with when analyzing spatial data. The role of spatial autocorrelation in spatial data sets is a central. Throughout the course we will address the following questions: how and why does spatial autocorrelation arise; how is it measured and understood; how does it relate to issues of spatial heterogeneity and spatial dependence; and how should it inform the specification and estimation of regression models? Specific modeling techniques include: indices of spatial autocorrelation (Moran's I, Geary's C, LISA), spatial regression models (SAR and SEM), geographically weighted regression (GWR), and conditional autoregressive models (CAR). We will examine how these models are similar/different and provide valuable information on the type of processes driving spatial relationships in data.
The course is structured around a combined lecture format (mornings) and computing lab exercises (afternoons). The focus of the course is on spatial statistical analysis, not Geographic Information Systems (GIS). Hands-on application of statistical methods in afternoon lab sessions will enable participants to pursue a broad range of social and behavioral science research topics. Software emphasis will be given to GeoDa and R for exploratory spatial data analysis and modeling. Some acquaintance with this software is helpful but is not a prerequisite. Detailed R code will be provided and discussed in labs, and we will learn how to interpret, visualize, and map model output.
Prerequisites: Prerequisites for maximizing learning in this course are a solid grounding in standard multivariate regression techniques and a minimal level of comfort with matrix notation and algebra..
Fee: Members = $1700; Non-members = $3200