Geographic Information Analysis (GIA)
For the UW Center for Demography & Ecology
The GIA program's overarching goal is to maintain CDE's excellence in GIA research, training, and production/management of georeferenced demographic data. We providing the following services to CDE members: (1) expert advice and access to appropriate tools for spatial aspects of existing CDE research projects; (2) active promotion of spatial data and analysis for projects in the planning stages; (3) individual user training and formal classroom teaching of spatial statistical theory and alternative strategies for dealing with spatial heterogeneity and spatial dependence in demographic research.
FAQs
What does it mean to do Spatial Analysis?
Why is Spatial Analysis important to demographers?
How can I tell if there are spatial processes affecting my data?
Where can I go online for more about spatial data analysis?
What does it mean to do Spatial Analysis?
"We consider explicitly the importance of locations, or the spatial arrangement of our observations through modifications, extensions and additions to standard statistical data analytical methods."
Paul Voss
Spatial Statistics for the Social Sciences Lecture (Fall, 2003)
Why is Spatial Analysis important to demographers?
“Sociologists study a wide variety of social, political, and economic phenomena. Many of these phenomena – for example, urbanization, political mobilization, economic development, diffusion of innovations – take place in and are distributed across geographical space. It is reasonable, therefore, to argue that sociologists are interested, indeed have long been interested, in social phenomena distributed in geographical space. Yet, in the main, our theoretical frameworks and data-analytic capabilities do not include the geography of social phenomena.”
Patrick Doreian
“Estimating linear models with spatially distributed data” (1981:359)
Overlooking spatial processes at work within the data often results in model mis-specification where parameter estimates become biased and/or inefficient.
How can I tell if there are spatial processes affecting my data?
In many analyses, examination of the residuals from the model can be a good first step. Spatial patterning such as autocorrelation in the residuals suggests that the assumption of independent identically distributed (i.i.d.) errors has been violated and a model that builds in spatial dependence may be more appropriate. To view a map of autocorrelated errors click here.
Where can I go online for more about spatial data analysis?
The Center for Spatially Integrated Social Sciences serves as an online clearing house for research, tools, and techniques relevant to research in the social sciences.