Map users often turn to maps (as opposed to a tabular version of the data) to help them achieve some understanding of the spatial or spatio-temporal patterns that are present in the data in which they are interested. Jacquez and Greiling (2003) have described a useful framework for thinking about the potential types of map pattern analysis that map users might perform. They identified three main types of questions that users may try to answer from their data: those about value (e.g., what values are present in the data and how are they arranged spatially?); change (e.g., how do values change through time (and space over time)?); and association (e.g., how are values related to one another over space?).
Maps are often used in conjunction with other statistical graphics for exploring patterns (e.g., scatterplots, parallel coordinate plots, time series graphs; see the Integration of Maps and Information Graphics concept gallery item in Lesson 8 for more information on this topic). Dynamic and interactive maps, as well as dynamic and interactive statistical graphics (i.e., electronic maps or graphs in which map readers can change classification and symbolization schemes) can help map readers spot patterns by highlighting or focusing on subsets of the data (see Figure 5.cg.16, below). Viewing multiple static maps that show combinations of variables or combinations of classification schemes can also prompt map readers to notice unusual or unexpected patterns.
Although maps may be quite useful for noticing unusual or unexpected patterns in the data, hopefully, by working on the project in this lesson, you will have experienced how changing the classification scheme or the symbolization scheme used for creating a map can have an impact upon a map reader's visual perception of a map pattern. This raises the question of whether the pattern is real or is only an artifact of the display parameters (i.e., is that cluster of similarly valued points really unusual, or could it have happened by chance?). For this reason, it is common for researchers to 'follow-up' on visual patterns they have spotted during data exploration with statistical hypothesis testing techniques to provide more evidence that the pattern is real. Some examples of techniques that are commonly used for spatial analysis include the Geographical Analysis Machine (GAM), which was used to try to demonstrate a connection between the number of cancer cases and potential exposure to radioactivity near nuclear power plants in Great Britain (Openshaw et al. 1987) and the spatial scan statistic, which can identify both spatial and spatiotemporal clusters of events (Kulldorf 1997).
If you are interested in investigating this subject further, I recommend the following:
- Andrienko, N. et al. 2001. "Exploratory Analysis of Spatial Data Using Interactive Maps and Data Mining." Cartography and Geographic Information Science. 28(3): 151-165.
- Boscoe, F.P. et al. 2003. "Visualization of the spatial scan statistic using nested circles." Health and Place. 9(3): 273-7.
- Guo, D. 2003. "Coordinating computational and visual approaches for interactive feature selection and multivariate clustering." Information Visualization. 2(4): 232-46.
- Jacquez, G.M. and D.A. Greiling. 2003. "Geographic boundaries in breast, lung and colorectal cancers in relation to exposure to air toxics in Long Island, New York." International Journal of Health Geographics. 2(4).