Introduction

Visualization Environments

Show and Tell

Problems and Solutions

Pyramid Exercise

Participants

Show and Tell

Mr S.M.Wise and Prof. R.P.Haining

Example of Visualization of Social Science Data

Our work has focused on methods and tools for the analysis of area-based data, particularly in connection with GIS. The image is a screen shot from a piece of software called SAGE (Spatial Analysis in a GIS Environment) developed by us with funding from an ESRC grant.

• GIS, which are good for display and for certain types of analysis (typically based on spatial queries and map manipulation)
• Statistical or specialist software, which provide good numerical analysis techniques, but are generally weak in display and visualization methods.

SAGE was written to bring together the strengths of both types of software, by linking a purpose written suite of spatial analysis tools with a GIS (ARC/INFO). An important element of SAGE is the provision of tools for supporting visualization, since such tools are extremely important in undertaking exploratory spatial data analysis, and in assessing some elements of model fit in confirmatory approaches.

The image shows a map of Sheffield on the left, divided into 200 regions. Within each region information is available on the proportion of women who have used the breast cancer screening service offered by the local health authority. All women over 30 are eligible to use this service, but the proportion who choose to do so varies widely across the city. One possible explanation is the time taken to reach the only screening unit, located just west of the city centre, from different parts of the city. In order to explore this hypothesis, SAGE has been used to look at the variation in uptake rate of the service with distance from the unit. Each of the boxplots on the right is a summary of the rates in regions at increasing spatial lags from the region containing the unit - the first is for immediate neighbours, the second for second order neightbours etc. The third boxplot has been selected in the graphics window - since the two windows are linked, this causes all the regions to which this boxplot relates to be highlighted on the map. This is a means of checking that lag order is a reasonable approximation of distance from the unit. The graph itself indicates that there is no clear relationship between lag order and uptake, suggesting that distance from the centre is not a major factor in women's decision whether to attend for a scan.

• SAGE has a regionalisation module which was used to group the original 1150 Enumeration Districts into 200 regions, in such a way that the regions are homogeneous in terms of levels of material deprivation, have equal numbers of women eligible for screening and are compact in shape.
• The ability to produce general statistical graphs (histograms, scatter plots, boxplots) and map views of the data
• A client server architecture which links the graphical windows with the map window drawn by ARC/INFO. Highlighting items in one window, causes the relvant items in all other graphical windows to be highlighted.
• Information from the GIS about the spatial contiguity of the areas is used to produce the lagged boxplots.

Contact Details