Mapping voting patterns at the 2004 Australia Federal Election with a web-based application
This paper develops a framework for spatial and statistical modelling and on-line visualisation of voting patterns at the 2004 Australia federal election at the polling booth level. The locations of polling booths are geocoded with national digital datasets. The objective is to show patterns of voter support for political parties, and to relate those patterns to demographic and socio-economic characteristics. This paper also describes a prototype Web geographical information system (GIS) for mapping voting patterns at the federal election. A classification functionality that consists of Equal Interval, Quantile, Natural Breaks and Location Quotient to generate different thematic displays has been added to the Web GIS application for identifying inherent patterns and level of voter support for a political party received through the primary votes or two party preferred casts for candidates standing for a party. The polling booths outcomes which may be expected on the basis of demographic and socio-economic data for booth catchments are modelled using discriminant analysis, and these expected outcomes can be visualised using the Web GIS application. This paper also compares the outcomes of discriminant analysis in distinguishing between political parties at the 2001 and 2004 federal elections. The development of the modelling and visualisation framework helps to explain Australian electoral geographies.