Constructing and Evaluating Contextual Indices Using GIS: A Case of Primary School Performance Tables

Abstract

School performance tables emphasising aggregate examination scores have become an enduring feature of the educational landscape. These tables are problematic, even flawed, as a guide, given the recognised broad link between pupil performance and the social and economic environment in which they live. There is continued interest in being able to contextualise school examination scores so as better to reflect relative achievement. The inherent spatiality of inequalities lends itself to analysis using geographical information systems (GIS), particularly in the task of creating context from geodemographic and lifestyle data. In this paper I explore a methodology for creating and analysing a contextual index of ambient disadvantage centred on robust normalisation of data and illustrate this by using census variables, pupil numbers, and test scores for 3687 primary schools in the north of England. Relevant census variables are interpolated using ordinary kriging with an element of smoothing so as to simulate, to some extent, the effect of school catchment areas. Key features of using robust normalisation are that variable weights can be tested and the internal level of support for an index, the weighted absolute deviation, can be calculated and mapped. This latter quantity provides a quality measure for an index. The methodology is critically assessed in relation to other recent approaches.