Uncertainty in town centre definition

The Defining Town Centre Statistics Project was aimed at defining town centre boundaries for all towns falling within the M25, and producing economic and statistical data for each boundary. The purpose of such data is for governmental monitoring of local town centre economies and to assist with supporting unsustainable and failing centres. A methodology has been developed at CASA for constructing statistically based boundaries which uses core government datasets: the Annual Business Inquiry---supplied by the Office for National Statistics the Rating Support Application database for floorspace---supplied by the Valuation Office Executive and University College London and unit postcode locations for geo-referencing purposes - supplied by Ordnance Survey. The methodology itself uses kernel density estimation to smooth the postcode referenced town centre activity data which produces a surface---called the Index of Town Centredness---which is then thresholded to produce the town centre boundaries---called Areas of Town Centre Activity. This thesis explores the errors in the process. It provides detailed background to the core datasets, describing their origins and collation processes, and demonstrates how they are used to produce the town centre model. Further to this a broad overview of the model is given, with a section on each type of error---covering errors in the raw data, geo-referencing problems, statistics issues and necessary caveats required to allow the end-user of the statistics to fully realise the potential and flaws of the product. Specialist chapters are provided looking at a number of the main problems in the model. Unfortunately, due to confidentiality, detailed studies of the accuracy of the government datasets are not covered. Therefore, these concentrate on the difference between the model and the reality of town centres improving the geo-referencing by using AddressPoint and assessing the problems with the thresholding method. Finally, recommendations are made for Stage 3 of the model - the national roll-out